(Advances in Experimental Medicine and Biology 826) Mindy F. Levin (Eds.) - Progress in Motor Control - Skill Learning, Performance, Health, and Injury-Springer-Verlag New York (2014)

Advances in Experimental Medicine and Biology
Series Editors
Irun R. Cohen
The Weizmann Institute of Science, Rehovot, Israel
N.S. Abel Lajtha
Kline Institute for Psychiatric Research, Orangeburg, New York, USA
Rodolfo Paoletti
University of Milan, Milan, Italy
John D. Lambris
University of Pennsylvania, Philadelphia, Pennsylvania, USA
Advances in Experimental Medicine and Biology presents multidisciplinary and
dynamic findings in the broad fields of experimental medicine and biology. The
wide variety in meeting proceedings it presents offers readers multiple perspectives
on a variety of disciplines including neuroscience, microbiology, immunology, bio-
chemistry, biomedical engineering and cancer research. Advances in Experimental
Medicine and Biology has been publishing exceptional works in the field for over 30
years and is indexed in Medline, Scopus, EMBASE, BIOSIS, Biological Abstracts,
CSA, Biological Sciences and Living Resources (ASFA-1), and Biological Sciences.
The series also provides scientists with up to date information on emerging topics
and techniques.
2013 Impact Factor: 2.012
More information about this series at http://www.springer.com/series/5584

Mindy F. Levin
Editor
Progress in Motor Control

Skill Learning, Performance, Health,
and Injury
Volume 826
2123
Editor
Mindy F. Levin
School of Physical and Occupational Therapy
McGill University
Montreal
QC
Canada
ISSN 0065-2598 ISSN 2214-8019 (electronic)

ISBN 978-1-4939-1337-4 ISBN 978-1-4939-1338-1 (eBook)
DOI 10.1007/978-1-4939-1338-1
Springer New York Heidelberg Dordrecht London
Library of Congress Control Number: 2014947961
© Springer Science+Business Media New York 2014

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the
material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology
now known or hereafter developed. Exempted from this legal reservation are brief excerpts in connection
with reviews or scholarly analysis or material supplied specifically for the purpose of being entered and
executed on a computer system, for exclusive use by the purchaser of the work. Duplication of this
publication or parts thereof is permitted only under the provisions of the Copyright Law of the Publisher’s
location, in its current version, and permission for use must always be obtained from Springer. Permissions
for use may be obtained through RightsLink at the Copyright Clearance Center. Violations are liable to
prosecution under the respective Copyright Law.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
While the advice and information in this book are believed to be true and accurate at the date of publication,
neither the authors nor the editors nor the publisher can accept any legal responsibility for any errors or
omissions that may be made. The publisher makes no warranty, express or implied, with respect to the
material contained herein.
Printed on acid-free paper
Springer is part of Springer Science+Business Media (www.springer.com)

This volume is dedicated to Dr. John P.
Scholz, (1950–2013), scholar, colleague,
teacher, mentor and friend. Forever in
our hearts.
Preface
This volume and all the previous volumes of the Progress in Motor Control series
are united by a common theme: understanding the basic principles and mechanisms
underlying the control and coordination of voluntary movement and relating these
findings to gain insights into the impact of pathology and injury on the sensorimotor
control system. This volume contains contributions by scientists invited to participate
in the meeting of the Ninth Progress in Motor Control, held in Montreal, Quebec,
Canada in July 2013. Progress in Motor Control international meetings are held
bi-annually and are the official scientific meetings of the International Society of
Motor Control (ISMC; www.i-s-m-c.org). At the 2013 meeting, there were 28 invited
presentations organized into 6 symposia, and a special pre-Congress symposium on
advances in cerebral palsy research organized by Annette Majnemer and Carol L.
Richards. During the meeting, the Society bestowed its highest recognition, The
Bernstein Award, to Dr. Gregory Orlovsky for his outstanding contribution to the
understanding of the organization of posture and locomotion.
The mission of the ISMC and the Progress in Motor Control conference series
is to promote basic and applied research that furthers our knowledge of the under-
standing of the control of movements in biological systems. An equally important
goal of the Society is to promote knowledge transfer and exchange between basic
and applied scientists, educators, and clinicians. Motor control scientists come from
diverse backgrounds, including bioengineering, biomechanics, biophysics, genetics,
kinesiology, neuropsychology, neuroscience, rehabilitation, speech and language,
and sports sciences. A relatively new addition to this multidisciplinary field, repre-
sented in the current volume, is the area of the performing arts, including dance,
music, and circus arts.
The current volume is divided into four parts. Part I, “Sensorimotor Integra-
tion” includes 6 chapters. Chapter 1 presents an integrated view of motor control
based on referent variables organized into a hierarchical system. The view is sup-
ported by evidence of how multiple motor elements, such as segment positions and
finger forces, produce feed-forward, anticipatory synergy adjustments. Chapter 2
continues on this theme and presents the notion of referent control, which describes
how movement is produced due to the difference between the referent position of
the body segments set by the nervous system and the actual position of the body
vii
viii Preface
segments due to its interaction with the physical environment. The concept is sup-
ported by empirical evidence of changes in corticospinal excitability during voluntary
and involuntary movement. The concept is also extended to provide a new explana-
tion of position sense that is not dependent on notions of ‘sense of effort’or ‘efference
copy’. Chapter 3 describes some of the properties of a basic element of the sensorimo-
tor system: motoneuronal firing behavior. This in-depth chapter describes persistent
inward currents and their neuromodulatory systems that are matched to the basic mo-
tor behaviors of posture, voluntary movement, and locomotion. Chapter 4 continues
on the theme of locomotion with a description of the regulation of limb stiffness
during locomotor tasks. Using a decerebrated cat model, the authors extend their
previous studies to show how pattern generation, endpoint stiffness, and propriocep-
tive feedback are modulated in a task-specific way during upslope and downslope
walking. The role of the vestibular system and neck proprioceptors on the modulation
of these properties is discussed to explain the regulation of whole-body locomotion.
The role of sub-cortical systems in the control of voluntary limb movement in humans
is further elaborated in Chap. 5. Using rapid unexpected target jumps during upper
limb reaching movement and lower limb stepping, the authors show that corrective
movements occur at very short latencies consistent with a sub-cortical visuomotor
mechanism. Finally, Chap. 6 returns to the notion of perception discussed in Chap. 2.
The authors propose a multisensory binding model of the perception of the location
of body segments and question the role of motor simulation in this process.
Part II of the volume deals with the concept of variability in motor control. All
three chapters in this part provide evidence to support the view that kinematic vari-
ability is not simply an undesirable manifestation of the noise within the system. On
the contrary, they suggest that variability is a necessary and desirable consequence
of purposeful control strategies. Chapter 7 is a review of the Uncontrolled Manifold
Model (UCM), by two of its original formulators. The UCM enables researchers
to classify the variability of a redundant number of kinematic degrees of freedom
with respect to their contribution or non-contribution tothe motor goal, and provides
a means by which motor equivalence and self-motion during movement can be in-
terpreted. Chapter 8 presents a methodological approach to study complex skills
that have extrinsic redundancy and dynamic characteristics. The authors provide
examples of how humans find stable solutions to the redundancy problem so that
movements are minimally affected by unexpected perturbations. The third chapter
in this part, Chap. 9, also uses a computational approach to provide an explanation
of how redundancy influences the structure and variability of repetitive movements,
using goal equivalent manifolds. Based on this theoretical approach, the authors
suggest that the nervous system uses a minimum intervention principle to produce
movement that minimizes variables at the level of the task goal.
Part III includes three chapters on the performing arts devoted to dance (Chap. 10),
music production (Chap. 11), and circus performance (Chap. 12). Chapter 10 extends
the concept of action-perception via mirror neurons to action-observation learning
of whole-body movements amongst professional dancers. Chapter 11 explores the
highly specialized training of musicians in multi-sensory and emotional domains.
Musician-specific injuries in the motor domain such as overuse injuries and dystonias,
Preface ix
as well as those linked to psychological mechanisms are invoked as contributing to

such injuries. This new classification, which takes into account both motor and
psychological mechanisms may help define prevention and treatment interventions.
Chapter 12, written by a professional artist and a medical doctor specializing in
prevention and treatment of performance-related injuries, reviews the specific skills
needed by highly specialized circus arts performers in an entirely new domain for
the field of motor control and learning, thereby providing important insights into the
care of these athletes.
Part IV of this volume includes three chapters about motor control approaches to
the understanding of musculoskeletal and neurological injury and recovery. Chap-
ter 13 reviews musculoskeletal disorders of the neck and shoulder region, especially
those related to occupational injuries after fatigue. The concepts of variability and
motor abundance are discussed as ways for the system to restore function in chronic
musculoskeletal disorders. Chapter 14 presents evidence of disorders of motoneu-
ronal threshold control, a concept based on the equilibrium-point theory of motor
control, as a primary mechanism underlying disorders of movement in patients with
acquired brain injury such as stroke. Evidence supporting the role of the corticospinal
tract in modulated motoneuronal thresholds is discussed. This concept may provide
a link between the observations of spasticity and disordered motor control that has
important implications for rehabilitation. Finally, Chap. 15 discusses the sensorimo-
tor contributions to balance and locomotor deficits in people with multiple sclerosis.
The chapter presents novel methods for quantifying stability and adaptability of pos-
ture and gait and discusses how these methods may inform the development of new
treatments for people with neurological dysfunction.
The chapters in this volume present state-of-the art methodologies and insights
into motor control concepts that are essential for the understanding of skilled move-
ment and motor learning in healthy individuals and in those with musculoskeletal and
neurological deficits. The chapters are written for the informed reader who has some
background in movement production, measurement methods, skilled performance
and/or rehabilitation. The volume will be a valuable reference for students and pro-
fessionals in both basic and applied areas of motor control such as neurophysiology,
biophysics, biomechanics and bioengineering, neuropsychology, kinesiology, sports
science, motor disorders, rehabilitation and the performing arts.
Montreal Mindy F. Levin

Acknowledgements
The Progress in Motor Control IX meeting would not have been possible without
the generous support of the Canadian Institutes of Health Research (CIHR), the Net-
work of Rehabilitation Researchers of Quebec (REPAR)—Quebec Health Research
Fund (FRQ-S), the Centre for Interdisciplinary Research in Rehabilitation (CRIR),
the Faculty of Medicine of McGill University, the Universite Laval Chair in Cere-
bral Palsy and NeuroDevNet. We also acknowledge our corporate sponsors: AMTI,
ANT Neuro, DelSys, Jintronix, Noraxon USA, Northern Digital, Magstim, Phoenix
Technology, Polhemus, Rogue Research and Vicon.
Planning and organizing of the meeting was assisted by members of the Sci-
entific Committee: Marcos Duarte, Anatol G. Feldman, Tamar Flash, Mark L.
Latash, John P. Rothwell and Michael Turvey; members of the local Organizing
Committee: Philippe Archambault, Andreanne Blanchette, Julie Cote, Paul Cisek,
David J. Ostry, Anatol G. Feldman, Roger Hobden, Therese Cadrin-Petit and stu-
dents Sandeep Subramanian, Aditi Mullick, Marcos Rodrigues, Maxime Robert and
Matthew Slimovich.
xi
Contents
1 Motor Control: On the Way to Physics of Living Systems . . . . . . . . . . 1

Mark L. Latash
2 Motor Control and Position Sense: Action-Perception Coupling . . . . 17

Anatol G. Feldman, Nabil Ilmane, Samir Sangani and Helli Raptis
3 Reconfiguration of the Electrical Properties of Motoneurons

to Match the Diverse Demands of Motor Behavior . . . . . . . . . . . . . . . . 33
C. J. Heckman and Michael D. Johnson
4 The Regulation of Limb Stiffness in the Context of Locomotor

Tasks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
T. Richard Nichols, Jinger S. Gottschall and Christopher Tuthill
5 Subcortical Visuomotor Control of Human Limb Movement . . . . . . . 55

Brian L. Day
6 Rethinking the Role of Motor Simulation in Perceptual Decisions . . . 69

Scott T. Grafton and Shivakumar Viswanathan
7 Use of the Uncontrolled Manifold (UCM) Approach to Understand

Motor Variability, Motor Equivalence, and Self-motion . . . . . . . . . . . . 91
John P. Scholz and Gregor Schöner
8 Acquisition of Novel and Complex Motor Skills: Stable Solutions

Where Intrinsic Noise Matters Less . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
Dagmar Sternad, Meghan E. Huber and Nikita Kuznetsov
9 The Dynamical Analysis of Inter-Trial Fluctuations Near Goal

Equivalent Manifolds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
Joseph P. Cusumano, Joseph M. Mahoney and Jonathan B. Dingwell
xiii
xiv Contents
10 Motor Control in Action: Using Dance to Explore the Intricate

Choreography Between Action Perception and Production
in the Human Brain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
Emily S. Cross and Anastassia Elizarova
11 Apollo’s Curse: Causes and Cures of Motor Failures in Musicians:

A Proposal for a New Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
Eckart Altenmüller, Christos I. Ioannou, Markus Raab and Babett
Lobinger
12 Motor Control and the Injured and Healthy Artist . . . . . . . . . . . . . . . . 179

Roger M. Hobden and Samuel Tétreault
13 Adaptations to Neck/Shoulder Fatigue and Injuries . . . . . . . . . . . . . . . 205

Julie N. Côté
14 Deficits in Spatial Threshold Control of Muscle Activation

as a Window for Rehabilitation After Brain Injury . . . . . . . . . . . . . . . . 229
Mindy F. Levin
15 Enhancing Postural Stability and Adaptability in Multiple

Sclerosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
Richard E. A. van Emmerik, Stephanie L. Jones, Michael A. Busa,
Jebb G. Remelius and Julianna L. Averill
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
Contributors
Eckart Altenmüller Institute of Music Physiology and Musicians’ Medicine,

Hannover University of Music, Drama and Media, Hannover, Germany
Julianna L. Averill Department of Kinesiology, University of Massachusetts,
Amherst, MA, USA
Michael A. Busa Department of Kinesiology, University of Massachusetts,
Amherst, MA, USA
Julie N. Côté Department of Kinesiology and Physical Education, McGill
University, Montreal, QC, Canada
Center for Interdisciplinary Research in Rehabilitation (CRIR), Jewish Rehabilitation
Hospital, Laval, QC, Canada
Emily S. Cross Wales Institute for Cognitive Neuroscience, School of Psychology,
Bangor University, Bangor, Gwynedd, Wales
Behavioural Science Institute and Donders Institute for Brain, Cognition and
Behaviour, Radboud University Nijmegen, Nijmegen, Gwynedd, The Netherlands
Joseph P. Cusumano Department of Engineering Science & Mechanics, Penn State
University, University Park, PA, USA
Brian L. Day Institute of Neurology, University College London, London, UK
Jonathan B. Dingwell Department of Kinesiology & Health Education, The
University of Texas at Austin, Austin, TX, USA
Anastassia Elizarova Wales Institute for Cognitive Neuroscience, School of
Psychology, Bangor University, Bangor, Gwynedd, Wales
Richard E. A. van Emmerik Department of Kinesiology, University of Mas-
sachusetts, Amherst, MA, USA
Anatol G. Feldman Department of Neuroscience, University of Montreal, Mon-
treal, QC, Canada
Center for Interdisciplinary Research in Rehabilitation (CRIR), Institut de readapta-
tion Gingras-Lindsay de Montreal, Montreal, QC, Canada
xv
xvi Contributors
Jinger S. Gottschall Department of Kinesiology, Pennsylvania State University,

State College, USA
Scott T. Grafton Department of Psychological Brain Sciences, University of
California, Santa Barbara, CA, USA
C. J. Heckman Departments of Physiology, Physical Medicine and Rehabilitation,
Physical Therapy and Human Movement Sciences, Feinberg School of Medicine,
Northwestern University, Chicago, IL, USA
Roger M. Hobden Department of Family Medicine and Emergency Medicine,
University of Montreal, Montreal, QC, Canada
Department of Family Medicine, CLSC des Faubourgs, Montreal, QC, Canada
Meghan E. Huber Department of Bioengineering, Northeastern University,
Boston, MA, USA
Nabil Ilmane Department of Neuroscience, University of Montreal, Montreal,
QC, Canada
Christos I. Ioannou Institute of Music Physiology and Musicians’ Medicine,
Hannover University of Music, Drama and Media, Hannover, Germany
Michael D. Johnson Departments of Physiology, Physical Medicine and Reha-
bilitation, Physical Therapy and Human Movement Sciences, Feinberg School of
Medicine, Northwestern University, Chicago, IL, USA
Stephanie L. Jones Department of Kinesiology, University of Massachusetts,
Amherst, MA, USA
Nikita Kuznetsov Department of Biology, Northeastern University, Boston, MA,
USA
Mark L. Latash Department of Kinesiology, The Pennsylvania State University,
University Park, PA, USA
Babett Lobinger Institute of Psychology, German Sport University Cologne, Köln,
Germany
Mindy F. Levin School of Physical and Occupational Therapy, McGill University,
Montreal, QC, Canada
Center for Interdisciplinary Research in Rehabilitation (CRIR), Jewish Rehabilitation
Hospital, Laval, QC, Canada
Joseph M. Mahoney Department of Engineering Science & Mechanics, Penn State
University, University Park, PA, USA
Contributors xvii
T. Richard Nichols School of Applied Physiology and Department of Biomedical

Engineering, Georgia Institute of Technology, Atlanta, GA, USA
Markus Raab Institute of Psychology, German Sport University Cologne, Köln,
Germany
Department of Applied Sciences, London South Bank University, London, UK
Helli Raptis Department of Neuroscience, University of Montreal, Montreal, QC,
Canada
Jebb G. Remelius Department of Kinesiology, University of Massachusetts,
Amherst, MA, USA
Samir Sangani Center for Interdisciplinary Research in Rehabilitation (CRIR),
Jewish Rehabilitation Hospital, Laval, QC, Canada
John P. Scholz Department of Physical Therapy and Biomechanics and Movement
Sciences Program, University of Delaware, Newark, DE, USA
Gregor Schöner Institut für Neuroinformatik, Ruhr-Universität Bochum, Bochum,
Germany
Dagmar Sternad Departments of Biology, Electrical and Computer Engineering,
Bioengineering, and Physics, Northeastern University, Boston, MA, USA
Samuel Tétreault Les 7 Doigts de la Main, Montreal, QC, Canada
Christopher Tuthill School of Applied Physiology and Department of Biomedical
Engineering, Georgia Institute of Technology, Atlanta, GA, USA
Shivakumar Viswanathan Department of Psychological Brain Sciences, Univer-
sity of California, Santa Barbara, CA, USA
Chapter 1
Motor Control: On the Way to Physics
of Living Systems
Mark L. Latash
1.1 Motor Control as a Subfield of Physics of Living Systems
Two approaches to the problem of the neural control of movement have dominated
the field recently. According to the first approach, the central nervous system (CNS)
performs computations in an attempt to predict the complex interactions both within
the body and between the body and the environment. Specific examples of such
computations have been addressed as “internal models” (reviewed in Kawato 1999;
Shadmehr and Wise 2005). The recent developments within the field of internal
models have been based on classical studies in the control theory developed for
inanimate objects directly applied to problems of motor control.
The alternative approach considers the CNS not as a computational system but
as a physical one (reviewed in Kugler and Turvey 1987; Latash 2010, 2012). Under
the word “physical,” I imply the totality of mechanical, chemical, and physiologi-
cal processes that obey laws of nature. Some of these laws may be unknown to us
at this time, particularly those related to such notions as intentionality, specific for
biological objects. While the established laws of physics of the inanimate nature are
expressed using mathematical equations, the existence of these equations does not
assume that planets, particles, neural structures, and other objects of study solve
these equations or perform any related computations. For example, no physicist
would assume that there is a structure on the Sun that computes all the forces that
have to be applied to the planets and other celestial objects within the solar system.
Such an assumption is, unfortunately, only all too common with respect to the brain:
Within the computational approach, the neural controller is frequently assumed to
solve mathematical equations. Within the physical approach, the term “neural con-
troller” is used in a different meaning, to represent decision-making processes with
currently unknown physics. Physics of living systems is a very poorly developed field
M. L. Latash ()
Department of Kinesiology, The Pennsylvania State University,
Rec. Hall-268N, University Park, PA 16802, USA
e-mail: [email protected]
© Springer Science+Business Media New York 2014 1

M. F. Levin (ed.), Progress in Motor Control, Advances in Experimental Medicine
and Biology 826, DOI 10.1007/978-1-4939-1338-1_1
2 M. L. Latash
of science. A major with this name was created in the Moscow Institute of Physics
and Technology (MIPT), the alma mater of the author, over 50 years ago. This was,
however, not a reflection of existence of this new subfield of physics, but rather a
promise that this subfield would be developed in future. The purpose of this chapter
is to try to persuade the reader that motor control may in fact be the most promising
area of research of biological systems to make physics of living systems a reality.
Studies of voluntary movements are very attractive for a physicist. Humans and
animals are usually very good at producing well-coordinated movements. One can set
tasks relatively easily and record all the relevant mechanical variables relatively ob-
jectively (in comparison, for example, to studies of such brain functions as thoughts,
emotions, memory, etc.). The mathematical apparatus of classical mechanics is also
of a great help, although its application to biological objects is never easy. Most
classical mechanical characteristics of moving objects (such as their mass-inertial
characteristics, parameters that define their elastic and damping properties, etc.)
change in the course of a movement. These characteristics also change in response
to an externally applied perturbation making commonly used system identification
techniques marginally useful. That is why, sometimes, I like to address the fields of
motor control and biomechanics as “physics of unobservable objects.”
1.2 Equilibrium-point Hypothesis and Control With Referent

Configurations
Arguably, the first physical approach to the problems of motor control was introduced
by another alumnus of physics of living systems in MIPT, Anatol Feldman (1966).
He developed the equilibrium-point (EP) hypothesis that views the neural control of
a muscle as a process of defining an input parameter, associated with the threshold
(λ) for activation of the alpha-motoneuronal pool of that muscle (Feldman 1986).
Further, the interaction of the muscle and all its reflex loops with the external load,
results in reaching an EP of that whole system characterized by certain values of
muscle length and force (and electrical activity, EMG). This simplified scheme is
illustrated in Fig. 1.1.
Within this simple scheme, two kinds of movements are possible. First, a change
in the external load (from L1 to L2 in Fig. 1.1a) leads to a change in the EP along
the tonic stretch reflex characteristic. Such a movement may be called “involuntary”
because no change in the parameter λ specified by the CNS is assumed to take place.
Note that activation patterns of many neurons within the CNS change during invol-
untary movements due to the action of reflex loops; such movements are obviously
associated with changes in muscle activation level and force. A voluntary movement
happens when the parameter λ changes (from λ1 to λ2 in Fig. 1.1b). Mechanical
effects of such a change in λ depend on the external load: In isometric conditions, a
change in force takes place, while in isotonic conditions, the muscle moves (compare
EP2,1 and EP2,2 in Fig. 1.1b).
1 Motor Control: On the Way to Physics of Living Systems 3
Fig. 1.1 An illustration of involuntary and voluntary movements within the equilibrium-point hy-
pothesis. a A change in the external load from L1 to L2 leads to a shift in the equilibrium point
(EP) of the system along the characteristic of the tonic stretch reflex shown by the dashed line
from EP1 to EP2 . This leads to a change in muscle length, muscle force, (and activation level, not
illustrated). b A change in the threshold of the tonic stretch reflex (λ) from λ1 to λ2 may lead to
different consequences depending on the external load. In isotonic condition, a movement takes
place (EP1 to EP2,1 ), while in isometric conditions, muscle force increases (EP1 to EP2,2 )
Generalization of the control with tonic stretch reflex thresholds to a two-muscle

system acting at a joint was relatively straightforward (Feldman 1980; Feldman and
Latash 1982). The development of the EP hypothesis to multi-muscle systems that
participate in all natural movements resulted in the referent configuration (RC) hy-
pothesis (Feldman and Levin 1995; Feldman 2009). According to this hypothesis, the
controller specifies neural parameters (associated with subthreshold depolarization
of neuronal pools) resulting in a referent body configuration defined as a configura-
tion, at which all muscles show zero activation levels and are at their thresholds for
activation via the tonic stretch reflex loops. Such configurations may be unattainable
leading to nonzero muscle activations because of both anatomical and external fac-
tors. In addition, nonzero muscle activations at the RC may be due to a coactivation
command (Feldman 1980, 1986). The notion of RC can be generalized to location
of the body in the environment, as shown in studies of locomotion (Feldman et al.
2011).
Let me emphasize a couple of important features of the EP hypothesis that make
it a truly physical hypothesis (these features are also parts of the RC hypothesis).
First, it does not try to deduce properties of the whole system from properties of its
elements such as individual neurons, peripheral receptors, and reflex loops. Instead,
it considers motor behavior as an emergent result of interactions among all elements
(muscle + receptors + feedback loops + neuronal machinery + external force field)
affected by a single parameter (λ) generated by a hierarchically higher controller.
Second, it appreciates that a controller dealing with a system evolving in time (a dy-
namic system) can only define parameters of this system but not its output variables.
So, this approach refutes all control schemes based on prescribing time patterns of
muscle forces or activation levels (cf. Gordon and Ghez 1987; Gottlieb et al. 1989;
Scheidt and Ghez 2007; Shadmehr and Mussa-Ivaldi 1994).
4 M. L. Latash
1.3 Equifinality and its Violations: Mass–Spring Analogy

and its Demise
One of the features of the original, simplified version of the EP hypothesis was
its rather direct association with mass–spring models. The muscle seemed to be
adequately represented as a nonlinear spring with modifiable zero length (reviewed
in Feldman 1986). This analogy led to predictions of equifinality of movements in
cases of unexpected transient perturbations acting in the course of a movement, if
the subject of this experiment does not react to the perturbation. Indeed, a mass on
a spring (with some kind of a dissipative element, a damper) is expected to come
to an equilibrium, which depends only on parameters of this system and external
force at the final state, but not on possible transient force changes. Phenomena of
equifinality have indeed been reported in several studies with transient changes in
the external forces (Bizzi et al. 1976; Kelso and Holt 1980; Schmidt and McGown
1980; Latash and Gottlieb 1990). However, violations of equifinality have also been
reported giving rise to doubts about the EP hypothesis as a whole (Lackner and DiZio
1994; DiZio and Lackner 1995; Hinder and Milner 2003).
In particular, when a person sits in a chair aligned with the vertical axis of a rotating
centrifuge, any arm movement is associated with the Coriolis force that acts only
during the voluntary movement but not at the final steady state. Experiments in such
conditions (performed in darkness to remembered visual targets) showed, however,
reproducible deviations of the arm end-state coordinates from the target (Lackner
and DiZio 1994; DiZio and Lackner 1995), that is, violations of equifinality. Another
study used a torque motor to generate external torques proportional to the rotational
velocity magnitude and acting to increase the velocity (Hinder and Milner 2003).
In other words, the torque motor simulated motion in a medium with (physically
impossible!) negative viscosity. In this experiment, the motor torque was applied
during the movement but not when velocity became zero. After the subjects learned
to move in such an unusual, destabilizing force field, the force field was turned off,
and the subjects showed consistent undershoots of the target.
These observations have been discussed in detail within the framework of the EP
hypothesis (Feldman and Latash 2005). In particular, it was necessary to dispel the
simplified view on the EP control and refute the mass–spring analogy as too simple
and leading to wrong expectations about the system’s behavior. This analogy was
used originally to illustrate the phenomenon of equifinality that could be expected
within the EP hypothesis under certain, but not all, conditions. In particular, the
EP hypothesis was sometimes viewed in an oversimplified way: The input into the
motoneuronal pool was supposed to be defined exclusively by descending signals
related to the task (Fig. 1.2a). There are other inputs that can change threshold of
the tonic stretch reflex, in particular those originating from proprioceptors in other
muscles. In addition, muscle spindle endings are sensitive not only to muscle length
but also velocity, which makes λ dependent on actual movement velocity. There are
also history effects on λ changes. So, descending signals define only a fraction of
actual λ changes as reflected by the equation in Fig. 1.2b.
Fig. 1.2 In the simplified version of the equilibrium-point (EP) hypothesis (a), the input into a
motoneuronal pool (λ) is defined exclusively by the brain. In a more recent version, this input
depends on the contribution from the brain (λ∗ ) and also on other inputs such as those produced
by sensory receptors in other muscles (ρ), related to velocity sensitivity of muscle spindle endings
(μV ), and history effects (f(t))
Note that these apparent complicating factors are not questioning the main princi-
ple of muscle control, although they potentially make movements not as accurate as
they could otherwise be. Indeed the central input into the motoneuronal pool defines
only a portion of signals leading to λ shifts and, as such, does not predetermine the
motor outcome. Since other contributors to λ shifts may be not perfectly predictable,
movements may be expected to show errors partly due to the varying contributions
of the three additional terms in the right side of the equation in Fig. 1.2. This mode of
control may be called “reasonably sloppy”. It requires corrections based on sensory
feedback signals as postulated by Bernstein (1967).
1.4 Redundancy and Abundance
At any level of analysis of the system for movement production, all natural move-
ments involve more elements than necessary. In other words, the number of
constraints associated with typical tasks is larger than the number of variables pro-
duced by the elements (elemental variables). This leads to one of the central problems
of motor control, the problem of motor redundancy (Bernstein 1947, 1967): How
does the CNS facilitate specific movement patterns from an infinite set? Examples of
this problem are: (1) How to select joint rotations leading to desired coordinates of
the endpoint in space? (2) How to select muscle forces leading to a desired resultant
joint moment of force? (3) How to select digit forces leading to a desired resultant
force applied to the hand-held object? (4) How to select a particular pattern of re-
cruitment of alpha-motoneurons to match a desired level of muscle activation? and
so on.
Most approaches to this problem have followed the tradition set by Bernstein
that the main problem of motor control is that of elimination of redundant degrees
of freedom (DOF; Bernstein 1967). In particular, optimization methods have been
6 M. L. Latash
broadly used to find unique solutions for this problem (reviewed in Seif-Naraghi
and Winters 1990; Prilutsky and Zatsiorsky 2002). Recently, the problem of motor
redundancy has been revisited and reformulated as bliss of motor abundance (Gelfand
and Latash 1998; Latash 2012). The principle of abundance states that the CNS
uses all the available DOFs and arranges task-related stability properties in different
directions within the redundant space of those elemental variables. The different
stability properties are reflected in the structure of variance across repetitive trials
(Scholz and Schöner 1999; Scholz et al. 2000; Latash et al. 2001) as well as in
reactions to perturbations (Yang et al. 2007; Mattos et al. 2011, 2013).
The two approaches seem incompatible since the former generates a single so-
lution while the latter generates families of solutions equally able to solve the task.
Recently, a way to reconcile the two approaches has been suggested (Park et al.
2010). The principle of abundance generates families of solutions—clouds of data
points at each phase of the action, while the centers of those clouds may be defined
based on an optimization principle.
1.5 Synergies and the Uncontrolled Manifold Hypothesis
The term “synergy” has been used in the field of movement studies for at least 100
years (Hughlings Jackson 1899; Babinski 1899). Most commonly, this term implied
something like “coordinated action of several elements” or “several elements that
change their outputs in parallel” (Ivanenko et al. 2004, 2005; Ting and Macpherson
2005; Tresch et al. 2006). Recently, an explicit, operational definition for this term
has been offered focused on stability of actions of redundant systems. Synergies
have been defined as neural organizations that make across-trials variance within the
space of elemental variables relatively small along directions that change task-related
performance variables and relatively large along orthogonal directions (reviewed in
Latash et al. 2007; Latash 2008). This definition is intimately linked to the principle of
abundance because it assumes that the neural control of natural actions is associated
with defining task-related stability properties in different directions within the space
of the elemental variables.
The suggested definition also provides direct links between the notion of syn-
ergies and the concept of the uncontrolled manifold (UCM; Schöner 1995; Scholz
and Schöner 1999). The UCM hypothesis assumes that variance is structured within
the redundant space of elemental variables to provide desired stability properties of
potentially important performance variables. For any performance variable, there is
a subspace within which this variable does not change; this subspace is the UCM
for that variable. If this performance variable is stabilized by covaried (across repet-
itive trials) adjustments of elemental variables, variance within the UCM (VUCM ) is
expected to be larger than that orthogonal to the UCM (VORT ), if both are quantified
per dimension in the corresponding subspaces.
Clearly, the two notions, synergy and UCM, are intimately linked. Quantita-
tive analysis of synergies has been developed using the framework of the UCM
Fig. 1.3 An illustration of a

task E1 + E2 = C. Two data
distributions are illustrated.
One of them is spherical and
has about equal amounts of
variance in the direction that
does not affect E1 + E2
(UCM) and orthogonal to that
direction. The second
distribution is elongated
along the UCM. Most of its
variance has no effect on
E1 + E2 . Synergy is defined
operationally as
VUCM > VORT
hypothesis. In particular, indices of synergy have been used reflecting the ratio
of VUCM to VORT and the normalized difference between VUCM and VORT (ΔV ).
Figure 1.3 illustrates two data distributions for a task of producing a constant sum
of two elemental variables, E1 and E2 . One of the distributions is elongated along
the UCM for the task-related variable (dashed, slanted line), and for this distribution
VUCM > VORT . This allows claiming that CNS actions organized variations in the E1
and E2 to stabilize their sum against spontaneous changes in the initial conditions
associated with repetitive tasks. This is not the case for the spherical data distribution
(VUCM = VORT ). Note that accuracy of performance is defined by VORT only; so, it
is possible to perform accurately without a synergy (a tiny spherical cloud of data
points, stereotypical performance) and to be sloppy with a synergy (a large data cloud
elongated along the UCM).
The notions of synergies and UCM do not imply any computations within the CNS,
although data distributions expected from the UCM hypothesis can be obtained us-
ing computational methods such as optimal feedback control (Todorov and Jordan
2002). Synergies and corresponding variance structure may be expected from purely
physical/physiological mechanisms such as those offered in two recent schemes
incorporating the idea of control with referent configurations and the idea of hier-
archical control of natural actions (Martin et al. 2009; Latash 2010). One of these
ideas explicitly assumes the existence of two types of parameters specified by the
hypothetical controller (a neural structure with the currently unknown physics). One
group of parameters defines target values (trajectories) of important, task-related per-
formance variables. The other group defines stability properties of those variables.
Two predictions follow from this general scheme. First, one can perform actions
of different magnitude and direction using the same synergies (same ΔV indices).
Second, one can change stability properties of an action without changing the value
(pattern) of associated performance variable. While the first prediction is intuitively
clear (e.g., Danna-dos-Santos et al. 2007), the second one is less obvious. It predicts
the existence of previously unknown phenomena, anticipatory synergy adjustments
(ASAs).
8 M. L. Latash
Fig. 1.4 An illustration of anticipatory synergy adjustment (ASA) for the task E1 +E2 = C.At steady
state, most variance is along the UCM corresponding to a strong synergy stabilizing (E1 + E2 ).
Prior to a self-paced quick change in (E1 + E2 ), VORT starts to increase, while VUCM may show a
parallel decrease. This process continues until movement initiation (time zero). The inset shows a
change in the synergy index during this process
1.6 Feed-forward Control of Synergies: Anticipatory Synergy

Adjustments
Stability of actions is one of the important features of motor control, particularly

given the poorly predictable external conditions for most everyday movements and
the continuously changing state of the body (for a critical review see Hasan 2005).
However, there is a trade-off between stability and ability to change the action quickly.
That is why, the ability of the CNS to modify stability of a performance variable
without changing the action is important for preparation to a quick change in that
variable when it becomes necessary. Consider a goalkeeper standing quietly in the
goals while the ball is on the opposite half of the field. The goalkeeper’s posture is
stable and shows little sway. Now consider the same goalkeeper waiting for a penalty
kick. Quite frequently, large body sway can be seen with the naked eye, increasing
as the opponent runs towards the ball. An ability to adjust synergies in preparation to
an action in an unknown direction has recently been documented (Zhou et al. 2013).
The phenomenon of ASAs represents attenuation of synergies stabilizing a per-
formance variable when a change in that variable is planned. Figure 1.4 illustrates
ASAs using the earlier example of the task E1 + E2 = C (as in Fig. 1.3). If a quick
change in C is planned, a gradual increase in VORT (sometimes associated with a
drop in VUCM ) is seen during the steady-state C production, 200–300 ms prior to the
actual change in C (Olafsdottir et al. 2005; Shim et al. 2005). Three ellipses of data
distributions are shown in Fig. 1.4 illustrating the time development of ASAs. The
changes in the two-variance components lead to a drop in the synergy index (ΔV, the
inset in Fig. 1.4) documented in a variety of multi-finger pressing task, multi-digit
prehensile tasks, and multi-muscle postural tasks (Shim et al. 2006; Klous et al.
2011; Zhou et al. 2013).
Other aspects of feed-forward control have been studied extensively, such as
anticipatory postural adjustment (APAs, reviewed in Massion 1992) and grip force
adjustments in preparation to changes in the load forces during prehensile tasks
(Johansson and Westling 1988). Such populations as healthy elderly and patients
with neurological disorders show an impaired ability to generate APAs and grip
force adjustments (Woollacott and Manchester 1993; Novak et al. 2007; Muratori
et al. 2008; Albert et al. 2010). The same groups of people show reduced and delayed
ASAs (Olafsdottir et al. 2007; Park et al. 2012, 2013) suggesting that different aspects
of feed-forward control may rely on similar neurophysiological mechanisms.
1.7 Synergies Within a Hierarchical Control Scheme
The idea of hierarchically organized control of movement is very old. Arguably,

the first hierarchy considered in the field of movement studies was that of the brain
(upper level) and the spinal cord (lower level). Note that the notion of hierarchical
control does not refute exchange of information across the levels of the hierarchy in
both directions. It implies, however, that higher levels are not as strongly driven by
information from the lower levels. In terms of the RC hypothesis, one can claim that
actual state of the system is driven by changes in the RC (and other factors such as
external forces) while RC may also not be 100 % immune to effects from changes
in the actual state of the system. Such effects have been modeled as back coupling
within a recent theoretical scheme developed by Schöner and his colleagues (Martin
et al. 2009).
While RC was originally defined as the body configuration when all muscles are at
their thresholds for activation, it has also been long accepted that the neural controller
does not micromanage at the individual muscle level (“the brain knows nothing about
muscles, it knows only movements,” Hughlings Jackson 1899) but likely unites
muscles into relatively stable groups (addressed as “modes” or “synergies,”d’Avella
et al. 2003; Krishnamoorthy et al. 2003; Ivanenko et al. 2005; Ting and Macpherson
2005). Such stable groups may reflect certain RC combinations that are used as
the basis for creation of more specific, task-related RC shifts. At the level of joint
mechanics, basic RCs may be reflected in sets of eigenmovements introduced for
whole-body actions (Alexandrov et al. 2001).
The scheme uniting the ideas of hierarchical control, control with RCs, and syn-
ergies has been proposed recently (Latash 2010). Within this scheme, at the higher
level of the hierarchy, only a handful of task-related variables define the top RC. Fur-
ther, a chain of few-to-many transformations leads to RCs at lower, effector-specific
levels (e.g., limb-specific, digit-specific, joint-specific, and muscle-specific). Each
of the transformations is organized in a synergic way, that is, it leads to variable RCs
10 M. L. Latash
Fig. 1.5 An illustration of a trade-off between two levels of a hypothetical hierarchy. In the task
E1 + E2 = C (a), each E variable is produced by two contributors, for example E1 = e1,1 + e1,2
(b). There is a synergy stabilizing (E1 + E2 ) (VUCM > VORT ). Variance of each element is defined
by VUCM : VE1 = VUCM • cos(45◦ ). At the lower level (b) VE1 defines VORT . So, high VUCM in panel
A is beneficial for a synergy at that level, but it produces high VORT at the lower level, which hurts
synergies
at the lower level compatible with stable RC at the higher level. This scheme results
in a few nontrivial, testable predictions. For example, it predicts a trade-off between
synergies at different levels of the hierarchy (Gorniak et al. 2007; Latash et al. 2010).
Indeed, consider Fig. 1.5 illustrating a task similar to the one described earlier
(E1 + E2 = C; see Fig. 1.3) but assuming that each E variable is produced by two
contributors, for example E1 = e1,1 + e1,2 . Assume also that, at the upper level of this
hypothetical hierarchy, there is a synergy reflected in covariation between E1 and E2
reducing variance of (E1 + E2 ) (VUCM > VORT , Fig. 1.5a). The ellipse of data points
in Fig. 1.5a can be projected on the two E-axes, and the length of each projection is
going to be defined by the long axis of the ellipse (for example, VE1 = VUCM • cos
(45◦ )). Now let us turn to the two contributors to E1 , e1,1 and e1,2 . In that space,
VE1 defines VORT . The comparison of the two panels suggests that high VUCM at the
upper level of the hierarchy is beneficial for a synergy at that level, but it produces
high VORT at the lower level, which hurts synergies. It is possible to have synergies
at both levels as demonstrated by studies of static prehension; however, only some
of the performance variables show synergies at both levels, while others fail to do so
(Gorniak et al. 2009).
In addition, the scheme suggests that transient perturbations applied to such a
system may not change the top level RCs resulting in equifinality of the action
(assuming that the subject is not reacting to the perturbation) while intermediate RCs
may show variations compatible with the top RC (violations of equifinality at the
level of elements). These predictions have been tested in a few recent experiments
including the study of arm–trunk coordination during reaching (Ma and Feldman
1995).
Fig. 1.6 A perturbation

applied during the task
E1 + E2 = C is likely to show
a large deviation of the
system along the UCM
because of the lower stability
in that direction (A1 to A2 ).
After the perturbation is
removed, the system returns
to a different point A3 on the
UCM for the task. There is
equifinality with respect to
the task-related variable
(E1 + E2 ) but not at the level
of elements
1.8 Equifinality in a Redundant System
As mentioned earlier, within the EP-(RC) hypothesis, movement and force produc-
tion are different peripheral outcomes of basically the same neural control process.
This allows testing the hypothesis formulated at the end of the previous section using
two types of tasks: A force-production task with transient positional perturbations,
and a positional task with transient force perturbations. The general idea is illus-
trated in Fig. 1.6 using the same minimalistic task, E1 + E2 = C. Imagine that this
redundant system shows relatively low stability along the UCM for this task and
relatively high stability in the orthogonal direction. Now imagine that in one par-
ticular trial, the system is in steady state at point A1 on the UCM. A perturbation
is likely to show deviations of the system both along the UCM and orthogonal to
it, for example to point A2 . The deviation along the UCM is expected to be larger
because of the lower stability in that direction (cf. Mattos et al. 2011, 2013). If now
a perturbation is removed, the system returns to a state compatible with the origi-
nal value of (E1 + E2 ) (defined by the referent value for that variable and external
steady-state forces). However, it does not have to return to A1 because this secondary
perturbation is expected to lead to another deviation along the UCM, which is not
attracted strongly to A1 . Assume that it ends at point A3 . This illustration suggests
that transient perturbations that lead to equifinality with respect to the task-related
variable may be associated with significant variance (violations of equifinality) at
the lower hierarchical level.
In one of the recent studies (Zhou et al. 2014), the subjects were asked to hold a
handle in a static position against a baseline force generated by a robot. Then the robot
applied a short-lasting change in the total force while the subjects were instructed
“not to intervene voluntarily” (let the robot move your arm). When the force changed,
the handle moved to a new position (and orientation); when the force returned to the
baseline level, the handle returned close to the initial position/orientation, which
is an example of equifinality. However, final positions after the perturbation were
12 M. L. Latash
associated with large variability at the level of joint rotations. Most of this variability
was within the UCM for the endpoint location (and orientation). In other words,
transient perturbations resulted in relative equifinality at the level of task variables,
which was achieved by using variable joint configurations, as predicted by Fig. 1.6.
In another study (Wilhelm et al. 2013), the subjects were asked to produce constant
level of total force while pressing with four fingers on four force sensors. Then, the
visual feedback was turned off and one of the sensors was lifted using the “inverse
piano” apparatus (Martin et al. 2011). This manipulation resulted in an increase in
the force of the lifted finger and a drop in the forces of the other three fingers (similar
to the results by Martin et al. 2011). Total force increased. When the perturbed finger
was lowered to its initial position, total force was close to that observed in similar
trials without perturbations. In all trials the total force dropped gradually, which
is common after turning off visual feedback (Slifkin et al. 2000; Vaillancourt and
Russell 2002). There were, however, significant variations in the force distribution
among the fingers. In other words, equifinality was observed at the task level but not
at the level of elemental variables.
1.9 What Now? Let Me Tell You What Now
The good news is that we have a working hypothesis on how coordinated natu-
ral movements are produced by the CNS based on laws of nature. We also have a
reasonably well-developed associated computational apparatus that allows estimat-
ing quantitatively relevant variables that describe not only patterns of performance
variables but also their characteristics related to action stability. This computa-
tional apparatus describes a physical/physiological system but does not assume
that somewhere within the system such computations (or any computations) are
performed.
The not-so-good news is that the available knowledge is miniscule, fragmented,
and not always reliable. In particular, mapping of the models onto neurophysiological
structures and mechanisms is all but missing. Even behavioral properties of the
system are known unreliably, not much is known about such crucial features of
movements as their stability. While the overall picture remains mosaic, recent works
offer a coherent physical approach to putting pieces together (Martin et al. 2009;
Latash 2010; Feldman 2011). Unfortunately, large efforts are still wasted on studies
that are not driven by a clear theoretical scheme compatible with the known physics
and physiology.
While a coherent, comprehensive theory of motor control is not here yet, there are
already elements of progress in applying elements of the theory to clinical problems.
Among those are, for example, the analysis of spasticity as a disorder in the control
of the threshold of the tonic stretch reflex (Jobin and Levin 2000), the analysis of
changes in motor synergies in stroke, Parkinson’s disease and cerebellar disorders
(Reisman and Scholz 2003; Park et al. 2012, 2013), the analysis of changes in
the structure of motor variance with specialized practice in persons with atypical
development (Scholz et al. 2003), young and older persons (Olafsdottir et al. 2008;
Wu et al. 2012, 2013). The developed tools are sensitive to aging, fatigue, training,
and neurological disorders; in some cases, they even show sensitivity to subclinical
states of neurological patients (Park et al. 2012).
Among questions that are in an urgent need for research are: Is there back coupling
between actual and referent trajectories? What factors define stability at different lev-
els of a hypothetical control hierarchy (cf. Gorniak et al. 2009)? Is there a distribution
of responsibilities among brain structures, for example are subcortical loops more
likely to define stability properties of actions (cf. Reisman and Scholz 2003; Park
et al. 2013)? What are the neurophysiological mechanisms that define stability of
performance (cf. Latash et al. 2005)? What is the role of information from peripheral
receptors versus back coupling within the CNS in defining synergies?
Acknowledgments Preparation of this paper was in part supported by NIH grants NS-035032
and AR-048563.
References
Albert F, Diermayr G, McIsaac TL, Gordon AM. Coordination of grasping and walking in
Parkinson’s disease. Exp Brain Res 2010; 202: 709–722.
Alexandrov AV, Frolov AA, Massion J. Biomechanical analysis of movement strategies in human
forward trunk bending. I. Modeling. Biol Cybern 2001; 84: 425–434.
Babinski F. De l’asynergie cerebelleuse. Revue Neurol 1899; 7: 806–816.
Bernstein NA. On the construction of movements. Medgiz: Moscow (in Russian); 1947.
Bernstein NA. The co-ordination and regulation of movements. Pergamon Press, Oxford; 1967.
Bizzi E, Polit A, Morasso P. Mechanisms underlying achievement of final head position. J
Neurophysiol 1976; 39: 435–444.
Danna-Dos-Santos A, Slomka K, Zatsiorsky VM, Latash ML. Muscle modes and synergies during
voluntary body sway. Exp Brain Res 2007; 179: 533–550.
d’Avella A, Saltiel P, Bizzi E. Combinations of muscle synergies in the construction of a natural
motor behavior. Nat Neurosci 2003; 6: 300–308.
DiZio P, Lackner JR. Motor adaptation to Coriolis force perturbations of reaching movements:
endpoint but not trajectory adaptation transfers to the nonexposed arm. J Neurophysiol 1995;
74: 1787–1792.
Feldman AG. Functional tuning of the nervous system with control of movement or maintenance
of a steady posture. II. Controllable parameters of the muscle. Biophysics 1966; 11: 565–578.
Feldman AG. Superposition of motor programs. I. Rhythmic forearm movements in man. Neurosci
1980; 5: 81–90.
Feldman AG. Once more on the equilibrium-point hypothesis (λ-model) for motor control. J Mot
Behav 1986; 18: 17–54.
Feldman AG. Origin and advances of the equilibrium-point hypothesis. Adv Exp Med Biol 2009;
629: 637–643.
Feldman AG, Latash ML. Interaction of afferent and efferent signals underlying joint position sense:
Empirical and theoretical approaches. J Mot Behav 1982; 14: 174–193.
Feldman AG, Latash ML. Testing hypotheses and the advancement of science: Recent attempts to
falsify the equilibrium-point hypothesis. Exp Brain Res 2005; 161: 91–103.
Feldman AG, Levin MF. Positional frames of reference in motor control: their origin and use. Behav
Brain Sci 1995; 18: 723–806.
14 M. L. Latash
Feldman AG, Krasovsky T, Bania MC, Lamontagne A, Levin MF. Changes in the referent body
location and configuration may underlie human gait, as confirmed by findings of multi-muscle
activity minimizations and phase resetting. Exp Brain Res 2011; 210: 91–115.
Gelfand IM, Latash ML. On the problem of adequate language in movement science. Motor Control
1998 2: 306–313.
Gordon J, Ghez C. Trajectory control in targeted force impulses. II. Pulse height control. Exp Brain
Res 1987; 67: 241–252.
Gorniak S, Zatsiorsky VM, Latash ML. Emerging and disappearing synergies in a hierarchically
controlled system. Exp Brain Res 2007; 183: 259–270.
Gorniak SL, Zatsiorsky VM, Latash ML. Hierarchical control of static prehension: II. Multi-digit
synergies. Exp Brain Res 2009; 194: 1–15.
Gottlieb GL, Corcos DM, Agarwal GC. Strategies for the control of voluntary movements with one
mechanical degree of freedom. Behav Brain Sci 1989; 12: 189–250.
Hasan Z. The human motor control system’s response to mechanical perturbation: should it, can it,
and does it ensure stability? J Mot Behav 2005; 37: 484–493.
Hinder MR, Milner TE. The case for an internal dynamics model versus equilibrium point control
in human movement. J Physiol 2003; 549: 953–963.
Hughlings Jackson J. On the comparative study of disease of the nervous system. Brit Med J 1899;
Aug. 17: 355–362.
Ivanenko YP, Poppele RE, Lacquaniti F. Five basic muscle activation patterns account for muscle
activity during human locomotion. J Physiol 2004; 556: 267–282.
Ivanenko YP, Cappellini G, Dominici N, Poppele RE, Lacquaniti F. Coordination of locomotion
with voluntary movements in humans. J Neurosci 2005; 25: 7238–7253.
Johansson RS, Westling G. Programmed and triggered actions to rapid load changes during precision
grip. Exp Brain Res 1988; 71: 72–86.
Jobin A, Levin MF. Regulation of stretch reflex threshold in elbow flexors in children with cerebral
palsy: a new measure of spasticity. Dev Med Child Neurol 2000; 42: 531–540.
Kawato M. Internal models for motor control and trajectory planning. Curr Opinion Neurobiol
1999; 9: 718–727.
Kelso JA, Holt KG. Exploring a vibratory systems analysis of human movement production. J
Neurophysiol 1980; 43: 1183–1196.
Klous M, Mikulic P, Latash ML. Two aspects of feed-forward postural control: Anticipatory postural
adjustments and anticipatory synergy adjustments. J Neurophysiol 2011; 105: 2275–2288.
Krishnamoorthy V, Latash ML, Scholz JP, Zatsiorsky VM. Muscle synergies during shifts of the
center of pressure by standing persons. Exp Brain Res 2003; 152: 281–292.
Kugler PN, Turvey MT. Information, natural law, and the self-assembly of rhythmic movement.
Hillsdale, NJ: Erlbaum; 1987.
Lackner JR, DiZio P. Rapid adaptation to Coriolis force perturbations of arm trajectory. J
Neurophysiol 1994; 72: 1–15.
Latash ML. Synergy Oxford University Press: New York; 2008.
Latash ML. Motor synergies and the equilibrium-point hypothesis. Motor Control 2010; 14:
294–322.
Latash ML. The bliss (not the problem) of motor abundance (not redundancy). Exp Brain Res
2012; 217: 1–5.
Latash ML, Gottlieb GL. Compliant characteristics of single joints: Preservation of equifinality
with phasic reactions. Biol Cybern 1990; 62: 331–336.
Latash ML, Scholz JF, Danion F, Schöner G. Structure of motor variability in marginally redundant
multi-finger force production tasks. Exp Brain Res 2001; 141: 153–165.
Latash ML, Shim JK, Smilga AV, Zatsiorsky V. A central back-coupling hypothesis on the organiza-
tion of motor synergies: a physical metaphor and a neural model. Biol Cybern 2005; 92: 186–191.
Latash ML, Scholz JP, Schöner G. Toward a new theory of motor synergies. Motor Control 2007;
11: 276–308.
Latash ML, Friedman J, Kim SW, Feldman AG, Zatsiorsky VM. Prehension synergies and control
with referent hand configurations. Exp Brain Res 2010; 202: 213–229.
Ma S, Feldman AG. Two functionally different synergies during arm reaching movements involving
the trunk. J Neurophysiol 1995; 73: 2120–2122.
Martin V, Scholz JP, Schöner G. Redundancy, self-motion, and motor control. Neural Comput
2009; 21: 1371–1414.
Martin JR, Budgeon MK, Zatsiorsky VM, Latash ML. Stabilization of the total force in multi-finger
pressing tasks studied with the ‘inverse piano’ technique. Hum Move Sci 2011; 30: 446–458.
Massion J. Movement, posture and equilibrium interaction and coordination. Prog Neurobiol 1992;
38: 35–56.
Mattos D, Latash ML, Park E, Kuhl J, Scholz JP. Unpredictable elbow joint perturbation during
reaching results in multijoint motor equivalence. J Neurophysiol 2011; 106: 1424–1436.
Mattos D, Kuhl J, Scholz JP, Latash ML. Motor equivalence (ME) during reaching: Is ME
observable at the muscle level? Motor Control 2013; 17: 145–175.
Muratori LM, McIsaac TL, Gordon AM, Santello M. Impaired anticipatory control of force sharing
patterns during whole-hand grasping in Parkinson’s disease. Exp Brain Res 2008; 185: 41–52.
Nowak DA, Topka H, Timmann D, Boecker H, Hermsdörfer J. The role of the cerebellum for
predictive control of grasping. Cerebellum 2007; 6: 7–17.
Olafsdottir H, Yoshida N, Zatsiorsky VM, Latash ML. Anticipatory covariation of finger forces
during self-paced and reaction time force production. Neurosci Lett 2005; 381: 92–96.
Olafsdottir H, Yoshida N, Zatsiorsky VM, Latash ML. Elderly show decreased adjustments of
motor synergies in preparation to action. Clin Biomech 2007; 22: 44–51.
Olafsdottir HB, Zatsiorsky VM, Latash ML. The effects of strength training on finger strength and
hand dexterity in healthy elderly individuals. J Appl Physiol 2008; 105: 1166–1178.
Park J, Zatsiorsky VM, Latash ML. Optimality vs. variability: An example of multi-finger
redundant tasks. Exp Brain Res 2010; 207: 119–132.
Park J, Wu Y-H, Lewis MM, Huang X, Latash ML. Changes in multi-finger interaction and
coordination in Parkinson’s disease. J Neurophysiol 2012; 108: 915–924.
Park J, Lewis MM, Huang X, Latash ML. Effects of olivo-ponto-cerebellar atrophy (OPCA) on
finger interaction and coordination. Clin Neurophysiol 2013; 124: 991 998.
Prilutsky BI, Zatsiorsky VM. Optimization-based models of muscle coordination. Exer Sport Sci
Rev 2002; 30: 32–38.
Reisman D, Scholz JP. Aspects of joint coordination are preserved during pointing in persons with
post-stroke hemiparesis. Brain 2003; 126: 2510–2527.
Scheidt RA, Ghez C. Separate adaptive mechanisms for controlling trajectory and final position in
reaching. J Neurophysiol 2007; 98: 3600–3613.
Schmidt RA, McGown C. Terminal accuracy of unexpected loaded rapid movements: Evidence
for a mass-spring mechanism in programming. J Mot Behav 1980; 12: 149–161.
Scholz JP, Schöner G. The uncontrolled manifold concept: Identifying control variables for a
functional task. Exp Brain Res 1999; 126: 289–306.
Scholz JP, Schöner G, Latash ML. Identifying the control structure of multijoint coordination
during pistol shooting. Exp Brain Res 2000; 135: 382–404.
Scholz JP, Kang N, Patterson D, Latash ML. Uncontrolled manifold analysis of single trials during
multi-finger force production by persons with and without Down syndrome. Exp Brain Res
2003; 153: 45–58.
Schöner G. Recent developments and problems in human movement science and their conceptual
implications. Ecol Psychol 1995; 8: 291–314.
Seif-Naraghi AH, Winters JM. Optimal strategies for scaling goal-directed arm movements.
In: Winters JM, Woo SL-Y (eds.) Multiple muscle systems: biomechanics and movement
organization. Springer-Verlag, New York, pp. 312–334; 1990.
Shadmehr R, Mussa-Ivaldi FA. Adaptive representation of dynamics during learning of a motor
task. J Neurosci 1994; 14: 3208–3224.
Shadmehr R, Wise SP. The computational neurobiology of reaching and pointing. MIT Press:
Cambridge, MA; 2005.
Shim JK, Olafsdottir H, Zatsiorsky VM, Latash ML. The emergence and disappearance of
multi-digit synergies during force production tasks. Exp Brain Res 2005; 164: 260–270.
16 M. L. Latash
Shim JK, Park J, Zatsiorsky VM, Latash ML. Adjustments of prehension synergies in response to
self-triggered and experimenter-triggered load and torque perturbations. Exp Brain Res 2006;
175: 641–653.
Slifkin AB, Vaillancourt DE, Newell KM. Intermittency in the control of continuous force
production. J Neurophysiol 2000; 84: 1708–1718.
Ting LH, Macpherson JM. A limited set of muscle synergies for force control during a postural
task. J Neurophysiol 2005; 93: 609–613.
Todorov E, Jordan MI. Optimal feedback control as a theory of motor coordination. Nat Neurosci
2002; 5: 1226–1235.
Tresch MC, Cheung VC, d’Avella A. Matrix factorization algorithms for the identification of
muscle synergies: evaluation on simulated and experimental data sets. J Neurophysiol 2006;
95: 2199–2212.
Vaillancourt DE, Russell DM. Temporal capacity of short-term visuomotor memory in continuous
force production. Exp Brain Res 2002; 145: 275–285.
Wilhelm L, Zatsiorsky VM, Latash ML. Equifinality and its violations in a redundant system:
Multi-finger accurate force production. J Neurophysiol 2013; 110: 1965–1973.
Woollacott MH, Manchester DL. Anticipatory postural adjustments in older adults: are changes in
response characteristics due to changes in strategy? J Gerontol 1993; 48: M64–M70.
Wu Y-H, Pazin N, Zatsiorsky VM, Latash ML. Practicing elements vs. practicing coordination:
Changes in the structure of variance. J Mot Behav 2012; 44: 471–478.
Wu Y-H, Pazin N, Zatsiorsky VM, Latash ML. Improving finger coordination in young and elderly
persons. Exp Brain Res 2013; 226: 273–283.
Yang J-F, Scholz JP, Latash ML. The role of kinematic redundancy in adaptation of reaching. Exp
Brain Res 2007; 176: 54–69.
Zhou T, Solnik S, WuY-H, Latash ML. Equifinality and its violations in a redundant system: Control
with referent configurations in a multi-joint positional task. Motor Control 2014; (in press).
Zhou T, Wu Y-H, Bartsch A, Cuadra C, Zatsiorsky VM, Latash ML. Anticipatory synergy adjust-
ments: Preparing a quick action in an unknown direction. Exp Brain Res 2013; 226: 565–573.
Chapter 2
Motor Control and Position Sense:
Action-Perception Coupling
Anatol G. Feldman, Nabil Ilmane, Samir Sangani and Helli Raptis
2.1 Introduction
This chapter focuses on the relationship between motor control and position sense
(PS) defined as the ability of humans to recognize the position of body segments
under different conditions. Studies showed that afferent signals from proprioceptive
and cutaneous and, to a lesser degree, articular receptors satisfactorily reflect pas-
sive changes in the position of body segments (Matthews 1972; Edin and Johansson
1995). In contrast, afferent signals deliver ambiguous positional information during
intentional motor actions. For example, the activity of a major contributor to PS—
muscle spindle afferents—increases with increasing isometric muscle torque (Vallbo
1974) whereas the arm position at which the torque is produced is perceived as un-
changed. During slow intentional finger motion in isotonic conditions, discharges
of spindle afferents virtually remain constant (Hulliger et al. 1982) and thus do not
reflect the perceived changes in the finger position. These and other examples, justi-
fied the conclusion that afferent signals are somehow integrated with central control
signals underlying motor actions to form PS (von Holst 1954; McCloskey 1981;
Feldman and Latash 1982a; Proske and Gandevia 2012). The nature of this integra-
tion in PS remains unclear. This situation results, in particular, from controversies in
the understanding of how motor actions are controlled.
A. G. Feldman () · N. Ilmane · H. Raptis

Department of Neuroscience, University of Montreal,
6300, Darlington, Montreal, QC H3S 2J4, Canada
Tel.: +(514) 340-2111
Center for Interdisciplinary Research in Rehabilitation (CRIR), Institut de readaptation
Gingras-Lindsay de Montreal, Montreal, QC, Canada
S. Sangani
Center for Interdisciplinary Research in Rehabilitation (CRIR),

and Biology 826, DOI 10.1007/978-1-4939-1338-1_2
18 A. G. Feldman et al.
This chapter reviews major hypotheses on how motor actions are controlled and
the respective views on how PS is formed. Data will be described that permit choosing
between alternative hypotheses of motor control. This approach may help advance
the understanding of how control signals underlying motor actions contribute to PS
in different conditions as well as of kinesthetic illusions, including the phantom limb
phenomenon—the feeling that the amputated limb is present.
2.2 Alternative Views of Motor Control
Historically, a departure point in thinking about neural control of actions was based
on the fact that motor actions obey and therefore are described in terms of the laws
of mechanics. Many outstanding scholars in the field of action and perception, in-
cluding von Holst (1954) combined this fact with the unconditional assumption that
control levels of the nervous system are directly involved in the specification of vari-
ables characterizing the motor outcome in terms of movement kinematics, muscle
forces and electromyogram (EMG) activity commonly called motor commands. One
problem associated with this assumption was recognized by Von Holst (1954). He
noticed that direct specification of motor commands to muscles to make intentional
motion from a stable posture would be met with the resistance of posture-stabilizing
reflexes (the posture-movement problem). To solve the problem, he suggested that
neural control centers use a copy of pre-programmed motor commands—efference
copy (EC)—to suppress movement-related afferent signals (reafference) and thus
prevent resistance of postural reflexes to movement. The EC concept introduced by
von Holst (1954) replaced similar concepts expressed under different names since
the nineteenth century (Bridgeman 2007). Except for some researchers (e.g. Gibson
1968), the solution for the posture-movement problem and the EC concept has been
accepted by the majority of researchers in the field without any critical analysis. It
became a key concept in theories of motor control, including those that postulated
the use of internal neural models for computations and specification of the motor
outcome (e.g. Bays and Wolpert 2007).
A recent analysis (Feldman and Latash 2005; Feldman 2009, 2011) has revealed
some drawbacks of the solution of the posture-movement problem offered by von
Holst (1954). Physically, by suppressing reafference or, in modern terminology,
position- and velocity-dependent feedback, would make the initial position unstable.
This proposal does not explain how stability of the final posture is regained. Before
the motion, postural mechanisms were tuned to stabilize the initial posture and, once
restored at the final posture, would react to the accumulated deviation from the initial
posture and drive the arm back to it. This prediction of von Holst’s proposal conflicts
with consistent findings that arm motion ends at a stable posture without any tendency
to return to the initial posture. His proposal also conflicts with physiological data that
postural reflexes are not suppressed but are reset to a new position when self-initiated
motion is made, thus transferring stability to the final posture (Asatryan and Feldman
1965; Adamovich et al. 1997; Feldman 2011).
2 Motor Control and Position Sense: Action-Perception Coupling 19
The resetting mechanism represents an empirically based solution to the posture-

movement problem that does not rely on the idea of preprogramming of motor
commands to muscles and the EC concept (Feldman 2011). When postural reflexes
are reset (re-addressed) to a new posture, the initial posture becomes deviated from
the new posture, and the same posture-stabilizing mechanisms that would resist self-
initiated motion, now drive the body segments to the new position to which stability
is transferred. In other words, by shifting the threshold position at which posture-
stabilizing mechanisms are centered, the nervous system converts postural reflexes
from a movement-resisting to a movement-producing mechanism.
EC-based theories of motor control have further been questioned by recent find-
ings that descending systems in humans, particularly the corticospinal system, do not
carry information about motor commands (EMG, forces) to muscles (Raptis et al.
2010; Sangani et al. 2011; Ilmane et al. 2013). Instead, these systems specify spatial
thresholds at which neuromuscular structures (including reflexes and intrinsic mus-
cle properties) begin to act, thus confirming the hypotheses that motor control is not
based on preprogramming of the motor outcome and EC. These thresholds define a
certain position, R, of body segments at which the activity of muscles is minimized.
Thereby, any deviation from this position is opposed by reflexes. The R is also
called the referent position since it plays the role of the origin of the spatial frame
of reference in which the muscles and reflexes are constrained to work (Feldman
2011). These studies confirmed the earlier demonstration that central shifts in the
referent position of body segments underlie intentional arm movements in humans
(Asatryan and Feldman 1965), an empirical finding underlying the equilibrium-point
hypothesis (for recent review see Feldman 2011).
2.2.1 Alternative Views on the Role of Central Control Processes

Underlying Motor Actions in Position Sense
The alternative theories of motor control motivated different answers to the question
of how PS is formed. One answer is based on the dominant view that the brain is
directly involved in preprogramming of the desired motor output, including EMG
patterns and muscle forces. It suggests that a copy of motor commands, i.e., EC, is
somehow integrated with afferent signals to form PS (Proske and Gandevia 2012). It
has also been assumed that EC underlies the sense of effort and therefore this sense
and PS are interrelated. Evaluations of PS before and after ischemic deafferentation
of the arm have been considered as supporting evidence of a contribution of sense
of effort to PS (Proske and Gandevia 2012). This proposal, however, conflicts with
results of the study by Allen et al. (2007) showing that under normal conditions, PS
is load- and effort-independent.
Threshold position control is an alternative to the EC-based view not only about
how motor actions are controlled but also about how they are perceived. Specifically,
as the threshold position for muscle activation, the R carries essential positional
information and as such, it can be used to form PS. The PS rule based on the R
concept was previously proposed by Feldman and Latash (1982a) and recently up-
dated (Feldman 2009, 2011; Feldman et al. 2013). Specifically, to determine the
actual position, Q, of body segments, the nervous system can use already available
information about the referent position, R, and combine it with sensory information
related to the deviation, P, of the actual position of body segments from the referent
position:
Q=R+P (2.1)
This formula describes a natural, physical relationship between variables Q, R, and P,
as illustrated in Fig. 2.1a for a single joint interacting with a load. To clarify, suppose
that a referent position R, is centrally set, say, for wrist flexor muscles. A load
counteracting these muscles will stretch them beyond the threshold, R. The muscle
activity and torque will increase until muscle and load torques become balanced at
some actual position, Q. To perceive this position, it is sufficient to know its deviation,
P, from R, as defined by formula 1.
By applying this formula to PS, we need to assume that proprioceptive, cutaneous,
and articular afferents do not convey direct information about the actual position, Q,
of body segments but instead signal the deviation, P, of the actual position of the
body segments from the referent position. In other words, it is assumed that PS
has central (R) and afferent (P) components that are integrated at some level of
the brain to perceive the actual position of body segments. The R is the control
variable underlying motor actions and the P is the afferent outcome conveyed by
ascending pathways to the brain. The PS rule thus represents an explicit expression
of action–perception coupling in kinesthesia (see also Ostry et al. 2010).
2.2.2 Explanations of Position Sense Based on the PS Rule
The PS rule provides an explanation of PS in different motor tasks as well as of

kinesthetic illusions and the phantom limb phenomenon (Feldman 2009; Feldman
et al. 2013)—the sense that the amputated limb is still present. In particular, the PS
rule suggests that passive changes in position that do not involve changing in the
central PS component are perceived from changes in afferent signals, P, from muscle
spindles and cutaneous receptors. In contrast, all active (self-initiated) changes in
position are associated with changes in the central component, R, and PS will result
from combined changes in both PS components. The relative contribution of these
components to PS may be different depending on the motor task. Consider several
examples.
Figure 2.1 shows not only how the PS rule is derived (A) but also how it works
under different conditions. The thick solid curve in each panel is the muscle-reflex
characteristic (torque versus joint angle) for a given R. The thin solid curve is the
characteristic following a shift in the R. Dashed lines are load (L) characteristics.
Subscripts i and f refer to initial and final values of variables, respectively, and filled
dots are the initial and open dots are final equilibrium points of the system.
a b
c d
Fig. 2.1 The position sense rule: its derivation and application to different motor tasks. a PS rule
for a single joint interacting with a load. When a referent joint angle (R) is set by descending
systems, the activity of muscles and the net joint torque (T ) increases with deviation of the joint
angle from the R position (solid curve). The load (L) stretches the muscles (flexors in this example)
until the system reaches an equilibrium point (filled circle; point of intersection of the two curves),
i.e., the combination of the joint angle and muscle torque that balances the load (dashed curve)
at some position, Q. To identify this position, it is sufficient to know the deviation, P, of the joint
from its referent position: Q = R + P. It is assumed that this deviation is transmitted by ascending
afferent pathways and converges with the referent joint angle at some central level where position
sense is formed. b–d PS in different motor tasks. b during the unloading reflex elicited by a sudden
change in the load, the same R is maintained. Changes in the joint angle, Q, are perceived from
changes in the afferent PS component, Q ≈ P. c During slow isotonic motion, changes in the
joint angle are predominantly perceived from changes in the central PS component, Q ≈ R.
d During isometric torque production, changes in the two PS components are equal but opposite,
such that the joint angle is perceived as unchanged, Q = 0. Thick solid curve in each panel is
the muscle-reflex characteristic (torque versus joint angle) for a given R; thin solid curve is the
characteristic following a shift in the R; dashed lines are load (L) characteristics; subscripts i and f
refer to initial and final values of variables, respectively; filled dots are the initial and open dots are
the final equilibrium points of the system. (Reproduced with permission from Feldman et al. 2013)
During the unloading reflex (B) elicited by a sudden change in the load of say,
preloaded elbow flexors, the same R is maintained (see Ilmane et al. 2013) and the
arm moves to another position at which the residual load and muscle torques are
balanced. According to the PS rule, changes in the joint angle, Q, in the unloading
reflex are perceived from changes in the afferent PS component alone, Q ≈ P,
since in this case, R ≈ 0.
In contrast, during slow isotonic motion (Fig. 2.1c), the afferent PS component re-
mains virtually constant, which is consistent with data by Hulliger et al. (1982) on the
absence of position-related changes in muscle spindle afferents during isotonic finger
motion. Thus, in isotonic conditions, changes in the joint angle are predominantly
perceived from changes in the central PS component, Q≈ R.
During isometric torque production, the position (Q) of the joint remains un-
changed but the deviation of this position increases when the threshold position,
R, decreases (Fig. 2.1d). As a result, the activity of muscles and isometric torque
increases. This process is associated with increasing proprioceptive signals (Vallbo
1974). The changes in the two PS components are equal but opposite such that the
joint angle is perceived as unchanged, Q = 0.
The alternative, EC-based PS hypothesis is not sufficiently specific to explain
all these cases. In particular, EC and sense of effort are virtually the same at dif-
ferent positions under isotonic conditions (Fig. 2.1c) and hardly contribute to PS.
It is unclear how afferent signals can be used to identify the changes in position in
isotonic conditions. Muscle spindle afferents in this condition do not signal changes
in position (Hulliger et al. 1982). However, the contribution of cutaneous afferents
to PS in isotonic conditions cannot be rule out. To our knowledge, positional sensi-
tivity of these afferents has been demonstrated during passive changes of different
joint angles (e.g., Edin and Johansson 1995). Cutaneous and proprioceptive afferent
influences usually converge on common interneurons (Jankowska 1992; Hultborn
2006) and it is likely that during active isotonic movements, cutaneous, like spindle
afferents, convey ambiguous positional information.
The unloading reflex and isometric torque production are associated with changes
in effort. The EC or effort-based hypotheses are also not specific enough to explain
why the sense of effort can contribute to the sensation of changes in position during
the unloading reflex and to the sensation of the absence of changes in position during
isometric torque generation. The sense of effort is primarily associated with muscle
torques and the EC- or effort-based hypotheses do not provide clues as to how it can
affect PS that is formed in the positional domain.
2.3 Testing Alternative Hypotheses on PS
The prediction of the effort-based hypothesis of PS was tested by asking subjects

to produce changes in elbow position against different isotonic loads. According
to the Weber–Fechner law (Weber 1834; Fechner 1860), the sense of effort and
the error in identification of the changes in the load are scaled with the magnitude
of the load. Therefore, if dependent on the sense of effort, PS would reflect such
properties of the law. It appeared that PS is load-independent (Feldman et al. 2013;
Fig. 2.2), confirming previous findings by Allen et al. (2007). These findings conflict
with the effort-based PS hypothesis, but are consistent with the PS rule that is based
on experimental findings that control signals underlying motor actions do not carry
information about motor commands or sense of effort.
In the unloading reflex (Figs. 2.1b, 2.3), the changes in the wrist position were
correlated with the amount of unloading and thus with the sense of effort. However,
the precision with which subjects reproduced the positional changes for different
Fig. 2.2 The sense of position a 6

is independent of the sense of
effort. a Different loads 5
counteracting elbow flexor
muscles were associated with 4
Effort
different efforts identified
3
with numbers 1–5 (group
data). b With eyes closed, the 2
subject (S6) reproduced the
same elbow angle regardless 1
of the load. c The precision of
the reproduction (coefficient 0
of variation) was also 0.4 0.8 1.2 1.6 2.0
load-independent in this and S6
other 11 subjects. b 60
(Reproduced with permission
from Feldman et al. 2013) 50
Angle (°)
40
30
20
10
0
0.4 0.8 1.2 1.6 2.0
c 10
8
Coefficient of
variation (%)
0
0.4 0.8 1.2 1.6 2.0
Load (Nm)
amounts of unloading was load-independent (Feldman et al. 2013). This finding also
conflicts with the assumption that PS depends on the sense of effort since if the PS
were effort-dependent, the precision of the perceived changes in the wrist position
would be scaled with the load according to the Weber–Fechner law (Weber 1834;
Fechner 1860).
Consider findings that have been regarded as evidence that PS depends on the
sense of effort (Gandevia et al. 2006; Smith et al. 2009; Walsh et al. 2010). In these
studies, PS at the wrist was measured before and after the arm was paralyzed by
an anesthetic or by ischemia produced by a blood pressure cuff. Before paralysis,
subjects accurately detected the position of the wrist when it was passively rotated.
After paralysis, subjects were unable to indicate the position of the hand at rest.
a 100
b 40° F^ = 1.005F - 2.2°
R2=0.96
Reproduced position (F^)

80
x ( ng
20
)
ΔP
fle di
60 re l o a Initial
position
Un
Torque (%)
40 Initial -40 -20 20 40°

position
Intentional Unloading amount
20 movements (ΔR) 100%
-20
75%
F E
50%
0
-30 -20 -10 0 10 20 30 40 25%
-40 0%
-20
Angle (°) Post-unloading position (F)
Fig. 2.3 Converting changes in the afferent component of position sense into intentional motion.
a The mean initial (filled circle) and final (open circles) wrist equilibrium points resulting from
different levels of unloading of wrist flexors from an initial position (E) in a representative subject,
in the absence of vision. After unloading, the subject intentionally moved the wrist from the post-
unloading position (F) to the initial, pre-unloading position and back to the memorized post-
unloading position (F ∧ ). According to previous analysis (Ilmane et al. 2013), motion resulting
from unloading was predominantly perceived from changes (P) in the afferent component of
position sense whereas intentional movements from position F to the pre-unloading and back to the
post-unloading position, F ∧ , required respective changes (R ≈ P) in the referent position. The
test was made for a different amount of unloading (100, 75, 50, and 25 % of the initial load; 0 %
is for trials in which unloading was not produced). b A plot of position F ∧ versus F for different
amounts of unloading. The slope of the regression line between positions F ∧ and F was close to 1
in this and other 9 subjects. (Reproduced with permission from Feldman et al. (2013)
Efforts to move the hand during paralysis produced large illusions of hand motion.
From these observations, the authors suggested that PS depends on the sense of effort.
Alternatively, these observations can be explained in terms of the PS rule. During
passive movements or relaxation, the R is shifted outside of the biomechanical range
of the joint (Levin and Feldman 1994). As a result, PS during muscle relaxation is
mainly based on changes in the afferent PS component. Naturally, when afferent
feedback is blocked, subjects are unable to perceive the changes in wrist position, as
observed by Gandevia et al. (2006). Based on the PS rule, the illusion of motion of the
hand during paralysis can be explained in the following way. In the absence of afferent
feedback, the position of body segments can still be sensed based on the central PS
component, R. In experiments by Walsh et al. (2010), subjects learned to produce
different efforts before paralysis. They thus learned to scale the R shifts with effort
(cf. Fig. 2.1d). Indeed, when subjects reproduced these R shifts during paralysis, they
perceived illusory movements correlated with the sense of effort. Since correlation
between the two senses was preconditioned by the experimental protocol, results of
the above studies cannot be considered as evidence for a dependency of PS on the
sense of effort. This conclusion is also supported by the obvious dissociation between
the two senses in isotonic, isometric, and reflex unloading conditions (Fig. 2.1b–
2.1d).
Experiments were also conducted to test specific predictions of PS rule. According

to the PS rule, the central and afferent components are equally important to position
sense. The relative contribution of the two components to PS depends on the con-
dition. In particular, it has been shown that descending influences on motoneurons
of wrist muscles and thus the referent position, R, are similar at the pre- and post-
unloading positions (Ilmane et al. 2013; Fig. 2.3). This means that changes in position
elicited by unloading are predominantly perceived from a change in the afferent PS
component, P: Q = P. In contrast, descending influences are modified when the
wrist is intentionally moved to another position, suggesting that intentional changes
in position in isotonic conditions are predominantly perceived from the change in
the referent position, R. We tested the hypothesis that the nervous system is able
to transform afferent information about changes in the wrist position resulting from
unloading into the R command to make intentional isotonic movement of the same
extent as that resulting from unloading (Fig. 2.3). Preloaded wrist flexors were fully
or partially unloaded, and subjects were required to reproduce the involuntary wrist
excursions by intentionally moving the wrist, in the absence of vision. Subjects had
no difficulty in performing the required transformation of the P into the R component
of PS. Indeed, such a sensorimotor transformation was facilitated by the parity in the
dimensions of the two PS components as well as by the fact that the central PS com-
ponent also underlies intentional motor actions. The finding of a strong correlation
between voluntary and involuntary changes in the wrist position (Fig. 2.3) seems
obvious, but only in the context of the PS rule. The initial and final positions in the
unloading reflex were associated with different tonic EMG levels (or ECs), different
senses of effort and different afferent feedback whereas these aspects, including af-
ferent feedback (Hulliger et al. 1982), were hardly distinguishable at wrist positions
established before and after intentional motion in isotonic conditions. In other words,
the effort- or EC-based PS theory is not helpful in the explanation of PS. Indeed,
intuitively, we take for granted that our sense of position is adequate regardless of
what we do, but the PS rule shows that this intuition is physiologically justified in
the context of the PS rule, rather than in the context of effort- or EC-based theory.
2.4 Kinesthetic Illusions
Usually, vibration of a muscle tendon elicits the illusion of muscle lengthening

(extension of the elbow joint if vibration is applied to elbow flexors; Matthews 1972)
in the absence of physical motion. If vibration is prolonged, the illusion can be
reversed (Roll et al. 1989, 1980; Feldman and Latash 1982b). In terms of the PS
rule, these illusions result from influences of vibration predominantly on the afferent
and central PS components, respectively.
Physical changes in position are apparently coupled with appropriate awareness
about these changes. The integration of the two components may just be a first step in
the formation of PS and it does not address the challenging question of how subjects
become consciously aware of physical or illusory changes in position. Normally,
subjects are well aware of the position of body segments but this awareness may
become inadequate during artificial stimulation (e.g. tendon vibration or electrical
stimulation of the brain). The illusion of motion elicited by tendon vibration is an
example of such inadequacy. Another example was observed in a patient undergoing
brain surgery (Desmurget et al. 2009). Electrical stimulation of inferior parietal areas
(Brodmann areas 39 and 40), elicited the sensation of arm movement although no
movement was made. The authors concluded that activation of these areas reflected
the intention to move even in the absence of motion. One can add that under normal
conditions, activation of these areas might be combined with opening of pathways
that transform the intention to move into changes in the referent position of the
appropriate body segments. The resulting motor action (a movement or, if move-
ment is prevented, isometric torque generation) is recognized by combining afferent
feedback with the referent command, as implied by the PS rule.
No motion illusions have been observed in two studies (Desmurget et al. 2009;
Feldman et al. 2013). In one, electrical stimulation of the premotor area (the dorsal
sector of Brodmann area 6) in patients undergoing brain surgery triggered limb
movements but patients denied that they had moved (Desmurget et al. 2009). One can
suggest that the physical motion was elicited by changes in the R command without
simultaneous influence of this command on the PS. In the other study (Feldman et al.
2013), no motion illusion has been observed during long-term tonic vibrations of
elbow flexors in the absence of vision when subjects were unaware of any physical
motion of the arm elicited by the vibration (Fig. 2.4).
It appeared that the no-motion illusion elicited by vibration is preserved when the
arm moves against different isotonic loads. The critical angle at which subjects start
perceiving physical motion does not depend on the magnitude of the load, further
confirming that PS does not depend on the sense of effort.
2.5 Phantom Limb Phenomenon
The phantom limb phenomenon was considered as supporting the assumption that
PS depends on the sense of effort (Proske and Gandevia 2012). However, a phantom
limb can be sensed in the absence of any physical or mental effort. The phantom
limb phenomenon can be explained without the assumption that it relies on the sense
of effort. According to the PS rule, when afferent sensitivity is deficient or absent,
the amputated limb can still be sensed as being present based on the central PS
component, R. Because of the absence or deficiency of the afferent PS component,
however, the shape of the phantom limb can be abnormal (e.g., twisted phantom
elbow or knee joint) and associated with pain or other abnormal sensations.
Although R may be controlled independently of the current position, the trans-
formation (“re-calibration”) of electrochemical descending influences into a spatial
(positional) variable R would be impossible in the absence of position-dependent
afferent feedback to α-motoneurons (for details see Fig. 2.4 in Feldman 2009). There-
fore, in the absence of such feedback, the association of descending influences with
Fig. 2.4 No-motion illusion

a No moon illusion
and its independency of the
sense of effort. a
Experimental setup. b Tonic
vibration of the biceps tendon
(frequency 100 Hz, amplitude
3 mm) elicited activation of
elbow flexors (BB, BR) and
extensors (TL, TM), resulting
in elbow flexion. This subject Vibrator
(and other 9 subjects) was not
aware of this motion 10.3°
(no-motion illusion) until the
change in the joint angle
began to exceed 10.3◦ . c The b Unperceived motion
mean (± SD) change in the
elbow joint angle after which 15°
Elbow
the subject began to feel the angle
motion, for different loads BB
counterbalanced by elbow
flexor muscles. (Reproduced
with permission from BR
Feldman et al. 2013) TL
TM
5 10 15 s
Vibration
S5
c 18
12
Angle(°)
0
0.4 0.8 1.2 1.6 2.0
Load (Nm)
spatial variables, like the R, gradually decays. This explains why phantom limb sen-
sation in amputees eventually disappears and why chronically deafferented subjects
do not have phantom limb sensations (Paillard 1991). The absence of referent con-
trol in deafferented humans may result in the loss of the body scheme and other
sensorimotor deficits (Paillard 1991).
2.6 Further Implications and Future Directions
The notion that somatosensory feedback conveys information about deviation of

the actual position of body segments from a centrally specified referent position
may be applicable to ascending somatosensory pathways as well as to neurons of
dorsal and ventral spinocerebellar tracts and propriospinal neurons having recurrent
projections to the cerebellum (Arshavsky et al. 1972, 1978; Alstermark et al. 2007).
According to the PS rule, this would be helpful in continuously monitoring changes
in the actual position of body segments to decide whether it is necessary to correct
the ongoing movement or switch to another pattern of referent shift in response
to destabilizing external perturbations, environmental events, or internal intentions.
Indeed, this is not the only function of ascending pathways. In particular, together
with other proprioceptive afferents, tendon organ afferents may transmit information
about muscle forces to form the sense of effort associated with the ongoing action.
Another implication of the present results is that they constrain the choice between
different theories of action–perception coupling. Previous studies (Raptis et al. 2010;
Sangani et al. 2011; Ilmane et al. 2013) have questioned the long held assumption
(von Holst 1954) that the brain preprograms motor commands to muscles or efference
copy (EC). Descending systems, particularly the corticospinal system, have been
shown to not be involved in direct specification of motor commands to muscles.
The same conclusion refers to theories of motor control based on the idea of internal
models in which EC is a core concept (e.g., Bays and Wolpert 2007). Results obtained
in our recent studies have also shown that PS is independent of motor commands
and the sense of effort, thus rejecting EC- or effort-based theories of PS.
Some emerging theories of action and perception have not integrated the notion
of referent position control. The sensory prediction theory (SPT) suggests that the
brain predicts the proprioceptive consequences of the intended movement and that
these predictions are delivered to motoneurons via descending pathways (Adams
et al. 2013). Motoneurons are activated or not depending on the difference (“predic-
tive error”) between the predicted and the actual proprioceptive feedback. The SPT
is reminiscent of a similar theory proposed several decades ago by Powers (1973)
with the complementary assumption that the predictive error is eliminated by the
neuromuscular system that works as a linear closed-loop servo-controller. To be
functional, such a controller would have to have a physiologically unrealistic high
gain, like in the servo-control hypothesis by Merton (Merton 1953; see also criti-
cisms of the Powers’ theory by Fowler and Turvey 1978). In any case, by assuming
that the nervous system preprograms the sensory consequences of the motor out-
come, SPTs do not help explain how PS is formed since these sensory consequences
carry ambiguous information about positions of body segments (see Introduction).
In addition, SPT seems to misrepresent how motor actions are controlled. Consider,
for example, isometric torque production. In terms of SPTs, this is achieved by pre-
programming of the sensory signals (“afference copy”) associated with the required
torque resulting from activation of motoneurons that function depending on the er-
ror in prediction of these signals. After several trials, the sensory signals associated
with the required torque could be identified correctly and reproduced with a minimal
error. Then, paradoxically, motoneurons that function depending on this error would
barely be activated to generate the required torque. Therefore, not only PS but also
motor control is inexplicable in the framework of SPTs.
Previous studies (Raptis et al. 2010; Ilmane et al. 2013) have shown that isotonic
wrist movements are accomplished by shifting the referent wrist position, R. Results
of these studies were interpreted based on the assumption that changes in the afferent
PS component in isotonic conditions were minimal and therefore the difference in
the initial and final wrist angles in these conditions was predominantly perceived
from the difference in the central PS component. Future experiments are necessary
to test whether this conclusion can be generalized to isotonic positions at other joints,
including the elbow joint. The contribution of cutaneous afferents to PS in isotonic
conditions cannot be rule out. To our knowledge, positional sensitivity of these
afferents was demonstrated during passive changes of different joint angles (e.g.,
Edin and Johansson 1995). Cutaneous and proprioceptive afferent influences usually
converge on common interneurons (Jankowska 1992) and it is likely that during
active movements, cutaneous, like spindle afferents, convey ambiguous position
information, which can also be tested in future studies.
The nature of information conveyed by ascending pathways to the cerebellum
and other brain areas remains unclear. Based on our present findings, future studies
can address the question of whether these pathways carry information about the
deviation of the actual position of body segments from their referent position, rather
than information about efference copy, as usually assumed.
2.7 Conclusions
There are several demonstrations that the corticospinal and other descending systems
specify the referent position, R, of body segments at which motor commands can
emerge without preprogramming. It is assumed that the referent position not only
underlies motor actions but also represents a component of position sense. In other
words, motor actions and position sense are controlled jointly, by setting and resetting
of the referent position of body segments. Thereby, feedback from proprioceptive,
cutaneous, and articular afferents delivers information about the deviation (P) of the
body segments from the referent position. Taken together, the central and afferent
components of position sense provide adequate information about the actual position
(Q) of body segments, as defined by the PS rule: Q = R + P. This rule is helpful in the
explanation of PS in many motor tasks as well as of kinesthetic illusions, including
the phantom limb phenomenon. Three suggestions derived from the PS rule were
confirmed by showing that: (1) changes in the afferent PS component resulting from
unloading can be converted to changes in the referent position to produce intentional
motor actions, (2) PS is independent of motor commands or sense of effort, (3)
subjects may be unaware of arm motion elicited by continuous tendon vibration
(no-motion illusion). Several theories of action and perception are rejected based on
the present results and previous findings of how motor actions are controlled and
perceived. Such are theories that assume direct preprogramming of the motor or/and
sensory outcome by the brain based on internal models. Our study thus advances the
understanding of how motor actions are controlled and perceived.
Acknowledgments Supported by Natural Sciences and Engineering Research Council (NSERC)

to AGF, Fonds de la recherché en santé du Québec (FRSQ) (Canada) and postdoctoral fellowship
for NI from team grant l’Équipe de Recherche en Réadaptation Sensori-Motrice, CIHR (Canada).
We thank Mindy Levin for the valuable commentaries on the paper and Gevorg Chilingaryan for
the statistical consultation.
References
Adamovich SV, Levin MF, Feldman AG. 1997. Central modifications of reflex parameters may
underlie the fastest arm movements. J Neurophysiol 77:1460–1469.
Adams RA, Shipp S, Friston KJ. 2013. Predictions not commands: active inference in the motor
system. Brain Structure Funct 218:611–643.
Allen TJ, Ansems GE, Proske U. 2007. Effects of muscle conditioning on position sense at the
human forearm during loading or fatigue of elbow flexors and the role of the sense of effort. J
Physiol 580:423–434.
Alstermark B, Isa T, Pettersson L-G, Sasaki S. 2007. The C3-C4 propriospinal system in the
cat and monkey: a spinal pre-motoneuronal centre for voluntary motor control. Acta Physiol
189:123–140.
Arshavsky YI, Berkinblit MB, Fukson OI, Gelfand IM, Orlovsky GN. 1972. Origin of modulation
in neurones of the ventral spinocerebellar tract during locomotion. Brain Res 43:276–279.
Arshavsky YI, Gelfand IM, Orlovsky GN, Pavlova GA. 1978. Messages conveyed by spinocerebel-
lar pathways during scratching in the cat. II Activity of neurons of the ventral spinocerebellar
tract. Brain Res 151:493–506.
Asatryan DG, Feldman AG. 1965. Functional tuning of the nervous system with control of
movements or maintenance of a steady posture: I. Mechanographic analysis of the work of the
joint on execution of a postural task. Biophysics 10:925–935.
Bays PM, Wolpert DM. 2007. Computational principles of sensorimotor control that minimize
uncertainty and variability. J Physiol 578:387–396.
Bridgeman B. 2007. Efference copy and its limitations. Comp Biol Med 37:924–929.
Desmurget M, Reilly KT, Richard N, Szathmari A, Mottolese C, Sirigu A. 2009. Movement
intention after parietal cortex stimulation in humans. Science 324:811–813.
Edin BB, Johansson N. 1995. Skin strain patterns provide kinesthetic information to the human
central nervous system. J Physiol 487:243–251.
Fechner GT. 1860. Elemente der Psychophisik. 2 vols. Leipzig, Breitkopf und Härtel. Engl transl.
(Vol. 1): Elements of Psychophysics. 1966. Adler HE, editor. NY: Holt, Rinehartand, Winston.
Feldman AG. 2009. New insights into action-perception coupling. Exp Brain Res 194:39–58.
Feldman AG. 2011. Space and time in the context of equilibrium-point theory. Wiley
Interdisciplinary Reviews. Cog Sci 2:287–304.
Feldman AG, Latash ML. 1982a. Afferent and efferent components of joint position sense;
interpretation of kinaesthetic illusion. Biol Cybern 42:205–214.
Feldman AG, Latash ML. 1982b. Inversions of vibration-induced senso-motor events caused by
supraspinal influences in man. Neurosci Lett 31:147−151.
Feldman AG, Latash ML. 2005. Testing hypotheses and the advancement of science: recent
attempts to falsify the equilibrium point hypothesis. Exp Brain Res 161:91–103.
Feldman AG, Ilmane N, Sangani S, Raptis H, Esmailzadeh N. 2013. Action-perception coupling:

a new approach. Neuropsychologia 51:2590–2599.
Fowler CA, Turvey MT. 1978. Skill acquisition: An event approach with special reference
to searching for the optimum of a function of several variables. In: Stelmach GE, editor.
Information Processing In Motor Control and Learning. New York: Academic Press, pp. 2–38.
Gandevia SC, Smith JL, Crawford M, Proske U, Taylor JL. 2006. Motor commands contribute to
human position sense. J Physiol 571:703–710.
Gibson JJ. 1968. The Senses Considered as Perceptual Systems. London: George Allen &
Unwin Ltd.
Hulliger M, Nordh E, Vallbo AB. 1982. The absence of position response in spindle afferent units
from human finger muscles during accurate position holding. J Physiol 322:167–179.
Hultborn H. 2006. Spinal reflexes, mechanisms and concepts: from Eccles to Lundberg and
beyond. Progress in Neurobiology, 78, 215–232.
Ilmane N, Sangani S, Feldman AG. 2013. Corticospinal control strategies underlying voluntary
and involuntary wrist movements. Behav Brain Res 236:350–358.
Jankowska E. 1992. Interneuronal relay in spinal pathways from proprioceptors. Prog Neurobiol
38:335–378.
Levin MF, Feldman AG. 1994. The role of stretch reflex threshold regulation in normal and
impaired motor control. Brain Res 657(1–2): 23–30.
Matthews PB. 1972. Mammalian muscle receptors and their central actions. Arnold: London.
McCloskey DI. 1981. Corollary discharges: motor commands and perception. In: Brookhart JM,
Mountcastle VB, editors. Handbook of Physiology: The Nervous System. Bethesda: American
Physiol Soc. Vol. 2 (Pt 2), p 1415–1447.
Merton PA. 1953. Servo action in human voluntary movements. In: Malcolm J L, and Gray JAB,
editors. The Spinal Cord. London: Churchill. p. 247.
Ostry DJ, Darainy M, Mattar AA, Wong J, Gribble PL. 2010. Somatosensory plasticity and motor
learning. J Neurosci 30:5384–5393.
Paillard J (1991). Motor and representational framing of space. In: Paillard J, editor. Brain and
Space. Chapter 10, Oxford: Oxford University Press. pp. 163–182.
Powers WT. 1973. Behavior: The Control of Perception. Chicago: Aldine.
Proske U, Gandevia SG. 2012. The proprioceptive senses: their role in signalling body shape, body
position and movement, and muscle force. Physiol Rev 82:1651–1697.
Raptis H, Burtet L, Forget R, Feldman AG. 2010. Control of wrist position and muscle relaxation
by shifting spatial frames of reference for motoneuronal recruitment: possible involvement of
corticospinal pathways. J Physiol 588:1551–1570.
Roll JP, Gilhodes JC, Tardy-Gervet MF. 1980. Perceptive and motor effects of muscular vibrations in
the normal human: demonstration of a response by opposing muscles. Arch Ital Biol 118:51–71.
Roll JP, Vedel JP, Ribot E. 1989. Alteration of proprioceptive messages induced by tendon vibration
in man: a microneurographic study. Exp Brain Res 76:213–222.
Sangani SG, Raptis HA, Feldman AG. 2011. Subthreshold corticospinal control of anticipatory
actions in humans. Behav Brain Res 224:145–154.
Smith JL, Crawford M, Proske U, Taylor JL, Gandevia SC. 2009. Signals of motor command bias
joint position sense in the presence of feedback from proprioceptors. JApp Physiol 106:950–958.
Vallbo AB. 1974. Human muscle spindle discharge during isometric voluntary contractions.
Amplitude relations between spindle frequency and torque. Acta Physiol Scand 90:319–336.
von Holst H. 1954. Relations between the central nervous system and the peripheral organs. Br J
Animal Beh 2:89–94.
Walsh LD, Gandevia SC, Taylor JL. 2010. Illusory movements of a phantom hand grade with the
duration and magnitude of motor commands. J Physiol 588:1269–1280.
Weber EH. (1834). De pulsu, resorptione, auditu et tactu. Engl translat of chapters De Tactu and
Der Tastsim, In: E.H. Weber’s the sense of touch. 1978. Ross HE, Marray DJ, editors and
translators. UK: Acad Press, Erlbaum, Taylor & Francis.
Chapter 3
Reconfiguration of the Electrical Properties
of Motoneurons to Match the Diverse
Demands of Motor Behavior
C. J. Heckman and Michael D. Johnson
3.1 Introduction
During the past 25 or so years, our studies, as well as those of many colleagues, have
revealed ever-greater complexity in the input–output properties of motoneurons. It
is thus no longer tenable to consider motoneurons to be simple “threshold and fire”
followers of motor commands. This complexity is puzzling though—why is such a
wide range of motoneuron input–output states needed to implement motor behaviors?
The guiding concept for our ongoing work is that this multiplicity of motoneuron
states is in fact an adaptation to match the great diversity of the normal movement
repertoire.
We propose the following correspondences between motoneuron states and motor
behaviors: For posture, motoneurons may act as integrators to provide memory of
brief inputs; for a wide range of voluntary movements, motoneurons may act as
variable gain amplifiers; and for repetitive movements driven by central pattern
generators, motoneurons may act as oscillators. Admittedly, these correspondences
are overly simplified, but they potentially provide clear guides for further studies of
this fundamental issue. The first two states are highly dependent on neuromodulatory
input from the brainstem, mediated via actions of serotonin (5HT) and norepinephrine
(NE) while the oscillatory state appears to depend primarily on glutamatergic N-
methyl-d-aspartate (NMDA) inputs, adding a whole new dimension to motoneuronal
behavior. The data supporting these correspondences are presented in the following
sections, after a brief review of the fundamental role of neuromodulation.
C. J. Heckman () · M. D. Johnson

Departments of Physiology, Physical Medicine and Rehabilitation,
Physical Therapy and Human Movement Sciences, Feinberg School of Medicine,
Northwestern University, Chicago, IL, USA
M. D. Johnson

and Biology 826, DOI 10.1007/978-1-4939-1338-1_3
34 C. J. Heckman and M. D. Johnson
3.2 Motoneuron Input–Output Processing Is Profoundly

Influenced by Brainstem Neuromodulatory Systems
Neuromodulation is the foundation of the diversity of motoneuronal states. The

two most powerful neuromodulatory inputs to motoneurons are mediated by axons
descending from the brainstem and releasing either 5HT or NE. Unlike ionotropic in-
puts, which open ion channels to generate excitatory postsynaptic potentials (EPSPs)
or inhibitory postsynaptic potentials (IPSPs), neuromodulatory inputs use G protein-
coupled receptors to activate signal transduction pathways (Hille 2001). The effects of
5HT and NE on motoneuron excitability are extraordinary (Hultborn 2002; Heckman
and Enoka 2012). Recruitment thresholds are greatly reduced by depolarization of the
rest potential and hyperpolarization of the spike voltage threshold. Rate modulation
is transformed due to facilitation of persistent inward currents (PICs) mediated by
voltage-sensitive Na and Ca channels. PICs are a primary focus of this proposal. Un-
doubtedly, motoneurons are subject to neuromodulation by other neurotransmitters
in addition to 5HT and NE (e.g., Svirskis and Hounsgaard 1998). The dopaminergic
input from the brainstem to the cord increases motoneuronal excitability, but has yet
only been studied in neonatal preparations (Garraway and Hochman 2001; Zhu et al.
2007). It is likely weaker than the 5HT and NE systems. Cervical motoneurons in
the cat have over 1500 synapses containing 5HT and over 1200 synapses containing
NE and these synapses cover their extensive dendritic trees (Montague et al. 2013).
Local spinal circuits are known to induce neuromodulatory effects on motoneurons
as well. Cholinergic interneurons projecting to C-terminals on motoneurons also
reduce the afterhyperpolarization (AHP) (Zagoraiou et al. 2009) and possibly this
is the source of the reduced AHP that occurs during the scratch reflex in spinalized
animals (Power et al. 2010). In contrast, 5HT and NE have very little effect on the
AHP in the adult (Lee and Heckman 1999; Li et al. 2007). Spinal circuits also lower
spike voltage threshold (Power et al. 2010). In this review, our focus is on neuromod-
ulation by 5HT and NE for two reasons: their effects on motoneurons are remarkably
powerful and reasonably well understood. Further states of motoneurons will likely
become apparent as our understanding of the full diversity of neuromodulatory inputs
increases.
3.2.1 Neuromodulatory Inputs are an Essential Component

of Normal Motor Commands
Numerous studies indicate that the brainstem neuromodulatory input is essential for
a wide range of motor behaviors (Heckman and Enoka 2012). To summarize: (1)
There exists steady neuromodulatory drive from the brainstem to the cord during the
waking state, which is increased during sustained motor output (Aston-Jones et al.
2001; Jacobs et al. 2002). (2) Motoneurons lacking this neuromodulatory drive are
severely hypoexcitable (Hounsgaard et al. 1988; Miller et al. 1996). Excitability is
3 Reconfiguration of the Electrical Properties of Motoneurons . . . 35
so low that even simultaneous, intense activation of all known inputs is far from
sufficient to produce the maximal motoneuronal output required for high forces
(Powers and Binder 2001; Cushing et al. 2005). (3) Motor unit firing patterns in
humans during slow increases in force closely resemble the firing patterns generated
by equally slow activation of cat motoneurons with moderate to large amplitude PICs
(Hornby et al. 2002; Heckman and Enoka 2012). In summary, normal motor behavior
cannot be achieved without a strong neuromodulatory input to motoneurons. This
input has a profound impact on motoneuron input–output processing, i.e., on its
“state.”
3.2.2 The PIC and Bistable Behavior as a Pattern Generator

for Postural Control
Bistable behavior reflects the inherent tendency of PICs to prolong inputs. Figure 3.1
shows an example of how the PIC transforms a synaptic current. The PIC converts
a brief excitatory current (green trace) into a prolonged “tail” current that persists
for many seconds (red trace; inward, depolarizing current is upward). The sustained
PIC tail current is the basis of self-sustained firing, which requires little or no tonic
input from descending or sensory systems. Recordings of electromyogram (EMG)
and force in response to brief inputs have documented prolonged motoneuronal
output lasting minutes (Hounsgaard et al. 1988; unpublished data, Heckman lab).
Bistable behavior has long been thought to be important for posture (Hounsgaard
et al. 1988). Consistent with this role, PICs with low-voltage thresholds and long-
lasting tail currents only exist in type S motoneurons that innervate fatigue resistant
muscle fibers, while PICs in type F motoneurons have much higher thresholds and
faster decays (Lee and Heckman 1998a, 1998a). This PIC “tuning” makes good
functional sense: Type F motoneurons do not participate in the low forces needed
for steady posture and their extremely low fatigue resistance seems well matched
to a more rapidly decaying PIC. The match between PIC properties and the fatigue
resistance of motor units was a key insight leading to our concept that different
motoneuronal states match different motor behaviors. In a sense, bistable behavior
can be considered a pattern generator for the steady forces required for posture.
Remarkably, this pattern generator is intrinsic to motoneurons themselves.
3.2.3 Does the Bistable State Provide Integration or Memory

of Transient Inputs?
Figure 3.1 suggests that a bistable motoneuron is not an integrator in the strict mathe-
matical sense—the prolonged current is less than the initial current during the pulse.
This history-dependent behavior, however, does have a strong memory-like aspect
to it. For posture, the key functional issue is how a motoneuron in the self-sustained
Fig. 3.1 Amplification and prolongation of synaptic input by persistent inward current. Intracellular
recording of a medial gastrocnemius motoneuron in the decerebrate cat preparation. The synaptic
current was generated by vibration of the Achilles tendon to steadily activate muscle spindle Ia
afferents for ∼ 0.75 s. The cell was voltage clamped using the switching single electrode technique.
The green trace shows the current generated by the steady Ia input when the cell was voltage clamped
to a very hyperpolarized level, preventing this input from activating the motoneuron’s persistent
inward current. The red trace shows the effect of the very same input when the cell was clamped
near its threshold for firing. Because the clamp is only effective at the soma, the Ia input activated
the PIC in dendritic regions. As a result, the input was both strongly amplified and prolonged—the
tail current persisted for many seconds (not shown). (Data from Lee and Heckman 1996)
firing state responds to further transients for postural corrections. Previous studies
of the effects of PICs on synaptic inputs have been done using a combination of
synaptic and injected currents (e.g., Bennett et al. 1998; Lee and Heckman 2000)
and have ignored how the cell behaves when primarily driven by the PIC to produce
self-sustained firing. Further studies are needed to define the behavior of the mo-
toneuron in the postural state. If the response during self-sustained firing to further
transients is prolonged, then the motoneuron is indeed acting much like an integrator
or memory device. If not, then bistable behavior may provide a steady background
that nonetheless allows for transient corrections.
3.2.4 PIC Amplification as the Basis for Gain Control of Motor

Output via the Brainstem
Figure 3.1 shows that the PIC not only prolongs input but also greatly enhances it
(note the large increase in current while the input is active). This enhancement is
generally thought to provide input amplification (Lee and Heckman 2000; Prather
et al. 2001; Hultborn et al. 2003) and this amplification is essential in increasing
the gain of motoneurons sufficiently so that they can be activated effectively by
descending and sensory inputs (see above). PICs are highly sensitive to inhibitory
inputs (Kuo et al. 2003; Bui et al. 2008b) and so motor commands can easily de-
couple PIC amplification from PIC prolongation by using inhibition to turn off the
PIC when it is not needed. The amplitude of the PIC is proportional to the level of
neuromodulatory input (Lee and Heckman 2000). Thus, brainstem neuromodulatory
control of PICs potentially provides motor commands with a mechanism to control
the input–output gain of motoneurons. We have speculated that a low motoneuronal
gain is advantageous for low-force, high-precision tasks but that gain should be
progressively increased as the force requirements of the task increase (Johnson and
Heckman 2010). Matching of motoneuronal gain to the widely varying forces of
voluntary motor tasks would constitute a clear example of the diversity of motoneu-
ronal behavior matching the diversity of motor behaviors, but this match has yet to
be clearly demonstrated.
3.2.5 Does Synaptic Inhibition of PICs Afford a Local, Highly

Specific Gain Control of Motoneurons?
Inhibitory synaptic input hyperpolarizes motoneurons, thereby increasing their

threshold for activation and opposing the depolarization from excitatory inputs.
Probably, most considerations of the role of inhibition in motor control assume this
fundamental model. Yet the effect of inhibition on the PIC suggests that it may also
alter motoneuron gain, because we have shown that inhibition can reduce the ampli-
tude of the PIC in a proportional manner (Kuo et al. 2003). Thus, inhibitory input
may provide a gain control mechanism that opposes the gain control from the brain-
stem. The brainstem system is remarkably diffuse—in fact some of these axons give
off branches along the entire length of the cord (Holstege and Kuypers 1987). In con-
trast, there are several highly focused sources of spinal inhibition—e.g., reciprocal
inhibition, recurrent inhibition, and low threshold cutaneous inhibition (Baldissera
et al. 1981; Jankowska 2001). A previous suggestion that recurrent inhibition may
provide gain control (Hultborn et al. 1979) faltered because of its relatively small
ionotropic currents (Lindsay and Binder 1991), but the effect of inhibition on the PIC
potentially provides a much larger impact on gain (Bui et al. 2008a). If inhibitory
inputs can reduce gain by reducing PIC amplitude, then local spinal circuits would
have the capacity for a focused, specific control of motoneuron gain (Heckman et al.
2008; Johnson and Heckman 2010).
3.2.6 Is the Oscillatory Motoneuron State Linked to CPGs

for Repetitive Behaviors?
NMDA receptors are ionotropic but their behavior is unique in that they are voltage-
sensitive because depolarization relieves the Mg2+ block of the channel (Hille
2001). The resulting synaptic currents behave much like a PIC and can generate
large plateau potentials. Unlike PICs, however, the NMDA plateau potential is self-
terminating, probably because of activation of an outward current (Manuel et al.
2012; Wang et al. 2013). The resulting NMDA-mediated oscillations have been con-
sistently demonstrated in neonatal interneurons and motoneurons (Harris-Warrick
2011). These oscillations are intrinsic to the cell, in that they persist in the presence
of tetrodotoxin (TTX) to block all inputs (Hochman et al. 1994; MacLean et al. 1997).
We have recently demonstrated that the response of motoneurons to NMDA does not
fade with maturation. Instead, adult rodent motoneurons exhibit very strong intrinsic
oscillations in response to bath administration of NMDA—exactly like the neonatal
results (Manuel et al. 2012). Thus, NMDA-mediated input may place the adult mo-
toneuron in an entirely new, oscillatory state. The actual input system that provides
the NMDA input to spinal motoneurons is unknown, but a recent study in phrenic
motoneurons (Enriquez Denton et al. 2012) supports the concept that spinal central
pattern generators (CPGs) for repetitive behaviors are a major source. If NMDA
mediated oscillations in spinal motoneurons are in fact driven from CPGs, then this
previously unknown “oscillation” state would match the electrical behavior of the
motoneuron to the demands of repetitive motor tasks like scratch and locomotion.
3.3 Summary
Thus far, considerable data indicate that the electrical properties of motoneurons are
reconfigured by the interactions between neuromodulation, inhibition, and NMDA-
mediated inputs. We present here the hypothesis that these different states are required
to match the properties of motoneurons to the extreme diversity of the normal move-
ment repertoire. Considerable further work is required to establish this hypothesis
more firmly but it appears to make good functional sense.
Acknowledgments The studies supporting the work reviewed here were supported by NIH NINDS
grants NS034382, NS071951, and NS077863. The authors thank Rochelle Bright for assistance
with writing.
References
Aston-Jones G, Chen S, Zhu Y, Oshinsky ML. 2001. A neural circuit for circadian regulation of
arousal. Nat Neurosci 4:732–738.
Baldissera F, Hultborn H, Illert M. 1981. Integration in spinal neuronal systems. In: Brooks VB,
editor. Handbook of Physiology. The Nervous System. Motor Control. Bethesda: American
Physiological Society. pp. 509–595.
Bennett DJ, Hultborn H, Fedirchuk B, Gorassini M. 1998. Synaptic activation of plateaus in hindlimb
motoneurons of decerebrate cats. J Neurophysiol 80:2023–2037.
Bui TV, Grande G, Rose PK. 2008a. Relative location of inhibitory synapses and persistent in-
ward currents determines the magnitude and mode of synaptic amplification in motoneurons.
J Neurophysiol 99:583–594.
Bui TV, Grande G, Rose PK. 2008b. Multiple modes of amplification of synaptic inhibition to
motoneurons by persistent inward currents. J Neurophysiol 99:571–582.
Cushing S, Bui T, Rose PK. 2005. Effect of nonlinear summation of synaptic currents on the
input-output properties of spinal motoneurons. J Neurophysiol 94:3465–3478.
Enriquez Denton M, Wienecke J, Zhang M, Hultborn H, Kirkwood PA. 2012. Voltage-dependent
amplification of synaptic inputs in respiratory motoneurones. J Physiol 590:3067–3090.
Garraway SM, Hochman S. 2001. Modulatory actions of serotonin, norepinephrine, dopamine, and
acetylcholine in spinal cord deep dorsal horn neurons. J Neurophysiol 86:2183–2194.
Harris-Warrick RM. 2011. Neuromodulation and flexibility in Central Pattern Generator networks.
Curr Opin Neurobiol 21:685–692.
Heckman CJ, Enoka RM. 2012. Motor Unit. Compr Physiol 2:2629–2682.
Heckman CJ, Hyngstrom AS, Johnson MD. 2008. Active properties of motoneurone dendrites:
diffuse descending neuromodulation, focused local inhibition. J Physiol 586:1225–1231.
Hille B. 2001. Ionic Channels of Excitable Membranes, 3rd Edition. Sunderland, MA: Sinauer
Assoc. Inc.
Hochman S, Jordan LM, Schmidt BJ. 1994. TTX-resistant NMDA receptor-mediated voltage
oscillations in mammalian lumbar motoneurons. J Neurophysiol 72:2559–2562.
Holstege JC, Kuypers HG. 1987. Brainstem projections to spinal motoneurons: an update. Neurosci
23:809–821.
Hornby TG, McDonagh JC, Reinking RM, Stuart DG. 2002. Motoneurons: a preferred firing range
across vertebrate species? Muscle Nerve 25:632–648.
Hounsgaard J, Hultborn H, Jespersen B, Kiehn O. 1988. Bistability of alpha-motoneurones in the
decerebrate cat and in the acute spinal cat after intravenous 5-hydroxytryptophan. J Physiol
405:345–367.
Hultborn H. 2002. Plateau potentials and their role in regulating motoneuronal firing. Adv Exp Med
Biol 508:213–218.
Hultborn H, Lindstrom S, Wigstrom H. 1979. On the function of recurrent inhibition in the spinal
cord. Exp Brain Res 37:399–403.
Hultborn H, Denton ME, Wienecke J, Nielsen JB. 2003. Variable amplification of synaptic input to
cat spinal motoneurones by dendritic persistent inward current. J Physiol 552:945–952.
Jacobs BL, Martin-Cora FJ, Fornal CA. 2002. Activity of medullary serotonergic neurons in freely
moving animals. Brain Res Rev 40:45–52.
Jankowska E. 2001. Spinal interneuronal systems: identification, multifunctional character and
reconfigurations in mammals. J Physiol 533:31–40.
Johnson MD, Heckman CJ. 2010. Interactions between focused synaptic inputs and diffuse
neuromodulation in the spinal cord. Ann NY Acad Sci 1198:35–41.
Kuo JJ, Lee RH, Johnson MD, Heckman HM, Heckman CJ. 2003. Active dendritic integration of
inhibitory synaptic inputs in vivo. J Neurophysiol 90:3617–3624.
Lee RH, Heckman CJ. 1996. Influence of voltage-sensitive dendritic conductances on bistable firing
and effective synaptic current in cat spinal motoneurons in vivo. J Neurophysiol 76:2107–2110.
Lee RH, Heckman CJ. 1998a. Bistability in spinal motoneurons in vivo: systematic variations in
rhythmic firing patterns. J Neurophysiol 80:572–582.
Lee RH, Heckman CJ. 1998b. Bistability in spinal motoneurons in vivo: systematic variations in
persistent inward currents. J Neurophysiol 80:583–593.
Lee RH, Heckman CJ. 1999. Enhancement of bistability in spinal motoneurons in vivo by the
noradrenergic alpha(1) agonist methoxamine. J Neurophysiol 81:2164–2174.
Lee RH, Heckman CJ. 2000. Adjustable amplification of synaptic input in the dendrites of spinal
motoneurons in vivo. J Neurosci 20:6734–6740.
Li X, Murray K, Harvey PJ, Ballou EW, Bennett DJ. 2007. Serotonin facilitates a persistent calcium
current in motoneurons of rats with and without chronic spinal cord injury. J Neurophysiol
97:1236–1246.
Lindsay AD, Binder MD. 1991. Distribution of effective synaptic currents underlying recurrent
inhibition in cat triceps surae motoneurons. J Neurophysiol 65:168–177.
MacLean JN, Schmidt BJ, Hochman S.1997. NMDA receptor activation triggers voltage oscilla-
tions, plateau potentials and bursting in neonatal rat lumbar motoneurons in vitro. Eur J Neurosci
9:2702–2711.
Manuel M, LiY, Elbasiouny SM, Murray K, Griener A et al. 2012. NMDA induces persistent inward
and outward currents that cause rhythmic bursting in adult rodent motoneurons. J Neurophysiol
108:2991–2998.
Miller JF, Paul KD, Lee RH, Rymer WZ, Heckman CJ. 1996. Restoration of extensor excitability
in the acute spinal cat by the 5-HT2 agonist DOI. J Neurophysiol 75:620–628.
Montague SJ, Fenrich KK, Mayer-Macaulay C, Maratta R, Neuber-Hess MS et al. 2013. Nonuni-
form distribution of contacts from noradrenergic and serotonergic boutons on the dendrites of
cat splenius motoneurons. J Comp Neurol 521:638–656.
Power KE, McCrea DA, Fedirchuk B. 2010. Intraspinally mediated state-dependent enhancement
of motoneurone excitability during fictive scratch in the adult decerebrate cat. J Physiol 588:
2839–2857.
Powers RK, Binder MD. 2001. Input-output functions of mammalian motoneurons. Rev Physiol
Biochem Pharmacol 143:137–263.
Prather JF, Powers RK, Cope TC. 2001. Amplification and linear summation of synaptic effects on
motoneuron firing rate. J Neurophysiol 85:43–53.
Svirskis G, Hounsgaard J. 1998. Transmitter regulation of plateau properties in turtle motoneurons.
Wang D, Grillner S, Wallen P. 2013. Calcium dynamics during NMDA-induced membrane potential
oscillations in lamprey spinal neurons—contribution of L-type calcium channels (CaV1.3). J
Physiol 591(Pt 10):2509–2521.
Zagoraiou L, Akay T, Martin JF, Brownstone RM, Jessell TM et al. 2009. A cluster of cholinergic
premotor interneurons modulates mouse locomotor activity. Neuron 64:645–662.
Zhu H, Clemens S, Sawchuk M, Hochman S. 2007. Expression and distribution of all dopamine
receptor subtypes (D(1)-D(5)) in the mouse lumbar spinal cord: a real-time polymerase chain
reaction and non-autoradiographic in situ hybridization study. Neurosci 149:885–897.
Chapter 4
The Regulation of Limb Stiffness in the Context
of Locomotor Tasks
T. Richard Nichols, Jinger S. Gottschall and Christopher Tuthill
4.1 Moving in the Real World: Requirements for Muscular

Activation and Limb Mechanics During Ramp Walking
Humans and other animals encounter varieties of terrain during terrestrial locomo-
tion, including irregular and sloped surfaces. To navigate the natural terrain and
its transitions, the motor system must be capable of adjusting to new physical re-
quirements and making these adjustments quickly. In the case of sloped surfaces,
alterations in muscular activation are necessary to deal with the change in orientation
with respect to the gravity vector, and also with changes in the mechanical interaction
of the limbs and the support surface. Furthermore, changes in the mechanical proper-
ties of the limbs are likely to be involved given that muscular function changes with
slope. Walking up a slope requires predominantly propulsive action with reinforce-
ment of force and stiffness, while walking down a slope requires some limb muscles
to operate in an eccentric mode (Maas et al. 2010) in which they provide a braking
action (Cham and Redfern 2002; Gregor et al. 2006; Lay et al. 2006; Prilutsky et al.
2011) with moderate stiffness.
This chapter provides a brief review of recent studies focused on these two as-
pects of motor control, namely, the manner in which patterns of muscular action
and limb mechanics are modulated for slope walking. In these investigations, we
used quadrupedal animal subjects. Since locomotion on sloped surfaces requires a
substantial reorientation of the quadrupedal body, we presumed that this comparative
approach would be helpful in discovering the underlying principles of control. As
discussed below, it turned out that these principles are readily applicable to bipedal
T. R. Nichols () · C. Tuthill

School of Applied Physiology and Department of Biomedical Engineering,
Georgia Institute of Technology, 555 14th Streeet NW, Atlanta, GA 30332-0356, USA
J. S. Gottschall
Department of Kinesiology, Pennsylvania State University, State College, USA

and Biology 826, DOI 10.1007/978-1-4939-1338-1_4
42 T. R. Nichols et al.
locomotion. The account presented here concerns the physical requirements and cor-
responding motor adaptations necessary for locomotion on sloped surfaces. Here we
provide evidence that the spinal cord changes state1 for different motor behaviors,
including modifications in pattern-generating circuits as well as proprioceptive cir-
cuits that regulate limb mechanics. A discussion of the manner in which humans
manage transitions in slope can be found in a recent publication (Gottschall and
Nichols 2011). Since transitions in slope constitute a major contributor to injury in
older individuals (Ashley et al. 1977; Tinetti et al. 1995; Polcyn et al. 1998), it is
important to understand both the differences in central neural states during walking
on different slopes and the manner and speed of the transitions between these states.
The relationships among limb mechanics, muscular activation patterns, and neural
control during locomotion have been studied extensively in the cat (Fowler et al. 1993;
Gregor et al. 2006; Maas et al. 2007; Prilutsky et al. 2011). During level walking, the
hind limb swings forward, briefly yields for weight acceptance upon paw contact,
then extends to provide propulsion for the next step during the subsequent stance
phase (Goslow et al. 1973). During upslope walking, there is little or no yield at
stance onset (Carlson-Kuhta et al. 1998; Nichols et al. 1999; Abelew et al. 2000;
Gregor et al. 2006), and there is an additional requirement for limb propulsion during
subsequent stance due to a reorientation of the body with respect to the gravity vector.
During downslope walking, the reorientation of the body with respect to gravity tends
to pitch the animal forward, requiring braking at the hip rather than propulsion.
Furthermore, the distal joints undergo more substantial yield as they participate in
braking (Nichols et al. 1999; Abelew et al. 2000; Gregor et al. 2006).
In a classical series of articles, Dr. Judith Smith and her colleagues described the
patterns of muscular activation in felines corresponding to these and other motor tasks
(Carlson-Kuhta et al. 1998; Smith et al. 1998) and the implications of these patterns
for the organization of pattern-generating networks in the spinal cord. These results
were later extended to include kinetic measurements (Gregor et al. 2006). Examples
of key findings are as follows. During level walking, the hip extensor/knee flexor
(anterior biceps femoris—aBF) subserves propulsion, and the ankle extensor (lateral
gastrocnemius—LG) provides compliance during the initial yield and propulsion
during the later phase of stance. The hip flexor (iliopsoas—ILIO) protracts the limb
during swing and is silent during stance. During upslope walking, LG and aBF
are recruited to a greater extent to provide additional propulsion, but maintain the
same pattern as observed during level walking. The rectus femoris muscle (RF), a hip
flexor and knee extensor, is recruited in late stance to contribute to propulsion. During
downslope walking, braking at the hip is mediated by a reversal in the recruitment
patterns of aBF and ILIO during stance, such that ILIO is recruited but not aBF
(the “downslope” pattern). In addition, RF is also recruited early in stance and its
activity late in stance is amplified, presumably to contribute to braking at the hip.
1
The term “state” refers to a given organization of synaptic interactions and active neuronal path-
ways. For example, during locomotion, pathways mediating excitatory force feedback become
active, altering the state of the spinal cord.
4 The Regulation of Limb Stiffness in the Context of Locomotor Tasks 43
The ankle extensor LG is recruited less during downslope walking, especially during
initial stance. Changes in body configuration were also recorded in these studies.
The animals tended to maintain their heads relatively level, such that the head would
be pitched downward for walking upslope and upward for walking downslope, so
that body orientation and neck angle changed systematically with slope.
4.2 Central Pattern-Generating Circuits are Modulated by a

Body Orientation Signal for Walking on Sloped Surfaces
It occurred to us that a signal coding the orientation of the animal’s body might
be responsible for mediating the altered patterns of activity for downslope walking
in addition to possible contributions from cutaneous and proprioceptive input from
the limbs. There is a substantial body of literature from the laboratories of Ottavio
Pompeiano and Victor Wilson revealing the presence of such an orientation signal
in the vestibular nuclei that reaches both the cerebellum and the spinal cord (Suzuki
et al. 1985; Manzoni et al. 2004). This signal is obtained by integration of vestibular
and afferent input from muscle spindles in the muscles of the neck (Fig. 4.1, left
panel). A signal corresponding to head orientation is detected by the otolith organs
in the vestibular apparatus. Muscle spindle receptors in the small muscles linking
the cervical vertebrae detect the configuration of the neck and thereby can supply
information concerning the spatial relationship of the head with respect to the body
(Lindsay et al. 1976). These two signals are combined in the vestibular nuclei to
provide an estimate of body orientation. These results were originally interpreted
in the context of the problem posed by von Holst and Mittelstaedt (1950), that
proprioceptive inputs resulting from voluntary movement could interfere with the
movement. These authors considered the “interaction of several afferences,” in which
two different sensory systems could interact to influence the ongoing movement
rather than a given sensory system interacting with an efference copy signal. In
this case, voluntary movements of the head would affect both vestibular and neck
afferent input in opposite directions resulting in no change in the body orientation
signal (von Holst and Mittelstaedt 1950; Boyle and Pompeiano 1981). Only by a
change in the pitch angle of the body would a change in the corresponding signal be
generated. This mechanism, a combination of “tonic neck” and “tonic labyrinthine”
pathways, was originally conceived as part of a postural support system (Magnus
and de Kleijn 1912; von Holst and Mittelstaedt 1950). We hypothesized (Gottschall
and Nichols 2007, 2011) that this signal representing body orientation might be
utilized to modify the pattern formation component of the central pattern-generating
networks (Lafreniere-Roula and McCrea 2005) as well. It was this hypothesis that
motivated our further studies.
To independently control head and body position, we employed the premammil-
lary decerebrate cat preparation. Under deep surgical anesthesia, the cortices, basal
ganglia, and a substantial portion of the thalamus were removed, with preservation
of the subthalamic nucleus (Grillner and Shik 1973). The head of the animal was
Fig. 4.1 The left panel shows a conceptual model providing the origin of a body orientation signal.
Physical interactions are shown in blue. The orientation of the head is detected by the otolith organs.
Proprioceptors in the neck muscles detect neck angle, which is determined by the difference between
head and body orientations. The neck angle and head orientation are recombined through neural
pathways to provide an estimate of body orientation, which is distributed to pattern-generating
networks in the spinal cord as well as proprioceptive circuits. Movements of the head alone result
in cancellation of the two signals. The right panel depicts the result of destruction of the vestibular
apparatus. In this case, movements of the head alone lead to modulation of pattern generation and
proprioceptive feedback. MLR the mesencephalic locomotor region
fixed in a stereotaxic apparatus mounted on the treadmill. The angle of the head
with respect to the treadmill surface could be adjusted by adjusting the angle of the
stereotaxic frame. Electromyographic (EMG) electrodes were implanted in selected
muscles to measure muscular activation. When anesthesia was withdrawn, this prepa-
ration exhibited well-coordinated stepping on a moving treadmill. The patterns of
EMG activity were in substantial agreement with the patterns observed previously
(Carlson-Kuhta et al. 1998; Smith et al. 1998; Gregor et al. 2006). During level
walking, LG and the posterior biceps femoris muscle (pBF, which contributes to hip
extensor as does aBF) were recruited during stance, and ILIO was recruited during
swing. When we changed the pitch angle of the treadmill (and therefore body ori-
entation) to correspond to downslope walking, we observed the downslope pattern,
namely activation of ILIO but not pBF during stance (Gottschall and Nichols 2007).
It should be noted that, since the head was fixed in the stereotaxic frame, altering the
pitch angle of the treadmill resulted in a change in the angle of the body and head
with respect to gravity, but preserved the angle of the head with respect to the body.
Therefore, the main sensory influence was on the vestibular apparatus with little or
Level
Downslope
Lateral gastrocnemius m posterior biceps femoris miliopsoas m
Swing
Stance
Fig. 4.2 Evidence that a body orientation signal regulates the pattern-generating networks in the
spinal cord. Electromyographic activity of three selected muscles in the step following tilting of the
head up are shown, for level walking (upper traces) and for walking down a slope (lower traces).
The anatomical sketch indicates the approximate locations of the posterior biceps and iliopsoas
(ILIO) muscles. Note that the hamstrings muscle posterior biceps was not activated during stance
in the downslope condition, and the hip flexor ILIO became active during stance. The change in
pattern corresponded to a switch from propulsion to braking between the two tasks. Although an
alteration in head position alone should evoke no change in muscular patterns since body orientation
is constant, the change occurred because neck afferent input is more rapid than vestibular input.
This pattern of activity reverted back to the level walking state within two to three steps as vestibular
and neck afferent feedback finally cancelled. After destruction of the vestibular system, the change
in pattern remained in effect as long as the head was tilted (data not shown)
no effect on neck angle (see Fig. 4.1). In addition, the recruitment of LG was reduced,
in agreement with previous findings (Gregor et al. 2006). These changes took place
over two to three steps. In the next experiment, we left the treadmill level and changed
the pitch of the head by altering the angle of the stereotaxic frame. Under these con-
ditions, both vestibular and neck afferent inputs would be affected. According to the
“interaction of several afferences” (von Holst and Mittelstaedt 1950), we should have
seen no change in the pattern of muscular activity, since the vestibular and the neck
afferent signals should have cancelled, corresponding to a voluntary head movement.
However, when we changed the pitch of the head upward, we observed a transient
change in muscular activation as appropriate to downslope walking (Fig. 4.2), but
a return to the pattern observed for level walking after a few steps. Combining the
results of the two manipulations, we reasoned that the afferent feedback from mus-
cle spindle receptors in the neck musculature had a relatively rapid effect on the
pattern formation network, while the vestibular input had a more delayed effect.
The expected cancellation of the two signals after tilt of the head alone did occur,
but only after a delay determined by the lag time of vestibular processing.
To test this hypothesis directly, we performed an additional set of experiments in

which the vestibular apparatus of the animal preparations was disrupted surgically
(Fig. 4.1, right panel; Gottschall and Nichols 2011). In these cases, changing the
pitch angle of the head alone resulted in ongoing changes in activation patterns of
ILIO and pBF. When the head was tilted up, muscular activation switched to the
downslope pattern and remained in that state as long as the head was tilted up.
Since the vestibular signal was no longer present to counterbalance the neck afferent
signal, the pattern formation network was influenced by neck afferents alone (refer to
Fig. 4.1). In summary, we had provided strong evidence that the modulation of central
pattern-generating networks for slope walking is mediated by a body orientation
signal derived from vestibular and neck afferent feedback.
An additional observation from the experiments described above was that the
kinematics of locomotion were not visibly altered when head pitch was altered. This
suggests that signals from muscle spindle receptors were not likely to change across
the three conditions, nor were signals from cutaneous receptors. It is likely, however,
that force feedback from Golgi tendon organs might have changed systematically
with changes in muscular activation, but not enough to alter limb kinematics. In any
case, it appeared that a body orientation signal derived from vestibular and neck
afferent feedback had a profound effect on the pattern formation networks in the
spinal cord.
Changes in body orientation are not as marked for human bipedal locomotion
as for quadrupeds, but the changes in muscular function during slope walking are
similar. That is, muscles are used in an eccentric or braking mode to a greater
extent during downslope locomotion than during level or upslope walking (Cham
and Redfern 2002; Lay et al. 2006). In addition, changes in trunk orientation have
also been observed, at least for upslope walking (Prentice et al. 2004). Although the
changes in trunk orientation of the cat are more dramatic than in the bipedal human,
these data suggest that the evidence from the experiments reported here are relevant
to human slope walking. It should also be noted that proprioceptive feedback from
within the limb could contribute to the regulation of muscular patterns during walking
on various surfaces. The mechanism we describe here appears to account for much
of the adjustment of motor patterns in the cat, but this issue remains to be explored
in the human. Indeed, EMG data were also reminiscent of the patterns of activity
observed in the cat (Lay et al. 2007). Activity of the biceps femoris (BF) muscle was
observed through a substantial portion of the stance phase during upslope walking,
but was absent during level and downslope walking. Apparently this muscle was
recruited when substantial propulsive force was needed. Although activity of ILIO
was not measured in these studies, RF, which has a hip flexor action, showed an
additional burst of activity during stance that was absent for both upslope and level
conditions. The patterns of activity of this muscle were somewhat different between
human and cat, but in both cases, the increased duration of its activity was likely to
result in a contribution to braking action at the hip during downslope walking (Smith
et al. 1998; Lay et al. 2007). It cannot be excluded that cutaneous and proprioceptive
feedback contribute to the regulation of the pattern formation network in either cat
or human, but the results discussed here indicate that the vestibular system and neck
afferents contribute importantly to this regulation.
Electrophysiological studies in the cat indicate that three descending systems
may contribute to the transmission of the body orientation signal to the spinal cord,
namely, the propriospinal, vestibulospinal, and reticulospinal systems (Srivastava
et al. 1984; Brink et al. 1985; Marchand et al. 1987). Since it had been shown
that the body orientation signal can be recorded in cells of the vestibular nuclei, the
vestibulospinal system is the most likely candidate for the effects observed in our
studies, but contributions from the other two systems cannot be excluded. Our obser-
vations that changes in muscular activation patterns in response to manipulations of
the body orientation signal without concomitant changes in kinematics, indicate the
importance of extrinsic feedback to the limb in the regulation of locomotion across
motor tasks. It should be noted that alterations in activation patterns of muscles in
the feline forelimb were also found during locomotion on different slopes (Smith
et al. 1998). Evaluation of the sensory mechanisms underlying changes in forelimb
muscular activity can be accomplished using the methods described here, and these
studies are planned for the future. Furthermore, the potential role of proprioceptive
inputs from within the limb remains to be explored.
4.3 Endpoint Stiffness Is Also Modulated Appropriately

by the Body Orientation Signal
In addition to reconfiguration of central pattern-generating networks for slope walk-

ing, changes in the mechanical properties of the limbs are also necessary. During
upslope walking, the antigravity muscles are used primarily for propulsion and un-
dergo predominantly active shortening. Under these conditions, reinforcement of
force output is important, and excitatory force feedback may contribute substantially
to this reinforcement (Duysens and Pearson 1980; Duysens et al. 2000; Donelan and
Pearson 2004; Donelan et al. 2009). Translated into mechanical terms, this reinforce-
ment would be measured as an increase in stiffness. During downslope walking, high
stiffness would not be appropriate given that the musculature is utilized for braking,
or eccentric mode where moderate stiffness would be more appropriate to absorb the
energy resulting from negative work. We reasoned that the same signals that regulate
pattern formation networks might also regulate limb stiffness through modulation of
proprioceptive pathways.
In more recent experiments, we have measured limb stiffness in the decerebrate
preparation in the level and downslope states. In this case, we employed the conven-
tional intercollicular decerebrate cat, since active locomotion was not required. In this
preparation, the brainstem is transected through the superior colliculi. Interruption of
the pathways from the subthalamic nucleus precludes spontaneous locomotion that
is characteristic of the premammillary preparation. Limb stiffness was measured by
connecting the pads of both hind paws to force–moment sensors mounted on multi-
axis robotic arms and by stabilizing the pelvis of the animal. In this way, we could
3
Head Level
Head Up (Downhill)
2.5
1.5
Force (N)
0.5
−0.5
−0.2 −0.1 0 0.1 0.2 0.3 0.4 0.5
Time (s)
Fig. 4.3 Demonstration that stiffness measured at the endpoint of the limb is reduced for the spinal
state corresponding to walking down a slope (head tilt up following labyrinthectomy). Traces
represent force trajectories in response to 2 cm perturbations of the limb imposed by a robotic arm
along the long axis of the limb in the direction of limb compression. A large number of responses
were collected over a range of background forces, and the means for the two conditions (head level
and head up) are shown. The main effect of condition is statistically significant in the shaded region
measure forces in all three directions in response to motions of the robotic arm. As
in the previous experiments described above, the two states were produced by alter-
ing head pitch following destruction of the vestibular system. Measurements were
obtained from each limb individually, since we desired to isolate the contribution of
sensory pathways within the limb. Figure 4.3 illustrates the force trajectory for the
limb when subjected to a ramp-and-hold perturbation delivered along the long axis
of the limb for the two states. The magnitude of the force is smaller in the downslope
state, in confirmation of our hypothesis. Each trajectory is the average of a large
number of individual responses obtained over a range of background forces, so the
result is a main effect of head position. Even more recently, we have measured stiff-
ness in different directions and confirmed that tilting the head-up results in a lower
stiffness in multiple directions (Tuthill and Nichols 2012). Endpoint stiffness of a
limb is usually expressed as an ellipsoidal field, with the major axis directed approx-
imately along the long axis of the limb. In our recent results, both major and minor
axes were reduced in magnitude in the downslope condition. These results were ob-
tained during static force production in intercollicular decerebrate preparations, so
the associated pathways apparently operate independently of locomotion.
4.4 Modulation of Stiffness Is Mediated by Changes

in Intermuscular Force Feedback
The observation of modulation of endpoint stiffness provided an opportunity to test

the hypothesis that muscular and therefore limb stiffness is regulated by the integra-
tion of force and length feedback in the spinal cord. The hypothesis was originally
proposed for the regulation of stiffness in individual muscles (Houk 1972). Excita-
tory length feedback from muscle spindle receptors converging with inhibitory force
feedback (Granit 1950) from Golgi tendon organs at the level of the motoneuron the-
oretically would result in the regulation of stiffness. Increased gain of autogenic (self)
length feedback would lead to increased stiffness, while increased autogenic force
feedback would result in the decrease in stiffness. Both pathways would compen-
sate for muscular fatigue and other nonlinear or nonstationary properties of muscle.
Evidence was subsequently obtained that stiffness is indeed regulated, since certain
nonlinear properties are compensated in the presence of autogenic feedback (Nichols
and Houk 1976; Huyghues-Despointes et al. 2003). Later research, however, showed
that the observed stiffness regulation was most likely mediated by length feedback
alone (Houk et al. 1981) for two reasons. First, little evidence was obtained for sub-
stantial autogenic force feedback under the same conditions. Second, the response
properties of muscle spindle receptors are appropriate for compensation. For exam-
ple, muscle spindles are more sensitive to increases in length than decreases, and
muscular stiffness is greater for shortening than lengthening. The two asymmetries
are complementary, so the result is that the stiffness exhibited by the muscle-reflex
system is more symmetrical for shortening and lengthening.
This evidence did not support the original hypothesis that length and force feed-
back around an individual muscle regulates the stiffness of that muscle, and as well
left the functional role of inhibitory force feedback poorly understood. A new version
of this hypothesis became apparent when we reviewed the research of Eccles and
coworkers (Eccles et al. 1957). In that study, the distribution of di- and trisynap-
tic pathways from group Ib afferents, presumably from Golgi tendon organs, was
mapped in the feline spinal cord. These studies revealed extensive intermuscular
linkages between the antigravity muscles of the hind limb. In a series of papers we
confirmed and extended these findings, and also showed that the pathways in ques-
tion were force dependent (Bonasera and Nichols 1994, 1996; Wilmink and Nichols
2003). Furthermore, we showed that these inhibitory pathways were active during
locomotion and polarized such that force feedback from more proximal muscles was
focused on the distal musculature (Nichols and Ross 2009; Ross and Nichols 2009).
Therefore, rather than a modular model in which each muscle is separately regulated,
the evidence suggested that limb stiffness is regulated by a proprioceptive network
in the spinal cord, and that the site of the regulation is the musculature of the distal
limb, muscles which are directly responsible for the mechanical interaction of the
body with the environment.
Data has also accumulated supporting the hypothesis that an excitatory pathway
from Golgi tendon organs is active during locomotion (Guertin et al. 1995; Angel et al.
a FHL Alone b c
18 FHL + G
16
FHL Response (N)
14
12
10
8
6
5N
4
2 50ms
0
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7
Baseline FHL Force (N) Baseline FHL Force (N) Baseline FHL Force (N)
Fig. 4.4 Evidence that the change in limb stiffness after head tilt in a labyrinthectomized, de-
cerebrate animal is due at least in part to modulation of inhibitory force feedback. The central
panel represents the control condition with the head level. Closed circles represent force responses
(obtained at the end of a ramp stretch (see inset, solid lines; vertical gray bar denotes time of
measurement)) of the flexor hallucis longus muscle (FHL) obtained at different background forces
(group 1). Force was modulated using a crossed-extension reflex, a response that naturally habitu-
ates with time, providing a range of background forces, and was obtained by electrical stimulation
of the tibial nerve in the contralateral limb. Open diamonds represent the responses of FHL when
the gastrocnemius (G) muscles were stretched simultaneously (group 2, dashed responses in in-
sets). The data points corresponding to the force trajectories shown in the insets are denoted by
open circles and closed diamonds, respectively, for groups 1 and 2. The lines signify quadratic least
squares fits to the data points. The extent of inhibition corresponds to the difference between the two
fitted lines. The left panel depicts the condition in which the head was tilted down, corresponding
to walking up a slope. Note that the magnitude of inhibition was reduced across the range of forces.
The right panel depicts the condition for downslope walking. Note that inhibition was greater than
in the other two conditions. In this experiment, the animal was not stepping, indicating that the
modulation of force feedback can occur under steady postural conditions as well
1996; Duysens et al. 2000; Donelan and Pearson 2004). We observed what appeared
to be excitatory force feedback during locomotion in premammillary preparations
(Ross et al. 2005), but this feedback was limited to autogenic feedback around the
gastrocnemius (G) muscles and coexisted with intermuscular inhibitory feedback.
Since the G muscles are biarticular, linking the ankle and knee joints, this excitatory
feedback would have the effect of coupling the two joints and influencing ankle
stiffness as well. Therefore, the net stiffness of the limb, and the distribution of
stiffnesses at the different joints of the limb would be strongly influenced by the
three sources of feedback, excitatory length feedback from muscle spindles, and
inhibitory and excitatory force feedback from Golgi tendon organs.
Since we had shown that limb stiffness varies with the motor task through the
body orientation signal, we sought to identify the underlying mechanisms. We un-
dertook a new project evaluating the magnitude of excitatory length and inhibitory
force feedback when the body orientation signal was varied. In order to do this, we
measured the magnitudes of the stretch reflex and of force-dependent inhibition in
decerebrate (intercollicular), labyrinthectomized animals (Tuthill and Nichols 2008).
The main result can be illustrated using the G muscles and the long toe flexor, the
flexor hallucis longus muscle (FHL). The tendons of these muscles were dissected
free from their insertions and connected to servo-controlled linear motors. The femur
and tibia of the limb were rigidly fixed to a baseplate. Stretches were applied to FHL
over a range of forces, and the effect of stretching G on those responses was tested
at matched forces. In order to control for fatigue and other time varying factors,
we alternated stretching FHL alone with stretches of both muscles simultaneously
as force declined during a crossed-extension reflex. Previous research in which this
protocol was used had shown strong inhibition from G to FHL (Bonasera and Nichols
1994). The central panel (b) of Fig. 4.4 illustrates the previous result, namely, that
stretch of G reduces the response of FHL over a wide range of forces, and this inhi-
bition increases with background force. In the condition shown in panel (a), the head
was pitched downward corresponding to upslope walking, and the inhibition was
found reduced. The most dramatic effect was observed when the head was pitched
up, corresponding to downslope locomotion (panel c), where inhibition substantially
increased. The magnitude of the stretch reflex was relatively little affected by the
pitch angle of the head, suggesting that autogenic length feedback was not greatly
affected by changes in the body orientation signal. Therefore, the evidence supports
the hypothesis that pathways mediating inhibitory force feedback constitute a major
site of modulation in the regulation of limb stiffness. Future work will be directed
toward generalizing this result to other intermuscular pathways.
4.5 Conclusions
Taken as a whole, these studies shed light on the manner in which pattern-generating
networks and proprioceptive pathways are modulated during slope walking in
quadrupeds to meet the mechanical demands of these tasks, including propulsion
and braking (Ahn and Full 2002). Upslope walking is characterized by the need for
propulsion and concentric muscular work. Hamstrings and antigravity muscles are
engaged for weight support and propulsion. During downslope walking, hip flex-
ors are recruited to oppose the propulsion of the animal due to gravitational force,
and the hamstrings are not recruited in order to prevent excessive propulsion. Distal
muscles such as the triceps surae group are recruited to provide braking as well as
weight support. Limb stiffness is reduced during downslope walking since a more
compliant interface is required to prevent damaging forces due to the interaction
of the limb with the support surface. Actions due to length and force feedback are
automatically tailored to the task (Maas et al. 2007). During downslope walking,
muscle spindles in the ankle extensors are likely to be particularly responsive and
would regulate muscular stiffness during active lengthening of these muscles. Golgi
tendon organs in the braking muscles are likely to provide regulation of distal limb
stiffness through intermuscular pathways to the distal musculature. The patterns of
activity of the hind limb muscles and their regulation through proprioceptive feed-
back are appropriately regulated according to their specific functions in these motor
tasks. These principles of regulation were investigated most directly in quadrupedal
animals, but, as discussed above, similar changes in the activation patterns of lower
limb muscles as well as trunk orientation occur in humans during or in anticipation

of ramp walking. Therapeutic approaches aimed at preserving these mechanisms in
elderly individuals would seem to be important in the prevention of instability and
falls in real-world locomotion.
Acknowledgments This work was supported by grants NS20855 and HD32571.
References
Abelew TA, Miller MD, Cope TC, Nichols TR. 2000. Local loss of proprioception results in
disruption of interjoint coordination during locomotion in the cat. J Neurophysiol 84:2709–2714.
Ahn AN, Full RJ. 2002. A motor and a brake: two leg extensor muscles acting at the same joint
manage energy differently in a running insect. J Exp Biol 205:379–389.
Angel MJ, Guertin P, Jimenez I, McCrea DA. 1996. Group I extensor afferents evoke disynaptic
EPSPs in cat hindlimb extensor motorneurones during fictive locomotion. J Physiol 494(Pt
3):851–861.
Ashley MJ, Gryfe CI, Amies A. 1977. A longitudinal study of falls in an elderly population II. Some
circumstances of falling. Age Ageing 6:211–220.
Bonasera SJ, Nichols TR. 1994. Mechanical actions of heterogenic reflexes linking long toe flexors
with ankle and knee extensors of the cat hindlimb. J Neurophysiol 71:1096–1110.
Bonasera SJ, Nichols TR. 1996. Mechanical actions of heterogenic reflexes among ankle stabilizers
and their interactions with plantarflexors of the cat hindlimb. J Neurophysiol 75:2050–2070.
Boyle R, Pompeiano O. 1981. Responses of vestibulospinal neurons to neck and macular vestibular
inputs in the presence or absence of the paleocerebellum. Ann N Y Acad Sci 374:373–394.
Brink EE, Suzuki I, Timerick SJ, Wilson VJ. 1985. Tonic neck reflex of the decerebrate cat: a role
for propriospinal neurons. J Neurophysiol 54:978–987.
Carlson-Kuhta P, Trank TV, Smith JL. 1998. Forms of forward quadrupedal locomotion. II. A
comparison of posture, hindlimb kinematics, and motor patterns for upslope and level walking.
Cham R, Redfern M. 2002. Changes in gait when anticipating slippery floors. Gait Posture 15:
159–171.
Donelan JM, Pearson KG. 2004. Contribution of sensory feedback to ongoing ankle extensor activity
during the stance phase of walking. Can J Physiol Pharmacol 82:589–598.
Donelan JM, McVea DA, Pearson KG. 2009. Force regulation of ankle extensor muscle activity in
freely walking cats. J Neurophysiol 101:360–371.
Duysens J, Pearson KG. 1980. Inhibition of flexor burst generation by loading ankle extensor
muscles in walking cats. Brain Res 187:321–332.
Duysens J, Clarac F, Cruse H. 2000. Load-regulating mechanisms in gait and posture: comparative
aspects. Physiol Rev 80:83–133.
Eccles JC, Eccles RM, Lundberg A. 1957. Synaptic actions on motoneurones caused by impulses
in Golgi tendon organ afferents. J Physiol 138:227–252.
Fowler EG, Gregor RJ, Hodgson JA, Roy RR. 1993. Relationship between ankle muscle and joint
kinetics during the stance phase of locomotion in the cat. J Biomech 26:465–483.
Goslow GE, Jr., Reinking RM, Stuart DG. 1973. The cat step cycle: hind limb joint angles and
muscle lengths during unrestrained locomotion. J Morphol 141:1–41.
Gottschall JS, Nichols TR. 2007. Head pitch affects muscle activity in the decerebrate cat hindlimb
during walking. Exp Brain Res 182:131–135.
Gottschall JS, Nichols TR. 2011. Neuromuscular strategies for the transitions between level and
hill surfaces during walking. Philos Trans R Soc Lond B Biol Sci 366:1565–1579.
Granit R. 1950. Reflex self-regulation of muscle contraction and autogenetic inhibition.

Gregor RJ, Smith DW, Prilutsky BI. 2006. Mechanics of slope walking in the cat: quantification of
muscle load, length change, and ankle extensor EMG patterns. J Neurophysiol 95:1397–1409.
Grillner S, Shik ML. 1973. On the descending control of the lumbosacral spinal cord from the
“mesencephalic locomotor region”. Acta Physiol Scand 87:320–333.
Guertin P, Angel MJ, Perreault MC, McCrea DA. 1995. Ankle extensor group I afferents excite
extensors throughout the hindlimb during fictive locomotion in the cat. J Physiol 487(Pt 1):
197–209.
Houk J. 1972. The phylogeny of muscular control configurations. In: Drischel P, Dettmar P, editors.
Biocybernetics. Jena: Fisher. pp 125–144.
Houk J, Crago P, Rymer W. 1981. Function of the spindle dynamic response in stiffness regulation
- a predictive mechanism provided by nonlinear feedback. In: Taylor A, Prochazka A, editors.
Muscle Receptors and Movement. London: Macmillan. pp 299–309.
Huyghues-Despointes CM, Cope TC, Nichols TR. 2003. Intrinsic properties and reflex compensa-
tion in reinnervated triceps surae muscles of the cat: effect of movement history. J Neurophysiol
90:1547–1555.
Lafreniere-Roula M, McCrea DA. 2005. Deletions of rhythmic motoneuron activity during fictive
locomotion and scratch provide clues to the organization of the mammalian central pattern
generator. J Neurophysiol 94:1120–1132.
Lay A, Hass C, Gregor R. 2006. The effects of sloped surfaces on locomotion: a kinematic and
kinetic analysis. J Biomech 39:1621–1628.
Lay AN, Hass CJ, Gregor RJ. 2007. The effects of sloped surfaces on locomotion: backward walking
as a perturbation. J Biomech 40:3050–3055.
Lindsay KW, Roberts TD, Rosenberg JR. 1976. Asymmetric tonic labyrinth reflexes and their
interaction with neck reflexes in the decerebrate cat. J Physiol 261:583–601.
Maas H, Prilutsky BI, Nichols TR, Gregor RJ. 2007. The effects of self-reinnervation of cat medial
and lateral gastrocnemius muscles on hindlimb kinematics in slope walking. Exp Brain Res
181:377–393.
Maas H, Gregor RJ, Hodson-Tole EF, Farrell BJ, English AW, Prilutsky BI. 2010. Locomotor
changes in length and EMG activity of feline medial gastrocnemius muscle following paralysis
of two synergists. Exp Brain Res 203:681–692.
Magnus R, de Kleijn A. 1912. Die Abhangigkeit des Tonus der Extremitatenmuskeln von der
Kopfstellung. Pflugers Arch 145:455–548.
Manzoni D, Andre P, Pompeiano O. 2004. Proprioceptive neck influences modify the information
about tilt direction coded by the cerebellar anterior vermis. Acta Otolaryngol 124:475–480.
Marchand AR, Manzoni D, Pompeiano O, Stampacchia G. 1987. Effects of stimulation of vestibular
and neck receptors on Deiters neurons projecting to the lumbosacral cord. Pflugers Arch 409:
13–23.
Nichols TR, Houk JC. 1976. Improvement in linearity and regulation of stiffness that results from
actions of stretch reflex. J Neurophysiol 39:119–142.
Nichols R, Ross KT. 2009. The implications of force feedback for the lambda model. Adv Exp Med
Biol 629:663–679.
Nichols TR, Cope TC, Abelew TA. 1999. Rapid spinal mechanisms of motor coordination. Exerc
Sport Sci Rev 27:255–284.
Polcyn AF, Lipsitz LA, Kerrigan DC, Collins JJ. 1998. Age-related changes in the initiation of
gait: degradation of central mechanisms for momentum generation. Arch Phys Med Rehabil
79:1582–1589.
Prentice SD, Hasler EN, Groves JJ, Frank JS. 2004. Locomotor adaptations for changes in the slope
of the walking surface. Gait Posture 20:255–265.
Prilutsky BI, Maas H, Bulgakova M, Hodson-Tole EF, Gregor RJ. 2011. Short-term motor compen-
sations to denervation of feline soleus and lateral gastrocnemius result in preservation of ankle
mechanical output during locomotion. Cells Tissues Organs 193:310–324.
Ross KT, Nichols TR. 2009. Heterogenic feedback between hindlimb extensors in the spontaneously
locomoting premammillary cat. J Neurophysiol 101:184–197.
Ross K, Duysens J, Smith V, Nichols T. 2005. Modulation of cutaneous and proprioceptive feedback
in the premammillary locomoting cat. Soc Neurosci Abstr 30:630–637.
Smith JL, Carlson-Kuhta P, Trank TV. 1998. Forms of forward quadrupedal locomotion. III. A com-
parison of posture, hindlimb kinematics, and motor patterns for downslope and level walking.
Srivastava UC, Manzoni D, Pompeiano O, Stampacchia G. 1984. Responses of medullary
reticulospinal neurons to sinusoidal rotation of neck in the decerebrate cat. Neuroscience
11:473–486.
Suzuki I, Timerick SJ, Wilson VJ. 1985. Body position with respect to the head or body position in
space is coded by lumbar interneurons. J Neurophysiol 54:123–133.
Tinetti ME, Doucette J, Claus E, Marottoli R. 1995. Risk factors for serious injury during falls by
older persons in the community. J Am Geriatr Soc 43:1214–1221.
Tuthill C, Nichols T. 2008. Context-dependent organization of force feedback in the decerebrate
cat. Soc Neurosci Abstr 34:377–373.
Tuthill C, Nichols TR. 2012. Task dependent changes to hindlimb impedance in the decerebrate
cat. Soc Neurosci Abstr 38:886–811.
von Holst E, Mittelstaedt H. 1950. Das reafferenzprincip. Naturwiss 47:464–476.
Wilmink RJ, Nichols TR. 2003. Distribution of heterogenic reflexes among the quadriceps and
triceps surae muscles of the cat hind limb. J Neurophysiol 90:2310–2324.
Chapter 5
Subcortical Visuomotor Control of Human
Limb Movement
Brian L. Day
Research into the visual control of movement over the past 30–40 years has focused
predominantly on cortical networks. These cortical networks are fed by information
travelling along pathways from retina to occipital cortex and then to motor cortical
areas via parietal cortex (dorsal “where” stream) or temporal cortex (ventral “what”
stream) (Ungerleider and Mishkin 1982). Although these cortical networks are very
powerful, they are not the only routes by which visual information can influence
action. This chapter reviews some behavioral, neurophysiological, and neuroimaging
data which converge on the idea that subcortical structures have a significant role to
play in the visual control of reaching, even in primates.
5.1 Fast Arm Reactions with Limited Modifiability by Intention
It was thought for a long time that visual information influences movement through a
process that is relatively sluggish compared to other sensory channels. For example,
proprioceptive signals can drive muscle responses in the human hand at latencies
of ∼25 ms through spinal circuitry, but even when the motor cortex is involved
proprioceptive input can evoke hand-muscle responses at a remarkably short latency
of 50 ms (Marsden et al. 1976). In contrast, the latency of visually driven reactions
was traditionally put at 200–300 ms (Keele and Posner 1968; Beggs and Howarth
1970). From the 1980s onwards, however, experiments investigating the speed at
which new visual information can modify ongoing limb movements revised this
value down dramatically to 100–150 ms (Carlton 1981; Soechting and Lacquaniti
1983; Zelaznik et al. 1983; Paulignan et al. 1990, 1991; Prablanc and Martin 1992).
The explanation for this halving of visuomotor latency was not immediately ap-
parent. It could have been due to methodological differences or possibly due to
B. L. Day ()
Institute of Neurology, University College London, London, UK

and Biology 826, DOI 10.1007/978-1-4939-1338-1_5
56 B. L. Day
technological advances offering more sensitive means of recording limb trajecto-

ries. Another and more physiological possibility is that these discrepant latencies
actually reflect the operations of distinct visuomotor processes. This was indeed
the conclusion that was reached based on the results of an experiment that investi-
gated the ability of human subjects to modify voluntarily their visually driven reach
adjustments (Day and Lyon 2000).
In this experiment, subjects were seated in a darkened room and were required
to reach out and touch an illuminated disk with the index finger. In random and
sparse trials (one in three), at the instant the finger was lifted, the illuminated disk
slid 10 cm either to the right or the left with equal probability. In one block of trials,
subjects were instructed to keep trying and touch the disk even if it moved. In a
separate block, they were instructed to touch the disk if it remained stationary, but
if they saw it move, then to move their finger in the opposite direction. In effect,
this was an anti-reach task not dissimilar to that of an oculomotor anti-saccade task
(Hallett 1978; Hallett and Adams 1980). The idea here was to see whether the fast
visual-evoked reach adjustments are truly voluntary. If so, subjects should have no
difficulty in replacing the fast response in the direction of target movement with an
equally fast response in the opposite direction during the anti-reach task. The results
of the experiment showed clearly that subjects were unable to do this.
When the target moved during the pro-reach task, subjects adjusted the trajectory
of the reaching finger at latencies of between 120–160 ms, ending up accurately on
the shifted target. In the anti-reach task, more often than not the finger’s trajectory
was also adjusted at short latency, but as in the pro-reach task this acted to take the
finger in the direction of the target rather than in the opposite, instructed direction
(Fig. 5.1). Despite this inappropriate early adjustment, the finger underwent a subse-
quent reversal at around 220 ms, this time to drive the finger in the opposite direction
to comply with the instructions.
The two sequential responses observed in the anti-reach task are highly suggestive
of two processes being active. One appears to be a fast-acting process that spatially
links the arm movement with the target movement in a compatible and relatively
inflexible way. The other process takes longer to act but is flexible allowing an
arbitrary spatial association between the visual stimulus and the action. It is tempting
to speculate that this slower, flexible process involves cortical processing (Wise et al.
1996) whereas the fast inflexible one is subcortically organized.
5.2 Does the Leg Behave like the Arm?
Most people spend a good proportion of their waking hours reaching for objects with
their hand. Perhaps with the exception of professional soccer players, the foot is not
routinely used in this way. However, ordinary walking often requires accurate foot
placement that relies upon on-line visual control (Reynolds and Day 2005b). It is
reasonable to enquire, therefore, whether the lower limb has access to the same fast-
acting visuomotor process that controls the upper limb. One might expect this to be the
5 Subcortical Visuomotor Control of Human Limb Movement 57
Fig. 5.1 Spatial and temporal properties of arm reach adjustments evoked by a moving target. a 3-D
superimposed trajectories of the finger from multiple trials of one representative subject when the
target remained stationary (green), moved right (red), or moved left (blue). The subject executed
either a normal reach (left panel) or an anti-reach (right panel) task (see text). Note the initial
deviation of the finger towards the target in both tasks. b Latency histograms of the initial deviation
of the finger in all trials in which the target moved. Responses grouped in 10 ms bins and shaded
according to whether in the correct (black bars) or incorrect direction (white bars). Note for the
anti-reach task most responses were initially in the incorrect direction, with the same latency as
correct response in the normal reach task. The small number of initially correct responses in the
anti-reach task tended to occur with longer and more variable latency. (Modified from Day and
Lyon 2000)
case if upper-limb reaching movements have evolved from stepping movements, as

has been suggested (Georgopoulos and Grillner 1989). On the other hand, mid-flight
adjustments of a stepping foot carry potentially serious consequences that upper limb
adjustments do not. This arises from the need to maintain balance when stepping. It
has been shown that the trajectory of the body during a step and the position of the
stepping foot at the end of the step are closely coordinated and determined before
the foot leaves the ground (Lyon and Day 1997, 2005). Any mid-flight alteration of
stepping-foot trajectory will violate this relationship and potentially leave the body
in an unstable state should the foot return to the ground at an unplanned location.
To examine this issue, experiments analogous to the arm-reaching experiments
were performed using a jumping floor-mounted target which subjects stepped onto
58 B. L. Day
Fig. 5.2 Spatial and temporal

properties of foot adjustments
when stepping onto a
floor-mounted jumping target.
a Group mean final foot
position at the end of the step
when the target remained
stationary (black), jumped
laterally (red), or jumped
medially (blue). Note the
smaller deviation of the foot
for medial target jumps.
b Group mean mediolateral
acceleration of the foot with
time (lateral positive), colour
coding as in a. Note the short
latency of foot adjustment
irrespective of target-jump
direction. (Modified from
Reynolds and Day 2005)
(Reynolds and Day 2005a). In one condition, subjects stepped in a natural fashion
onto the rectangular illuminated target, which in some trials could jump laterally
or medially at the instant the foot left the ground. In a second condition, the same
stimuli were employed with the subject upright, but now fully supported by an
external framework. In this latter condition, therefore, the usual balance constraint
was not present and the leg movement task more closely resembled an arm-reaching
task. The key result of this study was that the jumping target evoked short-latency
(< 150 ms) mid-flight adjustments of the leg indistinguishable in latency from those
of the arm (Fig. 5.2b). The amplitude of the leg adjustment, particularly for medial
target jumps (Fig. 5.2a), was less during free steps compared to frame-supported
steps, reflecting the limitations imposed by the need to maintain balance. However,
the latency was identical irrespective of whether or not the body was supported
by the frame. Thus, it seems that the leg, like the arm, has access to a fast-acting
visuomotor process, but with additional integration with balance processes. Given
the correspondence in latency, a parsimonious explanation is that the upper and lower
limbs share a common fast-acting visuomotor mechanism.
5.3 Testing the Subcortical Hypothesis
The opportunity arose to test the hypothesis that the fast (< 150 ms) visual-evoked
reach adjustment stems from a subcortical visuomotor process (Day and Brown
2001). This was provided by a patient who was discovered to have agenesis of the
corpus callosum and an absent or vestigial anterior commissure, and therefore no

direct connection between the left and right cerebral cortices. She was unaware of her
condition and was otherwise healthy (apart from a headache that initiated the brain
magnetic resonance imaging (MRI) scan to reveal the abnormality) with normal
functioning corticospinal tracts. The idea was to measure the crossed–uncrossed
latency difference in this patient, a method that was conceptualized 100 years ago
(Poffenberger 1912) and which had been used previously in the study of split-brain
patients (Jeeves 1969; Kinsbourne and Fisher 1971; Milner 1982; Milner et al. 1985;
Clark and Zaidel 1989). The method relies upon cortical lateralization whereby visual
information in each hemifield is transmitted to contralateral occipital cortex while
each upper limb is controlled by contralateral motor cortex. If the visual hemifield
and responding limb are on the same side (e.g., right arm reaching for right-sided
target while looking straight ahead), then all cortical processing can be done within
one cerebral hemisphere resulting in an “uncrossed” response. However, if the arm
and visual hemifield are on opposite sides (e.g., right arm reaching for left-sided
target), then information has to pass from one hemisphere to the other resulting in
a “crossed” response. In the absence of a corpus callosum, which is the most direct
route for communication between the left and right cortices, a more circuitous route
is required resulting in a crossed response with a longer latency than normal.
To establish the soundness of this theory in the acallosal subject, a task was
employed that required an arbitrary movement in response to a visual stimulus.
It was thought that the arbitrary association between visual stimulus and action
would engage a cortical network (Wise et al. 1996) and reveal a significant crossed–
uncrossed latency difference. The tasks were to lift vertically either the index finger
(distal movement) or the forearm (proximal movement) in response to illumination
of a disk randomly presented laterally in one or the other hemifields while fixating
a central light-emitting diode (LED). As predicted, the crossed reaction time in the
acallosal subject was 36 ms slower on average than the uncrossed reaction time
(Fig. 5.3a), a difference that was not present in healthy subjects.
Having established a clear crossed–uncrossed latency difference in the acallosal
subject for arbitrary visuomotor associations, the next step was to see whether the
same difference was present for the nonarbitrary task of reaching for a jumping target.
Importantly, the target and its locations for this task were identical to that used in the
arbitrary association task. The critical observations were that for this reaching task,
the reaction times were considerably faster than for arbitrary visuomotor associations
(120 vs. 320 ms) and were the same irrespective of which hand was used and which
hemifield the target jumped into (Fig. 5.3b). In other words, for the acallosal subject,
there was no apparent crossed–uncrossed latency difference when executing a mid-
flight reach adjustment to acquire a displaced target. This result could not be explained
by very early eye movements bringing the displaced target back into central vision
since for this subject, lateral saccades always occurred after the arm adjustment was
initiated (Fig. 5.3b). The compelling conclusion was therefore that the visual-evoked
reach adjustment did not use the cortical networks employed for arbitrary visuomotor
associations. This suggested strongly that subcortical centers were being employed
to drive the hand towards visually presented targets.
60 B. L. Day
Fig. 5.3 Manual responses to visual stimuli appearing in the right or left hemifield in a subject
with agenesis of the corpus callosum. a Reaction time to initiate arbitrary motor tasks (finger
extension, elbow flexion) in response to target appearance in right (red) or left (blue) hemifield
during central fixation. Note the significant interaction between arm side and target side for both
arbitrary tasks. b Mediolateral motion of the finger and eyes when reaching to a central target which
unpredictably jumped into right (red) or left (blue) hemifield during the reach. Top traces show mean
mediolateral velocity of finger in jump trials after subtraction of mean no-jump trials. Bottom traces
show mediolateral motion of the eyes in jump trials recorded electro-oculographically (high-pass
filtered). Note the same short latency of finger deviation irrespective of limb side or target side, and
that eyes deviated towards target after limb movement was initiated. P-values above targets (left
panels) denote probability of target appearance at respective locations. (Modified from Day and
Brown 2001)
5.4 Evidence for Involvement of the Superior Colliculus

in Visually Guided Reaching
The superior colliculus and underlying mesencephalic reticular formation are good
candidates for a subcortical visuomotor processing centre controlling the limbs. The
superior colliculus receives visual input directly from the retina (Kaas and Huerta
1988), and, although classically involved in gaze control through control of eye
(Sparks 1978) and head (Cowie and Robinson 1994), was discovered also to contain
arm-movement-related neurons (Werner et al. 1997a). These neurons occur through-

out the depth of the monkey’s superior colliculus, apart from the most superficial
layers, as well as in the underlying mesencephalic reticular formation. The “reach”
neurons have a different spatial distribution to visually responsive cells or saccadic
units in that they occur most frequently in the deeper layers (Werner et al. 1997b).
For retinal information to drive arm movements, visual signals need to undergo trans-
formation from an oculocentric coordinate frame to an arm-centered frame. There
is some evidence of this sort of transformation in the superior colliculus as shown
by two kinds of reach neurons (Stuphorn et al. 2000). One type, located in inter-
mediate layers of superior colliculus, has links to an oculocentric frame in that its
firing is gaze-related. The other type, located in the deep layers of superior col-
liculus and underlying mesencephalic reticular formation, is gaze-independent and
therefore associated with an arm-centered frame. Their firing patterns correlate well
with electromyography (EMG) patterns of arm muscles during reaching movements,
particularly for the proximal muscles of the shoulder girdle (Stuphorn et al. 1999).
Evidence in man for the involvement of these structures in reaching control comes
from a functional MRI (fMRI) study (Linzenbold and Himmelbach 2012). In this
study, human subjects simply reached with either arm to visual targets presented in
the left or right hemifields, whereupon increases in the BOLD signal were observed in
the region of the superior colliculus. Unsurprisingly, some of this signal change could
be attributed to saccadic activity, but there was another dissociable activity present.
This activity, in contrast to saccade-related activity, tended to be lateralized to the
reaching arm and occurred in deeper layer locations. Thus, the spatial distribution of
these human signals corresponds to the distribution of reach neurons in the monkey
superior colliculus and plausibly reflects the operation of an analogous process.
If activity within the superior colliculus is rapidly able to influence reaching
movements, there must be a fast-acting route that transmits the motor signal from
colliculus to spinal motoneurons. Courjon and colleagues showed that such a motor
pathway is functional in the cat during a visually guided reach (Courjon et al. 2004).
They stimulated in the deep layers of the superior colliculus using a 300 Hz train
of stimuli for 70–200 ms during a reach for a piece of food. This stimulus consis-
tently caused the cat’s reaching limb to deviate from its unstimulated trajectory with
an average latency of just 56 ms. In about half the stimulated trials, this deviation
occurred without a concomitant gaze or head perturbation thereby ruling out the
possibility of the limb deviation being indirectly evoked by a gaze shift. The output
pathway could be via direct tectospinal fibers, which project to C3–C4 propriospinal
neurons (Illert et al. 1978) thereby engaging spinal reach machinery. They are dense
in the cat (Nudo and Masterton 1989; Olivier et al. 1991), but much less so in other
mammals including primates (Nudo and Masterton 1989). An alternative and not
mutually exclusive possibility is that motor signals travel via the tectoreticulospinal
tract (Werner et al. 1997b). Alstermark and colleagues provided some support for the
involvement of this pathway by demonstrating a prolongation of the cat’s reach ad-
justment latency to a jumping target after a presumed lesion of the tectoreticulospinal
tract (Alstermark et al. 1987).
62 B. L. Day
Fig. 5.4 Effect of loud sound

on latency of foot adjustments
to a jumping target during a
step. a Group mean
mediolateral acceleration of
foot during trials with no
target jump (black), lateral
target jump (red) or medial
target jump (blue), and with
(dashed lines) or without
(continuous lines) a loud
sound at moment of foot lift
from floor. b Same as a but
with foot acceleration during
no-jump conditions
subtracted from all traces.
Note the hastening of
response to target jump when
combined with a loud sound.
(Modified from Reynolds and
Day 2007)
5.5 Indirect Evidence for Reticulospinal Involvement in Man
Reticulospinal tracts are thought to transmit the auditory startle response from the
caudal reticular formation of the human brainstem to spinal cord both in healthy
subjects (Brown et al. 1991a) and in patients with pathological exaggerated startle
or hyperekplexia (Brown et al. 1991b). This raises the possibility that the putative
subcortical mechanism controlling visually guided reaching may share some circuitry
with the auditory startle mechanism. This was investigated during stepping to visual
targets by interacting target jumps and loud sounds (Reynolds and Day 2007).
In this experiment, subjects stepped naturally onto an illuminated rectangle, which
in a third of trials jumped to the left or right with equal probability at the moment
the foot lifted from the ground. In just 18 of the total 240 trials, a startling loud
sound (120 dB SPL) was given through headphones at the point of foot lift, but
with the startle trials occurring with equal probability during no-jump, jump-right,
and jump-left trials. Therefore the loud sound conveyed no information about the
presence of a target jump or its direction. As in a previous study (Reynolds and Day
2005), the jumping target evoked a fast adjustment of the foot trajectory, at 134
ms on average when measured from acceleration traces, with medial displacement
magnitudes being less than lateral displacements. The loud sound had no effect on
control steps or on the magnitude of foot displacement following a target jump.
However, the loud sound did affect the latency of the foot correction, shortening it
by some 20–30 ms (Fig. 5.4).
This hastening of the visual-evoked foot trajectory adjustment was not a nonspe-
cific effect of a startle because the foot always moved in the direction of the target
jump when present and did not deviate during no-jump trials. Nor was it due to
release of a subcortically stored motor program (Valls-Sole et al. 1999; Carlsen et al.
2004) since the required adjustment was not known in advance of the sound stim-
ulus. Instead, it seemed that the sound stimulus interacted with the visual stimulus
provided by the target jump to cause a reduction in visuomotor processing time. By
taking account of likely afferent and efferent conduction delays, Reynolds and Day
(2007) estimated that even a modest hastening of 18 ms would represent a substan-
tial reduction of 30 % in central processing time. It is plausible that this interaction
occurs in the brainstem, possibly in the superior colliculus, but perhaps more likely
in the reticular formation.
5.6 Indirect Evidence for Superior Colliculus

Involvement in Man
When there is choice in selecting one action over another, and if the selected action
is determined by an external signal such as a visual stimulus, it is often found that the
time taken to react increases with the number of choices available. This relationship
between number of response choices and reaction time was formalized by Hick using
information processing theory (Hick 1952) and is often referred to as Hick’s law.
Although Hick’s law holds for many manual tasks, it does not hold for visually
guided saccades. Kveraga et al. (2002) showed that the latency of a saccade to a
visual target is the same irrespective of whether there are eight possible targets or
just one. They explained this violation of Hick’s law by referring to the anatomical
arrangement of the superior colliculus, where visual afferents form a retinotopic map
that lies in spatial register with an oculomotor map determining saccade direction
and amplitude. With this arrangement, a visual stimulus on the retina automatically
evokes a saccade that brings the visual target onto the fovea, making the central
processing time immune to the number of possible target locations. When this highly
compatible spatial relationship between stimulus and response was destroyed by
asking subjects to make eye movements in the opposite direction to a peripheral
visual stimulus (anti-saccade), Hick’s law was reinstated (Kveraga et al. 2002).
It could be argued that fast visually guided limb movements also should violate
Hick’s law if they utilize the neural machinery of the superior colliculus. This was
tested using a centre-out pointing task (Reynolds and Day 2012). For this, subjects
held a finger stationary in front of a central target on a vertical board and were
required to make a rapid discrete movement in response to the appearance of a
peripheral target (Fig. 5.5a). Four visual targets were located 15 cm above, below,
left and right of the central target. To eliminate temporal uncertainty about when
an event might happen, the subject initiated a trial by raising the opposite index
finger from a touch switch. In 67 % of trials this did nothing (null trials), but in the
remaining 33 % it caused the central target light to be extinguished and one of the
four peripheral targets to be illuminated. Subjects were given advance information
which of the peripheral targets might be illuminated in a trial (between one and four
targets) thereby manipulating the number of action choices. In one condition, the
stimulus–response compatibility was high in that subjects were required to move
64 B. L. Day
Fig. 5.5 Effect of

stimulus–response
compatibility and number of
choices on reaction time to
visual stimulus. a
Experimental set-up (see text
for details). b Reaction time
against number of choices for
high S–R compatibility
(normal pointing) and low
S–R compatibility
(orthogonal pointing)
conditions. Reaction times for
erroneous responses during
orthogonal pointing are
shown when the finger moved
in the direction of the target
(T-error) or in some other
incorrect direction (O-error).
Note the constancy and
brevity of normal pointing
reaction time regardless of
number of choices. (Modified
from Reynolds and Day 2012)
their finger in the direction of the illuminated target (normal). In a second condition,
the stimulus–response(S–R) compatibility was low as they were required to move
their finger in a direction 90◦ clockwise to the illuminated target (orthogonal).
As predicted, for normal pointing with high S–R compatibility, the reaction time
was fast at around 140 ms and remained the same regardless of the number of
choices. In contrast, for the orthogonal pointing task the reaction time was slower,
being 173 ms for the 1-choice condition (i.e., simple reaction time), and it increased
with number of choices, thus obeying Hick’s law (Fig. 5.5b). Therefore, in many
ways the pointing behavior was very similar to the saccade behavior described above
(Kveraga et al. 2002). As with the eyes, the violation of Hick’s law when there was
a direct spatial correspondence between the visual stimulus and the limb movement
suggests a mechanism with relatively hard-wired direct mapping between the two.
A parsimonious interpretation is that the superior colliculus lies at the heart of the
mechanism for visually guided pointing as it does for visually guided saccades. The
radically different behavior during orthogonal pointing suggests the possibility of a
slower mechanism being brought into play. Of course, for a slower mechanism to
dominate the faster mechanism, the latter would have to be suppressed in some way.
Interestingly, subjects tended to make errors by sometimes pointing incorrectly at
the illuminated target in the orthogonal condition (T-error in Fig. 5.5b), with these
erroneous movements usually being initiated faster than the correct movements.
This is compatible with two competing mechanisms where there is an incomplete
suppression of the faster process. As suggested in an earlier section to explain anti-
reach behavior (Day and Lyon 2000), orthogonal pointing could engage a cortical
mechanism, which would give the advantage of flexibly associating any action with
a visual signal thereby avoiding the necessity of a direct spatial correspondence
between stimulus and response. The cost is a longer response latency that grows
with choice (Hick’s law) because of the additional information processing required
for stimulus identification and action selection.
5.7 Communication Between Cortical and Subcortical

Visuomotor Processes
The picture painted so far in this chapter is of two processes for visual guidance
of limbs, one subcortical and the other cortical. The subcortical process is fast, but
rigid, and is well suited for direct interactions between the limb and an object when
both occupy a common space. In contrast, the slower cortical process conceivably
is infinitely flexible and may be better suited for visuomotor interactions during tool
use. This flexibility effectively detaches the limb from the object and allows any
arbitrary spatial relationship between the two, for instance, tracking vertical move-
ments of a target on a computer screen with a cursor controlled by forward–backward
movements of a mouse. However, it is unlikely that the two processes operate inde-
pendently of each other. As a minimum, the subcortical process would need to be
suppressible under certain circumstances. As we have seen, when a person is en-
gaged in interacting with an object directly with the limb, suppression is difficult to
achieve and often incomplete. Thus, during an anti-reach task, invariably the limb is
initially drawn towards the target’s new position, although with less vigor compared
to a standard reach (Day and Lyon 2000). Similarly, orthogonal pointing is possible
but at the expense of occasional errors towards the visual stimulus, indicating an in-
trusion of the fast process (Reynolds and Day 2012). Presumably, there are stronger
suppression signals available when the limb operates remotely from the near space of
an object or via an interposed tool. Communication between the two processes would
also be necessary if, as is likely, they act in concert during direct interactions be-
tween a limb and an object. There is a wealth of connections between action-related
areas of cortex and relevant subcortical structures such as the superior colliculus
(e.g., Kuypers and Lawrence 1967; Goldman and Nauta 1976; Catsman-Berrevoets
et al. 1979; Fries 1984, 1985) and the reticular formation (e.g. Catsman-Berrevoets
66 B. L. Day
and Kuypers 1976; Keizer and Kuypers 1984). These fibres could provide the com-
munication required for cooperation between cortical and subcortical visuomotor
processes.
References
Alstermark B, Gorska T, Lundberg A, Pettersson L-G, Walkowska M. Effect of different spinal

cord lesions on visually guided switching of target-reaching in cats. Neuroscience Research
1987;5:63–7.
Beggs WDA, Howarth CI. Movement control in a repetitive motor task. Nature 1970;225:752–3.
Brown P, Rothwell JC, Thompson PD, Britton TC, Day BL, Marsden CD. New observations on the
normal auditory startle reflex in man. Brain 1991a;114:1891–902.
Brown P, Rothwell JC, Thompson PD, Britton TC, Day BL, Marsden CD. The hyperekplexias and
their relationship to the normal startle reflex. Brain 1991b;114:1903–28.
Carlsen AN, Chua R, Inglis JT, Sanderson DJ, Franks IM. Can prepared responses be stored
subcortically? Experimental Brain Research 2004;159:301–9.
Carlton LG. Processing visual feedback information for movement control. Journal of Experimental
Psychology Human Perception and Performance 1981;7(5):1019–30.
Catsman-Berrevoets CE, Kuypers HGJM. Cells of origin of cortical projections to dorsal column
nuclei, spinal cord and bulbar medial reticular formation in the rhesus monkey. Neuroscience
Letters 1976;3:245–52.
Catsman-Berrevoets CE, Kuypers HGJM, Lemon RN. Cells of origin of the frontal projections to
magnocellular and parvocellular red nucleus and superior colliculus in cynomolgus monkey. An
HRP study. Neuroscience Letters 1979;12:41–6.
Clarke JM, Zaidel E. Simple reaction times to lateralized light flashes. Varieties of interhemispheric
communication routes. Brain 1989;112:849–70.
Courjon J-H, Olivier E, Pélisson D. Direct evidence for the contribution of the superior collicu-
lus in the control of visually guided reaching movements in the cat. Journal of Physiology
2004;556(3):675–81.
Cowie RJ, Robinson DL. Subcortical contributions to head movements in macaques I. Contrasting
effects of electrical stimulation of a medial pontomedullary region and the superior colliculus.
Journal of Neurophysiology 1994;72(6):2648–64.
Day BL, Lyon IN. Voluntary modification of automatic arm movements evoked by motion of a
visual target. Experimental Brain Research 2000;130:159–68.
Day BL, Brown P. Evidence for subcortical involvement in the visual control of human reaching.
Brain 2001;124:1832–40.
Fries W. Cortical projections to the superior colliculus in the macaque monkey: a retrograde study
using horseradish peroxidase. The Journal of Comparative Neurology 1984;230:55–76.
Fries W. Inputs from motor and premotor cortex to the superior colliculus of the macaque monkey.
Behavioural Brain Research 1985;18:95–105.
Georgopoulos AP, Grillner S. Visuomotor coordination in reaching and locomotion. Science
1989;245(4953):1209–10.
Goldman PS, Nauta WJH. Autoradiographic demonstration of a projection from prefrontal associ-
ation cortex to the superior colliculus in the rhesus monkey. Brain Research 1976;116:145–9.
Hallett PE. Primary and secondary saccades to goals defined by instructions. Vision Research
1978;18:1279–96.
Hallett PE, Adams BD. The predictability of saccadic latency in a novel voluntary oculomotor task.
Vision Research 1980;20:329–39.
Hick WE. On the rate of gain of information. Quarterly Journal of Experimental Psychology
1952;4(1):11–26.
Illert M, Lundberg A, Padel Y, Tanaka R. Integration in descending motor pathways controlling

the forelimb in the cat. 5. Properties of and monosynaptic excitatory convergence on C3-C4
propriospinal neurones. Experimental Brain Research 1978;33:101–30.
Jeeves MA. A comparison of interhemispheric transmission times in acallosals and normals.
Psychonomic Science 1969;16(5):245–6.
Kaas JH, Huerta MF. The subcortical visual system of primates. In: Steklis HD, Erwin J, editors.
Comparative primate biology, volume 4: neurosciences. New York: Wiley-Liss; 1988. pp.
327–91.
Keele SW, Posner MI. Processing of visual feedback in rapid movements. Journal of Experimental
Psychology 1968;77(1):155–8.
Keizer K, Kuypers HGJM. Distribution of corticospinal neurons with collaterals to lower brain stem
reticular formation in cat. Experimental Brain Research 1984;54:107–20.
Kinsbourne M, Fisher M. Latency of uncrossed and of crossed reaction in callosal agenesis.
Neuropsychologia 1971;9:471–3.
Kuypers HGJM, Lawrence DG. Cortical projections to the red nucleus and the brain stem in the
rhesus monkey. Brain Research 1967;4:151–88.
Kveraga K, Boucher L, Hughes HC. Saccades operate in violation of Hick’s law. Experimental
Brain Research 2002;146:307–14.
Linzenbold W, Himmelbach M. Signals from the deep: Reach-related activity in the human superior
colliculus. The Journal of Neuroscience 2012;32(40):13881–8.
Lyon IN, Day BL. Control of frontal plane body motion in human stepping. Experimental Brain
Research 1997;115:345–56.
Lyon IN, Day BL. Predictive control of body mass trajectory in a two-step sequence. Experimental
Brain Research 2005;161:193–200.
Marsden CD, Merton PA, Morton HB. Stretch reflex and servo action in a variety of human muscles.
Journal of Physiology 1976;259:531–60.
Milner AD. Simple reaction times to lateralized visual stimuli in a case of callosal agenesis.
Neuropsychologia 1982;20(4):411–9.
Milner AD, Jeeves MA, Silver PH, Lines CR, Wilson J. Reaction times to lateralized visual stimuli
in callosal agenesis: Stimulus and response factors. Neuropsychologia 1985;23(3):323–31.
Nudo RJ, Masterton RB. Descending pathways to the spinal cord: II. Quantitative study of the
tectospinal tract in 23 mammals. The Journal of Comparative Neurology 1989;286:96–119.
Olivier E, Chat M, Grantyn A. Rostrocaudal and lateromedial density distributions of superior
colliculus neurons projecting in the predorsal bundle and to the spinal cord: a retrograde HRP
study in the cat. Experimental Brain Research 1991;87:268–82.
Paulignan Y, Mackenzie C, Marteniuk R, Jeannerod M. The coupling of arm and finger movements
during prehension. Experimental Brain Research 1990;79:431–5.
PaulignanY, Mackenzie C, Marteniuk R, Jeannerod M. Selective perturbation of visual input during
prehension movements. Experimental Brain Research 1991;83:502–12.
Poffenberger AT. Reaction time to retinal stimulation with special reference to the time lost in
conduction through nerve centres. Arch Psychol 1912;3:1–73.
Prablanc C, Martin O. Automatic control during reaching at undetected two-dimensional target
displacements. Journal of Neurophysiology 1992;67(2):455–69.
Reynolds RF, Day BL. Rapid visuo-motor processes drive the leg regardless of balance constraints.
Current Biology 2005a;15(2):R48–R49.
Reynolds RF, Day BL. Visual guidance of the human foot during a step. Journal of Physiology
2005b;569(2):677–84.
Reynolds RF, Day BL. Fast visuomotor processing made faster by sound. Journal of Physiology
2007;583(3):1107–15.
Reynolds RF, Day BL. Direct visuomotor mapping for fast visually-evoked arm movements.
Neuropsychologia 2012;50:3169–73.
Soechting JF, Lacquaniti F. Modification of trajectory of a pointing movement in response to a
change in target location. Journal of Neurophysiology 1983;49(2):548–64.
68 B. L. Day
Sparks DL. Functional properties of neurons in the monkey superior colliculus: coupling of neuronal
activity and saccade onset. Brain Research 1978;156:1–16.
Stuphorn V, Hoffmann K-P, Miller LE. Correlation of primate superior colliculus and retic-
ular formation discharge with proximal limb muscle activity. Journal of Neurophysiology
1999;81:1978–82.
Stuphorn V, Bauswein E, Hoffmann K-P. Neurons in the primate superior colliculus coding for arm
movement in gaze-related coordinates. Journal of Neurophysiology 2000;83:1283–99.
Ungerleider LG, Mishkin M. Two cortical visual systems. In: Ingle DJ, Goodale MA, Mansfield
RJW, editors. Analysis of visual behavior. Cambridge (MA): MIT Press; 1982. pp. 549–86.
Valls-Sole J, Rothwell JC, Goulart F, Cossu G, Munoz E. Patterned ballistic movements triggered
by a startle in healthy humans. Journal of Physiology 1999;516(3):931–8.
Werner W, Dannenberg S, Hoffmann K-P. Arm-movement-related neurons in the primate superior
colliculus and underlying reticular formation: comparison of neuronal activity with EMGs
of muscles of the shoulder, arm and trunk during reaching. Experimental Brain Research
1997a;115:191–205.
Werner W, Hoffmann K-P, Dannenberg S. Anatomical distribution of arm-movement-related neu-
rons in the primate superior colliculus and underlying reticular formation in comparison with
visual and saccadic cells. Experimental Brain Research 1997b;115:206–16.
Wise SP, di Pellegrino G, Boussaoud D. The premotor cortex and nonstandard sensorimotor
mapping. Canadian Journal of Physiology and Pharmacology 1996;74:469–82.
Zelaznik HN, Hawkins B, Kisselburgh L. Rapid visual feedback processing in single-aiming
movements. Journal of Motor Behaviour 1983;15(3):217–36.
Chapter 6
Rethinking the Role of Motor Simulation
in Perceptual Decisions
Scott T. Grafton and Shivakumar Viswanathan
6.1 Introduction
Goal-oriented action demands a constant ebb and flow of information processing

between perception and movement. Decisions about whether or not a task can be
executed, such as jumping over a gap or reaching to switch a light, demand reliable
perceptual evidence. Likewise, perceptual decisions that are particularly subject
to optical distortion might benefit from physical knowledge gained through motor
experience. For example, the inaccurate estimation of angles and azimuths during
navigation could benefit from a motor-based simulation of the body moving in the
world. One of the ways these kinds of decisions can be achieved is by conscious
access to the motor system to form virtual movements. We can imagine jumping
over the gap, and use this to influence our decision to act. Or we can imagine
turning ourselves in the environment to get a better estimate of the relative bearing
between locations. Increasing evidence suggests that motor simulation of this sort can
aid subsequent overt motor performance, increase the rate of skill acquisition, and
potentially be used to solve cognitive or perceptual problems involving the physical
environment (Holmes and Collins 2001; Mulder et al. 2004). Understanding what
behavioral or neural signatures constitute a motor simulation is an essential step
for understanding how simulation might actually influence overt motor behavior
or enhance perception. In this chapter, we consider a widely used experimental
protocol that is broadly assumed in the literature to represent motor simulation, the
hand judgment or laterality task (Cooper and Shepard 1975; Parsons 1987a). We
provide overwhelming evidence that the behavioral phenomena associated with this
task are not a result of motor simulation, and provide an alternative interpretation
of these studies. The point of this review is not to undermine motor simulation as
an important process or concept. Rather, we argue that progress in motor simulation
S. T. Grafton () · S. Viswanathan

Department of Psychological Brain Sciences, University of California,
Santa Barbara, CA 93106-9660, USA
e-mail:[email protected]

and Biology 826, DOI 10.1007/978-1-4939-1338-1_6
70 S. T. Grafton and S. Viswanathan
research requires a higher standard of confirmatory evidence for what constitutes

motor simulation and that the behavioral effects that are commonly reported are not
necessarily sufficient to conclude that a motor simulation is occurring.
The general concept of simulation emerged in the 1970s during an explosion of
cognitive science directed at understanding how mental representations were manip-
ulated to solve problems (Shepard 1978). Simulation is now widely acknowledged
to be an integral component of cognition (Barsalou 2008) and as a mechanism for
manipulating or transforming internal representations to new states (Kosslyn et al.
2001). A classic example of how simulation is used in perceptual problem solving is
the mental rotation of complex visual objects in matching tasks. In this case, reaction
times (RTs) typically correlate with the rotation angle needed to match the orien-
tation of one object to another, revealing a mental operator with analog properties
acting on the internal representation of the object (Moulton and Kosslyn 2009). The
concept of motor simulation followed suit. In this form of simulation, knowledge
derived from the motor system might be used to organize virtual behavior. Motor
simulation has links to explicit motor imagery or emulation, where we can imagine
moving our body without actually displaying any overt action (Jeannerod and Decety
1995; Decety 1996; Jeannerod and Frak 1999; Jeannerod 2001). Similar to object
rotation, the time to complete imagined movements also displays analog properties
corresponding with real movement (Decety et al. 1989). However, in all of these early
studies of motor simulation (or equivalently, imagery and emulation), there was no
cognitive or motor problem in the experiments that actually needed to be solved by
the participant. Thus, it was unclear whether motor simulation is actually a meaning-
ful operation for solving cognitive problems. While one can imagine moving one’s
hand, it is challenging to find real examples where this sort of action imagery is
utilized for planning or execution of everyday behavior let alone for virtual motor
simulation.
One of the earliest and most widely cited examples of a motor simulation pro-
cess that might actually be used to solve a perceptual problem is the laterality task
(Cooper and Shepard 1975). In this wonderfully simple paradigm, the subject must
decide if a visual image of a hand constitutes an image of a right or left hand, and
then report their choice with a button press using their own right or left hand (Luria
1966). The three-dimensional (3-D) mirror symmetry of the two hands makes them
chiral objects, which could be potentially difficult for the visual system to distin-
guish from each other. In the laterality task, also referred to as the hand judgment
task, RTs depend on the angle of rotation between the hand stimulus and an unseen
vertically oriented “virtual” hand (palms directed away from the subject with the
fingers pointing upward). The analog property of RTs in the laterality task created a
strong conceptual link with other forms of simulation such as object rotation, sug-
gesting there is an internal representation of the hand stimulus that is transformed to
a new orientation. Cooper and Shepard interpreted the analog delay in the laterality
task as part of a decision-making process involving both perceptual discrimination
and response selection. They proposed that the visual system was making a holistic
analysis of the object, reorienting it, and comparing it to a canonical virtual left or
right hand. They wanted to know if right versus left discrimination could be solved
6 Rethinking the Role of Motor Simulation in Perceptual Decisions 71
Fig. 6.1 Traditional model of motor simulation in the hand judgment task: Laterality of the visual
hand stimulus is determined by comparing it to either of the “felt” hands, which are rotated by covert
motor activity to a matching spatial orientation. This provides a mapping between each stimulus
and the corresponding hand used to make the physical response
“without having either to preserve or to search through a store of fixed 2-D templates
corresponding to all possible retinal projections.” Mental rotation of the object (in
this case a hand) provided an algorithm for reducing the storage and search require-
ments, as shown in Fig. 6.1. The simulation (and virtual hand rotation) provided
new information that could be used to improve the decision-making process. While
compelling as a possible cognitive mechanism, the authors fully acknowledged the
fragilities of this interpretation. First, it was possible that the analog property in the
laterality task was not sufficient evidence on its own to demonstrate that an object
rotation was invoked to make the perceptual decision. Second, they acknowledged
that if it could be demonstrated that a parts-based analysis was sufficient to solve
hand identification then the use of simulated hand rotation to aid in the perceptual
decision was less likely. In parts-based analysis, the observer can use local features of
the object, such as the wrinkles on the palm or the pattern of the fingernails to make
the identification. The alternative, a whole-based analysis, relies on understanding
the complete 3-D structure of the object.
6.2 Motor Simulation and the Birth of a Conundrum
Uncertainties about what was actually being simulated or judged in the laterality task
appeared early in the literature. In an experiment testing the influence of head position
on laterality judgments, Sekiyama noticed that when the RTs, as a function of angle
Fig. 6.2 An example of the

correct-hand effect: The
distribution of reaction time is
different for right- and
left-hand stimuli. This can
only occur if the subject could
distinguish the stimuli
perceptually. (The plot is from
Fig. 2 in Sekiyama 1982)
of rotation, were plotted separately for each hand, the profiles were not symmetric
(Sekiyama 1982). They were skewed such that the RT profiles were mirror patterns
for right- and left-hand responses, as shown in Fig. 6.2. This pattern was not apparent
in the data from Cooper and Shepard, because they averaged the RTs for the two
hands together (Cooper and Shepard 1975). The pattern of separate RT data for each
hand in Sekiyama’s data raised a critical issue. It could only occur if the subject
was using a different mental process for each hand. This in turn implied that there
was already sufficient perceptual information in the stimuli for subjects to be able to
discriminate left and right hands. This possibility in turn raised the question of how
if at all mental rotation of the hand actually contributed to the perceptual decision-
making process. Sekiyama did not provide a clear answer. She interpreted the results
from a kinesthetic framework, such that the “felt” position of the hands might be
used to aid in the judgment.
The kinesthetic properties of the laterality task provided by Sekiyama were sub-
sequently overshadowed by Larry Parsons’ elegant experiments showing a range
of new effects that were interpreted as overwhelmingly motoric in nature (Parsons
1987a). First, he demonstrated a remarkable sensitivity of the RTs to biomechani-
cal constraints (Parsons 1994). RTs were highly sensitive to extreme positions such
that hand stimuli in postures that normally are not comfortable led to slower RTs.
Second, the analog properties of the laterality task were generalized to other body
parts. They were also observed in a foot version of the task (Parsons 1987b). Third,
many subjects (including the authors of this chapter) reported a sense of body motion
when performing the task. Fourth, the RTs for judging hands at different rotations
were proportional to the time to make actual movements from that position to a
neutral point with the joint angles at mid-position. These effects led Parsons to note,
“These imagined paths seemed to simulate the paths used for physically moving
the hand or foot between their task orientation and the orientation of the stimulus”
(Parsons 1987a). All of these findings have stood the test of time and continue to be
replicated 25 years later. In addition, follow-up studies demonstrated that the RTs
were sensitive to the posture of the participant as they performed the task (Sirigu
and Duhamel 2001; Ionta et al. 2007). For example, placing the left hand behind the
back would slow the RTs for left-hand judgments. They entrench the notion that the
perceptual decision making in the laterality task is based on motor simulation (Ionta
et al. 2007; ter Horst et al. 2010; Ferri et al. 2011; Ní Choisdealbha et al. 2011),
motor imagery (Grush 2004) or motor emulation (Moulton and Kosslyn 2009). The
motor simulation mechanism continues to be used to interpret findings in a variety
of patient populations as well (Deconinck et al. 2009; Helmich et al. 2009; Williams
et al. 2011).
However, as with Sekiyama’s data, Parsons’ work also reveals the remarkable
skewing of RT profiles when the two hands are plotted separately. This skewing
can only occur if the participants knew which hand to covertly rotate on each trial.
This correct-hand effect shows that perceptual identification of hand chirality is
orientation-invariant, even in the face of the added challenges imposed by the visual
mirror symmetry of the hand stimuli. Despite the seemingly impregnable interpre-
tation that a motor simulation is used in the laterality task, the correct-hand effect
observed in data from Sekiyama, Parsons, and countless follow-up studies intro-
duces a fundamental conundrum. A consistent limb-specific mirror skewing of the
RT profiles could only occur if subjects identified the correct hand, invariant to ori-
entation, prior to any putative motor simulation. If simulation is used in perceptual
decision making, then how does the motor simulator always choose the correct hand
in the first place and, more importantly, why bothr with motor simulation at all?
The conundrum clearly speaks to Cooper and Shepard’s original concern that if local
perceptual features could be used to discriminate the stimulus, then the argument
that simulation is used for perceptual decision making is flawed. In the face of this
uncertainty, Parsons proposed an alternative explanation that could sustain motor
simulation as a putative mechanism, as shown schematically in Fig. 6.3. He noted
that “Performance in the left-right judgment task appears to involve these operations,
some of which may occur concurrently: (a) analysis of the orientation and handed-
ness of the stimulus; (b) analysis of the orientation of the internally represented
corresponding hand; (c) planning a path for the internally represented hand to move
(within its joint constraints) to the orientation of the stimulus; (d) mental simulation
of planned action; and (e) exact-match confirmation of shape of imagined and per-
ceived hands” (Parsons 1994). His final point addresses the conundrum by invoking
a post hoc confirmation mechanism. This mechanism could take two forms. One
would be a direct confirmation of the percept. However, the entire notion of a post
hoc motor simulation to confirm what is already perceived is deeply problematic. It
is difficult to find any other examples in the cognitive science literature where motor
simulation or imagery is used to directly confirm an already correct percept. The
other explanation for the post-hoc confirmation step proposes that mental rotation
is needed to map the correctly perceived hand into the appropriate motor response,
that is, there is uncertainty in the mappings between “right” and “left,” and the sides
of the body used to make the response. In a recent set of experiments where local
features of the hand stimuli were manipulated, we discovered that neither of these
post-hoc confirmation mechanisms is tenable (Viswanathan et al. 2012).
Fig. 6.3 The conundrum introduced by the correct-hand effect. A correct percept of the visual
stimulus is always used as the referent for comparison in the motor simulation. Thus, it is unclear
what the subject is actually deciding since they already recognize the chirality of the stimulus
6.3 Multisensory Hand Binding
We propose that the behavioral properties of the laterality task are not due to motor
simulation. Instead, the analog and biomechanical features of the laterality task are
a consequence of what we refer to as “multisensory hand binding.” At the onset of
the stimulus, the observer has simultaneous access to sensory information about the
same entity (i.e., a hand) from different modalities—the visual representation of the
seen hand depicted by the stimulus, and the proprioceptive representations of the
observer’s own felt hands (Shenton et al. 2004). There are more than enough local
features in the visual stimulus used in a typical laterality experiment to uniquely iden-
tify it as the front or back of the hand. According to the multisensory hand-binding
hypothesis, there is a simultaneous, (orientation-invariant) cross-modal comparison
of the seen hand to the proprioceptive representation of the corresponding hand, as
shown schematically in Fig. 6.4. The outcome of cross-modal comparison is not a
match/mismatch signal that is used for perceptual decision making. Instead, it re-
sults in a binding of the visual representation of the seen hand to the proprioceptive
representation of the “matching” felt hand or vice versa. Successful binding pro-
duces an intermodal discrepancy due to the differing orientations of the bound seen-
and felt-hand representations (in a body-centered reference frame). This discrepancy
automatically initiates a sensorimotor recalibration process to align the spatial rep-
resentation of the felt hand to that of the seen hand—an aftereffect that can cause
a feeling of moving. Critically, the motor response that is required in the lateral-
ity task is delayed until this intermodal conflict is resolved. This delay leads to the
laterality specificity of the RTs. We speculate that the need to perform this recali-
bration between the seen and felt hand is evidence that the motor system requires
an internally consistent and unique representation of body position (or state) as part
of generating a motor command, although this consistency argument remains to be
tested directly. Critically, the recalibration in multisensory hand binding is achieved
by relating representations in different sensory reference frames with respect to each
other. A critical prediction is that this occurs without requiring a motor signal from
either premotor or motor cortex and does not require any sort of motor simulation,
emulation, or explicit strategy. With multisensory hand binding, the response delay is
related to recalibration of body position as a part of planning a normal movement and
not related in any way to perceptual decision making. Thus, there is no conundrum
as arises with the standard motor simulation account of laterality judgments.
Fig. 6.4 The multisensory binding process: In the laterality task, a visual stimulus of the hand
is evaluated in terms of hand shape and orientation (palm up/down) and this information is used
to establish chirality of the visual stimulus. This is matched to the spatial representation of the
corresponding felt hand. The planning of a motor response requires that this spatial discrepancy
between two estimates of body position be reconciled
6.4 Experimental Evidence for Multisensory Hand Binding
Unlike the traditional motor simulation and confirmation model, the multisen-
sory hand-binding hypothesis imposes strong constraints on when and whether the
correct-hand effect will occur. Here, we summarize a study that used an attention
manipulation to selectively induce a “wrong-hand effect” (Viswanathan et al. 2012).
The experiment selectively extinguished the correct-hand effect by eliminating the
contextual relevance of the proprioceptive inputs. Neither effect is plausible with
Fig. 6.5 Stimuli in the standard form of the laterality task have sufficient local features to determine
chirality (a). In an ambiguous silhouette, each shape can correspond to either hand (c). With simple
symbolic cues, chirality can be restored (b)
motor simulation. Finally, we describe a recent study that requires a simple percep-
tual decision of determining if two ambiguous hand silhouettes have the same or
different outlines. The results show that there is an obligatory multisensory binding
of the stimuli to the observer, even in a task when no right/left judgments are required.
Inducing the Wrong Hand According to the hand-binding hypothesis, only sensory
information about the spatial configuration (“shape”) of the visual hand represen-
tation can be correlated with the proprioceptive hand representations. The afferent
proprioceptive representation of a hand codes the relative spatial position of the dig-
its and hand, but does not itself contain information about visual attributes such as
the color, lines, and textural patterns of the skin covering that hand. Therefore, the
“shape” of the seen hand and the “patterns” on the seen hand should have dissocia-
ble effects on laterality identification. To test this prediction, we exploited the fact
that visual hand “patterns” are indispensable in identifying the laterality of certain
hand shapes. In Fig. 6.5a, the visual patterns on each hand are diagnostic of whether
the palm or the back of that hand is being viewed. Without these patterns, the 2-D
hand shapes exhibit multiple symmetries. The shapes of the palm-up and palm-down
views of the same hand are mirror-symmetric, as are the shapes of the right and left
hands having the same view as shown in Fig. 6.5c. Furthermore, the palm-up view
of one hand has the same shape as the palm-down view of the other hand. Due to
these symmetries, the laterality of a hand cannot be uniquely determined based on
shape alone, without the view information from the visual patterns. The Cooper–
Shepard paradigm was altered to include a task-set manipulation, using the hands in
Fig. 6.5b as stimuli. Black silhouettes depicted hand shapes and colored dots were
used to denote hand orientation. A red dot indicated a palm-down view of the hand,
and a green dot indicated the palm-up view. Participants readily learned to recog-
nize this color-to-view mapping before the experiment. The paradigm involved two
independent conditions—the View-first condition and the Shape-first condition.
Each condition consisted of cued and uncued trials that occurred with (approxi-
mately) equal frequency. On the cued trials of the View-first condition, participants
first saw a colored dot (red or green) indicating the view of the forthcoming test
stimulus. After a brief offset, a test stimulus depicting a hand shape (without view
information) was presented at a variable picture-plane orientation. On the uncued

trials, the advance cue (a gray dot) did not provide any task-relevant information,
and the subsequent test stimulus contained both view and shape information together.
Participants had to combine view and shape information, whether presented serially
(cued trials) or simultaneously (uncued trials), to identify the laterality of the hand.
The stimulus ordering on the cued trials was reversed in the Shape-first condition—
the advance cue was a hand shape, followed by a test stimulus depicting that hand’s
view. Critically, the uncued trials were identical in the Shape-first and View-first
conditions. We assumed that participants would selectively attend to one and then
the other visual attribute of the stimulus in the same sequence used in the cued tri-
als of that condition. With this assumption, the hand-binding hypothesis predicts
that response times on the uncued trials should differ in a very specific manner be-
tween the View-first and Shape-first conditions. On the View-first condition, “view”
should be processed before “shape.” According to the hand-binding hypothesis, view
representations should therefore be available to mediate the subsequent perceptual
correlation of hand shape to the proprioceptive hand representations, thus leading to a
unique and “correct” laterality binding solution. Consistent with this, the RT profiles
for these stimuli were mirror-reversed in the View-first condition, consistent with
the correct-hand effect reported by previous studies. On the Shape-first condition,
“shape” would be processed before “view.” A hand shape without view information
has an ambiguous laterality. Due to the participant’s palm-down hand position, each
shape should be correlated with the spatial configuration of a unique proprioceptive
hand representation. Consequently, the ambiguous hand shapes are automatically
bound to the correlated felt-hand representation, according to the hand-binding hy-
pothesis. This “premature” binding, without incorporating stimulus-defined view
information, leads to a critical result. Stimuli having identical shapes but different
views were bound to the very same felt hand. RT profiles for these stimuli were
not mirror-reversed in the Shape-first condition, unlike the predictions for the View-
first condition. Due to this premature binding, palm-up stimuli were consistently
bound to the “wrong” felt hand, in violation of the correct-hand effect. Note that
this “wrong-hand effect,” while eliminating the asymmetry of RT profiles, does not
necessarily result in “wrong” laterality judgments. When the binding enters aware-
ness, participants can evaluate whether the stimulus-defined view corresponds to a
palm-down or a palm-up view, and respond accordingly with either the bound hand
or the opposite hand.
Binding Without Laterality Judgments We performed a variant of the Cooper–
Shepard paradigm to demonstrate that hand binding could be tightly controlled by
the experimenter. Participants had to interpret an ambiguous hand silhouette palm
up or palm down, rather than left or right hand. We tested if this perceptual deci-
sion, which had no laterality judgment, would again interact with the responding
“felt” hand. On each trial, participants first received an advance cue specifying the
laterality of the forthcoming test stimulus. That is, subjects knew in advance if a
right- or left-hand stimulus would be presented, eliminating hand judgment alto-
gether. The left/right instruction also indicated which hand to use to indicate their
palm-up/palm-down decision. In doing so, the decision making has no uncertainty

in the mapping between “right” and “left” stimuli and the side of the body to be
used to respond. After a short delay, a test stimulus depicting only a hand shape was
presented. Participants judged whether the test stimulus depicted the palm-up or the
palm-down view of the hand of known laterality with an index or ring finger button
press. Despite explicit prior knowledge of the expected hand laterality, pre-attentive
mechanisms involved in laterality identification were predicted to still be involved.
We found that the advance preparation of the response hand induced a “selective at-
tention” to the felt representation of the responding hand, while inhibiting inputs from
the other (nonresponding) hand. Consistent with the hand-binding hypothesis, only
shapes correlated with the palm-down position of the response hand led to successful
binding, but not shapes corresponding to the palm-up view, even though these latter
shapes are correlated with the palm-down position of the nonresponding/unattended
hand. The palm-down stimuli of each hand were associated with the characteristic
RT mirror asymmetry of the correct-hand effect, but not the palm-up stimuli. This
systematic difference in RTs between the palm-down and palm-up stimuli supports
the assumption of the hand-binding hypothesis that successful binding is the basis
for recalibration and the associated illusory movement reported by many subjects.
These results further argue that the mechanism causing the illusory movements is
not under direct strategic control, despite the extensive voluntary control over the
sensory inputs and the interpretation of the outputs. If it were a “strategy,” then it
would require participants to pre-identify the palm-up and for palm-down stimuli in
order to apply a different strategy to each. Furthermore, since the palm-down stimuli
for one hand are identical to the palm-up stimuli for the other hand, these results are
inconsistent with the exclusive use of a strategy exploiting the visual position of the
thumb. Additionally, if participants used “simulated” movements to either confirm
or disconfirm their decisions, the RT asymmetries based on the correct-hand effect
should be present on both palm-down and palm-up stimuli, contrary to the observed
results (Fig. 6.6).
Automatic Multisensory Hand Binding When It Is Irrelevant We further tested if
multisensory hand binding is an automatic process, unrelated to hand judgment, in
a two-handed version of the laterality task. Subject’s observed two ambiguous hand
silhouettes positioned one above the other (Fig. 6.7a). The task was to decide as
quickly as possible if the two silhouettes were rotated versions of the same shape,
or different, mirror-symmetric versions of the two shapes. The main axes of the two
shapes were always oriented 120◦ apart (Fig. 6.7b). Extensive research in perceptual
matching studies of rotated objects has established that the solution to this problem
is to mentally rotate one object to match the orientation of the other and the time to
do this depends on the arc length or angle of the rotated object (Cooper and Shepard
1984). Because the two orientations are always 120◦ apart, the time should be the
same irrespective of the absolute orientation of the images (as shown on the circle
in Fig. 6.7c). The orientations were chosen such that subjects could not use visual
symmetry as a perceptual “shortcut” for detecting two mirror shapes. Critically, if
there is multisensory hand binding, then the automatic binding of either of the seen
Fig. 6.6 Binding without

laterality judgment: In this
task, subjects are told if the
stimulus is a left or right hand
and told to respond with the
corresponding left or right
hand and report if the
stimulus corresponds to a
palm-up or palm-down
orientation. Thus, there is no
laterality judgment and no
uncertainty over the laterality
of the motor response.
Subjects demonstrate a
correct-hand effect for
palm-down hand stimuli that
match the responding hand,
but not for matching palm-up
stimuli
hands and the observer’s responding hand will lead to biomechanically constrained
delays in response time and a dependency of RT as a function of where the axes are
positioned (Fig. 6.7d). This RT dependency on position was found to be particularly
dramatic for orientations associated with the most uncomfortable position to rotate
the hands into (Fig. 6.7e). When results of this two-handed version of the laterality
task are combined with the traditional single-hand judgment studies, the generality
of the results as well as the automaticity of the behavior constitute strong evidence
that multisensory binding is an ecologically valid process (De Gelder and Bertelson
2003).
6.5 Reinterpreting the Laterality Task
Multisensory hand binding and recalibration of body position are sufficient to ex-
plain much of the behavioral phenomenology typically found in experiments that
use the laterality task or its variations. First, it is well known that conflicts between
seen and felt body positions can induce a sense of motion without any overt motor
command or covert motor simulation. Anyone who has experienced aftereffects of
motion from a carnival ride or experienced “sea-legs” after returning to the shore can
attest to this (Cohen 1996). In other words, a sense of motion by itself is insufficient
evidence to conclude that a motor simulation is occurring. Second, the sensitivity
of the laterality task to extreme body postures could be based on biomechanically
constrained boundary conditions placed on an internal model of the body schema
rather than on limitations to the range of possible simulated motor commands. That
is, our ability to mentally represent the position of our body in space (irrespective
Fig. 6.7 In the two-handed matching task, two hand silhouettes with ambiguous chirality are
presented, and the subject must decide if they are the same hand or a mirror (different) pair (a). The
stimuli are presented one above the other to avoid perceptual mirror symmetries. The two hands
are always presented with the primary axis rotated 120◦ with respect to each other (b). Mental
rotation of one object to another would lead to response times that are invariant to the absolute
angle instances (c). In contrast, if hand binding occurs, there will be delays in response times as
a function of angle instance (d). Data from subjects performing this task confirm there is a strong
hand-binding effect (e)
of our generation of motor commands) is strongly constrained by real biomechan-

ics. Furthermore, as amplified below, any clinical disorder that disturbs the body
schema (amputation, locked-in syndrome, chronic pain) could distort the internal
representation of body position and influence the time needed to relate this internal
schema with visual hand stimuli. Third, the sensitivity of response times to the static
posture of the observer could be due to peripheral, proprioceptive influences on the
internal model of the body position, irrespective of any motor involvement (Riemer
et al. 2010). Fourth, experimentally distorting hand position (using methods such as
vibrotactile stimulation of the biceps tendon) should influence RTs in the laterality
task (McCormick et al. 2007). Fifth, the multisensory hand-binding process appears
to be pre-attentive or automatic. Laterality judgment delays are resistant to explicit
control such as instructions that tell the subject a stimulus is their own or another
person’s (Ferri et al. 2011).
6.6 Neural Mechanisms of Multisensory Binding
The multisensory hand-binding model proposes that the registration of different

estimates of body position plays a dominant role in the preparation for action. Pro-
prioceptive representations of one’s hands are continuously accessible as they provide
a running estimate of the position of the hands relative to the body, both when the
hand is at rest and in motion, and even when the hands cannot be seen. It is known
that the cross-modal comparison of the representations of a seen hand with that of the
felt hand is critical for maintaining the coherence of one’s body schema (Graziano
and Botvinick 2001; Ehrsson et al. 2004), and essential for enhancing motor perfor-
mance (Desmurget et al. 1995). Intersensory conflicts can result in a variety of body
schema distortions such as the rubber hand illusion (Botvinick and Cohen 1998; Pa-
vani et al. 2000), extracorporeal perception, and errors of agency attribution (Petkova
and Ehrsson 2008; Slater et al. 2009).
The need for coherence across sensory modalities implies that there is a high level
of specificity to the perceived visual stimulus, as a visually seen right hand should not
be confused with a felt left hand, and vice versa. Neurons identified in Brodmann’s
area 5 of the superior parietal lobule (SPL) of the nonhuman primate exhibit such
specificity (Graziano et al. 2000). They have a graded response depending on the
angular disparity between the felt arm and a seen artificial arm. Furthermore, these
neurons respond in the same manner to an inverted hand (the back of the same hand)
but not when the visual cues imply that the inverted hand belongs to the opposite
arm. In this latter case, the neurons responded as if there was no stimulus present
at all. Similar properties are observed in dorsal premotor cortex (PMd) (Graziano
1999). These two areas form a network that transforms vision and body-centered
reference frames (Caminiti et al. 1996; Crawford et al. 2011). The properties of
neurons within these two areas in nonhuman primates form a compelling substrate
for multisensory hand binding. That said, further studies are needed to understand
the relationship between the time it takes to resolve multisensory conflict and the
computational processes within SPL neurons that could be mediating this process.
While direct neuronal evidence of the hand-binding process is not available in
humans, a wealth of indirect evidence using functional imaging with both positron
emission tomography (PET) and functional magnetic resonance imaging (fMRI),
spanning 15 years of study in 10 experiments and over 120 subjects, demonstrate
localization to the SPL and to a lesser degree PMd (Hetu et al. 2013). The SPL loci
are remarkably similar in location to nonhuman primate recordings in Brodmann’s
area 5. The specific involvement of SPL in imaging is particularly apparent when
hand judgment is contrasted with object rotation, which effectively controls for motor
cortex activity related to making the motor response (Bonda et al. 1995; Kosslyn et al.
1998; de Lange et al. 2005). Using parametric event-related fMRI, SPL activity scales
with biomechanical complexity during the laterality task. We propose this graded
response is related to the multisensory binding, analogous to what is observed in
nonhuman primate neurons rather than to a simulation (de Lange et al. 2006).
Hand binding in the laterality task appears to be a highly specialized property of the
superior parietal cortex, based on a comparison with other functional determinants
of body representation. In a factor analysis of patients with parietal lobe stroke, a
laterality task deficit was distinct from other tests of body representation including (1)
chronometric scaling while imagining moving the fingers contralateral to the lesion,
(2) localization and identification of body parts, and (3) understanding the relative
position of different segments of the arm (e.g., that the wrist is distal to the elbow).
Furthermore, lesion overlap maps demonstrated increased involvement of the SPL
for patients with laterality deficits that were different from probable lesion locations
for the other body representation disorders (Schwoebel et al. 2001). The behavioral
and lesion location differences of laterality judgment and imagined finger movements
are quite important, as the latter is generally considered to be a prototypical example
of motor simulation (Sirigu et al. 1995).
Each parietal-premotor cortex represents to a greater degree the proprioceptive
state of the contralateral hand (Bernier and Grafton 2010). Thus, binding should only
occur between a seen hand and the felt hand of only one hemisphere. This might be
reflected in PET and fMRI experiments by greater brain activity for one hand than the
other. However, most imaging studies show bilateral recruitment of SPL and PMd
cortices in the laterality task, even when activity for each hand is modeled separately
(de Lange et al. 2006). This might be expected, given that the hemispheres are re-
ciprocally connected, and activation of one will co-recruit the other. Patient studies
could possibly provide evidence that only one hemisphere is “bound” per trial. In
stroke patients, lesions of the left hemisphere lead to slowing of judgments to ei-
ther hand stimulus whereas right brain damage slows neither (Tomasino et al. 2003;
Daprati et al. 2010). Irrespective of the side of the lesion, there might be mechanisms
unrelated to multisensory binding influencing patient performance. The lesion lo-
cation is diverse across subjects, there is an influence of hemispheric specialization
for hand dominance, and patients may under go reorganization post stroke. The best
evidence that the SPL is specialized to bind vision and proprioception of the opposite
hand can be found in patients with a sectioning of the corpus callosum (Parsons et al.
1998). Using hemifield stimuli and the hand laterality task, it is clear from error rates
that each hemisphere only matches the contralateral hand.
6.7 Is the Motor Cortex Involved in the Laterality Task?
One possible acid test of motor simulation in the laterality task would be a demon-
stration that the motor cortex of healthy subjects is involved during performance
of the task. Motor cortex can clearly be activated in tasks explicitly cueing imag-
ined movement (Lotze et al. 1999; Sharma et al. 2008). However, the evidence for
motor cortex engagement in the laterality task is by no means apparent. Early PET
imaging experiments of the laterality task using block designs reported activity of
the motor cortex (Kosslyn et al. 1998). With well-balanced experimental designs
and higher-resolution imaging, it became evident that motor cortex activity is not
greater for hand judgment than object rotation (Bonda et al. 1995). Furthermore,
motor cortex activity only appeared at the time of a response, presumably after any
perceptual decision had occurred (de Lange et al. 2005). A recent meta-analysis of
75 imaging studies of motor simulation compared localization in the laterality task
with other forms of motor simulation, such as kinesthetic or visual motor imagery
(Hetu et al. 2013). With the laterality task, motor cortex activation is rare whereas
the other forms of motor imagery task demonstrated motor cortex involvement about
25 % of the time. When single-pulse transcranial stimulation (Ganis et al. 2000) or
slow, 1 Hz, repetitive transcranial magnetic stimulation (TMS; Pelgrims et al. 2011)
was used to create a “virtual” lesion of motor cortex, there was a general slowing of
RTs in the laterality task. However, these results did not influence the correct-hand
effect and only described a generalized slowing of responses. Furthermore, in a de-
tailed single-pulse TMS experiment in which virtual lesions of motor cortex were
chronometrically manipulated and the correct-hand effect was carefully measured,
there was a complete failure to disrupt hand judgment performance at any stimu-
lation time post stimulus, establishing that there is no time window during which
the motor cortex makes a contribution to mental rotation of the hand (Sauner et al.
2006). Although motor cortex may be involved in other motor simulation paradigms
such as kinesthetic motor imagery, there is no compelling evidence that it is used in
hand laterality judgments, undermining the argument for a motor simulation process
in this task.
6.8 The Malleable Body Schema
State estimation is a computational process that has been proposed as a way to in-
tegrate sensory and motor information of position in real time (Tin and Poon 2005;
Mulliken et al. 2008). Retinotopic-based reference frames appear to play a dominant
role in representing space and planning action under typical experimental conditions.
Vision provides a powerful and accurate map of limb position and the environment
and may serve as a default reference frame for organizing body state and action
(Cohen and Andersen 2002). There is growing consensus that the reference used to
determine position is not absolute, but contingent on available evidence from dif-
ferent sensory channels (Denève and Pouget 2004; Faisal et al. 2008; McGuire and
Sabes 2009). While vision-based information typically dominates, body-based ref-
erence frames can also be used under degraded vision. However, maps of the body
position based on proprioception as well as tactile input are far less veridical than
visual inputs (Kammers et al. 2009). They are easily distorted by illusions induced by
visual–tactile (Botvinick and Cohen) or visual–proprioceptive (Burrack and Brugger
2005) mismatches and do not clearly map onto the body topology precisely (Longo
and Haggard 2010). Nevertheless, the correct-hand effect of the laterality task sug-
gests that recalibration relies on a body-centered reference frame with a “default”
proprioceptive configuration: palms down and fingers pointed upwards.
Notably, the default configuration that is “bound” in the laterality task can be re-
shaped by additional information such as repositioning of the subject (Riemer et al.
2010). It is also influenced by the illusion of wrist flexion induced by vibrotactile
stimulation to the wrist extensor tendons. In this case, RT delays were only observed
when judging hand stimuli corresponding to the side of the stimulation (McCormick
et al. 2007). It also appears to be overridden by biases induced by experimental
context, such as constraining motor responses to one side or the other, as well as
handedness (Ní Choisdealbha et al. 2011). Other contextual influences, such as on-
going observed or executed action, also influences the estimated body configuration
(Conson et al. 2009). Interestingly, in young children, motor behavior is strongly
shaped by a body representation that is close to the default position described in the
laterality task and they are less influenced by their own body position (Saimpont et al.
2009). As they develop and acquire knowledge of the dynamics of their bodies, their
body representation becomes more closely aligned with the actual physical position
and this is reflected in a sensitivity of laterality judgment times to the participant’s
position. It is possible then that in the face of uncertainty about limb position, adults
might resort to this developmentally more primitive map of body position.
6.9 Multisensory Binding in Patients
Many patient populations have been tested with the laterality task and abnormal
findings have invariably interpreted as evidence for deficits of motor simulation,
even when there is no lesion of the motor system. In many cases, a more direct
explanation can be drawn from multisensory binding. An illustrative case is patients
who have undergone a limb amputation. It is well established that they can have
a significantly distorted body schema (Ramachandran and Hirstein 1998). In the
framework of multisensory binding, it is the poor alignment of a visual hand to a
distorted body map that leads to the delays of laterality judgment, particularly for
the hand stimulus corresponding to the distorted body schema (Reinersmann et al.
2010).
Another illustrative population who has abnormalities of laterality judgment is
patients with pain disorders. For those with unilateral, chronic limb pain, there is
a delay in laterality judgments to the hand stimuli corresponding to the affected
side. The delay is also related to the duration of symptoms and to the pain that
would be evoked by executing the movement (Moseley 2004b; Coslett et al. 2010b;
Reinersmann et al. 2010). When a picture of a foot is used as the stimulus, delays in
judgment are also observed for the stimuli corresponding to the painful foot, even
when the subject responds with their intact hands (Coslett et al. 2010a). The fidelity
of the laterality task for tracking pain severity has led to the proposal that it be used
as an objective independent measure of subjective pain (Coslett et al. 2010b).
Schwoebel interpreted the response delays for stimuli corresponding to the painful
limb (whether hand or foot) as evidence that there was a sensitivity of the internal
representation of body position, particularly the nonvisual representation, to pain
(Schwoebel et al. 2001). The multisensory binding model draws on the same general
explanation that the delay is related to a distortion of the body schema. However,
with multisensory binding, there is an important new distinction worth amplifying.
In most of the studies of chronic pain, including the original paper by Schwoebel,
the “movement” of the body representation is assumed to be a motor simulation or
emulation, where the subject volitionally manipulates a body schema to solve the
laterality task. This of course reintroduces the conundrum of the correct-hand effect
and is inconstant with new experimental evidence (van Elk et al. 2012). In contrast,
with multisensory binding the rotation of the body schema is automatic and driven
by the incongruence with the observed hand position. This actually simplifies the
explanation.
It has been found that patients with chronic neuropathic pain can actually gain
some relief as measured by subjective pain scales when “treated” with the later-
ality task. Explanations for this improvement have been uncertain, and based on
recruitment of motor and premotor networks (which we now know is not an accurate
account of the networks used in this task) or perhaps due to focused attention on the
affected hand (which could just as readily make the pain worse) (Moseley 2004a).
The multisensory binding model provides an alternative explanation: One of the
consequences of chronic pain could be a distortion of the body schema, particularly
for the affected limb. Normalizing this distortion either by overt motor activity or
through coregistration of seen and felt limb positions in the laterality task might
reduce the distortion by image and associated pain.
6.10 When Does Binding Occur?
The laterality task reveals an illusion where a seen and felt hand are perceptually
bound and spatially matched. When does this illusion generalize? Evidence to date
would suggest the phenomenon also occurs with foot stimuli (Parsons 1987b). How-
ever, it is not at all clear that a correct-hand effect is observed when the hand stimulus
is replaced with a glove (Daprati et al. 2010) or is attached to a body (Zacks et al.
2002). In this case, RT effects are more consistent with typical object rotation tasks
or spatial compatibility effects. This would suggest that there is something special
about disarticulated hands and feet as sources of perceptual confusion. Binding for
these stimuli appears irrespective of the 3-D orientation of the stimulus (Parsons
1987b, 1987a). This undermines an assumption in some experiments that uncom-
fortable positions of visual stimuli would be treated by the observer as allocentric,
whereas comfortable positions would be egocentric (Brady et al. 2011).
Another important and unanswered question is whether an analogous binding
mechanism might be involved in the matching of the proprioceptive felt hand to
an object in the environment. Consider a primate swinging through the trees. Limb
grasping would require rapid accurate matching of limb orientation with a palm-
centric reference frame. This could be performed through a rotation of visual and
proprioceptive reference frames, with planning time influenced by biomechanical
effects including end-state comfort. As with the hand-binding model, there is no

need for motor cortex “simulation” to solve this realignment problem. Of note, when
subjects make perceptual decision about how their hand should be oriented to make
a power grasp on a dowel, the decision times have all these properties (Johnson
2000b). Furthermore, imaging studies demonstrate that both the SPL and PMd are
associated with the simulated hand rotation, in regions that are remarkably similar
to those found in the laterality task (Johnson et al. 2002). Finally, patients with
hemiparesis and motor cortex damage can continue to perform this grasp selection
task, further undermining a motor simulation account (Johnson 2000a). We speculate
that the object itself provides a visual reference frame that the felt hand can be bound
to without any need for motor simulation. For more complex behavior, including
the manipulation of tools, additional control mechanisms and estimates of position
would certainly need to be invoked.
6.11 What Should We Call Motor Simulation?
A consequence of the hand-binding model is that the perceptual judgment of a hand

does not rely on a motor simulation process. At a certain level, this is not surprising.
The perceptual demands of making a hand laterality judgment are not particularly
difficult when considered in the context of many other difficult perceptual problems
we face every day, and where no motor simulation is required. The hand-binding
model also suggests that much of the evidence that is used to make a case for mo-
tor simulation, whether it is an objective behavioral measure or subjective report of
movement, is inadequate. The observation that RTs are sensitive to hand orientation,
to biomechanical constraints, to body posture, to physical injury, or to a distorted
body scheme are insufficient evidence on their own to conclude that a motor sim-
ulation is occurring in a task. Similarly, a sense of motion could be based purely
on sensory mismatch. This does not mean we are proposing that motor simulation
cannot occur. Instead, we simply argue that the range of tasks involving motor sim-
ulation may be smaller than is commonly stated. The outcome of our work with the
laterality task is that a higher burden of proof is needed to conclude that a motor
simulation is involved in cognitive problem solving. The most convincing experi-
ments suggesting that a motor simulation is occurring require subjects to explicitly
imagine they are achieving specific action goals, such as walking to a target, tapping
fingers in a specific sequence, drawing, reaching, tracking, and making bimanual
actions (Decety et al. 1989; Sirigu et al. 1995; Decety 1996; Vargas et al. 2004). In
all these cases, there are chronometric links between real and imagined actions. The
tasks clearly involve willful, explicit motor emulation centered on a goal, rather than
on a perceptual decision.
Acknowledgments The work was supported by a grant from the James S. McDonnell Foundation
and the Institute for Collaborative Biotechnologies through contract no: W911NF-09-D-0001 from
the US Army Research Office.
References
Barsalou LW. 2008. Grounded cognition. Annu Rev Psychol 59:617–645.

Bernier PM, Grafton ST. 2010. Human posterior parietal cortex flexibly determines reference frames
for reaching based on sensory context. Neuron 68:776.
Bonda E, Petrides M, Frey S, Evans A. 1995. Neural correlates of mental transformations of the
body-in-space. Proc Natl Acad Sci U S A 92:11180–11184.
Botvinick M, Cohen J. 1998. Rubber hands ‘feel’ touch that eyes see. Nature 391:756.
Brady N, Maguinness C, Ní ChoisdealbhaA. 2011. My hand or yours? Markedly different sensitivity
to egocentric and allocentric views in the hand laterality task. PLoS ONE 6:e23316.
Burrack A, Brugger P. 2005. Individual differences in susceptibility to experimentally induced
phantom sensations. Body Image 2:307–313.
Caminiti R, Ferraina S, Johnson PB. 1996. The sources of visual information to the primate frontal
lobe: a novel role for the superior parietal lobule. Cereb Cortex 6:319–328.
Cohen H. 1996. Vertigo after sailing a nineteenth century ship. J Vestib Res 6:31–35.
Cohen YE, Andersen RA. 2002. A common reference frame for movement plans in the posterior
parietal cortex. Nat Rev Neurosci 3:553–562.
Conson M, Sarà M, Pistoia F, Trojano L. 2009. Action observation improves motor imagery: specific
interactions between simulative processes. Exp Brain Res 199:71–81.
Cooper LA, Shepard RN. 1975. Mental transformations in the identification of left and right hands.
J Exp Psychol Hum Percept Perform 104:48–56.
Cooper LA, Shepard RN. 1984. Turning something over in the mind. SciAm 251:106–107, 110–104.
Coslett HB, Medina J, Kliot D, Burkey A. 2010a. Mental motor imagery and chronic pain: the foot
laterality task. J Int Neuropsychol Soc 16:603–612.
Coslett HB, Medina J, Kliot D, Burkey AR. 2010b. Mental motor imagery indexes pain: the hand
laterality task. Eur J Pain 14:1007–1013.
Crawford JD, Henriques DYP, Medendorp WP. 2011. Three-dimensional transformations for goal-
directed action. Annu Rev Neurosci 34:309–331.
Daprati E, Nico D, Duval S, Lacquaniti F. 2010. Different motor imagery modes following brain
damage. Cortex 46:1016–1030.
De Gelder B, Bertelson P. 2003. Multisensory integration, perception and ecological validity. Trends
Cogn Sci 7:460–467.
de Lange FP, Hagoort P, Toni I. 2005. Neural topography and content of movement representations.
J Cogn Neurosci 17:97–112.
de Lange FP, Helmich RC, Toni I. 2006. Posture influences motor imagery: an fMRI study.
Neuroimage 33:609–617.
Decety J. 1996. Do imagined and executed actions share the same neural substrate? Brain Res Cogn
Brain Res 3:87–93.
Decety J, Jeannerod M, Prablanc C. 1989. The timing of mentally represented actions. Behav Brain
Res 34:35–42.
Deconinck FJA, Spitaels L, Fias W, Lenoir M. 2009. Is developmental coordination disorder a motor
imagery deficit? J Clin Exp Neuropsychol 31:720–730.
Denève S, Pouget A. 2004. Bayesian multisensory integration and cross-modal spatial links. J
Physiol Paris 98:249–258.
Desmurget M, Rossetti Y, Prablanc C, Stelmach GE, Jeannerod M. 1995. Representation of hand
position prior to movement and motor variability. Can J Physiol Pharmacol 73:262–272.
Ehrsson HH, Spence C, Passingham RE. 2004. That’s my hand! Activity in premotor cortex reflects
feeling of ownership of a limb. Science 305:875–877.
Faisal AA, Selen LP, Wolpert DM. 2008. Noise in the nervous system. Nat Rev Neurosci 9:292–303.
Ferri F, Frassinetti F, Costantini M, Gallese V. 2011. Motor simulation and the bodily self. PLoS
ONE 6:e17927.
Ganis G, Keenan JP, Kosslyn SM, Pascual-Leone A. 2000. Transcranial magnetic stimulation of
primary motor cortex affects mental rotation. Cereb Cortex 10:175–180.
Graziano MS. 1999. Where is my arm? The relative role of vision and proprioception in the neuronal
representation of limb position. Proc Natl Acad Sci U S A 96:10418–10421.
Graziano M, Botvinick M. 2001. How the brain represents the body: insights from neurophysiology
and psychology. Common mechanisms in perception and action 19:136–157.
Graziano MS, Cooke DF, Taylor CS. 2000. Coding the location of the arm by sight. Science
290:1782–1786.
Grush R. 2004. The emulation theory of representation: motor control, imagery, and perception.
Behav Brain Sciences 27:377–396; discussion 396–442.
Helmich RC, Aarts E, de Lange FP, Bloem BR, Toni I. 2009. Increased dependence of action
selection on recent motor history in Parkinson’s disease. J Neurosci 29:6105–6113.
Hetu S, Gregoire M, Saimpont A, Coll MP, Eugene F, Michon PE, Jackson PL. 2013. The neural
network of motor imagery: an ALE meta-analysis. Neurosci Biobehav Rev 37:930–949.
Holmes PS, Collins DJ. 2001. The PETTLEP approach to motor imagery: A functional equivalence
model for sport psychologists. Journal of Applied Sport Psychology 13:60–83.
Ionta S, Fourkas AD, Fiorio M, Aglioti SM. 2007. The influence of hands posture on mental rotation
of hands and feet. Exp Brain Res 183:1–7.
Jeannerod M. 2001. Neural simulation of action: a unifying mechanism for motor cognition.
Neuroimage 14:S103–109.
Jeannerod M, Decety J. 1995. Mental motor imagery: a window into the representational stages of
action. Curr Opin Neurobiol 5:727–732.
Jeannerod M, Frak V. 1999. Mental imaging of motor activity in humans. Curr Opin Neurobiol
9:735–739.
Johnson SH. 2000a. Imagining the impossible: intact motor representations in hemiplegics.
Neuroreport 11:729–732.
Johnson SH. 2000b. Thinking ahead: the case for motor imagery in prospective judgements of
prehension. Cognition 74:33–70.
Johnson S, Rotte M, Grafton S, Hinrichs H, Gazzaniga M, Heinze H. 2002. Selective activation of
a parietofrontal circuit during implicitly imagined prehension. Neuroimage 17:1693–1704.
Kammers MP, Longo MR, Tsakiris M, Dijkerman HC, Haggard P. 2009. Specificity and coherence
of body representations. Perception 38:1804–1820.
Kosslyn SM, DiGirolamo GJ, Thompson WL, Alpert NM. 1998. Mental rotation of objects ver-
sus hands: neural mechanisms revealed by positron emission tomography. Psychophysiology
35:151–161.
Kosslyn SM, Ganis G, Thompson WL. 2001. Neural foundations of imagery. Nat Rev Neurosci
2:635–642.
Longo MR, Haggard P. 2010. An implicit body representation underlying human position sense.
Proc Natl Acad Sci U S A 107:11727–11732.
Lotze M, Montoya P, Erb M, Hulsmann E, Flor H, Klose U, Birbaumer N, Grodd W. 1999. Activation
of cortical and cerebellar motor areas during executed and imagined hand movements: an fMRI
study. J Cogn Neurosci 11:491–501.
Luria AR. 1966. Higher cortical functions in man. London: Tavistock.
McCormick K, Zalucki N, Hudson M, Moseley GL. 2007. Faulty proprioceptive information
disrupts motor imagery: an experimental study. Aust J Physiother 53:41–45.
McGuire LM, Sabes PN. 2009. Sensory transformations and the use of multiple reference frames
for reach planning. Nat Neurosci 12:1056–1061.
Moseley GL. 2004a. Graded motor imagery is effective for long-standing complex regional pain
syndrome: a randomised controlled trial. Pain 108:192–198.
Moseley GL. 2004b. Why do people with complex regional pain syndrome take longer to recognize
their affected hand? Neurology 62:2182–2186.
Moulton ST, Kosslyn SM. 2009. Imagining predictions: mental imagery as mental emulation. Philos
Trans R Soc Lond, B, Biol Sci 364:1273–1280.
Mulder T, Zijlstra S, Zijlstra W, Hochstenbach J. 2004. The role of motor imagery in learning a
totally novel movement. Exp Brain Res 154:211–217.
Mulliken GH, Musallam S, Andersen RA. 2008. Forward estimation of movement state in posterior
parietal cortex. Proc Natl Acad Sci U S A 105:8170–8177.
Ní Choisdealbha Á, Brady N, Maguinness C. 2011. Differing roles for the dominant and non-
dominant hands in the hand laterality task. Exp Brain Res 211:73–85.
Parsons LM. 1987a. Imagined spatial transformation of one’s body. J Exp Psychol Gen 116:172–
191.
Parsons LM. 1987b. Imagined spatial transformations of one’s hands and feet. Cognit Psychol
19:178–241.
Parsons LM. 1994. Temporal and kinematic properties of motor behavior reflected in mentally
simulated action. J Exp Psychol Hum Percept Perform 20:709–730.
Parsons LM, Gabrieli JD, Phelps EA, Gazzaniga MS. 1998. Cerebrally lateralized mental
representations of hand shape and movement. J Neurosci 18:6539–6548.
Pavani F, Spence C, Driver J. 2000. Visual capture of touch: out-of-the-body experiences with
rubber gloves. Psychol Sci 11:353–359.
Pelgrims B, Michaux N, Olivier E, Andres M. 2011. Contribution of the primary motor cortex to
motor imagery: a subthreshold TMS study. Hum Brain Mapp 32:1471–1482.
Petkova VI, Ehrsson HH. 2008. If I were you: perceptual illusion of body swapping. PLoS ONE
3:e3832.
Ramachandran VS, Hirstein W. 1998. The perception of phantom limbs. The D. O. Hebb lecture.
Brain 121:1603–1630.
Reinersmann A, Haarmeyer GS, Blankenburg M, Frettlöh J, Krumova EK, Ocklenburg S, Maier
C. 2010. Left is where the L is right. Significantly delayed reaction time in limb laterality
recognition in both CRPS and phantom limb pain patients. Neurosci Lett 486:240–245.
Riemer M, Trojan J, Kleinböhl D, Hölzl R. 2010. Body posture affects tactile discrimination and
identification of fingers and hands. Exp Brain Res 206:47–57.
Saimpont A, Pozzo T, Papaxanthis C. 2009. Aging affects the mental rotation of left and right hands.
PLoS ONE 4:e6714.
Sauner D, Bestmann S, Siebner HR, Rothwell JC. 2006. No evidence for a substantial involvement
of primary motor hand area in handedness judgements: a transcranial magnetic stimulation
study. Eur J Neurosci 23:2215–2224.
Schwoebel J, Friedman R, Duda N, Coslett HB. 2001. Pain and the body schema: evidence for
peripheral effects on mental representations of movement. Brain 124:2098–2104.
Sekiyama K. 1982. Kinesthetic aspects of mental representations in the identification of left and
right hands. Percept Psychophys 32:89–95.
Sharma N, Jones PS, Carpenter TA, Baron JC. 2008. Mapping the involvement of BA 4a and 4p
during Motor Imagery. Neuroimage 41:92–99.
Shenton JT, Schwoebel J, Coslett HB. 2004. Mental motor imagery and the body schema: evidence
for proprioceptive dominance. Neurosci Lett 370:19–24.
Shepard R. 1978. The mental image. Am Psychol 33:125.
Sirigu A, Duhamel JR. 2001. Motor and visual imagery as two complementary but neurally
dissociable mental processes. J Cogn Neurosci 13:910–919.
Sirigu A, Cohen L, Duhamel JR, Pillon B, Dubois B, AgidY, Pierrot-Deseilligny C. 1995. Congruent
unilateral impairments for real and imagined hand movements. Neuroreport 6:997–1001.
Slater M, Perez-Marcos D, Ehrsson HH, Sanchez-Vives MV. 2009. Inducing illusory ownership of
a virtual body. Front Neurosci 3:214–220.
ter Horst AC, van Lier R, Steenbergen B. 2010. Mental rotation task of hands: differential influence
number of rotational axes. Exp Brain Res 203:347–354.
Tin C, Poon CS. 2005. Internal models in sensorimotor integration: perspectives from adaptive
control theory. Journal of Neural Engineering 2:S147–163.
Tomasino B, Toraldo A, Rumiati RI. 2003. Dissociation between the mental rotation of visual
images and motor images in unilateral brain-damaged patients. Brain Cogn 51:368–371.
van Elk M, Viswanathan S, van Schie HT, Bekkering H, Grafton ST. 2012. Pouring or chilling a
bottle of wine: an fMRI study on the prospective planning of object-directed actions. Exp Brain
Res 218:189–200.
Vargas CD, Olivier E, Craighero L, Fadiga L, Duhamel JR, Sirigu A. 2004. The Influence of
Hand Posture on Corticospinal Excitability during Motor Imagery: A Transcranial Magnetic
Stimulation Study. Cereb Cortex 14:1200–1206.
Viswanathan S, Fritz C, Grafton ST. 2012. Telling the Right Hand From the Left Hand Multisensory
Integration, Not Motor Imagery, Solves the Problem. Psychological Science 23:598–607.
Williams J, Anderson V, Reddihough DS, Reid SM, Vijayakumar N, Wilson PH. 2011. A com-
parison of motor imagery performance in children with spastic hemiplegia and developmental
coordination disorder. J Clin Exp Neuropsychol 33:273–282.
Zacks JM, Ollinger JM, Sheridan MA, Tversky B. 2002. A parametric study of mental spatial
transformations of bodies. Neuroimage 16:857–872.
Chapter 7
Use of the Uncontrolled Manifold (UCM)
Approach to Understand Motor Variability,
Motor Equivalence, and Self-motion
John P. Scholz and Gregor Schöner
7.1 Characterizing Variability in Motor Performance
Variability of motor output often has been considered a form of noise that interferes
with reliable performance. This assumption, however, depends on the level of the
motor system under consideration. For targeting tasks, variability of the end-effector
position will affect the consistency of targeting, depending on the task requirements
and the size of the target. Variability of coordination patterns used in artistic perfor-
mance may impact the aesthetics of performance. However, variability at the level of
the motor elements, including small variations in coordination patterns, often reflect
task flexibility that is only possible when the motor system exhibits sufficient mo-
tor abundance (Latash 2012). Consider, for example, performing cardiopulmonary
resuscitation (CPR) on an infant, with the index and middle fingers exerting the
necessary force to produce adequate chest compression. If fluctuation in the com-
pression force of one finger leads to a tendency for higher total force output, then
reduction in the forces exerted by the other fingers is necessary to compensate and
maintain a consistent total compression force. This is only possible because there
are two fingers contributing to a single total force output. Note that an alternative ap-
proach would be to attempt to control precisely, to the extent possible, the variability
of individual finger forces such that each finger generates approximately the same
force on each repetition. The presence of compensatory finger forces, however, has
been well documented (Latash et al. 2001, 2002a, 2002b), and is consistent with the
notion of a functional synergy among the motor elements (Latash et al. 2007).
Determining whether variability of motor output reflects motor noise or flexible
motor patterns is not trivial, however. Bernstein observed that when blacksmiths hit
G. Schöner ()
Institut für Neuroinformatik, Ruhr-Universität Bochum, Bochum, Germany
J. P. Scholz
Department of Physical Therapy and Biomechanics and Movement Sciences Program,
University of Delaware, Newark, DE 19716, USA

and Biology 826, DOI 10.1007/978-1-4939-1338-1_7
92 J. P. Scholz and G. Schöner
the chisel with the hammer there appeared to be more variability of the trajectories
of individual joints than there was for the trajectory of the hammer (Bernstein 1967).
This led him to conclude that a movement is never repeated in exactly the same
manner. Although his intuition was correct, there was no way to clearly establish
this fact. For one thing, the trajectory of the joints and the end effector (hammer) are
measured in different units and the number of degrees of freedom of each is quite
different. How does one compare variability of up to 10 joint motions (including
scapular motion) measured in radians to the variability of three dimensions of end-
effector motion measured in meters? Schöner and Scholz (Schöner 1995; Scholz and
Schoner 1999) developed the uncontrolled manifold (UCM) approach to overcome
this problem and to quantify statistically the extent to which variability of motor
elements tends to lead to noise or error in performance versus reflecting the use of
flexible patterns of coordination. In this approach, all analysis to answer this question
is performed at the level of the motor components, e.g., joint motions, finger forces,
and muscle modes.
To accomplish this, the UCM approach requires a model that relates how changes
in elemental variables affect the task level (e.g., hand position in space, total force
output). This model can be obtained formally, as when relating joint motions to
movement of the hand in space (e.g., x = l1 cos θ1 + cos (θ1 + θ2 ) + · · · ) or via
regression analysis when a formal model is not readily available or excessively com-
plicated (Freitas and Scholz 2010). The null space of the equation relating the task
space to the space of motor elements provides a linear estimate of all combinations
of the motor elements that do not affect the value of the task variable at that point
in a movement trajectory or in time (e.g., J (θmean ) θi = 0 where J is the Jaco-
bian matrix of partial derivatives relating small changes in the elemental variables
to changes in the task variable). The null space is computed around the mean value
of the motor elements (θmean ) at each point in the movement trajectory. Experimen-
tally measured mean-free values of the motor elements (θi = θi − θmean ) for each
movement repetition at a given point in the normalized movement are projected into
the null space and its compliment, or range space (the subspace of motor elements
in which different combinations of the motor elements lead to different values of
the task variable of interest). This is done for each repetition and the variance of
the projection lengths is then computed and normalized to the dimensions of the
subspace to make the analysis more conservative. Greater variance in the null space
or UCM subspace than in the range space suggests a control strategy in which the
central nervous system provides stabilizing control signals that restrict variations of
the motor elements when they affect the desired value of the task variable but allows
for some degree of variability in combinations of those variables if they have no
effect on the task variable (i.e., variations within the UCM).
Studies of many different motor tasks have shown that variability at the level of
motor elements is more consistent with the use of flexible combinations of those
elements that preserve a stable state (e.g., posture) or produce a consistent trajectory
(e.g., reaching) of a task-level variable (e.g., center of mass position, trajectory of
the hand). In contrast, range space variability typically is shown to be significantly
smaller (Scholz and Schoner 1999; Scholz et al. 2000; Latash et al. 2001; Scholz
7 Use of the Uncontrolled Manifold (UCM) . . . 93
et al. 2001, 2002; Krishnamoorthy et al. 2003, 2007; Latash et al. 2002a). UCM
analysis also has been able to differentiate between movement synergies in persons
with neurological dysfunction and healthy control subjects (Reisman and Scholz
2006; Park et al. 2012, 2013). In addition, different hypothesized task variables
can be evaluated with this approach to help determine what variables are of greatest
importance to task performance (Scholz et al. 2000). The results are in agreement
with the minimum intervention principle, which suggests that allowing variability
in redundant (abundant) dimensions is the optimal control strategy in the face of
uncertainty (Todorov and Jordan 2003), although a control structure more in line
with the UCM hypothesis than optimal control has been shown to better account for
detailed characteristics of movement trajectories (Martin et al. 2009).
The dependence of the outcome of a UCM analysis of motor variability on the
variables used to describe the effector system has recently been criticized (Sternad
et al. 2010). For example, the authors suggest that movements may equally well be
planned in joint or segment angle coordinates. They provide an example of a mini-
mally redundant effector system in which the UCM method leads to quite different
results when either of these sets of variables is used. This criticism is relevant and
emphasizes that researchers must make considered choices of the variables used for
analysis. Three clarifications are in order, however:
1. A choice of variables fixes the space in which configurations of the effector are
described. If we choose joint angles as variables, then we describe the effector
in joint space. A point in that space represents one particular configuration of
the effector. A metric must be fixed as well, which assesses the distance between
any two points in that space, that is, between two configurations of the effector.
Typically, the Euclidian metric is used, in which the squared distances along each
of a set of orthogonal coordinate axes are summed and the square root is taken.
The UCM approach is then actually invariant under any change of coordinate
frame that leaves the metric invariant (Schöner and Scholz 2007). This includes,
in particular, rotations of the coordinate frame. This invariance reflects the fact that
the UCM analysis is based on a geometrical view of variance, in which the shape
of the cloud of points in joint space is observed across trials at a particular point
during a movement. If that shape is elongated along the direction of the null space,
then the UCM hypothesis is confirmed. The shape of the cloud of points is invariant
under any coordinate transform that preserves the metric of the space spanned
by the chosen variables. This is useful in some cases, such as for the shoulder
joint, for which there is no principled way to select a particular coordinate frame
to represent the three degrees of freedom that reside in that joint. Any orthogonal
set of coordinate axis is equally meaningful, so that invariance under rotation of
a coordinate frame is desirable. The randomization method (Müller and Sternad
2004), in contrast, is essentially a form of nonlinear, multivariate correlation. In
that approach, the coordinate frames matter as they do for correlation. If the cloud
of points is elongated along a coordinate axis, for instance, then that shape is not
picked up as correlation but as inherent variability of that particular degree of
freedom.
2. Transformations that do not leave the metric invariant matter in the analysis of
variance and this is how the choices of variables that set up the configuration
space come into play. For instance, representing joint configurations through
segment angles or through joint angles does not lead to the same shape of the
cloud of points. Mathematically, going from segment angles to joint angles is a
transformation that does not leave distances invariant: It is not a metric preserv-
ing transformation, unlike the rigid rotations that may be used to link different
orthogonal axes for joint angles anchored in the shoulder. This dependence on
the embedding space is shared, of course, by all approaches to the analysis of
multidimensional variance (e.g., Müller and Sternad 2004; Cusumano and Cesari
2006).
3. Fortunately, the choice of embedding space can be guided by what we know
about physiology. Specifically, the choice of joint over segment angles is not
arbitrary. Known sensory receptors provide information to the nervous system
about changes in joint angles (Grigg 1994). There are no known sensory receptors
signaling orientation of a limb segment in external space, although transformation
of sensory receptor information can be used to estimate limb orientation (Poppele
et al. 2001). Segment angles are inherently dependent on each other. For example,
a flexion of the ankle brought about by a signal sent to muscles that act on the
ankle joint leads to changes of all segment angles along the kinematic chain of the
upright body, including joints not linked to the ankle by any muscle. Similarly,
changing the segment angle that the humerus forms with an external reference
frame also changes the segment angles of the forearm and hand, without any
activity by muscles acting on these more distal joints. Thus, distal segment motion
in an open kinematic chain is not independent of proximal segment motion. This
is not the case with joint angles. Changing one joint angle does not necessarily
affect another joint angle, unless there are particular mechanisms that bring about
such dependence like multi-articular muscles or coordinated neural signals.
Thus, it seems to us that the best of two worlds is achieved by combining the ge-
ometrical view of the UCM approach, which is conceptually attractive, with the
analytical power of the correlational approach. This is now routinely done by re-
searchers who select a set of variables and a particular coordinate frame based on
substantive hypotheses. They can then use the surrogate data procedure of the cor-
relational approach to verify if the shape of the variance in the UCM analysis truly
comes from covariation among the variables identified as meaningful rather than
from inherent differences in variance among the different degrees of freedom (see,
for example, Yen and Chang 2009; Verrel et al. 2010).
7.2 Quantifying Motor Equivalence
More recently, the geometrical perspective of the UCM approach has been used
to address additional issues in motor control, namely, motor equivalence and self-
motion. The term motor equivalence has been used in a variety of ways, but is defined
here as a change in the configuration of motor elements after a perturbation that tends
to preserve the outcome of a task or the stability of a task-relevant variable. Kelso
and colleagues (Kelso et al. 1984) performed a seminal study of motor equivalence
in the context of the control of speech utterances. They found that adjustments in
the articulators were task-specific, dependent on the nonsense syllable that subjects
spoke when a perturbation was delivered to depress the jaw. Further evidence was
provided when naı̈ve subjects were unable to distinguish utterances performed during
perturbed and nonperturbed trials. However, in many cases, distinguishing between
adjustments of motor coordination that lead to disturbance of the task versus being
a reflection of motor equivalent adjustments to preserve the task is not trivial. For
example, when reaching to a target, a transient perturbation of a joint will lead to at
least some effect on the motion of the end effector. If the end effector still reaches
the target, this suggests that motor equivalence must be present. What if the hand,
however, hits the target but deviates from its position on nonperturbed trials or what
effect does the perturbation have on the path of the hand itself? Presuming that
there are readjustments in the joint configuration or muscle firing patterns due to the
perturbation, how can one determine the extent to which the adjustments account for
observed deviations in the hand path or whether more of the adjustments tend to act
to preserve the hand path? A modification of the UCM approach allows this question
to be addressed quantitatively. If one takes a set of nonperturbed trials, the null space
of the Jacobian matrix relating small changes in the motor elements to changes in a
hypothesized task variable (e.g., the hand position) can be computed using the mean
value of the motor elements across trials. This null space corresponds, as in UCM
variance analysis, to a linear estimation of all combinations of joint configurations
that lead to the same value of the task variable. Again, this analysis is performed at
each point in the normalized (to 100 %) movement.
One can then obtain the vector of the configuration of motor elements from a
perturbed
(pert) trial, subtract it from the mean of the nonperturbed (nonpert) trials
θpert − θ̄non-pert , and project this difference vector into the null space (UCM) and
range space of the nonperturbed trials. If the null space projection is significantly
larger than the range space projection, then this suggests that more of the adjustment
in the joint configuration due to the perturbation is motor equivalent, tending to pre-
serve the nonperturbed value of the relevant task variable. This approach, therefore,
provides a statistical method for determining the extent to which motor equivalence
is present.
In a collaborative study with Fay Horak and John Jeka, motor equivalence relative
to the position of the center of mass of the body was measured in persons standing
on a moveable force platform that was perturbed by different amplitudes, keeping
the velocity of perturbation constant (Scholz et al. 2007). An example of the results
obtained immediately after the transient perturbation are presented in Fig. 7.1. Note
that the projection into the UCM subspace or null space was larger, and significantly
so, than the projection into the range space, both computed based on the nonperturbed
trials. This difference increased with greater amplitudes of perturbation. Thus, most
of the change in the joint configuration as a result of the perturbation was motor
equivalent, tending to preserve the pre-perturbation position of the center of mass of
the body.
Fig. 7.1 Left: Participants stood upright when the support platform was abruptly moved by a
varied amount (perturbation amplitude). Right: Six degrees of freedom were included in a motor
equivalence analysis. The difference vector between joint configurations in perturbed trials from the
mean configuration in unperturbed trials was projected into the UCM subspace (solid line) and the
range spaced (dashed line). The plot shows the average length of these differences vectors across
perturbation trials per degree of freedom together with the SME (error bars) as functions of the
perturbation amplitude. The data are from Scholz et al. 2007 (Fig. 4 there)
7.2.1 Quantifying Self-Motion
UCM variance analysis indicates how fluctuations of the motor elements across
repetitions or, in the case of relatively steady state behavior, across time are structured.
For example, in upright posture, to what extent do joint fluctuations over time lead to
postural sway of the center of mass versus being coordinated to flexibly stabilize the
center of mass location? Nonetheless, fluctuations of the motor elements could be
relatively small or large, depending on the task. The amount of self-motion provides
an estimate of the magnitude of changes in the motor elements that lie in the UCM
subspace and, therefore, do not affect the value of a task variable and those that
change the value of the task variable. The concept of self-motion comes from the
robotics of redundant effectors and, in that context, refers to the magnitude of the
vector of time-dependent changes (e.g., velocity) in the motor elements that lies
in the null space or UCM subspace. In contrast, range space motion refers to that
component of this vector that lies in the range space, or actually moves the task
variable with reference to which the analysis is applied.
Consider reaching to a target with the hand. One might suspect that the most
efficient movement of the hand would occur if the joints were coordinated such that
their velocities were directed primarily toward that goal. If so, then the range space
velocity component of the joints, which actually moves the hand in space, would be
expected to be substantially larger than the self-motion or UCM component. Self-
motion might be advantageous in some circumstances, however. For example, when
carrying a relatively full glass of Guinness stout, if a fly lands on your elbow, you
Fig. 7.2 Reaching with ten degrees of freedom to a target was analyzed with respect to self-motion,
using a Jacobian that took into account both the 3D end-effector position and the 3D orientation
of the hand for a total of six degrees of freedom in task space. Mean self-motion and range space
motion per degree of freedom averaged across participants is shown for the early, middle, and late
phase of reaching when reaching at a fast speed to a target. Data are from Scholz et al. (2011;
compare to Fig. 3 there, which includes other conditions as well)
might want to flick the elbow to get it to go away without spilling the precious
stout! Internal motion of the arm joints that does not affect the hand position is, by
definition, self-motion. But how much self-motion is present for a given task when
motion to achieve a secondary task is not required?
To answer this question, we developed a method of analysis that can be carried out
on individual trials or on means across trials. The method was adapted from the UCM
approach to the analysis of variance. The Jacobian, J, that relates the joint velocity
vector, θ̇ , to the velocity of the hand, ẋ = J (θ )θ̇ , is computed from a geometrical
model of the effector. It is evaluated at the instantaneous joint configuration, θ , at
each point in time. Based on this Jacobian, the null (UCM) and range spaces can
be determined. The joint velocity vector, θ̇ , is computed from the time series by
numerical differentiation and is projected into either subspace. Finally, the length
of these projections is computed and divided by the number of dimensions of each
subspace.
Figure 7.2 presents an example of this analysis from a recent article investigat-
ing self-motion at different speeds of reaching (Scholz et al. 2011). The results for
reaching at a self-selected fast speed are depicted. Note that although range space
motion (component of the joint velocity vector projection that moves the hand in
space) generally was larger than self-motion, self-motion was nonetheless quite sub-
stantial. The results were similar at slow and moderate speeds of reaching, although
self-motion was not quite as large as compared to range space motion. Of interest
was that at all speeds of reaching, self-motion actually was larger than range space
motion at the early stage of reaching, probably because the arm had to be adjusted
to exit the trough in which it rested as the reach was initiated.
The results suggest that even during the performance of goal-oriented targeting
tasks there is a substantial amount of self-motion, i.e., joint configuration motion
within the UCM subspace of joint space, even when an obvious secondary task is not
involved. It is hypothesized that this amount of self-motion is a reflection of the nature
of the control system that minimally restricts combinations of motor elements for a
task that do not interfere with successful performance even when the configurations
deviate from the initially planned configuration.
7.3 Conclusion
The UCM approach was developed originally to investigate the role of motor vari-
ance and has been successfully applied to a variety of motor tasks from finger force
production to reaching to postural control. In most cases, variance consistent with
flexible combinations of the motor elements that maintain a consistent value of an
important task variable has been shown to be significantly greater than variance
leading to variability of the task variable. The approach allows one to test hypothe-
ses about the importance of different task-relevant variables based on the structure
of variance of the underlying motor elements and to evaluate how different motor
elements contribute to that structure.
Recently, the approach has been extended to address additional important issues
in motor control such as motor equivalence in the presence of a perturbation and
the extent to which motor abundance is used in the control of motor tasks through
self-motion analysis. A model of a control strategy based on the UCM hypothesis
and consistent with the recent results was developed by Martin, Scholz, and Schöner
(Martin et al. 2009) and applied to postural control by Reimann et al. (2011). We
believe that the geometrical perspective offered by UCM thinking will be useful both
to interpret experimental signatures of control hypotheses and to investigate possible
neural processes that bring about the coordination of the many degrees of freedom of
the motor system. Ultimately, we will need to understand how spatial information and
timing constraints for the motion of effectors in space can be translated to control
signals at the level of each muscle (Bullock et al. 1993; Butz et al. 2007). That
transformation sets up the geometry uncovered by the UCM method.
Acknowledgments Work reviewed here was supported by NINDS Grant R01-NS050880 to John
Scholz and DFG (Germany) Grant SCHO 336/7-1 to Gregor Schöner. Thanks to Dr. Hendrik
Reimann for a critical reading. A draft of this paper was completed by John Scholz shortly before
he passed away.
References
Bernstein N. 1967. The co-ordination and regulation of movements. Oxford: Pergamon Press.
Bullock D, Grossberg S, Guenther F. 1993. A self-organizing neural model of motor equivalence
reaching and tool use by a multijoint arm. J Cogn Neurosci 5:408–435.
Butz MV, Herbort O, Hoffmann J. 2007. Exploiting redundancy for flexible behavior: unsupervised
learning in a modular sensorimotor control architecture. Psychol Rev 114(4):1015–1046.
Cusumano JP, Cesari P. 2006. Body-goal variability mapping in an aiming task. Biol Cybern
94:367–379.
Freitas SMSF, Scholz JP. 2010. A comparison of methods for identifying the Jacobian for
uncontrolled manifold analysis. J Biomech 43:775–777.
Grigg P. 1994. Peripheral neural mechanisms in proprioception. J Sport Rehabil 3:2–17.
Kelso JA, Tuller B, Vatikiotis-Bateson E, Fowler CA. 1984. Functionally specific articulatory coop-
eration following jaw perturbations during speech: evidence for coordinative structures. J Exp
Psychol Hum Percept Perform 10(6):812–832.
Krishnamoorthy V, Latash ML, Scholz JP, Zatsiorsky VM. 2003. Muscle synergies during shifts of
the center of pressure by standing persons. Exp Brain Res 152(3):281.
Krishnamoorthy V, Scholz JP, Latash ML. 2007. The use of flexible arm muscle synergies to perform
an isometric stabilization task. Clin Neurophysiol 118(3):525–537.
Latash ML. 2012. The bliss of motor abundance. Exp Brain Res 217(1):1–5.
Latash ML, Scholz JF, Danion F, Schoner G. (2001). Structure of motor variability in marginally
redundant multifinger force production tasks. Exp Brain Res 141:153–165.
Latash ML, Scholz JF, Danion F, Schoner G. (2002a). Finger coordination during discrete and
oscillatory force production tasks. Exp Brain Res 146(4):419–432.
Latash ML, Scholz JF, Schoner G. (2002b). Motor control strategies revealed in the structure of
motor variability. Exerc Sport Sci Rev 30(1):26–31.
Latash ML, Scholz JF, Schoner G. 2007. Toward a new theory of motor synergies. Motor Control
11(3):276–308.
Martin V, Scholz JP, Schöner G. 2009. Redundancy, self-motion, and motor control. Neural Comp
21(5):1371–1414.
Müller H, Sternad D. 2004. Decomposition of variability in the execution of goal-oriented tasks:
Three components of skill improvement. J Exp Psychol Hum Percept Perform 30:212–233.
Park J, Wu Y-H, Lewis MM, Huang X, Latash ML. 2012. Changes in multi-finger interaction and
coordination in Parkinson’s disease. J Neurophysiol 108:915–924.
Park J, Lewis MM, Huang X, Latash ML. 2013. Effects of olivo-ponto-cerebellar atrophy (OPCA)
on finger interaction and coordination. Clin Neurophysiol 124:991–998.
Poppele RE, Bosco G, Rankin AM. 2001. Independent representations of limb axis length and
orientation in spinocerelbellar response components. J Neurophysiol 87:409–422.
Reimann H, Schöner G, Scholz JP. (2011). Visual information is sufficient for maintaining upright
stance – a multi-joint model of human posture. Program No. 184.05 2011 Neuroscience Meeting
Planner. Washington: Society for Neuroscience. Online
Reisman D, Scholz JP. 2006. Workspace location influences joint coordination during reaching in
post-stroke hemiparesis. Exp Brain Res 170:265–276.
Scholz JP, Schoner G. 1999. The uncontrolled manifold concept: identifying control variables for
a functional task. Exp Brain Res 126(3):289–306.
Scholz JP, Schöner G, Latash ML. 2000. Identifying the control structure of multijoint coordination
during pistol shooting. Exp Brain Res 135:382–404.
Scholz JP, Reisman D, Schöner G. 2001. Effects of varying task constraints on solutions to joint
coordination in a sit-to-stand task. Exp Brain Res 141(4):485.
Scholz JP, Danion F, Latash ML, Schoner G. 2002. Understanding finger coordination through
analysis of the structure of force variability. Biol Cybern, 86(1):29–39.
Scholz JP, Schöner G, Hsu WL, Jeka JJ, Horak FB, Martin V. 2007. Motor equivalent control of the
center of mass in response to support surface perturbations. Exp Brain Res 180:163–179.
Scholz JP, Dwight-Higgin T, Lynch JE, Tseng Y, Martin V, Schöner G. 2011. Motor equivalence
and self-motion induced by different movement speeds. Exp Brain Res 209:319–332.
Schöner G. 1995. Recent developments and problems in human movement science and their
conceptual implications. Ecol Psych 7(4):291–314.
Schöner G, Scholz JP. 2007. Analyzing variance in multi-degree-of-freedom movements: uncover-
ing structure versus extracting correlations. Motor Control 11:259–275.
Sternad D, Park S-W, Müller H, Hogan N. 2010. Coordinate dependence of variability analysis.
PLoS Comp Biol 6(4 (e1000751)), 1–16.
Todorov E, Jordan MI. 2003. A minimal intervention principle for coordinated movement. In T.
Becker, Obermayer (Ed.), Advances in Neural Information Processing (Vol. 15, pp. 27–34).
Boston: MIT Press.
Verrel J, Lövdén M, Lindenberger U. 2010. Motor-equivalent covariation stabilizes step parameters
and center of mass position during treadmill walking. Exp Brain Res 207(1–2):13–26.
Yen JT, Chang Y-H. 2009. Rate-dependent control strategies stabilize limb forces during human
locomotion. J R Soc Interface 7(46):801–810.
Chapter 8
Acquisition of Novel and Complex Motor Skills:
Stable Solutions Where Intrinsic Noise
Matters Less
Dagmar Sternad, Meghan E. Huber and Nikita Kuznetsov
8.1 Introduction: Acquisition of Skill and Experimental Control
While professional quarterbacks like Peyton Manning make passing a football look
effortless, it takes a lifetime of disciplined practice for these elite athletes to reach
such a high level of skill. The principal responsibility of the quarterback in American
football is to throw a forward pass to his teammate. This skill, however, is not as
trivial as it may appear. To achieve a successful pass, he must accurately throw to
a target that is moving, while he himself is in motion. But even before the pass is
thrown, he must quickly decide which receiver will have the best chance of catching
the ball, judging from the continuous movements of the defensive team. He must
also choose what style of throw and what initial conditions (i.e., position, velocity)
are appropriate for where he aims the ball. This decision-making process cannot last
more than a fraction of a second or he risks being tackled by opponents twice his
size. This ability to choose from a vast—in fact infinite—set of actions is an integral
part of many sport skills. Such complexity is what makes events like the Olympic
Games fascinating, not only to spectators around the globe, but also to movement
scientists, who want to unravel how humans achieve coordinated actions. How does
the brain acquire and control such complex spatiotemporal skills?
It is somewhat sobering to then look into the laboratories and see how motor skill
is typically studied. The vast majority of experiments have focused on highly sim-
plified tasks, such as finger-to-thumb opposition (Karni et al. 1998), finger tapping
(Wing and Kristofferson 1973), sequenced isometric contractions (Shim et al. 2005;
D. Sternad ()
Departments of Biology, Electrical and Computer Engineering, Bioengineering, and Physics,
Northeastern University, Boston, MA 02115, USA
M. E. Huber
Department of Bioengineering, Northeastern University, Boston, MA 02115, USA
N. Kuznetsov
Department of Biology, Northeastern University, Boston, MA 02115, USA

and Biology 826, DOI 10.1007/978-1-4939-1338-1_8
102 D. Sternad et al.
Waters-Metenier et al. 2014; Wu et al. 2012), or reaching and pointing (Landi et al.
2011; Shabbott and Sainburg 2010; Shadmehr and Mussa-Ivaldi 1994). Unfortu-
nately, the requirements of experimental control, measurement, and quantification
have stripped motor skills of most aspects that have made them interesting in the first
place. Simplified tasks permit precise assessment of performance variables, show rel-
atively consistent patterns across subjects, and thereby permit scientifically stringent
testing of hypotheses. Such focal movements are also well suited for neuroimaging
studies, such as fMRI, where the subject’s head must be still and more extensive
movements distort the magnetic field. Yet, all these benefits come with the risk of
oversimplification of task demands, to the point that findings from these tasks may no
longer generalize to more complex and realistic scenarios. These simple movements
shed light on fundamental neural underpinnings, but they show little or no improve-
ments with practice, as the movements are already part of the behavioral repertoire
of the participants. How can learning and neuroplasticity be studied to look closer at
the spectacular agility and complexity of movements in real life?
Motor Adaptation One line of research that has been extremely successful and gone
beyond the simple movement to study learning and neuroplasticity is the experi-
mental paradigm on reaching with adaptation to external perturbations. Following a
seminal study on adaptation to a force field by Shadmehr and Mussa-Ivaldi (1994),
a prolific line of research ensued that addressed questions on adaptation and gen-
eralization (Criscimagna-Hemminger et al. 2003; Shadmehr and Moussavi 2000),
consolidation (Baraduc et al. 2004), and trial-by-trial learning (Donchin et al. 2003;
Thoroughman and Shadmehr 2000), to name only a few research directions. The
paradigm employed simple horizontal reaching movements that were perturbed via
an external complex force field or a prismatic distortion of its path. It is important to
point out that subjects initially perform straight-line reaches, as is intrinsic to their
behavioral repertoire (Morasso 1981).When the perturbations were applied, subjects
committed large deviations from the direct path, but soon learned to adapt and com-
pensate, and their performance returned to the initial straight path. When the force
field or visual distortion was subsequently removed, performance temporarily wors-
ened, as the compensation to the perturbation persisted. These so-called aftereffects
indicated that the nervous system has learned something. In the absence of further
perturbations, however, the return to the level prior to the experiment was relatively
fast and persistent.
Can we consider these results to reveal mechanisms underlying the acquisition of
skill? In our thinking, the answer is no, because at the end of the experiment, there is
no net change in skill (Fig. 8.1a). Even though there are aftereffects and even though
in a second encounter with the force field, adaptation may proceed faster (reflecting
“savings”), these compensatory processes only reestablish, or maintain, previous
behavior. The subject has not acquired any truly novel skill that she did not know
before.
Skill Acquisition Compare this motor adaptation to the scenario of learning how to
throw a Frisbee. Here, the learner attempts a novel combination of arm, hand, and
body movements, with a novel object that has novel flight properties that he or she
8 Acquisition of Novel and Complex Motor Skills 103
a MOTOR ADAPTATION b SKILL ACQUISITION
Error
Error
ΔSkill
ΔSkill=0
0 0
Pre Perturbation Retention Pre Practice Retention
Fig. 8.1 a Motor adaptation paradigm. The participant starts with a highly learned movement, such
as reaching, which is performed with zero error in the pre-stage (red). A perturbation is applied
during a block of trials (yellow), until the participant adapts and reestablishes the initial performance.
When the perturbation is removed, there is a short aftereffect, indicating that something has been
learned. However, the original movements are resurrected quickly (blue); thus, there is no net
change in skill. b Skill acquisition. In the pre-stage, the participant performs a novel motor skill
with high error (red). The participant decreases error over practice (yellow). Error in the retention
stage (blue) is less than the error in the pre-stage (red), reflecting the acquisition of the novel skill
has never encountered before. With practice, he or she learns to release the Frisbee to
lead to an accurate return, reflecting increased proficiency or skill (Fig. 8.1b). Note
that there is no imposed perturbation. The skill is new and persists after training,
indicating retention. There is much anecdotal and some experimental evidence that
once a motor skill is learned, it persists for a long time over years of no practice
(Draganski et al. 2004; Nourrit-Lucas et al. 2013; Park et al. 2013; Swift 1910).
While in motor adaptation there are also savings, it remains to be tested whether
such savings persist for a similarly long time. If not, this would suggest that the
temporary adaptation processes involve different forms of neuroplasticity than the
acquisition of a novel skill. Hence, we conclude that the first desideratum for studying
skill acquisition is that the movement task poses a novel challenge to the neuromotor
system.
Intrinsic and Extrinsic Redundancy However, novel skills such as throwing a
Frisbee are complex and difficult to study in the laboratory without at least some
experimental reduction. Therefore the question is: What are the minimal require-
ments for an experimental task to still allow for the study of skill, but in a controlled
and hypothesis-driven fashion? We propose that the essence of skill is redundancy.
As exemplified by the quarterback’s forward pass, a skill should allow for multiple
solutions to achieve the task. Unlike in adapting a reaching movement in an exter-
nal force field to restore the initial straight path, subjects should be able to explore
different solutions to find the best one with practice. Such solutions may also differ
between individuals.
Two types of redundancy can be distinguished: intrinsic to the organism and
extrinsic in the task. The role of intrinsic redundancy has been widely recognized in
motor control. Our complex neuromuscular system is hierarchical with an infinite
number of degrees of freedom at any one level. The most frequently used example for
redundancy is at the level of joints: Pointing to an object with a multi-joint arm can
be achieved with an infinite number of joint angle configurations (Bernstein 1967).
Equivalently, at the level of muscles, multiple muscle contraction patterns may lead
to the same kinematic pattern (d’Avella et al. 2003; Ting 2007). This excess of
degrees of freedom for any movement goal repeats itself at the level of motor units
and further. However, the task remains unique: The finger needs to point to one
location in space.
Extrinsic redundancy, in contrast, would allow pointing to a whole region in
space. For example, grasping and bringing a cup of coffee can be achieved using
multiple approach paths, multiple finger contact points, and multiple contact points
with the mouth. Yet, the task is only successful if the person brings the cup to the
mouth, drinks, and does not spill the coffee. There are infinitely many combinations
of execution that lead to success with zero error. This kind of task redundancy has
received less attention in motor learning than the intrinsic redundancy. Returning to
the quarterback example, not only can different throwing styles achieve the same
target but also different throwing releases may lead to the same target. Extrinsic, or
task redundancy allows the actor to “decide” what combination of execution-relevant
variables to use in pursuing the task goal. Note that “decision” does not always imply
a conscious weighing of alternatives with high-level cortical involvement (Araujo
et al. 2006). Rather, the situation forces the quarterback to make extremely fast
decisions, which may happen at much lower levels of the neural hierarchy. Hence,
the second desideratum for studying skill and skill acquisition is that the task should
include redundancy to allow for choices between many solutions that achieve a given
task goal.
How can we capture this redundancy and study the acquisition of novel skills in
the laboratory without losing experimental control and precise measurement? Unlike
several decades ago, measurement of complex movements is no longer the problem,
as both technology and analysis methods have significantly advanced. New camera
systems and wearable sensors have outdated tedious digitizing of video recordings.
In addition, virtual technology and programmable robotic devices have created a
whole new arena for experimental manipulations and measurement. However, simply
measuring trajectories of Peyton Manning’s throwing arm, even in truly ecological
settings, does not tell us much about his skilled performance. Without experimental
control and hypotheses, the multitude of data simply redescribe the performance.
A hypothesis-driven approach to skill requires some simplification—but without
“throwing the baby out with the bathwater.”
8.2 A Methodological Approach to Understanding

Skill Acquisition
We developed an approach using mathematical modeling and virtual technology to

study the acquisition of novel skills with extrinsic redundancy. In the following, we
first describe our methodological steps and then present three examples following
this approach: throwing a ball to a target, rhythmically bouncing a ball, and carrying
a cup of coffee:
1. Select a skill: The first step is to choose a skill that represents some core aspects
germane to many other tasks. In throwing, for example, the timing of release is one
such core aspect. Carrying a cup filled with coffee exemplifies manual interaction
with a complex object. Next, it is essential that the experimental task has a well-
defined goal and yet allows for a variety of solutions to achieve this goal: Task
redundancy should be center stage. For example, the quarterback throwing a ball
to a receiver can use a variety of strategies to throw to the intended wide receiver.
Further, the task should be novel and sufficiently challenging to require practice to
achieve success. Improvement should be visible within one or few experimental
session(s), but also require fine-tuning at a longer time scale.
2. Mathematically model the task: The next step is to model the relevant physics that
govern the task. A model is a useful simplification of the real phenomenon because
it formalizes the assumptions about the task and prunes away the irrelevant aspects
of the real-life task. What is the simplest system that captures the challenge to
manipulate a dynamically complex object? What exactly determines timing of
release? One core element in the modeling stage is to distinguish between the
execution variables and the result variables: The result variable(s) are defined
by the task goal and capture the quality of performance; this is typically an
error measure. Execution variables are under control of the performer and fully
determine the task result. The functional relationship between execution and result
is the essence of the model.
3. Formulate hypotheses: Based on the mathematical model, the space of all possible
solutions to the task can be derived. As the model system is typically nonlinear,
the space of solutions, or result space, is complex and reveals additional properties
of solutions, such as risk or dynamic stability, as we explain below. Depending on
the model, different mathematical tools can be used to derive predictions about
stability or robustness to perturbations. Importantly, exact quantitative hypotheses
can be formulated about which solutions have the greatest probability of success.
4. Render the model task in a virtual environment: Based on the explicit mathe-
matical understanding, the task can be rendered in a virtual environment. The
execution variables are the ones that the subject controls via an interface with
the virtual system. The result variables can be precisely measured in the virtual
world. For example, while the subject performs a throwing task, the arm trajec-
tory is real and controls the ball release, but the ball and target are virtual. The
virtual rendering has the advantage that it confines the task to exactly the variables
and parameters that were analyzed. There are no uncontrolled aspects as would
occur in the real world. Further, the execution and the result variables can be
manipulated at will to test hypotheses about the performance strategies.
5. Measure subjects’ performance and test hypotheses: Subjects interact with the
virtual physics of the task via a manipulandum. The variables measured from the
subjects’ movements should correspond to the execution variables in the model.
The measured execution variables and the task result are then evaluated against
the space of all solutions. Hypotheses about solutions derived from the model can
be evaluated.
8.3 Hypothesis: Exploit Redundancy and Seek Stable Solutions

that Make Intrinsic Noise Matter Less
Before detailing this approach, it is necessary to be more specific about the overall
research aim. Given the intrinsic and extrinsic redundancy of the task and performer,
we pursued the hypothesis that individuals improve their performance by seeking
solutions that are robust with respect to perturbations to make their intrinsic noise
matter less. Due to the intrinsic redundancy of the hierarchical and complex neu-
romuscular system, subjects always show variability or intrinsic noise, even when
they want to repeat the same movement under fixed external conditions (Faisal et al.
2008). In the presence of extrinsic redundancy with infinite equivalent task solu-
tions, we hypothesize that the neuromotor system chooses those solutions that are
most tolerant to this ever-present noise. As highlighted above, tasks with redundancy
provide options that subjects can explore and exploit. This overall hypothesis will
be operationalized for each task on the basis of the model. The following review
exemplifies how we pursued this approach in three different paradigms.
8.3.1 Paradigm 1: Skittles, a Discrete Throwing Skill
The Skill This experimental paradigm was motivated by the British pub game “skit-
tles,” which is a table version of the American game “tetherball.” The actor throws
a ball that is tethered to a post by a string like a pendulum; the goal is to hit a target
skittle on the opposite side of the pole (Fig. 8.2a). Accurate throwing requires a
controlled hand trajectory that prepares the ball release at exactly the right position
and velocity that sends the ball onto a trajectory that hits the target skittle. The task
has redundancy as elaborated below and timing is one essential element of this skill.
The Model To simplify the task, the movement of the ball was confined to two hori-
zontal dimensions, eliminating the elevation due to the pendular excursion (Fig. 8.2b).
In the model, the ball is attached to two orthogonal, massless springs with its rest
position at the center post. To execute a throw, the ball is deflected from its rest
position by the hand—a virtual lever arm that moves in correspondence to the real
arm movements of the participant (Fig. 8.2b). Upon release, the ball traverses an
elliptic path generated by the restoring forces of the two springs (Müller and Sternad
2004b). The equation for ball position in the x- and y-directions at time t is:
⎛ ⎞ ⎛ ⎞ ⎛ ⎞⎛ ⎞
x(t) xp cos φr − sin φr l cos ωt
⎝ ⎠ = ⎝ ⎠ cos ωt + ⎝ ⎠⎝ ⎠.
y(t) yp − sin φr cos φr vr /ωt
The frequency ω denotes the natural frequency of the springs, and (xp , yp ) denotes
the pivot point and l the length of the arm. For a given throw, two execution variables,
angle φr and velocity vr , of the virtual hand at the moment of release fully determine
the ball trajectory.
a b
Target
Error
Center Post
Ball
2D Model
c d
800 800 6
Release Velocity (°/s)
400 Release Velocity (°/s) 5

600
4
Error (cm)
0
400 3
800
2
400 200
1
Day 1
0 0 Day 11 0
0 90 180 0 90 180 0 30 60 90 120
Release Angle (°) Release Angle (°)
Fig. 8.2 a Real-life skittles game. b Two-dimensional mechanical model of skittles. The ball (white
circle) is attached to two orthogonal springs centered at the origin of the center post (red circle). The
ball trajectory (dotted line) upon release is determined by the springs and depends on two execution
variables: release angle φr and velocity vr of the ball. Error is defined as the minimum distance
between the ball trajectory and the target center (yellow circle). c Execution space for four different
target locations. Error is calculated for each point in execution space and indicated in shades of
gray; the lighter shades denote lower errors. The black line represents the one-dimensional solution
manifold. d Execution space with nonlinear solution manifold with two data sets representing 20
throws collected on the first day (blue) and after 11 days of practice (yellow)
Similar to the real skill, the actor’s goal is to throw the ball to hit the target
skittle, without hitting the center post. The location of the target is defined in (x, y)-
coordinates; the error is defined as the minimum distance between the ball trajectory
and the center of the target (Fig. 8.2b). Thus, the result variable is the error and
is fully determined by the two execution variables. Importantly, there is more than
one combination of angle φr and velocity vr that leads to zero error. Figure 8.2c
illustrates this functional relationship: For each point in execution space, spanned
by release angle and velocity, error is depicted by shades of gray. Lighter shades
indicate smaller errors; black signifies that releases hit the center post, which is a
penalty area in the experiment. The thin black lines denote the one-dimensional set of
zero-error solutions or solution manifold. The solution manifold can be analytically
derived and is defined by the following equation:

vr (−l sin φr − yp )xt + (l cos φr + xp )yt
= .
ω (l + cos φr xp + sin φr yp )2 − ( cos φr xt + sin φr yt )2
An interesting feature of this task is that the result space and the solution manifold
depend on the target location in ways that would be hard to predict without the
model (Fig. 8.2c). For example, for target position at x = 0.6 m and y = 1.5 m, the
result space is nonlinear and U-shaped, as shown in the top left example (Cohen and
Sternad 2009). For the target position x = 0.05 m, y = 1.055 m, the solution manifold
is a vertical line, such that only the angle of release determines the error, as shown in
the top right example (Sternad et al. 2011). In the two examples in the lower panels,
the solution manifolds terminate on the penalty area, which makes those solutions
risky. Importantly, the difficulty of the task is determined by the shape of the solution
manifold and can vary significantly for different target locations.
Hypotheses In principle, subjects can release the ball with any of the position–
velocity combinations on the solution manifold. However, Fig. 8.2d illustrates that
this is not what subjects do: The two data sets represent 20 throws collected on
the 1st day (blue) and after 11 days of practice (yellow). Each data point on this
figure corresponds to a throw and, as to be expected, the data show significant
scatter, especially in the beginning. The two distributions also show that the subject
improved, as the throws early in practice have visibly higher errors on average (in dark
colored area). The throws late in practice are aligned with the solution manifold, with
smaller errors. What drives this change? What solutions do humans prefer and why?
The hypothesis is that individuals improve their performance by seeking solutions
that are robust with respect to perturbations to make their intrinsic noise matter less.
The sample data illustrate how three conceptually different routes to performance
improvement can be distinguished: (1) Subjects shift their solutions in execution
space to those regions on the solution manifold where small changes in the angle and
velocity of release do not lead to large decreases in the magnitude of error. Wider
bands of light gray on the right branch indicate that deviations from the solution
manifold do not incur large errors. We refer to this as error tolerance. (2) Subjects
reduce the amplitude of their dispersion. We refer to this as noise reduction. (3)
Subjects rotate or reshape their distribution to align with the solution manifold. We
refer to this as covariation. All three routes channel variability to make their intrinsic
noise matter less for the quality of the performance. The specific hypothesis is that
learners first improve their performance by shifting their throws to error-tolerant
solutions in result space. This first stage includes exploration of options. The second
stage involves rotating, or covarying, the data with respect to the solution manifold.
Having exploited these two routes, the final stage is to reduce the dispersion. This
latter route corresponds to reducing intrinsic noise.
Quantitative measures of tolerance, noise, and covariation (TNC) were first
introduced by Müller and Sternad (2003, 2004a, 2004b, 2009) and subsequently re-
formulated as cost measures by Cohen and Sternad (2009). The TNC-cost measures
a b
4
Switch 3
TNC-Cost (cm)
Poteniometer
PC 0
1 10 20 30 40
Blocks of Practice
Fig. 8.3 a Participant interacting with the virtual environment in the skittles setup. The center post
(red) and target (yellow) and the movements of the lever arm (magenta) and the ball (white) are
presented to subjects on the rear projection screen in front of them. b Mean T -cost (red), N-cost
(green), and C-cost (blue) for three expert participants over 45 blocks (15 days) of practice. Error
bars show standard errors across participants
assess to what degree the result of a set of data could be improved, if one of these
components were optimized. The numerical analysis method was detailed in Co-
hen and Sternad (2009). Note the similarity with the uncontrolled manifold (UCM)
method (Scholz and Schöner 1999), but Müller and Sternad (2009) also outlined
some key conceptual differences between the methods. One critical methodological
issue is that the UCM method, as one of the many covariance-based methods, is
more prone to coordinate transformations, although the TNC method, in particular
the covariation cost, is not immune to this problem either (Sternad et al. 2010).
Virtual Environment To test the hypothesis, the mechanical model was rendered in
a virtual setup, where subjects interacted with the model via a manipulandum. In the
experiment, the actor grasped a ball attached to the distal end of a manipulandum
and performed a forearm rotation to throw a virtual ball to a target in the virtual
environment (Fig. 8.3a). Initially, the subject closed a contact switch with her index
finger; lifting the finger opened the switch and triggered the release of the virtual ball.
A potentiometer continuously sampled the angular position of the manipulandum,
and the movements of the arm were mapped onto the displayed virtual arm; angular
velocity of the arm was estimated online. The ball traversed around the center post
as determined by the angular position and angular velocity at the moment of release
(Fig. 8.3a).
Experimental Findings Cohen and Sternad (2009) tested the hypothesis that learners
improve their performance by first locating error-tolerant solutions in result space and
then by channeling and reducing their intrinsic noise. In their study, nine participants
with average and three participants with extensive throwing experience practiced
three blocks of 60 throws each day for 6 and 15 days, respectively. Each block of
a b
800 6 25 25
Time in Hit Zone (ms)

Release Velocity (°/s)
Timing Error (ms)

5
600 20 20
4
Error (cm)
400 3 15 15
2
200 10 10
Day 1 1
0 Day 11 0 5 5
0 30 60 90 120 1 5 10 15
Release Angle (°) Days of Practice
Fig. 8.4 a Execution space with two sets of arm trajectories representing 20 trials collected on the
first day (blue) and after 11 days of practice (yellow). b Mean timing error (green) and time in hit
zone for three expert participants over 15 days of practice. Error bars show standard errors across
participants
60 throws was analyzed as a distribution and the three components tolerance, noise,
and covariation were quantified: Each data set was translated to obtain T-cost, scaled
for N-cost, or oriented via recombination for C-cost to find the optimal mean result
for this component. Note that each transformation left all other aspects of the data
invariant. The cost was then defined as the difference between the mean result, error,
of the original data set and the optimally transformed data set.
The TNC-cost analysis showed that both novices and expert throwers revealed
similar patterns of change in T -, N-, and C-costs over practice (Cohen and Sternad
2009). Figure 8.3b shows all three costs for the expert throwers. T-cost (red) de-
creased very early in practice followed by a considerably slower decrease in C-cost
(blue). N-cost (green) decreased at the slowest rate and remained the highest cost to
performance. The overall rank ordering of these costs was similar across participants
and expertise levels. This pattern of results suggests that early in practice participants
tried out different performance strategies and searched for the most error-tolerant re-
gion of the solution manifold. They then improved covariation to better align with
the solution manifold. The stochastic Noise component was reduced only modestly,
suggesting that intrinsic noise is relatively inaccessible to practice.
How does the actor achieve these changes? Particularly, how can covariation
between position and velocity be achieved? Extending initial work by Müller and
Loosch (1999), Cohen and Sternad (2012) tested the hypothesis that subjects learn
to align their trajectory of the arm with the solution manifold. For this analysis, it is
interesting to recognize that the execution space, spanned by position and velocity, is
equivalent to state space. Hence, not only the release points but also the continuous
trajectory can be analyzed in this space (Fig. 8.4a). The solution manifold corresponds
to an ideal trajectory, which would achieve the desired result at every point. Evidently,
the arm trajectory cannot exactly follow the solution manifold, but a trajectory that
at least temporarily follows the solution manifold gives the subject a time window to
release the ball that achieves a successful hit. Cohen and Sternad (2012) hypothesized
that with practice subjects shape their trajectories to travel along the solution manifold
so that the ball can be released within this time window and achieve a successful
hit. This strategy exploits the redundancy of the task and ameliorates the effects of
intrinsic noise in timing.
The same participants’ data from the first study were analyzed. Each continuous
arm trajectory was converted to a trajectory of error over time, assuming that every
angle–velocity point on the trajectory was a release (its associated error was computed
using the model equations). To measure the shaping of the arm trajectory with respect
to the solution manifold, time in hit zone was defined as the time that the trajectory
spent below a given error threshold. Timing accuracy or timing error was defined as
the time difference between actual and ideal release time. The ideal release time was
the time when the arm trajectory crossed the solution manifold. Cohen and Sternad
(2012) found that with extended practice, performance improved due to continued
shaping of the trajectory (Fig. 8.4a). Interestingly, timing error decreased first but
then asymptoted at 9 ms at approximately Day 6, while time in hit zone continued to
improve (Fig. 8.4b). This suggests two interpretations: When timing error reached a
plateau, performance improvement could only be achieved by aligning the trajectory
with the solution manifold. Alternatively, subjects did not need to further improve
their timing accuracy, as their trajectory was in the hit zone and made any further
improvement in timing superfluous.
Interim Conclusions Analysis of the throwing skill exemplified our approach to skill
acquisition using a discrete task with redundancy. The results showed that redundancy
allows the performers to choose their performance strategy to minimize the poten-
tially detrimental effects of intrinsic noise on performance. These findings support the
overall hypothesis that when performing a new skill, humans exploit the redundancy
of the task and find the error-tolerant regions to enhance their performance.
8.3.2 Paradigm 2: Rhythmically Bouncing a Ball: Humans

Exploit Dynamically Stable Solutions
The Skill Rhythmically bouncing a ball on a racket is a seemingly simple and playful
task. Yet a closer look reveals that it requires a high level of perceptually guided
coordination to repeatedly intercept the ball to hit a target amplitude, i.e., perform
in a rhythmic fashion (Fig. 8.5a). As in skittles, success is determined at one critical
moment when the racket hits the ball, as this impact fully determines the trajectory
of the ball. Hence, a core feature of this task is the control of collisions, which
is germane to numerous other behaviors, ranging from running to clapping, and
numerous sport skills, such as volleyball. One key difference to skittles, however,
is that these impacts are performed in continuous rhythmic fashion, and errors from
one bounce propagate to the next. Via the repeated interactions with the ball, the
actor becomes part of a continuous dynamical system.
a b
Target
Error
Ball
Racket
c d
Target
Ball Velocity (m/s)
-3
Tri
al
Position
1
-4
Tri
-5 al 2
0
0
0.1 .2
Ra siti )
0 1 0.8 0.6 0.4 0.2 0

Po (m
Time
ck on
et
Racket Velocity (m/s)
Fig. 8.5 a Real-life rhythmic ball bouncing skill. b Model of the racket–ball system. The vertical
ball position between each instantaneous impact follows ballistic flight, which depends on three
execution variables: ball velocity vb− and racket velocity vr− velocities just before impact, and racket
position xr at impact. Error is defined as the absolute difference between the target height and
maximum ball height. c Execution space of the racket–ball system. The gray plane represents the
two-dimensional solution manifold. Two data sets represent bounces in first trial (blue) and after 20
trials of practice (yellow). d Illustration of dynamic stability associated with negative acceleration.
The black trajectory of the ball exemplifies stable performance. A positive perturbation to racket
velocity just before impact was added to the green ball trajectory at the arrow. The error from the
perturbation is corrected after three bounces without any changes in the sinusoidal racket trajectory
The Model The model for this task is a well-studied nonlinear dynamical system,
originally developed for a particle bouncing on a vibrating surface (Guckenheimer
and Holmes 1983; Tufillaro et al. 1992). The simple model consists of a planar
surface moving sinusoidally up and down, and a ball or point mass that impacts the
surface with instantaneous contact (Fig. 8.5b). The vertical position of the ball xb
between the kth and the k + 1th racket–ball impact follows ballistic flight:
xb (t) = xr (tk ) + vb+ (t − tk ) − g/2(t − tk )2 ,

where tk is the time of the kth ball–racket impact, xr is racket position, vb+ is the
ball velocity just after impact, and g is the acceleration due to gravity (9.81 m/s2 ).
With the assumption of instantaneous impact, the ball velocity just after impact vb+
is determined by
vb+ = (1 + α)vr− − αvb− ,
where vb− and vr− are the ball and racket velocities just before impact, and the energy
loss at the collision is governed by the coefficient of restitution α. The maximum
height of the ball between time tk and tk+1 depends on the ball and racket velocities
just before impact and the position at impact xr , as shown by the following equation:
max xb (t) = xr (tk ) + (((1 + α)vr− − αvb− )(t − tk )) /2g.

2
tk ≤t≤tk+1
The task goal is to bounce the ball to a target height, and the error is defined as the
difference between the maximum height of the ball and the target height (Fig. 8.5b).
Even in this simplified form, the task has redundancy, as the result variable error is
determined by three execution variables: vb− , vr− , and xr . Figure 8.5c shows execution
space with the solution manifold, i.e., the planar surface that represents all solutions
leading to zero height error. The blue and yellow data points are two data sets from
early and late in practice, respectively; each data point corresponds to one ball–racket
contact. As to be expected, the early data show a lot of scatter, while the late practice
data cluster around the solution manifold.
Hypotheses The representation of bounces in execution space affords analysis of the
data using the TNC method. As in the two-dimensional version in skittles, the data can
be transformed to estimate the contributions of tolerance, noise, and covariation. The
same hypothesis as in skittles can be pursued in ball bouncing. However, an important
difference to skittles is that only racket position and velocity at impact are under
the actor’s direct control; the third execution variable, ball velocity, is determined
by the previous contact (which in turn is determined by the previous contact, etc.).
Therefore, continuous rhythmic control of the ball depends on the preceding contacts,
which adds an additional challenge to the controller. As mentioned above, the task
is a dynamical system that lends itself to different analyses and predictions.
To model the racket and ball system as a continuous dynamical system, the racket
movements need to be specified. To this end, the model assumes simple sinusoidal
motion of the racket. With this assumption, racket position and velocity at impact
collapse into a single state variable, racket phase, or acceleration at impact. Further,
the continuous system can be discretized at the ball–racket contact, where ball and
racket positions are identical. A discrete map can be derived based on two state
variables, the ball velocity just after impact vk+ and the racket phase at impact θk :
+
vk+1 = (1 + α)Aω cos θk+1 − αvk+ + gα(θk+1 − θk )/ω
0 = Aω2 ( sin θk − sin θk+1 ) + vk+ ω(θk+1 − θk ) − g/2(θk+1 − θk )2 ,

where A and ω are the amplitude and frequency of the sinusoidal racket movements
(Dijkstra et al. 2004; Schaal et al. 1996).
This ball–racket system is a nonlinear system that displays dynamic stability and,
despite its simplicity, shows the complex dynamics of a period-doubling route to
chaos (Guckenheimer and Holmes 1983; May 1976; Tufillaro et al. 1992). As the
task requires simple periodic bouncing, only periodic behavior was considered. Local
linear stability analysis of this discrete map identifies a fixed-point attractor, when
racket acceleration at impact ar satisfies the inequality (Schaal et al. 1996; Sternad
et al. 2001):
(1 + α 2 )
−2g < ar < 0.
(1 + α)2
In this dynamically stable state, small perturbations of the racket or ball die out
without requiring any corrections. To make this property of the task more intuitive,
consider the illustration in Fig. 8.5d: While hitting the ball with negative racket
acceleration (the upwardly decelerating phase), a ball that contacts the racket earlier
is hit with relatively higher velocity, which leads to a higher ball amplitude following
the contact. Conversely, a ball that contacts the racket later, such as after an amplitude
overshoot, is hit with relatively lower velocity, which leads to a lower maximum
ball height on the next bounce. Note that this purely physical relation automatically
compensates for an over- or undershoot in the previous ball amplitude. This automatic
compensation via covariation among the execution variables is the underpinning to
the error compensation in dynamically stable performance.
This property leads to an interesting hypothesis for human control: If subjects
establish such a dynamically stable regime, they need not correct for small pertur-
bations that may arise from the persistent motor noise. Thus, we hypothesized that
subjects learn this “smart” solution of hitting the ball with negative racket acceler-
ation to exploit dynamic stability. This dynamically stable solution makes intrinsic
noise matter less.
Virtual Environment In the experiment, the participant stood in front of a rear pro-
jection screen and was instructed to rhythmically bounce the virtual ball (white) to a
target line (yellow) using a real table tennis racket (Fig. 8.6a). The projected racket
movements (red) were shown on the screen impacting the ball. One trial usually
lasted 40 s. A light rigid rod with two hinge joints was attached to the racket and run
through a wheel, whose rotations were registered by an optical encoder. While the
joints allowed the racket to move and tilt with minimal friction in all three dimen-
sions, the encoder only measured the vertical displacement of the racket, in analogy
with the model. Racket velocity was continuously estimated. The displacements of
the virtual racket were controlled by the measured position of the real racket, and the
vertical position of the virtual ball between impacts was determined using the ballis-
tic flight equation. To simulate the haptic sensation of a real ball hitting the racket, a
mechanical brake, attached to the rod, was activated at the ball–racket impact of each
bounce. Racket acceleration at impact was analyzed after the experiment and served
as the primary measure of dynamic stability to test the hypothesis (Wei et al. 2007).
a b
20 20
Racket Accel at Impact (m/s2)

15 15
Racket
Error (cm)
10 10
5
Encoder
0 0
-5 -5
PC 1 5 10 15 20
Trial Number
Fig. 8.6 a Participant interacting with the virtual environment. The movements of the racket (red)
and ball (white) and the target height (yellow) are presented to subjects on the large screen in front
of them. b Median error (red) and racket acceleration at impact (blue) for 48 novice participants
over 20 trials of practice. Error bars show the 25th and 75th percentiles across participants
Experimental Findings The hypothesis was tested in an experiment that involved

48 novice subjects, who performed 20 trials of 40 s each (which corresponds to
approximately 60 bounces per trial; Ehrlenspiel et al. 2010). Figure 8.6b shows
the medians over subjects in each trial. The error (red) declined with practice as to
be expected. More importantly, this improvement was paralleled by a decrease in
racket acceleration at impact (blue), starting from positive values at the beginning
to assuming negative values after about eight trials. This result highlights that the
intuitive solution for novices is to accelerate to impart energy to the ball in the upward
direction. Yet, this is not the most effective approach, when the ball is hit repeatedly
with periodic racket movement. Further, from a biomechanical perspective, hitting
the ball with maximum velocity, i.e., at the moment of zero acceleration, would be
the most energy-efficient solution. Maximum velocity would result in the highest
ball amplitude for a given rhythmic racket trajectory. Nevertheless, subjects reliably
converge to solution where the racket is in its decelerating phase at ball contact.
Additional support for this strategy was revealed by estimating covariation among the
three execution variables at impact (Ehrlenspiel et al. 2010). These results highlight
that adopting solutions that afford dynamic stability is not trivial, but is a smart
solution that requires practice.
A series of studies tested the generality of this result in different experimental
renderings to rule out the possibility that this could be the fortuitous effect of the
tight constraints in the virtual setup. The experimental setups ranged from having the
participant manipulate a pantograph linkage that allowed precise control of the haptic
contact (Schaal et al. 1996; Sternad et al. 2001), a real tennis racket to bounce a real
ball attached to a weighted boom that confines the ball to a vertical path (Katsumata
et al. 2003; Sternad et al. 2000, 2001), to having the participant freely bounce a
real ball with a real tennis racket in 3D (Sternad et al. 2000, 2001). The findings
were robust: Experienced performers hit the ball with negative racket acceleration.
Further support that humans seek dynamically stable solutions came from a study by
Morice et al. (2007), where the visuomotor phase between the real racket and ball
movements was shifted, performers still learned to exploit dynamic stability, where
intrinsic noise matters less.
Interim Conclusions Just as in the skittles task, participants were shown to find
stable or error-tolerant solutions in the redundant space of solutions to achieve the
task. Further analysis of covariation and noise is still to come. The commonality
of the two paradigms is that the subject’s control of the ball was confined to a very
short moment. A first extension from this momentary interaction was made in an
experiment that extended the duration of contact: A beanbag replaced the elastic
ball, such that the contact duration became long enough to afford corrections and
essentially became a catch followed by a resetting of initial conditions to throw
(Katsumata et al. 2003). However, real-life skills include numerous examples that
involve continuous interaction with objects. Hence, the next step is to design an
experimental paradigm that features continuous interaction with a complex object:
a cup of coffee.
8.3.3 Paradigm 3: Complex Object Manipulation: Carrying

a Cup of Coffee
The Skill Grasping a cup of coffee and leading it to one’s mouth is a seemingly mun-
dane task. Yet, it exemplifies a class of movements that require continuous interaction
with an object that has internal dynamics (Mayer and Krechetnikov 2012). While
one moves the cup, the coffee is only indirectly controlled via moving its container
(Fig. 8.7a). How the CNS controls interactions with such objects is intriguing, as the
moving liquid creates time-varying reaction forces that have to be accounted for to
keep the cup under control and not spill the coffee. The essential challenge of this
task is to perceive these time-varying dynamics and establish a safety margin for
transporting the cup of coffee.
The Model Leading a cup of coffee to one’s mouth is a tall order to model mathe-
matically. Hence, we again simplified the situation by modeling of the cup as a bowl
with a ball moving inside it—representing the internal degree of freedom that affect
the system’s dynamics (Hasson et al. 2012a, b, Hasson and Sternad 2014) as shown
in Fig. 8.7a. Assuming that the cup’s movements are confined to the horizontal direc-
tion and the ball can swerve within the cup, this reduced system is approximated by
the well-studied cart-and-pendulum system (Fig. 8.7b). The cart corresponds to the
cup and the bob’s pendular movements represent the sloshing liquid. The governing
equations of the system dynamics are
(m + M)ẍ = F
θ̈ = (ẍ/ l) cos θ − (g/ l) sin θ ,
a b
‫ݔ‬ ‫ܯ‬
݈ ߠ
݉
c d
Cup Acceleration (m/s2)
−1
−200
0 −40
0
Velo Ball 200 40
gle (°)
city
(° /s) Ball An
Fig. 8.7 a Holding a cup of coffee simplified as holding a cup with a ball inside. b Model of
the cart-and-pendulum system. In the mathematical model, x is the cart/cup position, θ is the
ball angular position, M is the cup mass, m is the ball mass, and l is the pendulum length. c An
exemplary successful trajectory (trial) in three-dimensional execution space, spanned by ball angle
and velocity and the acceleration of the cup, which together specify the energy margin of the ball.
The energy margin describes a two-dimensional manifold (meshed blue surface). Assuming the cup
acceleration is unchanged, the trajectory inside the manifold has a positive safety margin and the
ball will remain in the cup (red trajectory); if the trajectory is outside (black trajectory), the energy
margin is negative and the ball will escape unless a correction is made. d Participant interacting
with the virtual environment. A haptic manipulandum provides the mechanical interaction with the
object model and the movements of ball and cup appear on the rear projection screen. The task is
to transport the cup from the left green start box to the right green target box as fast as possible
(minimum-time condition) or in exactly 2 s (target-time condition) without letting the ball escape
from the cup
where θ , θ̇ , and θ̈ are angular position, velocity, and acceleration of the ball; x, ẋ,
and ẍ are the cart/cup position, velocity, and acceleration, respectively; F is the net
force acting on the cart/cup. Parameters of the system are m and M representing the
masses of ball and cup, respectively; l is the length of the rod (pendulum length);
and g is the gravitational acceleration. Hence, the model has four state variables
x, ẋ, θ , and θ̇ and the externally applied force F that determines the behavior of the
ball-and-cup system. The task is determined by five variables. The new problem is
that only one variable (force) is under direct control of the subject, and even this
variable is continuously influenced by the reaction forces of the ball. In fact, F is
the net force, as the ball or pendulum bob moves under the influence of the cart
dynamics and gravity and imparts a time-dependent force to the cup in addition to
the active force supplied by the actor. This makes the distinction into execution and
result variables significantly more complicated than in the previous two examples.
Hypotheses Nevertheless, the core feature of redundancy is present also in this task:
Transporting the cup from one position to another can be achieved by multiple trajec-
tories. How can one conceive of a result space determined by execution variables?
One important constraint of the task is to keep the ball within the cup, which re-
duces the range of all available trajectories by limiting the forces that the actor can
successfully apply. This task constraint can be expressed via the energy of the ball
and its distance from escape (see more detail in Hasson et al. 2012b). Briefly, the
current energy of the ball is a sum of its potential, kinetic, and inertial energy (energy
conferred to the ball due to the acceleration of the cup). Potential energy is defined
by the height of the ball within the cup, kinetic energy is defined by the velocity of
the ball, and the inertial energy is provided by the instantaneous acceleration of the
cup (Hasson et al. 2012b). At each sampled point in the movement, one can define an
energy margin as the difference between the current energy of the ball and the energy
that would make the ball escape. This energy margin is the dynamic analogue to the
safety margin, as defined in studies on grip force in object manipulation (Flanagan
and Wing 1997; Johansson and Westling 1984). The energy margin is positive, when
participants use accelerations that keep the current energy smaller than the energy
level that would make the ball escape—a safe strategy. A more risky strategy would
yield values of energy margin near zero, getting the ball close to the rim of the cup.
Negative energy margin values indicate that the ball would escape in the near future,
assuming no change in the currently applied acceleration.
When the task result is defined in these terms, execution space spans only three
variables: acceleration of the cup/cart, position, and velocity of the ball/pendulum.
Figure 8.7c visualizes this execution space including a result surface that specifies
the energy margin of zero. The area inside this manifold captures those executions
that have positive energy margins (from 0 to 1); all solutions outside the manifold
would lead to ball escape (assuming no change in acceleration). The plotted trajectory
demonstrates an example evolution of the execution variables during a single trial of
moving the cup. Note that the trajectory may leave the volume bounded by the blue
surface temporarily, but as long as participants quickly adjust their acceleration, they
can prevent ball escape. However, larger deviations will lead to ball escape.
Extending previous arguments that performers acquired solutions with stability
to make intrinsic noise matter less, we tailored this hypothesis to the current task. As
subjects moved the cup from A to B, and not “spill the coffee” (lose the ball), we hy-
pothesized that with practice subjects will seek solutions that have large energy mar-
gins. Further, we hypothesized that individuals with greater movement variability,
or intrinsic noise, would seek solutions that were less risky (and vice versa).
Virtual Environment The cart-pendulum system was rendered in a virtual environ-

ment where participants only saw the bob of the pendulum and the arc of its trajectory
that represented the cup (Fig. 8.7d). Participants controlled the virtual cup using a
robotic manipulandum that was constrained to move the cup in the horizontal direc-
tion (Haptic Master, Moog, Netherlands). The task goal was to transport the virtual
cup from an initial position to a target position at 40 cm distance (green boxes). In
two conditions, subjects moved the cup either as fast as possible (minimum-time
group) or at the specified time of 2 s (target-time group). The time window was
signaled to subjects by a white box that began descending onto the target box at
constant velocity upon leaving the start box. Two seconds passed when the white
box was exactly overlaid on the target box.
Empirical Findings Two groups of nine participants each practiced this task for
300 trials in each of the two conditions (Hasson et al. 2012b). The motivation for
the target-time condition was to create a condition where there was redundancy: 2
seconds gave subjects ample time to develop different strategies. We hypothesized
that they would seek stable strategies that optimized the energy margins to allow for
noise and perturbations to matter less. For comparison, the minimum-time condition
only presented one optimal solution. To increase speed, subjects were expected to
decrease their energy margins. In the two conditions, changes in both directions
would signal that subjects were aware of their variability.
With practice, all participants in the target-time group became better in placing
the cup in the target box in the specified 2-second interval; similarly, the minimum-
time group decreased their movement time. Further, in both conditions, participants
lost the ball less often and their trial-to-trial variability decreased. The most central
variable was the energy margin, which was calculated for each point along the cup
trajectory, yielding a time-varying pattern. Figure 8.8a, 8.8b exemplifies the pattern
of results in both conditions and their change with practice. In the target-time con-
dition, participants increased their energy margin, indicating that strategies became
safer with practice. This result is consistent with the hypothesis, because higher
energy margin values provide more tolerance to intrinsic noise. In contrast, the
minimum-time group decreased their energy margin, indicating that their strategies
became more risky as a compromise to increase speed. Yet, the number of ball losses
still decreased, reflecting that subjects were sufficiently sensitive to their variability.
Figure 8.8c summarizes the change in energy margin with practice in all individuals
in the two conditions.
To test the second hypothesis stating that the preferred energy margin should
be adjusted to each individual’s variability, we examined the correlation between
the variability of the total energy throughout the movement and the energy margin
during that same movement in the last 30 trials of the experiment (for details, see
Hasson et al. 2012b). The results showed that participants in the minimum-time
group (blue circles) who had more variability also showed greater energy margins,
suggesting that more variable individuals used safer movement strategies (Fig. 8.9).
However, this correlation was not present in the minimum-time group (red triangles).
One explanation is that the variability in the total energy did not simply reflect the
a Target Time c
1.0
Target Minimum
Time Time
Energy Margin
0.5
0.5
0
Early 0.45
Late
-0.5
0 50 100
Movement Time (%) 0.4
Energy Margin 0.35

b Minimum Time
1.0
0.3
Energy Margin
0.5
0.25
0
Early
Late 0.2
-0.5
0 50 100
Movement Time (%)
Increased Energy Margin:

Decreased Energy Margin: Early Late Early Late
Fig. 8.8 Energy margin over the time of a trial (normalized to 100 %) for an exemplary participant
from the a target-time group and the b minimum-time group. Dark/light shading indicates a de-
crease/increase in energy margin between the average for first 30 versus last 30 trials. c Summary
of changes in energy margin with practice. Each data point represents the average for the first (early
practice) or last (late practice) 30 trials for the target-time (triangles) and minimum-time group
(circles)
Fig. 8.9 Correlations 5

between energy margin and R2 = .86
SD Total Energy (J)
trial-to-trial variability 4
(standard deviation) of total
energy during the last 30
trials for the target-time group 3
(triangles) and the
minimum-time group 2
(circles)
1
0.2 0.4 0.6
Energy Margin
variability at the “physiological limit,” but also captured the greater variability in
the movements when less constrained. Despite this partially inconsistent result, the
findings lend support to the hypothesis that actors take stability properties of the task
and their own intrinsic noise characteristics into account when learning to manipulate
a “cup of coffee.”
8.4 Conclusions and Outlook
This chapter aimed to present a methodological approach to study the acquisition

and control of novel and more complex skills that capture coordination challenges
inherent to real-life motor skills. This work differs from much of the more recent
research on motor learning or adaptation, where an existent movement pattern is
adapted to external manipulations. We advocated to study more complex tasks to
understand the acquisition of novel skills, which is an essential and ubiquitous process
that enables humans not only to perform extraordinary actions, such as skateboarding,
but also to eat with knife and fork or drive a car. How the brain changes and stores
such infinitely many new skills that essentially define our human existence remains
one of the core open questions in human neuroscience.
We proposed that the minimal ingredient for skill is redundancy, in particular
extrinsic or task redundancy that allows individuals to make choices or even decisions
about how they perform the task. This contrasts with much of the previous work in
motor neuroscience that has tended to focus on simple and well-defined movements
with a single goal. These experimental paradigms are well suited to study simpler
forms of learning, including habituation and sensitization, classical and operant
conditioning, and motor adaptation. In order to extend the realm of scientific study
and address more high-level skills, we developed a methodological approach that
still afforded hypothesis-driven research. In three examples, we exemplified our
model-based approach.
All three experimental paradigms involved interaction with an object and thereby
exemplified different aspects of tool use, a fundamental human skill. They form a
logical progression in the physics they involve: In skittles, the trajectory of the thrown
ball is fully determined at the moment of release; there is no further interaction with
the ball. Every trial is a new start, setting new initial conditions. In ball bouncing,
subjects again contact the ball at a very brief moment that fully determines the
subsequent ball trajectory. However, unlike in skittles, the successive ball contacts
are not independent: The ball’s trajectory from the previous impact influences the
next ball impact. Lastly, in the cup-and-coffee task, the force applied to the cup and
the force exerted by the sloshing coffee onto the cup are in continuous and complex
interplay. In this continuous dynamical system, every moment in human control
matters. Due to this increasing complexity, the methods for analysis in the three
paradigms need to be different. Nevertheless, even though the models are different
and render different variables, all three experiments pursued one core hypothesis:
Humans seek stable solutions where intrinsic noise matters less.
Each experimental task was designed to allow an infinite number of solutions to a

given task. Given these multiple options, it is necessary to have a model as reference
to evaluate the observed variability within subjects across practice, but also across
subjects. It is essential that the model described not only the physics of the task,
but also how the subject interacted with the physics, i.e., how the subject executed
the task to achieve a result. If the functional relationship between execution and
result is known, the space of all solutions is known and the manifold of solutions can
be calculated. With this knowledge, not only can the practice-induced progression
in performance be characterized but individual differences can also be studied in a
more principled way. For example, a person with high variability may prefer safer
strategies, while a person with low intrinsic noise may prefer a more risky strategy.
While this is a first start, more work is needed to better understand interindividual
differences (Cesqui et al. 2012).
While our current research has focused on developing and testing the approach
on healthy individuals, the experimental platforms are ideal to take the next step:
diagnose coordination deficits and design interventions to accelerate learning and
modify behavior of patient populations. For example, the virtual rendering of the task
lends itself to perceptual modifications that may shape the characteristics of behavior.
One study exemplifies the initial step in this direction: Using a modified skittles task,
we showed how a decrease in perceived variability could affect movement strategies
in children with dystonia (Chu et al. 2013). Following such first forays, the arena of
questions and applications is wide open.
Acknowledgments This work was supported by the National Institute of Child Health and Human
Development (NICHD) R01 HD045639, National Science Foundation NSF-DMS0928587, and
American Heart Association AHA-11SDG7270001, awarded to Dagmar Sternad. Fellowships from
the Northeastern University Graduate School of Engineering and The Mathworks supported Meghan
Huber.A Ruth L. Kirschstein National Research Service Award supported Christopher J. Hasson
(1F32AR061238).
References
Araujo D, Davids K, Hristovski R. 2006. The ecological dynamics of decision making in sport.
Psychol Sport Exerc 7(6):653–76.
Baraduc P, Lang N, Rothwell JC, Wolpert DM. 2004. Consolidation of dynamic motor learning is
not disrupted by rTMS of primary motor cortex. Curr Biol 14(3):252–6.
Bernstein NA. The co-ordination and regulation of movements. Oxford, UK: Pergamon; 1967.
Cesqui B, d’Avella A, Portone A, Lacquaniti F. 2012. Catching a ball at the right time and place:
individual factors matter. PLoS One 7(2):e31770.
Chu VWT, Sternad D, Sanger TD. 2013. Healthy and dystonic children compensate for changes in
motor variability. J Neurophysiol 109:2169–78.
Cohen RG, Sternad D. 2009. Variability in motor learning: relocating, channeling and reducing
noise. Exp Brain Res 193(1):69–83.
Cohen RG, Sternad D. 2012. State space analysis of timing: exploiting task redundancy to reduce
sensitivity to timing. J Neurophysiol 107(2):618–27.
Criscimagna-Hemminger SE, Donchin O, Gazzaniga MS, Shadmehr R. 2003. Learned dynamics

of reaching movements generalize from dominant to nondominant arm. J Neurophysiol 89(1):
168–76.
d’Avella A, Saltiel P, Bizzi E. 2003. Combinations of muscle synergies in the construction of a
natural motor behavior. Nature Neurosci 6(3):300–8.
Dijkstra TMH, Katsumata H, De Rugy A, Sternad D. 2004. The dialogue between data and model:
passive stability and relaxation behavior in a ball bouncing task. Nonlinear Stud 11(3):319–45.
Donchin O, Francis JT, Shadmehr R. 2003. Quantifying generalization from trial-by-trial behavior
of adaptive systems that learn with basis functions: theory and experiments in human motor
control. J Neurosci 23(27):9032–45.
Draganski B, Gaser C, Busch V, Schuierer G, Bogdahn U, May A. 2004. Neuroplasticity: changes
in grey matter induced by training. Nature 427(6972):311–2.
Ehrlenspiel F, Wei K, Sternad D. 2010. Open-loop, closed-loop and compensatory control:
performance improvement under pressure in a rhythmic task. Exp Brain Res 201(4):729–41.
Faisal AA, Selen LPJ, Wolpert DM. 2008. Noise in the nervous system. Nature Rev Neurosci
9(4):292–303.
Flanagan JR, Wing AM. 1997. The role of internal models in motion planning and control: evidence
from grip force adjustments during movements of hand-held loads. J Neurosci 17(4):1519–28.
Guckenheimer J, Holmes P. Nonlinear oscillations, dynamical systems, and bifurcations of vector
fields. New York, NY: Springer; 1983.
Hasson CJ, Hogan N, Sternad D. Human control of dynamically complex objects. Proceedings of the
4th IEEE RAS/EMBS International Conference on Biomedical Robotics and Biomechatronics;
2012 June 24–28; Rome, Italy: IEEE; 2012a.
Hasson CJ, Shen T, Sternad D. 2012b. Energy margins in dynamic object manipulation. J
Neurophysiol 108(5):1349–65.
Hasson CJ, Sternad D. 2014. Safety margins in older adults increase with improved control of a
dynamic object. Front Comp Neurosci 6: doi: 10.3389/fnagi.2014.00158
Johansson RS, Westling G. 1984. Roles of glabrous skin receptors and sensorimotor memory in
automatic control of precision grip when lifting rougher or more slippery objects. Exp Brain
Res 56(3):550–64.
Karni A, Meyer G, Rey-Hipolito C, Jezzard P, Adams MM, Turner R, et al. 1998. The acquisition of
skilled motor performance: fast and slow experience-driven changes in primary motor cortex.
Proc Nat Acad Sci USA 95(3):861–8.
Katsumata H, Zatsiorsky V, Sternad D. 2003. Control of ball-racket interactions in rhythmic
propulsion of elastic and non-elastic balls. Exp Brain Res 149(1):17–29.
Landi SM, Baguear F, Della-Maggiore V. 2011. One week of motor adaptation induces structural
changes in primary motor cortex that predict long-term memory one year later. J Neurosci
31(33):11808–13.
May RM. 1976. Simple mathematical models with very complicated dynamics. Nature
261(5560):459–67.
Mayer HC, Krechetnikov R. 2012. Walking with coffee: why does it spill? Phys Rev E 85(4):046117.
Morasso P. 1981. Spatial control of arm movements. Exp Brain Res 42(2):223–7.
Morice AHP, Siegler IA, Bardy BG, Warren WH. 2007. Learning new perception–action solutions
in virtual ball bouncing. Exp Brain Res 181(2):249–65.
Müller H, Loosch E. 1999. Functional variability and an equifinal path of movement during targeted
throwing. J Hum Mov Studies 36:103–26.
Müller H, Sternad D. 2003. A randomization method for the calculation of covariation in multi-
ple nonlinear relations: illustrated with the example of goal-directed movements. Biol Cybern
89(1):22–33.
Müller H, Sternad D. 2004a. Accuracy and variability in goal oriented movements–decomposing
gender differences in children. J Human Kinetics 12:31–50.
Müller H, Sternad D. 2004b. Decomposition of variability in the execution of goal-oriented tasks:
three components of skill improvement. J Exp Psychol: Hum Percept Perform 30(1):212–33.
Müller H, Sternad D. Motor learning: changes in the structure of variability in a redundant task.
In: Sternad D, editor. Progress in Motor Control—A Multidisciplinary Perspective. New York,
NY: Springer; 2009. p. 439–56.
Nourritt-Lucas D, Zelic G, Deschamps T, Hilpron M, Delignières D. 2013. Persistent coordination
patterns in a complex task after 10 years delay. Hum Mov Sci 32(6):1365–78.
Park S-W, Dijkstra TMH, Sternad D. 2013. Learning to never forget—time scales and specificity of
long-term memory of a motor skill. Front Comp Neurosci 7: doi: 10.3389/fncom.2013.00111.
Schaal S, Atkeson CG, Sternad D. 1996. One-handed juggling: a dynamical approach to a rhythmic
movement task. J Motor Behav 28(2):165–83.
Scholz JP, Schöner G. 1999. The uncontrolled manifold concept: identifying control variables for
a functional task. Exp Brain Res 126(3):289–306.
Shabbott BA, Sainburg RL. 2010. Learning a visuomotor rotation: simultaneous visual and
proprioceptive information is crucial for visuomotor remapping. Exp Brain Res 203(1):75–87.
Shadmehr R, Moussavi ZMK. 2000. Spatial generalization from learning dynamics of reaching
movements. J Neurosci 20(20):7807–15.
Shadmehr R, Mussa-Ivaldi FA. 1994. Adaptive representation of dynamics during learning of a
motor task. J Neurosci 14(5):3208–24.
Shim JK, Olafsdottir H, Zatsiorsky VM, Latash ML. 2005. The emergence and disappearance of
multi-digit synergies during force-production tasks. Exp Brain Res 164(2):260–70.
Sternad D, Duarte M, Katsumata H, Schaal S. 2000. Dynamics of a bouncing ball in human
performance. Phys Rev E 63(1):011902.
Sternad D, Duarte M, Katsumata H, Schaal S. 2001. Bouncing a ball: tuning into dynamic stability.
J Exp Psychol: Hum Percept Perform 27(5):1163.
Sternad D, Park S-W, Müller H, Hogan N. 2010. Coordinate dependence of variability analysis.
PLoS Comp Bio 6(4):e1000751.
Sternad D, Abe MO, Hu X, Müller H. 2011. Neuromotor noise, error tolerance and velocity-
dependent costs in skilled performance. PLoS Comp Bio 7(9):e1002159.
Swift EJ. 1910. Relearning a skillful act. An experimental study in neuro-muscular memory. Psychol
Bull 7(1):17.
Thoroughman KA, Shadmehr R. 2000. Learning of action through adaptive combination of motor
primitives. Nature 407(6805):742–7.
Ting LH. 2007. Dimensional reduction in sensorimotor systems: a framework for understanding
muscle coordination of posture. Prog Brain Res 165:299–321.
Tufillaro NB, Abbott T, Reilly J. An experimental approach to nonlinear dynamics and chaos.
Redwood City, CA: Addison-Wesley; 1992.
Waters-Metenier S, Husain M, Wiestler T, Diedrichsen J. 2014. Bihemispheric transcranial di-
rect current stimulation enhances effector-independent representations of motor synergy and
sequence learning. J Neurosci 34(3):1037–50.
Wei K, Dijkstra TMH, Sternad D. 2007. Passive stability and active control in a rhythmic task. J
Wing AM, Kristofferson AB. 1973. Response delays and the timing of discrete motor responses.
Percept Psychophys 14(1):5–12.
Wu Y-H, Pazin N, Zatsiorsky VM, Latash ML. 2012. Practicing elements versus practicing
coordination: changes in the structure of variance. J Motor Behav 44(6):471–8.
Chapter 9
The Dynamical Analysis of Inter-Trial
Fluctuations Near Goal Equivalent Manifolds
Joseph P. Cusumano, Joseph M. Mahoney and Jonathan B. Dingwell
9.1 Introduction
The human body possesses a number of degrees of freedom far in excess of that
needed to accurately execute a typical goal-directed movement. It is natural to ex-
pect this redundancy to play an important role in the regulation of motor variability
as well as to influence its experimentally-observed structure. An important class of
data analysis methods, based on the notion of task manifolds (Scholz and Schoner
1999; Müller and Sternad 2004; Cusumano and Cesari 2006), has been developed
to examine the effect of this redundancy. In this chapter, we present a data anal-
ysis paradigm that integrates a consideration of redundancy at the task level with
the dynamical analysis of inter-trial fluctuations arising from repeated goal-directed
movements. We ground our discussion by presenting a study of variability in a virtual
shuffleboard task, and model inter-trial fluctuations as the output of a perception-
action loop whose primary function is to reduce error from one trial to the next. We
show that the fluctuation dynamics in the vicinity of the task’s goal equivalent man-
ifold (GEM) (Cusumano and Cesari 2006; John and Cusumano 2007; Dingwell et
al. 2010; Dingwell and Cusumano 2010; Dingwell et al. 2013; Cusumano and Ding-
well 2013) allow us to characterize not only the “static,” geometrical distribution of
the variability, but also its temporal structure. This combined space-time analysis of
observed variability yields an improved understanding of how goal-level errors are
regulated and generated.
Task manifolds are surfaces in an appropriate body state space (e.g., joint kine-
matic variables) that contain all possible task solutions. Since, by definition, every
J. P. Cusumano () · J. M. Mahoney

Department of Engineering Science & Mechanics, Penn State University,
University Park, PA 16802, USA
Tel.: +1.814.865.3179
J. B. Dingwell
Department of Kinesiology & Health Education,
The University of Texas at Austin, Austin, TX 78712, USA
and Biology 826, DOI 10.1007/978-1-4939-1338-1_9
126 J. P. Cusumano et al.
point in a task manifold corresponds to body states that result in perfect task exe-
cution, only deviations off of the manifold result in error at the goal level. While
this same basic idea underlies multiple methods of variability analysis, a different
approach has been taken to implement it in each case, largely motivated by a differ-
ence in analytical focus. Uncontrolled manifold (UCM) analysis (Scholz et al. 2000;
Scholz and Schoner 1999; Latash et al. 2002; Schöner and Scholz 2007) assumes that
the task manifold is defined at each instant along a given movement trajectory, and
uses an average movement in a time-normalized set of trials to represent the task’s
goal. Given the hypothesis that control will only be applied to correct deviations
off of the manifold, ratios of normalized variances perpendicular and tangent to a
candidate manifold are used as a test: the expectation is that, for a true UCM, there
should be greater variance along it than there is normal to it. With a primary focus
on motor learning, the tolerance, noise, and covariation (TNC) method (Cohen and
Sternad 2009; Müller and Sternad 2004; Ranganathan and Newell 2010; Sternad
et al. 2011) statistically decomposes observed variability into the three empirical
“costs” in its name, each of which are defined relative to a task manifold. The TNC
approach defines the task manifold in a minimal space of variables needed to specify
the outcome of a task, such as the position and velocity of a ball at release during
a throwing task. TNC analysis does not just focus on the orientation of variability
with respect to the task manifold (via the covariation cost) but also takes into con-
sideration the total body-level variability (the noise cost), and relates the goal-level
error to variability at the body level (via the tolerance cost).
The GEM concept (Cusumano and Cesari 2006) was initially developed to carry
out experimental sensitivity analyses explicitly relating variability at the body and
goal levels. The GEM approach defines the task manifold in a manner similar to that
of the TNC approach, however it does so by emphasizing the role of a goal function,
a mathematical hypothesis on the task strategy that encodes the relationship between
the body and goal needed for perfect task execution. The zeros of the goal function are
used to analytically define the GEM. In addition, the derivative of the goal function
gives the body-goal matrix, the singular values of which characterize the task’s
sensitivity to body-level errors (Cusumano and Cesari 2006; John and Cusumano
2007), independent of any control considerations. Motivated by the fact that optimal
control, particularly in the form of the minimum intervention principle (MIP), has
been proposed as a theoretical basis for modeling the neuromotor system (Scott 2004;
Todorov and Jordan 2002; Todorov 2004), optimal control ideas were incorporated
with the GEM approach. The resulting dynamical data analysis framework allows one
to create models of inter-trial fluctuations that can be tested against movement data
from human subjects (Dingwell and Cusumano 2010; Dingwell et al. 2010; Dingwell
et al. 2013), providing an analysis of human motor variability data that combines
task manifold, optimal control, and time series analysis approaches (Cusumano and
Dingwell 2013).
In what follows, we describe GEM-based fluctuation analysis, and, as an illus-
tration, use it to study data from a virtual shuffleboard task. After defining the key
concepts, we obtain the geometric stability properties of the inter-trial dynamics for
skilled players operating near the shuffleboard GEM. The eigenvalues and eigen-
vectors of a linear update equation estimated from the data characterize the way
9 The Dynamical Analysis of Inter-Trial Fluctuations Near Goal Equivalent Manifolds 127
inter-trial fluctuations are organized around the GEM. We show that subjects exhibit
strong control of fluctuations in an eigendirection transverse to the GEM, but weak
control of fluctuations in an eigendirection nearly tangent to it. Furthermore, we
demonstrate that our dynamical analysis is robust under coordinate transformations
in a way that non-temporal variance-based methods are not. We conclude by dis-
cussing how our results support a generalized interpretation of the MIP, and in doing
so, suggest the possible involvement of competing costs other than pure goal-level
error minimization.
9.2 The Goal Equivalent Manifold
Consider a task for which we can express the goal-level error, e, in terms of a goal
function as:
f(x) = e (9.1)
where x ∈ RB (the body space), f ∈ RE (the goal space). For redundant systems,
B > E, that is the dimension of the body state space is greater than that of the space
of goal-level errors. This introduces the possibility that there are many (possibly
infinite) x that result in the goal-level error being zero. This particular set of body
states is called the goal equivalent set (GES), and is expressed mathematically as the
set G:
G {x | f(x) = 0}.
If the GES forms a surface in the body space, then it is referred to as a GEM. By
definition, changes in the body state variables that remain in G do not change the
performance at the goal, because the error remains zero.
Fluctuations in the body state are mapped to their resulting fluctuations at the
goal. Given an operating point on the GEM, x∗ ∈ G, we write the state for small
body-level fluctuations, ξ , away from a state on the GEM, x∗ ∈ G. The sensitivity
of the goal-level error to the body-level fluctuations ξ is found using a Taylor series
expansion of the error:

e = f(x∗ + ξ ) = f(x∗ ) + Df(x∗ )ξ + O ||ξ ||2 ,
where Df(x∗ ) J is the Jacobian of the goal function evaluated at x∗ . By hypothesis,

f(x∗ ) = 0, and, for skilled task performance, we expect small errors and, hence, small
fluctuations away from the GEM ( ξ
1). Thus, the error is expected to be well
approximated by the linear relationship:
e ≈ Df(x∗ )ξ = Jξ . (9.2)
In this context, J is referred to as the body-goal variability matrix (Cusumano and

Cesari 2006): it maps the fluctuations at the body level, ξ , to the error at the target, e.
Start Target
x0 v0 μ
x L
Fig. 9.1 Schematic of shuffleboard task. The subject releases the puck at position x0 with an initial
velocity v0 . The goal is for the puck to come to a stop at the target. The coefficient of kinetic friction
is constant along the length of the board
Since B > E (by hypothesis), the E × B matrix J has more columns than rows,
and so can be decomposed into orthogonal subspaces: the null space, N , and the
column space, C (Poole 2010; Golub and van Loan 1996), where:
N {ξ | Jξ = 0}, (9.3a)
and
C {ξ | ξ T w = 0 ∀w ∈ N }. (9.3b)
As defined in Eq. (9.3a), all perturbations away from x∗ that belong to the null
space will not cause the error to change from zero, to linear order in ξ . In contrast,
perturbations in the column space (Eq. 9.3b) will create nonzero error at the target.
Thus, in the context of movement variability analysis, we call N and C, the goal
equivalent and goal relevant subspaces, respectively. Geometrically speaking, the
null space is tangent to the GEM at x∗ , whereas the column space is orthogonal to
the GEM.
As an illustration, consider the simple shuffleboard task shown in Fig. 9.1, for
which B = 2 and E = 1. The goal of this task is to release a puck at some initial
distance, x0 , with some initial velocity, v0 , and have it stop on the target line that is
located at a distance L from the start position, x = 0. Once released, the puck is
decelerated by a Coulomb friction force until it comes to rest, so that, by Newton’s
second law (Greenwood 1988):
ẍ(t) = −μg, (9.4)
where μ is the coefficient of kinetic friction and g is the local acceleration of gravity.
Solving the above differential equation gives the final position, xf as
v02
xf = x0 + . (9.5)
2μg
For convenience, we rescale the release position and velocity into dimensionless
quantities, x and v, using
x0 v0
x= and v= √ , (9.6)
L 2μgL
which makes our analysis universal for any length of board, coefficient of friction,
or local acceleration of gravity. The goal function for the system is then obtained,
following Eq. (9.1), by noting that xf − L is the scalar error, where xf is as in
Eq. (9.5). After substituting Eq. (9.6), we find the goal-level error, e, in dimensionless
form as:
xf − L
e= = v2 + x − 1. (9.7)
L
Thus, for the shuffleboard task, the goal function is a scalar valued function, e =
f (x) ≡ f (x, v), so that the 2D body state is x = (x, v). That is, the performance
in one trial, as measured by e, is determined exclusively by the value of (x, v), the
position and velocity of the puck at release. √
The GEM is then obtained from Eq. (9.7) as G = {(x, v) | v = 1 − x } (for
0 ≤ x ≤ 1; see Fig. 9.2), and the 1 × 2 body-goal variability matrix is computed by
the gradient of Eq. (9.7) as
J = (1, 2v). (9.8)
Using the definitions of Eqs. (9.3a) and (9.3b), we find the unit normal and unit
tangent, n̂ and t̂, respectively, to be

T
1 2v
n̂ = √ , √ (9.9a)
1 + 4v2 1 + 4v2
and

T
2v 1
t̂ = − √ , √ , (9.9b)
1 + 4v2 1 + 4v2
where the superscript T indicates the matrix transpose. The unit vectors are shown
in Fig. 9.2.
9.3 Inter-Trial Control
Fluctuations in movement arise from inherent physiological noise that is present at

multiple scales (Eldar and Elowitz 2010; Faisal et al. 2008; Osborne et al. 2005;
Stein et al. 2005; McDonnell and Ward 2011). Thus, during repeated task execution,
a subject will strive to make adjustments from one trial to the next in an attempt
to maximize performance. We model this behavior as an iterative, error-correcting,
perception-action process (Warren 1990, 2006) in which perceived error from trial
k is used to estimate the body state at trial k + 1 in an attempt to drive the goal-level
error to zero. Among the simplest models are update equations with the form
xk+1 = xk + (I + Nk ) u(xk ) + ν k , (9.10)

1
n
›
t
›
0.8
x*
0.6
v
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1
x
Fig. 9.2 The goal equivalent manifold (GEM) for the shuffleboard task. The solid gray line denotes
the GEM (body states x = (x, v) leading to perfect task performance); dashed gray lines are ±5 %
error contours. Also shown (black arrows) are the unit normal n̂ (Eq. 9.9a) and unit tangent t̂
(Eq. 9.9b) at an operating point x∗ . Deviations from the GEM in the t̂ direction cause no error at
the target (i.e., they are goal equivalent), whereas deviations in the n̂ direction result in error at the
target (i.e., they are goal relevant)
in which u(xk ) is an inter-trial, error-correcting controller depending on the current

state xk , Nk is a matrix representing signal-dependent noise in the motor outputs
(Harris and Wolpert 1998), and ν k is an additive noise vector representing unmodeled
effects from perceptual, sensory, and motor sources. For skilled movements, which
are the focus of this chapter and for which the fluctuations ξ are small, similarly
small multiplicative noise terms drop out of the leading order analysis (Dingwell
et al. 2010; John and Cusumano 2007; Cusumano and Dingwell 2013), and so the
random matrix Nk will not be included in what follows.
To motivate the above model, consider that, for a skilled subject, we expect xk
to lie very close to the GEM, so if there were no noise there would be no need for
control (i.e., u = 0), and Eq. (9.10) would give xk+1 ≈ xk . However, some noise, ν k ,
is always present, and if there were no control, the fluctuations in xk would display
a random walk in the body space, which is not observed in experiments. These two
extreme limits illustrate that the job of the controller is to keep a multi-trial perfor-
mance closer to that of a perfect repetition than to a noise-induced random walk.
Models of this type implicitly assume a hierarchical structure to the overall motor
control system: the controller of Eq. (9.10) makes error-correcting adjustments, be-
tween trials, to an approximately “feed forward” controller that executes the task
within trials. In this view, the body state xk of Eq. (9.10) plays the role of a control
parameter for each individual goal-directed action.
Similar models have found use for the study of motor learning (van Beers 2009;
Burge et al. 2008; Diedrichsen et al. 2005), with the difference that the controllers
were formulated to depend directly on the goal-level error ek , instead of the body
state xk . However, in our approach, the body and goal-level fluctuations are related
by Eq. (9.1), so the fact that Eq. (9.10) depends only on body states yields a system
that is amenable to dynamical analysis in the presence of task redundancy. However,
it leaves open the question of how the controller incorporates error information,
something which we now address.
As before, we let ξ k be a fluctuation (at trial k) from an operating point on the
GEM, x∗ ∈ G, so that ξ k = xk − x∗ . For skilled performance, these fluctuations will
be small and so we assume a proportional linear controller
uk = Bξ k , (9.11)
where B is a constant proportional gain matrix. Subtracting x∗ from both sides of

Eq. (9.10) and substituting in Eq. (9.11), we directly relate the fluctuations from one
trial to the next:
ξ k+1 = (I + B)ξ k + ν k , (9.12)
where I is the identity matrix. Taking an optimal control approach, we further assume
that the controller acts to specify the state at the next trial, xk+1 , to minimize the
expected value of a cost function
with the following general form:

M

= α0 e(xk+1 )2 + αj pj (xk+1 ) + ξ Tk BT KBξ k , (9.13)
j =1
with terms described as follows. The first term in e(xk+1 ) = f(xk+1 ) (via Eq. 9.1),
represents the cost of goal-level error. The summation term represents additional
possible costs, as appropriate for a given problem, that may arise from biomechanical
(e.g., range of motion), physiological (e.g., energy minimization), psychological
(e.g., risk avoidance), or other considerations. The last term represents the cost of
controller “effort.” Also in Eq. (9.13), the parameters αj (j = 0, 1, . . . , M) as well
as the matrix K, are adjustable weights that can influence details of the system such
as, for example, the location of the operating point x∗ .
Thus, the goal-level error is incorporated into the controller of Eq. (9.10) implic-
itly, via the first cost of
. Furthermore, because the leading terms of
are evaluated
at the next step, substitution of the update equation (Eq. 9.10) into Eq. (9.13) im-
mediately shows how the noise ν k enters into the optimization, explaining why we
minimize the expected value, E[
], to determine the controller.
For an ideal MIP controller, α0 = 0 and αj = 0 for j ≥ 1 in Eq. (9.13), so the
sole cost being minimized at the next trial is the goal-level error. Since the goal-
level error deviates from zero only for goal relevant fluctuations (Fig. 9.2), such
a “perfect” MIP controller would push subsequent body states onto the GEM, but
would exert absolutely no control along it. The result is that in the presence of noise,
the model Eq. (9.10) predicts an unbounded random walk along the GEM. However,
such behavior is yet to be observed in multi-trial experiments. For this reason, in
Dingwell et al. (2010), an additional cost related to minimizing the distance from
a “preferred operating point” on the GEM was included as p1 in Eq. (9.13) (with
α1 = 0 and M = 1 in the sum). This allowed the associated models to capture the
localization of multi-trial experimental data around a location on the GEM in a way
that an ideal MIP model could not.
9.4 Local Geometric Stability
We can analyze the local geometric stability of the inter-trial control process using
Eq. (9.12). For convenience, we let A = I + B and rewrite Eq. (9.12) slightly as
ξ k+1 = Aξ k + ν k . (9.14)
The eigenvalues and eigenvectors of A, which satisfy Aê = λê, determine the sta-
bility properties of the inter-trial control applied to fluctuations near the GEM. The
magnitude of an eigenvalue indicates the action of the system on fluctuations in the
direction of its corresponding eigenvectors: after one trial, the size of the fluctuation
is reduced by a factor of (1 − |λ|) × 100 %. Thus having at least one eigenvalue
with |λ| > 1 indicates instability, whereas if all eigenvalues satisfy 0 ≤ |λ| ≤ 1,
the system is stable (Guckenheimer and Holmes1983; Hirsch et al. 2004). Heuris-
tically speaking, smaller values of |λ| correspond to greater stability. To facilitate
our discussion, in the remainder we focus on the case where the body states xk (and
hence the fluctuations ξ k ) are 2D, as is true for the shuffleboard task. In this case,
A is a 2 × 2 matrix, so there will be two eigenvalues. We also limit our discussion
to the case of real, distinct eigenvalues, which has been found to be sufficient in
experimental applications to date.
We label the the eigenvalues as λw and λs to classify them as “weak” and “strong,”
respectively, according to their magnitude, so that |λs | < |λw |. The corresponding
eigenvectors are labeled as êw and ês . Thus, after one trial, the size of fluctuations
along êw is reduced much less than those along ês . A typical situation is shown
schematically in Fig. 9.3, which shows the orientation of the eigenvectors with respect
to the GEM. Also shown are the weakly and strongly stable subspaces, span{êw } and
span{ês }, respectively. As illustrated in the figure, in general êw · ês = 0.
Note that, in general, the stability subspaces are not aligned with the normal and
tangential directions of the GEM. However, for an ideal MIP controller that pushes
states onto the GEM but has absolutely no control along it, one finds that êw is identical
to t̂, the unit tangent to the GEM (Fig. 9.2). Furthermore, the eigenvalues of an ideal
single-step MIP controller are λw = 1 and λs = 0, indicating that fluctuations along
the GEM are allowed to persist indefinitely, whereas those transverse to it are, in the
absence of noise, eliminated after a single trial. However, as discussed at the end of
the last section, such a scenario is not expected in experiments, since in the presence
of noise, it would result in a random walk of body states along the GEM from trial
to trial. Thus, we do not anticipate the controller to satisfy a strict interpretation of
the MIP, or UCM hypothesis, but rather a generalized version for which goal-level
error is still the dominant cost to be minimized (Eq. 9.13), but there is at least one
1
e
abl
y st
ngl
stro
0.8 ew x* es
›
wea
kly
stab
0.6 le
v
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1
x
Fig. 9.3 Schematic showing the typical orientation of the weak and strong eigenvectors of matrix
A (Eq. 9.12), êw and ês , respectively (black arrows) with respect to the goal equivalent manifold
(GEM) (gray curve). The weakly and strongly stable subspaces (dashed lines) span the respective
eigenvectors. The associated eigenvalues satisfy |λs | < λw ; in general, λs ≈ 0 and λw is closer to 1.
The weakly stable subspace for a generalized minimum intervention principle (MIP) controller will
be close to tangent to the GEM. The subspaces are so named because deviations off of the GEM
are strongly corrected, whereas deviations along it are only weakly corrected
additional cost, pj (xk+1 ), with a corresponding αj = 0. A small perturbation of this

type will only slightly change the local stability, so that we still expect |λs | ≈ 0 and
0
λw < 1. In addition, the eigenvectors will be slightly rotated from those for the
ideal MIP controller, so that, in particular, êw will no longer be exactly tangent to the
GEM.
As discussed in Cusumano and Dingwell (2013), consideration of the above leads
to a set of generic experimental hypotheses. Here, we present only those needed for
the experimental study of the 2D shuffleboard task to be discussed in the remainder
of the chapter:
H1 Consistent with the hypothesis of weak (but not zero) control along the GEM,
one of the eigenvectors, êw , will be close to the GEM tangent, and so will make
an angle θw ≈ 90◦ with n̂, the GEM normal. The corresponding eigenvalue,
λw , will be less than one but much greater than zero (0
λw < 1).
H2 The fluctuation dynamics transverse to the GEM are strongly stable: i.e., we
expect 0 ≈ |λs |
λw . The associated eigenvector, ês , will make a much
smaller angle with n̂ than does êw (θs
θw ).
H3 Consistent with the stability properties of H1 and H2, fluctuations in the weakly
stable subspace will exhibit statistical persistence, whereas those in the strongly
stable subspace will be non-persistent (uncorrelated). In terms of the exponent
α from detrended fluctuation analysis (DFA) (Hausdorff et al. 1995; Peng et
al. 1992), we expect 0.5 ≈ αs
αw .
Table 9.1 Experimental

Younger Older
subject demographics (mean
± standard deviation). n 8 8
Assuming normal Age (years) 23.7 ± 3.8 66.2 ± 7.2
distributions, no measures,
except age, had significantly Body mass (kg) 68.6 ± 13.0 72.5 ± 8.4
different means Height (cm) 172.9 ± 8.4 168.0 ± 11.1
Upper arm length (cm) 30.3 ± 2.3 30.0 ± 1.9
Forearm length (cm) 26.1 ± 1.4 26.5 ± 2.8
Hand length (cm) 18.3 ± 1.3 18.4 ± 1.2
Taken together, these hypotheses can be summarized as the expectation of

dynamic anisotropy in the fluctuation dynamics, consistent with the action of a
generalized MIP controller that has goal-level error as its main, but not only, cost.
9.5 Experimental Implementation
In this section, we apply the ideas and methods developed in the previous sections
to study the experimentally-observed variability in a virtual shuffleboard task.
9.5.1 Apparatus and Protocol
Sixteen subjects (Table 9.1) were recruited as two cohorts, eight subjects between
18 and 35 (“younger”) and eight subjects older than 60 (“older”). Subjects were
screened and asked to report any neurological conditions, upper-body arthritis and
joint replacement, or uncorrected vision in either eye for exclusion. Subjects were
also given the Edinburgh Handedness Inventory (Oldfield 1971), and all participants
were deemed right handed, having a score of at least 60 on a −100 to 100 scale.
Testing was approved by The Pennsylvania State University Institutional Review
Board. No statistically significant differences for the analyses carried out for this
chapter were found between the younger and older cohorts, so the results presented
below were computed with the cohorts merged.
A custom experimental virtual shuffleboard system was used for the task. The
experimental system was composed of a manipulandum on a linear bearing attached
to a linear variable differential transformer (LVDT) (Daytronic Corporation; Dayton,
OH) that measured the manipulandum’s position. A linear accelerometer (ADXL320;
Analog Devices, Inc.; Norwood, MA) was mounted on the manipulandum. Both the
LVDT and the accelerometer were routed into a low-pass filter with a 20 Hz cutoff.
Out of the filter, the data were sampled by a 12-bit PCMCIA data acquisition card
(NI DAQCard-6024E; National Instruments; Austin, TX) at 1 kHz, and sent to a
laptop where it was read into simulation and control software written in MATLAB
(The Mathworks; Natick, MA).
The control software was used to generate a 2D, top-down view of a virtual
shuffleboard court that was projected onto a screen located in front of the subjects’
seat. The projected image displayed a virtual cue that moved in sync with physical
motions of the manipulandum, together with a puck and a target line (recall Fig. 9.1).
Each trial started with the puck at rest at its zero position. While the puck was being
pushed by the cue and subsequently released, its position was displayed in real time.
The puck was released when the computed virtual contact force between the puck and
cue fell to zero. The velocity at release, v, was found by integrating the acceleration
data, and the release position, x, was found directly from the LVDT. A numerical
simulation then took over to calculate the released puck’s position and velocity, based
on the equation of motion (Eq. 9.4), and the puck’s motion was displayed in real time.
After the puck came to a stop, in addition to being able to visually see the puck’s
final location, subjects were shown the error e (Fig. 9.1), which was displayed at the
top-right of the screen. When the cue was retracted to the start position, the next trial
began.
Subjects were asked to adjust the seat to their comfort, but were required to have
their forearm parallel to the ground and their right hand above their mid-thigh when
the cue was at its initial position. All participants used their right hand to operate the
manipulandum. Subjects were given the instruction: “On each trial, have the center
of the puck stop on the black target line.” Subjects were given a training session of 50
trials to familiarize themselves with the system, and then were asked to complete 5
blocks of 50 trials each that were used for analysis. Between each block, the subject
was given a minimum of 90 s to rest. The release position and velocity, xk = (xk , vk ),
and the error ek (via Eq. 9.7) for each trial were recorded and stored.
9.5.2 Data Processing
All 250 trials were grouped together for each of the 16 subjects. Because we assumed
that subjects performed at a skilled steady-state, the first trial of each block after the
training block was removed to allow for refamiliarization after a break, and the
final trial was removed in case the subject performed differently knowing it was
the last trial. Any trials that involved subject errors or technical malfunctions were
also removed: there were a total of only six such trials removed across all subjects,
with a maximum of three for one subject. All data analyses were performed using
MATLAB.
For skilled subjects, we expect the average performance to be nearly perfect. That
is, the mean body state over all trials is very nearly on the GEM, and so we defined an
empirical operating point as x∗ = x. Then the fluctuation time series was computed
with respect to this mean operating point (MOP) from ξ k = xk − x.
Because of the limited length of each inter-trial time series, as well as the known
sensitivity of eigenanalyses for matrices contaminated by noise (Golub and van Loan
1996), bootstrapping (Mooney and Duval 1993; Freedman 1981; Efron and Tibshi-
rani 1994) was used to produce a confidence interval (CI) for the eigenvalue and
eigenvector calculations. A set of 250 fluctuations, ξ k , and the following fluctuation,

ξ k+1 , was selected randomly with replacement from the original trial data on each
iteration of the bootstrap. The matrix A was estimated via least squares, using the
pseudo-inverse (Moore and Barnard 1939; Golub and van Loan 1996) for each set
of ξ k and ξ k+1 . The eigenvalues and eigenvectors for each A were then calculated
and stored. This process was repeated over 105 iterations of the bootstrapping pro-
cedure. Occasionally, an A matrix with imaginary eigenvalues and eigenvectors was
produced: in this case, such matrices are numerical artifacts stemming from the fact
that we are trying to estimate a nearly-singular matrix with one eigenvalue near zero.
However, across all subjects, a mean of only 0.06 % of matrices generated during
the bootstrapping iteration fell in this category, with a maximum of 0.93 % for one
subject. Because of the relatively small number of instances, these matrices were
simply removed from the sample and rerun with new entries.
The above bootstrapped eigenanalyses were used to test hypotheses H1 and H2.
To test hypothesis H3, we employed DFA (Hausdorff et al. 1995; Peng et al. 1992).
The output of the DFA algorithm is a positive exponent, α: when α < 0.5, a time
series is said to be “antipersistent,” meaning that fluctuations in one direction are
likely to be followed by fluctuations in the opposite direction; when α > 0.5, the
time series is “persistent,” meaning that fluctuations are likely to be followed by
subsequent fluctuations in the same direction; finally, when α = 0.5, the time series
is “non-persistent” or uncorrelated. DFA has been widely used to search for long-
range persistence (also called long-range correlations) in experimental time series, a
topic that is beyond the scope of this chapter. However, as discussed at some length
in Cusumano and Dingwell (2013), when used for this purpose, DFA is prone to
false positives (Delignières and Torre 2009; Maraun et al. 2004; Gao et al. 2006).
Accordingly, we emphasize that we are not using DFA to claim that observed fluctu-
ations exhibit long-range persistence, but merely as a convenient tool to characterize
persistence in general. In previous work (Dingwell and Cusumano 2010), we have
shown that antipersistent time series are consistent with the overcorrecting control
of fluctuations normal to a GEM, whereas persistent fluctuations suggest weak or
“indifferent” control of fluctuations along it. In this chapter, instead of projecting the
fluctuations onto the vectors normal and tangent to the GEM, we examine fluctuations
in the weakly and strongly stable directions.
Direct application of DFA within each iteration of bootstrapping was compu-
tationally costly, so we first computed the mean value, A, of the 105 matrices A
generated during the bootstrapping procedure for one subject. Using A, we found
the “average” weak and strong eigenvectors êw and ês . Then we transformed the
fluctuations using the linear coordinate transformation
ξ = Pζ , (9.15)
where P is the matrix containing êw and ês as its columns. This transformation
projects the fluctuations ξ , originally expressed in dimensionless position and veloc-
ity coordinates, onto the average eigendirections, so that we can write ζ = (ζw , ζs )T .
Finally, DFA was performed on the resulting ζw and ζs fluctuation time series. These
results were used to test H3.
0.9
0.8
0.1 0.15 0.2 0.2

x
Fig. 9.4 Plot in the body space of all 250 trials (light gray dots) for one subject, showing the weakly
and strongly stable eigenvectors, êw and ês , respectively, obtained from the average, A, of all 105
matrices A estimated via regression during bootstrapping. Also shown are the mean operating point
(MOP, black circle) and goal equivalent manifold (GEM) (dark gray line). Normal and tangential
dashed reference lines are spanned by n̂ and t̂, respectively (Fig. 9.2). The angles θw and θs to each
eigenvector are measured counterclockwise from the normal, as shown
While the bootstrapped empirical probability density functions (EPDFs) obtained

for this study were close to Gaussian, the aggregate distributions were typically
not. Nor were they symmetric about their median. Therefore, across all subjects, a
comparison of medians was done using a two-sample, two-sided sign test (Wilcoxon
1945), with significance level set to 0.05. In the box plots used throughout, the
center line shows the median value, the lower and upper box edges encompass
the interquartile range (IQR), and whiskers extend a maximum of 1.5 IQR. Note,
however, that all data, including outliers, were included in the data analyses described
above.
9.6 Results
To illustrate the geometrical features of the data analysis, a scatter plot of an ensemble
of body states (release positions and velocities), collected for all 250 trials for one
subject, is shown in Fig. 9.4. For reference, the GEM and the directions normal
and tangential to it are also displayed. We see that the MOP is not located on the
GEM, but is close to it, as expected. Also plotted are the weakly and strongly stable
eigenvectors, êw and ês , respectively, for this figure computed from the average
matrix, A, estimated during the bootstrapping process. Also displayed are the angles
θw and θs that each eigenvector makes with the normal. Note also that the apparent
orientation of the cloud of trials does not align with the normal and tangential axes.
Across all subjects and trials, the goal-level error was e = −0.0204
[−0.1828, 0.1713], where the stated value is the median, and the range in square
0.07
λs
0.06
λw
0.05
0.04
EPDF
0.03
0.02
0.01
0
−0.4 −0.2 0 0.2 0.4 0.6 0.8 1
λ
90 0.2 90
120 60 120 60
0.04
150 0.1 30 150 30
0.02
180 0 180 0
210 330 210 330
240 300 240 300

270 270
θs θw
Fig. 9.5 Typical empirical probability density functions (EPDFs) for the local stability analysis,
obtained via bootstrapping for one subject: (top) eigenvalues λw and λs ; (bottom) angles θw and θs
made by the eigenvectors êw and ês , respectively, with the normal n̂ to the gold equivalent manifold
(GEM) (see Fig. 9.4 for angle definitions). In the angular EPDFs, the radial direction is the relative
frequency, the circumferential direction is the angle, and the italicized numbers represent the relative
frequency
brackets is the 90 % CI. The corresponding median distance from the GEM in the
body space was δ = −0.0118 [−0.0221, 0.0046]. Thus the aggregate performance
displayed, roughly speaking, about as much overshooting (e > 0) as undershooting
(e < 0), though with some indication of a small bias toward undershooting. This
small bias is more apparent in the stopping distance, since the CI shows a significant
bias toward trials “under” the GEM curve (δ < 0). Nevertheless, these overall results
are consistent with the assumption that skilled performance would involve operation
close to the GEM.
In Fig. 9.5 we show typical results from the bootstrapped local stability analysis
using the data from all 250 trials for one subject. Figure 9.5 (top) displays the EPDFs
obtained for the eigenvalues, λw and λs , of the matrix A (Eq. 9.14); in Fig. 9.5
(bottom) are the EPDFs for the angles, θw and θs that the eigenvectors êw and ês
make with n̂, the unit normal to the GEM (Fig. 9.4). Observe that the displayed
EPDFs are unimodal and symmetric about the median, so that the mean, median,
and mode are close to equal: this was found to be true for all bootstrapped distributions
0.8
100
(degrees)
0.6
50
0.4
0.2 0
0
λw λs θw θs
Fig. 9.6 Aggregate geometric stability results: magnitude of mean eigenvalues , λw and λs , obtained
from each subject’s bootstrapped distributions (left); mean eigenvector angles θ w and θ s (right) . The
median value of λw (0.5597) was significantly larger than that of λs (0.1076), with p = 3.05×10−5 .
The aggregate median value of θ w (85.83◦ ) was significantly larger than that of θ s (20.33◦ ), with
p = 3.05 × 10−5 . Since λw was positive for all subjects, these results confirm hypotheses H1 and
H2: 0 ≈ |λs |
λw and θs
θw ≈ 90◦
generated for this study. For this subject, we find λs = −0.1039 [ − 0.2191, 0.0143]
and λw = 0.4947 [0.3997, 0.5871], so that 0 ≈ |λs |
λw as hypothesized. The fact
that λs < 0 indicates that this subject slightly overcorrected deviations transverse
to the GEM, suggesting a slightly suboptimal controller (Dingwell et al. 2010). We
also find θw = 93.62◦ [73.68◦ , 119.74◦ ], indicating that êw is nearly tangent to the
GEM, whereas θs = 20.78◦ [16.82◦ , 24.45◦ ]. Thus, again as hypothesized, we find
θs
θw ≈ 90◦ .
Aggregate geometric stability results across all subjects are shown in Fig. 9.6. To
compare eigenvalues, we look at their absolute values to directly compare the relative
strength of control without regard to possible overcorrection (λ < 0) or undercorrec-
tion (λ > 0). Note, however, that for all subjects it was found that the bootstrapped
distribution of λw was strictly positive, as in Fig. 9.5. As a comparison statistic, for
each subject, we used the mean value of the eigenvalues and angles obtained from
each subject’s bootstrapped EPDF. In Fig. 9.6 (left) are box plots showing the dis-
tribution of |λs | and |λw | ≡ λw , whereas Fig. 9.6 (right) displays box plots for the
corresponding mean eigenvector angles θ w and θ s . Applying the two-sided sign test
showed that the aggregate median value of λw (0.5597 [0.3026, 0.9194]) was signif-
icantly larger than that of λs (0.1076 [0.0141, 0.3564]), with p = 3.05 × 10−5 . It
was also found that the aggregate median value of θ w (85.83◦ [53.72◦ , 115.59◦ ])
was significantly larger than that of θ s (20.33◦ [ − 13.39◦ , 33.35◦ ]), again with
p = 3.05 × 10−5 . Since λw > 0 in all cases, we find that 0 ≈ |λs |
λw and
θs
θw ≈ 90◦ . Taken together, these results confirm hypotheses H1 and H2.
The persistence properties of the inter-trial fluctuations, computed using DFA,
are presented in Fig. 9.7. For each subject, the average matrix A was obtained from
bootstrapping, and the fluctuations ξ were projected onto the average weakly and
strongly stable directions using the transformation Eq. (9.15). The DFA algorithm
was then applied to the resulting ζw and ζs fluctuation components. We found that the
1.5
α
0.5
0
ζw ζs
Fig. 9.7 Aggregate results for the detrended fluctuation analysis (DFA) exponent α, for fluctuations
in the weakly and strongly stable directions, ζw and ζs , respectively. For ζw , we found αw =
0.9433 [0.7500, 1.2950] indicating statistical persistence consistent with weak control (Dingwell
and Cusumano 2010). For ζs we found αs = 0.4948 [0.2773, 0.6863], indicating approximately
uncorrelated fluctuations, consistent with strong control. The difference between the median values
of α in both cases was significant (p = 3.06 × 10−5 ). These results confirm hypothesis H3:
0.5 ≈ αs
αw
ζw fluctuations were statistically persistent, with an αw = 0.9433 [0.7500, 1.2950],

suggesting weak control in the weakly stable direction, which Fig. 9.6 shows is nearly
tangent to the GEM. In contrast, for ζs , we found αs = 0.4948 [0.2773, 0.6863],
indicating approximately uncorrelated fluctuations. This is consistent with strong
control that suppresses substantial deviations from the GEM, leaving only noise.
The difference between the median values of α in both cases was significant (p =
3.06 × 10−5 ). Thus, 0.5 ≈ αs
αw , confirming hypothesis H3.
Taken together, these results demonstrate that subjects exhibited strong dynamical
anisotropy around the GEM, precisely as anticipated. This supports the claim that
inter-trial error correction for the shuffleboard task is organized around a “GEM
aware” generalized MIP controller.
9.7 Comparison with “Static” Variability Analysis
In this section, we contrast our dynamical analysis with an approach that does not
consider the temporal order of the data, but instead analyzes only the “static” inter-
trial variability structure. Here, we do this by applying principal component analysis
(PCA) (Mardia et al. 1979; Daffertshofer et al. 2004) to the ensemble of body states.
PCA calculates a new set of coordinates for the data set such that the covariance
between the coordinates is zero. A principal component (PC) is an eigenvector of
the data’s covariance matrix, and its associated eigenvalue, the principal value (PV)
100 270
80 225
% total variance
θPCA (degrees)
180
60
135
40
90
20 45
0 0
λ1 λ2 θ1 θ2
Fig. 9.8 Aggregate results of principal component analysis (PCA) applied to each subject’s Inter-
Trial data: (left) percent of total variance explained by each PV, λ1 and λ2 ; (right) the angle, θ1
and θ2 , between each PC and n̂ (Fig. 9.2). We see that the first principal value (PV ) explains about
80 % of the total variance (81.66 % [61.66 %, 95.58 %]), however, both PCs are approximately the
same angular distance (about 27◦ ) from the tangent, t̂, and hence do not locate the goal equivalent
manifold (GEM) in body space
indicates how much variance lies along that direction. Thus, each PV represents
a fraction of the total variance that lies in the direction of its associated PC. With
2D data, as was collected for this study, a possible interpretation of PCA results
is that the first PC, which has the larger PV and, therefore, the greater variability,
indicates a weakly-controlled direction because fluctuations are larger along it. The
second, smallest variance PC then indicates a strongly-controlled direction. If true,
that would imply that variance analysis alone could identify the organization of inter-
trial variability about a candidate GEM. We here show that this interpretation does
not hold, in general.
Applying PCA to our fluctuation data, we obtain the aggregate results shown in
Fig. 9.8. We labeled the two PVs as λ1 and λ2 , where λ1 > λ2 , so that the associated
PCs are the unit vectors p̂1 and p̂2 . We then computed the angles that each PC made
with the unit normal to the GEM, as was done for Fig. 9.6. We see that λ1 accounted for
81.66 % [61.66 %, 95.58 %] of the total variance, whereas λ2 accounted for 18.34 %
[4.416 %, 38.23 %]. We also see that θ1 = 27.23◦ [7.714◦ , 116.6◦ ] and θ2 = 117.2◦
[97.71◦ , 206.6◦ ]. Thus, neither PC was close to either n̂ or t̂, and so they did not help
locate the GEM. In particular, the angular distance of the first and second PCs from
the GEM tangent was roughly 63◦ and 27◦ , respectively, thus neither could be said
to identify a weakly controlled direction.
To further explore this issue, we transformed the data from one subject using a
coordinate transformation that rendered the variability isotropic, for which, that is,
the variance ellipse would be a circle. This is easily accomplished using a linear
transformation similar to Eq. (9.15), however instead of constructing the matrix P
by using the eigenvectors êw and ês as its columns, we use the eigenvectors of the
covariance matrix, that is, the PCs, p̂1 and p̂2 . This projects the original fluctuations
ξ onto the principal directions. After this step, each coordinate was normalized to
unit variance, giving fluctuations in rescaled principal coordinates, q = (q1 , q2 )T .
8
1
q2
v 0.9
4
0.8
0.1 0.2 16 18 20
x q1
Fig. 9.9 Illustration showing the effect of a coordinate transformation that renders the variability
ellipse circular: (left) data from one subject (light gray dots) together with the weakly and strongly
stable eigenvectors, êw and ês (black arrows), for a matrix A estimated via regression during a single
iteration of the bootstrap; (right) the same ensemble (gray dots) after transforming into rescaled
principal coordinates (q1 , q2 ) that have isotropic variance. In the right plot, the eigenvectors were
calculated from a new matrix A estimated from the transformed data. Also shown in both figures is
the goal equivalent manifold (GEM) (dark gray line) in the local coordinates, and the variance ellipse
(dashed line). We observe that while the variability is radically changed by the transformation, the
local geometry of the inter-trial dynamics is fundamentally preserved
Figure 9.9 shows the effect of this transformation on a typical geometric stability
analysis, carried out using the data from one subject. In Fig. 9.9 (left), we see the
results for the data in the original position–velocity fluctuation coordinates. For this
figure, the eigenvectors êw and ês were computed from a single, typical matrix A, as
would be estimated during a single bootstrap iteration. Also included in the figure
is the standard variance ellipse, which, as would be expected from the results of
Fig. 9.8, is not particularly well aligned with the GEM. However, the weakly and
strongly stable subspaces are oriented in a way that helps understand how control
is organized around the GEM. In Fig. 9.9 (right), we see the same data, however
transformed into (q1 , q2 ) coordinates, together with the eigenvectors. These eigen-
vectors were not merely transformed from Fig 9.9 (left), rather, the transformed data
were used to estimate a new matrix A, and a new eigenanalysis was then performed.
In the transformed picture, it is clear that the variability cloud provides no useful
information about the location of the GEM since, by design, its variability ellipse
is circular. However, the new eigenvectors continue to define weakly and strongly
stable subspaces that correctly demonstrate how control is organized about the GEM.
9.8 Discussion
In this chapter, we have described a dynamical approach to the experimental anal-

ysis of inter-trial variability in skilled movement tasks. Our approach combines
consideration of task manifolds, in the particular form of GEMs, with optimal con-
trol concepts, to formulate experimentally testable models of inter-trial fluctuations.
Because of this fundamentally dynamical perspective, we are not forced to make

inferences about control based solely on the “static” structure of variability observed
in some body-level state space, but rather can characterize the way variability, at
both the body and goal levels, is organized and generated in time.
While the general GEM-based framework discussed in this chapter can be used
to create mathematical models suitable for theoretical analysis, our approach is phe-
nomenological and experimentally oriented. Thus, the primary significance of the
theoretical discussion presented here is that it allows us to arrive at a set of quite
generic hypotheses on the structure of variability observed with skilled performers:
namely, that we expect to see strong evidence of dynamic anisotropy in the inter-
trial fluctuations near a GEM. Furthermore, while motivated by both the MIP and
the UCM hypothesis, consideration of our relatively simple mathematical models
in the light of available experimental evidence leads us to the conclusion that nei-
ther can be true in an ideal form, since “perfect” MIP controllers would result in
nonlocalized fluctuations that drift randomly along the GEM. Thus, our hypotheses
include the assumption of a generalized MIP for which control along the GEM is
not zero, but weak. This seemingly minor observation has an important experimental
consequence: while this weakly controlled direction might be expected to be nearly
tangent to the GEM, it will not, in general, be perfectly so.
To illustrate the theory, as well as its experimental implementation, we applied
our analysis approach to study the variability observed in a virtual shuffleboard ex-
periment. We found that all of our general hypotheses regarding dynamic anisotropy
were supported by the analysis results. In particular, we found that subjects exhibited
strong control of fluctuations in a direction transverse, but not typically normal, to
the GEM, whereas they only weakly controlled fluctuations in a direction nearly, but
not exactly, tangent to it. The assessment of control “strength” in these two directions
was not determined by the relative magnitude of variances, but was provided both
by local stability analysis and DFA. In both of these cases, the differences between
the weakly and strongly stable directions were found to be as predicted and highly
significant.
We compared our dynamical analysis with a non-temporal, variance-based anal-
ysis carried out using PCA. We found that the orientation of the variability cloud
did not help identify how inter-trial control was organized around the GEM. Further-
more, we showed that our dynamical approach is robust to coordinate changes that
result in isotropic variability. Under such circumstances, variability alone provides
no information about the GEM, whereas the orientation of weakly and strongly stable
subspaces still helps explain how variability is structured around a GEM, as well as
how control is temporally organized.
The theoretical discussion in this chapter, together with the experimental demon-
stration that the weakly stable directions are not exactly directed along the GEM,
suggest that while goal-level error minimization is likely a dominant cost being min-
imized by the human motor system, it is not the only one. In particular, our work
suggests that there are relatively small additional costs—such as might be related to
ergonomic, physiological, or even psychological needs of the performer—that need
to be considered in combination with error reduction so that we might completely
characterize experimentally-observed motor variability. It is hoped that dynamical

data analysis methods like those presented here might help us to identify such costs
in future experiments.
Acknowledgements Partial funding for this project was provided by National Institutes of Health
grant 1-R03-HD058942-01 (to JBD) and by National Science Foundation grant 0625764 (to JPC).
References
Burge J, Ernst MO, Banks MS. The Statistical Determinants of Adaptation Rate in Human Reaching.
Journal of Vision. 2008;8(4):1–19.
Cohen R, Sternad D. Variability in Motor Learning: Relocating, Channeling and Reducing Noise.
Experimental Brain Research. 2009;193(1):69–83.
Cusumano JP, Cesari P. Body-Goal Variability Mapping in an Aiming Task. Biological Cybernetics.
2006;94(5):367–379.
Cusumano JP, Dingwell JB. Movement Variability Near Goal Equivalent Manifolds: Fluctuations,
Control, and Model-based Analysis. Human Movement Science. 2013;32(5):899–923.
Daffertshofer A, Lamoth CCJ, Meijer OG, Beek PJ. PCA in Studying Coordination and Variability:
A Tutorial. Clinical Biomechanics. 2004;19(4):415–428.
Delignières D, Torre K. Fractal Dynamics of Human Gait: A Reassessment of the 1996 Data of
Hausdorff et al. Journal of Applied Physiology. 2009;106(4):1272–1279.
Diedrichsen J, HashambhoyY, Rane T, Shadmehr R. Neural Correlates of Reach Errors. The Journal
of Neuroscience. 2005;25(43):9919–9931.
Dingwell JB, Cusumano JP. Re-Interpreting Detrended Fluctuation Analyses of Stride-to-Stride
Variability in Human Walking. Gait & posture. 2010;32(3):348–353.
Dingwell JB, John J, Cusumano JP. Do Humans Optimally Exploit Redundancy to Control Step
Variability in Walking? PLoS Computational Biology. 2010;6(7):e1000856.
Dingwell JB, Smallwood RF, Cusumano JP. Trial-to-trial Dynamics and Learning in a Generalized,
Redundant Reaching Task. Journal of Neurophysiology. 2013;109(1):225–237.
Efron B, Tibshirani RJ. An Introduction to the Bootstrap. vol. 57 of CRC Monographs on Statistics
& Applied Probability. Boca Raton, FL: Chapman & Hall; 1994.
Eldar A, Elowitz MB. Functional Roles for Noise in Genetic Circuits. Nature. 2010;467(7312):
167–173.
Faisal AA, Selen LPJ, Wolpert DM. Noise in the Nervous System. Nature Reviews Neuroscience.
2008;9(4):292–303.
Freedman DA. Bootstrapping Regression Models. The Annals of Statistics. 1981;9(6):1218–1228.
Gao J, Hu J, Tung WW, Cao Y, Sarshar N, Roychowdhury VP. Assessment of Long-Range
Correlation in Time Series: How to Avoid Pitfalls. Physical Review E. 2006 January;73:016117.
Golub GH, van Loan CF. Matrix Computations. Baltimore, MD: The John Hopkins University
Press; 1996.
Greenwood DT. Principles of Dynamics. Upper Saddle River, NJ: Prentice-Hall; 1988.
Guckenheimer J, Holmes P. Nonlinear Oscillations, Dynamical Systems, and Bifurcations of Vector
Fields. vol. 42 of Applied Mathematical Sciences. New York: Springer-Verlag; 1997.
Harris CM, Wolpert DM. Signal-Dependent Noise Determines Motor Planning. Nature.
1998;394(6695):780–784.
Hausdorff JM, Peng C, Ladin Z, Wei JY, Goldberger A. Is Walking a Random Walk? Evidence
for Long-range Correlations in Stride Interval of Human Gait. Journal of Applied Physiology.
1995;78:349–349.
Hirsch MW, Smale S, Devaney RL. Differential Equations, Dynamical Systems, and an Introduction
to Chaos. San Diego: Academic Press; 2004.
John J, Cusumano JP. Inter-Trial Dynamics of Repeated Skilled Movements DETC2007-35380.

ASME 2007 International Design Engineering Technical Conferences. 2007;p. 707–716.
Latash ML, Scholz JP, Schöner G. Motor Control Strategies Revealed in the Structure of Motor
Variability. Exercise & Sport Sciences Reviews. 2002;30(1):26–31.
Maraun D, Rust HW, Timmer J. Tempting Long-Memory: on the Interpretation of DFA Results.
Nonlinear Processes in Geophysics. 2004;11(4):495–503.
Mardia KV, Kent JT, Bibby JM. Multivariate Analysis. London: Academic Press; 1979.
McDonnell MD, Ward LM. The Benefits of Noise in Neural Systems: Bridging Theory and
Experiment. Nature Reviews Neuroscience. 2011;12(7):415–426.
Mooney CZ, Duval RD. Bootstrapping: A Nonparametric Approach to Statistical Inference. vol. 95.
Newbury Park, CA: Sage Publications; 1993.
Moore EH, Barnard RW. General Analysis. Philadelphia, PA: American Philosophical Society;
1939.
Müller H, Sternad D. Decomposition of Variability in the Execution of Goal-Oriented Tasks: Three
Components of Skill Improvement. Journal of Experimental Psychology: Human Perception
and Performance. 2004;30(1):212–233.
Oldfield RC. The Assessment and Analysis of Handedness: The Edinburgh Inventory. Neuropsy-
chologia. 1971;9(1):97–113.
Osborne LC, Lisberger SG, Bialek W. A Sensory Source for Motor Variation. Nature.
2005;437(7057):412–416.
Peng CK, Buldyrev SV, Goldberger AL, Havlin S, Sciortino F, Simons M, et al. Long-Range
Correlations in Nucleotide Sequences. Nature. 1992;356(6365):168–170.
Poole D. Linear Algebra: A Modern Introduction. Boston, MA: Brooks/Cole; 2010.
Ranganathan R, Newell K. Motor Learning Through Induced Variability at the Task Goal and
Execution Redundancy Levels. Journal of Motor Behavior. 2010;42(5):307–316.
Scholz JP, Schoner G. The Uncontrolled Manifold Concept: Identifying Control Variables For A
Functional Task. Experimental Brain Research. 1999;126(3):289–306.
Scholz JP, Schoner G, Latash ML. Identifying the Control Structure of Multijoint Coordination
During Pistol Shooting. Experimental Brain Research. 2000;135(3):382–404.
Schöner G, Scholz JP. Analyzing Variance in Multi-Degree-of-Freedom Movements: Uncovering
Structure Versus Extracting Correlations. Motor Control. 2007;11(3):259–275.
Scott SH. Optimal Feedback Control and the Neural Basis of Volitional Motor Control. Nature
Reviews Neuroscience. 2004;5(7):532–546.
Stein RB, Gossen ER, Jones KE. Neuronal Variabiltiy: Noise or Part of the Signal? Nature Reviews
Neuroscience. 2005;6(5):389–397.
Sternad D, Abe MO, Hu X, Müller H. Neuromotor Noise, Error Tolerance and Velocity-Dependent
Costs in Skilled Performance. PLoS Computational Biology. 2011;7(9):e1002159.
Todorov E. Optimality Principles in Sensorimotor Control. Nature Neuroscience. 2004;7(9):
907–915.
Todorov E, Jordan MI. Optimal Feedback Control as a Theory of Motor Coordination. Nature
Neuroscience. 2002;5(11):1226–1235.
van Beers RJ. Motor Learning Is Optimally Tuned to the Properties of Motor Noise. Neuron.
2009;63(3):406–417.
Warren WH. The Perception-Action Coupling. In: Bloch H, Bertenthal BI, editors. Sensory-Motor
Organizations and Development in Infancy and Early Childhood. Dortrecht, The Netherlands:
Kluwer Academic; 1990. pp. 23–37.
Warren WH. The Dynamics Of Perception and Action. Psychological Review. 2006;113(2):
358–389.
Wilcoxon F. Individual Comparisons by Ranking Methods. Biometrics Bulletin. 1945;1(6):80–83.
Chapter 10
Motor Control in Action: Using Dance to Explore
the Intricate Choreography Between Action
Perception and Production in the Human Brain
Emily S. Cross and Anastassia Elizarova
10.1 Introduction
Cue the crescendo: An audience watches as a ballerina takes center stage and pre-
cisely executes 32 fouettés en tournant—en pointe. Each revolution is performed
with athletic deftness, spatial precision, and corporeal poise. Underlying the smooth
delivery, the dancer’s mind and body reach an apex: maintaining balance with arched
feet, pointed toes, grounded core muscles, and postural stability. She generates im-
petus for the turns while coordinating her arms, shoulders, and hands, to appear
light and effortless, with a smile on her face and perfect synchronization with the
music. Complex coordination in a dance context extends well beyond the movement
vocabulary of classical ballet dancers. It is cultivated among the many cultures of
the world and their respective dance traditions. For the purposes of the present chap-
ter, we define dance as bodily movements—whether choreographed or impromptu,
rhythmic or asymmetric, with or without musical accompaniment—as a medium for
artistic expression for individuals or groups. The ubiquity and purpose of dance have
been examined in scientific contexts for over a century. First postulated by Darwin
and corroborated by recent research, the origins of dance are speculated to be dis-
plays of courtship and mate selection, including indicators of symmetry, testosterone
exposure in males, and synchrony with partner(s) (Fusani 2008; Shuster 2009).
Scientific investigation of dance can illuminate more than just its origins or the
evolutionary function of coordinated movement, however. Of interest in the present
chapter is how the behavioral and brain sciences are using dance paradigms in exper-
imental contexts to better understand the neurophysiological substrates supporting
E. S. Cross () · A. Elizarova

Wales Institute for Cognitive Neuroscience, School of Psychology, Bangor University,
Adeilad Brigantia, Bangor, Gwynedd LL57 2AS, Wales
Behavioural Science Institute and Donders Institute for Brain,
Cognition and Behaviour, Radboud University Nijmegen, Adeilad Brigantia, Nijmegen,
Gwynedd LL57 2AS, The Netherlands

and Biology 826, DOI 10.1007/978-1-4939-1338-1_10
148 E. S. Cross and A. Elizarova
links between the physical or affective experience of an observer and how the ob-
server perceives a movement. We have chosen to focus on two particular points of
convergence between dance and neuroscience, namely (1) how dance informs our
understanding of action expertise, and (2) how dance advances our understanding of
neuroaesthetics.
Nearly as many dance styles and traditions exist as there are human cultures: from
the highly stylized, minimalist movements of Kabuki dancers in Japan to the rigorous,
structured movement vocabulary of classical ballet, or the dramatized isolation of the
eye muscles, neck, and knuckles in the Indian dance form of Kathakali, to the whole-
body, poetic articulation of the dances of the Yoruba people in West Africa, humans
all over the world create, perform, and watch dance. Only recently have scientists
turned to this art form as a resource for research into motor control and aesthetics. In
the first section of this chapter, we consider how early neurophysiological work with
nonhuman primates gave rise to the notion of neural mechanisms that link action with
perception, and how recent work with dancers has helped to refine our knowledge
of how physical experience and expertise shape perception. In the second section,
we delve into the realm of empirical aesthetics, specifically into the budding field of
neuroaesthetics, to explore how this nascent field seeks to link subjective appraisals
of stimuli with the neurobiological foundations that support such appraisals. Here as
well, we highlight how research with dance and dancers is helping to move forward
our understanding of the biological substrates of action perception and appreciation.
10.2 How Dance Informs Research Into Action Expertise
A question that has been of particular interest to philosophers for centuries, and to
psychologists and neuroscientists ever since the advent of their fields, is how we are
able to integrate physical and perceptual experiences to learn new movements. Such
abilities are quite pronounced in professional dancers, whose livelihoods depend on
rapid and adept movement, reproduction, and refinement. Before discussing how
research with dancers advances our understanding of complex action learning and
perception, it is first necessary to ground these questions within a broader, histori-
cal context. With the advent of cognitive psychology in the late nineteenth century,
William James introduced the idea that action and perception processes likely con-
verge or overlap in the human mind (James 1890). Neurophysiological investigation
into the ability of the brain to use perceptual information to shape movement began
in the mid-twentieth century, when information-processing explanations proposing
complex transformations from perception to the organization and execution of action
gathered momentum (Welford 1968; Sanders 1967, 1983; Mountcastle et al. 1975;
Massaro and Friedman 1990). In the mid-1970s, Mountcastle and colleagues began
the first investigations into putative neurophysiological mechanisms linking action
with perception (Mountcastle 1975; Mountcastle et al. 1975; Yin and Mountcastle
1977). Through these seminal studies, Mountcastle and colleagues determined that
neurons in parietal cortex respond to visual cues associated with specific actions.
10 Motor Control in Action: Using Dance to Explore . . . 149
Mountcastle and colleagues suggested that parietal cortex is a likely candidate for
cross-modal convergence of action and perception, and established the foundation
for the next several decades of further research into how these modalities interact
(Andersen et al. 2004).
In the 1990s, research into the interface between action perception and production
experienced an extraordinary surge in interest that was sparked by the discovery of
so-called mirror neurons within the ventral premotor cortex of the macaque monkey.
These neurons fired in a similar manner both when a monkey performed an action and
when it observed another monkey or a human perform the same action (di Pellegrino
et al. 1992; Gallese et al. 1996; Rizzolatti et al. 1996). A similar firing pattern when
perceiving or performing actions was subsequently observed in the inferior parietal
lobule (Fogassi et al. 2005), which suggested that perceiving others’ actions engages
a similar fronto-parietal circuit as executing actions.
One dominant explanation of these response profiles is tied to the notion of
motor simulation (Jacob and Jeannerod 2005; Jeannerod 2001). The concept of
motor simulation is consistent with other forms of mental simulation, which involve
the reenactment of mental processes associated with previously experienced states
(Barsalou 2008). Such states can be perceptual, motoric, or introspective. Hence, mo-
tor simulation is an instance of a more general concept of mental simulation (Jacob
and Jeannerod 2005). Many researchers implicitly or explicitly hold that the mirror
system is a key neural substrate for processes of motor simulation (e.g., Decety and
Grezes 1999, 2006). In addition, motor simulation accounts of action comprehension
have proposed that perceiving and understanding the action of another might involve
the observer’s brain simulating observed movements using their own motor system
(Fadiga et al. 1995, 1999; Grafton et al. 1996; Rizzolatti et al. 2001). Thus, it is
argued that through a process of motor simulation, action comprehension occurs by
engaging specific parts of the motor system that would be used to perform the same
action that is being observed (Rizzolatti et al. 2001).
Interest in mirror neurons has sparked hundreds of studies and an ongoing de-
bate among researchers regarding the specific parameters, scope, and limitations of
a possible action simulation system within the human brain (Gallese et al. 2011;
Gallese and Sinigaglia 2011). While an exhaustive review of the past two decades
of research performed in this domain is beyond the scope of this chapter (for re-
views, see Grosbras et al. 2012; Molenberghs et al. 2012; Rizzolatti and Sinigaglia
2010), what a burgeoning corpus of research demonstrates is that neural tissue found
within parietal and premotor cortices of the human brain is engaged when actions
are performed and when they are observed. This work has given rise to the notion
of a human mirror system, which consists of multiple cortical regions and shows
evidence for behaving in a similar manner to individual mirror neurons found within
nonhuman primate brains (Molenberghs, et al. 2012). A schematic representation of
the human mirror system is depicted in Fig. 10.1.
The implications of a neural link between perception and action are broad-reaching
and have been explored by scientists from disciplines and perspectives beyond neu-
rophysiology and cognitive neuroscience, including the sport sciences, cognitive
psychology, philosophy, and physical and occupational therapy (Johnson-Frey 2004;
Fig. 10.1 Schematic representation of the human mirror system. The core nodes of the human
mirror system are bilateral ventral premotor cortex (PMC) and inferior parietal lobule (IPL). While
other regions within the human brain have been reported to show mirror-like properties, the strictest
definition of the human mirror system includes only those brain regions where so-called mirror
neurons, or cells that respond when performing an action or watching another perform an action,
have been reported through neurophysiological investigations in the nonhuman primate brain
Braun et al. 2006; Jacob 2008; Goldman 2009; Yarrow et al. 2009; Guillot and Collet
2010; Moran et al. 2012). Moreover, some researchers have found that experimental
paradigms that borrow from the arts, namely dance, offer an ideal means for ad-
vancing knowledge of how action and perception are linked (Bläsing et al. 2012).
Paradigms incorporating dance training or the study of professional dancers have
shed light on how motor expertise is manifest in the human brain, as evidenced
by the two original studies to approach questions of action expertise using expert
dancers as participants.
The first study to investigate how physical expertise with a particular dance style
shapes neural activity when watching dance was conducted by Calvo-Merino et al.
(2005). The authors studied a group of expert male capoeira dancers, expert male
ballet dancers, and a group of age and sex-matched control participants. In this
functional magnetic resonance imaging (fMRI) experiment, participants from all
three groups watched 12 different 3-s videos depicting capoeira movements, and
12 different 3-s videos depicting ballet movements. The task was to indicate via a
button press how tiring they thought each movement was. The authors found strong
evidence in support of the idea that one’s individual motor repertoire strongly shapes
perception of others in action. Specifically, they found that when ballet dancers
watched ballet or when capoeira dancers watched capoeira, they showed greater
activity across brain regions associated with the mirror system, including left dorsal
and ventral premotor cortices and posterior superior temporal sulci and bilateral
intraparietal sulci. This finding is illustrated by the parameter estimates that quantify
the relative amplitude of brain responses to watching the different kinds of dance
videos among the three groups of participants. Crucially, no differentiation between
the dance styles was seen in these regions within the brains of control participants.
The authors concluded that long-held motor expertise markedly shapes perception,
such that brain regions implicated in linking up performed with perceived actions are
more active when an observer watches a particular movement he or she is physically

familiar performing.
The study by Calvo-Merino et al. (2005) forged new territory in understanding
how highly complex action and perception are linked in a number of ways. First, the
data suggested that the mirror system codes complete action patterns, as opposed to
individual movements. This was evidenced by the fact that movements from each
dance style contained similar elements (such as swinging arms or jumping), yet
the brains of capoeira and ballet dancers responded preferentially to their embodied
movement style. The authors interpret this as evidence that neural architecture linking
action with perception is sensitive to abstract levels of organization (such that two
different dance styles can be distinguished). The findings are also important, in
that they demonstrate the sensitivity of mirror system representations to learned
movements, as opposed to innate action patterns that might be performed more or
less by different groups of people.
A subsequent experiment more closely investigated this latter point in a longi-
tudinal training study performed with a company of contemporary dancers (Cross
et al. 2006). In this study, the authors scanned the brains of ten expert contemporary
dancers once a week across 5 weeks of the rehearsal process as they learned a new
piece of choreography that was rhythmically complex and lasted 25 min. The dancers
rehearsed the new choreography for 4–5 days each week, and on each weekend, they
underwent fMRI while watching 18 short-movement sequences from the work they
were learning, or 18 kinematically similar control movements (all performed by the
same dancer). While the dancers were in the scanner, their task was to watch each
movement, imagine themselves performing it, and to evaluate how well they could
perform each movement segment. At the end of each video clip, dancers reported on
a 1–4 scale how well they thought they could physically perform the sequence they
just watched.
The authors found that across the testing sessions, how the dancers rated their
performance ability for the rehearsed movements significantly increased while their
performance ability ratings for the (unrehearsed) control movements did not change.
In terms of the neuroimaging findings, when the dancers watched movement that they
had physically practiced compared to the kinematically similar control movement,
more activity was seen within bilateral mirror system regions. This finding provided
a clear replication and extension of what was reported by Calvo-Merino et al. (2005)
with the ballet and capoeira dancers, demonstrating that de novo learning also results
in marked differences in how similar movements with different experiential profiles
are perceived. The most novel finding from this study, however, is what the authors
found when they ran a parametric analysis querying brain regions whose activity
increased with higher ratings of performance proficiency. This analysis revealed
two small foci—left inferior parietal lobule and left ventral premotor cortex, which
showed the strongest responses when dancers watched the movements that they were
(individually) best at performing. What this finding suggests is that the better a dancer
is at performing a movement he or she is watching, the more he or she simulates
the observed action, and core mirror system regions within the left hemisphere are
engaged by such processing.
These two highlighted studies using expert dancers to explore the neural signatures
of physical expertise have been joined by a small but growing number of subsequent
studies also using dancers to further knowledge on the impact of expertise on the
brain (cf. Calvo-Merino et al. 2006; Orgs et al. 2008; Fink et al. 2009; Hänggi et al.
2009). Taken as a whole, these studies demonstrate how combining brain-imaging
approaches with populations in possession of a highly specialized and skilled motor
repertoire can advance our understanding of how action perception and performance
are linked at behavioral and brain levels. However, the utility of dancers and dance
paradigms for understanding how we perceive others is not limited to questions of
motor expertise. In the following section, we explore how the dance domain can
inform other aspects of the psychological and brain sciences, such as our affective
appraisal of the movements of others.
10.3 How Dance informs Understanding of Aesthetics

at Behavioral and Neural Levels
Another way in which research with dancers and the use of dance-based experimental
paradigms has advanced understanding of how the brain perceives others in action is
via empirical aesthetics. Classically, the study of aesthetics is associated with scholars
in the humanities, such as philosophers, visual artists, and performing artists. For
this reason, it is useful to define what, precisely, we mean by “aesthetics” when
we discuss it in a scientific context. In consideration of the breadth of meanings
and ambiguities the term “aesthetics” may carry, Brown and Dissanayake (2009)
suggested that “aesthetics” has been used in two ways. In one sense, “the aesthetic”
has been used to account for emotional responses that emerge when perceiving
works of art, wherein art works (i.e., paintings, sculptures, sketches, dances, music)
and the arts (i.e., corresponding activities or expressions) are fundamental to this
usage. Another use of “aesthetics” is in reference to any system of valuation for the
appreciation of beauty; the sensory preferences that humans and/or animals hold,
with positive emotional responses towards objects resulting from selective attention
(Orians 2001; Voland and Grammer 2003).
For this chapter, we will consider the term “aesthetics” as a heuristic for consid-
ering the experiences that result from the perception, appreciation, and/or creation
of an artistic work. Aesthetics are studied and appreciated across myriad contexts
including (but not limited to) faces, landscapes, music, dance, sculpture, food, paint-
ings, mate selection, and machinery (Jacobsen 2006). For thousands of years, the
only way in which aesthetic processing could be studied was to directly ask an in-
dividual about his or her experience with a stimulus, or observe their overt behavior
toward a stimulus (Jacobsen et al. 2004, 2006). With the advent of cognitive neu-
roscience, new tools and techniques have enabled scholars from the psychological
and brain sciences to explore the neurophysiological underpinnings of aesthetic per-
ception. As such, neuroaesthetics offers another means of investigation into one’s
experience of the sublime. By simply being in the world, humans are in a constant
state of aesthetic appraisal of ordinary objects, people, and experiences. The domain
of neuroaesthetics aims to shed light on the nature of the aesthetic experience; that
is, the relationship between the observer and the object, and the aesthetic processes
that coalesce to create a rich relationship between stimulus and perceiver. As neu-
roaesthetics is a relatively new discipline, only a small number of empirical studies
yet explored the affective relationship between an observer and a stimulus. More-
over, most of the work to date has focused on aesthetic responses to (static) visual
artworks or music (cf. Cinzia and Gallese 2009; Blood and Zatorre 2001), with just
a handful of studies that have explored the aesthetic responses in brain and behavior
to dance. To place the dance neuroaesthetics research within a broader context, we
first consider what we learn from the wider literature on empirical aesthetics and
neuroaesthetics.
When a spectacle and its spectator meet, what the spectator beholds can vary
from a fleeting sublimity to a sense of humdrum to a mere, neutral acknowledg-
ment of an artwork. Each experience is, of course, personal to the beholder; one’s
personal preferences, expertise, and exposure, for example, may reflect in the eval-
uative judgments that follow, giving rise to inter-individual variability (Palmer et al.
2013). Underlying the aesthetic processing of a stimulus is an assortment of neural
processing, including (but not limited to) activity within the medial orbitofrontal cor-
tex (mOFC), a brain region associated with decision making, affective judgments,
and expectation (Kawabata and Zeki 2004; Vartanian and Goel 2004; Kringelbach
2005). It thus appears that beauty is not so much in the eye, but instead the brain, of
the beholder. In one recent study (Ishizu and Zeki 2011), volunteers drawn from a
range of cultural/ethnic backgrounds were asked to rate a series of works—paintings
or segments of music—as belonging to the categories of “beautiful”, “neutral,” or
“ugly” while undergoing fMRI. The goal of the study was to determine where in the
brain cortical activity correlated with experiences that participants rated as beautiful
or ugly. The most striking finding from this study was that a portion of the mOFC
was active when participants experienced a musically or visually beautiful stimulus.
Of note, the intensity of activation within this region was correlated with ratings of
the intensity of beauty. This region is thought to make rapid assessments as to the
reinforcing properties of a stimulus, offering one possible pathway between reward,
judgment and aesthetic pleasure upon perception. The authors concluded by offering
a brain-based theory of beauty wherein the mOFC codes the aesthetic or rewarding
value of a stimulus, regardless of its modality.
When further considering the relationship between the observed and the observer
in an artistic context, a number of scholars have argued that understanding the role of
empathy is crucial. Robert Vischer (1873; as referenced in Mallgrave and Ikonomou
1994) initially discussed empathy and its role in aesthetic experience as the notion of
Einfühlung, which refers to the physical response when observing forms within art
works (primarily paintings). This notion has subsequently been defined as “aesthetic
empathy,” and naturally applies to the dance domain as well, in what has been termed
kinesthetic empathy (Hagendoorn 2004). In distinguishing from a passive mode of
“seeing” and an active mode of “looking at,” Vischer described how certain depicted
forms, i.e., bodily postures or the function of muscles and limbs, could arouse certain
feelings in response. Such nonpassive perceiving was proposed to involve a profound

response of empathy. As a result, the body of the observer would resonate with the
observed or depicted forms in the work. In turn, certain forms should evoke certain
emotions. Hildebrand (1893) elucidated upon this idea, suggesting that in order to
decipher a work, one has to implicitly grasp its creative process. Warburg proceeded
to acknowledge this in his notion of Pathosformel, or form-evoking Pathos, wherein
certain gestures, bodily forms, actions and expressions can be consistently detected
throughout art history (Warburg 1999).
The array of dimensions involved in an aesthetic response of the observer is
reflected by the body of research addressing the structure of this experience. One
seminal theory that attempts to draw together notions of mirroring between the artist’s
work and the observer, grounded within contemporary neuroscientific understanding
of the human mirror system, was proposed by Freedberg and Gallese (2007). In this
account, termed the embodied simulation theory of aesthetics, the authors suggest
that embodiment plays a significant role in appraising an artwork. Somewhat surpris-
ingly, in many formal art criticism circles, empathy towards or bodily resonance with
a work of art was previously dismissed as entirely intuitive subjective phenomena
(Collingwood 1938). Due to accumulating research on the function and scope of the
mirror system, however, this claim has been convincingly challenged. For example,
when an observer watches a particular body part of someone else being touched,
activation is seen within the brain of the observer as if the observer’s own body were
itself being touched (cf. Keysers et al. 2004). Freedberg and Gallese (2007) suggest
that empathetic simulation is evoked by the representation of figures in an image,
and a high degree of sensorimotor resonance with an agent depicted in an artwork is
what leads to an aesthetically pleasing experience for the observer. The authors go
on to propose that such resonance between an artwork and observer need not end at
figural representations. Instead, an observer might experience some kind of somatic
resonance with the artistic medium or technique used in the work. For example, an
observer might be drawn in on a visceral level by the artistic gestures used to create
a work, such as the brisk brushstrokes of Pollock, or the slashed cuts of a Lucio
Fontana piece. This is what allows an observer to, as described by Freedberg and
Gallese (2007), “[feel] the movement behind the mark” (p. 201).
Following on from the proposal that embodiment processes play a role in aesthetic
appraisal, several research groups have turned to dance paradigms to further explore
the role of kinesthetic empathy between the observer and artwork (in this case, the
work of art is the movements performed by a dancer). The first study to do this was
performed by Calvo-Merino et al. (2008) with a group of dance-naı̈ve participants
watching short-dance segments while undergoing fMRI. Several months after the
fMRI experiment, the participants returned to the laboratory and rated the individual
movements on a number of established aesthetic dimensions (after Berlyne 1974).
In contrast to the mOFC findings that have been reported by experiments looking at
aesthetic appraisal of static works of art and music (cf. Kawabata and Zeki 2004;
Ishizu and Zeki 2011), the brain regions found to track with increasing aesthetic
ratings of dance were located within bilateral occipital cortices and right premotor
cortex. Thus, it appears that visual and sensorimotor cortices are more engaged
when watching aesthetically pleasing dance, a finding that fits well with Freedberg
and Gallese’s (2007) embodied simulation account of aesthetics.
A subsequent study by Cross et al. (2011) sought to tie together the findings dis-
cussed in the previous section on embodiment and physical experience with what was
reported by Calvo-Merino et al. (2008) concerning the involvement of sensorimotor
brain regions in the aesthetic appraisal of dance. In this study, dance-naı̈ve partici-
pants watched a range of ballet and contemporary dance movements performed by
professional dancers while undergoing functional neuroimaging. Participants were
asked to rate each movement on how well they could physically reproduce it (same
question/scale used by Cross et al. 2006), and also how much they liked watch-
ing each movement. Behaviorally, the authors found a strong negative correlation
between perceived physical ability and liking, such that participants most enjoyed
watching those movements they rated as the most difficult to physically perform
(Cross et al. 2011). To analyze the brain-imaging data, the authors ran parametric
analyses to query brain regions that showed increasing responses based on increasing
(or decreasing) ratings of physical ability or aesthetic value. The most interesting
finding emerged from the interaction between parameters of perceived physical abil-
ity and aesthetic value. This analysis evaluated brain regions showing a stronger
response when participants rated a movement as difficult to perform, but highly
enjoyable to watch. The authors found that bilateral occipitotemporal cortices and
right inferior parietal lobule showed increasing responses the more a participant liked
watching a movement, and the less he or she could perform it.
When the findings of Calvo-Merino et al. (2008) and Cross et al. (2011) are con-
sidered together, they begin to build a compelling case in support of Freedberg and
Gallese’s embodied simulation account of aesthetics (2007). It appears that when
watching dance, even dance-naı̈ve observers engage sensorimotor brain regions as-
sociated with action perception and action performance to a larger degree when they
find the movement aesthetically pleasing. One might conclude from these findings
that embodied simulation of a dance piece is integral to aesthetic experience, al-
though this hypothesis has not yet been tested directly. Along these lines, however,
recent work from our laboratory is seeking to more fully characterize the role and
impact of physical experience on aesthetic experience in dance contexts (Kirsch et al.
2013; Kirsch et al., in preparation). One of the patterns of findings most relevant to
consider in light of Freedberg and Gallese’s embodied simulation account of aes-
thetics concerns the relationship between the perception of one’s ability to perform
a movement, or the inherent complexity of a movement, and how much an observer
enjoys watching the movement. Figure 10.2 illustrates findings from dance-naı̈ve
observers who watched a series of dance movements and were asked to rate their
physical ability to perform the observed movements (panel A; Cross et al. 2011) or
the perceived complexity of the movements (panel B; Kirsch et al. 2013), as well as
how much they enjoyed watching them.
The plot in panel A and the first plot in panel B illustrate that dance-naı̈ve observers
enjoy watching most of those movements that they perceived as difficult or complex
to perform. This pattern of findings might suggest that complex dance movements
156
Fig. 10.2 Comparison of findings from Cross et al. (2011, panel A) and Kirsch et al. (2013, panel B) regarding the relationship between perceived ability
to physically perform a movement or perceived complexity of a movement, and how enjoyable a movement is to watch. The relationship between perceived
complexity/difficulty and liking is manifest as a positive correlation among dance-naı̈ve observers, but once physical experience is gained, this relationship
attenuates. (Figure adapted from Cross et al. 2011 and Kirsch et al. 2013)
E. S. Cross and A. Elizarova
engage observers more, perhaps by challenging them to embody the movement in

a manner. However, the study illustrated in panel B (Kirsch et al. 2013) incorpo-
rated a training manipulation where participants learned to perform a subset of the
movements they observed and rated before having any kind of dance training or ex-
perience. What the right plot in panel B illustrates is that after learning to physically
embody a movement, the relationship between perceived complexity and enjoyment
is no longer present. Of course, this finding raises more questions than answers,
concerning how embodiment and aesthetic value are linked in the action perception
and production domain. Ongoing work in our laboratory aims to further delineate
the behavioral and neural consequences of the relationship between embodiment and
affective processing of others’ movement.
10.4 Conclusion
The aim of the present chapter was to illustrate the utility of dancers to scientists
seeking to explore how action and perception are linked in learning, expertise, and
aesthetics. We have attempted to place the action expertise and neuroaesthetics re-
search with dance in a broader context that clarifies how and why scientists have
worked with dance and dancers in their experiments. Neuroscientists initially turned
to the domain of dance to use it as a means of addressing fundamental questions of
action and perception, rather than to study its representation as an artistic medium
in the human mind or body per se (Cross and Ticini 2012). With this being said, it is
perhaps unsurprising that ever since the first neuroscience work with dancers started
to emerge nearly a decade ago, many individuals in the dance community have been
interested in what such research might reveal (or not) about the performance and
perception of their art form. A number of cross-disciplinary forums and research
projects are starting to emerge that attempt to find ways for artists and scientists
to work together on research that mutually informs and benefits both the scientific
and artistic domains (cf. Motion Bank (www.motionbank.org) and Watching Dance
(www.watchingdance.org)). From a dance perspective, these endeavors have the
possibility to illuminate factors that influence the performer–observer relationship,
and to map the boundaries of an aesthetic experience. From a scientific perspective,
such collaborations will undoubtedly reveal new ways of exploring fundamental as-
pects of human behavior and brain function with dance, both within and beyond the
domains of action expertise and neuroaesthetics.
Acknowledgments The authors would like to thank Dr. Alison Wiggett for critical feedback on
an earlier version of this chapter. E. S. Cross gratefully acknowledges research funding from the
Netherlands Organisation for Scientific Research (NWO Veni Award: 451–11-002), the Economic
and Social Research Council (ESRC Future Research Leader Award: ES/K001892/1), and a Marie
Curie Career Integration Award (FP7-PEOPLE-2012-CIG-322256).
References
Andersen RA, Meeker D, Pesaran B, Breznen B, Buneo C, Scherberger H. 2004. Sensorimotor trans-
formations in the posterior parietal cortex. In M. Gazzaniga (Ed.), The Cognitive Neurosciences
III. Cambridge, MA: The MIT Press.
Barsalou LW. 2008. Grounded cognition. Ann Rev Psych 59:617–645.
Berlyne DE. 1974. Studies in the new experimental aesthetics: Steps toward an objective psychology
of aesthetic appreciation. Washington: Hemisphere Co.
Bläsing B, Calvo-Merino B, Cross ES, Jola C, Honisch J, Stevens CJ. 2012. Neurocognitive control
in dance perception and performance. Acta Psych 139(2): 300–308.
Blood AJ, Zatorre RJ. 2001. Intensely pleasurable responses to music correlate with activity in brain
regions implicated in reward and emotion. Proc Natl Acad Sci 98:11818–11823.
Braun SM, Beurskens AJ, Borm PJ, Schack T, Wade DT. 2006. The effects of mental practice in
stroke rehabilitation: a systematic review. Arch Phys Med Rehabil 87(6):842–852.
Brown S, Dissanayake E. 2009. The arts are more than aesthetics: Neuroaesthetics as narrow
aesthetics. In: Skov M, Vartanian O, editors. Neuroaesthetics. Amityville: Baywood. pp. 43–57.
Calvo-Merino B, Glaser DE, Grezes J, Passingham RE, Haggard P. 2005. Action observation and
acquired motor skills: an fMRI study with expert dancers. Cereb Cortex 15(8):1243–1249.
Calvo-Merino B, Grezes J, Glaser DE, Passingham RE, Haggard P. 2006. Seeing or doing? Influence
of visual and motor familiarity in action observation. Curr Biol 16(19):1905–1910.
Calvo-Merino B, Jola C, Glaser DE, Haggard P. 2008. Towards a sensorimotor aesthetics of
performing art. Conscious Cogn 17(3):911–922.
Cinzia DD, Gallese V. 2009. Neuroaesthetics: a review. Curr Op Neurobiol 19(6):682–687.
Collingwood RG. 1938. The Principles of Art. London: Oxford University Press.
Cross ES, Ticini LF. 2012. Neuroaesthetics and beyond: New horizons in applying the science of
the brain to the art of dance. Phenomenology Cog Sci 11(1):5–16.
Cross ES, Hamilton AF, Grafton ST. 2006. Building a motor simulation de novo: Observation of
dance by dancers. Neuroimage 31(3):1257–1267.
Cross ES, Kirsch L, Ticini LF, Schutz-Bosbach S. 2011. The impact of aesthetic evaluation and phys-
ical ability on dance perception. Front Human Neurosci 5:102. doi: 10.3389/fnhum.2011.00102.
Decety J, Grezes J. 1999. Neural mechanisms subserving the perception of human actions. Trends
Cog Sci 3(5):172–178.
Decety J, Grezes J. 2006. The power of simulation: Imagining one’s own and other’s behaviour.
Cog Brain Res 1079:4–14.
di Pellegrino G, Fadiga L, Fogassi L, Gallese V, Rizzolatti G. 1992. Understanding motor events:
a neurophysiological study. Exp Brain Res 91:176–180.
Fadiga L, Fogassi L, Pavesi G, Rizzolatti G. 1995. Motor facilitation during action observation: a
magnetic stimulation study. J Neurophysiol 73(6):2608–2611.
Fadiga L, Buccino G, Craighero L, Fogassi L, Gallese V, Pavesi G. 1999. Corticospinal excitability
is specifically modulated by motor imagery: a magnetic stimulation study. Neuropsychologia
37(2):147–158.
Fink A, Graif B, Neubauer AC. 2009. Brain correlates underlying creative thinking: EEG alpha
activity in professional vs. novice dancers. Neuroimage 46(3):854–862.
Fogassi, L., Ferrari, P. F., Gesierich, B., Rozzi, S., Chersi, F., & Rizzolatti, G. (2005). Parietal lobe:
from action organization to intention understanding. Science 308(5722):662–667.
Freedberg D, Gallese V. 2007. Motion, emotion and empathy in esthetic experience. Trends Cog
Sci 11(5):197–203.
Fusani, L. 2008. Testosterone control of male courtship in birds. Hormones and Behavior 54(2):
227–233.
Gallese V, Sinigaglia C. 2011. What is so special about embodied simulation? Trends Cogn Sci
15(11):512–519.
Gallese, V., Fadiga, L., Fogassi, L., & Rizzolatti, G. (1996). Action recognition in the premotor
cortex. Brain 119(Pt 2), 593–609.
Gallese V, Gernsbacher MA, Heyes C, Hickock G, Iacoboni M. 2011. Mirror Neuron Forum.
Perspect Psych Sci 6:369–407.
Goldman A I. 2009. Mirroring, simulating and mindreading. Mind Lang 24(2):235–252.
Grafton ST, Arbib MA, Fadiga L, Rizzolatti G. 1996. Localization of grasp representations in
humans by positron emission tomography. 2. Observation compared with imagination. Exp
Brain Res 112(1):103–111.
Grosbras MH, Beaton S, Eickhoff SB. 2012. Brain regions involved in human movement perception:
a quantitative voxel-based meta-analysis. Hum Brain Map 33(2):431–454.
Guillot A, Collet C. 2010. The Neurophysiological Foundations of Mental and Motor Imagery.
Oxford: Oxford University Press.
Hagendoorn I. 2004. Some speculative hypotheses about the nature and perception of dance and
choreography. J Consciousness Studies 11:3–4.
Hildebrand, A. (1893). The problem of form in painting and sculpture (M. Meyer and R. M. Ogden,
Trans.). Strassburg: Heitz.
Hänggi J, Koeneke S, Bezzol L, Jancke L. 2009. Structural neuroplasticity in the sensorimotor
network of professional female ballet dancers. Hum Brain Map, 1196–1206.
Ishizu T, Zeki S. 2011. Toward a brain-based theory of beauty. PloS One 6(7):e21852.
Jacob P. 2008. What do mirror neurons contribute to human social cognition? Mind Lang 23:
190–223.
Jacob P, Jeannerod M. 2005. The motor theory of social cognition: a critique. Trends Cog Sci
9(1):21–25.
Jacobsen T. 2006. Bridging the arts and sciences: A framework for the psychology of aesthetics.
Leonardo 39(2):155–162.
Jacobsen T, Buchta K, Köhler M, Schröger E. 2004. The primary of beauty in judging the aesthetics
of objects. Psych Rep 94(3c):1253–1260.
Jacobsen T, Schubotz RI, Höfel L, Cramon DYV. 2006. Brain correlates of aesthetic judgment of
beauty. Neuroimage 29(1):276–285.
James W. 1890. Principles of Psychology, Vol. 1. New York: Holt.
Jeannerod M. 2001. Neural simulation of action: A unifying mechanism for motor cognition.
NeuroImage 14:S103–S109.
Johnson-Frey SH. 2004. Stimulation through simulation? Motor imagery and functional reorgani-
zation in hemiplegic stroke patients. Brain Cogn 55(2):328–331.
Kawabata H, Zeki S. 2004. Neural correlates of beauty. J Neurophysiol 91(4):1699–1705.
Keysers C, Wicker B, Gazzola V, Anton JL, Fogassi L, Gallese V. 2004. A touching sight: SII/PV
activation during the observation and experience of touch. Neuron 42(2):335–346.
Kirsch L, Drommelschmidt KA, Cross ES. 2013. The impact of sensorimotor experience on affective
evaluation of dance. Front Hum Neurosci 7. doi: 10.3389/fnhum.2013.00521
Kringelbach ML. 2005. The orbitofrontal cortex: linking reward to hedonic experience. Nat Rev
Neurosci 6:691–702.
Mallgrave HF, Ikonomou E. 1994. Empathy, Form, and Space: Problems in German Aesthetics,
1873–1893 (Santa Monica). p 1–85.
Massaro DW, Friedman D. 1990. Models of integration given multiple sources of information.
Psychol Rev 97(2):225–252.
Molenberghs P, Cunnington R, Mattingley JB. 2012. Brain regions with mirror properties: a meta-
analysis of 125 human fMRI studies. Neurosci Biobehav Rev 36(1):341–349.
Moran A, Campbell A, Holmes P, MacIntyre T. 2012. Observation, learning and skill acquisition
in sport. In: Hodges N, Williams MA, editors. Skill Acquisition in Sport (2nd ed.). London:
Routledge. pp. 94–111.
Mountcastle VB. 1975. The view from within: pathways to the study of perception. Johns Hopkins
Med J 136(3):109–131.
Mountcastle VB, Lynch JC, Georgopoulos A, Sakata H, Acuna C. 1975. Posterior parietal asso-
ciation cortex of the monkey: command functions for operations within extrapersonal space. J
Orgs G, Dombrowski JH, Heil M, Jansen-Osmann P. 2008. Expertise in dance modulates alpha/beta
event-related desynchronization during action observation. Eur J Neurosc 27(12):3380–3384.
Orians GH. 2001. An evolutionary perspective on aesthetics. Bull Psych Arts 2(1):25–29.
Palmer SE, Schloss KB, Sammartino J. 2013. Visual aesthetics and human preference. Ann Rev
Psych, 64:77–107.
Rizzolatti G, Sinigaglia C. 2010. The functional role of the parieto-frontal mirror circuit:
interpretations and misinterpretations. Nat Rev Neurosci 11(4):264–274.
Rizzolatti G, Fadiga L, Gallese V, Fogassi L. 1996. Premotor cortex and the recognition of motor
actions. Brain Res Cogn Brain Res 3(2):131–141.
Rizzolatti G, Fogassi L, Gallese V. 2001. OPINION. Neurophysiological mechanisms underlying
the understanding and imitation of action. Nat Rev Neurosci 2(9):661–670.
Sanders AF. 1967. Some aspects of reaction processes. Acta Psychol (Amst) 27:115–130.
Sanders AF. 1983. Towards a model of stress and human performance. Acta Psychol (Amst)
53(1):61–97.
Shuster SM. 2009. Sexual selection and mating systems. Proc NatAcad Sci U SA 106:10009–10016.
Vartanian O, Goel V. 2004. Neuroanatomical correlates of aesthetic preference for paintings.
Neurorep 15(5):893–897.
Voland E, Grammer K. 2003. Evolutionary Aesthetics. Berlin: Springer-Verlag.
Warburg A. 1999. The Renewal of Pagan Antiquity (translated by D. Britt). Los Angeles: The Getty
Research Institute for the History of Art and Humanities.
Welford AT. 1968. Fundamentals of Skill. London: Methuen.
Yarrow K, Brown P, Krakauer JW. 2009. Inside the brain of an elite athlete: the neural processes
that support high achievement in sports. Nat Rev Neurosci 10(8):585–596.
Yin TC, Mountcastle VB. 1977. Visual input to the visuomotor mechanisms of the monkey’s parietal
lobe. Science 197(4311):1381–1383.
Chapter 11
Apollo’s Curse: Causes and Cures of Motor
Failures in Musicians: A Proposal for a New
Classification
Eckart Altenmüller, Christos I. Ioannou, Markus Raab and Babett Lobinger
11.1 Becoming a Horowitz: Challenges in Acquiring Superior

Motor Skills in Musical Performance
Performing music at a professional level is one of the most complex human accom-
plishments. Playing an instrument requires integration of multimodal sensory and
motor information, generation of appropriate action plans, selection, and retrieval
of highly refined movement patterns from procedural motor memory, and initiation
of the movement. In most instances, these movements are highly overlearned, and
depend on feed-forward programming of the anticipated—mostly audible—results
and on real-time feedback.
Auditory and kinesthetic feedback is needed to improve and perfect performance.
Music making therefore relies primarily on a highly developed auditory–sensory–
motor integration capacity, which has been theoretically conceptualized in the
“common-coding” model by Prinz (1984). Simplifying this model, movements are
represented as sound patterns, and sound patterns as movements (see for a review
Zatorre et al. 2007). Subtle perturbations of this auditory feedback will have a major
impact on motor control. For example, playing on a keyboard with slightly delayed
sound production will compromise regularity of scale playing even in highly accom-
plished pianists (Cheng et al. 2013). Furthermore, the kinesthetic senses constitute
E. Altenmüller () · C. I. Ioannou

Institute of Music Physiology and Musicians’ Medicine, Hannover University of Music,
Drama and Media, Emmichplatz 1, Hannover 30175, Germany
Tel.: 0049 (0) 511 3100 552
M. Raab · B. Lobinger
Institute of Psychology, German Sport University Cologne, Am Sportpark Müngersdorf 6,
Köln 50933, Germany
Tel.: 0049 (0) 221 4982 5700
M. Raab
Department of Applied Sciences, London South Bank University,
Am Sportpark Müngersdorf 6, London 50933, UK

and Biology 826, DOI 10.1007/978-1-4939-1338-1_11
162 E. Altenmüller et al.
another basis of high-level performance. They allow for control and feedback of
muscle and tendon tension as well as joint positions enabling continuous monitoring
of finger, hand, or lip position in the frames of body and instrument coordinates
(e.g., the keyboard, the mouthpiece). Again, subtle changes in kinesthetic feedback
result in alteration of the motor program, and long-term loss will cause, in some
individuals, severe disturbances of the motor program and even focal dystonia (for
a theoretical account and review on this topic, see Konzcak and Abruzzese 2013).
In order to acquire these specialized sensory–motor skills, extensive training peri-
ods over many years, starting in early infancy and passing through stages of increasing
physical and strategic complexities, are necessary. This process of practicing involves
assembling, storing, and constantly improving sensorimotor programs through pro-
longed and repeated execution of motor patterns under the controlled monitoring of
auditory and kinesthetic senses. To attain a professional level, as a rule of thumb,
10,000 h in 10 years of deliberate practice are needed (Ericsson et al. 1993). Of
course, time invested into the acquisition of motor skills is only a necessary, how-
ever, not a sufficient precondition of becoming an outstanding artist. Quality of
practice as well as communication skills and expressive gesturing, rendering a per-
formance “speaking,” are equally important in the process of artistic perfection of
publicly acclaimed and valued interpretations (Hallam in press).
In music, learning through experience and training is accompanied by remarkable
plastic adaptations of the central nervous system (CNS), which are not only reflected
in modifications of the neuronal networks of the brain, as a result of a strengthen-
ing of neuronal connections, but also in its overall gross structure. It is known, for
example, that music practice enhances myelination, grey matter growth, and fiber
formation of brain structures involved in the specific musical task (for a review see
Münte et al. 2002; Wan and Schlaug 2010). For example, Gaser and Schlaug (2003)
could demonstrate enhancement of grey matter density in cortical sensorimotor re-
gions, auditory regions, the left dorsolateral prefrontal cortex, and in the cerebellum
in professional instrumentalists as compared to nonmusicians and amateurs. Interest-
ingly, these plastic adaptations depend on critical periods: Musicians, who start early,
before age 7, do not display these adaptations of the brain structure at least in the
sensorimotor cortices and the callosal fibers. However, they seem to have an “early
optimized network,” which allows superior performance of motor tasks without en-
larged anatomical structures (Steele et al. 2013; Vaquero et al. 2014). In contrast, late
starters, after age 7, do show the abovementioned structural adaptations accounting
for the effects observed in many morphological brain imaging studies (e.g., Bangert
and Schlaug 2006; Gärtner et al. 2013).
When discussing skilled motor behaviour in music performance and its deteriora-
tion or even loss, it should be emphasized what is unique to making music and what
renders it particularly challenging and fragile in terms of motor control:
1. Musical training usually starts very early, sometimes before age 6, when the
adaptability of the CNS is highest. This feature is not unique to music, since other
skilled activities, for example classical ballet dancing also requires an early start.
11 Apollo’s Curse: Causes and Cures of Motor Failures in Musicians 163
2. Making music is linked to sound production. As in speech, the auditory system

provides very precise feedback of the movement effects, with a temporal reso-
lution superior to kinesthetic and visual feedback. Furthermore, in the frame of
classical music, which is notated and available as sheet music, the target param-
eters, namely temporal accurateness (correct tempo, accuracy of rhythm, swing,
beat, pulse etc.), and spatial accurateness (correct pitch on key or fingerboard,
correct sound quality), are predefined. Therefore, a highly reliable reproduction
of movements meeting these targets is required. This feature is unique to music,
since for example in classical ballet, visual feedback is less critical in terms of
temporo-spatial precision.
3. Most musicians work at the upper limit of their sensorimotor capabilities and
strive to push their limits even further, in order to be faster, louder, and more
expressive. Given the complexity of music, the demands of composers, especially
in the past 100 years, and the role of outstanding peers as models, for example
the Chinese pianist Lang Lang, or the “record breaking” violinist David Garrett,
the theoretical limit of movement accuracy and speed is the temporal and spatial
resolution of the auditory system. As musicians say “there is always a colleague
who plays this piece faster, louder and more beautifully.” This is in part also true
for sports, but in music, fine motor skills predominate. Musicians therefore are
colloquially frequently denoted as “small muscle athletes.”
4. The societal pressure and expectancies concerning the quality of musical per-
formances has definitely grown over the past centuries, due to the ubiquitous
availability of music recorded in media such as YouTube and CDs, as well as due
to collective learning processes leading to higher standards of music appreciation.
This process augments anxiety, tension, and competition amongst musicians ren-
dering their life increasingly stressful. Frequently even outstanding soloists have
to cope with severe performance anxiety (Wilson 1997).
On the other hand, making music is frequently linked to highly positive emotions,
to feelings of joy, satisfaction, and even to strong emotional reactions, known as
“chill responses” (Altenmüller et al. 2013). These qualities are known to enhance
plastic adaptations of the brain and can even lead to a sort of addictive behavior,
causing younger musicians to over-practice and ignoring their bodily limits which
are indicated by fatigue and musculoskeletal pain.
In the following section, we will focus on the loss of motor control in highly
performing musicians and will argue that these problems need to be classified in a
more refined manner.
11.2 Deterioration of Motor Control in Musicians: Towards

a Refined Classification
In this section, we will not deal with problems of acquisition of new and refined motor
skills. Instead, we will focus on the deterioration of skilled movements, which had
been acquired previously through long-term practice and automation. The former is
familiar to all music teachers: some students are simply unable to attain a certain
tempo, to perform a certain articulatory technique (e.g., “double-tongue,” in flute
players), or to control the execution of phasic movements (e.g., the vibrato is too
fast, or too slow etc.). Its causes are not well understood and will require substantial
research in the future.
Furthermore, we will exclude disturbances of motor control due to tissue damage,
be it peripheral, such as cervical root injury, trigger finger etc., or be it of CNS origin,
such as stroke, brain hemorrhage, or tumors. We will not cover these topics here, but
we will concentrate on proposing six different categories of deterioration or loss of
motor control in musicians with intact neural substrates.
For classifying deterioration of motor control in musicians, based on the current
state-of-the-art, we have applied the following classification criteria: (1) duration and
development of the problem, (2) triggering mechanisms, (3) psychological profiles,
(4) response to treatment, (5) accompanying symptoms, and (6) genetics.
11.2.1 Motor Fatigue
Most musicians know instances of loss of motor control when playing their instru-
ment. Frequently this phenomenon is short-lasting and due to either lack of practice
or overuse. It is accompanied by mental or bodily fatigue and obviously linked to re-
duced attention and insufficient movement monitoring. Usually these deteriorations
of movement coordination are short-term, disappear overnight, and do not compro-
mise overall performance. Although these incidents seem to be frequent, there is
almost no scientific data available. A recent questionnaire study in German orches-
tral brass instrumentalists identified temporary crises of embouchure coordination
in a percentage as high as 30 % of musicians (Steinmetz et al. 2013). In clinical
practice, symptoms are lack of regularity in scales, trills and other fast repetitive
movements, wind players complain of loss of sound quality and occasionally early
fatigue. Usually, pain is not reported. There is no information available whether any
medical intervention other than rest is useful to treat this condition.
11.2.2 Overuse Injury
In contrast to the former conditions, overuse in the narrow sense of the word should
only be diagnosed when pain is a dominating symptom and a history of either pro-
longed or unaccustomed practice exists. Here, local inflammation of overstrained
tissues, release of pain mediators, and spinal reflexes with increased or occasionally
lowered muscle tone may lead to a deterioration of motor control, mostly accompa-
nied by relieving movements (Szeto and Lin 2011). This condition may last a couple
of days, and under rest subsides after a few weeks.
11.2.3 Choking Under Pressure
Another condition leading to deterioration of motor control is “choking under

pressure.” Choking under pressure (CuP) figures widely in the sports psychology lit-
erature but has not yet been adopted by experts in musician’s medicine and performing
arts psychologists. We therefore devote a longer section to this condition.
In sports, the current definition characterizes CuP as an acute performance failure
due to a perceived mismatch between the individual resources of an athlete and the
demands of the situation (Hill et al. 2009). CuP describes the situation in which the
individual perceives a subjectively unmanageable situation that is accompanied by
fear of failure, anxiety, and increased arousal, leading to reduced motor control and
worse performance outcome. In musicians, CuP is a relatively frequent experience,
especially in performing novices, and is usually subsumed under the term “perfor-
mance anxiety.” According to questionnaires, between 15 and 60 % of performing
musicians occasionally suffer from this condition (Brugués et al. 2011a; Steptoe and
Fidler 1987). It may lead to loss of agility, heightened muscular stiffness accom-
panied by increased co-contraction of antagonist muscles (Yoshie et al. 2009), and,
as a consequence, reduction of temporo-spatial precision of movements and sound
quality. In brass players, the so-called tongue stopper prior to an important cue is a
typical manifestation.
In sports, different theories have been proposed to explain the appearance of a
choke. They can be divided into drive theories and attentional theories (for a review
see Hill et al. 2010). Drive theories are based on the assumption that increased arousal,
resulting from striving to perform well under pressure, will have a detrimental effect
on performance (Spence and Spence 1966). Drive theories account for the classical
inverted-U model of performance anxiety. This model predicts that increased arousal
will first lead to an improved performance; however, further increase will reduce
performance quality (Yerkes and Dodson 1908). However, drive theories cannot
explain why some athletes or musicians thrive but others fail under pressure.
Attentional theories try to account for individual differences. Attentional the-
ories can be separated into distraction (Carver and Scheier 1981) and self-focus
(Baumeister 1984) theories. Distraction theories assume that pressure-induced anx-
iety will occupy the performer’s working memory, restricting the processing of
task-relevant information (Eysenck and Calvo 1992). Self-focus theories propose
that performance anxiety causes the performer to shift the focus of attention inward
or to consciously monitor the skill, which detrimentally affects the well-learned auto-
mated skill (Baumeister 1984). Crucial for the occurrence of choking is the perceived
control of a situation, which is influenced by both situational and personality factors
(Otten 2009). Personality factors that increase the likelihood of choking include low
self-confidence (Baumeister and Showers 1986) and high trait anxiety (Wilson 2008).
11.2.4 Dynamic Stereotype in Musicians
When motor incoordination and lack of motor control persist for more than 4 weeks,
even though rest has been observed and careful rehabilitation under the guidance
of a therapist or teacher has been attempted, one can assume a more grave alter-
ation of sensorimotor networks leading to a deterioration of motor programs in the
CNS. We have called this condition “dynamic stereotype” (DS), a term borrowed
from the eminent Russian physiologist Ivan Pavlov (1951; for a critical review, see
Windholz 1996). Originally, this condition can be understood as a reflection of fatal
compensation strategies, which became automated. In the words of the exponents
of Russian behaviorism, “dynamic stereotypes” are defined as a type of integral
activity performed by the cerebrum of higher animals and man and manifested by
a fixed, or stereotyped, succession of conditioned reflexes. The DS is influenced
by external factors that are repeated in a certain order. Accordingly, the DS is the
most vivid manifestation of the extremely subtle analyzing and synthesizing activity
of the cerebral cortex and is the product of very complex interactions between the
areas of the cortex. It can at least partly be conceptualized as a consequence of long-
term CuP when these dysfunctional movements are stored in procedural memory
traces, maybe as a consequence of conditioned reactions to previous choking experi-
ences and procedural memory formation under stress (Klämpfl et al. 2013; Lobinger
et al. 2014).
The phenomenology of DS resembles in many aspects focal, task-specific dysto-
nia (see below). However, in contrast to the latter, it seems to be more modifiable and
more fluctuating, especially during stressful performances. Sometimes, “islands” of
well-being and complete motor control occur, although only for hours or a few days.
It responds occasionally to trick sensory maneuvers, such as alterations of either
tactile input from the body parts affected by dystonia or alterations of auditory input
for example by delay of the produced sound. We could demonstrate that improve-
ment of motor control, when playing with a latex glove is related to better outcome
of retraining and behavioral therapies (Paulig et al. 2014). It should be mentioned,
however, that responses to sensory tricks and objective improvement are rare and
highly variable (Cheng et al. 2013). We therefore prefer to consider this phenomenon
a “soft-sign” with respect to classification of problems of motor control.
11.2.5 Focal Dystonia in Musicians
The most severe movement disorder among instrumental performers is task-specific

musician’s dystonia (MD;Altenmüller 2003), also known as musician’s cramp. Com-
monly, two major forms, focal hand dystonia (FHD) and embouchure dystonia (ED)
are distinguished. It is a movement disorder characterized by persistent muscular
incoordination or loss of voluntary motor control during task-specific highly trained
movements, such as playing a musical instrument (Altenmüller 2003; Jankovic and
Ashoori 2008). In most cases, pain does not accompany the disorder. Occasionally
Fig. 11.1 Typical patterns of

dystonic postures in a pianist,
a violinist, a flutist, and a
trombone player suffering
from musician’s dystonia
some muscular strain can occur due to attempts of compensating for the dystonic
movement by over-activation of the antagonist muscles. However, the lack of pain
distinguishes it from the abovementioned overuse injury. It is important to make this
distinction bearing in mind that prolonged pain syndromes may lead to symptomatic
dystonia. MD frequently terminates professional careers and is among musicians’
most highly disabling condition (Altenmüller 2003; Brandfonbrener and Robson
2004; Lederman 1991; Altenmüller and Jabusch 2010).
Various symptoms can mark the beginning of the disorder: subtle loss of control
in fast passages, finger curling (cf. Fig. 11.1), lack of precision in forked fingerings
in woodwind players, irregularity of trills, fingers sticking on the keys, involuntary
flexion of the bowing thumb in strings, impairment of control of the embouchure
in woodwind and brass players in certain registers, are among the most frequent
descriptions given by patients. At this stage, most musicians believe that the reduced
precision of their movements is due to a technical problem or lack of practice. As
a consequence, they intensify their efforts, but this reaction often exacerbates the
problem. The loss of muscular coordination is frequently accompanied by a co-
contraction of antagonist muscle groups. For example, in pianist’s cramp, the co-
activation of wrist flexor and wrist extensor muscles is frequently observed, and we
have documented a case of a task-specific dystonia in the leg of a bass drummer,
leading to pronounced co-activation of ankle flexor and extensor muscles (Lee and
Altenmüller 2014).
There are special cases of task-specific loss of motor control in musicians, which
should be mentioned here since they are strongly related to MD or even constitute
subgroups: Dystonic tremor is characterized by task-specific tremulous movements
of the supporting arm in woodwinds or in the bowing arm of string players (Lee et al.
2013). A very rare condition is a task-specific inability to recruit motor programs
required for a specific overlearned movement, termed “negative dystonia” (Mezaki
2007).
According to recent estimates, 1 % of all professional musicians are affected
(Altenmüller and Jabusch 2010). In contrast, in the general population, prevalence
of focal dystonias, including writer’s cramp, blepharospasm, and cervical dystonia,

is estimated as 29.5 per 100,000 in the USA and 6.1 per 100,000 in Japan (Nutt et al.
1988; Nakashima et al. 1995). In comparison to other activities producing dystonic
movements, such as writing, playing golf (the “yips”), or dart (“dartism”), classical
musicians are at highest risk to develop focal dystonia (Frucht 2009).
Demographic data demonstrate a preponderance of male musicians with a male–
female ratio of about 4:1 (Lim and Altenmüller 2003). Hereditary factors play a role
in the etiology of MD, since a positive family history of dystonia exists in up to 36 %
of affected musicians (Schmidt et al. 2009). According to epidemiological data, the
probability of developing MD depends on the instrument played, with guitar players,
pianists, and brass instrument players having the highest risk of developing dystonia
(Altenmüller and Jabusch 2010). Repetitive use, controllability of motor actions,
temporo-spatial demands, and extra-instrumental fine motor burdens, such as writing,
are triggering factors (Altenmüller et al. 2012; Baur et al. 2011). Furthermore, those
musicians who start later than age 10 with their instrumental practice are at a much
higher risk to develop MD (Schmidt et al. 2013).
With respect to a differentiation of MD from the above-described DS, there is
probably a certain overlap. Generally, focal dystonia is more severe and the dysfunc-
tional movements are more obvious, and more resistant to any attempt to correct
them voluntarily. Furthermore, movements respond to a lesser degree to the above-
mentioned sensory tricks. Sudden spontaneous improvements are rare exceptions
and psychological stressors do not influence the loss of motor control to a major
degree. Finally, in contrast to DS, in MD there is a tendency to generalize from
the specific movement when playing the instrument to other daily-life movements.
For example, the problem may appear first only in the ring and little finger of the
pianist when playing scales, as depicted in Fig. 11.1, but later extends to typing on a
computer keyboard and to buttoning up a shirt, and in the end may lead to permanent
cramping of the hand. In neurology, this condition is termed “dystonic cramp” and it
seems to affect about 35 % of musicians suffering from primarily task-specific dysto-
nia (Rosset-Llobet et al. 2007). In the same line, loss of motor control may progress
from the forearm to the upper arm, leading in rare cases to segmental dystonia. When
dystonia propagates from one hand to the other, which is the case in about 3 % of the
musicians suffering from FHD, a bifocal dystonia has to be assumed (Rosset-Llobet
et al. 2007).
11.2.6 Symptomatic Task-specific Dystonias in Musicians
Symptomatic task-specific dystonias in musicians are a rare heterogeneous group

of dystonias. By definition, the leading symptom is isolated deterioration and loss
of motor control when playing a musical instrument. In contrast to the abovemen-
tioned “idiopathic” focal dystonias, this disorder is caused by either an underlying
neurological or psychological pathology. For example, in rare cases it may mark
the beginning of Parkinson’s disease, albeit we have to acknowledge that we have
only seen two MD patients who developed Parkinson’s disease several years later.
When taking the numbers of our whole group of presently 850 MD patients into
account, this is in the range of the incidence of Parkinson in the general population
in Germany. In neurological textbooks, furthermore morbus Wilson, Huntington’s
chorea, and other neurodegenerative diseases have been mentioned as a cause of
symptomatic dystonia; however, an isolated task-specific loss of motor control in
these conditions, specifically when playing the instrument, has to our knowledge not
yet been described.
MD is sometimes triggered by bodily trauma. Here, it is unclear whether the
lesioning of peripheral nerves and the concomitant sensory degradation or the
trauma-induced rest or change in practice schedules, playing postures, etc. con-
stitute triggers. We have seen two flutist patients suffering from inferior alveolar
nerve lesion and numbness of the lower lip who developed embouchure dystonia
(unpublished clinical data).
Psychogenic dystonias in musicians are extremely rare conditions. We have never
seen any musician malingering, probably due to the rewarding nature of music and the
lack of secondary benefits when losing the ability to perform. However, fluctuating or
longer-lasting loss of motor control when playing an instrument may occur in some
psychiatric diseases, such as schizophrenia, obsessive–compulsive disorders, and
constraint and anxiety disorders. In schizophrenia, long-term consequences of an-
tipsychotic drugs have to be considered as a cause. In obsessive–compulsive disorder,
dysfunctional working behavior and, as a consequence, motor fatigue and overuse
may be the underlying pathogenic mechanisms. Anxiety has been frequently found
as a risk factor in MD (see below). This might hypothetically be linked to chronic
choking under pressure, as discussed above in the “DS” section.
Finally, psychogenic MD in the narrow sense of the word is, as far as our expe-
rience goes, extremely rare. The underlying theory assumes a conversion disorder,
which in turn is caused by fundamental unresolved psychological conflicts, leading
to subconscious alterations of motor behavior which may express the nature of the
underlying conflict in a “converse” manner. It is very difficult to diagnose and there-
fore frequently subsumed under the term “unexplained medical conditions” (Lang
and Voon 2011). As a rule of thumb, musicians suffering from psychogenic dystonia
usually have a history of previous physical or psychological trauma or disorder, show
the symptoms in an exaggerated, “speaking” manner, seem to be more emotionally
distant to the motor disorder, and display less despair. Sometimes, a triggering event
can be identified, and in the follow-up, very fluctuating symptoms with prolonged
periods of complete remission or miracle healing are suspicious of a psychogenic
cause (Czarnecki and Hallett 2012).
Summarizing this section, an outline of a provisional “classification tree” is
sketched in Fig. 11.2. The idea is that motor disturbances in musicians are gradually
developing under given circumstances, such as high workload, underlying anxiety,
and stress.
Still, many details of this classification are hypothetical. Our thoughts were guided
by clinical experience and our goal is to find a better diagnostic algorithm in order
to improve client-tailored treatment.
Fig. 11.2 “Flow chart,” displaying the different types of motor disturbances sorted according to
duration and severity of the motor deficit. Factors promoting the transition from one class to the
next one are shown in italics, therapeutic strategies in bold italics at the right part of the panel. CBT
stands for “cognitive-behavioral therapy”
11.3 Can Psychological Profiles and Neurophysiological

Mechanisms Contribute to a Refined Classification
of Motor Degradation in Musicians?
In the following section, we will examine how a classification based on phenomenol-

ogy and underlying neurophysiological mechanisms may lead to client- tailored
treatment. As outlined above, the six types of motor disorders can be classified
according to the duration of motor degradation, the accompanying symptoms, the un-
derlying neurophysiological mechanisms, and the response to treatment. In Fig. 11.3,
we have depicted a heuristic model, assuming a continuous worsening of motor
control from temporary subtle awkwardness to increasingly unstable motor control
and finally, fully developed focal dystonia. Furthermore, we have added trigger-
ing factors, identified in our previous epidemiological studies (Altenmüller 2003;
Altenmüller and Jabusch 2010).
In the beginning, motor fatigue may cause a temporary degradation of motor
skills, or, in highly skilled experts, additional/alternative recruitment of muscles
contributing to dysfunctional movements. This mechanism has been convincingly
demonstrated in skilled table tennis players (Aune et al. 2008), and most probably
also applies for instrumental musicians. For example, in skilled piano players, fatigue
of the long flexor muscles in the forearm may be compensated by activation of the
intrinsic muscles in the hand, which in turn results in dysfunctional movements in the
Fig. 11.3 Heuristic model demonstrating the proposed interplay of sensory–motor and psychologi-
cal triggering factors and their contribution to the different motor disturbances. The assumed degree
of “psychological” triggering factors is displayed as the dotted space; the increasing grey shade
symbolizes the increasing degree of loss of motor control
metacarpophalangeal (MCP) joints with lack of fine control of touch and degraded
sound quality. These changes are accompanied by central nervous adaptations, due
to short-term plasticity, resulting in reduced amplitude of movement-related poten-
tials and in an alteration of the topography of motor and premotor cortex activations
(Dirnberger et al. 2004). Interestingly, in a study with musicians suffering from dys-
tonia, a fatiguing muscular contraction significantly improved motor performance. In
contrast, in healthy musicians, performance consistently worsened following fatigue
(Pesenti et al. 2004).
In the overuse condition, dysfunctional CNS plasticity probably plays an impor-
tant role in motor degradation (Byl et al. 1996; see also Flor 2012 for a review).
Although in most instances prognosis is good and quick recovery is common, under
conditions with heightened anxiety and other stressors, such as high professional
workload, these dysfunctional motor patterns may stabilize in procedural memory.
Here, psychological stress might induce the cascade of emotionally induced memory
consolidation, which has been described previously for different forms of memory:
It mainly relies on noradrenergic activation of the basolateral amygdala (BLA) (Mc-
Gaugh 2000; Packard et al. 1994). The primary motor cortex, which is essentially
involved in the storage of motor memories (Karni et al. 1998), receives a BLA pro-
jection (Sripanidkulchai et al. 1984). Thus, it may be assumed that consolidation
of these dyscoordinated movements as dysfunctional motor programs is a BLA-
mediated process in the primary motor cortex (Jabusch and Altenmüller 2004). This
may also be the link to conditions discussed below, namely CuP and DS.
CuP is not uncommon in sports and has been investigated in golfers, suffering from
the yips. This condition is defined as involuntary movements during the execution
of putting strokes, resulting in a serious decrease in the success rate in putting. It has
been classified by some authors as a task-specific focal dystonia (Adler et al. 2005);
however, many authors believe that the yips is more related to CuP (Lobinger et al.
2014). Arguments against the classification as a focal dystonia are:
1. The yips occur not only in professionally trained golfers but also in golfing be-
ginners. This renders a pathophysiological mechanism similar to that of MD very
unlikely. In musicians, we could demonstrate that dysfunctional brain plasticity
induced by prolonged practice leads to distorted topographies of receptive fields
in the somatosensory cortex, which in turn leads the observed degradation in
motor control, due to abnormal somatosensory feedback (Elbert et al. 1998).
2. The yips look more like a sudden jerk, or an anticipatory tremor prior to hitting
the ball. In MD, reduced inhibition in the motor output is another proven cause of
dysfunctional movements, leading to prolonged co-activation and typical dystonic
postures, for example, the curling of fingers in Fig. 11.1.
3. The yips are relatively frequent in the golfing community with prevalence between
16 and 24 % (Klämpfl et al. 2013). This is unusual for any kind of dystonia. The
most common task- specific dystonia up to now is MD, which affects 1–2 % of
professional musicians.
In music, the term CuP has not been applied, but the phenomenon clearly exists. The
abovementioned tongue stopper in brass players and the short-action bowing tremor
in violin players playing soft notes with the tip of the bows are good examples. Here,
most probably anxiety induced “reinvestment” leads to cognitive interference, result-
ing in dysfunctional movements due to the attempt to prevent or even correct feared
errors. In an electroencephalography (EEG) study on highly trained professional pi-
anists, we could demonstrate that their brain anticipates errors of motor execution,
and that these “pre-error”-related brain waves arise about 50 ms prior to the wrong
keystroke (Herrojo-Ruiz et al. 2009). Such a rapid error anticipating mechanism
cannot be cognitively controlled. Rather, is highly susceptible to disturbances via
cognitive control. Therefore, it is plausible that reinvestment and CuP may lead to a
deterioration of motor control.
DS is characterized by a more permanent reduction of motor control. As outlined
above, it differs from focal dystonia only gradually, but probably is more linked to
psychological triggering factors as to underlying genetic factors (Ioannou and Alten-
müller 2014). In terms of underlying neurophysiological mechanisms, we speculate
that these musicians have a deficit in the so-called limbic loops of the basal ganglia,
linking movements and motor control to emotions. However, this question remains
to be addressed in future investigations in a patient population, which a priori is
correctly classified (Ioannou and Altenmüller 2014).
MD can be distinguished from the former by the more pronounced worsening of
motor control, the lack of “islands” of well-being, and the tendency to progress to
dystonic cramps. The pathophysiological basis of MD can only briefly be summa-
rized. For more details, we refer to a review by Altenmüller and Jabusch (2010).
Numerous studies are revealing abnormalities in three main areas: (a) reduced inhi-
bition in the sensorimotor system, (b) altered sensory perception, and (c) impaired
sensorimotor integration. In recent years, an increasing number of brain-imaging
studies in musicians—and other focal dystonias—have demonstrated that these al-
terations are probably not task specific. Functional connectivity (Moore et al. 2012),
cortical activation patterns (Haslinger et al. 2010), and basal ganglia anatomy (Wal-
ter et al. 2012) have proven to be abnormal, although behavioral tests are more in
favor of task specificity when testing musicians in other skilled motor tasks (Rosset-
Llobet 2007). We argue that musician’s focal dystonia is the product of a hereditary
susceptibility, probably related to a general lack of central nervous inhibition and
the abovementioned triggering factors. This leads not only to task specific functional
alterations of CNS networks—as we predict in patients suffering from a dynamic
stereotype—but also to structural alterations found in the abovementioned studies.
In previous publications, we have emphasized psychological conditions as an un-
derlying triggering factor. In several questionnaire studies, we found elevated anxiety
and extreme perfectionism in MD patients (Altenmüller and Jabusch 2009; Enders
et al. 2012). According to Jabusch et al. (2004), these psychological characteristics
had already been present before onset of dystonia according to personal recall. We
now argue that musicians with a major psychological burden most probably can be
subsumed under the classification of a DS. We are presently conducting a study ad-
dressing the impact of stress and anxiety on motor performance in musicians suffering
from loss of motor control who are demonstrating either high or low psychological
burdens. Subsequently, we will apply specific interventions aiming at improving psy-
chological conditions and preventing dysfunctional reinvestment in order to improve
motor control in those musicians who suffer from dynamic stereotypes.
11.4 Consequences for Treatment of Motor Disturbances

in Musicians
What do psychological and neurophysiological findings imply for the differential

treatment of motor disturbances in musicians? We have already mentioned that mo-
tor fatigue should probably be best treated with rest, although no scientific data are
available. For treatment of overuse injuries, besides rest, pain medication, muscle
relaxants, and physiotherapy with gentle stretching exercises have proven to be use-
ful (for a review see Helliwell and Trainor 2004). In these pain syndromes, a prompt
intervention is required to avoid dysfunctional plastic adaptations of the CNS seen
in chronic pain conditions (Jensen et al. 2013). CuP in musicians is probably best
prevented by mental training and specific cognitive strategies, already applied in
sports psychology (e.g., “hemispheric priming,” Beckmann et al. 2013). Efficient
treatment probably includes cognitive-behavioral therapy, mindfulness training, and
various breathing and relaxation techniques (for a review, see Brugués 2011b). As
medications, beta-adrenergic receptor blockers (such as propanolol) and benzodi-
azepines are potentially helpful. Both drugs act on preventing dysfunctional motor
memory formation (Soeter and Kindt 2013).
As far as treatment of DS is concerned, we predict that psychological tech-

niques such as prevention of dysfunctional reinvestment and cognitive interference
will be helpful. The latter, for example, could be influenced by guiding attention
from an internal (body-related) to an external (sound-related) focus as have been
shown to be efficient in complex motor studies of normal and patient groups (Wulf
2007, for an overview). Furthermore, psychotherapeutic techniques reducing anx-
iety and perfectionism will probably improve the condition. Specific interventions
are learning-based sensorimotor training (LBST) based on redefining spatial and
temporal processing capacities in the sensory and motor cortices in order to restore
task-specific skills (Byl and McKenzie 2002). Prolonged pedagogical retraining has
been successfully applied in pianists suffering from loss of motor control (van Vugt
et al. 2014). Useful medications include selective serotonin reuptake inhibitors (SS-
RIs; e.g., escitalopram) to overcome reinvestment, overfocussing, and depression,
anticholinergic drugs to reduce dysfunctional motor memories, and finally as a symp-
tomatic treatment, local injections of botulinum toxin into the cramping muscles
(Schuele et al. 2005). These medications are also applied in MD.
Finally, treatment for MD is mostly symptomatic and depends on the type of
dystonia. Psychological interventions seem to be less effective as compared to DS.
Various oral medications have been used and the anticholinergic drug trihexyphenidyl
has proven to be the most effective agent (Jabusch et al. 2005). Occasionally, pa-
tients may benefit from baclofen, or antiepileptic drugs such as phenytoin, primidon,
or rivotril. Chemical denervation using botulinum toxin has been used for many
forms of MD with considerable success. Botulinum toxin blocks the transmission
of nerve impulses to the muscle and weakens the overactive muscles involved. Re-
sults in musician’s cramp depend on the dystonic pattern, on the injection technique,
and on the precise localization of the dystonic muscles. In our series, injections of
botulinum toxin were applied in 71 musicians suffering from hand dystonia. Fifty-
seven percent of patients reported long-term improvement (Schuele et al. 2005).
A new promising therapy is bihemispheric transcranial direct current stimulation
during execution of functional movements at the instrument. In a double-blind ran-
domized prospective trial, we could demonstrate that stimulation with inhibition
of the dysfunctional network over the “dystonic” motor cortex, and activation of
the “healthy” network over the contralateral motor cortex improved performance
significantly (Furuya et al. 2014).
Preventing MD is important, since successful treatment is still a challenge. With
the refined heuristic model outlined in this article, assuming a progression of increas-
ing motor disturbances from fatigue, pain, anxiety to dysfunctional motor memories,
and finally to MD with structural memory consolidation and brain adaptations, we
now have the theoretical means at hand to intervene at an early stage. From the first
lesson on, music educators should strive to create a friendly, supportive atmosphere,
introduce reasonable practice schedules, teach economic techniques, prevent overuse
and pain by including mental practice, and vary movement patterns. Students should
be encouraged to maintain motivation by avoidance of mechanical repetitions and
frustration. They should be taught to adopt healthy living habits, including warm-ups
and cool-down exercises, regular physical exercise, sufficient breaks, and sleep as
the cornerstones of healthy musical practice.
Acknowledgments The authors are grateful to the IMMM members for their beneficial feedback
and insights, as well as to Martin K. Klämpfl from the German Sport University Cologne for his
stimulating work on yips and the many discussions on the topic of classification.
This work has been funded by the German Research Foundation (DFG, AL269/8-1).
References
Adler CH, Crews D, Hentz JG, Smith AM. Caviness JN (Abnormal co-contraction in yips-affected
but not unaffected golfers: Evidence for focal dystonia. Neurology 2005; 64: 1813–1814
Altenmüller E. Focal dystonia: Advances in brain imaging and understanding of fine motor control
in musicians. Hand Clinics 2003; 19: 523–538.
Altenmüller E, Jabusch HC. Focal hand dystonia in musicians: Phenomenology, etiology, and
psychological trigger factors. J Hand Ther 2009; 22: 144–155.
Altenmüller E, Jabusch, HC. Focal Dystonia in Musicians. Eur J Neurol 2010;17 Suppl 1: 31–36.
Altenmüller E, Baur V, Hofmann A, Lim VK, Jabusch HC. Musician’s cramp as manifestation of
maladaptive brain plasticity: arguments from instrumental differences. Ann N Y Acad Sci 2012;
1252: 259–265.
Altenmüller E, Kopiez R, Grewe O. A contribution to the evolutionary basis of music: Lessons
from the chill response. In: E. Altenmüller E, S. Schmidt, E. Zimmermann (eds.) Evolution of
Emotional Communication From Sounds in Nonhuman Mammals to Speech and Music in Man.
Oxford University Press, Series in Affective Sciences, Oxford: 13–335 (2013)
Aune TK, Ingvaldsen RP, Ettema GJ. Effect of physical fatigue on motor control at different skill
levels. Percept Mot Skills 2008 Apr;106(2): 371–86.
Bangert M, Schlaug G. Specialization of the spezialized in features of external brain morphology.
Eur J Neurosci 2006; 24: 1832–1834.
Baumeister RF. Choking under pressure: Self-consciousness and paradoxical effects of incentives
on skillful performance. J Personality Soc Psychol 1984; 46: 610–620.
Baumeister RF, Showers C J. A review of paradoxical performance effects: Choking under pressure
in sports and mental tests. European Journal of Social Psychology, 1986; 16: 361–383.
Baur V, Jabusch HC, Altenmüller E. Behavioral factors influence the phenotype of musician’s
dystonia. Mov Disord 2011; 26: 1780–1781.
Beckmann J, Gröpel P, Ehrlenspiel F. Preventing motor skill failure through hemisphere-specific
priming: cases from choking under pressure. J Exp Psychol Gen. 2013 Aug;142(3): 679–91.
Brandfonbrener AG, Robson C. A review of 111 musicians with focal dystonia seen at a performing
artist’s clinic 1985–2002. Adv Neurol 2004; 94: 255–256.
Brugués AO. Music performance anxiety-part 1. A review of its epidemiology. Med Probl Perform
Art. 2011a Jun; 26(2): 102–5.
Brugués AO. Music performance anxiety-part 2. A review of treatment options. Med Probl Perform
Art. 2011b Sep;26(3): 164–71.
Byl NN, Merzenich MM, Jenkins WM. A primate genesis model of focal dystonia and repetitive
strain injury: I. Learning-induced dedifferentiation of the representation of the hand in the
primary somatosensory cortex in adult monkeys. Neurology 1996; 47: 508–520.
Byl, N.N., McKenzie, A. The effect of sensory discrimination training on structure and function in
patients with focal hand dystonia: 3 case series. Arch Phys Med Rehab 2002; 84: 1505–1514.
Carver C S. Scheier MF. Attention and self-regulation: A control theory approach to human behavior.
New York, NY: Springer; 1981
Cheng FP, Großbach M, Altenmüller E. Altered sensory feedbacks in pianist’s dystonia: the altered
auditory feedback paradigm and the glove effect. Front Hum Neurosci 2013 Dec 17;7:868. doi:
10.3389/fnhum.2013.00868.
Czarnecki K, Hallett M. Functional (psychogenic) movement disorders. Curr Opin Neurol. 2012
Aug;25(4): 507–12
Dirnberger G, Duregger C, Trettler E, Lindinger G, Lang W. Fatigue in a simple repetitive motor task:
a combined electrophysiological and neuropsychological study. Brain Res 2004 Nov 26;1028(1):
26–30.
Elbert T, Candia V, Altenmüller E, Rau H, Rockstroh B, Pantev C, Taub E. Alteration of dig-
ital representations in somatosensory cortex in focal hand dystonia. NeuroReport 1998; 16:
3571–3575.
Enders L, Spector JT, Altenmüller E, Schmidt A, Klein C, Jabusch, HC. Musician’s dystonia and
comorbid anxiety: Two sides of one coin? Mov Dis 2012;26(3): 539–542.
Ericsson KA, Krampe RT, Tesch-Römer C. The role of deliberate practice in the acquisition of
expert performance. Psychol Rev 1993; 100: 363–406.
Eysenck M, Calvo M. Anxiety and performance: The processing efficiency theory. Cognition &
Emotion 1992; 6: 409–434.
Flor H. New developments in the understanding and management of persistent pain. Curr Opin
Psychiatry 2012 Mar;25(2): 109–13.
Frucht SJ. Focal task-specific dystonia of the musicians’ hand-a practical approach for the clinician.
J Hand Ther 2009;22(2): 136–143.
Furuya S, Nitsche MA, Paulus W. Altenmüller E Surmounting retraining limits in musicians’ dysto-
nia by transcranial stimulation. Ann Neurol 2014 Apr 7: doi: 10.1002/ana.24151. [Epub ahead
of print]
Gärtner H, Minnerop M, Pieperhoff P, Schleicher A, Zilles K, Altenmüller E, Amunts K. Brain
morphometry shows effects of long-term musical practice in middle-aged keyboard players.
Front Psychol 2013 Sep 23; 4: 636. doi: 10.3389/fpsyg.2013.00636.
Gaser C, Schlaug G. Brain structures differ between musicians and non-musicians. Journal of
Neuroscience, 2003; 23: 9240–9245.
Hallam S. Cultural perceptions of musicality and musical expertise. In: M. Barrett (ed.), A Cultural
Psychology for Music Psychology. Oxford: Oxford University Press (in press)
Haslinger B, Altenmüller E, Castrop F, Zimmer C, Dresel C. Sensorimotor overactivity as a
pathophysiologic trait of embouchure dystonia. Neurology 2010; 74: 1790–1797.
Helliwell PS Taylor WJ. Repetitive strain injury. Postgrad Med J. 2004 Aug;80(946): 438–43.
Herrojo-Ruiz M, Jabusch HC, Altenmüller E. Detecting Wrong Notes in Advance: Neuronal
Correlates of Error Monitoring in Pianists. Cerebral Cortex 2009; 19: 2625–39.
Hill DM, Hanton S, Fleming S, Matthews N. A re-examination of choking in sport. Eur J Sport Sci
2009; 9: 203–212.
Hill DM, Hanton S, Matthews N, Fleming S. Choking in sport: A review. Int Rev Sport Exerc
Psychol 2010; 3: 24–39.
Ioannou CI, Altenmüller E. Psychological characteristics in Musician’s Dystonia: A new diagnostic
classification. Neuropsychol. 2014; 61: 80–88.
Jabusch HC, Altenmüller E. Anxiety as an aggravating factor during onset of focal Dystonia in
musicians. Med Probl Perf Arts 2004; 19: 75–81.
Jabusch HC, Müller SV, Altenmüller E. High levels of perfectionism and anxiety in musicians with
focal dystonia. Mov Disord 2004:19; 990–991.
Jabusch HC, Zschucke D, Schmidt A, Schuele S, Altenmüller E. Focal dystonia in musicians:
treatment strategies and long-term outcome in 144 patients. Mov Disord. 2005; 20: 1623–1626.
Jankovic J, Ashoori A. (Movement disorders in musicians. Move Disord 2008;23(14): 1957–1965.
Jensen KB, Srinivasan P, Spaeth R, Tan Y, Kosek E, Petzke F, Carville S, Fransson P, Marcus
H, Williams SC, Choy E, Vitton O, Gracely R, Ingvar M, Kong J. Overlapping structural and
functional brain changes in patients with long-term exposure to fibromyalgia pain. Arthritis
Rheum 2013 Dec;65(12): 3293–303.
Karni A, Meyer G, Rey-Hipolito C, et al. The acquisition of skilled motor performance: Fast and
slow experience-driven changes in primary motor cortex (M1). Proc Natl Acad Sci U S A 1998;
95: 861–868.
Klämpfl MK, Lobinger BH, Raab M. How to detect the yips in golf. Human Mov Sci 2013; 32:
1270–1287.
Konczak J, Abbruzzese G. Focal dystonia in musicians: linking motor symptoms to somatosensory

dysfunction. Front Hum Neurosci 2013, 25;7:297. doi: 10.3389/fnhum.2013.00297.
Lang AE, Voon V. Psychogenic movement disorders: past developments, current status, and future
directions. Mov Disord 2011;26(6): 1175–86.
Lederman RJ. Focal dystonia in instrumentalists: clinical features. Med Probl Perform Art 1991; 6:
132–136.
Lee A, Altenmüller E. Heavy-Metal-Curse: A task-specific dystonia in the lower limb of a
professional percussionis Med Prob Perform Art (2014): in revision.
Lee A, Tominaga K, Furuya S, Miyazaki F, Altenmüller E. Task Specific Tremor in String Play-
ers: Evidence of Coactivation in a specific frequency range. Mov Disord 2013; Jul 12: doi:
10.1002/mds.25569.
Lim V, Altenmüller E. Musicians Cramp: Instrumental and Gender Differences. Med Probl Perform
Art 2003; 18: 21–27.
Lobinger B, Klämpfl M, Altenmüller E. We are able, we intend, we act-but we do not succeed: A
theoretical framework for a better understanding of paradoxical performance in sports. 2014.
Submitted
McGaugh JL. Memory-a century of consolidation. Science 2000; 287: 248–251.
Mezaki T (2007) Dystonia redefined as central non-paretic loss of control of muscle action: a
concept including inability to activate muscles required for a specific movement, or ’negative
dystonia’. Med Hypotheses. 2007;69(6): 1309–12.
Moore RD, Gallea C, Horovitz SG, Hallett M. Individuated finger control in focal hand dystonia:
an fMRI study. Neuroimage 2012 Jul 16; 61(4): 823–31.
Münte TF, Altenmüller E, Jäncke L. The musician’s brain as a model of neuroplasticity. Nature
Reviews Neuroscience 2002; 3: 473–478
Nakashima K, Kusumi M, Inoue Y, Takahashi K. Prevalence of focal dystonias in the western area
of Tottori Prefecture in Japan. Mov Disord 1995; 10: 440–443.
Nutt JG, Muenter MD, Melton LJ, Aronson, A. Kurland LT Epidemiology of dystonia in Rochester,
Minnesota. Advances in Neurology 1988; 50: 361–365.
Otten M. Choking vs. clutch performance: A study of sport performance under pressure. J Sport
Exerc Psychol 2009; 31: 583–601.
Packard MG, Cahill L, McGaugh JL. Amygdala modulation of hippocampal-dependent and caudate
nucleus-dependent memory processes. Proc Natl Acad Sci U S A 1994; 91: 8477–8481.
Paulig J, Jabusch, HC, Großbach M, Boullet L, Altenmüller E. Sensory trick phenomenon improves
motor control in pianists with dystonia: prognostic value of glove-effect “. Frontiers in Cognitive
Neuroscience 2014 (in revision).
Pavlov IP. Dinamicheskaya stereotipiya vysshego otdela golovnogo mozga [The dynamic stereotype
of the higher section of the brain]. In: I. P Pavlov, polnoe sobraniie sochineniy (2d ed., enlarged,
Vol. 3, Pt. 2, pp. 240–244). Moscow: Izdatel’stvo Akademii Nauk SSSR 1951.
PesentiA, Barbieri S, PrioriA. Limb immobilization for occupational dystonia: a possible alternative
treatment for selected patients. Adv Neurol 2004; 94: 247–254.
Prinz, W. Modes of linkage between perception and action. In W. Prinz & A.-F. Sanders (Eds.),
Cognition and motor processes (pp. 185–193). Berlin: Springer 1984.
Rosset-Llobet J, Candia V, Fàbregas S, Ray W, Pascual-Leone A. Secondary motor disturbances in
101 patients with musician’s dystonia. J Neurol Neurosurg Psychiatry 2007 Sep;78(9): 949–53.
Schmidt A, Jabusch HC, Altenmüller E, Hagenah J, Brüggemann N, Lohmann K, Enders L, Kramer
PL, Saunders-Pullman R, Bressman SB, Münchau A, Klein C. Etiology of musicians dystonia:
Familial or environmental? Neurology 2009;72(14): 1248–1254.
Schmidt A, Jabusch HC, Altenmüller E, Kasten M, Klein C. Challenges of making music: What
causes musician’s dystonia? JAMA Neurol 2013;70(11): 1456–9.
Schuele SU, Jabusch HC, Lederman RJ, Altenmüller E. Botulinum toxin injections of musician’s
dystonia. Neurology 2005; 64: 341–343.
Soeter M, Kindt M. High trait anxiety: a challenge for disrupting fear memory reconsolidation.
PLoS One 2013 18;8(11):e75239. doi: 10.1371/journal.pone. 0075239.
Spence JT, Spence KW. The motivational component of manifest anxiety: Drive and drive stimuli.
In C. Spielberger (Ed.), Anxiety and behavior (pp. 291–326). New York, NY: Academic Press
1966.
Sripanidkulchai K, Sripanidkulchai B, Wyss JM: The cortical projection of the basolateral amyg-
daloid nucleus in the rat: A retrograde fluorescentdye study. J Comp Neurol 1984;229:
419–431.
Steele CJ, Bailey JA, Zatorre RJ, Penhune VB. Early musical training and white-matter plasticity
in the corpus callosum: evidence for a sensitive period. J Neurosci. 2013 Jan 16;33(3):1282–90.
doi: 10.1523/JNEUROSCI.3578-12.2013.
Steinmetz, A., Stang, A., Kornhuber, M., Röllinghoff, M., Delank. K.S., Altenmüller E. From
embouchure problems to embouchure dystonia?—A survey of self-reported embouchure dis-
orders of 585 professional orchestra brass players. Int Arch Occup Envi Health 2013; DOI:
10.1007/s00420-013-0923-4.
Steptoe A, Fidler H (1987) Stage fright in orchestral musicians: A study of cognitive and behavioural
strategies in performance anxiety. Br J Psychol 1987; 78: 241–249.
Szeto GP, Lin JK. A study of forearm muscle activity and wrist kinematics in symptomatic of-
fice workers performing mouse-clicking tasks with different precision and speed demands. J
Electromyogr Kinesiol. 2011;21(1): 59–66.
van Vugt FT, Boullet L, Jabusch HC, Altenmüller E. Musician’s dystonia in pianists: Long-term
evaluation of retraining and other therapies. Parkinsonism Rel Disord 2014;20(1): 8–12.
Vaquero L, Hartmann K, Ripollés P, Rojo N, Sierpowska J, Cámara E, Mohammadi B, Samii A,
Münte TF, Rodriguez-Fornells A, Altenmüller E. The Earlier, the Smaller: Neuroplastic Changes
in Professional Pianists Depend on Age of Onset 2014: (Submitted to cerebral Cortex).
Walter U, Buttkus F, Benecke R, Grossmann A, Dressler D, Altenmüller E. Sonographic Alteration
of Lenticular Nucleus in Focal Task-Specific Dystonia of Musicians. Neurodegener Dis 2012;
9: 99–103.
Wan CY, Schlaug G. Music making as a tool for promoting brain plasticity across the life-span. The
Neuroscientist 2010; 16: 566–577.
Wilson GD. Performance anxiety. In D.J. Hargreves & A.C. North (Eds.), The social psychology
of music (pp 229–248). Oxford: Oxford University Press. 1997
Wilson M. From processing efficiency to attentional control: A mechanistic account of the anxiety-
performance-relationship. Intern Rev Sport Exerc Psychol 2008; 1: 184–201.
Windholz H. Pavlov’s conceptualization of the dynamic stereotype in the theory of higher nervous
activity Am J Psychol 1996;109(2): 287–295.
Wulf, G. (2007). Attention and motor skill learning. Champaign, IL: Human Kinetics.
Yerkes RM, Dodson JD. The relation of strength of stimulus to rapidity of habit-formation. J Comp
Neurol Psychol 1908; 18: 459–482.
Yoshie M, Kudo K, Murakoshi T, Ohtsuki T. Music performance anxiety in skilled pianists: effects
of social-evaluative performance situation on subjective, autonomic, and electromyographic
reactions. Exp Brain Res 2009;199(2): 117–26.
Zatorre RJ, Chen JL, Penhune VB. When the brain plays music: auditory-motor interactions in
music perception and production. Nat Rev Neurosci. 2007 Jul;8(7): 547–58.
Chapter 12
Motor Control and the Injured
and Healthy Artist
Roger M. Hobden and Samuel Tétreault
12.1 The Origin of Art
The origin of art is probably related to the origin of symbolic thought (Lorblanchet
2007). The human being, like other animals, is able to partly modify his environment
and the objects contained therein. By modifying objects, human beings are able to
increase the information contained in those objects (Schneider 2006; Koch 2014).
These objects acquire information that they would have never been able to acquire
by themselves by being exposed to the forces in the environment. Modified objects
become tools that can be used by their creator and by other humans. The information
contained within these objects is thus also transferred in part to other humans. Human
beings can also create abstract objects, like words and symbols, which also can be
used to transfer information. Some of these words and symbols can be used to describe
the laws of nature, and thus there is a relation between art and science through the
action of the creative mind.
Art is present in all aspects of life. Virtually all man-made objects that we interact
with during normal life have a specific artistic stamp—the bed we sleep in, the
shower we use, the clothes we wear, the plate and glasses we use to eat breakfast, the
vehicle we travel with, the building where we work, the novels we read, the music
we listen to, etc. Every man-made object in a city, town, or village has an artistic
R. M. Hobden ()
Department of Family Medicine and Emergency Medicine,
University of Montreal, Montreal, QC, Canada
Tel.: 514-527-2361
Department of Family Medicine, CLSC des Faubourgs, 1705 de la Visitation,
Montreal, QC H2L 3C3, Canada
S. Tétreault
Les 7 Doigts de la Main, 225, Roy est (bureau #205), Montreal, QC H2W 1M5, Canada
Tel.: 514-521-4477

and Biology 826, DOI 10.1007/978-1-4939-1338-1_12
180 R. M. Hobden and S. Tétreault
component. An artless existence is simply unthinkable. Art is part and parcel of the
human existence. There is no human being that is untouched by art.
12.1.1 Lessons from Artists
In the area of neurophysiology, much progress has occurred from working with in-
jured patients. Correlating damaged areas of the body with loss of function has led to
many insights about function. The disadvantage of this approach is that documenting
a lesion will only tell you that a specific area of the nervous system plays an important
role in a specific disease, but not clearly what that role might be. To use an image: if
you remove a transistor from a radio, and hear a screeching sound when the radio is
turned on again, this does not mean that you have removed “the screech inhibitor.”
The advantage of doing clinical and experimental research on high-level perform-
ers like elite athletes, dancers, musicians, circus artists, singers, etc., is that these
performers have become experts in the mastery of their chosen discipline. As soon as
something is “not quite to right,” they are able to describe pain and loss of function
that interfere with performance. These verbal descriptions and practical demonstra-
tions are usually very accurate and precise (valid and reliable). Since high-level
performing artists function at the extreme limits of human capacity, they can be a
unique and useful source of information for the clinician and the researcher who
wants to understand how the human body can work at an optimum level. We will
recall that Nikolai Bernstein not only studied skilled workers but also elite pianists,
for instance, documenting with kymocyclography that keyboard musicians could
play hundreds of notes per minute (Kursell 2006).
The interest in the injuries of artists has a long history. The first textbook that
mentions injuries in musicians is Bernardino Ramazzini’s 1713 treatise, Diseases of
Tradesmen (Ramazzini 1933). With the advent of the pianoforte in the nineteenth
century, composers started taking advantage of the properties of this new instrument
to create musical pieces that required increased speed, power, and sound intensity.
With increasing performance requirements came increasing injury rates, and a few
physicians started to develop an expertise in the treatment of these ailments (Poore
1887). Some of these injuries would now be recognized as tendinopathies, and others,
known as musician’s cramp, now fall under the heading of focal dystonias. Different
medical approaches were devised to address these various ailments and obstacles to
enhanced performance. For example, it was well acknowledged by piano teachers
and students that the fourth finger had less mobility and power than the others and,
eventually, physicians invented an “ingenious” solution to this problem: tenotomy (a
surgical procedure) of the secondary tendons of the ring finger. A few surgeons spe-
cialized in this technique and traveled by invitation to practice their art on those who
wished. The long-term results were less than convincing, however, and this practice
was eventually abandoned (Sataloff et al. 2010). The famous Romantic piano com-
poser, Robert Schumann, developed a hand injury that prevented him eventually from
practicing his instrument. Current research indicates that this condition, affecting his
12 Motor Control and the Injured and Healthy Artist 181
third finger, may have been a focal dystonia (Garcia de Yebenes 1995). Indeed, in the
early 1830s, Schumann created the Toccata in C Major Op. 7, an extremely difficult
piece that can nonetheless be entirely played without the use of the third finger of
the right hand (Alltenmüller et al. 2005).
The rapprochement between artists and health professionals has led to interesting
discoveries. For instance, the technique of indirect laryngoscopy was invented by a
professional opera singer, Manuel Garcia, who used a dental mirror to examine the
vocal cords of his students. Garcia eventually presented his findings at the Royal
Society of Medicine in 1855 (Sataloff et al. 2010). In the second half of the twentieth
century, there was a major increase of scientific activities in relation to health, disease,
and performance of the professional artist. There are presently annual conferences
devoted to arts medicine in North America, Europe, and Asia. Some cover all of
the performing arts, and others are more focused, for example, on the injuries of
musicians or dancers. Scientific journals dedicated to health and disease in music,
voice, and dance have been in existence for over a quarter of a century. Indeed, the
effect of music on the brain is an area of research in neurology that has increased by
many orders of magnitude since the early 90s.
The visual arts field also has not been left out, as more researchers are exploring
the links between visual perception and artistic creation in the areas of painting and
other visual arts (Conway and Livingstone 2007; Pinna 2011).
12.2 How Artists Learn and Train Depends on the Type

of Discipline Considered
Let us take the example of dance. In dance, students will acquire very specific skills
in classroom situations. For instance, ballet students will typically start training when
they are 8 years old or even earlier. A teacher will demonstrate how to perform ballet
movements correctly and the student will attempt to reproduce these movements,
while the ballet teacher corrects the student until the proper movement has been
achieved. In the case of ballet, the quality of movement is of the utmost importance.
It is not simply a question of standing still with the feet turned out and of bending
the knees. This action must be done in a very fluid and controlled way that appears
effortless and flawless. The dance student must learn to minimize jerkiness by the
fine control of the equilibrium of the posture of the body and the smooth activation
of the muscles. The student first learns to do this on both legs, and then on one
leg (the supporting leg), during which the other leg (the gesture leg) will describe
different patterns in space. These patterns have different names and form part of the
ballet vocabulary. There are a few variations in terminology, depending on which
ballet method is used (Vaganova, Cecchetti, Bournonville, etc.), but in essence the
similarities are greater than the differences. Learning in a ballet class depends a lot
on visual cues and the ubiquitous presence of the mirror, which helps the student to
self-correct the attempted movements.
The comparison of the ballet dancer’s movement with those of others in the class
can be stressful. Because of the values of perfectionism that are part and parcel of
ballet, the student will always perceive herself as imperfect: too tall, too short, too
fat, too slow, etc. One of the challenges of a good dance teacher is to convince the
student that she must strive to be the best of what she is and not a failed copy of
which she is not.
Eating disorders are commonly associated with the word “ballet,” even though
research has failed to demonstrate that anorexia and bulimia are more prevalent in
ballet schools than in the general population. The scientific consensus at this time
is that eating disorders are influenced by genetic and epigenetic factors, and family
dynamics, but are triggered by specific psychosocial events and interactions (Treasure
et al. 2010). An example of trigger would be a boyfriend or a dance teacher telling the
student that it would be desirable for her to lose weight. One of the central aspects of
certain eating disorders is that the individual has a distorted visual image of herself.
This distortion is in fact an enhanced visual perception of the body that is associated
with the notion of being overweight (Keizer et al. 2013). Any minute bulge of the
skin will be detected with the eyesight of an eagle. The action–perception issues
associated with this medical condition have recently started to attract the attention
of researchers (Guardia et al. 2010).
One of the challenges of teaching ballet is that even though the ballet teacher
demonstrates exactly the same sequence of movements to a dozen different pupils,
each of these students will perceive what has been shown and explained differently.
This perception will be based on the psychological and physical properties of the
student (Gibson 1987). Thus, a sequence of movements will be seen as doable,
partially doable, or not doable. This will depend on the state of the components of
fitness of the student: coordination, flexibility, strength, endurance, power, etc., but
also on the genetic potential of the student, similar to that of any other individual, to
acquire the required performance level (Bouchard 2012).
One aspect, which differentiates ballet from sports training, is that there is no
individualization of the training program. No block of time is set aside in the weekly
schedule so that the student can work privately to improve her own specific motor
skills. The dance student is never assessed individually to identify her training needs.
Corrections are usually given “on the fly” within the framework of the dance class.
12.2.1 The Challenges of Effortless Appearance
Another aspect that differentiates dance from most sports is that the motor actions
need to appear effortless. Let us take, for example, a “grand rond de jambe en l’air.”
The dancer starts the movement standing on one leg, with the other leg extended
in front of her, parallel to the ground. Keeping the knee straight, the dancer will
then gradually bring the gesture leg outwards towards abduction, and then bring the
leg backwards, always staying in the same horizontal plane. At the beginning of the
movement, the sole of the foot is facing the ground, and, at the end, it is facing
the ceiling. What happens is that, as the leg travels from forward to backwards,
the leg rotates so that the load is transferred, apparently without any effort, from
one group of muscles to another, along the upper arc that is part of the cylinder of
continuous muscles that extend from the thigh to the ankle. Naturally, all the muscle
groups must be co-activated to maintain the shape of the leg, even though the greatest
muscle activation will be provided by a band of motor units that are supporting the
mass of the leg at any given time.
The appearance of effortlessness must also be maintained during ballistic move-
ments. As the dancer performs a “grand battement devant,” the leg must be shot high
up in the air as fast as possible, yet the rest of the body must appear completely im-
mobile. Rapid movements are more challenging because they require more kinetic
energy. Kinetic energy is related to the mass and the square of the speed of an object.
So if the leg is travelling at a speed of three units, the mass will be multiplied by
nine, whereas if the leg travels at a speed of ten units, the mass will be multiplied
by 100, which is an order of magnitude higher than a movement three times as slow.
The body of the dancer must not only be able to generate such high kinetic energy
but must also be able to control, as the movement is taking place, the exact trajectory
of the leg according to the demands of the piece.
Naturally, the dancer must be able to simultaneously mobilize all necessary forms
of energy to stabilize the rest of the body and give this appearance of perfect immobil-
ity of the trunk and the standing leg. All dance students have different characteristics
(power, flexibility, endurance, etc.), and so the movement solutions chosen by the
dancer must be “good enough” to accomplish the task according to the demands of
the teacher, and not necessarily “optimal.” As the student learns to handle movement
of all her body parts as an integrated whole, less energy will be wasted to stabilize
limb trajectory and body shape, and more attention can be focused on decomposing a
global movement into the smaller details that make the dance piece more interesting
for the spectator, including the possibility that the movement is voluntarily jerky, if
wished for by the choreographer.
12.3 Differences and Similarities Between Artists

and the General Population
The professional artist needs to be performing at the highest skill level. His employ-
ability depends essentially on this characteristic. Artists have to be different from the
rest of the population. The saying that “it’s the little differences that make the big
difference” applies especially well to the arts world. The artist can be seen as an ideal
model of what can be accomplished by the human body with specialized training.
Motor coordination will develop slowly based on exercise and experimenting with
the task at hand. At the early stages of training, the artist will be overwhelmed with
the excessive number of degrees of freedom related to the task: playing this note
versus that note, landing on this space and not another, putting the hand here instead
of there to catch the ball, etc. Sensory feedback given by the outcome of a determined
movement will permit self-corrections and stabilization of the intended task. Over-
coming peripheral indeterminacy by coordination of movement will lead to relative
stability of goal-directed performance, accompanied by a sense of self-confidence
(Bernstein 1967; Reppa et al. 2012).
When the artist has achieved mastery of his given discipline, the motor impulses
used to control movement will have become more efficient and will be used to
guide the passive elements (bone, water, fat, cells, etc.) of the limbs and digits
in the direction of the intended goal. The inertial properties, reaction forces, and
interaction torques of these passive elements then cease to be a problem, and in fact
become part of the solution of the required movement. All this is accomplished by
online constant afferent feedback from the mechanoreceptors (Bernstein 1967; Silva
2012). The information that arises from the periphery is also transmitted through
the body by mechanical waves that travel at the speed of sound through the various
tissues of the organs. These waves travel much faster than that of neural transmission.
To what extent this mechanical information can be utilized by the central nervous
system (CNS) is as yet unknown. Certainly, most of the cells of the human body,
including neurons, are equipped with primary cilia (Anderson et al. 2008). Many of
these primary cilia have a role in mechanical perception at the cellular level (Tobin
and Beales 2009). The existence of such primary cilia may also explain in part the
favorable effects of strength and aerobic training on health and disease (Trilk and
Phillips 2014).
12.4 Circus Disciplines
In contrast to dance, voice, and music, up until now, the circus disciplines have not
attracted a lot of attention on the part of health and fundamental researchers, and the
existing studies are mostly of an epidemiological nature (Shrier et al. 2009; Wanke
and Hamilton 2012). This may be in part due to the fact that many circus injuries are
also commonly found in various sport disciplines, and thus the transfer of basic and
clinical knowledge can be made more readily.
Still, the unique blending of motor and psychological skills with the goal of
activating an artistic effect on the audience adds a new layer of complexity to the
accomplishment of the circus movements. Contrary to his athletic counterpart, the
circus artist cannot afford to be in a psychological cocoon, oblivious to the reactions
of the spectators. Quite the opposite, the successful circus artist must be able to reach
out and connect with each person in the audience.
In this section, we will review the main circus disciplines and the motor control
issues that arise within each of them. Circus disciplines can be classified into five main
families: aerial disciplines, equilibristic disciplines, acrobatic disciplines, juggling
and object manipulation, and clowning disciplines.
Fig. 12.1 Aerial hoop.

(Photo: Lionel Montagnier,
artist: Alexandra Royer in
Sequence 8 by Les 7 Doigts
de la Main)
12.4.1 Aerial Disciplines
Aerial disciplines require specific abilities and motor skills. They require a predom-
inance of upper body work and pulling against gravity. The construction of these
movements could be addressed by motor control levels A and B, as proposed by
N.A. Bernstein (Bernstein et al. 1996). Examples include aerial hoop (Fig. 12.1),
static trapeze, flying trapeze, swinging trapeze, and vertical rope. A high level of pro-
prioception and spatial orientation is required. These disciplines present a significant
level of danger and risk, and psychological issues including the fear of falling.
Fig. 12.2 Hand balancing.

(Photo: Louis Ducharme;
artist: Samuel Tétreault)
12.4.2 Equilibristic Disciplines
Equilibristic disciplines require, in contrast, the ability to push against gravity. Con-
stant postural corrections are required on the part of the stabilizing muscles. These
might correspond to Bernstein’s levels B and C1. Examples include hand balancing
(Fig. 12.2), tightwire, high wire, slack rope, rola bola, and hand to hand. The per-
fect balance that is required is not immobility but rather a dynamic balance (active
correction of posture and the ability to perform a task while maintaining a stable
position). Psychologically, this type of discipline requires from the artist the ability
to sustain concentration at every instant.
Fig. 12.3 Chinese hoop

diving. (Photo: Valérie
Remise; artist: William
Underwood in TRACES by
Les 7 Doigts de la Main)
12.4.3 Acrobatic Disciplines
Acrobatic disciplines require explosive movements (fast-twitch fibers are more in

demand than slow-twitch fibers) and are characterized by a predominance of lower
body work. These might be assigned to Bernstein’s levels C2 and D. Examples of
these disciplines include Chinese hoop diving (Fig. 12.3), Korean board, Russian
bar, and Chinese pole. Spatial orientation and proprioception are very important,
and there is a high level of risk, so fear of injury needs to be addressed.
12.4.4 Juggling and Object Manipulation
Juggling and object manipulation requires excellent hand–eye coordination skills.

These skills would be an example of control level D according to Bernstein. There
Fig. 12.4 Cigar box juggling. (Photo: Sylvie-Ann Paré; artist: Eric Bates in Sequence 8 by Les 7
Doigts de la Main)
is a high demand on upper limb dexterity. Examples of these disciplines include

cigar box juggling (Fig. 12.4), club juggling, and diabolo. Juggling can also be done
with partners. A large amount of practice is necessary to accomplish juggling acts.
Jugglers tend to have mathematical and musical types of mind (patterns, precision,
timing; Polster 2003). Indeed, the first recorded mathematician-juggler is believed to
have been Abu Sahl al-Quhi, a mathematician living in Baghdad in the tenth century
(Buhler et al. 1994).
12.4.5 Clowning Disciplines
Clowning disciplines find their origins in the grotesque characters of ancient cultures
and in the early forms of theatre such as commedia dell’arte. Both classical and
contemporary circus clowns very often combine their specific artistic performance
with other circus disciplines and therefore there are no specific motor control skills
or body types required.
12.5 Training in the Circus Arts
Learning is more a question of differentiation rather than a question of enrichment

(Gibson and Gibson 1955). There is no reason to believe that artists use differ-
ent strategies to learn than other human beings. Based on experience, coaches are
adept in identifying those individuals with enhanced motor-perceptual skills that are
“trainable” within a specific discipline. The existence of individuals with distinct
or exceptional qualities is well-recognized, and even if there are many competitive
descriptive frameworks that address this (see, for example, Gardner 1983), satisfying
explanations from a neurophysiological, biomechanical, and biochemical point of
view are still lacking. Naturally, some disciplines, like contortion, require a genetic
and epigenetic makeup that is truly exceptional. Identifying to what extent these
super-flexible individuals are similar or different to the majority could enhance our
understanding of the interaction between gene expression and training with regard to
flexibility in the general population, and could lead to insights about injury prevention
in sports and in the workplace.
While the artist trains, he or she compares qualitative differences (Verri and Pog-
gio 1989) such as bigger than, farther than, faster than, etc. For instance, when a
juggler tries to catch a ball that is thrown in the air, the optical flow to his eye is
compared with previous experiences of watching movable objects transiting in every
manner possible. The immediate perception he has of his own body will tell him
instantly if he has even the remotest possibility of catching that ball. This is based
not only on the relative speed and direction that the ball is traveling but also on his
training and how his body perceives the state of health of his limbs (injury in the
shoulder, etc.). Eventually, these tasks will become well-learned habits that the cir-
cus artist can count on to be accomplished nearly automatically, whether the artist is
immobile, or traveling along a well-known trajectory. When confronted with the new
situation, more areas of the brain will be solicited to be able to compare known ex-
periences versus the necessity of inventing new tasks. As the circus student becomes
a professional, the possibilities afforded by the environment and its objects will be
modified because the artist’s capabilities have been modified. This training process
entails an updating of the brain’s active operative representation (modeling) of the
external world (Bernstein 1967). Whether this process is qualitative or quantitative
is presently the object of much debate (see, for example, Ostry and Feldman 2003).
When exposed to the same array of objects (ropes, balls, ladders, etc.), a nonartist
will perceive different affordances from a circus artist.
General principles of circus training include physical preparation in strength and
flexibility, basic acrobatic classes and biomechanics (floor and trampoline), de-
composition of complex movements into simpler ones, progression using educative
movements, use of safety lines and crash mats, repetition until patterns and sequences
become integrated, consistency, and endurance.
Psychological issues are also very important, as the artist must learn to plan
near into the future the complex movements related to his or her discipline. “Death
defying” acts have always been part and parcel of the traditional circus trademark
Fig. 12.5 Swinging trapeze. (Photo: Yann Boyenval; artist: Danica Plamondon in PSY by Les 7
Doigts de la Main)
while in the contemporary circus arts the focus is less on sensationalism and more
on the artistic research and individual expression. Commonly, especially during the
first half of the twentieth century, many aerial artists boasted about their capacity
to work without a safety net. Current research is now looking into these and other
psychological skills (Shrier and Halle 2011). Learning to master the fear of death is
intimately tied with the human capacity to plan ahead (Bernstein 1967).
12.5.1 Training in Aerial Disciplines
Morphological qualities include a smaller and lighter body type, good flexibility,
and good lines, and the proportion of slow-twitch versus fast-twitch fibers depends
on the type of aerial discipline. Psychological skills include being comfortable with
heights, courage, a high tolerance to pain (for instance, caused by the friction of
the rope or twine), the capacity to visualize complex sequences of movements and
thereby their mechanics, risk management (security), and rigging knowledge. Action
and perceptual skills include hanging and suspension postures (from hands, knees,
feet, heels, etc.), spatial orientation without contact with the ground, swinging on
a trapeze, for example, (Fig. 12.5) and tempo movements, feeling the “dead point”
(moment of weightlessness), and flipping and twisting movements.
12.5.2 Training in Equilibristic Disciplines
Morphological qualities are believed to require a preponderance of slow-twitch mus-

cle fibers, and the ideal body type will depend on the discipline. For example, good
shoulder and elbow flexibility is required for hand balancing or hand-to-hand ac-
tions (Fig. 12.6). Psychological skills include attention to details, being meticulous,
having the capacity to concentrate over long periods of time, patience and perse-
verance, calmness, understanding the biomechanics of balance, and the capacity to
visualize shapes and postures in terms of balance structures. Motor control skills
include consciousness of the base (anchor) points and extremities, consciousness
of body alignment, posture and muscle tone, proprioception accuracy, standing and
balancing postures using different body parts as anchors (hands, feet, head, knees,
etc.), dynamic balance (active correction of posture), isolation of movement within
a fixed shape, and steady movement.
12.5.3 Training in Acrobatic Disciplines
Morphological qualities include a shorter, muscular body type with fast-twitch fiber
predominance. Psychological skills include a courageous and daredevil spirit, the ca-
pacity to visualize complex sequences of movement and their biomechanics, greater
attention to the whole and less detail-oriented, and risk management (security). Motor
control skills include jumping and landing from and to different surfaces, spatial ori-
entation and fast movement combinations, quick trajectory and velocity corrections,
initiating and stopping rotation, explosive movements, and flipping and twisting
movements such as those seen in Fig. 12.7 on the Korean board.
12.5.4 Training in Juggling and Object Manipulation
No specific morphological qualities are required for this discipline (body type, flexi-
bility, muscle type). Psychological skills include patience, perseverance to the point
of obsessiveness, a mathematical/musical mind type, and the capacity to visualize
complex patterns of intricate upper limb movement (Beek and Turvey 1992). Of
note, there is a predominance of male jugglers. Motor control skills include a high
level of hand–eye coordination and dexterity; relation with objects in space and time;
throwing, catching, and manipulating objects of different shapes and weights using
different body parts; and consistency in highly repetitive movements.
Fig. 12.6 Hand to hand. (Photo: ODC Photo; artists: Sebastien Soldevila and Émilie Bonnavaud
in La Vie by Les 7 Doigts de la Main)
Fig. 12.7 Korean board. (Photo: Sylvie-Ann Paré; artists: Maxim Laurin and Ugo Dario in
Sequence 8 by Les 7 Doigts de la Main)
12.5.5 Training in the Clowning Discipline
What mostly defines the clown artists is found on a psychological level: A great sense
of humor and self-derision is an essential personality trait to any good clown. Clowns
often provoke laughter by exposing their own flaws and vulnerabilities, appearing
Fig. 12.8 Russian bar. (Photo: Lionel Montagnier; artists: Eric Bates, Tristan Nielsen and
Alexandra Royer in Sequence 8 by Les 7 Doigts de la Main)
clumsy or failing in what they attempt. They are a mirror for what we usually do
not want to see in ourselves or try to hide from others and therefore clowns need to
have an acute sense of observation for their own behaviors and those of others. It
is not rare to see some of the funniest clowns on stage being rather moody or even
depressed off stage.
12.5.6 Partnering and Group Disciplines
In terms of morphological quality, a “flyer” (the acrobat who is being carried, thrown,
and caught) will usually have a shorter and lighter body type, while the “base”
or “catcher” has a taller and stronger body type (Fig. 12.8). Psychological skills
include the capacity and interest for teamwork, capacity to put one’s trust in someone
else and to be trustworthy, verbal and nonverbal communication skills, and the
capacity to visualize movement sequences involving the simultaneous movement of
others. Motor control skills include the capacity of throwing and catching people
using different body parts (hands, feet, shoulders, back), giving and taking weight
and counterweight, letting go of the automatic reflex of balancing yourself (flyer)
and letting the “base” balance you, developing synergistic movement with someone
(synchronized “tempo”) to combine pushing actions, high adaptability, and requiring
quick adjustments (the other person is a variable you cannot control).
12.5.7 The Live Performance Has Specific Issues of Its Own
The artist has to deal with psychological factors like the stress of performing in
front of the public, the fear of failing or falling, mishaps, and last-minute changes.
Outside factors include the conditions of the show: lighting, musical cues, different
stage sizes, floor types and ceiling heights, etc. This is where experience and training
allow the artist to transcend the acrobatic elements and motor skills and to reach a
sense of freedom in his artistic expression.
12.6 Basic Considerations About Motor Control and Injury
In the past 150 years, much progress has been made in the area of motor control. In
particular, the area of synergies and their origin have been the object of many pub-
lications and discussions, including the proper definition of the word itself (Latash
2012; Turvey 2007). The creation of a neologism that captures exactly “what is
meant” might contribute usefully to the standardization of motor control terminol-
ogy. For the moment, let us consider that there may be three kinds of synergies:
inorganic synergies like those of the lattice that unites the molecules of a rock, or-
ganic synergies characterizing the structure and metabolism of living things, and
action synergies, which are the goal-directed movements of living things and their
interaction with the environment. The last two categories may be corollaries of each
other. Action synergies are driven by the demands of the organism and its interactions
with the environment. The similarities between the goals of a unicellular organism
and those of a complex animal like a mammal are obvious. All living organisms
need to consume molecules to generate energy, need to gain access to sources of
energy, need to maintain their structural integrity, and need to reproduce themselves
(Pross 2012). Indeed, it may even be hard to draw the line between metabolism and
structure. For instance, the cytoplasm also contributes to the mass, shape, and struc-
ture of the cell, through both osmotic and hydrostatic pressure (Myers et al. 2007).
Another consideration is that all living organisms are constrained by physical factors
like pressure (the pressure of water or of the atmosphere), temperature (molecular
agitation and the threat of loss of information), electromagnetic waves, mechanical
waves, and gravity.
What the CNS seems to learn are correlations between changes of body sensation
and visualization of body movement, so that a given dynamic image of the change
of the body configuration is associated with a change of body perception. The unit of
control of the CNS is probably not the muscle, but rather the motor unit. Muscles do
not seem to exist as far as the controller is concerned. The controller would seem to
recruit, through threshold modification, the number and location of the motor units
necessary to accomplish the required task. The elementary component of movement
confined in the organism is most probably the motor unit and its feedback loop (motor
loop). It is through this motor loop that the animal probably perceives part of the
environment, and notably deformation of matter, which is interpreted as “force.” It is
thought that similar loops are related to the existence of consciousness in humans and
animals (for a review, see Dehaene 2014; see also Tononi 2008; Koch and Tsuchiya
2012; Koch 2014).
Injury is a structural breakdown of the normal components of the organism. This
breakdown can affect the metabolism, the external or internal mechanical structure,
the communication system, and any or all of the above. Inflammation, a common
component of injury, will create localized changes in pressure, volume, and tem-
perature. These localized inflammatory changes will disrupt the organic synergies
of the animal or human. For instance, thermodynamic measurements indicate that
native proteins are only marginally stable under physiological conditions (Voet et al.
2012). These localized inflammatory changes may also disrupt action synergies.
Injury can also be accompanied by structural hypermobility, hypomobility, or
both. The coexistence of both is not at all rare: For instance, in the case of a severe
injury to the knee, the joint may have not only an increased mobility because of
a partial ligamentous tear but also a decreased range of motion because of a torn
meniscus.
The existence of organic synergies is not trivial. The proper way that each mechan-
ical part of a living animal will interact is already included in the way that the animal
is constructed. This greatly simplifies motor control, as movements that are forbidden
and movements that are permitted are already built into the biomechanical structure
of the animal. For instance, the knee is constrained to mostly perform flexion and
extension, and the healthy artist does not have to think about producing “impossible”
movements. Indeed, if an injury affects the normal length or tensile properties of the
ligaments, fasciae, etc. of the knee, the artist, because of his enhanced training, will
become very worried because he can instantly tell that “something” is not quite right,
and that he cannot count on his knee or leg to perform flawlessly. Quite often, the
instability can be very subtle, and there is a real risk that the injury will be dismissed
out of hand by an examiner who is not used to dealing with dancers and circus artists,
for instance.
This does not contradict the fact that action synergies are probably controlled
only at the most distal part of the motor arc (or motor circle), the muscle activation
threshold of the muscles of the body (Feldman 2009). According to Bernstein, “the
co-ordination reflex is not an arc but a closed circle with functional synapses at both
ends of the arcs.” Recent studies have supported this point of view (Mattos et al.
2013).
Naturally, the bones that meet at the knee joint have, in reality, six degrees of
freedom, like every other bone of the body, if viewed on a sufficiently small scale.
The debate between those who see the helical axes of the knee as more ligament-
controlled or more muscle-controlled (Blankevoort et al. 1991; Zatsiosrky 1998) is
probably artificial, as the tensegrity (Ingber 2008; Swanson 2013) model, based on
the concept of pre-stress within biological entities (cells, organs, etc.), apparently
reconciles these seemingly opposing viewpoints.
12.7 Injuries in the Artistic Disciplines
The causes of injury in the artistic disciplines are very similar to those in sports.
Injury is usually multicausal. Let us take yet again the example of dance. It is useful
to parse the causes of injury into intrinsic and extrinsic risk factors (Emery 2003;
Liederbach 2010; Ojofeitimi and Bronner 2011).
Intrinsic factors of injuries can be subdivided into modifiable and nonmodifiable
factors. Examples of nonmodifiable factors include sex, age, genetic, epigenetics
training history, and previous injury. Modifiable factors include fitness level, flexi-
bility, strength, stability, proprioception, nutrition, total workload, and psychosocial
factors.
Extrinsic risk factors can also be subdivided into nonmodifiable and modifiable
factors. Some of these nonmodifiable factors would include type of dance, dance
level, and unpredictable accidents. Modifiable factors include the disregard of sound
training principles, predictable accidents, temperature, surface, and choreography.
The modifiable character of this last factor depends naturally on the goodwill of
the choreographer. Indeed, in more than a few instances, one would be excused in
believing that the definition of a choreographer is “someone who creates new ways
of getting injured.”
12.7.1 Injuries and Their Impact on Motor Control of an Artist
Let us take, for example, the case of an artist with an ankle sprain (Gehring et al.
2013a, b; Hubbard-Turner et al. 2013; van der Wees et al. 2006). When a person
sprains his ankle, common clinical findings include the following: (1) inflammation
and swelling, (2) hypermobility of the ligaments, (3) loss of mobility of the ankle
joint, (4) pain, (5) loss of muscle strength, and (6) loss of proprioception. In fact,
most of the injuries of the various joints of the body will generate the same clinical
findings. These findings can be used as outcome variables to help therapists determine
when the injury has resolved. In the case of an ankle sprain, the sprain will be healed
when there is no more inflammation; the ligaments have healed with a length and
stiffness that is as close as possible as the pre-injury state; the range of motion of the
ankle has been restored; the pain is gone; the strength, speed, and endurance of the
muscles are back to normal; and proprioception has been restored.
Let us consider the effect of these six variables on normal motor control.
1. Inflammation will cause abnormal neural messages to be sent to the nervous
system through two main types of stimuli: chemical stimuli, caused by specialized
molecule, and physical stimuli, caused by abnormal increases of volume and
pressure on the nerve endings (Pongratz and Straub 2013). This array of abnormal
stimuli caused by inflammation will serve as a negative feedback loop to inhibit
muscle activity.
2. Hypermobility between two bones will have an inhibiting effect on muscle action
through a different pathway than the previous one. Recall that the relative length
of muscles is regulated through muscle spindles and their relative strength through
the Golgi tendon organ (Kistemaker et al. 2013). The laws of mechanics tell us
that the array of mechanoreceptors measure not only the “action” (force) of the
muscles involved but also the “reaction” (including the inertial properties (heav-
iness) of the bones and other tissues) on which the muscles act. Recent research
suggests that perceived heaviness is related to mass, volume, and symmetry of
the perceived object (Turvey et al. 1999).
When excessive (abnormal) mobility is perceived between two bones, the nervous
system will send instructions to the muscles to protect the body from this excessive
mobility. Thus, muscles will not be allowed to stretch to their maximal length.
The same muscles will also not be allowed to exert their maximal force. Based on
his clinical experience, one of the authors has found that trained artists (circus,
dance, music) are able to sense these effects in great detail and can give very
useful feedback to the health professionals that treat them.
These changes may be triggered because the body perceives that the combined
inertial properties of the bones and other attached structures have changed. Natu-
rally, at the cellular level, the array of Golgi tendon organs involved probably does
not measure “force,” but rather sends out signals in proportion to the deformation
of their physical and chemical structure, signals that are interpreted qualitatively
(“heavier than,” “lighter than,” etc.).
How simultaneous loss of strength and loss of maximal muscle length are me-
diated can be predicted and explained by the equilibrium point theory of motor
control (Feldman 2008). The equilibrium point theory (or threshold control the-
ory) seems to be the only theory that is presently capable of explaining why the
same muscle can be measured as being weak at one end of the range and spastic
at the opposite end of the range, in stroke patients (Levin et al. 2000).
3. Hypomobility is often also accompanied by a loss of strength and a loss of maxi-
mal muscle length. If the muscle is directly damaged, the cause of both the loss of
mobility and of strength is readily apparent. However, there are common injuries
that are accompanied by a misalignment between the bones in the joint space. Sur-
prisingly, this can coexist with the fact that the joint is not necessarily painful, yet
the muscles seem to be “switched off” or “toned down.” In these cases, restora-
tion of anatomical congruence by the means of manual treatments is accompanied
by an instantaneous return of maximal muscle length and strength (Cibulka et al
1986; Suter et al. 1999). Research is ongoing as to the local and regional effects
of these treatments (Dunning and Rushton 2009). Again, the equilibrium point
theory brings a neurophysiological explanation to this common response, which
is observed by health professionals on a daily basis in the clinical setting. Be-
cause of their heightened proprioceptive awareness, artists are able to guide the
therapist in a step-by-step manner during the treatment session to restore proper
mobility of each joint as required for optimal performance.
4. Pain in the tendon or in the muscle (due to a muscle tear, for instance), despite the
pain component being a “subjective” sensation created by either increased focal
pressure or tissue damage (Moayedi and Davis 2013), will directly generate a loss
of strength and resistance to lengthening through “nociceptive” stimuli (antalgic
paresis).
5. Loss of strength and endurance can be caused either directly through damage of
the muscles or tendons or indirectly through damage of the nerves, or through
inhibition caused by inflammation, hypermobility, or hypomobility.
6. Loss of proprioception can be caused either directly through damage of the
nerve endings (mechanoreceptors) or indirectly, because of inflammation,
hypermobility, or hypomobility.
All the joints of the body generate the same reactions to tissue damage. These
phenomena are scale invariant, whether the damage involves a joint of a finger,
the knee joint, or the intervertebral segments. In the case of a microscopic tear of
the intervertebral disc, even without any bulging or change of shape of the disc,
inflammation and disruption of the nerve endings within the annulus fibrosis will
affect muscle function. Certain muscles will lose their maximal strength and
maximal length, and this will disrupt the overall synergies of the trunk muscles.
12.7.2 The Special Properties of the Spine
If the joints of the locomotor system are all the same, as they are indeed histologically,
being composed of bone, cartilage, synovial fluid, joint capsules, ligaments, enthesis
(Benjamin et al. 2004), tendons, muscle, nerves, and skin, why is the spine perceived
to be somehow different?
Yet, the trunk is indeed different, for three main reasons. These reasons are
physical, anatomical, and neurological.
1. Neurologically, the spine and trunk muscles are the oldest part of the locomotor
system, from a phylogenetic point of view. Human consciousness does not have
a clear image of the mobility of the vertebrae, compared to the consciousness
that we have of our upper or lower limbs. With our eyes closed, it is easy for
us to visualize the motion between the phalanxes of the finger, or the motion
between the thigh and the leg. If we try to visualize the rotation of the vertebra
L4 versus L5, we are incapable of doing so. When we turn the trunk to the
right, we will only have a general idea of the motion that takes place between the
vertebrae. Obviously, part of us knows which muscles to activate and how, but our
consciousness does not have access to that information. When we injure a knee,
for instance, it is possible for us to adopt a strategy to reduce nociception, such
as deciding to walk with the knee extended to avoid painful flexion. When we
injure the joint space between L4 and L5, however, there is no way we can have
conscious access to muscle strategies that permit us to dissociate the movement
of L4–L5 from that of the rest of the trunk.
2. From an anatomical point of view, all the trunk movements are multi-muscle
movements that are united through synergies between agonists and antagonists
(Robert et al. 2008; Vora 2010). We would be tempted to say that the trunk
muscles are enslaved to one another, in a similar way and for the same reasons
as the muscles of the fingers (Zatsiorsky et al. 2000). When there is an injury
between L4 and L5, all of the 20 or so muscles that attach to these vertebrae on
each side of the body are affected by whatever is taking place at the joint space.
Thus, all the muscle synergies of the trunk will be affected by this very localized
injury at only one level. Some muscles will receive instructions telling them to
activate less strongly and to not lengthen to the maximal length, but other muscles
will receive the usual instructions they would get to accomplish the given task.
This conflict between normal functioning muscles and abnormal ones will create,
for instance, the so-called spasm that takes place because a muscle is extended
beyond its minimized range, trying to follow its companions. The spasm caused
by palpation is due to the same mechanism: The palpating finger lengthens the
muscle beyond its permitted range, generating a painful reaction.
3. Physically, the trunk is where the center of gravity of the body is situated when
the human being is standing. The gravity line passes through the trunk and the
center of mass projects between the feet. As the body moves around in space, so
does its center of mass (Roussouly and Pinheiro-Franco 2011). Each displace-
ment in space will require activation of the trunk muscles and mobility of the
intervertebral spaces, possibly triggering an increase of inflammation, increase
of hypermobility, etc. It is not possible to physically isolate these vertebral joints
in the same way that one can isolate an injured knee or ankle during locomotion.
Every movement of the whole body requires necessarily a movement involving
all of the vertebrae of the body.
So, for these neurological, anatomical, and physical reasons, spinal dysfunctions
are much more debilitating than injuries of the upper and lower limbs. These are
probably the main reasons why spinal injuries are intuitively perceived by patients and
health professionals to be different from limb injuries, even if the basic histological
structures and physiopathological mechanisms are similar.
Circus artists and dancers develop an exquisite sense of body awareness of their
spinal movements. This enhanced spinal proprioception makes them ideal subjects
to study the normal and abnormal function of the trunk, as they are able to pinpoint
exactly the intervertebral segments that are not functioning correctly. In the situation
where manual treatments are required, they are able to give immediate feedback
as to the effectiveness of a given manipulation. Through the course of a half-hour-
or an hour-long treatment, all the intervertebral segments that are hypomobile can
have their proper motion restored. This is accompanied by a gradual and significant
increase in mobility of the gross movements of the trunk (flexion, extension, rotation,
etc.) throughout the treatment session. Naturally, if there is an underlying segmental
hypermobility associated with deep muscle weakness, this will need to be treated by
an appropriate series of exercises.
In addition, undiagnosed inflammation is a common finding that may lead to
failure of treatment if not addressed correctly. The role of the nervous system at the
level of the joint has been identified as a significant factor in the modulation of the
inflammatory response (Schaible and Straub 2014). In particular, the intravertebral

disc has attracted a lot of attention, and many recent publications demonstrate the
ongoing interest in this area (Adams and Dolan 2012; Rajan et al. 2013; Ulrich et al.
2007).
The experience of one of the authors is that practically all dancers and circus
artists affected by common spinal dysfunctions (“low back pain”) are able to return
to performance with no functional limitations whatsoever, even after being off work
and in treatment for 1 or 2 years. Even taking into account the “well-worker effect,”
understanding the reasons for this may lead to a better understanding of how to treat
spinal dysfunctions in the general population.
12.8 The Healthy and Injured Artist as an Object of Study
As was discussed initially, arts and science are related to each other like both sides
of a piece of paper. In the same way, understanding injury and disease is another
way of understanding health and performance. Both go hand in hand. Developing
research projects to understand how high-level artists achieve the mastery of their art
form and how injury or disease interferes with artistic performance has the potential
to lead to better insights about how the human body works in general.
References
Adams MA, Dolan P. Intervertebral disc degeneration: evidence for two distinct phenotypes. J Anat.
2012;221(6):497–506.
Alltenmüller E., Bogousslavsky J, Boller F (eds): Neurological Disorders in Famous Artists. Front
Neurol Neurosci. Basel, Karger, 2005, vol 19, pp 1–10
Anderson CT, Castillo AB, Brugmann SA, Helms JA, Jacobs CR, Stearns T. Primary cilia: cellular
sensors for the skeleton. Anat Rec. 2008;291(9):1074–1078.
Beek PJ, Turvey MT. Temporal patterning in cascade juggling. J Exp Psych Hum Percep Perform.
1992;18(4):934–947.
Benjamin M, Moriggl B, Brenner E, Emery P, McGonagle D, Redman S. The “enthesis organ”
concept: why enthesopathies may not present as focal insertional disorders. Arth Rheum.
2004;50(10):3306–3313.
Bernstein N. The Co-ordination and Regulation of Movements. London: Pergamon Press Ltd.;
1967.
Bernstein N, Latash M, Turvey M. Dexterity and Its Development: Routledge; 1996.
Blankevoort L, Huiskes R, de Lange A. Recruitment of knee joint ligaments. J Biomech Eng.
1991;113(1):94–103.
Bouchard C. Genomic predictors of trainability. Exp Physiol. 2012;97(3):347–52.
Buhler JE, Graham R, Wright C. Juggling drops and descents. Amer Math Monthly.
1994;101:507–519.
Cibulka MT, Rose SJ, Delitto A, Sinacore DR. Hamstring muscle strain treated by mobilizing the
sacroiliac joint. Phys Ther 1986;66(8):1220–1223.
Conway BR, Livingstone MS. Perspectives on science and art. Cur Op Neurobiol.
2007;17(4):476–482.
Dehaene S. Consciousness and the Brain: Deciphering how the brain codes our thoughts. New
York. Viking Penguin; 2014
Dunning J, Rushton A. The effects of cervical high-velocity low-amplitude thrust manipu-
lation on resting electromyographic activity of the biceps brachii muscle. Manual Ther.
2009;14(5):508–513.
Emery CA. Risk factors for injury in child and adolescent sport: a systematic review of the
literature. Clin J Sports Med 2003;13(4):256–268.
Feldman AG. Threshold position control signifies a common spatial frame of reference for motor
action and kinesthesia. Brain Res Bull. 2008;75(5):497–499.
Feldman AG. Origin and advances of the equilibrium-point hypothesis. Adv Exp Med Biol.
2009;629:637–643.
Garcia de Yebenes J. Did Robert Schumann have dystonia? Movement Dis. 1995;10(4):413–417.
Gardner H. Frames of Mind: The Theory of Multiple Intelligences. New York: Basic Books; 1983.
Gehring D, Wissler S, Mornieux G, Gollhofer A. How to sprain your ankle-a biomechanical case
report of an inversion trauma. J Biomech. 2013a;46(1):175–178.
Gehring D, Faschian K, Lauber B, Lohrer H, Nauck T, Gollhofer A. Mechanical instability desta-
bilises the ankle joint directly in the ankle-sprain mechanism. Brit J Sports Med. 2013b(0): 1–7.
Gibson EJ. What does infant perception tell us about theories of perception? J Exp Psych Hum
Percep Perform. 1987;13(4):515–523.
Gibson JJ, Gibson EJ. Perceptual learning; differentiation or enrichment? Psych Rev.
1955;62(1):32–41.
Guardia D, Lafargue G, Thomas P, Dodin V, Cottencin O, Luyat M. Anticipation of body-scaled
action is modified in anorexia nervosa. Neuropsychologia. 2010;48(13):3961–3966.
Hamilton GM, Meeuwisse WH, Emery CA, Shrier I. Examining the effect of the injury definition
on risk factor analysis in circus artists. Scand J Med Sci Sports. 2012;22(3):330–334.
Hubbard-Turner T, Wikstrom EA, Guderian S, Turner MJ. Acute ankle sprain in a mouse model.
Med Sci Sports Exer. 2013;45(8):1623–1628.
Ingber DE. Tensegrity-based mechanosensing from macro to micro. Prog Biophys Mol Biol.
2008;97(2–3):163–179.
Keizer A, Smeets MA, Dijkerman HC, Uzunbajakau SA, van Elburg A, Postma A. Too fat to fit
through the door: first evidence for disturbed body-scaled action in anorexia nervosa during
locomotion. PloS One. 2013;8(5):e64602.
Kistemaker DA, Van Soest AJ, Wong JD, Kurtzer I, Gribble PL. Control of position and movement
is simplified by combined muscle spindle and Golgi tendon organ feedback. J Neurophysiol.
2013;109(4):1126–1139.
Koch C. Ubiquitous minds. Scientific Am Mind. 2014;Jan–Feb:26–29.
Koch C, Tsuchiya N. Attention and consciousness: related yet different. Trends Cog Sci.
2012;16(2):103–105.
Kursell J. Piano Mécanique and Piano Biologique: Nikolai Bernstein’s neurophysiological study
of piano touch. Configurations. 2006;14(3):245–273.
Latash ML. The bliss (not the problem) of motor abundance (not redundancy). Exp Brain Res.
2012;217(1):1–5.
Levin MF, Selles RW, Verheul MH, Meijer OG. Deficits in the coordination of agonist and
antagonist muscles in stroke patients: implications for normal motor control. Brain Res.
2000;853(2):352–369.
Liederbach M. Perspectives on dance science rehabilitation understanding whole body mechanics
and four key principles of motor control as a basis for healthy movement. J Dance Med Sci.
2010;14(3):114–124.
Lorblanchet M. The Origin of Art. Diogenes. 2007;54(May):98–109.
Mattos D, Kuhl J, Scholz JP, Latash ML. Motor equivalence (ME) during reaching: is ME
observable at the muscle level? Motor Control. 2013;17(2):145–175.
Moayedi M, Davis KD. Theories of pain: from specificity to gate control. J Neurophysiol.
2013;109(1):5–12.
Myers KA, Rattner JB, Shrive NG, Hart DA. Hydrostatic pressure sensation in cells: integration
into the tensegrity model. Biochem Cell Biol. 2007;85(5):543–551.
Ojofeitimi S, Bronner S. Injuries in a modern dance company effect of comprehensive management
on injury incidence and cost. J Dance Med Sci. 2011;15(3):116–122.
Ostry DJ, Feldman AG. A critical evaluation of the force control hypothesis in motor control. Exp
Brain Res. 2003;153(3):275–288.
Pinna B. The organization of shape and color in vision and art. Front Hum Neurosci. 2011;5:104.
Polster B. The Mathematics of Juggling. New York: Springer Verlag; 2003. 226 p.
Pongratz G, Straub RH. Role of peripheral nerve fibres in acute and chronic inflammation in
arthritis. Nat Rev Rheum. 2013;9(2):117–126.
Poore GV. Clinical lecture on certain conditions of the hand and arm which interfere with the
performance of professional acts, especially piano-playing. Brit Med J. 1887;1(1365):441–444.
Pross A. What is Life? How Chemistry Becomes Biology. Oxford. Oxford University Press. 2012.
Rajan NE, Bloom O, Maidhof R, Stetson N, Sherry B, Levine M, et al. Toll-Like Receptor 4 (TLR4)
expression and stimulation in a model of intervertebral disc inflammation and degeneration.
Spine. 2013;38(16):1343–1351.
Ramazzini. Diseases of Tradesmen. H G, editor. New York: Medical Lay Press; 1713 (1933).
Reppa I, Schmidt WC, Ward R. Informational affordances: evidence of acquired perception-action
sequences for information extraction. Psychol Bull Rev. 2012;19(3):418–428.
Robert T, Zatsiorsky VM, Latash ML. Multi-muscle synergies in an unusual postural task: quick
shear force production. Exp Brain Res. 2008;187(2):237–253.
Roussouly P, Pinheiro-Franco JL. Biomechanical analysis of the spino-pelvic organization and
adaptation in pathology. Eur Spine J 2011;20 Suppl 5:609–618.
Sataloff RT, Brandfonbrener AG, Lederman RJ, editors. Performing Arts Medicine. 3rd edition.
Narberth: Science & Medicine, Inc; 2010.
Schaible HG, Straub RH. Function of the sympathetic supply in acute and chronic experimental
joint inflammation. Autonom Neurosci: Basic & Clinical. 2014;182:55–64.
Schneider TD. Claude Shannon: biologist. The founder of information theory used biology to
formulate the channel capacity. IEEE Engin Med Biol. 2006;25(1):30–33.
Shrier I, Halle M. Psychological predictors of injuries in circus artists: an exploratory study. Brit
J Sports Med. 2011;45(5):433–436.
Shrier I, Meeuwisse WH, Matheson GO, Wingfield K, Steele RJ, Prince F, et al. Injury patterns
and injury rates in the circus arts: an analysis of 5 years of data from Cirque du Soleil. Am J
Sports Med. 2009;37(6):1143–1149.
Silva PL, Turvey MT. The role of haptic information in shaping coordination dynamics: inertial
frame of reference hypothesis. Hum Movement Sci. 2012;31(5):1014–1036.
Suter E, McMorland G, Herzog W, Bray R. Decrease in quadriceps inhibition after sacroiliac joint
manipulation in patients with anterior knee pain. J Manip Physiol Ther. 1999;22(3):149–153.
Swanson RL, 2nd. Biotensegrity: a unifying theory of biological architecture with applications to
osteopathic practice, education, and research-a review and analysis. J Am Osteopath Assoc.
2013;113(1):34–52.
Treasure J, Claudino AM, Zucker N. Eating disorders. Lancet. 2010;375(9714):583–593.
Tobin JL, Beales PL. The nonmotile ciliopathies. Genetics Med. 2009;11(6):386–402.
Tononi G. Consciousness as integrated information: a provisional manifesto. Biol Bull.
2008;215(3):216–242.
Trilk JL, Phillips EM. Incorporating ‘Exercise is Medicine’ into the University of South Car-
olina School of Medicine Greenville and Greenville Health System. Brit J Sports Med.
2014;48(3):165–167.
Turvey MT. Action and perception at the level of synergies. Human movement science.
2007;26(4):657–697.
Turvey MT, Carello C. Obtaining information by dynamic (effortful) touching. Phil Trans Roy Soc
London Series B, Biol Sci. 2011;366(1581):3123–132.
Turvey MT, Shockley K, Carello C. Affordance, proper function, and the physical basis of
perceived heaviness. Cognition. 1999;73(2):B17–26.
Ulrich JA, Liebenberg EC, Thuillier DU, Lotz JC. ISSLS prize winner: repeated disc injury causes
persistent inflammation. Spine. 2007;32(25):2812–2819.
van der Wees PJ, Lenssen AF, Hendriks EJ, Stomp DJ, Dekker J, de Bie RA. Effectiveness of
exercise therapy and manual mobilisation in ankle sprain and functional instability: a systematic
review. Aus J Physiother. 2006;52(1):27–37.
Verri A Poggio T. Motion Field and Optical Flow: Qualitative Properties. IEEE Trans Pattern Anal
Machine Intel. 1989;11(5):490–498.
Voet D, Voet, JG, Pratt, CW. Fundamentals of biochemisty: life at the molecular level. Hoboken:
John Wiley & Sons, Inc. 2012.
Vora AJ, Doerr KD, Wolfer LR. Functional anatomy and pathophysiology of axial low back pain:
disc, posterior elements, sacroiliac joint, and associated pain generators. Phys Med Rehabil
Clin North Amer. 2010;21(4):679–709.
Wanke EM, McCormack M, Koch F, Wanke A, Groneberg DA. Acute injuries in student circus
artists with regard to gender specific differences. Asian J Sports Med. 2012;3(3):153–160.
Zatsiosrky VM. Kinematics of Human Motion. Champaign: Human Kinetics; 1998. 418 p.
Zatsiorsky VM, Li ZM, Latash ML. Enslaving effects in multi-finger force production. Exp Brain
Res. 2000;131(2):187–195.
Chapter 13
Adaptations to Neck/Shoulder
Fatigue and Injuries
Julie N. Côté
13.1 Neck/Shoulder Musculoskeletal Disorders:

A Considerable Societal Problem
In the past three decades, the incidence of musculoskeletal disorders (MSDs) has
grown to affect a large portion of the population, accounting for considerable amounts
of compensation claims, lost time, and health-care costs in most industrialized coun-
tries (Feuerstein et al. 2003). In many countries, MSDs consistently cause more
work absenteeism or disability than any other group of diseases (Badley et al. 1994;
Feeney et al. 1998; Leijon et al. 1998; Woolf et al. 2010), even though the pro-
portion of the workforce working in sectors characterized by heavy physical work
has steadily declined (Madeleine 2010). Although there can be various causes for
MSDs, repetitive work has often been cited as one of the leading causes for reporting
symptoms. For instance, it has been previously reported that 55 % of MSD attributed
to repetitive movements were thought to be caused by work-related activities (Tjep-
kema 2003). Among the European workforce, it has been shown that the two highest
exposures to physical risks are a painful/tiring position and repetitive movement
(Fourth European Working Conditions Survey 2007). Indeed, these statistics show
that the proportion of workers exposed to repetitive hand or arm movement for at
least one-quarter of the working day has increased over the past 5 years, reaching
62 % of the working population. Further, 45 % of the population indicated experienc-
ing painful, tiring positions for at least one-quarter of the working day. While many
studies have focused on MSDs of the low-back or distal upper limb, MSDs affecting
the neck and/or shoulder region have substantial socioeconomic consequences in
J. N. Côté ()
Department of Kinesiology and Physical Education, McGill University,
475 Pine Avenue West, Montreal, QC H2W 1S4, Canada
Tel.: (514) 398-4184

and Biology 826, DOI 10.1007/978-1-4939-1338-1_13
206 J. N. Côté
many countries (Silverstein et al. 1998; Bongers 2001; Hamberg-van Reenen et al.
2007). Although distinctive risk factors have been associated with either neck or
shoulder MSDs, most studies agree that complaints often affect the neck/shoulder
area as a unit (Visser and van Dieen 2006), especially in the case of low-load, work-
related symptoms, so that the etiology of neck/shoulder MSDs is thought to revolve
around one common mechanism that is not yet fully understood (Andersen et al.
2003). Neck/shoulder MSDs have been linked to the development of chronic pain,
dysfunction, and a high prevalence of secondary health problems (Nordlund and Ek-
berg 2004). Studies have suggested that neck/shoulder MSD can be caused by highly
repetitive work, forceful exertions, highly prolonged static loads, static or extreme
postures, and combinations of these factors (Larsson et al. 2007). One of the most
important occupational risk factors for neck/shoulder pain is prolonged repetitive
upper limb movement. A study showed that working with neck and/or body bent
forward, arms above shoulders, precision work tasks and female gender were pre-
dictors of neck/shoulder MSDs (Wahlstedt et al. 2010). Despite the frequency, high
cost, and range of initiatives implemented to prevent neck/shoulder MSDs, little is
known about which interventions are the most effective, so that neck/shoulder MSDs
remain a considerable problem (Buckle and Devereux 2002).
13.2 Towards a Model of Neck/Shoulder MSD Production
It is generally agreed that the mechanisms of MSDs are multifactorial, as suggested

by the International Classification of Functioning (ICF) model of the World Health
Organization (www.who.int/classifications/icf/), and few theoretical models succeed
in conceptualizing the pathways by which injury can occur. With the goal of focus-
ing more specifically on factors that can lead to the development of work-related
neck/shoulder MSDs, we have developed a simple conceptual model (Fig. 13.1). In
this model, personal factors combine with work-related risk factors to determine the
level of exposure to damage of some selective neck/shoulder structures, which can
result in injury if a balance between tissue damage and repair mechanisms is not met.
Whereas force is a known MSD risk factor that can in itself have a direct effect on
injury risk (Karsh 2006), in low-force tasks, the balance of personal and work-related
factors is highly sensitive to the accumulation of damage caused by sustained efforts,
which suggests a link between fatigue and the development of muscle injury.
13.3 Fatigue as a Precursor to MSD in the Injury Trajectory?
One of the most often studied risk factors for MSD, especially in association with
repetitive work, is fatigue. Enoka and Stuart (1992) define fatigue as a decrease in
functional capacity associated with an increase in the perceived difficulty in main-
taining force production. In low-force tasks, fatigue is an ongoing process, which
may not necessarily result in a decline in target submaximal force, but is typically
13 Adaptations to Neck/Shoulder Fatigue and Injuries 207
Fig. 13.1 General theoretical model of potential physical risk factors for neck/shoulder injury
production. Injury to musculoskeletal structures can be caused by a combination of factors in the
first row, affecting the person or his/her exposure to a risk of injury. Ultimately, injury production will
depend on whether the balance between selective damage (caused by person- or exposure-related
factors) and selective tissue repair processes can be maintained
manifested as an increase in perceived effort (Jones and Hunter 1983) and a reduction
in the maximal force generating capacity of the neuromuscular system that occurs
during sustained activity (Bigland-Ritchie et al. 1983; Vollestad 1997). Previous
research shows increased amplitude of the electromyographic (EMG) signal (e.g.,
root mean square, RMS) and decreased mean power frequency (MPF) as fatigue
develops during submaximal contractions (Vollestad 1997). Changes encumbered
by the motor apparatus as fatigue develops may be attributable to impairment at any
point in central or peripheral neuromuscular pathways involved in muscle contrac-
tion (Stokes et al. 1989). For instance, fatigue may be caused by changes at the
central level, due to impaired motivation or failure of motoneuronal drive, or by
peripheral changes, due to impairment at or beyond the motor nerve, including the
neuromuscular junction and the contractile apparatus itself (Edwards 1983). When
the task involves a submaximal contraction, an individual is able to increase the cen-
tral motor command to counteract the reduction in force resulting from peripheral
fatigue mechanisms (Maton and Gamet 1989), supporting the idea that fatigue is a
mechanism that develops over time and that involves the coordinated action of many
elements of the neuromuscular pathway.
Figure 13.2 shows a schematic representation of the temporal progression that
could lead a worker exposed to a new work task to eventually develop MSD and
associated disability. When first exposed to a new task, a worker or athlete would
initially improve as a result of repetition and learning to attain a higher level of perfor-
mance. If the worker continues to work beyond this optimal level and/or experiences
insufficient recovery time, somewhat manageable symptoms such as fatigue and
discomfort may lead to the development of less manageable and more debilitating
and threatening ones such as delayed-onset muscle soreness (DOMS) and lesion-
causing microtrauma, which can in turn cause permanent damage. Indeed, several
208 J. N. Côté
Fig. 13.2 Temporal sequence of events that can lead a worker to develop a musculoskeletal disorder
and absence from work. The time scale can represent weeks to months to years
authors have suggested that muscle fatigue plays an important triggering role in
neck/shoulder MSD (Rempel et al. 1992; Sundelin and Hagberg 1992; Hermans and
Spaepen 1995; Takala 2002). The exact pathophysiological mechanisms by which
fatigue can trigger pain and injury has been thought to involve the sensitization of
nociceptors caused by a release of mediators from the free nerve endings (FRNs),
which in turn may be due to peripheral mechanisms such as lowering of the firing
threshold of the nociceptors (Madeleine 2010). However, the way in which sustained
or recurrent fatigue, discomfort, and pain episodes may lead to chronic symptoms
and impairment remains poorly understood. It is well known in the sports training
world that a certain level of fatigue is necessary to trigger a training and tissue mod-
eling effect, so that the difference between “effective” and “threatening” fatigue is a
puzzle that has posed a challenge to many coaches and therapists. Moreover, what
distinguishes the trajectory of one worker over that of another, impacting on the
speed of this downward slope displayed in Fig. 13.2, remains in large part a mystery
and likely depends on the balance between damage and repair mechanisms, which
in itself may depend on many work- and nonwork-related factors.
13.4 Low-Force Efforts and MSD: The Cinderella Hypothesis
Even though it is conceivable that injury can be avoided with low-force efforts in-
terspersed with sufficient rest, it is difficult to predict the exact work–rest regimen
that can effectively achieve this goal. Even efforts as low as 10 % maximum effort
have documented endurance time limits, in the order of 1 h (Fallentin et al. 1985).
In such cases, the primary physiological mechanism leading to fatigue is largely
unknown (McLean and Goudy 2004). It has been argued that since there is little
blood flow occlusion at this low-force level (Parker et al. 1984), muscle fibers in-
volved in this action should theoretically be capable of sustaining this force nearly
indefinitely. However, evidence suggests that even such low-force activities, when
sustained and/or repeated, may contribute to muscle injury (Winkel and Westgaard
1992; Kilbom 1994; van der Windt et al. 2000; Sluiter et al. 2001). To resolve this
issue, over the past decades, a number of theories mention unfavorable motor control
strategies as a risk factor in the etiology of MSD (van Dieen et al. 1993; Sjogaard et al.
1986; Mathiassen and Aminoff 1997), and recent prospective studies have provided
evidence for the link between some motor control patterns and the development of
MSDs (Veiersted 1994; Madeleine et al. 2003).
The most recognized hypothesis for neck/shoulder MSD associated with low-
force efforts is the Cinderella hypothesis (Hagg 1991). This hypothesis points to the
behavior of the low-threshold motor units (MUs) and their associated type 1 fibers as
a central aspect of the mechanism of injury associated with sustained, low-intensity
efforts (Visser and van Dieen 2006). Cinderella fibers are the type 1 (slow, fatigue-
resistant) muscle fibers connected to the lowest recruitment threshold MUs that,
according to the size principle of MU recruitment (Henneman 1957; Bawa and Mur-
naghan 2009), are the first ones to be recruited and remain activated for the longest
periods of time during sustained low-force efforts. Muscle injury would result from
overload, accumulation of calcium ions preventing muscle relaxation, and structural
damage to these fibers (Lexell 1993; Hagg and Astrom 1997), implying insufficient
regenerative processes at the injury site (Hagg and Suurkula 1991; Visser and van
Dieen 2006; Voerman et al. 2007). These mechanisms would then trigger secondary
ones, including stimulation of nociceptors, triggering of sympathetic nervous system
responses and alterations in blood flow (Larsson et al. 2004; Shiro et al. 2012). Fi-
nally, sensitization would trigger injury chronicity. Some biopsy studies lend support
to the Cinderella hypothesis (Lindman et al. 1990; Hagg 1991, 2000) and as such, it
is one of the few MSD injury hypotheses linking experimental evidence with tissue
damage mechanisms. Another prediction of the Cinderella hypothesis is that the in-
jury trajectory is dependent on the duration of use but could also be accelerated by
deficits in normal MU rotation and load-sharing mechanisms (Visser and van Dieen
2006). Indeed, it has been shown that some derecruitment of low-threshold MUs
can take place during low-intensity sustained muscle contractions (Westgaard and
DeLuca 1999). As such, the neck/shoulder muscles, particularly the trapezius, often
display continuous activity of some MUs over a wide range of tasks (Thorn et al.
2002; Zennaro et al. 2003), but some studies have shown that some subjects display
less derecruitment than others (Westgaard and DeLuca 1999). Moreover, these par-
ticular patterns do not seem to be related with any known specific injury risk factor,
suggesting that their presence could predispose certain people to develop overload
and injury in an unpredictable way.
The Cinderella hypothesis has mostly been used as a framework for studies of
low-load repetitive upper-limb tasks targeting output from the neck and upper limb
musculature. For instance, studies of trapezius muscles of cleaners showed experi-
mental evidence for the predictions of the Cinderella hypothesis with the presence of
210 J. N. Côté
presumably overloaded type 1 necrotic, “ragged red” fibers in workers with symp-
toms (Sjogaard and Sogaard 1998). Two notable characteristics of this work situation
are (1) the involvement of the complex neck/shoulder musculature, encompassing
many muscles that can theoretically produce force in many directions, and motion of
considerable amplitude, (2) the prolonged, repetitive nature of the associated upper
limb task. These two characteristics can be generalized to two aspects of motor be-
havior that can be linked to the risk of tissue overload and injury: motor abundance
and motor variability. We illustrate these two concepts within the Cinderella hypoth-
esis by suggesting that motor abundance, through the potential contribution of many
muscles to one given action, can be used as a way to provide help to vulnerable
Cinderella fibers, whereas motor variability, through variation of one movement to
the next, can be used to provide Cinderella fibers some breaks.
13.5 Give Cinderella Some Help: Abundance
13.5.1 The Concept of Abundance
One movement can be accomplished in many different ways. At the mechanical level,
this concept can be modeled as a system taking advantage of its redundant degrees
of freedom (Bernstein 1967). At the control level, redundancy can be considered as
a consequence of having many muscles, having many MUs that can be triggered
separately to produce a motor action. How does one select a “motor solution” in a
given condition? In recent years, this apparent “problem” has been reformulated by
introducing the concept of abundance (Gelfand and Latash 1998). In this view, the
redundancy of the human system does not pose a problem but rather offers a theo-
retically rich set of solutions that can be used when accomplishing a motor action
(Latash 2000). The concept of abundance is particularly attractive when applied to
tasks related to neck/shoulder MSD, since as stated previously, the neck/shoulder
musculoskeletal system has a high degree of structural complexity that can easily be
underestimated by simple mechanical models (Hill et al. 2008). Thus, it is conceiv-
able that in tasks involving a highly abundant system, even more so for tasks that
allow for the contributions of other additional muscles and joints beyond those of the
neck/shoulder region, taking advantage of this motor abundance could have impli-
cations on overload, and therefore injury risk, on the neck/shoulder musculoskeletal
structures. More specifically in association with the Cinderella hypothesis, having
access to the contributions of many joints and muscles to produce a task could help
reduce the burden on a particularly vulnerable set of neck/shoulder Cinderella muscle
fibers.
13.5.2 Abundance at the Level of the Neck/Shoulder Musculature
The most widely documented illustration of the concept of abundance has been shown
to occur at the individual muscle level, with selective recruitment, derecruitment and
substitution of MUs during the prolonged performance of a task.MU substitution
typically describes a mechanism whereby higher-threshold MUs are recruited to re-
place lower-threshold fatigued units that have stopped firing (Westgaard and DeLuca
1999). Using this definition, MU substitution would be consistent with Henneman’s
size principle (Henneman 1957); however, MU substitution is often used synony-
mously with MU rotation (Sale 1987), which describes cyclical activity between
MUs, such that substitution of one MU for another would be followed by back sub-
stitution of the original unit. While MU substitution is well described, MU rotation is
controversial in that it suggests that a violation of the size principle of MU activation
may occur. Moreover, MU substitution was previously believed to be more plausi-
ble in muscles that are anatomically subdivided (e.g., quadriceps femoris); however,
recent evidence for the ability to control subdivisions of individual muscles can be
used to argue against this suggestion and as such, MU substitution has been shown
to occur within the upper trapezius muscle (Mathiassen and Winkel 1990; Westgaard
and DeLuca 1999; Zennaro et al. 2003). Westgaard and DeLuca (1999) studied the
activity of the human trapezius muscle during contractions lasting 10 min and found
that low-threshold MUs showed periods of inactivity during which they were sub-
stituted by MUs with higher recruitment thresholds. It has been suggested that a
purpose of MU rotation could be to maintain a desired level of muscle output, such
as force (Sjogaard et al. 1986).
Few have attempted to relate phenomena of MU rotation with other pathological
mechanisms related to injury. Sjogaard et al. (1986) investigated myoelectric sig-
nal activity during isometric knee extension performed continuously for 1 h at 5 %
of maximal voluntary effort. It was found that blood flow was distributed hetero-
geneously throughout the muscle during the entire duration of the task, and EMG
recordings suggested that this heterogeneity was related to the alternating recruitment
of muscle fibers. However, currently there is little evidence related to the mecha-
nism through which MU substitution occurs. In most studies, MU substitution was
not seen during the first few minutes of a contraction, even among units that later
displayed this phenomenon. It may be that a target force can be maintained while
particular MUs (or even muscle fiber bundles) experience periods of inactivity to
allow for the replenishment of metabolic needs (Sjogaard et al. 1986; Westgaard
and DeLuca 1999). Indeed, the concept of MU substitution has long been postu-
lated to offset the effects of fatigue (Person 1974), and researchers have reported the
observation of substitution activity of MUs within a muscle that was performing a
sustained, low-level static contraction (Sale 1987; Westgaard and DeLuca 1999), and
have speculated that this is a natural strategy to resist fatigue. Van Dieen et al. (1993)
even noted that muscles of subjects who demonstrated MU rotation had longer en-
durance times. However, in some studies, substitution and rotation were observed
in muscles of some, but not all subjects (McLean and Goudy 2004), suggesting an
212 J. N. Côté
individual predisposition for this mechanism that could be dependent on the ability
to detect a need to trigger tissue regeneration mechanisms. In turn, this mechanism
could be related to one’s predisposition to develop MSD (Veiersted and Westgaard
1993). This is supported in studies from Thorn et al. (2002) and Zennaro et al. (2003)
who used long-duration MU decomposition techniques to show that in a minority of
subjects, at least one single MU remained continuously active over the duration of
the contraction.
Other multisite EMG spatial distribution techniques have recently been devel-
oped to illustrate within-muscle changes in activation patterns with fatigue and with
experimental pain. Previous studies have documented relative changes in the EMG
amplitudes of different sections of the trapezius muscle after sustained isometric
contractions and in response to experimental pain (Madeleine et al. 2006; Falla et al.
2007a). A recently developed technique quantifies the amount of shared information
contained between two EMG time series, accounting for both linear and nonlinear
signal characteristics (Madeleine et al. 2011), with the assumption that high levels
of shared information illustrate functional connectivity between both time series.
Indeed, within-trapezius functional connectivity has been observed to increase with
DOMS (Madeleine et al. 2011). However, in another study of asymptomatic men
and women accomplishing a prolonged repetitive upper-limb task, no effect on task
time was found (which could be explained by a lower level of fatigue induced by
the task). Moreover, women displayed a higher level of functional connectivity than
men in two of the three muscle subdivisions studied (Johansen et al. 2013), confirm-
ing previous studies showing gender differences in trapezius muscle behavior (Falla
et al. 2008).
13.5.3 Abundance and Multimuscle Patterns During Tasks

Targeting the Neck/Shoulder Musculature
To this day, the description of patterns of multi-muscle behavior poses a significant

methodological challenge, even despite recent advances in instrumentation and data
processing capabilities (Hug 2011). This could explain why comparatively fewer
studies have described patterns of groups of agonist, antagonist, and synergist mus-
cles, even less so across many joints, during the production of complex upper limb
movements. For one, cocontraction of single-joint agonist–antagonist muscle pairs
has been well documented in situations of fatigue and pain (Psek and Cafarelli 1993).
These patterns have been explained as an attempt to stabilize and protect vulnera-
ble body areas from further damage, in line with the pain adaptation model (Lund
et al. 1991). Moreover, findings from recent studies of low-back pain show that
elevated coactivation patterns displayed by initially asymptomatic people were asso-
ciated with the development of low-back pain, suggesting a causal injury mechanism
(Nelson-Wong and Callaghan 2010). A few studies have also documented modifi-
cations of inter-muscle coordination among agonist and synergistic muscle groups
of individual joints (Danion et al. 2000, 2001) and reduced trans-joint inhibition of
synergistic muscles (Aymard et al. 1995) in prolonged tasks. In addition, Falla and
colleagues showed that healthy subjects with experimental pain, and groups with
whiplash and idiopathic neck pain showed increased use of accessory neck/shoulder
muscles when performing repetitive upper-limb tasks (Falla et al. 2004, 2007b).
We recently showed higher activity in stabilizers of the shoulder girdle in people
with neck/shoulder pain performing a repetitive upper-limb task (Lomond and Côté
2011). We also applied the mutual information technique (Madeleine et al. 2011) to
describe functional connectivity in neck/shoulder synergistic muscle pairs, to show
decreased connectivity among the neck/shoulder group as fatigue developed during
the repetitive task, and that this was more noticeable in men (Fedorowich et al. 2013).
Even fewer studies have extended their analyses beyond the level of individual
joints. Some studies conducted in our lab showed that fatigued, noninjured people
seemingly take advantage of multijoint abundance during the performance of whole-
body tasks. (Côté et al. 2002, 2008; Fuller et al. 2009). In studies in which fatigue
was induced by repetitive hammering, sawing, and pointing, subjects showed re-
duced hammering and sawing force output and movement amplitude, and increased
activity of the fatigued upper limb muscles, in parallel with increased multidirec-
tional movement amplitude of the trunk, whole-body center of mass (CoM) and
center of pressure (CoP), and trunk EMG activity. Together, these changes seemed
to target an increased contribution of trunk to endpoint motion; indeed, an increase
of only a few degrees of trunk flexion, or even rotation could increase the ampli-
tude of the upper limb endpoint by several centimeters. Trunk movement could be
used to compensate for changes encumbered by fatigue at the neck/shoulder region.
Moreover, most of the observed whole-body changes began to occur about halfway
to exhaustion, suggesting an ongoing fatigue adaptation strategy (Fuller et al. 2011).
Despite these many changes developing in parallel with fatigue, here again, as with
overall muscle force despite MU rotation shown previously to occur during knee
extensions (Sjogaard et al. 1986), some general task characteristics such as endpoint
motion amplitude and timing, were kept constant, suggesting adaptations tailored to
the perceived task goals. As summarized in Côté and Hoeger Bement (2010), studies
of three low-force tasks performed to exhaustion provides evidence for (1) fatigue in
areas mainly involved in each task (e.g., increased muscle activity, decreased motion
amplitude at the elbow and shoulder), (2) increased involvement and contributions
toward the overall task from adjacent body segments such as the trunk (e.g., in-
creased muscle activity, increased motion amplitude), and (3) maintenance of the
consistency of some overall task characteristics (e.g., endpoint movement amplitude
and movement speed). Taken together, these findings not only provide evidence that
changes in whole-body movement patterns occurred with low-force muscle fatigue
but also suggest that these adaptations involve the use of the overall abundance of the
musculoskeletal system even in areas remote to the one directly fatigued. The ob-
servation that global task measures remain largely unaltered after low-force fatigue
despite the many kinematic changes occurring across the body, supports the hypoth-
esis of a task-specific global adaptation strategy, of which parts can be modulated,
but as a whole, remains largely unaffected by low-force fatigue. This was further
challenged and confirmed in recent studies where we showed that most of the global
214 J. N. Côté
task-specific adaptations that can compensate for localized fatigue also occur in the
presence of a second perturbing factor added on top of fatigue (i.e., sudden support
surface translations (Fuller et al. 2013), wearing an added external weight (Cantu
et al. accepted). It should be noted that the task-specific characteristic that is thought
to be used to guide the choice of motor solutions can also be related to the concept
of synergies, defined as neural organizations that ensure task-specific covariation
of elemental variables providing desired stability properties of important output, or
performance variables (Latash 2012; see also Latash, this volume). According to this
concept, synergies would serve to simplify the organization and selection of a given
motor solution that would satisfy, with more or less room for variation from one trial
to another, what are perceived to be important task goals.
In situations of experimental or acute pain, previous studies have documented
similarities in whole-body adaptation patterns displayed by healthy fatigued people
(Madeleine et al. 1999; Latash and Anson 2006). In comparison, the range of whole-
body adaptations to repetitive movements seems smaller in people with MSD. We
have previously shown that there was a significant correlation between scores of the
shoulder pain and disability index, and reductions in kinematic (motion, speed, ac-
celeration) amplitudes at the upper limb joints during hammering (Côté et al. 2005).
Participants in the symptomatic group also displayed reduced trunk as well as overall
endpoint motion amplitude. To challenge this finding, we performed a study where
the task required subjects to maintain a specific endpoint motion amplitude during
repetitive reaching between two targets, and found that in this case, patients again
show less upper limb joint motion, with significant negative correlations between
initial pain level and shoulder range of motion (Lomond and Côté 2010). However,
in this case, participants developed a strategy of increasing whole-body CoM motion
amplitude, suggesting a prioritization of compensatory strategies, with perhaps the
most cost-effective ones prioritized over the least cost-effective ones and/or the most
painful ones (Lomond and Côté 2011; see also Levin, this volume). Figure 13.3
shows previously unpublished data on the same repetitive reaching task, but this
time accomplished in the seated position, further reducing the abundance of move-
ment solutions to accomplish the repetitive task under the same spatial (i.e., moving
the finger forward between the two targets adjusted for height and distance, while
maintaining the elbow above a barrier adjusted to force the entire arm to move in a
horizontal plane) and temporal (moving at one movement per second) constraints.
The data shows that in the seated position, the symptomatic group took less advan-
tage of trunk (in this case, head, arms, and trunk CoM) compensatory strategies,
possibly to avoid destabilizing the upper body or placing additional load on the ex-
tensors of the upper trunk. This reinforces the notion that (in this case, mechanically)
more advantageous solutions are prioritized, especially when dealing with repetitive
motions in the presence of pain.
Fig. 13.3 Head, arms and trunk center of mass trajectory (mm) in time (s) during repetitive
forward–backward reaching at shoulder height, in the sitting position (representative subjects).
CTRL asymptomatic, PAIN chronic neck–shoulder pain, NF during the first minute, FT during the
last minute right before exhaustion
13.6 Give Cinderella a Break: Variability
13.6.1 Defining the Concept and Scope of Variability
While abundance, as defined here, addresses the various ways in which the system
can produce a given sensorimotor action, the concept of variability addresses the re-
peatability of these patterns from one action to the next. Variability, the variation of
behavioral outcomes over repetitions or time, is an inherent feature that plays a role
in the central organization of voluntary movements (Latash et al. 2002). The amount
of variability in any characteristic of posture or movement can be estimated using
simple statistics such as standard deviation (SD), coefficient of variation (CV), or
interquartile range (IQR) (Srinivasan and Mathiassen 2012), each measure providing
a slightly different illustration of variability (e.g., CV is impacted by the mean mea-
surement taken over repeated trials, whereas SD is not). In motor learning research,
variability has been considered not just as undesirable noise (Newell and Slifkin
1998) but also as an indicator of motor development and skill acquisition (Bartlett
et al. 2007). In this paradigm, skill acquisition does not merely imply developing an
adequate movement pattern but also developing a certain optimal variability of that
pattern, that is neither too low (stereotypical) nor too high (unstable) (Stergiou et al.
2006). Motor variability, in turn, ensures that new motor solutions can be explored in
216 J. N. Côté
response to changes in the external environment or the internal physiology (Dingwell

et al. 2001; Riley and Turvey 2002). One important issue to keep in mind when nav-
igating through the literature is that variability has been studied as various levels of
the motor output (Newell and Corcos 1993; Stergiou 2004; Davids et al. 2006). For
instance, variability in performance outcome measures may have important implica-
tions on the quality of performance and in fact may not easily be modified in real-life
situations such as during work at an imposed rhythm. Variability in joint or segment
kinematic characteristics displayed while performing a task may be more easily ma-
nipulated without affecting performance outcome; however, these adjustments may
be limited by the inherent mechanical constraints associated with the architecture
of the human body (Newell and Slifkin 1998). Variability in activity patterns and
in more elemental control variables may also operate. Ultimately, the system must
account for the possibility that some level of variability exists at all of these levels
to produce adequate actions. A large body of literature related to the uncontrolled
manifold (UCM) hypothesis (Scholz and Schöner 1999; Latash and Scholz 2007;
Martin et al. 2009) has focused on the structure of variability, distinguishing between
variability that does not affect performance quality (“good variability”) and variabil-
ity that (negatively) does (“bad” variability). Recent contributions to this literature
notably further highlight good variability as beneficial towards overall performance
stability (Rosenblatt et al. 2013). While the UCM literature undoubtedly provides
an important contribution to the literature on variability in motor control, it is exten-
sively reviewed elsewhere (see Latash, this volume) and is thus not expanded upon
further in this chapter, which will focus on the amount, and not the structure, of
motor variability.
13.6.2 Variability as a Strategy to Find Solutions

to an Acute Problem
Repetitive tasks imply the potential development of fatigue, and fatigue has been
shown to have an effect on movement variability. For instance, Cignetti et al. (2009)
examined variability of gait during cross-country skiing and found arm and leg move-
ments to become more variable with fatigue. However, other studies suggest that
motor variability may have a positive effect on mitigating fatigue development. In
an examination of a target-tracking task, Selen et al. (2007) showed that despite
increased kinematic variability with fatigue, overall task performance (accuracy)
was maintained. In a repetitive reaching study (Fuller et al. 2011), movement-to-
movement variability of the shoulder and elbow joints and the CoM average position
increased with fatigue, but these were not associated with any deterioration in perfor-
mance. What is more, some studies have associated variability in movement patterns
with high endurance. Studies of repeated knee extensions (Skurvydas et al. 2010)
and back muscle exertions (van Dieen et al. 1993) show associations between joint
and muscle pattern variability in the nonfatigued state with high endurance during
Fig. 13.4 Movement-to-movement variability (coefficient of variation, CV ) in trapezius EMG RMS

between consecutive forward-reach movements at shoulder height, accomplished in the seated
position, in asymptomatic participants. Each symbol represents the CV of one participant, in each
trapezius subdivision (UT upper trapezius, MT middle trapezius, LT lower trapezius), at each of
the nonfatigued (NF) and fatigue-terminal (FT ) states
the performance of tasks engaging the respective muscles and joints. Similar find-
ings were shown recently by our group, with initially higher variability in trapezius
muscular patterns showing a positive association with longer endurance time dur-
ing the repetitive reaching task, but with this association being significant only in
our female subsample (Fedorowich et al. 2013). This study was, to our knowledge,
one of very few documenting movement-to-movement variability of electromyo-
graphic (EMG) characteristics of muscles of the neck/shoulder region, including
muscle (trapezius) subsections. These findings were confirmed in a recent (unpub-
lished) study (Fig. 13.4) where subjects performed the same task in a seated position,
further targeting the load during the repetitive upper-limb reaching task on this mus-
culature. Here again, we observed significant increases in reach-to-reach variability
in the fatigued state in all three subsections of the trapezius muscle despite the fact
that frequency and trajectory of reaching were still preserved.
The preservation of overall performance despite increased variability with fatigue
implies that some level of proficiency may be required in order to develop such
compensatory strategies (Bosch et al. 2011). This is supported by studies that have
shown that only expert table tennis players were able to increase variability and still
maintain a high level of performance (Aune et al. 2008), and other studies that have
shown that increased variability with fatigue negatively affected the performance
218 J. N. Côté
of complex upper-limb tasks (Huysmans et al. 2008; Missenard et al. 2008). As

a whole however, most studies agree that during repetitive, fatiguing tasks, where
some level of performance can theoretically be maintained over a certain number of
repetitions, variability in some movement characteristics can be useful to manage
fatigue. Yet, we still have little understanding of the mechanisms underlying these
observations. It has been suggested that whole-body compensatory strategies could
be triggered by the detection of time-related changes at the muscle level, such as in-
creased muscle activity, decreased proprioception, or accumulation of metabolites.
Since these localized changes have been shown to occur well before exhaustion in
low-force tasks (Fuller et al. 2011), and consistently with the notion that variability
could be used as a compensatory strategy, we could expect to observe fluctuations
in variability characteristics over the duration of a task performed to exhaustion,
although this has not been well-documented to date. Nevertheless, this would have
the advantage of providing fatiguing muscles with windows of opportunity to rest,
or at least with some degree of load variations (Visser and van Dieen 2006), which
in itself has been shown to promote derecruitment of MUs (Westad et al. 2003),
thereby favoring intermittent recovery during prolonged tasks. These observations
feed beliefs held by practitioners in occupational health that biomechanical variation
decreases the risk of developing MSDs as well as in sport biomechanics in a formu-
lation of the “variability-overuse hypothesis” (Bartlett et al. 2007). It should also be
noted that the hypothesis that increased variability during repetitive tasks illustrates
a search for solutions to an acute problem, up to now in this section represented as
fatigue, is supported by similar findings in situations of experimental pain induced
during a simulated cutting task (Madeleine et al. 2008) and during a prolonged,
discomfort-provoking sitting task (Søndergaard et al. 2010).
13.6.3 Variability in People with Musculoskeletal Disorders
The association between chronic pain conditions and reduced motor variability has
been shown in several studies. In their series of studies on butcher work, Madeleine
et al. showed that as opposed to acute pain, chronic pain was associated with de-
creased variability. In subjects with unilateral knee impairment, the injured limb
exhibited less motor variability than the intact one (Hamill et al. 1999; Heiderscheit
et al. 2002; Georgoulis et al. 2006). Recently, it was shown that people with low back
pain exhibited reduced trunk movement variability during gait in comparison to a
control group with no pain (van den Hoorn et al. 2012). In interpreting their findings,
authors suggest that less variable patterns illustrate protective strategies where one,
or a small set of motor solutions, representing the least amount of pain and/or small-
est risk of aggravation or injury, is found and maintained. However, some authors
suggest that more stereotyped (i.e., less variable) movement patterns may actually
represent a greater risk of injury (Mathiassen et al. 2003; Madeleine et al. 2008),
either by reducing the adaptability of the system to suddenly changing conditions or
by posing a risk of overload and injury to the consistently loaded structures (Côté
and Hoeger Bement 2010).
The above-cited studies likely would lead to the assumption that low variabil-
ity would emerge as a consequence of experiencing pain or injury. This is further
supported in a study on low gait variability in a pain group that showed increased
variability to almost that of a healthy group when pain was temporarily reduced by
the application of patellar taping (Heiderscheit 2000). However, other studies have
provided evidence to argue that variability may be a part of one’s inherent motor
pattern and as such would have an impact on one’s predisposition to pain maladap-
tive strategies. This hypothesis is supported by the observation that initially larger
movement variability was associated with higher probabilities of returning to normal
postural strategies after experimental pain (Moseley and Hodges 2006).
Even more so than in studies of variability with fatigue or acute pain, findings
from studies of variability between healthy groups and others with chronic pain are
largely impacted by the choice of metrics to quantify variability. For instance, given
the same amount of movement-to-movement variability (SD) in degrees of joint
motion amplitude, the finding that people with chronic pain usually show smaller
ranges of motion would mean that as a result, their coefficient of variation would be
larger. This was observed in a previous study of our group (Lomond and Côté 2011)
and is replicated with new data below on variability patterns displayed during the
repetitive pointing task performed in the sitting position (Fig. 13.5). This observation
has two major implications: first, at the methodological level, this implies that great
care should be taken when interpreting findings based on different ways to compute
variability. Second, at the theoretical level, this plays in favor for a recommendation to
consider both concepts of abundance (which has a direct impact in the contribution of
a specific joint or muscle towards the production of a task) and variability together, so
as to draw a more complete picture of whole-body adaptations, especially in chronic
pain.
13.7 Cinderella Versus Cinderello: A Quick Word on Gender

Differences
Most research agrees that in general, both the work conditions and the worker’s per-
sonal features may predict the likelihood of developing MSDs (Messing et al. 2003;
Messing and Stellman 2006; Théberge 2012). As such, both may potentially explain
why women report MSDs more often. Indeed, recent statistics in Quebec show that
work-related MSDs affect 16 % of men and 25 % of women (Stock et al. 2011).
Higher reports of neck/shoulder MSDs in women are especially well-documented
in the literature (Hooftman et al. 2009). A more detailed understanding of these
gender differences may help tailor more effective injury rehabilitation and preven-
tion approaches. Moreover, in the context of this chapter, a greater understanding
of factors underlying this difference could help gain a better understanding of the
role played by motor control mechanisms in the initiation and aggravation of MSDs.
220 J. N. Côté
Fig. 13.5 Anteroposterior shoulder amplitude and variability characteristics during the performance
of a repetitive reaching task at shoulder height, in the seated position. Upper panels: 30-s traces,
representative asymptomatic (CTRL, left) and chronic neck–shoulder injured (PAIN, right) partic-
ipants. Bottom panel: Variability index (coefficient of variation, CV ) computed for each subject,
during the first (NF, black) and last (FT ) minutes of the task. Despite the largest motion amplitude
displayed by CTRL at FT, their variability, computed using CV (SD/mean), is the smallest
This was the approach taken in a recent review (Côté 2012), which led to a first
attempt at conceptualizing a model based on anatomical and neuromuscular factors
to explain why the incidence and prevalence of neck/shoulder MSD is consistently
higher in women. With recent data on gender differences that can be related to the
concepts of variability and abundance, we propose a new version of this evolving
model (Fig. 13.6).
The main elements of this conceptual model have been described previously (Côté
2012), and the additions in this current version of the model were introduced in
previous sections of this chapter. To summarize briefly: a few studies have shown a
reduced ability to modify muscle activity patterns in women (Ge et al. 2005), notably
with women showing continued use of the same regions of the trapezius muscle even
in the presence of experimental pain, whereas men responded by using different
ones. Similar results were shown in our recent study on neck/shoulder patterns during
fatigue adaptation (Fedorowich et al. 2013). However, recent studies show higher
functional connectivity among neck/shoulder muscle parts during a low-fatigue task
(Johansen et al. 2013). This rather suggests that the reason for the seemingly lower
adaptability in women may be due to higher baseline functional connectivity. In other
words, it could be that for the same task, women initially already make optimal use
of abundance, such that their adaptability margin is lower. However, other studies
Fig. 13.6 Conceptual model to explain mechanisms underlying the physical basis for sex/gender
differences in the prevalence of neck/shoulder MSDs. Based on a previously published model
(Côté 2012), this new version integrates the concepts of abundance and variability at the core of
this chapter
are needed to confirm or disprove this hypothesis. Nevertheless, this line of findings
does shed an original light on the concept of abundance: similarly to findings from
the literature on synergies, the key could be the flexibility in creating and breaking
down synergies (Latash 2010), and in making use of abundance, that could predict
effective MSD prevention. Conversely, in a few recent studies, it was shown that
women displayed lower movement-to-movement force variability (Svendsen and
Madeleine 2010), which can be linked to reported higher accuracy in women (Côté
2012). However, a recent study from our group showed a positive association between
initial variability and endurance time in women (Fedorowich et al. 2013), further
reinforcing the importance of benefitting from flexibility to optimize the response to
stimuli such as fatigue and pain. The next logical step could then be to determine what
predicts the higher flexibility and adaptability in women compared to men? Again
using the gender comparison as a theoretical model, starting with the underlying
premise that there are well-documented differences in anthropometric and strength
characteristics between men and women, the key could lie in the size of the margin
between one’s capabilities and the demands of the task.
13.8 Conclusion: The Concept of Motor Repertoire
The notions of safety margin, leeway, and margin of maneuver have been introduced
in relation to health and performance, with slight distinctions according to their field
of application (Côté 2012; Durand et al. 2011; Côté et al. 2013). Generally, all of these
222 J. N. Côté
refer to some ability to perform a task, and this ability is introduced here as a motor
repertoire. The idea is that the possibility to benefit from a comparatively large motor
repertoire is desirable, we argue, for two purposes: safety and performance. Let us
compare the human body to a large orchestra displaying some musical repertoire. This
repertoire can be constituted of simple and complex pieces involving the contributions
of varying numbers of musicians. Similarly, at least some pieces of this repertoire
can be produced to some degree of quality even in the absence of some musicians.
The repertoire can be enriched with the introduction of some variations of musical
pieces, as long as they satisfy the overall quality requirements of the orchestra. All
of these characteristics can find parallels with what we know about the functioning
of the human body, including the notions of abundance and variability. In turn, one
objective pursued by ergonomists and therapists alike, could be to build a motor
repertoire as large as possible within the existing capabilities, which could in turn
serve to produce safe, durable, effective, and flexible motor performance. As such,
one last important element to address is the task-dependency characteristic of this
repertoire development. For instance, it is conceivable that equal instructions (e.g.,
move the finger as fast as possible; perform a data entry job for 40 h a week) can
be perceived differently from one person to the next, depending on the person’s
(possibly changing) disposition and the (possibly changing) conditions in which the
task is to be performed. Ultimately, the key could lie in creating a repertoire as
large as possible between the perceived capabilities and the perceived task demands
(Fig. 13.1), while accounting for flexibility for the capabilities and demands to evolve
through time. For instance, this repertoire could be smaller for a woman compared to
a man if the performance is related in any way to force capabilities, but this difference
could be smaller if performance was rather associated with high accuracy demands.
Still, despite these documented guiding principles that help understand how motor
capabilities can differ between two individuals, ergonomists and therapists can design
job modifications and training interventions to impact on health and performance.
Interventions based on the concepts of abundance and variability can in turn have a
positive impact on reducing and preventing MSDs.
Acknowledgments Research contributions from all past and current members of OBEL; funding
from NSERC and CFI; J. Côté currently holds a CIHR-IRSST Gender, Work and Health Research
Chair, and is a member of CRIR and REPAR.
References
Andersen JH, Kaergaard A, Mikkelsen S, Jensen UF, Frost P, Bonde JP, et al. Risk factors in the onset
of neck/shoulder pain in a prospective study of workers in industrial and service companies.
Occup Environ Med. 2003 Sep;60(9):649–54.
Aune TK, Ingvaldsen RP, Ettema GJ. Effect of physical fatigue on motor control at different skill
levels. Percept Mot Skills. 2008 Apr;106(2):371–86.
Aymard C, Katz R, Lafitte C, Le Bozec S, Penicaud A. Changes in reciprocal and transjoint inhibition
induced by muscle fatigue in man. Exp Brain Res. 1995;106(3):418–24.
Badley EM, Rasooly I, Webster GK. Relative importance of musculoskeletal disorders as a cause of
chronic health problems, disability, and health care utilization: findings from the 1990 Ontario
Health Survey. J Rheumatol. 1994 Mar;21(3):505–14.
Bartlett R, Wheat J, Robins M. Is movement variability important for sports biomechanists? Sports
Biomech. 2007 May;6(2):224–43.
Bawa P, Murnaghan C. Motor unit rotation in a variety of human muscles. J Neurophysiol. 2009
Oct;102(4):2265–72.
Bernstein NA (1967) The co-ordination and regulation of movements. Pergamon Press, Oxford
Bigland-Ritchie B, Johansson R, Lippold OC, Woods JJ. Contractile speed and EMG changes during
fatigue of sustained maximal voluntary contractions. J Neurophysiol 1983 Jul;50(1):313–24.
Bongers PM. The cost of shoulder pain at work. BMJ. 2001 Jan 13;322(7278):64–65.
Bosch T, Mathiassen SE, Visser B, de Looze MP, van Dieën JH. The effect of work pace on
workload, motor variability and fatigue during simulated light assembly work. Ergonomics.
2011 Feb;54(2):154–68.
Buckle PW, Devereux JJ. The nature of work-related neck and upper limb musculoskeletal disorders.
Appl Ergon. 2002 May;33(3):207–17.
Cantu HA, Emery K, Côté JN. Effects if additional external weight on posture and movement
adaptations to fatigue induced by a repetitive pointing task. Hum Mov Sci. (accepted)
Cignetti F, Schena F, Rouard A. Effects of fatigue on inter-cycle variability in cross-country skiing.
J Biomech. 2009 Jul 22;42(10):1452–9.
Côté JN. A critical review on physical factors and functional characteristics that may ex-
plain a sex/gender difference in work-related neck/shoulder disorders. Ergonomics. 2012
Aug;55(2):173–82.
Côté JN, Mathieu PA, Levin MF, Feldman AG. Movement reorganization to compensate for fatigue
during sawing. Exp Brain Res. 2002 Oct;146(3):394–8.
Côté JN, Levin MF, Raymond D, Mathieu PA, Feldman AG. Differences in multijoint kinematic
patterns of repetitive hammering in healthy, fatigued and shoulder-injured individuals. Clin
Biomech. 2005 Jul;20(6):581–90.
Côté JN, Feldman AG, Mathieu PA, Levin MF. Effects of fatigue on intermuscular coordination
during repetitive hammering. Motor Control. 2008 Apr;12(2):79–92.
Côté JN, Hoeger Bement MK. Update on the relation between pain and movement: consequences
for clinical practice. Clin J Pain. 2010 Nov–Dec;26(9):754–62.
Côté JN, Ngomo S, Stock S, Messing K, Vézina N, Antle D, Delisle A, Bellemare M, Laberge
M, St-Vincent M (2013) Quebec Research on Work-related Musculoskeletal Disorders:
Deeper Understanding for Better Prevention. Relations industrielles/Industrial Relations (RI/IR)
68(4):643–60.
Danion F, Latash ML, Li ZM, Zatsiorsky VM. The effect of fatigue on multifinger coordination in
force production tasks in humans. J Physiol. 2000 Mar 1;523 Pt 2:523–32.
Danion F, Latash ML, Li VM, Zatsiorsky VM. The effect of a fatiguing exercise by the index finger
on single- and multi-finger force production tasks. Exp Brain Res. 2001 Jun;138(3):322–9.
Davids K, Bennett S, Newell KM, 2006. Movement system variability. Human Kinetics,
Champaign, IL.
Dingwell JB, Cusumano JP, Cavanagh PR, Sternad D. Local dynamic stability versus kine-
matic variability of continuous overground and treadmill walking. J Biomech Eng. 2001
Feb;123(1):27–32.
Durand MJ, Vézina N, Baril R, Loisel P, Richard MC, Ngomo S. Relationship between the margin of
manoeuvre and the return to work after a long-term absence due to a musculoskeletal disorder:
an exploratory study. Disabil Rehabil. 2011;33(13–14):1245–52.
Edwards RHT (1983) Biochemical basis of fatigue in exercise performance: catastrophe theory
of muscular fatigue. In: Knuttgen HG, Vogel JA, Poortmans J (eds) Biochemistry of exercise.
Human Kinetics, Champaign, Ill., pp 3–28
Enoka RM, Stuart DG. Neurobiology of muscle fatigue. J Appl Physiol 1992 May;72(5):1631–48.
224 J. N. Côté
Falla D, Bilenkij G, Jull G. Patients with chronic neck pain demonstrate altered patterns of muscle
activation during performance of a functional upper limb task. Spine (Phila Pa 1976). 2004 Jul
1;29(13):1436–40.
Falla D, Farina D, Graven-Nielsen T. Experimental muscle pain results in reorganization of coordi-
nation among trapezius muscle subdivisions during repetitive shoulder flexion. Exp Brain Res.
2007a Apr;178(3):385–93.
Falla D, Farina D, Dahl MK, Graven-Nielsen T. Muscle pain induces task-dependent changes in
cervical agonist/antagonist activity. J Appl Physiol (1985). 2007b Feb;102(2):601–9.
Falla D, Arendt-Nielsen L, Farina D. Gender-specific adaptations of upper trapezius muscle activity
to acute nociceptive stimulation. Pain. 2008 Aug 15;138(1):217–25.
Fallentin N, Sidenius B, Jorgensen K. Blood pressure, heart rate, and EMG in low-level contractions.
Acta Physiol Scand. 1985 Oct;125(2):265–75.
Fedorowich L, Emery K, Gervasi B, Côté JN. Gender differences in neck/shoulder muscular
patterns in response to repetitive motion induced fatigue. J Electromyogr Kinesiol. 2013
Oct;23(5):1183–9.
Feeney A, North F, Head J, Canner R, Marmot M. Socioeconomic and sex differentials in reason
for sickness absence from the Whitehall II Study. Occup Environ Med. 1998 Feb;55(2):91–8.
Feuerstein M, Shaw WS, Lincoln AE, Miller VI, Wood PM. Clinical and workplace factors
associated with a return to modified duty in work-related upper extremity disorders. Pain. 2003
Mar;102(1-2):51–61.
Fourth European working conditions survey 2007. European Foundation for the Improvement
of the Living and Working Conditions. Office for Official Publications of the European
Communities, Luxembourg.
Fuller JR, Lomond KV, Fung J, Côté JN. Posture-movement changes following repetitive
motion-induced shoulder muscle fatigue. J Electromyogr Kinesiol. 2009 Dec;19(6):1043–52.
Fuller JR, Fung J, Côté JN. Time-dependent adaptations to posture and movement character-
istics during the development of repetitive reaching induced fatigue. Exp Brain Res. 2011
May;211(1):133–43.
Fuller JR, Fung J, Côté JN. Posture-movement responses to stance perturbations and upper limb
fatigue during a repetitive pointing task. Hum Mov Sci. 2013 Aug;32(4):618–32.
Ge HY, Arendt-Nielsen L, Farina D, Madeleine P. Gender-specific differences in electromyographic
changes and perceived pain induced by experimental muscle pain during sustained contractions
of the upper trapezius muscle. Muscle Nerve. 2005 Dec; 32(6):726–733.
Gelfand IM, Latash ML. On the problem of adequate language in movement science. Motor
Control. 1998 Oct;2(4):306–13.
Georgoulis AD, Moraiti C, Ristanis S, Stergiou N. A novel approach to measure variability in the
anterior cruciate ligament deficient knee during walking: the use of the approximate entropy in
orthopaedics. J Clin Monit Comput. 2006 Feb;20(1):11–8.
Hagg G (1991) Static work loads and occupational myalgia—a new explanation model. In: P.A.
Anderson, D.J. Hobart, and J.V. Danhoff, eds. Electromyographical kinesiology. Amsterdam:
Elsevier Science Publishers, 141–144.
Hagg GM. Human muscle fibre abnormalities related to occupational load. Eur J Appl Physiol.
2000 Oct;83(2–3):159–65.
Hagg GM, Astrom A. Load pattern and pressure pain threshold in the upper trapezius muscle and
psychosocial factors in medical secretaries with and without shoulder/neck disorders. Int Arch
Occup Environ Health. 1997;69(6):423–32.
Hagg GM, Suurkula J. Zero crossing rate of electromyograms during occupational work and
endurance tests as predictors for work related myalgia in the shoulder/neck region. Eur J Appl
Physiol Occup Physiol. 1991;62(6):436–44.
Hamberg-van Reenen HH, Ariëns GA, Blatter BM, van Mechelen W, Bongers PM. A systematic
review of the relation between physical capacity and future low back and neck/shoulder pain.
Pain. 2007 Jul;130(1–2):93–107.
Hamill J, van Emmerik RE, Heiderscheit BC, Li L. A dynamical systems approach to lower
extremity running injuries. Clin Biomech (Bristol, Avon). 1999 Jun;14(5):297–308.
Heiderscheit BC. Movement variability as a clinical measure for locomotion. J Appl Biomech.
2000 Nov;16(4):419–427.
Heiderscheit BC, Hamill J, van Emmerik REA. Variability of Stride Characteristics and Joint
Coordination Among Individuals With Unilateral Patellofemoral Pain. J Appl Biomech. 2002
May;18:110–121.
Henneman E. Relation between size of neurons and their susceptibility to discharge. Science. 1957
Dec 27;126(3287):1345–7.
Hermans V, Spaepen AJ. Perceived discomfort and electromyographic activity of the upper
trapezius while working at a VDT-station. Int J Occup Saf Ergon. 1995;1(3):208–214.
Hill AM, Bull AM, Wallace AL, Johnson GR. Qualitative and quantitative descriptions of
glenohumeral motion. Gait Posture. 2008 Feb;27(2):177–88.
Hooftman WE, van der Beek AJ, Bongers PM, van Mechelen W. Is there a gender difference in
the effect of work-related physical and psychosocial risk factors on musculoskeletal symptoms
and related sickness absence? Scand J Work Environ Health. 2009 Mar;35(2): 85–95.
Hug F. Can muscle coordination be precisely studied by surface electromyography? J Electromyogr
Kinesiol. 2011 Feb;21(1):1–12.
Huysmans MA, Hoozemans MJ, van der Beek AJ, de Looze MP, van Dieën JH. Fatigue effects on
tracking performance and muscle activity. J Electromyogr Kinesiol. 2008 Jun;18(3):410–9.
Johansen TI, Samani A, Antle DM, Côté JN, Madeleine P Gender effects on the coordination of
subdivisions of the trapezius muscle during a repetitive box-folding task. Eur J Appl Physiol.
2013 Jan;113(1):175–82.
Jones LA, Hunter IW. Effect of fatigue on force sensation. Exp Neurol 1983 Sept;81(3):640–50.
Karsh BT (2006) Theories of work-related musculoskeletal disorders: Implications for ergonomic
interventions. Theoretical Issues in Ergonomics Science 7(1) Jan–Feb, 71–88.
Kilbom A. Repetitive work of the upper extremity: part II-the scientific basis (knowledge base) for
the guide. Int J Ind Ergon 1994;14:59–86
Larsson B, Bjork J, Kadi F, Lindman R, Gerdle B. Blood supply and oxidative metabolism in muscle
biopsies of female cleaners with and without myalgia. Clin J Pain. 2004 Nov–Dec;20(6):440–6.
Larsson B, Sogaard K, Rosendal L. Work related neck-shoulder pain; a review on magnitude, risk
factors, biochemical characteristics, clinical picture and preventive interventions. Best Pract
Res Clin Rheumatol. 2007 Jun;21(3):447–63.
Latash ML. There is no motor redundancy in human movements. There is motor abundance. Motor
Control. 2000 Jul;4(3):259–60.
Latash ML. Two archetypes of motor control research. Motor Control. 2010 Jul;14(3):e41–e53.
Latash ML. The bliss (not the problem) of motor abundance (not redundancy). Exp Brain Res.
2012 Mar;217(1):1–5.
Latash ML, Anson JG. Synergies in health and disease: relations to adaptive changes in motor
coordination. Phys Ther. 2006 Aug;86(8):1151–60.
Latash ML, Scholz JP, Schöner G Toward a new theory of motor synergies. Motor Control. 2007
Jul;11(3):276–308.
Latash ML, Scholz JP, Schoner G. Motor control strategies revealed in the structure of motor
variability. Exerc Sport Sci Rev. 2002 Jan;30(1):26–31.
Leijon M, Hensing G, Alexanderson K. Gender trends in sick-listing with musculoskeletal
symptoms in a Swedish county during a period of rapid increase in sickness absence. Scand J
Soc Med. 1998 Sep;26(3):204–13.
Lexell J. Ageing and human muscle: observations from Sweden. Can J Appl Physiol. 1993
Mar;18(1):2–18.
Lindman R, Eriksson A, Thornell LE. Fiber type composition of the human male trapezius muscle:
enzyme-histochemical characteristics. Am J Anat. 1990 Nov;189(3):236–44.
226 J. N. Côté
Lomond KV, Côté JN. Movement timing and reach-to-reach variability during a repetitive reaching
task in persons with chronic neck/shoulder pain and healthy subjects. Exp Brain Res. 2010
Oct;206(3):271–82.
Lomond KV, Côté JN. Differences in posture-movement changes induced by repetitive arm
motion in healthy and shoulder-injured individuals. Clin Biomech (Bristol, Avon). 2011
Feb;26(2):123–9.
Lund JP, Donga R, Widmer CG, Stohler CS. The pain-adaptation model: a discussion of the
relationship between chronic musculoskeletal pain and motor activity. Can J Physiol Pharmacol.
1991 May;69(5):683–94.
Madeleine PM. On functional motor adaptations: from the quantification of motor strategies to the
prevention of musculoskeletal disorders in the neck-shoulder region. Acta Physiol (Oxf). 2010
Jun;199 Suppl 679:1–46.
Madeleine P, Lundager B, Voigt M, Arendt-Nielsen L. Shoulder muscle coordination during
chronic and acute experimental neck-shoulder pain. An occupational pain study. Eur J Appl
Physiol Occup Physiol. 1999 Jan;79(2):127–40.
Madeleine P, Lundager B, Voigt M, Arendt-Nielsen L. The effects of neck-shoulder pain develop-
ment on sensory-motor interactions among female workers in the poultry and fish industries.
A prospective study. Int Arch Occup Environ Health. 2003 Feb;76(1):39–49.
Madeleine P, Leclerc F, Arendt-Nielsen L, Ravier P, Farina D. Experimental muscle pain changes
the spatial distribution of upper trapezius muscle activity during sustained contraction. Clin
Neurophysiol. 2006 Nov;117(11):2436–45.
Madeleine P, Mathiassen SE, Arendt-Nielsen L. Changes in the degree of motor variability
associated with experimental and chronic neck-shoulder pain during a standardised repetitive
arm movement. Exp Brain Res. 2008 Mar;185(4):689–98.
Madeleine P, Samani A, Binderup AT, Stensdotter AK. Changes in the spatiotemporal organization
of the trapezius muscle activity in response to eccentric contractions. Scand J Med Sci Sports.
2011 Apr;21(2):277–86.
Martin V, Scholz JP, Schöner G. Redundancy, self-motion, and motor control. Neural Comput.
2009 May;21(5):1371–414.
Mathiassen SE, Aminoff T. Motor control and cardiovascular responses during isoelectric
contractions of the upper trapezius muscle: evidence for individual adaptation strategies. Eur
J Appl Physiol Occup Physiol. 1997;76(5):434–44.
Mathiassen SE, Winkel J. Electromyographic activity in the shoulder-neck region according to arm
position and glenohumeral torque. Eur J Appl Physiol Occup Physiol. 1990;61(5–6):370–9.
Mathiassen SE, Möller T, Forsman M. Variability in mechanical exposure within and between
individuals performing a highly constrained industrial work task. Ergonomics. 2003 Jun
20;46(8):800–24.
Maton B, Gamet D. The fatigability of two antagonistic muscles in human isometric voluntary
submaximal contraction: an EMG study II-motor unit firing rate and recruitment. Eur J Appl
Physiol Occup Physiol. 1989;58(4):369–74.
McLean L, Goudy N. Neuromuscular response to sustained low-level muscle activation: within- and
between-synergist substitution in the triceps surae muscles. Eur J Appl Physiol. 2004 Mar;91
(2–3):204–16
Messing K, Stellman JM. Sex, gender and women’s occupational health: The importance of
considering mechanism. Environ Res. 2006 Jun;101(2):149–162.
Messing K, Punnett L, Bond M, Alexanderson K, Pyle J, Zahm S, Wegman D, Stock S, De
Grosbois S. Be the fairest of them all: challenges and recommendations for the treatment of
gender in occupational health research. Am J Ind Med. 2003 Jun;43(6):618–629.
Missenard O, Mottet D, Perrey S. The role of cocontraction in the impairment of movement
accuracy with fatigue. Exp Brain Res. 2008 Feb;185(1):151–6.
Moseley GL, Hodges PW. Reduced variability of postural strategy prevents normalization of
motor changes undiced by back pain: a risk factor for chronic trouble? Behav Neurosci. 2006
Apr;120(2):474–6.
Nelson-Wong E, Callaghan JP. Is muscle co-activation a predisposing factor for low back pain
development during standing? A multifactorial approach for early identification of at-risk
individuals. J Electromyogr Kinesiol. 2010 Apr;20(2):256–63.
Newell KM, Corcos DM, 1993. Variability and Motor Control. Human Kinetics, Champaign, IL.
Newell KM, Slifkin AB, 1998. The Nature of Movement Variability. In: Piek, J.P. (Ed.) Motor
behavior and human skill. A multidisciplinary approach. Human Kinetics.
Nordlund A, Ekberg K. Self reported musculoskeletal symptoms in the neck-shoulders and-or
arms and general health (SF-36): eight year follow up of a case-control study. Occup Environ
Med. 2004 Mar;61(3):e11.
Parker PA, Korner L, Kadefors R. Estimation of muscle force from intramuscular total pressure.
Med Biol Eng Comput. 1984 Sep;22(5):453–7.
Person RS. Rhythmic activity of a group of human motoneurons during voluntary contraction of
a muscle. Electroencephalogr Clin Neurophysiol. 1974 Jun;36(6):585–95.
Psek JA, Cafarelli E. Behavior of coactive muscles during fatigue. J Appl Physiol (1985). 1993
Jan;74(1):170–5.
Rempel DM, Harrison RJ, Barnhart S. Work-related cumulative trauma disorders of the upper
extremity. JAMA. 1992 Feb 12;267(6):838–42.
Riley MA, Turvey MT. Variability and determinism in motor behavior. J Mot Behav. 2002
Jun;34(2):99–125.
Rosenblatt NJ, Hurt CP, Latash ML, Grabiner MD. An apparent contradiction: increasing variability
to achieve greater precision? Exp Brain Res. 2013 Oct 27.
Sale DG (1987) Influence of exercise and training on motor unit activation. In: Pandolf KB (ed)
Exercise and sport science reviews, vol 15. MacMillan, London, pp 95–151
Scholz JP, Schöner G. The uncontrolled manifold concept: identifying control variables for a
functional task. Exp Brain Res. 1999 Jun;126(3):289–306.
Selen LP, Beek PJ, van Dieën JH. Fatigue-induced changes of impedance and performance in
target tracking. Exp Brain Res. 2007 Jul;181(1):99–108.
Shiro Y, Arai YC, Matsubara T, Isogai S, Ushida T. Effect of muscle load tasks with maximal iso-
metric contractions on oxygenation of the trapezius muscle and sympathetic nervous activity in
females with chronic neck and shoulder pain. BMC Musculoskelet Disord. 2012 Aug 13;13:146.
Silverstein B, Welp E, Nelson N, Kalat J. Claims incidence of work-related disorders of the upper ex-
tremities: Washington State, 1987 through 1995. Am J Public Health. 1998 Dec;88(12):1827–33.
Sjogaard G, Sogaard K. Muscle injury in repetitive motion disorders. Clin Orthop Relat Res. 1998
Jun;(351):21–31.
Sjogaard G, Kiens B, Jorgensen K, Saltin B. Intramuscular pressure, EMG and blood flow during
low-level prolonged static contraction in man. Acta Physiol Scand. 1986 Nov;128(3):475–84.
Skurvydas A, Brazaitis M, Kamandulis S. Prolonged muscle damage depends on force variability.
Int J Sports Med. 2010 Feb;31(2):77–81.
Sluiter JK, Rest KM, Frings-Dresen MH. Criteria document for evaluating the work-relatedness
of upper extremity musculoskeletal disorders. Scand J Work Environ Health. 2001;27 Suppl
1:1–102.
Søndergaard KH, Olesen CG, Søndergaard EK, de Zee M, Madeleine P. The variability and
complexity of sitting postural control are associated with discomfort. J Biomech. 2010 Jul
20;43(10):1997–2001.
Srinivasan D, Mathiassen SE. Motor variability in occupational health and performance. Clin
Biomech (Bristol, Avon). 2012 Dec;27(10):979–93.
Stergiou N, 2004. Innovative analyses of human movement. Human Kinetics, Champaign, IL.
Stergiou N, Harbourne R, Cavanaugh J. Optimal movement variability: a new theoretical
perspective for neurologic physical therapy. J Neurol Phys Ther. 2006 Sep;30(3):120–9.
Stock S, Funes A, Delisle A, St-Vincent M, Turcot A, Messing M (2011) Chapter 7. Troubles
musculo-squelettiques. In Vézina M et al. Enquê te québécoise sur des conditions de travail,
d’emploi, et de santé et de sécurité du travail (EQCOTESST), Québec, Institut de recherche
228 J. N. Côté
Robert-Sauvé en santé et sécurité du travail - Institut national de santé publique du Québec et

Institut de la statistique du Québec.
Stokes MJ, Edwards RHT, Cooper RG. Effect of low frequency fatigue on human muscle strength
and fatigability during subsequent stimulated activity. Eur J Appl Physiol Occup Physiol.
1989;59(4):278–83.
Sundelin G, Hagberg M. Electromyographic signs of shoulder muscle fatigue in repetitive arm
work paced by methods-time measurement system. Scand J Work Environ Health. 1992
Aug;18(4):262–8.
Svendsen JH, Madeleine P. Amount and structure of force variability during short, ramp and
sustained contractions in males and females. Hum Mov Sci. 2010 Feb;29(1):35–47.
Takala EP. Static muscular load, an increasing hazard in modern information technology. Scand J
Work Environ Health. 2002 Aug;28(4):211–3.
Theberge N. Studying gender and injuries: a comparative analysis of the literatures on women’s
injuries in sport and work. Ergonomics. 2012 Aug;55(2):183–193.
Thorn S, Forsman M, Zhang Q, Taoda K. Low threshold motor unit activity during a 1-h static
contraction in the trapezius muscle. In. J Ind Ergon. 2002 30:225–236.
Tjepkema M. Repetitive strain injury. Health Rep. 2003 Aug;14(4):11–30.
van den Hoorn W, Bruijn SM, Meijer OG, Hodges PW, van Dieen JH. Mechanical coupling
between transverse plane pelvis and thorax rotations during gait is higher in people with low
back pain. J Biomech. 2012 Jan;45(2):342–7.
van der Windt DA, Thomas, Bouter LM et al. Occupational risk factors for shoulder pain: a
systematic review. Occup Environ Med. 2000 Jul;57(7):433–42.
van Dieen JH, Vrielink HHEO, Toussaint HM. An investigation into the relevance of the pattern of
temporal activation with respect to erector spinae muscle endurance. Eur J Appl Physiol Occup
Physiol. 1993;66(1):70–5.
Veiersted KB, Westgaard RH. Development of trapezius myalgia among female workers performing
light manual work. Scand J Work Environ Health. 1993 Aug;19(4):277–83.
Veiersted KB. Sustained muscle tension as a risk factor for trapezius myalgia. Int J Ind Ergon.
1994;14:333–339
Visser B, van Dieen JH. Pathophysiology of upper extremity muscle disorders. J Electromyogr
Kinesiol. 2006 Feb;16(1):1–16.
Voerman GE, Vollenbroek-Hutten MM, Hermens HJ. Upper trapezius muscle activation patterns
in neck-shoulder pain patients and healthy controls. Eur J Appl Physiol. 2007 Dec;102(1):1–9.
Vollestad NK. Measurement of human muscle fatigue. J Neurosci Methods 1997 Jun;74:219–27.
Wahlstedt K, Norbäck D, Wieslander G, Skoglund L, Runeson R. Psychosocial and ergonomic
factors and their relation to musculoskeletal complaints in the Swedish workforce. Int J Occup
Saf Ergon. 2010;16(3):311–21.
Westad C, Westgaard RH, De Luca CJ. Motor unit recruitment and derecruitment induced by brief
increase in contraction amplitude of the human trapezius muscle. J Physiol. 2003 Oct 15;552
(Pt 2):645–56.
Westgaard RH, DeLuca CJ. Motor unit substitution in long duration contractions of the human
trapezius muscle. J Neurophysiol. 1999 Jul;82(1):501–4.
Winkel J, Westgaard RH. Occupational and individual risk factors for shoulder-neck complaints:
part II-the scientific basis (literature review) for the guide. Int J Ind Ergon. 1992;10(1–2):85–104.
Woolf AD, Vos T, March L. How to measure the impact of musculoskeletal conditions. Best
practice & research. Best Pract Res Clin Rheumatol. 2010 Dec;24(6):723–32.
World Health Organization International Classification of Functioning
(http://www.who.int/classifications/icf/)
Zennaro D, Laubli T, Krebs D, Klipstein A, Krueger H. Continuous intermitted and sporadic motor
unit activity in the trapezius muscle during prolonged computer work. J Electromyogr Kinesiol.
2003 Apr;13(2):113–24.
Chapter 14
Deficits in Spatial Threshold Control of Muscle
Activation as a Window for Rehabilitation After
Brain Injury
Mindy F. Levin
14.1 Background
14.1.1 The Problem of Spasticity and Disordered Motor Control
Spasticity is associated with an upper motor neuron lesion affecting recovery of motor
function (Kim and Park 2011; Sunnerhagen et al. 2013). It is estimated that spasticity
affects over 12 million patients worldwide who have had a stroke, spinal cord injury,
head trauma, multiple sclerosis, cerebral palsy (CP) or other disorder of the central
nervous system (CNS; Burke 1988; www.aans.org). At the root of the motor deficit
after stroke is hemiparesis characterized by a diminished capacity to recruit agonist
muscles (Hammond et al. 1988; Gemperline et al. 1995; Chang et al. 2013), or
the use of abnormal recruitment patterns (Bourbonnais et al. 1989; Dewald et al.
1995; Kamper and Rymer 2001; but see Gowland et al. 1992; Fellows et al. 1994a,
1994a) and weakness (Kamper et al. 2006; Chang et al. 2013). Agonist–antagonist
coactivation may be responsible for a reduced ability to selectively activate arm and
hand muscles (Lang and Schieber 2004). This disability can also be associated with
abnormal timing of agonist and synergist muscle activation, and failure to deactivate
antagonist muscles (Hoffman and Strick 1995).
Spasticity, weakness, and motor impairments have traditionally been considered
as separate phenomena (Fig. 14.1). However, the idea that they are interrelated was
alluded to in the definition suggested by Lance in 1980. This definition, now fre-
quently cited in the literature, describes spasticity as “a motor disorder characterized
by a velocity-dependent increase in stretch reflexes (muscle tone with exaggerated
tendon jerks) as one component of the upper motor neuron syndrome.” Lance’s def-
inition suggests that spasticity and motor deficits are related to problems in stretch
M. F. Levin ()
School of Physical and Occupational Therapy,
McGill University, 3654 Promenade Sir William Osler, Montreal, QC H3G 1Y5, Canada
and Biology 826, DOI 10.1007/978-1-4939-1338-1_14
230 M. F. Levin
BrainLesion
Spasticity Weakness
Motor Impairments
-excessive co-activation
-lack of appropriate co-activation
-difficulty maintaining sustained
contraction
-difficulty relaxing muscles (excessive
prolonged contraction)
-abnormal force/EMG relationship
-etc.
Fig. 14.1 Schematic diagram of the interrelationship between spasticity, weakness, and disordered
motor function after a brain lesion
14 Deficits in Spatial Threshold Control of Muscle Activation . . . 231
reflex excitability. However, the precise relationship between spasticity, hyperac-

tive stretch reflexes, and the abnormalities in the production of voluntary movement
remains unclear.
Studies attempting to determine the degree to which altered muscle tone and
stretch reflex hyperexcitability may influence voluntary motor control have yielded
equivocal results. Corcos et al. (1986) studied ballistic ankle dorsiflexion movements
in subjects with spasticity of mixed etiology and showed that deficits in voluntary calf
muscle activation were related to hyperactive stretch reflexes. However, other studies
of upper limb hemiparesis reported that the disorder in voluntary motor control may
not be related to spasticity or hyperactive stretch reflex activity (Sahrmann and Norton
1977; O’Dwyer et al. 1996; Burne et al. 2005; Chang et al. 2013). The controversy
is associated with a lack of a coherent view of the relationship between spasticity
and disordered voluntary muscle activation.
A major problem in the understanding of this relationship is the method used to
quantify spasticity. Although the presence or absence of spasticity can be identified
using current clinical scales (Malhotra et al. 2009), such clinical scales are insufficient
to determine the relationship between spasticity, deficits of voluntary movements,
and functional ability (for reviews, see Elovic et al. 2004; Wood et al. 2005; Calota
and Levin 2009). There is general agreement that current scales of spasticity, even if
objective, measure biomechanical variables (e.g., the resistance to imposed muscle
stretch) that are effects, rather than causes of spasticity (Malhotra et al. 2009), and that
various physiological measures (e.g., inhibition/facilitation mediated by cutaneous
and muscular afferents) do not provide a comprehensive understanding of the nature
of spasticity and disordered movement (Malhotra et al. 2008, but see Levin et al.
2000, Krakauer 2005; Musampa et al. 2007; Mullick et al. 2013).
Different components of spasticity have been identified using physiological mea-
sures (e.g., disorders of reflex pathways, Mazevet et al. 2003), biomechanical
measures (e.g., deficits in interactive joint torque control, Dewald et al. 1995; Beer
et al. 2000), or a combination of both (e.g., reflex and nonreflex components of ankle
spasticity, Zhang et al. 2013); based on the assumption that these components are
controlled separately. However, there is increasing evidence to suggest that not only
spasticity and movement impairments but also muscle weakness and loss of dexterity
(skilled muscle actions) are all consequences of the same underlying control deficit.
It is proposed that this common control deficit is the specification and regulation of
spatial thresholds (STs) of the stretch and other proprioceptive reflexes (Levin and
Feldman 1994; Feldman and Levin 1995; Musampa et al. 2007; Feldman 2011).
14.1.2 ST Regulation in the Healthy Nervous System
The ST corresponds to the muscle length or joint angle (R) at which muscle activity
begins. This expresses the muscle activation threshold in the spatial (angular) rather
than the temporal (e.g., in terms of latencies) domain. The ST is the position of a
body segment at which postural reflexes begin to act—such that postural reflexes are
232 M. F. Levin
“centered” at the ST (Levin and Feldman 1994; Feldman 2011). Associated with
the ST is a torque/angle characteristic representing the dependency of muscle re-
flex force on muscle length for a given threshold. The angular range through which
the ST and its associated torque/angle relationship can be regulated is shown as a
horizontal band in Fig. 14.2a. In the healthy nervous system, ST can be regulated
throughout and beyond the biomechanical range of the joint (shown as the physical
limits of joint flexion and extension lying between θ− and θ+ , which correspond to
the minimal and maximal biomechanical muscle length, respectively). The limits
of ST regulation are shown by R− and R+ . By shifting ST, the brain resets (“re-
addresses”) posture-stabilizing mechanisms to a new limb or body position. When
the ST is shifted entirely to the left as shown in Fig. 14.2a (the lower limit, R− ), the
muscle is activated even at very short muscle lengths. When it is shifted beyond the
biomechanical range to the upper limit (R+ ), the muscle cannot be activated and is
fully relaxed. By regulating ST and its associated torque/angle characteristic (diago-
nal line in Fig. 14.2a) within these limits, the CNS can produce any physiologically
possible combination of muscle activity, torque, and position without the need to
specify these variables directly (Levin et al. 2000; Feldman 2011). Resetting of ST
is also associated with the conversion of movement-resisting to movement-producing
forces, providing a solution to the fundamental posture-movement problem; origi-
nally formulated by von Holst (1954; see also Ilmane et al. 2013 and Feldman, this
volume).
ST regulation is an important mechanism explaining motor control in the healthy
nervous system (Feldman and Levin 1995; Feldman 2009, 2011). ST control of
stretch reflexes (in cats and humans) and intentional movements (in humans) is well
established (Matthews 1959; Asatryan and Feldman 1965; Feldman and Orlovsky
1972; Nichols and Steeves 1986; Capaday 1995; Raptis et al. 2010; Sangani et al.
2011; Ilmane et al. 2013). The ST of a given muscle is controlled by descending
inputs directly or indirectly influencing the membrane potential or electrical thresh-
old of α-motoneurons (pre and postsynaptically via interneurons or γ-motoneurons;
Matthews 1959; Feldman and Orlovsky 1972; Nichols and Steeves 1986; Hultborn
and Kiehn 1992; Capaday 1995; McClelland et al. 2001). Changes in ST can be
mediated by cutaneous afferents or those responsible for reflex intermuscular in-
teractions, including reciprocal inhibition of agonist–antagonist muscles (Matthews
1959; Feldman and Orlovsky 1972). Segmental mechanisms including presynaptic
and reciprocal inhibition are themselves modulated by biomechanical factors leading
to the recruitment of different subsets of motoneurons (Ter Haar Romeny et al. 1984;
van Zuylen et al. 1988) as well as by task-related descending influences mediated
by neuromodulators, such as serotonin and norepinephrine, which change motoneu-
ronal intrinsic properties, thus causing depolarization, plateau potentials, and shifts
in their electrical thresholds (Lundberg 1967; Hultborn et al. 1987; Hultborn and
Kiehn 1992; Meunier and Pierrot-Deseilligny 1998; McPherson et al. 2008). These
mechanisms combine to regulate STs in a multi-muscle system according to the
configuration of the system and specific task demands (Nichols and Steeves 1986;
Feldman and Levin 1995; McClelland et al. 2001).
a 36
32
28
Torque (Nm)
24
20
16
12
8
4
0
0 20 40 60 80 100 120 140 160 180 200 220
b 36
32
28
Torque (Nm)
24
20
16
12
8
4
0
0 20 40 60 80 100 120 140 160 180 200 220
Active
Weakness Spasticity
Control
Zone Zone
Zone
Joint angle (º)
Fig. 14.2 Schematic diagram of central control of spatial thresholds (STs) of proprioceptive
reflexes and motor deficits resulting from impairments of ST control. In this scheme, it is assumed
that flexor muscles lengthen and extensor muscles shorten with increasing joint angle. a Normally,
the range (R− , R+ ) of threshold regulation exceeds the biomechanical range of the joint (θ− , θ+ ),
thus enabling the system to relax (when R+ is shifted to the right beyond θ+ ) or to activate muscles
at any angle (when R+ is shifted anywhere between θ+ and R− ) to generate motion within the
234 M. F. Levin
14.1.3 ST Regulation After CNS Lesions
As reviewed above, in healthy subjects normal muscle activation is associated with

an ability to regulate the ST in a range that exceeds the biomechanical joint range.
Consequently, a decrease in the range of ST regulation following CNS injury or
disease would result in motor control deficits (Fig. 14.2b and 14.2c). Deficits in
descending and spinal mechanisms resulting from CNS injury, possibly together with
changes in intrinsic motoneuronal properties such as in chloride reversal potentials
and properties of serotonin receptors, in combination and isolation may contribute
to limitations in ST regulation and the appearance of spasticity (Nielsen et al. 2007).
Empirical support for a common deficit in ST regulation underlying spasticity
and motor control problems in single- and double-joint systems has been described
in subjects with stroke (Lee et al. 1987; Powers et al. 1989; Levin and Dimov 1997;
Levin et al. 2000; Mihaltchev et al. 2005; Musampa et al. 2007) and in children
with hemiplegic CP (Jobin and Levin 2000). In particular, poststroke subjects with
spasticity have deficits in the regulation of STs within the physiological joint range
(Fig. 14.2b). Figure 14.2b shows what would occur if the lower and upper physi-
ological limits of ST regulation (R− , R+ ) were reduced so that they lie within the
biomechanical range of the joint. This would result in a narrowing of the range in
which movement can be produced using “typical” reciprocal muscle innervation pat-
terns. This range is called the “active control zone” since it is characterized by muscle
activation patterns observed in people without neurological injury or disease. In the
joint range between R+ and the end of the joint excursion, muscles would be unable
to relax at rest and the amount of muscle activation would be related to the velocity
of stretch. This spatial joint range is called the “spasticity zone,” which is not usually
observed in people without neurological injury. Conversely, when the lower ST limit
(R− ) is increased (i.e., shifted to the right of θ− ), the spatial zone between θ− and R−
would be characterized by muscle weakness (“weakness zone”) due to inadequate
muscle activation (Levin et al. 2000). A combination of changes in the regulation of
the upper and lower limits of R can result in different deficits occurring in different
joint ranges (zones) accounting for abnormal muscle activation patterns observed
when patients attempt voluntary movement (Levin et al. 2000; Lang and Schieber
2004).
Patterns of spasticity, weakness, and active control zones in flexor and extensor
muscles around the elbow joint from 12 subjects with chronic stroke (> 0.8–7.1
years post stroke) and different levels of arm paresis ranging from mild to severe
biomechanical range. b Following a brain lesion, the upper limit (R+ ) of threshold regulation can
abnormally fall within the biomechanical range, resulting in an inability to relax flexor muscles at
joint angles exceeding the position of R+ . Clinically, this deficit in poststroke subjects is identified
as spasticity. c Following a brain lesion, the lower limit (R− ) can also fall within the biomechanical
range, resulting in an inability to activate flexor muscles if the joint angle is less than the R−
(weakness zone)
(Fugl-Meyer Arm Assessment Score 15–65 out of 66; Fugl-Meyer et al. 1975) are
shown in Fig. 14.3 (Levin et al. 2000). Zones were determined by stretching pas-
sive muscles at different velocities to identify STs of elbow flexors Rf+ and elbow
extensors Re+ and then by identifying patterns of agonist and antagonist muscle coac-
tivation during slow full-range voluntary elbow extension and flexion. Each subject
had a unique pattern of active control (blue), spasticity (black/yellow), and coacti-
vation (black) zones within the biomechanical joint range. The range of movement
in which active movement was possible is indicated from M− to M+ . The range of
M could extend beyond the active control zone, but movements made in this angular
zone would be characterized by agonist/antagonist muscle coactivation. The figure
shows that different patients had different angular ranges in which active movement
was accomplished using reciprocal muscle activation (blue zones) and further move-
ment was possible only with muscle coactivation (black zones). In all cases, the
borders between reciprocal and coactivation zones coincided with STs identified in
flexor (Rf+ ) and extensor (Re+ ) muscles, shown for the first four subjects in Fig. 14.3.
14.1.4 Methodology Used to Identify the Range

of Regulation of STs
In a series of studies, we sought to determine the upper limit of ST regulation for flexor
and extensor muscles of the elbow in patients with spasticity. However, by definition,
the tonic ST is the value of the threshold when the system is at rest (e.g., at zero
velocity of stretch) and thus cannot be measured directly. Therefore, we developed
a method to determine the velocity-dependent or dynamic stretch reflex thresholds
(DSRTs), and then to extrapolate the value of the tonic stretch reflex threshold (TSRT)
from these values. Each DSRT is the joint angle and velocity corresponding to the
onset of the EMG in the stretched muscles at each stretch velocity. A regression line
is plotted through the set of DSRTs and the intercept of this regression line with
zero velocity is the TSRT (Fig. 14.4b and 14.4c; Mullick et al. 2013). The TSRT
thus corresponds to the ST for this muscle. In nondisabled individuals, DSRTs can
usually only be evoked in noncontracting muscles if the stretch is performed at very
high velocities (> 300◦ /s; Thilmann et al. 1991; Levin et al. 2000). However, in
adults with stroke and children with CP, muscle stretches at speeds as low as 8◦ /s,
applied to the elbow joint, can generate DSRTs.
Previous studies in patients with spasticity due to stroke or CP have shown that
(1) the TSRT lies within the physiological range of motion of the elbow; (2) the
TSRT value may or may not be correlated with the degree of resistance produced by
the same muscles when they are stretched; (3) the TSRT may be a better measure
of spasticity than muscle resistance measures (Jobin and Levin 2000; Levin et al.
2000).
Jobin and Levin (2000) measured TSRTs in elbow flexor muscles of children
with CP and tested the reliability of this value as an estimate of clinical spasticity.
Fourteen children with CP and eight typically developing children participated in the
236 M. F. Levin
M- M+
Least S1
severe M- Rf+ M+
S2
Re+
M- M+
S3 M-
M- Rf+ M+
S4
M- M+
S5
M+
S6
M- M+
S7
M- M+
S8
M- M+
S9
M- M+
S10
M- M+
S11
Most
M- M+
severe
S12
40 60 80 100 120 140 160 180

Angle (deg)
Active Control Coactivation Spasticity/no
movement
Fig. 14.3 The profiles of active control (blue), spasticity + coactivation (black + yellow) and
spasticity + weakness (yellow) zones in 12 patients with chronic stroke (> 0.8–7.1 years post stroke)
and different levels of arm paresis ranging from mild to severe (Fugl-Meyer Arm Assessment Score
15–65 out of 66). Each subject had a unique pattern of disability. Subject 1 (S1) had full range of
movement with typical muscle innervation patterns. All other subjects had limited active control
zones, while S9 and S10 had no active control zones throughout the entire joint range. Movements
were still possible in these two subjects, but these were accomplished with muscle coactivation
(black zones). (Adapted with permission from Levin et al. 2000)
200
180 BR
160
140 r2=0.98
120
Elbow angle 100
80
20 deg
60
40
Tonic ST:
Velocity 20
158.3°
0
20 60 100 140 180
40 deg/s b
200
180 BB
160 r2=0.98
140
Velocity (°/s)
120
100
EMG BR
80
1s
60
40
a 20
Tonic ST:
153.2 °
0
20 60 100 140 180
c Joint angle (°)
Fig. 14.4 Spasticity in a poststroke subject is associated with the presence of velocity-dependent
threshold angles beyond which the muscle cannot be relaxed. a Single traces of elbow displacements
(top) through a 90◦ range of extension at seven different velocities (middle) ranging from 8◦ /s
(purple) to 160◦ /s (dark blue). Stretches were applied by a torque motor. EMG activity from Biceps
Brachii (BB) evoked by each stretch is shown in bottom traces. Dynamic stretch reflex thresholds
(DSRTs) are defined as the angle and velocity values at which EMG activation in the stretched muscle
begins (arrows). b, c Example of dependencies of reflex spatial thresholds (STs) on velocity for
two elbow flexors (BR, Brachioradialis; BB) in stroke subjects. Each point on the graphs represents
the DSRTs for different velocities of stretch. The intercept and slope of the regression line through
the DSRTs are the tonic stretch reflex threshold (TSRT ) and dynamic sensitivity (μ) of the system,
respectively. Horizontal strips below each graph show the flexor spasticity zone (red) and active
control zone (white). (Adapted with permission from Mullick et al. 2013)
trial. DSRTs were evaluated by performing eight sets of stretches at seven, randomly
selected velocities between 8 and 160◦ /s using a torque motor. For the elbow flexors,
TSRT angles occurred later (closer to full elbow extension) in children with less-
severe spasticity and TSRTs were only weakly correlated with clinically measured
resistance to stretch of the passive muscle (r = 0.39). The test–retest reliability of
the computed TSRTs was estimated as good (ICC = 0.73, p < 0.001).
238 M. F. Levin
14.1.5 TSRTs in Multi-Joint Muscles and Double-Joint Systems
Although the range of regulation of TSRTs can predict the location of ranges of
spasticity and normal and abnormal movement patterns in a single joint, everyday
gestures involve movements of more than one joint. Indeed, the TSRT value in a
given muscle can be modulated by reflex interactions between muscles crossing
adjacent joints (Matthews 1993; Bonasera and Nichols 1994; Ginanneschi et al.
2006; Musampa et al. 2007; Roberts et al. 2008). Interjoint interactions include
autogenic facilitation; recurrent inhibition; presynaptic inhibition; reciprocal inhi-
bition and facilitation; and heteronymous inhibition and facilitation (Cavallari and
Katz 1989; Gracies et al. 1991; Katz et al. 1991; Cavallari et al. 1992; Créange
et al. 1992). Interjoint reflex interactions have been identified in arm muscles in
healthy subjects (Marchand-Pauvert et al. 2000; Archambault et al. 2005) and may
lead to changes in net motoneuronal excitability of the stretched muscle from the
convergence of information from afferents of heteronymous muscles via spinal, pro-
priospinal, and/or transcortical pathways. In the intact nervous system, intermuscular
reflex interactions have been reported for double-joint elbow muscles (Biceps Brachii
and Triceps Brachii) when muscles were preactivated and placed in three different
positions (McClelland et al. 2001).
In patients with poststroke spastic hemiparesis, Musampa et al. (2007) evalu-
ated the influence of intermuscular interactions on TSRTs in uni and biarticular
muscles of the double-joint elbow–shoulder system using a similar methodology as
described above. They characterized the relationship between muscle length, TSRTs,
and voluntary muscle activation patterns in elbow flexors and extensors and identi-
fied spasticity zones in the space of elbow–shoulder configurations from three initial
positions of the shoulder (Fig. 14.5a). They also investigated how the presence of
spasticity zones influenced voluntary arm movement. Similar to the previous identi-
fication of spasticity zones in a single joint (elbow; Levin et al. 2000), the limitations
in the regulation of TSRTs in the double-joint arm system were shown to result
in a subdivision of all-possible arm configurations into spatial spasticity zones and
no-spasticity zones (Fig. 14.5b). All patients had ranges of shoulder–elbow arm con-
figurations in which muscles could not be relaxed (spatial spasticity zones), which in
some patients covered a substantial part of the biomechanically defined range of all
possible arm configurations. It was also observed that these zones were practically
the same for synergistic muscles and could overlap for antagonist muscle groups,
such that there could be a zone in which both flexor and extensor muscles would
show spasticity.
When patients attempted to voluntarily extend or flex their arm into the spas-
ticity zones, significant abnormal agonist–antagonist muscle coactivation patterns
occurred. Of note is that the angular value of the TSRTs evoked by stretching of
passive muscles and the borders of the spasticity zone identified by active movement
were highly correlated (r = 0.64–0.86), indicating a relationship between spasticity
and disordered movement confined to specific areas in joint space. Thus, during
active elbow extension, when the elbow reached the specific joint angle associated
with the elbow flexor TSRT, elbow flexors became active and coactivation occurred.
This relationship was less marked in the extensors during active flexion.
a Extension
Flexion
60° 90° 145°
b 180
S1
140
100
60
BR AN
20
180
S2
140
100
60
20
180
S3
140
100
60
20
20 60 100 140 180 20 60 100 140 180
Fig. 14.5 a Methodology used to stretch single- and double-joint elbow muscles in order to identify
the influence of initial muscle length (changes in initial shoulder position) on evoked responses.
Initial position of the shoulder was varied from 60◦ to 145◦ before stretching elbow flexors (top row)
and elbow extensors (bottom row). b Spatial thresholds (STs) evoked in four elbow muscles (Biceps
Brachii, BB, black squares; Brachioradialis, BR, black circles; Triceps Brachii, TB, white squares;
Anconeus, AN, white circles). Spatial spasticity zones are shown separately for the flexor muscles
(left panels) and extensor muscles (right panels). Data from three patients with stroke-related upper
240 M. F. Levin
In summary, our studies have shown that limitations in the ranges of regulation
of the TSRT correspond to the appearance of abnormal muscle activation patterns,
such as excessive coactivation, when patients attempt to make voluntary movements
(Levin et al. 2000; Musampa et al. 2007). In addition, ranges in which typical patterns
of muscle activation can occur, such as reciprocal activation, have also been described
using the TSRT approach. The use of this fundamental concept of threshold control
based on equilibrium-point theory has led to a new and more in-depth understanding
of the mechanisms underlying motor control deficits in patients with CNS lesions.
Considering these findings, it may be inappropriate to view spasticity independently
from the motor control deficit. Indeed, the concept of threshold control allows us
to account for the spatial structure of deficits in the regulation of muscle activation
and reflexes by indicating where in the biomechanical range, descending control of
movement is mostly affected. Thus, it describes the control deficit in a functional
context. This spatial structure of motor deficits is usually not taken into account in
clinical evaluations and research on spasticity and disordered motor control, which
may explain the heretofore elusive explanation of the relationship between them.
Accounting for the spatial structure of motor deficits may benefit both researchers
and clinicians by advancing the understanding of the mechanisms underlying unim-
paired and impaired motor control. Based on this understanding, clinicians may
more accurately measure these phenomena and establish more effective medical and
physiotherapeutic interventions for their management.
14.2 Corticospinal Origin of Spasticity and Disordered

Motor Control
In the healthy nervous system, cortical descending fibers mainly cross at the pyra-
midal decussation and project to muscles on the opposite side of the body (Kuypers
1985). Uncrossed fibers from the ipsilateral corticospinal tract mainly innervate axial
and proximal muscles (Chen et al. 1997; Harris-Love et al. 2007). Previous studies
using transcranial magnetic stimulation (TMS) in healthy subjects have shown that
the corticospinal system exerts control over STs (Raptis et al. 2010; Sangani et al.
2011; Ilmane et al. 2013). Other crossed and uncrossed descending systems, such as
the reticulospinal, rubrospinal, and tectospinal tracts, have also been shown to reg-
ulate STs in animal models (e.g., Feldman and Orlovsky 1972; Nichols and Steeves
1986). Of particular interest is that the crossed corticospinal tract is involved not
only in the production of intentional movements but also in muscle relaxation in the
entire biomechanical joint range (Raptis et al. 2010) and in anticipatory preparation
of muscles to perturbations (Petersen et al. 2009; Sangani et al. 2011).
limb hemiparesis are shown (S1, S2, S3). The STs are points on the border in shoulder–elbow joint
space beyond which muscles are activated abnormally at rest. Dark shaded areas indicate joint
space where flexors or extensors are activated. Light shaded areas indicated joint spaces in which
only one of the two flexor or extensor muscles is active. (Adapted with permission from Musampa
et al. 2007)
Stroke-related damage to cortical and subcortical systems results in an imbalance

of excitatory and inhibitory activity at different levels of the CNS (Ward 2011). The
presence of a tonic stretch reflex response within the biomechanical range (ST) at
rest in spastic muscles implies that the resting membrane potential of α-motoneurons
innervating the muscle is higher than normal (> − 70mV; i.e., closer to threshold).
This is supported by findings that TSRT excitation and inhibition are altered in mo-
toneuronal pools due to interruption of the inhibitory cortico-reticulospinal pathway
(Lance 1980; Powers et al. 1989) and due to changes in other descending pathways
from the brainstem (e.g., Fedirchuk and Dai 2004).
With respect to voluntary movement, deficits in muscle activation result from
damage to the same descending pathways mentioned above. Indeed, transmission
in cortical (Shimizu et al. 2002; Bütefisch et al. 2003) and spinal (Zehr et al. 2012)
neuronal circuits is impaired after stroke. The uncrossed corticospinal, rubrospinal,
and reticulospinal tracts all could provide alternate routes for motor cortical output
to reach the contralateral spinal cord and may result in the appearance of motor
compensations (Lawrence and Kuypers 1968a, 1968a; Woolsey et al. 1972; Kuypers
1985; Fisher 1992; Fries et al. 1993; Cao et al. 1998; Belhaj-Saif and Cheney 2000).
The motor cortex also sends uncrossed corticospinal projections to the spinal cord
(Colebatch and Gandevia 1989; Kuypers and Brinkman 1970; Nirkko et al. 2001).
Alterations in rubrospinal and reticulospinal pathways influence predominantly the
proximal musculature (Nathan and Smith 1982; Nathan et al. 1996). In addition to
spinal and muscle mechanisms, damage to the corticospinal system leads to an im-
balance in afferent influences ascending to different brain areas, including sensory
and motor cortices in ipsi and contralateral hemispheres (Lindberg et al. 2009). After
stroke, greater ipsilateral corticospinal tract involvement is associated with greater
compensations and poorer recovery (Perez and Cohen 2009) and an imbalance be-
tween projections from both hemispheres due to altered interhemispheric inhibition
(Misawa et al. 2008).
14.2.1 Functional Consequences of Limitation

in Regulation of STs
Motor deficits and abnormal movement synergies after stroke are traditionally
thought to be expressions of excessive reflex activity because of the loss of descend-
ing modulatory control (Burke 1988). However, these physiological mechanisms by
themselves do not explain how movement deficits arise. The concept of ST regulation
allows us to describe how excessive reflex activity may lead to the appearance of
abnormal movement synergies and motor compensations. In other words, it provides
a means to describe how the deficits can be expressed in a spatial frame of reference,
or within the context of movements made by the body. Figure 14.6 is a schematic
diagram of a model of disordered motor control giving rise to abnormal synergy
patterns. Each joint range, shown by horizontal bands, is characterized by spatial
ranges in which active control is possible and where spasticity is present in either
242 M. F. Levin
b
Wrist
Flex Extend
Rf+ M+
Elbow
Flex Extend
Re+ Rf+
Shoulder
Extend Flex
Rf+ M+
Fig. 14.6 Hypothetical model of how disordered control of spatial thresholds (STs) may give rise
to abnormal muscle synergies. When a patient attempts to make a reaching movement a, the ability
to activate muscles depends on the location of the ST within the joint range of each degree of
freedom involved in the movement b. STs are indicated in each joint range as Rf + for flexor
muscles and Re+ for extensor muscles and mark the angles at which spasticity begins. Ranges
of movement required at each joint are indicated by horizontal arrows. For the movement shown
in a, shoulder flexor activation is normal throughout the movement (first horizontal band—black
range), while elbow extensors are unopposed initially (second horizontal band, blue zone) but
coactivation of flexors occurs when Rf + is reached (second horizontal band—right yellow zone)
and wrist muscles are coactivated throughout the movement (third horizontal band, blue zone).
Abnormal interjoint couplings and restricted ranges of regulation of STs may explain limitations in
the kinematic redundancy of the affected limbs
agonist or antagonist muscles. Active control zones are limited in each of the joints
so that when movements are attempted that require one or several joints to move
into spasticity zones, abnormal couplings between joints or abnormal synergies may
arise. These abnormal interjoint couplings and restricted ranges of regulation of
STs may explain limitations in the kinematic redundancy of the affected limbs. An
example is shown in Fig. 14.5 by overlapping spasticity zones in elbow–shoulder
joint space (Musampa et al. 2007). Thus, individuals who have deficits in multiple
ST regulation have a limited number of possible joint combinations with which to
produce different actions. If one extends this concept to more than two joints in the
seven degree of freedom upper limb, then one would have multiple areas within the
Fig. 14.7 Example of limited

a PUSH
redundancy in upper limb
movements in patients with 145°
stroke. a Subjects
165° 20°
isometrically pushed or
pulled a handle against a load 185° 0°
in three different directions.
The initial load of 30 % MVC -20°
was fully or partially removed
in six steps for each load PULL
direction. b Hand trajectories
after unloading for one b PUSH PULL
healthy subject (top panels)
Healthy
and for two patients with
stroke (middle, lower panels).
Spatial dispersions of hand
trajectories were smaller and
more restricted in stroke
compared to healthy subjects.
(Adapted with permission
from Mihaltchev et al. 2005)
Mild stroke
Y position (m)
Moderate stroke
0.02
0.02
X position (m)
3D arm workspace in which there would be restrictions in the number of available

combinations of joint movements.
Another example of limited redundancy in stroke is shown for an upper limb
pushing/pulling task. The subject was asked to resist a load while either pushing
(Fig. 14.7, left panels) or pulling (Fig. 14.7, right panels) against it in three different
directions (see Mihaltchev et al. 2005 for full details). The initial load of 30 % of their
maximal voluntary contraction (MVC) was fully or partially removed in six steps
for each load direction. Figure 14.7 also shows the hand trajectories after unloading
for one healthy subject (Fig. 14.7b, top panels) and for two patients with stroke
(Fig. 14.7b, middle and bottom panels). Compared to the healthy subject, the spatial
dispersions of the hand trajectories were smaller and more restricted in terms of their
variability in most participants with stroke. In this example and others (Reisman and
244 M. F. Levin
Scholz 2003; van Kordelaar et al. 2012), arm movements made by subjects with
stroke are characterized by limitations in the variability of joint rotations, leading to
stereotypical endpoint trajectories, as seen here, which are more marked in people
with more severe sensorimotor impairment.
The presence of stereotypical movement patterns is related to the severity of the
hemiparesis. Brunnstrom (1970) described five stages of recovery from stroke in
which severe pathology is characterized by the presence of hyperactive segmental
reflexes, spasticity and, when present, movement of the limbs only within defined
flexor and/or extensor synergistic patterns, called “abnormal synergy patterns.” In
the upper limb, the abnormal flexor synergy consists of shoulder retraction, upper
arm flexion, abduction and external rotation coupled with elbow flexion, forearm
supination and finger flexion. The extensor synergy includes shoulder extension, ad-
duction, and internal rotation coupled with elbow extension, forearm pronation and
finger extension. Patients at various levels of recovery demonstrate all or some of the
elements of abnormal synergies when they attempt to make voluntary movements.
Movements are often also characterized by compensatory movement patterns such
as excessive shoulder elevation and/or shoulder protraction when attempting to flex
the upper arm (Merdler et al. 2013; Niessen et al., 2008) and excessive trunk dis-
placement when attempting to move the hand away from the body (Cirstea and Levin
2000; Levin et al. 2002). Thus, there is evidence for both limited redundancy in the
joints affected by reduced ST regulation leading to a restricted range of movement
with limited variability and a preserved redundancy in joints not affected by such
limitations leading to the appearance of motor compensations.
Indeed, recovery of upper limb movement is often evidenced by the ability of
system to express more appropriate redundancy: The appearance of fractionated
movements and movements out of pathological synergies due to a decrease in syn-
ergistic coupling between joints and/or by the use of fewer compensatory motor
strategies (Lang and Schieber 2004; Michaelsen et al. 2006).
14.3 Summary
Despite detailed knowledge of the anatomo-physiological changes in descending,

ascending, and spinal pathways after stroke, the relationship between spasticity and
disordered motor control remains elusive. Part of the problem is that these have
been considered to be separate phenomena. However, the role of the corticospinal
tract in regulating STs provides a mechanism by which these motor impairments
can be considered together. An important consideration is that motor deficits are
present when subjects with stroke move or intend to move paretic limbs while most
physiological examinations of spasticity are conducted at rest. In addition, previous
studies have not considered the spatial nature of the impairments when measuring
corticospinal output.
The extent to which the corticospinal tract is involved in ST regulation merits
further investigation. By analyzing motor deficits within the experimentally based
context of ST control, it is possible to go beyond the usual characterization of motor

impairments after CNS injury as separate phenomena. Consideration of the motor
impairment in spasticity as being derived from a common deficit in the descending
regulation of STs at which neuromuscular elements, including reflexes, begin to act,
provides a way to understand the relationships between them and possibly provide
an explanation for the reduced kinematic redundancy in the motor system typically
seen in patients after CNS lesions. Identification of spasticity zones can also be used
to monitor improvements in patient status due to pharmacological, medical, and
physical treatment interventions.
References
Archambault PS, Mihatchev P, Levin MF, Feldman AG. 2005. Basic elements of arm postural
control analysed by unloading. Exp Brain Res 164:225–241.
Asatryan DG, Fel’dman AG. 1965. Functional tuning of the nervous system with control of move-
ment or maintenance of a steady posture—I. Mechanographic analysis of the work of the joint
during the performance of a postural task. Biophysics 10:925–935 (Engl translation of Biofizika
10:837–846).
Beer RF, Dewald JPA, Rymer WZ. 2000. Deficits in the coordination of multi-joint arm movements
in patients with hemiparesis: evidence for disturbed control of limb dynamics. Exp Brain Res
131: 305–319.
Belhaj-Saif A, Cheney PD. 2000. Plasticity in the distribution of the red nucleus output to forearm
muscles after unilateral lesions of the pyramidal tract. J Neurophysiol 83:3147–3153.
Bonasera SJ, Nichols TR. 1994. Mechanical actions of heterogenic reflexes linking long toe flexors
with ankle and knee extensors of the cat hindlimb. J Neurophysiol 71(3):1096–1110.
Bourbonnais D, Vanden Noven S, Carey KM, Rymer WZ. 1989 Abnormal spatial patterns of elbow
muscle activation in hemiparetic human subjects. Brain 112:85–102.
Brunnström S. Movement therapy in hemiplegia: a neurophysiological approach. NewYork: Harper
and Row; 1970.
Burke D. 1988. Spasticity as an adaptation to pyramical tract injury. Adv Neurol 47:401–423.
Burne JA, Carleton VL, O’Dwyer NJ. 2005. The spasticity paradox: movement disorder or disorder
of resting limbs? J Neurol, Neurosurg Psychiatry 76:47–54.
Bütefisch CM, Netz J, Wessling M, Seitz RJ, Hömberg V. 2003. Remote changes in cortical
excitability after stroke. Brain 126:470–481.
Calota A, Levin MF. 2009. Tonic stretch reflex threshold as a measure of spasticity: implications
for clinical practice. Top Stroke Rehabil 16(3):177–188.
CaoY, D’Olhaberriague L, Vikingstad EM, Levine SR, Welch KMA. 1998. Pilot study of functional
MRI to assess cerebral activation of motor function after poststroke hemiparesis. Stroke 29:
112–122.
Capaday C. 1995. The effects of baclofen on the stretch reflex parameters of the cat. Exp Brain Res
104:287–296.
Cavallari P, Katz R. 1989. Pattern of projections of group I afferents from forearm muscles to
motoneurones supplying biceps and triceps muscles in man. Exp Brain Res 78(3):465–478.
Cavallari P, Katz R, Penicaud A. 1992. Pattern of projections of group I afferents from elbow
muscles to motoneurones supplying wrist muscles in man. Exp Brain Res 91(2):311–319.
Chang SH, Francisco GE, Zhou P, Rymer WZ, Li S. 2013. Spasticity, weakness, force variability,
and sustained spontaneous motor unit discharges of resting spastic-paretic biceps brachii muscles
in chronic stroke. Muscle Nerve 48(1):85–92.
246 M. F. Levin
Chen R, Cohen LG, Hallett M. 1997. Role of the ipsilateral motor cortex in voluntary movement.
Can J Neurol Sci 24(4):284–288.
Cirstea MC, Levin MF. 2000. Compensatory strategies for reaching in stroke. Brain 123:940–953.
Colebatch JG, Gandevia SC. 1989. The distribution of muscular weakness in upper motor neuron
lesions affecting the arm. Brain 112:749–763.
Corcos DM, Gottlieb GL, Penn RD, Myklebust B, Agarwal GC. 1986. Movement deficits caused
by hyperexcitable stretch reflexes in spastic humans. Brain 109:1043–1058.
Créange A, Faist M, Katz R, Penicaud A. 1992. Distribution of heteronymous Ia facilitation and
recurrent inhibition in the human deltoid motor nucleus. Exp Brain Res 90:620–624.
Dewald JP, Pope PS, Given JD, Buchanan TS, Rymer WZ. 1995. Abnormal muscle coactivation
patterns during isometric torque generation at the elbow and shoulder in hemiparetic subjects.
Brain 118:495–510.
Elovic EP, Simone LK, Zafonte R. 2004. Outcome assessment for spasticity management in the
patient with traumatic brain injury: the state of the art. J Head Trauma Rehabil 19:155–177.
Fedirchuk B, DaiY. 2004. Monoamines increase the excitability of spinal neurons in the neonatal rat
by hyperpolarizing the threshold for action potential production. J Physiol 557(Pt 2):355–361.
Feldman AG. 2009. New insights into action-perception coupling. Exp Brain Res 194:39–58.
Feldman AG. 2011. Space and time in the context of equilibrium-point theory. Wiley Interdisci-
plinary Reviews. Cognitive Sci 2:287–304.
Feldman AG, Levin MF. 1995. The origin and use of positional frames of reference in motor control.
Behav Brain Sci 18:723–806.
Feldman AG, Orlovsky GN. 1972. The influence of different descending systems on the tonic stretch
reflex in the cat. Exp Neurol 37:481–494.
Fellows SJ, Kaus C, Ross HF, Thilmann AF. 1994a. Agonist and antagonist EMG activation dur-
ing isometric torque development at the elbow in spastic hemiparesis. Electroencephal Clin
Neurophysiol 93:106–112.
Fellows SJ, Kaus C, Thilmann AF. 1994b. Voluntary movement at the elbow in spastic hemiparesis.
Ann Neurol 36:397–407.
Fisher CM. 1992. Concerning the mechanism of recovery in stroke hemiplegia. Can J Neurol Sci
19:57–63.
Fries W, Danek A, Scheidtmann K, Hamburger C. 1993. Motor recovery following capsular stroke:
role of descending pathways from multiple motor areas. Brain 116:369–382.
Fugl-Meyer AR, Jaasko L, Leyman I, Olsson S, Steglind S. 1975. The post-stroke hemiplegic
patient. 1. A method for evaluation of physical performance. Scand J Rehabil Med 7(1):13–31.
Gemperline JJ, Allen S, Walk D, Rymer WZ. 1995. Characteristics of motor unit discharge in
subjects with hemiparesis. Muscle Nerve 18:1101–1114.
Ginanneschi F, Dominici F, Biasella A, Gelli F, Rossi A. 2006. Changes in corticomotor excitability
of forearm muscles in relation to static shoulder positions. Brain Res 1073–1074:332–338.
Gowland C, deBruin H, Basmajian JV, Plews N, Burcea I. 1992. Agonist and antagonist activity
during voluntary upper-limb movement in patients with stroke. Phys Ther 72:624–633.
Gracies JM, Meunier S, Pierrot-Deseilligny E, Simonetta M. 1991. Pattern of propriospinal-like
excitation to different species of human upper limb motoneurones. J Physiol 434:151–167.
Hammond MC, Fitts SS, Kraft GH, Nutter PB, Trotter MJ, Robinson LM. 1988. Co-contraction in
the hemiparetic forearm: quantitative EMG evaluation. Arch Phys Med Rehabil 69:348–351.
Harris-Love ML, Perez MA, Chen R, Cohen LG. 2007. Interhemispheric inhibition in distal and
proximal arm representations in the primary motor cortex. J Neurophysiol 97(3):2511–2515.
Hoffman DS, Strick PL. 1995. Effects of a primary motor cortex lesion on step-tracking movements
of the wrist. J Neurophysiol 73:891–895.
Hultborn H, Kiehn O. 1992. Neuromodulation of vertebrate motor neuron membrane properties.
Curr Opp Neurobiol 2:770–775.
Hultborn H, Meunier S, Pierrot-Deseilligny E, Shindo M. 1987. Changes in presynaptic inhibition
of Ia fibres at the onset of voluntary contraction in man. J Physiol 389:757–772.
Ilmane N, Sangani S, Feldman AG. 2013. Corticospinal control strategies underlying voluntary and
involuntary wrist movements. Beh Brain Res 236:350–358.
Jobin A, Levin MF. 2000. Regulation of stretch reflex threshold in elbow flexors in children with
cerebral palsy: a new measure of spasticity. Dev Med Child Neurol 42:531–540.
Kamper DG, Rymer WZ. 2001. Impairment of voluntary control of finger motion following stroke:
role of inappropriate muscle coactivation. Muscle Nerve 24:673–681.
Kamper DG, Fischer HC, Cruz EG, Rymer WZ. 2006. Weakness is the primary contributor to finger
impairment in chronic stroke. Arch Phys Med Rehabil 87:1262.
Katz R, Penicaud A, Rossi A. 1991. Reciprocal Ia inhibition between elbow flexors and extensors
in the human. J Physiol 437:269–286.
Kim WH, Park EY. 2011. Causal relation between spasticity, strength, gross motor function, and
functional outcome in children with cerebral palsy: a path analysis. Dev Med Child Neurol
53:68–73.
Krakauer JW. 2005. Arm function after stroke: From physiology to recovery. Semin Neurol
25(4):384–394.
Kuypers HGJM. 1985. The anatomical and functional organization of the motor system. In: Swash
M, Kennard C, editors. Scientific basis of clinical neurology. London: Churchill Livingstone;
1985. pp. 3–18.
Kuypers HGJM, Brinkman J. 1970. Precentral projections to different parts of the spinal
intermediate zone in the rhesus monkey. Brain Res 24:29–48.
Lance JW. 1980. The control of muscle tone, reflexes and movement. Robert Wartenberg Lecture.
Neurol 30:309–326.
Lang CE, Schieber MH. 2004. Reduced muscle selectively during individuated finger movements in
humans after damage to the motor cortex or corticospinal tract. J Neurophysiol 91:1722–1733.
Lawrence DG, Kuypers HGJM. 1968a. The functional organization of the motor system in the
monkey. I. The effects of bilateral pyramidal lesions. Brain 19:1–14.
Lawrence DG, Kuypers HGJM. 1968b. The functional organization of the motor system in the
monkey. II. The effects of lesions of the descending brain-stem pathways. Brain 19:15–36.
Lee WA, Boughton A, Rymer WZ. 1987. Absence of stretch gain enhancement in voluntarily
activated spastic muscle. Exp Neurol 98: 317–335.
Levin MF, Dimov M. 1997. Spatial zones for muscle coactivation and the control of postural
stability. Brain Res 757:43–59.
Levin MF, Feldman AG. 1994. The role of stretch reflex threshold regulation in normal and impaired
motor control. Brain Res 657:23–30.
Levin MF, Selles RW, Verheul MH, Meijer OG. 2000. Deficits in the coordination of agonist
and antagonist muscles in stroke patients: implications for normal motor control. Brain Res
853:352–369.
Levin MF, Michaelsen S, Cirstea C, Roby-Brami A. 2002. Use of the trunk for reaching targets
placed within and beyond the reach in adult hemiparesis. Exp Brain Res 143:171–180.
Lindberg PG, Gaverth J, Fagergren A, Fransson P, Forssberg H, Borg J. 2009. Cortical activity
in relation to velocity dependent movement resistance in the flexor muscles of the hand after
stroke. Neurorehabil Neural Repair 23(8):800–810.
Lundberg A. 1967. The supraspinal control of transmission in spinal reflex pathways. Electroenceph
Clin Neurophysiol, Suppl 25:35–46.
Malhotra S, Cousins E, Ward A, Day C, Jones P, Roffe C, Pandyan A. 2008. An investigation into
the agreement between clinical, biomechanical and neurophysiological measures of spasticity.
Clin Rehabil 22(12):1105–1115.
Malhotra S, Pandyan AD, Day CR, Jones PW, Hermens H. 2009. Spasticity, an impairment that is
poorly defined and poorly measured. Clin Rehabil 23(7):651–658.
Marchand-Pauvert V, Mazevet D, Nielsen J, Petersen N, Pierrot-Deseilligny E. 2000. Distribution
of non-monosynaptic excitation to early and late recruited units in human forearm muscles. Exp
Brain Res 134(2):274–278.
248 M. F. Levin
Matthews PB. 1959. A study of certain factors influencing the stretch reflex of the decerebrate cat.
J Physiol 147:547–564.
Matthews PB. 1993. Interaction between short- and long-latency components of the human stretch
reflex during sinusoidal stretching. J Physiol 462:503–527.
Mazevet D, Meunier S, Pradat-Diehl P, Marchand-Pauvert V, Pierrot-Deseilligny E. 2003. Changes
in propriospinally mediated excitation of upper limb motoneurons in stroke patients. Brain
126:988–1000.
McClelland VM, Miller S, Eyre JA. 2001. Short latency heteronymous excitatory and inhibitory
reflexes between antagonist and heteronymous muscles of the human shoulder and upper limb.
Brain Res 299:82–93.
McPherson JG, Ellis MD, Heckman CJ, Dewald JPA. 2008. Evidence for increased activation
of persistent inward currents in individuals with chronic hemiparetic stroke. J Neurophysiol
100(6):3236–3243.
Merdler T, Liebermann DG, Levin ML, Berman S. 2013. Arm-plane representation of shoulder
compensation during pointing movements in subjects with stroke. J Electromyogr Kinesiol
21:938–947.
Meunier S, Pierrot-Deseilligny E. 1998. Cortical control of presynaptic inhibition of Ia afferents in
humans. Exp Brain Res 119:415–426.
Michaelsen SM, Dannenbaum R, Levin MF. 2006. Task-specific training with trunk restraint on
arm recovery in stroke. Randomized control trial. Stroke 37:186–192.
Mihaltchev P, Archambault PS, Feldman AG, Levin MF. 2005. Control of double-joint arm posture
in adults with unilateral brain damage. Exp Brain Res 163:468–486.
Misawa S, Kuwabara S, Matsuda S, Honma K, Ono J, Hattori T. 2008. The ipsilateral cortico-spinal
tract is activated after hemiparetic stroke. Eur J Neurol 15(7):706–711.
Mullick AA, Musampa NK, Feldman AG, Levin MF. 2013. Stretch reflex threshold measure
discriminates between spasticity and rigidity. Clin Neurophysiol 124(4):740–751.
Musampa NK, Mathieu PA, Levin MF. 2007. Relationship between stretch reflex thresholds and
voluntary arm muscle activation in patients with spasticity. Exp Brain Res 181:579–593.
Nathan PW, Smith MC. 1982. The rubrospinal and central tegmental tracts in man. Brain 105:
223–269.
Nathan PW, Smith MC, Deacon P. 1996. Vestibulospinal, reticulospinal and descending pro-
priospinal nerve fibres in man. Brain 119:1809–1833.
Nielsen JB, Crone C, Hultborn H. 2007. The spinal pathophysiology of spasticity—from a basic
science point of view. Acta Physiol 189(2):171–180.
Niessen M, Janssen T, Meskers C, Koppe P, Konijnenbelt M, Veeger D. 2008. Kinematics of the
contralateral and ipsilateral shoulder: A possible relationship with post-stroke shoulder pain. J
Rehabil Med 40(6):482–486.
Nichols TR, Steeves JD. 1986. Resetting of resultant stiffness in ankle flexor and extensor muscles
in the decerebrate cat. Exp Brain Res 62:401–410.
Nirkko AC, Ozdoba C, Redmond SM, Burki M, Schroth G, Hess CW, Wiesendanger M. 2001.
Different ipsilateral representations for distal and proximal movements in the sensorimotor
cortex: activation and deactivation patterns. Neuroimage 13:825–835.
O’Dwyer NJ, Ada L, Neilson PD. 1996. Spasticity and muscle contracture following stroke. Brain
119:1737–1749.
Perez MA, Cohen LG. 2009. The corticospinal system and transcranial magnetic stimulation in
stroke. Topics in Stroke Rehabil 16:254–269.
Petersen TH, Rosenberg K, Petersen NC, Nielsen JB. 2009. Cortical involvement in anticipatory
postural reactions in man. Exp Brain Res 193(2):161–171.
Powers RK, Campbell DL, Rymer WZ. 1989. Stretch reflex dynamics in spastic elbow flexor
muscles. Ann Neurol 25(1):32–42.
Raptis H, Burtet L, Forget R, Feldman AG. 2010. Control of wrist position and muscle relaxation
by shifting spatial frames of reference for motoneuronal recruitment: possible involvement of
corticospinal pathways. J Physiol 588:1551–1570.
Reisman DS, Scholz JP. 2003. Aspects of joint coordination are preserved during pointing in persons
with post-stroke hemiparesis. Brain 126:2510–2527.
Roberts LV, Stinear CM, Lewis GN, Byblow WD. 2008. Task-dependent modulation of pro-
priospinal inputs to human shoulder. J Neurophysiol 100:2109–2114.
Sahrmann SA, Norton BJ. 1977. The relationship of voluntary movement to spasticity in the upper
motor neuron syndrome. Ann Neurol 2:460–465.
Sangani SG, Raptis HA, Feldman AG. 2011. Subthreshold corticospinal control of anticipatory
actions in humans. Beh Brain Res 224(1):145–154.
Shimizu T, Hosaki A, Hino T, Sato M, Komori T, Hirai S, Rossini. PM. 2002. Motor cortical
disinhibition in the unaffected hemisphere after unilateral cortical stroke. Brain 125:1896–1907.
Sunnerhagen KS, Oliver J, Francisco GE. 2013. Assessing and treating functional impairment in
poststroke spasticity. Neurology 80:S35–S44.
Ter Haar Romeny BM, Denier van der Gon JJ, Gielen CCAM. 1984. Relation between location
of a motor unit in the human biceps brachii and its critical firing levels for different tasks. Exp
Neurol 85:631–650.
Thilmann AF, Schwarz M, Topper R, Fellows SJ, North J. 1991. Different mechanisms underlie
the long-latency stretch reflex response of active human muscle at different joints. J Physiol
444:631–643.
Van Kordelaar J, van Wegen EEH, Kwakkel G. 2012. Unraveling the interaction between patho-
logical upper limb synergies and compensatory trunk movements during reach-to-grasp after
stroke: a cross-sectional study. Exp Brain Res 221:251–262.
Van Zuylen EJ, Gielen CCAM, Denier Van Der Gon JJ. 1988. Coordination and in- homogeneous
activation of human arm muscles during isometric torques. J Neurophysiol 60:1523–1548.
Von Holst H. 1954. Relations between the central nervous system and the peripheral organs. Br J
Animal Behav 2:89–94.
Ward AB. 2011. A literature review of the pathophysiology and onset of post-stroke spasticity. Eur
J Neurol 19:21–27.
Wood DE, Burridge JH, Van Wijck FM, et al. 2005. Biomechanical approaches applied to the lower
and upper limb for the measurement of spasticity: A systematic review of the literature. Disabil
Rehab 27(1–2):19–32.
Woolsey CN, Gorska T, Wetzel A, Erickson TC, Earls FJ, Allman JM. 1972. Complete unilateral
section of the pyramidal tract at the medullary level in the macaca mulatta. Brain Res 40:119–123.
Zehr EP, Loadman PM, Hundza SR. 2012. Neural control of rhythmic arm cycling after stroke. J
Neurophysiol 108:891–905.
Zhang LQ, Chung SG, Ren Y, Liu L, Roth EJ, Rymer WZ. 2013. Simultaneous characterization
of reflex and nonreflex dynamic and static changes in spasticity hemiparesis. J Neurophysiol
110:418–430.
Chapter 15
Enhancing Postural Stability and Adaptability
in Multiple Sclerosis
Richard E. A. van Emmerik, Stephanie L. Jones, Michael A. Busa,

Jebb G. Remelius and Julianna L. Averill
15.1 Introduction
Multiple sclerosis (MS) is a chronic demyelinating disease of the central nervous sys-
tem and is the most common neurological disease in 20 to 50 year-old individuals.
People living with MS consistently rate balance and walking impairments as having
the greatest negative impacts on their quality of life, yet our understanding of the
cause of instability in these individuals remains limited (Ford et al. 2001; Hemmett
et al. 2004). A growing body of research aims to understand the sensorimotor contri-
butions to balance dysfunction and difficulty with walking in people with MS. Our
own research on MS aims to investigate how fatigue, muscle weakness, and sensory
loss interact to limit physical function and mobility (see Fig. 15.1). In this chapter,
we relate aspects of somatosensory loss and symptomatic fatigue to balance function,
and provide new insights into the mechanisms of balance and gait dysfunction in MS,
through the use of novel analytical methods and experimental paradigms. We first re-
view existing methods and paradigms to assess postural and gait stability in research
on MS. Next, we introduce novel measures to assess the stability and adaptability of
posture and gait in people with MS that are based on nonlinear and complex systems
methods. These novel methods include (1) boundary-relevant measures of postural
stability and control (postural “time to contact”) and (2) entropy measures for as-
sessing postural and gait adaptability that emphasize the functional role of variability
in motor control. We also discuss how these methods and paradigms may help to
develop innovative treatments for balance and gait dysfunction in people with MS.
R. E. A. van Emmerik () · S. L. Jones · M. A. Busa · J. G. Remelius · J. L. Averill

Department of Kinesiology, University of Massachusetts,
30 Eastman Lane, Amherst, MA, USA

and Biology 826, DOI 10.1007/978-1-4939-1338-1_15
252 R. E. A. van Emmerik et al.
Fig. 15.1 Contribution of different subsystems to postural instability and mobility problems in
multiple sclerosis (MS)
15.2 Changes in the Control of Posture and Locomotion in MS
Among the most devastating effects of MS is impairment of balance and mobility.

Reports of balance dysfunction in individuals with MS have reached 90 % by some
estimates, and roughly three-quarters of people with MS report balance impairments
as having a moderate-to-severe impact on their daily functions (Ford et al. 2001;
Hemmett et al. 2004). The pervasive effects of demyelination in MS can lead to
complex and multifactorial impacts on balance function involving all aspects of
sensorimotor function (see Fig. 15.1). Understanding the relative contributions and
interactions between these systems will provide vital information with which to
identify targets and modalities for interventions to slow or stall the progress of the
disease-related impairments.
15 Enhancing Postural Stability and Adaptability in Multiple Sclerosis 253
15.2.1 Balance Impairment in MS
The majority of individuals with MS report balance and/or mobility impairments that
often worsen as the disease progresses. It has been reported that 50 % of individuals
with MS will require ambulatory assistance within 15 years of their initial diagnosis
(Weinshenker et al. 1989), providing a significant social and economic burden (Pike
et al. 2012). Roughly half of those with MS reported falls within a 3 to 6 month pe-
riod (Finlayson et al. 2006; Nilsagard et al. 2009), and despite having no increased
prevalence of falls among those using ambulatory aids (Coote et al. 2013), use of
these aids has been identified as a significant risk factor for falls (Cattaneo et al.
2002; Nilsagard et al. 2009). Contributors to fall risk are considered multifactorial,
as fall risk has also been linked to fatigue (Coote et al. 2013), increased fear of falling
(Finlayson et al. 2006), spasticity (Nilsagard et al. 2009), and impaired propriocep-
tion (Nilsagard et al. 2009). However, equivocal results on clinical balance and gait
tests have been reported (Cattaneo et al. 2002; Coote et al. 2013; Nilsagard et al.
2009) that may reflect the significant heterogeneity of the MS population. In addi-
tion, it is plausible that many of the clinical tests used to assess balance and mobility
in this population may not be sensitive enough to quantify the decrements that can
be assessed using sophisticated, laboratory-based measures (Findling et al. 2011;
Martin et al. 2006). Balance decrements, as indicated by increasing postural sway,
have been shown to worsen with increased disability and progressive MS subtypes
(Boes et al. 2012; Corporaal et al. 2013; Fjeldstad et al. 2011; Gutierrez et al. 2005;
Soyuer et al. 2006), lending further support to heterogeneity as the source of the
equivocal findings regarding balance and gait impairments. This heterogeneity may
relate to the multifactorial etiology of balance dysfunction in individuals with MS
that can result from a combination of underlying impairments of sensory and motor
functioning.
Laboratory-based balance assessments have shown that people with MS demon-
strate diminished performance on timed balance tasks (Frzovic et al. 2000; Soyuer
et al. 2006), increased center of pressure (CoP) velocity (Boes et al. 2012; Porosinska
et al. 2010), and increased CoP or trunk sway during quiet standing and reach-
ing/leaning tasks (Boes et al. 2012; Corporaal et al. 2013; Findling et al. 2011;
Porosinska et al. 2010; Spain et al. 2012; Van Emmerik et al. 2010). Balance in MS
has also been shown to worsen with increasing difficulty of the task, either through
base of support restrictions (Frzovic et al. 2000), reaching and leaning (Frzovic et al.
2000; Karst et al. 2005; Van Emmerik et al. 2010), dual tasks (Negahban et al. 2011),
or altered sensory conditions (Cattaneo and Jonsdottir 2009; Fjeldstad et al. 2009;
Porosinska et al. 2010; Spain et al. 2012).
Individuals with MS, particularly those with vestibular or proprioceptive impair-
ment, demonstrate increased reliance on the visual system during balance tasks
(Soyuer et al. 2006; Van Emmerik et al. 2010), since removal of vision has one of
the most profound effects on balance performance during standing postural tasks.
This compensation may be necessary, given the high prevalence of somatosensory
impairments demonstrated by individuals with MS (Citaker et al. 2011; Kelleher
et al. 2010; Soyuer et al. 2006; Van Emmerik et al. 2010). Indeed, balance re-
sponses demonstrated by those with MS show alterations consistent with an etiology
of somatosensory impairment (Cameron et al. 2008; Corporaal et al. 2013). Postu-
ral responses to unexpected balance disruption are significantly delayed, resulting
in hypermetric responses to successfully maintain balance (Cameron et al. 2008).
These delays correlate with delays in spinal transmission of somatosensory-evoked
potentials, suggesting that individuals with MS may have impaired transmission,
and/or processing of somatosensory information. In addition, strong relationships
between impaired somatosensation and balance ability in people with MS have been
reported (Citaker et al. 2011; Corporaal et al. 2013).
15.2.2 Motor Impairments, Strength, and Balance in MS
Deficits in somatosensory function may require greater control from higher brain
centers in those with MS. Enhanced predictive scaling of postural response mag-
nitudes, to unexpected perturbations when the perturbation amplitude is known
a priori (Cameron et al. 2008), has been demonstrated among individuals with
MS, which may be indicative of increased reliance on the cerebellum and corti-
cal structures. Feedforward control of posture is also disrupted in those with MS,
possibly reflecting the impact of MS on balance mediated by higher brain centers
(Krishnan et al. 2012b). Delayed onset of anticipatory postural adjustments (APAs)
(Jacobs and Kasser 2012a; Krishnan et al. 2012a), as well as reduced magnitude
of the APA and ability to move the center of mass (CoM) to its stability limits
(Krishnan et al. 2012a), have been identified in those with MS. These delays are
exacerbated under dual task conditions (Jacobs and Kasser 2012b), further support-
ing a greater involvement of the higher brain centers in control of APAs in those
with MS.
Despite these severely delayed postural responses, the magnitude of postural ad-
justments is not reduced in MS, suggesting little impairment in the efferent pathways
that compose balance responses (Cameron et al. 2008). This is puzzling, given ev-
idence that individuals with MS demonstrate strength reductions (De Haan et al.
2000; Newsome et al. 2011; Rice et al. 1992) that can be traced to both peripheral
and central deficits (Newsome et al. 2011; Ng et al. 2004). Most studies demonstrate
impaired strength in those with MS, although this may be more evident in dynamic
rather than isometric paradigms (Chung et al. 2008). It has been suggested that these
strength differences, particularly those that correlate with walking or balance impair-
ments, may be related to central impairments in motor drive (Chung et al. 2008; De
Haan et al. 2000; Ng et al. 1997, 2004) as demonstrated by reduced central activation
ratio and reduced ability to perform rapid foot taps. Although peripheral motor im-
pairments in the form of reduced relaxation time (Ng et al. 2004), slower contraction
speed (Ng et al. 2004), and perhaps lower oxidative capacity (Kent-Braun et al. 1994)
and atrophy (Kent-Braun et al. 1997) have been identified in those with MS, these
may be secondary to reduced central activation leading to decreased motor neuron
discharge (Kent-Braun et al. 1997; Ng et al. 2004). Importantly, there appears to be a

link between motor impairments and balance performance, given that a greater knee
joint power asymmetry was related to reduced balance performance during quiet
standing (Chung et al. 2008).
15.2.3 Relationship Between Fatigue and Balance

and Mobility in MS
The above-described motor impairments may be exacerbated by both central and

peripheral fatigue. Fatigue experienced by those with MS, termed “symptomatic
fatigue,” differs from what is experienced by healthy adults and has been defined as
“reversible motor and cognitive impairment with reduced motivation and desire to
rest, either appearing spontaneously or brought on by mental or physical activity,
humidity, acute infections, and food ingestion. It is relieved by daytime sleep or rest
without sleep. It can occur at any time, but is usually worse in the afternoon” (Mills
and Young 2008). Fatigue has not been shown to correlate with measures of central
activation or other measures of muscle function (Ng et al. 2004).
Symptomatic fatigue in those with MS has been correlated with perceived walk-
ing impairment and balance confidence (Noguiera et al. 2013), and with a decline
in walking speed during the 6- and 12-min walk tests (Burschka et al. 2012). These
declines occurred at a greater rate in those with moderate MS compared to healthy
controls. Symptomatic fatigue has also been negatively correlated to performance on
standing balance assessments. Both acute (Visual Analog Fatigue Scale) and chronic
(Fatigue Severity Scale) measures of symptomatic fatigue correlate modestly with
increased postural sway during quiet standing in those with MS (Chung et al. 2008).
However, symptomatic fatigue appears to have a stronger relationship with balance
as postural challenge increases. Specifically, the Modified Fatigue Impact Scale has
strong negative correlations for both the physical and cognitive subscales with bal-
ance as assessed by the sensory organization test (posturography) (Hebert and Corboy
2013), particularly during tasks in which somatosensory feedback was nullified and
vision manipulated. Indeed, 62 % of the variance in balance scores was explained
by symptomatic fatigue, pointing to a significant relationship between fatigue and
balance. Standing balance assessment, following a fatiguing protocol, supports a
greater effect on balance in those with MS as task difficulty increases (Van Emmerik
et al. 2010). Individuals with MS demonstrate reduced forward lean performance
and decreased postural stability in backward leans, but no effect on quiet standing
following fatiguing protocols (Van Emmerik et al. 2010). These findings suggest
that symptomatic fatigue has the potential to greatly impact balance in those with
MS and may require functional adaptations to operate safely within the environ-
ment, particularly as fatigue heightens throughout the day and following physical
exertion.
15.2.4 Altered Walking Patterns in MS
To combat reduced balance ability that is exacerbated by fatigue, individuals with

MS may rely on proactive behavior during voluntary movements to compensate for
peripheral sensory and motor impairments. Individuals with MS walk in a more cau-
tious way, typified by slowed walking velocity, shorter stride lengths, increased stride
width, and a prolonged double-support phase (Benedetti et al. 1999; Gianfrancesco
et al. 2011; Remelius et al. 2012; Sosnoff et al. 2012). These stride changes are
accompanied by decreased ankle, knee, hip, and trunk ranges of motion (Corporaal
et al. 2013; Huisinga et al. 2013; Martin et al. 2006) in the sagittal plane and increased
trunk range of motion in the frontal and transverse planes, with the frontal plane be-
ing the most discriminative between individuals with and without MS (Spain et al.
2012). Kinetic alterations also provide evidence of a more cautious walking pattern
with reduced frequencies of vertical forces (indicating a reduced vertical oscillation
and perhaps a less dynamic walking pattern) (Wurdeman et al. 2011) and reduced
torques at the lower limbs, particularly when torque would contribute to propelling
the body forward (e.g., reduced dorsiflexor torque during early stance and plantar
flexor torque during late stance) (Huisinga et al. 2013). Increased co-contraction
of ankle musculature through the stance phase, accomplished through early onset
of gastrocnemius and late relaxation of the tibialis anterior muscles, has also been
observed and may act to stiffen the ankle joint during the stance phase (Benedetti
et al. 1999).
While these gait alterations are widely reported, the majority of these research
studies have made comparisons between individuals with and without MS in which
the groups differed in their walking speeds. Thus, the altered gait parameters reported
could represent speed-related group differences as opposed to alterations that were
related to MS. Recent studies have supported that at least some of these parameters
appear to be different between MS and non-MS independent of walking speed;
specifically, increased stride width and prolonged double-support phase have been
identified even when MS and control cohorts walk at matched speeds (Remelius et al.
2012). Kinematics, kinetics, and muscle activation have yet to be studied between
those with and without MS at matched speeds. However, when using walking speed
as a covariate, reduced joint ranges of motion and torque appear to be MS-related
gait characteristics (Huisinga et al. 2013).
Gait modifications that are characteristic of a protective strategy may reflect an
attempt to increase stability. Wider strides enable the walker to maintain a larger base
of support, and increased double support time could be an attempt to prolong foot
contact with the ground, both of which could reflect an attempt to increase walking
stability. Prolonged foot contact would also allow for increased peripheral sensory
feedback from the foot-surface interface, which may be of vital importance in those
with MS who have reduced cutaneous sensitivity. Individuals with MS also alter
their body position by leading with their trunk and head, as opposed to their swing
foot, particularly at higher walking speeds, which may represent an attempt to gain
further sensory information by enhancing visualization of step placement (Remelius
et al. 2012). Although these strategies may be an attempt to enhance stability of the
movement, they may precipitate detrimental consequences that could instead increase
the likelihood of balance loss. Prolonging double support time requires a shortened
swing time if walking speed is to be maintained; however, hastening swing time
may prove destabilizing, potentially increasing the likelihood of balance loss during
the single support phase of walking (Remelius et al. 2012). In addition, attempting
to gain more visual information during walking may necessitate the more forward
directed head/trunk position demonstrated by individuals with MS (Remelius et al.
2012). However, this body position could prove to be less stable, given that the body
CoM is closer to the forward stability boundary, providing less time to respond to
unexpected balance perturbations.
15.3 Boundary-Relevant Measures of Posture and Gait:

Time to contact
We now turn to a discussion of boundary-relevant measures of postural stability and

control (postural “time to contact”) that have emerged from complex and nonlin-
ear systems approaches. A central element in assessing stability and adaptability is
the relation between essential movement variables such as body CoM and stability
boundaries formed by the base of support at the ground. It is anticipated that this
approach will offer novel insights into MS-related changes in static and dynamic
postural control and help determine whether these changes represent a functional
adaptation to maintain stability or reflect MS-related limitations.
15.3.1 Stability Boundaries and Postural Fluctuations
Postural instability in MS during upright standing is often associated with increased

CoP standard deviation, sway area, or average velocity (Chung et al. 2008; Fjeldstad
et al. 2011; Van Emmerik et al. 2010). However, as we and others have argued,
postural fluctuations can provide functional information regarding (1) geometry of
support surface, (2) state of disequilibrium, and (3) approach to boundaries (Riccio
1993). From this perspective, postural fluctuations can be an expression of both
exploratory and performatory aspects of postural sway (Riley et al. 1997; Slobounov
et al. 1997; Van Emmerik and Van Wegen 2002).
Standing balance is maintained when the CoM of the body remains within the
boundaries of the base of support at the foot–ground interface (Fig. 15.2a). While
in standing balance this base of support is fixed, under dynamic conditions, such as
locomotion, these boundaries continually shift (Fig. 15.2b). Fluctuations in the foot
CoP during upright stance control the CoM and stabilize the body under different
postural conditions (Winter 1995). To quantify dynamic balance with respect to
these boundaries, the time to contact the stability boundary can be obtained by
Fig. 15.2 Static (a) and dynamic (b) base of support
assessing the instantaneous distance, velocity, and/or acceleration of the CoP or

CoM (Hasson et al. 2008, 2009; Remelius et al. 2008; Van Wegen et al. 2001,
2002). Research on CoP and CoM control, relative to these stability boundaries,
has provided novel insights regarding the role of postural fluctuations. For example,
postural control research in Parkinson’s disease has shown that the time-to-contact
variability of the CoP to the stability boundary at the feet is lower in people with
Parkinson’s disease compared to controls (Van Wegen et al. 2001; Van Emmerik
et al. 2013). In contrast, CoP displacement variability was larger in the Parkinson
group, a result consistent with previous findings in the literature. These postural
fluctuations in relation to the stability boundary, as measured by time to contact,

may facilitate the pickup of perceptual information regarding boundaries between
stable and unstable postures and therefore enhance postural stability and adaptations
in response to changing conditions. Time to contact the postural support boundaries
may therefore be a sensitive measure in assessing postural stability.
The original time-to-contact concept emerged from vision research that sought
to characterize the visual perception of a distance closure (gap closure) as a directly
perceivable temporal variable called tau (τ) (Lee and Lishman 1977; Lee 2009).
Lee’s general tau theory states that: (1) all purposeful (prospective) movements entail
controlling the closure of action gaps (between effectors and goals), (2) the time to
closure of the action-gap (at current rate), tau, is what is sensed and controlled
and directly perceivable in any sensory modality, and (3) the principal method of
movement guidance is coupling and coordinating tau’s of different action gaps. In the
remainder of this section, we will address all three aspects of Lee’s General Tau theory
by examining time to contact in the postural and locomotor control domains. This was
originally developed by Riccio (1993) and formalized by Slobounov et al. (1997).
15.3.2 Boundary-Relevant Analysis of Postural Control
Research assessing spatial limits with respect to the base of support has been in
existence for some time (e.g., McCollum and Leen 1989), and work by Said et al.
(2008) assessed the spatial separation of the CoM and CoP in relation to the stance
foot–heel boundary during walking over obstacles. Pai and Patton (1997) and Espy
et al. (2010) used a phase space (position–velocity) representation as a “dynamic”
boundary to identify regions of initiation for forward and backward falls during
upright stance and walking. Spatiotemporal proximity to the stability boundary based
on Lee’s (2009) visual tau and Riccio’s (1993) expansion to the postural domain has
recently inspired a rapidly developing line of research on postural time to contact,
with applications to aging (Forth et al. 2007; Slobounov et al. 1998; Van Wegen et al.
2002), neurological disorders (Van Wegen et al. 2001; Remelius et al. 2008; Van
Emmerik and Van Wegen 2002; Van Emmerik et al. 2010), orthopedic injuries (Hertel
and Olmsted-Kramer 2007), as well as conceptual/methodological developments
(Haddad et al. 2006; Hof et al. 2005; Slobounov et al. 1997).
In research on people with MS, studies have found reduced postural time to
contact in upright standing tasks with sensory manipulations (Van Emmerik et al.
2010; Cattaneo 2012). Van Emmerik et al. (2010) observed decreased time to contact
of the CoM to the support boundaries under a variety of postural tasks, including
reaching and leaning in different directions. Cattaneo (2012) found time to contact to
be a sensitive and valid method to assess balance changes in MS under static balance
conditions (comparing sold and compliant surfaces) and sensory manipulations (eyes
open and closed). The static time to contact, however, was not found to be a predictor
of falls in the MS group in this study (Cattaneo 2012). These results suggest that time
to contact may be a more sensitive measure when used under dynamic conditions,
such as in balance perturbations or locomotion.
15.3.3 Time to Contact and the Initiation of Stepping
The majority of studies to date using postural time to contact have assessed differ-
ences between populations. During gait initiation, Remelius et al. (2008) observed
reduced gait initiation speed, a wider stance base, longer time to contact, and reduced
anticipatory postural reactions during the initiation of gait in MS. These results in-
dicate functional adaptations in people with MS that reduce proximity to stability
boundaries prior to stepping. Whether or not these adaptations are possible and
observed during gait initiation at different speeds is not known.
Little research exists addressing Lee’s (2009) claim that time to contact is sensed
and controlled in any modality. Slobounov et al. (2009) asked participants to stand
on one leg as long as possible and assessed the transition to stepping back on two
legs. They found that minimum time to contact (and variability) increased during the
transition to instability. Time to contact may also function as a control parameter in
the transition from a stationary base of support to taking a step (Hasson et al. 2008,
2009). In earlier research, we investigated the postural responses elicited by balance
perturbations of increasing magnitude applied to the upper body in younger and older
adults (Hasson et al. 2008, 2009). A key variable in the analysis of postural responses
was the time to contact the anterior stability boundary. We found that minimum time
to contact predicts the perturbation level that elicits a stepping response that acts to
create a larger base of support and maintains postural stability. Healthy older adults
(without a history of falls) stepped at lower perturbation levels than younger adults,
but this response was functional as they experienced greater CoM accelerations. In
both age groups similar time to contact and muscular response patterns to increasing
perturbation magnitude were observed, suggesting that time to contact may be used as
an informational variable to regulate muscle activity in maintaining postural stability
(Hasson et al. 2009).
15.3.4 Dynamic Time to Contact: Changing Boundaries During

Locomotion
Gait research has found that people with MS walk with a slower preferred speed,
shorter stride length, altered limb kinematics, and longer dual-support times
(Benedetti et al. 1999; Martin et al. 2006; Kelleher et al. 2010; Remelius et al.
2012). Many of the altered gait parameters that have been observed during walking
could be due to different gait speeds between MS and control groups. Remelius
et al. (2012) controlled for gait speed and found that the increased dual-support time
at preferred speed is also present when speed is controlled for. At slow (0.6 m/s),
medium (1.0 m/s), and fast speeds (1.4 m/s), the MS group showed increased dual
support times compared to age-matched controls. This increased dual support time
may be seen as a protective gait strategy in which people with MS adapt their gait
to optimize dual support to enhance stability. However, Remelius et al. (2012) also
Fig. 15.3 a Partitioning of swing into phases relative to the boundaries of the base of support. Early
swing: from toe-off to center of mass (CoM) reaching heel boundary. Mid swing: CoM moves from
heel-to-toe boundary. Late swing: from CoM passing the toe boundary to next heel contact. b
Differences in timing (ms) of phases of swing between an individual with MS compared to age-
and gender-matched control
found that at slower and faster imposed speeds the MS group altered the kinematics
of swing to accommodate these dual-support adaptations. These alterations in swing
could provide different challenges to postural stability, as discussed below.
Phases of swing are usually defined on the basis of lower extremity kinematics,
such as maximal knee joint angle (Winter 1983). However, the phases of swing can
also be defined by the position and movement of the CoM relative to the stability
boundary formed by the base of support at the foot–ground interface. In this approach,
two events serve to partition swing: the point in swing when the CoM moves in front
of the posterior limit of the stance foot (heel), and when it moves beyond the anterior
limit of the stance foot (toe) in the sagittal plane. Thus, the phases of swing can
be defined as: early swing (CoM of body behind the stance foot), mid swing (CoM
of body between heel and toe of the stance foot), and late swing (CoM in front of
the stance foot) (see Fig. 15.3a). Our proposed description of swing thus defines the
phases of swing in organism–environment relevant terms, here represented by the
movement of the CoM in relation to the limits of the base of support at the body–
ground interface. Figure 15.3b shows the changes in these three phases of swing
when comparing walking in a person with MS to a healthy age-matched control.
In the person with MS, there is a decrease in early swing time and an increase in
mid swing, with the late swing timing left relatively intact compared to the healthy
control participant. These changes are representative of group differences observed
by Remelius et al. (2012) and highlight the increased importance of the mid-swing
part of gait when the CoM is positioned over the single base of support of the stance
leg.
As indicated earlier, increased dual-support times during walking in people with
MS may be adaptive and increase stability during gait, but Remelius et al. (2012)
suggested that these adaptations may also require potentially destabilizing modifi-
cations to the swing-phase dynamics of walking. In the Remelius et al. (2012) study,
we found that individuals with MS positioned their head and body CoM closer to
the anterior support boundary compared to non-MS controls when entering the un-
stable equilibrium of the swing phase (the point at which the swing foot passes the
anterior boundary of the base of support). At faster velocities, the head even tended
to lead the swing foot into the unstable equilibrium for individuals with MS (see
Fig. 15.4a). This altered sequence in the approach to the unstable equilibrium may
indicate an attempt to better visualize step placement, as we have also found that
people with MS maintain a head orientation (and presumably point of gaze) closer to
the body compared to age-matched controls (Fig. 15.4b). This increased reliance on
visual control of foot placement and swing formation likely results from decreased
somatosensory function. However, these adaptations could have serious implications
for stability during this phase because the swing foot may be in a less optimal posi-
tion to reestablish the base of support, should balance be perturbed (Remelius et al.
2012). Further study is needed to determine whether the functional adaptation to
lengthen dual-support time is related to increased risk of falls for individuals with
MS.
The CoM time to contact (TtC CoM ) of the approach towards the stability bound-
aries can be computed by dividing the first derivative of CoM position (velocity,
Eq. 15.1) by the distance between the current position of the CoM and the physical
boundary (Bp : the stance foot toe) or the anticipated boundary (Ba : heel of the next
foot contact):
Bp,a − CoM(m)
T tCCoM (s) = (15.1)
velocityCoM (m/s)
where Bp,a can either be the physical (Bp ) or anticipated (Ba ) boundary. The swing
foot time to contact (TtC swing foot ) is computed similarly. The Bp and Ba locations are
obtained from kinematic records post-hoc. Examples of the time-to-contact approach
to the physical boundary for a participant with MS and an age-matched control are
presented in Fig. 15.5. To investigate Lee’s (2009) third aspect of his general tau
theory, that movement guidance emerges from coupling different tau’s, coupling
ratios can be computed to describe the relationship between temporal closures and
coordination of the CoM and swing foot with the physical boundary (Bp ) or the
anticipated boundary (next step landing area; Ba ):
T tCswing foot
= Kp,a (15.2)
T tCCoM
where Kp,a can either be the coupling ratio for the approach to the physical (Kp )
or anticipated (Ka ) boundary. The point in swing when the coupling ratio (Kp )
becomes less than 1.0 indicates when the TtC swing foot has dropped below TtC CoM in
the approach to the Bp , signifying that the swing foot will arrive before the CoM
enters the unstable equilibrium, and the swing foot will lead the body in the approach
to the next anticipated boundary. A Kp = 1 that occurs later in the swing phase may
Fig. 15.4 a Timing of center of mass (head and body) and swing foot relative to crossing the anterior
stability boundary (Bp ) while walking at 1.4 m/s. Negative values: CoM crossing after swing foot;
positive values: CoM crossing before swing foot (adapted from Remelius et al. (2012). b Projection
of head orientation on the surface of support in the direction of locomotion. Distances indicated are
from the intersection of the head vector on the ground to the body (the vertical projection of the
CoM)
indicate a less stable final part of swing and less controlled foot placement. This can
be observed in the time-to-contact time series and coupling ratio in the case of a
person with MS (see Fig. 15.5); the Kp = 1 occurs later in the swing phase compared
to the age-matched control (24 % versus 9 % of swing).
Fig. 15.5 Time to contact the physical (Bp ) boundary for swing leg and body CoM and their coupling
ratio for a control and MS participant
The coupling ratio Ka (Eq. 15.2) reflects the approach to the anticipated boundary
and quantifies the difference in contact estimates of the swing foot and the CoM.
A high Ka ratio indicates the CoM arriving shortly after the swing foot, whereas
a low Ka ratio indicates a longer estimated time between swing foot arrival at the
Ba and the eventual CoM arrival. Espy et al. (2010) provided evidence that a higher
CoM velocity (associated with greater overground gait speeds) increases stability and
reduces falls when exposed to a slip upon foot contact. As a higher CoM velocity
is likely linked to a reduced time to contact, this would mean that a higher Ka ratio
results in a safer step-landing pattern, reducing the probability of balance loss during
a slip. However, a ratio approaching 1 would indicate simultaneous arrival of the
CoM and swing foot and less of a controlled fall during late swing with little time
for proper foot placement before full weight acceptance. The relation between this
ratio and gait stability may therefore be U-shaped with optimal stability between
0 and 1. People with mild-to-moderate impairment due to MS were found to have
similar Ka ratios compared to age-matched controls, suggesting that the approach
to the step-landing area is preserved during walking (Remelius 2012). This may be
the result of different body and head orientations during walking that enable greater
visual input of foot placement and the step-landing area (Remelius 2012). This ratio
may be affected in those with greater disability and balance disturbances due to
their MS.
In summary, the boundary-relevant measures of postural and gait control presented

here, together with the time-to-contact analysis, have provided novel insights into
the formation of swing during walking in people with MS. These findings indicate
systematic changes in the approach to the unstable equilibrium (the “controlled fall”
phase of gait) that may reflect instability and potentially greater vulnerability to
postural perturbations. In contrast, the analysis also showed that the approach to the
step-landing area is preserved in people with mild-to-moderate MS, possibly due to
greater use of vision of the immediate area under the feet.
The approach presented here assesses postural stability in relation to the stability
boundaries and has significant potential in furthering our understanding of changes in
postural stability due to aging and various pathologies. Another promising approach
is the assessment of fluctuations in movement variables such as the CoP or the CoM
on the basis of entropy analysis that can quantify the complexity and adaptability of
the movement dynamics.
15.4 Nonlinear Entropy Analysis of Posture and Gait:

Assessing Movement Adaptability in MS
Current research in the biological and physical sciences stresses the beneficial and
adaptive aspects of variability that emerge from the interaction of the many degrees
of freedom in a complex system (Glass 2001; Lipsitz 2002; Van Emmerik and Van
Wegen 2002). These developments question the previously dominant perspective
in which decreased variability is universally associated with increased competence,
skill, and health. Instead, the path to frailty or disease is identified by a loss of vari-
ability in fundamental variables reflecting biological function, as well as reduced
“complexity” related to the interactions between the multiple degrees of freedom
(Glass 2001; Lipsitz 2002). Over time, reductions in effective degrees of freedom
as well as loss of interactivity between them may become associated with a loss of
variability. When these reductions in degrees of freedom and variability reach a crit-
ical threshold, injury or disease may emerge. Various nonlinear analysis tools have
been developed to assess these changes in system complexity or adaptability, includ-
ing Lyapunov analysis, detrended fluctuation analysis, and various assessments of
entropy (Glass 2001; Lipsitz 2002). In the next sections, we discuss developments
related to entropy measures, and how these have been used to assess changes in the
nature of movement control in people with MS.
15.4.1 Introduction to Entropy Analysis for Posture and Gait
The movement science literature has seen a recent emergence of various entropy
analysis techniques to assess the complexity and adaptability of human motion by
quantifying fluctuations in physiological time series data. Currently, there are three
different entropy measures used in the analysis of postural and gait control: Ap-
proximate entropy (ApEn; Pincus 1991), sample entropy (SampEn; Richman and
Moorman 2000), and multiscale entropy (MSE; Costa et al. 2002), all of which
are estimates of Kolmogorov entropy. While each measure operates with slightly
different computational mechanics, all are nonlinear techniques intended to assess
time series fluctuations. Richman and Moorman (2000) outlined limitations of the
ApEn method and provided an alternative measure, SampEn, for the estimation of
the complexity of biological time series. A major issue with ApEn is that entropy
measures can be inflated due to self-matching bias (i.e., comparing a single data point
to itself) when assessing similarity between points in the time series. The power of
SampEn has been expanded by the work of Costa and colleagues (Costa et al. 2002,
2005, 2007) into MSE, which determines SampEn over many time scales, and then
calculates the integral of the resulting curve to compute the complexity index (Ci ), a
measure that is informative of how complexity changes over the time scales in ques-
tion. A meaningful way to evaluate the outcomes of entropy calculations is through
the “loss of complexity hypothesis” (Lipsitz and Goldberger 1992), which views loss
of system complexity as a pathway to disability or injury. Analysis of postural and
locomotor fluctuations based on entropy analysis has the potential to augment our
understanding of how different diseases, such as MS, constrain the movement output
and limit adaptability of the system.
15.4.2 Loss of Complexity
Lipsitz and Goldberger (1992) introduced the loss of complexity hypothesis in the
early 1990s, and have provided evidence for the framework in light of some chal-
lenges to its premises (Vaillancourt and Newell 2002). The insight conferred by
studies using MSE analysis provides a robust way of evaluating the complexity of
physiological systems, where reduced complexity can lead to maladaptive outcomes,
such as loss of balance and falls. Various studies examining aging and frailty have
demonstrated loss of complexity in postural control using entropy measures (Costa
et al. 2002, 2005, 2007; Kang et al. 2009; Manor et al. 2010). The inclusion of
multiple time scales in the MSE analysis, rather than a single time scale approach,
allows for a more complete view of the dynamics of the relevant physiological pro-
cesses that impact postural control in individuals with MS (Goldberger et al. 2002).
Single time-scale approaches are limited, as the observed differences or similarity
in entropy at a particular time scale may not be representative for other relevant time
scales that impact system dynamics.
The strength of entropy-based methods compared to other measures of time se-
ries variability is that point-to-point fluctuations in noncyclical tasks can be analyzed,
such as those observed in postural control of the CoP. Entropy analysis of the CoP al-
lows assessment of postural fluctuations through estimates of the amount of Shannon
Information (Shannon and Weaver 1949) within the signal. Signals with higher en-
tropy values, regardless of method used, contain greater information and are thereby
deemed more complex. The particular advantage of the MSE entropy method is that
it provides the most comprehensive insight into how the complexity varies over the
multiple time scales that are of physiological relevance to a particular biological
signal.
15.4.3 Complexity Changes due to MS
As mentioned earlier, balance and gait impairments in people with MS are com-
mon and can have major impact on quality of life and daily activities. Increased
understanding of adaptability through entropy analysis during static and dynamic
movements may provide insights into how the disease impacts a person’s ability
to suitably adapt responses to postural challenges. ApEn has been used to evaluate
changes in postural control in people with MS (Huisinga et al. 2012), and shown that
people with MS exhibit reduced ApEn values during quiet standing in the medio-
lateral CoP time series (with both eyes open and eyes closed). These changes in
the point-to-point fluctuations as measured by ApEn may be indicative of a reduced
ability to manage unexpected perturbations, especially in the mediolateral direc-
tion. These changes in postural control fit well with the conclusion from Maki et al.
(1994) that impaired mediolateral control of posture may place subjects at increased
fall risk. The lack of differences observed in anterior–posterior CoP fluctuations may
arise from previously noted issues with the ApEn measure (Richman and Moorman
2000).
Kaipust et al. (2012) examined spatial aspects of the walking-gait cycle at a self-
selected speed, and found reductions in ApEn in those with MS compared to controls.
There was, however, no ability to differentiate those with mild from moderate im-
pairment due to MS, indicating that ApEn may lack the sensitivity to distinguish
between MS subpopulations. MSE has shown to be effective in distinguishing be-
tween clinical subgroups in those with adolescent idiopathic scoliosis (Gruber et al.
2011) as well as identifying systematic differences between older individuals with
and without a history of falls (Costa et al. 2007), frailty (Kang et al. 2009), and
sensory impairments (Manor et al. 2010). The outcomes of these postural studies
indicate that MSE analysis may have the ability to discriminate between those with
and without MS and/or different MS subtypes.
15.4.4 Multiscale Entropy: An Example
Here, we will present a comparison of the mediolateral CoP time series during quiet
upright standing for a person with mild MS and an age- and gender-matched person
without MS. The purpose of this comparison is to demonstrate how MS impacts CoP
fluctuations. The MSE analysis represents data across ten time scales of interest,
ranging from 0.11–0.2 s. This analysis uses an r = 0.15 of the standard deviation
Fig. 15.6 Example of multiscale entropy (MSE) of the medial–lateral foot center of pressure (CoP)
for one MS and one control participant.Time scale in (s). Inset: Complexity index across all time
scales
across the entire time series and m = 2, where r is the tolerance for accepting matches
and m is the length of the sequence to be compared. From Fig. 15.6, it is apparent
that the person with MS exhibits reduced SampEn across the time scales of interest,
resulting in a reduced MSE complexity index compared to the person without MS.
The observed lower entropy is consistent with other findings from the literature
(Costa et al. 2007; Manor et al. 2010) that demonstrate a reduction in functional
degrees of freedom controlling the system, resulting in less adaptable movements.
These preliminary results using entropy analysis suggest that even in those with
minimal physical symptoms due to their MS, systematic changes in the nature and
adaptability of their postural control can be revealed. Epidemiological evidence in-
dicates that those with MS are at increased risk for falls (Cattaneo et al. 2002),
demonstrating a strong need for sensitive tests to assess disease progression that
may also inform rehabilitative strategies. The MSE analysis technique presented
here has the potential to perform such assessments by measuring subtle changes in
the complexity and adaptability of postural control in MS.
In summary, entropy techniques are beginning to provide sensitive analyses of the
postural control changes in people with MS. The different entropy analysis techniques
are all rooted in similar theoretical principles (Richman and Moorman 2000), though
SampEn and MSE algorithms are more valid as they avoid self-matching bias. The
MSE analysis also allows assessment of many time scales of a particular time series,
which could provide insights into how posture is regulated using input from the
somatosensory, visual, or vestibular systems with different temporal dynamics. It is
therefore important to consider a variety of time scales when evaluating the impact of
a disorder on the postural control system, as postural instability and loss of balance
result from the deterioration and loss of interaction of many subsystems.
15.5 Novel Treatments for Balance and Gait in MS: Tai Chi
Methods that aim to improve the overall complexity or adaptability of all of the
subsystems (e.g., somatosensation, muscle strength, flexibility) that affect postu-
ral control may prove to be a more robust way of improving balance compared to
interventions that aim to improve single subsystems. We now turn our focus to in-
terventions aimed at improving balance and mobility in people with MS, and in
particular the effects of using Tai Chi as an intervention that integrates coordination,
strength, and sensory function to enhance adaptive postural control.
15.5.1 Current Intervention Research in MS
Several different rehabilitation and training protocols have been used in the MS pop-
ulation to reduce symptom severity. The most common exercise interventions have
included aerobic, resistance, and balance training, alone or in combination. Aerobic
training interventions have demonstrated improvements in mobility, level of fatigue,
and perceived balance (Learmonth et al. 2013; Dettmers et al. 2009). Similarly, resis-
tance training interventions have shown improvements in balance and locomotion, as
well as reductions in fatigue (Cakt et al. 2010; Gutierrez et al. 2005). Balance training
interventions have reported improvements in fatigue, standing balance, and quality
of life (Hebert et al. 2011; Tarakci et al. 2013). An 8-week intervention that used a
combination of aerobic, resistance, and balance training to reduce symptom severity
reported improved 25-ft walk times, functional ambulatory scores, and Timed Up
and Go times (Motl et al. 2012), indicating improved mobility following the 8-week
intervention.
After a review of 11 different exercise interventions for MS (studies were di-
vided into four categories: aerobic, flexibility, strengthening/resistance training, or
yoga), Asano et al. (2009) concluded that while each program has its own benefits,
an exercise intervention that integrates aerobic capacity, flexibility or coordination,
strength, and balance training may prove to be the most effective treatment. Tai Chi
is an exercise form that combines all these aspects and has the potential to enhance
both static and dynamic balance in individuals with MS.
15.5.2 Tai Chi as a Comprehensive Intervention
Tai Chi is an ancient Chinese martial art that consists of gentle flowing move-
ments that have made it an adaptable and efficient exercise for a wide range of
people. Tai Chi has been used as an intervention to reduce symptom severity and
increase quality of life in older individuals and in diabetic, stroke, and Parkinson’s
populations. In healthy older adults Tai Chi practice has been reported to reduce
fear of falling and improve joint proprioception, reaction time, and static balance
(Fong and Ng 2006; Sattin et al. 2005; Tsang and Hui-Chan 2004). Individuals with
diabetes have demonstrated increased peripheral nerve conduction and improvements
in balance and plantar sensation following a 6-month Tai Chi intervention (Hung et al.
2009; Li and Manor 2010; Richerson and Rosendale 2007). Participants with mild-
to-moderate Parkinson’s disease showed improved performance on clinical measures
such as the 6-minute walk, Timed Up and Go test, and backward walking (Hackney
and Earhart 2008) following a 3-month Tai Chi training program. Chronic stroke
patients who participated in a 12-week Tai Chi program showed improved static and
dynamic balance compared to non-Tai Chi stroke controls (Au-Yeung and Hui-Chan
2009).
15.5.3 Tai Chi as a Balance Intervention in MS
Husted et al. (1999) found improvements in 25-ft walking speed, hamstring flexibil-
ity, and psychosocial well-being after an 8-week Tai Chi intervention in individuals
with MS, while Mills et al. (2000) found positive effects on balance and reductions
in depression following a 2-month Tai Chi intervention. We recently completed a
3-week Tai Chi pilot intervention study to assess changes in functional status and
static balance in individuals with mild-to-moderate MS (Averill 2013). The Tai Chi
intervention consisted of 9 h of Tai Chi standing meditation and Tai Chi slow walking
training practiced in a group setting. Following the intervention, participants with
MS increased neural drive (rapid foot tapping), leg strength (repeated sit-to-stand
movements), and improved their standing balance (Fig. 15.7). For static balance, an
increase in the CoP time to contact during tandem stance (a task that was not prac-
ticed during the intervention) was observed following the intervention, indicating
increased temporal margins to the stability boundary. For dynamic balance during
Tai Chi standing meditation with arm movements, participants were more likely to
challenge their stability boundaries as shown by a decreased time to contact during
this condition (see Fig. 15.7).
The significance of Tai Chi as an intervention is its generalizability in treating a
wide variety of impairments experienced by those with MS. Larger and more com-
prehensive intervention studies evaluating Tai Chi and other interventions are needed
to assess the most effective way to enhance balance control and somatosensory func-
tion in people with MS. Once treated, improvements in balance control and mobility
may help to increase physical activity levels, reduce fatigue, and optimize quality of
life, allowing people with MS to retain their independence for as long as possible.
15.6 Summary and Conclusions
Understanding the nature and mechanisms of balance impairments in MS is essential

because of the major impact that reduced mobility has on quality of life, combined
with the fact that MS affects people at relative early stages in their lives. A growing
Fig. 15.7 a Changes in muscular strength (time to complete five-chair rises) and neural drive
(number of foot taps completed in 15 s) following a 3-week Tai Chi intervention for three participants
with MS. Data represent means and standard deviations. b Changes in time to contact (s) of the
instantaneous center of foot pressure (CoP) to the base of support following a 3-week Tai Chi
intervention for three participants with MS during standing meditation with arm movement and
tandem stance. Data represent means and standard deviations
body of literature is emerging that addresses the nature of these postural and loco-
motor impairments in MS. Here, we reviewed the major changes in postural and gait
control in MS and showed how strength, fatigue and somatosensory loss impact bal-
ance function. We also provided new insights into conditions under which postural
instability emerges through the application of novel analytical methods and exper-
imental paradigms based on nonlinear and complex systems methods. These novel
methods included (1) boundary-relevant measures of postural stability and control
(postural “time to contact”) and (2) entropy measures for assessing postural and gait
adaptability. The promise of these novel methods is that they allow us to differentiate
between postural and gait variability caused by dysfunction and variability that pro-
vides adaptability to respond to postural challenges. Finally, we discussed how these
methods and paradigms may help to develop innovative treatments for balance and
gait dysfunction in people with MS. In particular, we showed that using a Tai Chi
intervention, which integrates aspects of aerobic exercise, coordination and flexibil-
ity training, somatosensory function, as well as strength and balance, is a promising
exercise intervention that may improve balance and mobility and therefore have a
major impact on quality of life in people living with MS.
Acknowledgements The research presented here was supported by a Grant from the National
Multiple Sclerosis Society (NMSS; RG3974A2)
References
Asano M, Dawes D, Arafah A, Moriello C, Mayo N. What does a structured review of the effective-
ness of exercise interventions for persons with multiple sclerosis tell us about the challenges of
designing trials? Mult Scler 2009;15:412.
Au-Yeung S, Hui-Chan C. Short-form Tai Chi improves standing balance of people with chronic
stroke. Neurorehabil Neural Repair 2009; 23:515–22.
Averill J. Effect of a Tai Chi Chuan Slow Walking Intervention on Balance and Mobility in In-
dividuals with Multiple Sclerosis. Unpublished Master’s Thesis. University of Massachusetts
Amherst, 2013.
Benedetti MG, Piperno R, Simoncini L, Bonato P, Tonini A, Giannini S. Gait abnormalities in
minimally impaired multiple sclerosis patients. Mult Scler 1999;5(5):363–8.
Boes MK, Sosnoff JJ, Socie MJ, Sandroff BM, Pula JH, Motl RW. Postural control in multiple
sclerosis: effects of disability status and dual task. J Neurol Sci 2012 Apr 15;315(1–2):44–8.
Burschka JM, Keune PM, Menge U, Hofstadt-van Oy U, Oschmann P, Hoos O. An exploration of
impaired walking dynamics and fatigue in multiple sclerosis. BMC Neurol 2012;12:161.
Cakt BD, Nacir B, Genç H, Saraçoğlu M, Karagöz A, Erdem HR, Ergün U. Cycling progressive
resistance training for people with multiple sclerosis: a randomized controlled study. Am J Phys
Med Rehabil 2010; 89(6):446–57.
Cameron MH, Horak FB, Herndon RR, Bourdette D. Imbalance in multiple sclerosis: a result of
slowed spinal somatosensory conduction. Somatosenory Motor Res 2008;25(2):113–22.
Cattaneo D. The virtual time to contact in the evaluation of balance disorders and prediction of falls
in people with multiple sclerosis. Disabil Rehabil 2012;34(6):470–477.
Cattaneo D, Jonsdottir J. Sensory impairments in quiet standing in subjects with multiple sclerosis.
Mult Scler 2009;15(1):59–67.
Cattaneo D, De Nuzzo C, Fascia T, Macalli M, Pisoni I, Cardini R. Risks of falls in subjects with
multiple sclerosis. Arch Phys Med Rehabil 2002;83(6):864–7.
Chung LH, Remelius JG, Van Emmerik REA, Kent-Braun JA. Leg power asymmetry and postural
control in women with multiple sclerosis. Med Sci Sport Exer 2008;40(10):1717–24.
Citaker S, Gunduz AG, Guclu MB, Nazliel B, Irkec C, Kaya D. Relationship between foot sensation
and standing balance in patients with multiple sclerosis. Gait Posture 2011;34(2):275–8.
Coote S, Hogan N, Franklin S. Falls in people with multiple sclerosis who use a walking aid:
prevalence, factors, and effect of strength and balance interventions. Arch Phys Med Rehabil
2013;94(4):616–21.
Corporaal SH, Gensicke H, Kuhle J, Kappos L, Allum JH, Yaldizli O. Balance control in multiple
sclerosis: correlations of trunk sway during stance and gait tests with disease severity. Gait
Posture 2013;37(1):55–60.
Costa M, Goldberger AL, Peng CK. Multiscale entropy analysis of complex physiologic time series.
Phys Rev Lett 2002;89(6):068102.
Costa M, Goldberger AL, Peng CK. Multiscale entropy analysis of biological signals. Phys Rev
2005;71:021906.
Costa M, Priplata AA, Lipsitz LA, Wu Z, Huang NE, Goldberger AL, Peng CK. Noise and poise:
enhancement of postural complexity in the elderly with a stochastic-resonance-based therapy.
Europhys Lett 2007;77(6);68008.
De Haan A, de Ruiter CJ, van Der Woude LH, Jongen PJ. Contractile properties and fatigue of
quadriceps muscles in multiple sclerosis. Muscle Nerve 2000;23(10):1534–41.
Dettmers C, Sulzmann M, Ruchay-Plössl A, Gütler R, Vieten M. Endurance exercise improves
walking distance in MS patients with fatigue. Acta Neurol Scand 2009;120(4):251–7.
Espy DD, Yang F, Bhatt T, Pai YC. Independent influence of gait speed and step length on stability
and fall risk. Gait Posture 2010;32(3):378–82.
Findling O, Sellner J, Meier N, Allum JHJ, Vibert D, Lienert C, et al. Trunk sway in mildly
disabled multiple sclerosis patients with and without balance impairment. Exp Brain Res
2011;213(4):363–70.
Finlayson ML, Peterson EW, Cho CC. Risk factors for falling among people aged 45 to 90 years
with multiple sclerosis. Arch Phys Med Rehabil 2006;87(9):1274–9.
Fjeldstad C, Pardo G, Frederiksen C, Bemben D, Bemben M. Assessment of postural balance in
multiple sclerosis. Int J MS Care 2009;11:1–5.
Fjeldstad C, Pardo G, Bemben D, Bemben M. Decreased postural balance in multiple sclerosis
patients with low disability. Int J Rehabil Res 2011 Mar;34(1):53–8.
Fong S, Ng G. The effects on sensorimotor performance and balance with Tai Chi training. Arch
Phys Med Rehabil 2006;87: 82–7.
Ford HL, Gerry E, Johnson MH, Tennant A. Health status and quality of life of people with multiple
sclerosis. Disabil Rehabil 2001;23(12):516–21.
Forth KE, Metter EJ, Paloski WH. Age associated differences in postural equilibrium con-
trol: a comparison between EQscore and minimum time to contact (TTCmin). Gait Posture
2007;25(1):56–62.
Frzovic D, Morris ME, Vowels L. Clinical tests of standing balance: performance of persons with
multiple sclerosis. Arch Phys Med Rehabil 2000;81(2):215–21.
Gianfrancesco MA, Triche EW, Fawcett JA, Labas MP, Patterson TS, Lo AC. Speed- and cane-
related alterations in gait parameters in individuals with multiple sclerosis. Gait Posture
2011;33(1):140–2.
Glass L. Synchronization and rhythmic processes in physiology. Nature 2001;410:277–84.
Goldberger AL, Peng CK, Lipsitz LA. What is physiologic complexity and how does it change with
aging and disease? Neurobiol Aging 2002;23(1):23–6.
Gruber AH, Busa MA, Gorton GE, Van Emmerik REA, Masso PD, Hamill J. Time-to-contact and
multiscale entropy identify differences in postural control in adolescent idiopathic scoliosis.
Gait Posture 2011;34(1):13–8.
Gutierrez GM, Chow JW, Tillman MD, McCoy SC, Castellano V, White LJ. Resistance train-
ing improves gait kinematics in persons with multiple sclerosis. Arch Phys Med Rehabil
2005;86(9):1824–9.
Hackney M, Earhart G. Tai Chi improves balance and mobility in people with Parkinson disease.
Gait Posture 2008;28:456–60.
Haddad JM, Gagnon J, Hasson CJ, Van Emmerik REA, Hamill, J. The use of time-to-contact
measures in assessing postural stability. J Appl Biomech 2006;22:155–61.
Hasson CJ, Van Emmerik REA, Caldwell GE. Predicting dynamic postural instability using center
of mass time-to-contact information. J Biomech 2008;41:2121–9.
Hasson CJ, Caldwell GE, Van Emmerik REA. Scaling of plantarflexor muscle activity and postural
control time-to-contact in response to upper-body perturbations in young and older adults. Exp
Brain Res 2009;196(3):413–27.
Hebert JR, Corboy JR. The association between multiple sclerosis-related fatigue and balance as a
function of central sensory integration. Gait Posture 2013;38(1):37–42.
Hemmett L, Holmes J, Barnes M, Russell N. What drives quality of life in multiple sclerosis? Qjm
2004;97(10):671–6.
Hertel J, Olmsted-Kramer LC. Deficits in time-to-boundary measures of postural control with
chronic ankle instability. Gait Posture 2007;25(1):33–9.
Hof A, Gazendam M, Sinke W. The condition for dynamic stability. J Biomech 2005;38:1–8.
Huisinga JM, Yentes JM, Filipi ML, Stergiou, N. Postural control strategy during standing is altered
in patients with multiple sclerosis. Neurosci Lett 2012;524(2):124–8.
Huisinga JM, Schmid KK, Filipi ML, Stergiou N. Gait mechanics are different between healthy
controls and patients with multiple sclerosis. J Appl Biomech 2013;29(3):303–11.
Hung J, Liou C, Wand P, Yeh S, Lin L, Lo S, Tsai F. Effect of 12-week tai chi chuan exercise on
peripheral nerve modulation in patients with type 2 diabetes mellitus. J Rehab Med 2009;41:
924–9
Husted C, Pham L, Hekking A, Niederman R. Improving quality of life for people with chronic
conditions: the example of t’ai chi and multiple sclerosis. Altern Ther Health Med 1999;5:70–4.
Jacobs JV, Kasser SL. Balance impairment in people with multiple sclerosis: preliminary evidence
for the balance evaluation systems test. Gait Posture 2012a;36(3):414–8. (a)
Jacobs JV, Kasser SL. Effects of dual tasking on the postural performance of people with and without
multiple sclerosis: a pilot study. J Neurol 2012b;259(6):1166–76. (b)
Kaipust JP, Huisinga JM, Filipi M, Stergiou N. Gait variability measures reveal differences between
multiple sclerosis patients and healthy controls. Motor Control 2012;16(2):229–44.
Kang HG, Costa MD, Priplata AA, Starobinets OV, Goldberger AL, Peng CK, Lipsitz LA. Frailty
and the degradation of complex balance dynamics during a dual-task protocol. J Gerontol A
Biol Sci Med Sci 2009;64(12):1304–11.
Karst GM, Venema DM, Roehrs TG, Tyler AE. Center of pressure measures during standing tasks in
minimally impaired persons with multiple sclerosis. J Neurol Phys Ther 2005 Dec;29(4):170–80.
Kelleher KJ, Spence WD, Solomonidis S, Apatsidis D. The effect of textured insoles on gait patterns
of people with multiple sclerosis. Gait Posture 2010;32(1):67–71.
Kent-Braun JA, Sharma KR, Miller RG, Weiner MW. Postexercise phosphocreatine resynthesis is
slowed in multiple sclerosis. Muscle Nerve 1994;17(8):835–41.
Kent-Braun JA, Ng AV, Castro M, Weiner MW, Gelinas D, Dudley GA, et al. Strength, skeletal
muscle composition, and enzyme activity in multiple sclerosis. J Appl Physiol 1997;83(6):1998–
2004.
Krishnan V, Kanekar N, Aruin AS. Anticipatory postural adjustments in individuals with multiple
sclerosis. Neurosci Lett 2012;506(2):256–60. (a)
Krishnan V, Kanekar N, Aruin AS. Feedforward postural control in individuals with multiple
sclerosis during load release. Gait Posture 2012;36(2):225–30. (b)
Learmonth Y, Motl R, Sandroff B, Pula J, Cadvid D. Validation of patient determined disease steps
(PDDS) scale scores in persons with multiple sclerosis. BMC Neurol 2013;13–37.
Lee DN. General Tau theory: evolution to date. Perception 2009;38:837–58.
Lee DN, Lishman R. Visual control of locomotion. Scand J Psychol 1977;18(3):224–30.
Li L, Manor B. Long term Tai Chi exercise improves physical performance among people with
peripheral neuropathy. Am J Chin Med 2010;38:449–59.
Lipsitz LA. Dynamics of stability: the physiologic basis of functional health and frailty. J Gerontol
A Biol Sci Med Sci 2002;57:B115–25.
Lipsitz LA, Goldberger AL. Loss of ‘Complexity’ and Aging. JAMA 1992;267(13):1806–9.
Maki BE, Holliday PJ, Topper AK. A prospective study of postural balance and risk of falling in an
ambulatory and independent elderly population. J Gerontol 1994;49(2):M72–84.
Manor B, Costa MD, Hu K, Newton E, Starobinets O, Kang HG, et al. Physiological complexity and
system adaptability: evidence from postural control dynamics of older adults. J Appl Physiol
2010;109(6):1786–91.
Martin CL, Phillips BA, Kilpatrick TJ, Butzkueven H, Tubridy N, McDonald E, et al. Gait and
balance impairment in early multiple sclerosis in the absence of clinical disability. Mult Scler
2006;12(5):620–8.
McCollum G, Leen TK. Form and exploration of mechanical stability limits in erect stance. J Mot
Behav 1989;21:225–44.
Mills N, Allen J, Carey-Morgan S. Does Tai Chi/Qi Gong help patients with multiple sclerosis?
J Bodywork Mvt Therap 2000;4(1):39–48.
Mills RJ, Young CA. A medical definition of fatigue in multiple sclerosis. Qjm 2008 Jan;101(1):
49–60.
Motl RW, Smith DC, Elliott J, Weikert M, Dlugonski D, Sosnoff JJ. Combined training improves
walking mobility in persons with significant disability from multiple sclerosis: a pilot study.
J Neurol Phys Ther 2012;36(1):32–7.
Negahban H, Mofateh R, Arastoo AA, Mazaheri M, Yazdi MJ, Salavati M, et al. The effects
of cognitive loading on balance control in patients with multiple sclerosis. Gait Posture
2011;34(4):479–84.
Newsome SD, Wang JI, Kang JY, Calabresi PA, Zackowski KM. Quantitative measures detect
sensory and motor impairments in multiple sclerosis. J Neurol Sci 2011;305(1–2):103–11.
Ng AV, Miller RG, Kent-Braun JA. Central motor drive is increased during voluntary muscle
contractions in multiple sclerosis. Muscle Nerve 1997;20(10):1213–8.
Ng AV, Miller RG, Gelinas D, Kent-Braun JA. Functional relationships of central and peripheral
muscle alterations in multiple sclerosis. Muscle Nerve 2004;29(6):843–52.
Nilsagard Y, Lundholm C, Denison E, Gunnarsson LG. Predicting accidental falls in people with
multiple sclerosis—a longitudinal study. Clin Rehabil 2009;23(3):259–69.
Noguiera LA, Teixeira L, Sabino P, Filho HA, Alvarenga RM, Thuler LC. Gait character-
istics of multiple sclerosis patients in the absence of clinical disability. Disabil Rehabil
2013;35(17):1472–8.
Pai YC, Patton J. Center of mass velocity-position predictions for balance control. J Biomech
1997;30(4):347–54.
Pike J, Jones E, Rajagopalan K, Piercy J, Anderson P. Social and economic burden of walking and
mobility problems in multiple sclerosis. BMC Neurol 2012;12:94.
Pincus SM. Approximate entropy as a measure of system complexity. Proc Natl Acad Sci U S A
1991;88(6):2297–301.
Porosinska A, Pierzchala K, Mentel M, Karpe J. Evaluation of postural balance control in patients
with multiple sclerosis—effect of different sensory conditions and arithmetic task execution. A
pilot study. Neurol Neurochir Pol 2010;44(1):35–42.
Remelius JG. Adaptations to stride patterns and head movements during walking in persons with and
without multiple sclerosis. Doctoral Dissertation, University of Massachusetts Amherst, Schol-
arWorks@UMass Amherst, http://scholarworks.umass.edu/dissertations/AAI3498368, 2012.
Remelius JG, Hamill J, Kent-Braun J, Van Emmerik REA. Gait initiation in multiple sclerosis.
Motor Control 2008;12(2):93–108.
Remelius JG, Jones SL, House JD, Busa MA, Averill JL, Sugumaran K, Kent-Braun JA, Van
Emmerik REA. Gait impairments in persons with Multiple Sclerosis across preferred and fixed
walking speeds. Arch Phys Med Rehabil 2012;93:1637–42.
Riccio GE. Information in movement variability. About the qualitative dynamics of posture and
orientation. In K.M. Newell and D.M. Corcos (Eds.), Variability and Motor Control 1993,
317–57. Champaign, IL: Human Kinetics.
Rice CL, Vollmer TL, Bigland-Ritchie B. Neuromuscular responses of patients with multiple
sclerosis. Muscle Nerve 1992;15(10):1123–32.
Richerson S, Rosendale K. Does Tai Chi improve plantar sensory ability? A pilot study. Diabetes
Technol Ther 2007;9:276–86.
Richman JS, Moorman JR. Physiology time-series analysis using approximate entropy and sample
entropy. Am J Physiol Heart Circ Physiol 2000;278:2039–49.
Riley AR, Wong S, Mita S, Turvey MT. Common effects of touch and vision on postural parameters.
Exp Brain Res 1997;117:165–70.
Said CM, Goldie PA, Patla AE, Culham E, Sparrow WA, Morris ME. Balance during obstacle
crossing following stroke. Gait Posture 2008;27(1):23–30.
Sattin R, Easley K, Wolf S, Chen Y, Kutner M. Reduction in fear of falling through intense Tai Chi
exercise training in older, transitionally frail adults. J Am Geriatr Soc 2005;53:1168–78.
Shannon C, Weaver W. The Mathematical Theory of Information, 1949, Urbana, IL: University of
Illinois Press.
Slobounov SM, Slobounova ES, Newell KM. Virtual time-to-collision and human postural control.
J Mot Behav 1997;29:263–81.
Slobounov SM, Moss SA, Slobounova ES, Newell KM. Aging and time to instability in posture.
J Gerontol Biol Sci 1998;53A(1):B71–8.
Slobounov SM, Cao C, Jaiswal N, Newell KM. Neural basis of postural instability identified by
VTC and EEG. Exp Brain Res 2009;199:1–16.
Sosnoff JJ, Sandroff BM, Motl RW. Quantifying gait abnormalities in persons with multiple sclerosis
with minimal disability. Gait Posture 2012;36(1):154–6.
Soyuer F, Mirza M, Erkorkmaz U. Balance performance in three forms of multiple sclerosis. Neurol
Res 2006;28(5):555–62.
Spain RI, St. George RJ, Salarian A, Mancini M, Wagner JM, Horak FB, et al. Body-worn motion
sensors detect balance and gait deficits in people with multiple sclerosis who have normal
walking speed. Gait Posture 2012;35(4):573–8.
Tarakci E, Yeldan I, Huseyinsinoglu BE, Zenginler Y, Eraksoy M. Group exercise training for
balance, functional status, spasticity, fatigue and quality of life in multiple sclerosis: randomized
controlled trial. Clin Rehabil 2013;27(9):813–22.
Tsang W, Hui-Chan C. Effect of a 4 and 8 week Tai Chi Training on balance control in the elderly.
Med Sci Sports Exer 2004;36:648–57.
Vaillancourt DE, Newell KM. Changing complexity in human behavior and physiology through
aging and disease. Neurobiol Aging 2002;23(1):1–11.
Van Emmerik REA, Van Wegen EEH. On the functional aspects of variability in postural control.
Exercise Sport Sci Rev 2002;30(4):177–83.
Van Emmerik REA, Remelius JG, Johnson MB, Chung LH, Kent-Braun JA. Postural control in
women with multiple sclerosis: effects of task, vision and symptomatic fatigue. Gait Posture
2010;32(4):608–14.
Van Emmerik REA, Jones SL, Busa MA, Baird JA. Systems perspective on postural and gait
stability: implications for physical activity in aging and disease. Kinesiol Rev 2013;2:17–28.
Van Wegen EEH, Van Emmerik REA, Wagenaar RC, Ellis T. Stability boundaries and lateral postural
control in Parkinson’s Disease. Motor Control 2001;3:254–69.
Van Wegen EEH, Van Emmerik REA, Riccio GE. Postural orientation: age-related changes in
variability and time-to-boundary. Hum Mov Sci 2002;21:61–84.
Weinshenker BG, Bass B, Rice GP, Noseworthy J, Carriere W, Baskerville J, et al. The natural
history of multiple sclerosis: a geographically based study. I. Clinical course and disability.
Brain 1989;112(Pt 1):133–46.
Winter DA. Biomechanical motor patterns in normal walking. J Motor Behav 1983;15(4):302–30.
Winter DA. Human balance and posture control during standing and walking. Gait Posture
1995;3:193–214.
Wurdeman SR, Huisinga JM, Filipi M, Stergiou N. Multiple sclerosis affects the frequency content
in the vertical ground reaction forces during walking. Clin Biomech 2011;26(2):207–12.
Index
A Averill, J. 270
Abelew, T.A., 42, Aymard, C., 213
Abundance, 6, 210–215, 219, 221, 222
Action, 7, 12, 26, 28, 41, 56, 65, 70, 81, B
83, 130, 148–150, 151, 157, 179, Babinski, F., 6
181, 215 Badley, E.M., 205
Adamovich, S.V., 18 Balance, 58, 146, 191, 241, 252–257, 259, 264,
Adams, B.D., 56 269, 270, 272
Adams, M.A., 201 Baldissera, F., 37
Adams, R.A., 28 Bangert, M., 162
Adler, C.H., 172 Baraduc, P., 102
Afferent feedback, 25, 26, 45, 46, 184 Barsalou, L.W., 70, 149
Ahn, A.N., 51 Bartlett, R., 215, 218
Albert, F., 9 Baumeister, R.F., 165
Alexandrov, A.V., 9 Baur, V., 168
Allen, T.J., 19, 22 Bawa, P., 209
Alstermark B, 28, 61 Bays, P.M., 18, 28
Altenmüller, E., 163, 166, 167, 170, 181 Beales, P.L., 184
Analysis of variance, 97 Beckmann, J., 173
Andersen, J.H., 206 Beek, P.J., 191
Andersen, R.A., 83l, 149 Beer, R.F., 231
Anderson, C.T., 184 Beggs, W.D.A., 55
Angel, M.J., 49 Belhaj-Saif, A., 241
Anson, J.G., 214 Benedetti, M.G., 256, 260
Anticipatory synergy adjustments, 7 Benjamin, M., 199
Araujo, D., 104 Bennett, D.J., 36
Archambault, P.S., 238 Berlyne, D.E. 154
Arshavsky, Y.I., 28 Bernier, P.M., 82
Artist, 152, 154, 157, 162, 180–184, 186, 189, Bernstein, N., 92, 184, 185, 189, 190
195, 197, 201 Bernstein, N.A., 5, 103, 210
Asano, M., 269 Bertelson, P., 79
Asatryan, D.G., 18, 19, 232 Bigland-Ritchie, B., 207
Ashley, M.J., 42 Binder, M.D., 37
Ashoori, A., 166 Bizzi, E., 4
Aston-Jones, G., 34 Blankevoort, L., 196
Astrom A., 209 Bläsing, B., 150
Aune, T.K., 170, 217 Blood, A.J., 153

and Biology 826, DOI 10.1007/978-1-4939-1338-1
278 Index
Body Cesqui, B., 122

Boes, M.K., 253 Cham, R., 41, 46
Bonasera, S.J., 49, 51 Chang, S.H., 229, 231
Bonda, E., 81, 83 Chang, Y.-H., 94
Bongers, P.M., 206 Chen, R., 240
Bosch, T., 217 Cheney, P.D., 241
Botvinick, M., 81, 83 Cheng, F.P., 166, 167
Bourbonnais, D., 229 Choking under pressure, 165,
Boyle, R., 43 Chu, V.W.T., 122
Brady, N., 85 Chung, L.H., 254, 255, 257
Brandfonbrener, A.G., 167 Cibulka, M.T., 198
Braun, S.M., 150 Cignetti, F., 216
Bridgeman, B., 18 Cinderella, 208–210, 215, 219
Brink, E.E., 47 Cinzia, D.D., 153
Brinkman, J., 241 Circus arts, 189, 190
Bronner, S., 197 Cirstea, M.C., 244
Brown, P., 58, 62 Citaker, S., 253, 254
Brown, S., 152
Clark, J.M., 59
Brugger, P., 83
Classification of movement disorders, 166
Brugués, A.O., 165
Cohen, H., 79
Brunnström, S., 244
Cohen, J., 81, 83
Buckle, P.W., 206
Cohen, R.G., 108–111, 126
Buhler, J.E., 188
Cohen, Y.E., 83
Bui, T.V., 37
Bullock, D., 98 Colebatch, J.G., 241
Burke, D., 241 Collet, C., 150
Burne, J.A., 231 Collingwood, R.G., 154
Burrack, A., 83 Collins, D.J., 69
Burschka, J.M., 255 Complex skill, 121
Bütefisch, C.M., 241 Conson, M., 84
Butz, M.V., 98 control, 1, 2, 5, 12, 17–19, 28, 29, 37, 94,
Byl, N.N., 171, 174 98, 103, 161–169, 172–174, 185, 188,
194, 196, 197, 209, 232
Conway, B.R., 181
C
Cakt, B.D., 269 Cooper, L.A., 69–72, 78
Callosal agenesis, 59, 162 Coordination, 91, 92, 95, 98, 121, 147, 164,
Calota, A., 231 166, 168, 183, 184, 262, 269, 272
Calvo, M., 165 Coote, S, 243
Calvo-Merino, B., 150–152, 154, 155 Corboy, J.R., 255
Cameron, M.H., 254 Corcos, D.M., 231
Caminiti, R., 81 Corporaal, S.H., 253, 254, 256
Cantu, H.A., 214 Corticospinal tract, 59, 240, 241, 244
Cao, Y., 241 Coslett, H.B., 84
Capaday, C. 232 Costa, M., 266–268,
Carello, C., 198 Courjon, J.-H., 61
Carlsen, A.N., 62 Cowie, R.J., 60
Carlson-Kuhta, P., 42, 44 Crawford, J.D., 81
Carlton, L.G., 55 Créange, A., 238
Carver, C.S., 165 Criscimagna-Hemminger, S.E., 102
Catsman-Berrevoets, C.E., 65, 66 Cross, E.S., 155, 157
Cattaneo, D., 253, 259, 268 Cushing, S., 35
Cavallari, P., 238 Cusumano, J.P., 94, 125, 126
Cesari, P., 94, 125, 126 Czarnecki, K., 169
Index 279
D Elbert, T., 172

Dai, Y., 241 Elovic, E.P., 231
Dance, 147–157, 181–183, 197 Emery, C.A., 197
Danion, F., 121 Emulation, 70, 73, 74, 85, 86
Danna-Dos-Santos, A., 7 Enders, L., 173
Daprati, E., 82, 85 Enoka, R.M., 34, 35
d’Avella, A., 9, 104 Enriquez Denton, M., 38
Davids, K, 216 Entropy, 265–268
Davis, K.D., 199 Equifinality, 4, 10–12
Day, B.L., 57, 65 Equilibrium-point hypothesis, 2, 3
De Gelder, B., 79 Ericsson, K.A., 162
De Haan, A., 254 Espy, D.D., 259, 264
de Kleijn, A., 43 Eysenck, M., 165
de Lange, F.P., 81–83
Decerebrate Cat, 43, 47 F
Decety, J., 70, 86 Fadiga, L., 149
Deconinck, F.J.A., 73 Faisal, A.A., 83, 106, 129
Degree of freedom problem, 92, 103, 196, Falla, D., 212, 213
266, 268 Fallentin, N., 208
Dehaene, S., 196 Fatigue, 35, 51, 163, 169, 174, 206–209,
DeLuca, C.J., 209, 211, 212 212–214, 216–219, 255, 269
Denève, S., 83 Fechner, G.T., 22, 23
Fedirchuk, B., 241
Desmurget, M., 26
Fedorowich, L., 113, 217, 221
Dettmers, C., 269
Feeney, A., 205
Devereux, J.J., 206
Feldman, A.G., 2–4, 12, 19–23, 26, 27,
Dewald, J.P., 229, 231
231, 232
di Pellegrino, G., 149
Fellows, S.J., 229
Dijkstra, T.M.H., 114
Ferri, F., 73, 80
Dimov, M., 234
Feuerstein, M., 204
Dingwell, J.B., 125, 126, 130
Fidler, H., 165
Dirnberger, G., 171
Findling, O., 253
Disordered motor control, 229, 240, 241, 244
Fink, A., 152
Dissanayake, E., 152
Finlayson, M.L., 253
DiZio, P., 4
Fisher, C.M., 241
Dodson, J.D., 165 Fisher, M., 59
Dolan, P., 201 Fjeldstad, C., 253, 257
Donchin, O., 102 Flanagan, J.R., 118
Donelan, J.M., 47, 50 Flor, H., 171
Draganski, B., 103 Fogassi, L., 149
Duhamel, J.R., 72 Fong, S., 270
Dunning, J., 198 Force Feedback, 46, 47, 49–51
Durand, M.J., 221 Ford, H.L., 251, 252
Duysens, J., 47 Forth, K.E., 259
Duysens, J., 47, 50 Fowler, C.A., 28, 42
Dynamic stereotype, 166, 173 Fowler, E.G., 42
Dynamical system, 111–113, 121 Frak, V., 70
Freedberg, D., 154, 155
E Freitas, S.M.S.F., 92
Eccles, J.C., 49 Friedman, D. 148
Edin, B.B., 17, 22, 29 Fries, W., 65, 241
Edwards, R.H.T., 207 Frucht, S.J., 168
Ehrlenspiel, F., 115 Frzovic, D., 253
Ehrsson, H.H., 81 Fugl-Meyer, A.R., 235
280 Index
Fuller, J.R., 213, 214, 216, 218 Guertin, P., 49

Furuya, S., 174 Guillot, A., 150
Fusani, L., 147 Gutierrez, G.M., 253, 269
H
G
Hackney, M., 270
Gallese, V., 154, 155
Haddad, J.M., 259
Gamet, D., 207
Hagberg, M., 208
Gandevia, S.C. 241
Hagendoorn, I., 153
Ganis, G., 83
Hagg, G., 209
Garcia de Yebenes, J., 181
Hagg, G.M., 209
Gardner, H., 189
Hallam, S., 162
Garraway, S.M., 34
Halle, M., 190
Gärtner, H., 162
Hallett, M., 169
Gaser, C., 162
Hallett, P.E., 56
Ge, H.Y., 220
Hamberg-van Reenen, H.H., 206
Gehring, D., 197
Hamill, J., 218
Gelfand, I.M., 6, 210 Hamilton, G.M., 184
Gemperline, J.J., 229 Hammond, M.C., 229
Gender differences, 212, 219, 220 Hänggi, J., 152
Georgopoulos, A.P., 57 Harris-Love, M.L., 240
Georgoulis, A.D., 218 Harris-Warrick, R.M., 38
Ghez, C., 3 Hasan, Z., 8
Gianfrancesco, M.A., 256 Haslinger, B., 173
Gibson, E.J., 189 Hass, C., 41, 46
Gibson, J.J., 18, 182, 189 Hasson, C.J., 116, 118, 119, 258, 260
Ginanneschi, F, 238 Health, 181, 184, 198, 201, 265
Glass, L., 265 Hebert, J.R., 255
Goel, V., 153 Heckman, C.J., 34–37
Goldberger, A.L., 266 Heiderscheit, B.C., 218, 219
Goldman, A.I., 150 Helliwell, P.S., 173
Goldman, P.S., 65 Helmich, R.C., 73
Gordon, J., 3 Hemmett, L, 251, 252
Gorniak, S., 10 Henneman, E., 209, 211
Gorniak, S.L., 10, 13 Hermans, V., 208
Goslow, G.E., Jr., 42 Herrojo-Ruiz, M., 172
Gottlieb, G.L., 3, 4 Hertel, J, 259
Gottschall, J.S., 42–46 Hetu, S, 81, 83
Goudy N., 209, 211 Hick, WE., 63
Gowland, C., 229 Hildebrand, A., 154
Gracies, J.M., 238 Hill, A.M., 210
Grafton, S.T., 149 Hill, D.M., 165
Grammer, K., 152 Hille, B., 34, 37
Granit, R., 49 Himmelbach, M., 61
Graziano, M., 81 Hinder, M.R., 4
Gregor, R.J., 42, 44 Hochman, S., 34, 38
Grezes, J., 149 Hodges, P.W., 219
Grigg, P., 94 Hof, A., 259
Grillner, S., 43 Hoffman, D.S., 229
Grosbras, M.H., 149 Holmes, P., 112, 114
Gruber, A.H., 267 Holmes, P.S., 69
Grush, R., 73 Holstege, J.C., 37
Guardia, D., 182 Holt, K.G., 4
Guckenheimer, J., 112, 114 Hooftman, W.E., 219
Index 281
Hornby, T.G., 35 Johnson-Frey, S.H., 149

Houk, J., 49 Jones, L.A., 207
Hounsgaard, J., 34, 35 Jonsdottir, J., 253
Howarth, C.I., 55 Jordan, M.I., 7, 93, 126
Hubbard-Turner, T., 197
Huerta, M.F., 60 K
Hug, F., 212 Kaas, J.H., 60
Hughlings Jackson, J., 6, 9 Kaipust, J.P., 267
Hui-Chan, C., 270 Kammers, M.P., 83
Huisinga, J.M., 256, 267 Kamper, D.G., 229
Hulliger, M., 17, 21, 22, 25 Kang, H.G., 266, 267
Hultborn, H., 22, 34, 37 Karni, A., 101, 171
Human, 46, 55, 56, 62, 121, 126, 147, 149, Karsh, B.T., 206
161, 179, 201, 222 Karst, G.M., 253
Hung, J., 270 Kasser, S.L., 254
Hunter, I.W., 207 Katsumata, H., 115, 116
Husted, C., 270 Katz, R., 238
Huyghues-Despointes, C.M., 49 Kawabata, H., 154
Huysmans, M.A., 218 Kawato, M., 1
Keele, S.W., 55
I
Keizer, A., 182,
Ikonomou, E., 153
Keizer, K., 66
Illert, M., 61
Kelleher, K.J., 253, 260
Ilmane, N., 19, 21, 25, 28, 29
Kelso, J.A., 95
Imagery, 70, 73, 83
Kent-Braun, J.A., 254, 255
Ingber, D.E., 196
Keysers, C., 154
Injury, 164, 180, 189, 196, 197, 200, 201,
Kiehn, O., 232
206–209, 211, 219, 245, 265
Kilbom, A., 209
Interaction, 2, 43, 49, 63, 116, 121, 155, 189,
Kindt, M., 173
195, 265, 268
Kinesthetic illusions, 18, 20, 25, 26, 29
Ioannou, C., 172
Kinsbourne, M., 59
Ionta, S., 72
Kirsch, L., 155–157
Ishizu, T., 153, 154
Kistemaker, D.A., 198
Ivanenko, Y.P., 6, 9
Klämpfl, M.K., 166, 172, 175
J Klous, M., 9
Jabusch, H.C. 167, 168, 170, 173, 174 Koch, C., 196
Jacob, P., 149, 150 Kosslyn, S.M., 70, 81, 82
Jacobs, B.L., 34 Krakauer, J.W., 231
Jacobs, J.V., 254 Krechetnikov, R., 116
Jacobsen, T., 152 Kringelbach, M.L., 153
James, W., 86 Krishnamoorthy, V., 9, 93
Jankovic, J., 166 Krishnan, V, 254
Jankowska, E., 22, 29 Kristofferson, A.B., 101
Jeannerod, M., 70, 149 Kugler, P.N., 1
Jeeves, M.A., 59 Kuo, J.J., 37
Jensen, K.B., 173 Kursell, J., 180
Jobin, A., 14, 234, 235 Kuypers, H.G., 37
Johansen, T.I., 212. 220 Kuypers, H.G.J.M., 65, 66, 240, 241
Johansson, N., 17, 22, 29 Kveraga, K., 63, 64
Johansson, R.S., 9, 118
Johnson, M.D., 37 L
Johnson, S., 86 Lackner, J.R., 4
Johnson, S.H., 86 Lacquaniti, F., 55
282 Index
Lafreniere-Roula, M., 43 Maki, B.E., 267

Lance, J.W. 229, 241 Malhotra, S., 231
Landi, S.M., 102 Mallgrave, H.F., 153
Lang, A.E., 169 Manchester, D.L., 9
Lang, C.E., 229, 234, 244 Manor, B., 266–268, 270
Larsson, B., 206, 209 Manuel, M., 38
Latash, M.L., 3, 17, 18, 20, 25 Manzoni, D., 43
Latash, M.L., 1, 6, 7, 9, 10, 12, 13, 91–93, Marchand, A.R., 47
126, 215 Marchand-Pauvert, V., 238
Laterality, 69, 70, 74, 76–78, 79, 82–84 Martin, C.L., 253, 256, 260
Lawrence, D.G., 65, 241 Martin, J.R., 12
Lay, A., 41, 46 Martin, O., 55
Learmonth, Y., 269 Martin, V., 7, 9, 12, 93, 98, 216
learning, 104, 121, 126, 132, 215 Massaro, D.W., 148
Lederman, R.J., 167 Massion, J. 9
Lee, A., 167, 234 Masterton, R.B., 61
Lee, D.N., 259 Mathiassen, S.E., 209, 211
Lee, R.H., 34–37 Maton, B., 207
Lee, W.A., 234 Matthews PB., 232, 238
Leijon, M., 205 Mattos, D., 11, 6, 196
Levin, M.F., 198, 231, 234–236, 240, 244, May, R.M., 114
Lexell J., 209 Mayer, H.C., 116
Li, L., 270 Mazevet, D., 231
Li, X., 34 McClelland, V.M., 232, 238
Liederbach, M., 197 McCloskey, D.I., 17
Lim, V., 168 McCollum, G, Leen T.K., 259
Limb stiffness, 47, 49–51 McCormick, K., 80, 84
Lin, J.K., 164 McCrea, D.A., 43
Lindberg, P.G., 241 McGuire, L.M., 83
Lindman, R., 209 McKenzie, A., 174
Lindsay, A.D., 37 McPherson, J.G., 232
Lindsay, K.W., 43 Merdler, T., 244
Linzenbold, W., 61 Merton, P.A., 28,
Lipsitz, L.A., 265, 266 Messing, K., 219,
Lobinger, B., 166, 172 Meunier, S., 232
Locomotion, 3, 38, 42, 46, 48, 50, 52, 200, 269 Mezaki, T. 167
Lomond KV, Côté JN., 213, 214, 219 Michaelsen, S.M., 244
Longo MR, Haggard P., 83 Mihaltchev, P., 234, 243
Lorblanchet, M. 179 Miller, J.F., 34
Lotze, M., 82 Mills, N., 270
Lower limb, 56 Mills, R.J., 255
Lund, J.P., 212 Milner, A.D., 59
Lundberg, A., 232 Milner, T.E., 4
Luria, A.R., 70 Mirror system, 149, 150, 151, 154
Lyon, I.N., 57, 65 Misawa, S., 241
Mishkin, M., 55
M Missenard, O., 218
Ma, S., 10 Mittelstaedt, H., 43, 45
Maas, H., 41, 42, 51 Moayedi, M., 199
MacLean, J.N., 38 Modulation, 34, 46, 47, 49, 51, 200
Macpherson, J.M., 6, 9 Molenberghs, P., 149
Madeleine, P.M., 221 Montague, S.J., 34
Magnus, R., 43 Moore, R.D., 173
Index 283
Moran, A., 150 O

Morice, A.H.P., 116 O’Dwyer, N.J., 231
Moseley, G.L., 84, 85, 219 Ojofeitimi, S., 197
Motl, R.W., 269 Olafsdottir, H., 101
Motoneuron, 4, 25, 28, 29, 33–38, 49, 232 Olafsdottir, H.B., 9, 13
Motor Olivier, E., 61
Moulton, S.T., 70, 73 Orgs, G., 152
Mountcastle, V.B., 148 Orians, G.H., 152
Moussavi, Z.M.K., 102 orientation, 43, 46, 47, 50, 51
Mulder, T., 69 Ostry, D.J., 20, 189
Müller, H., 93, 94, 106, 108, 109 Otten, M., 165
Mullick, A.A., 231, 235, 237
Mulliken, G.H., 83 P
Multi-degree of freedom motor control, 98, 265 Packard, M.G., 171
Multisensory integration, 74, 82, 84, 85 Pai, Y.C., 259
Münte, T.F., 162 Paillard, J., 27
Muratori, L.M., 9 Palmer, S.E., 153
Murnaghan. C., 209 Parietal, 26, 54, 81, 82, 148–151, 155
Musampa, N.K., 231, 234, 238 Park, E.Y., 229
Musculoskeletal disorders, 205, 218 Parker, P.A., 209
Musicians’ dystonia, 166–169 Parsons, L.M., 69, 72, 73, 82, 85
Mussa-Ivaldi, F.A., 3, 102, Patton, J., 259
Myers, K.A., 195 Paulig, J., 166
Paulignan, Y., 55
N Pavlov, I.P., 166
Nakashima, K., 168 Pearson, K.G., 47, 50
Nathan, P.W., 241 Pelgrims, B., 83
Nauta, W.J.H., 65 Perception, 29, 69, 81, 129, 148–155, 157,
Neck afferents, 46 173, 189
Negahban, H., 253 Perez, M.A., 241
Nelson-Wong, E., 212 Performance, 8, 12, 13, 73, 82, 91, 92, 98, 108,
Neuroaesthetics, 148, 152, 153, 157 109, 111, 138, 151, 157, 165, 181, 213
Neurological injury, 234 Performing arts, 181
Neuromotor noise, 103, 106, 126 Person, R.S., 211
neuron, 229, 254 Pesenti, A., 171
Neuroscience, 121, 148, 149, 157 Petersen, T.H., 240
Newell, K.M., 126, 215, 216, 266 Petkova, V.I., 81
Newsome, S.D., 254 Phantom limb, 18, 20, 26, 27, 29
Ng, A.V., 254, 255, 270 Phillips, E.M., 184
Ng, G., 270 Physics, 1, 2, 7, 12, 121, 122
Ní Choisdealbha, Á., 73, 84 Pike, J., 253
Nichols T.R., 48 Pincus, S.M., 266
Nichols, T., 50 Pinheiro-Franco, J.L., 200
Nichols, T.R., 42, 43, 46, 49, 232, 238 Pinna, B., 181
Nielsen, J.B., 234 Poffenberger, A.T., 59
Nilsagard, Y., 253 Poggio, T., 189
Nirkko, A.C., 241 Polcyn, A.F., 42
N-methyl-d-aspartate (NMDA), 33, 37, 38 Polster, B., 188
Noguiera, L.A., 255 Pompeiano, O., 43
Nordlund, A., 206 Pongratz, G., 197
Norepinephrine, 33, 232 Poon, C.S., 83
Nudo, R.J., 61 Poore, G.V., 180
Nutt, J.G., 168 Poppele, R.E., 94
284 Index
Porosinska, A., 253 Sauner, D., 83

Position sense, 17, 19, 29 Schaal, S., 115, 116
Posner M.I., 55 Schaible, H.G., 201
Pouget, A., 83 Scheidt, R.A., 3
Power, K.E., 34 Scheier, M.F., 165
Powers, R.K., 35, 234, 241 scheme, 27, 86
Powers, W.T., 28 Schieber, M.H., 229, 234, 244
Prablanc, C., 55 Schlaug, G., 162
Prather, J.F., 36 Schneider, T.D., 179
Premotor, 26, 74, 81, 82, 85, 149, 150, Scholz, J.P., 92, 109
151, 154 Schöner, G., 92, 109
Prentice, S.D., 46 Schuele, S.U., 174
Prilutsky, B.I., 6, 41, 42 Schwoebel, J., 82, 84, 85
Prinz, W., 161 Seif-Naraghi, A.H., 6
Sekiyama, K., 72
R Selen, L.P., 216
Redfern, M., 41, 46 Sense of effort, 19, 22–24, 26, 28, 29
Redundancy, 5, 6, 103, 104, 106, 111, 118, Serotonin, 32, 174, 232, 234
121, 243, 244 Shabbott, B.A., 102
Referent body configuration, 3 Sharma, N., 82
Referent configuration, 3, 7 Shenton, J.T., 74
Reflexes, 18, 19, 240, 245 Shepard, R., 69, 70
Reticulospinal, 47, 62, 240, 241 Shepard, R.N., 69–72, 78
Reynolds, R.F., 56, 62, 63, 65 Shik, M.L., 43
Riccio, G.E., 259 Shim, J.K., 101
Rice, C.L., 254 Shim, J.K., 8, 9
Richerson, S., 270 Shimizu, T., 241
Richman, J.S., 266 Shiro, Y., 209
Riemer, M., 80, 84 Showers C J. 165
Riley, M.A., 216 Shrier, I., 184
Rizzolatti, G., 149 Shrier, I., 190
Robert, T., 200 Shuster, S.M., 147
Roberts, L.V., 238 Silva, P.L., 184
Robinson, D.L., 60 Silverstein, B., 206
Robson, C., 167 Simulation, 70, 71, 73, 74, 81, 86, 134,
Roll, J.P., 25 135, 149
Rosenblatt, N.J., 216 Sinigaglia, C., 149
Rosset-Llobet, J., 168 Sirigu, A., 72, 82, 86
Roussouly, P., 200 Sjogaard, G., 210
Rushton, A., 198 Skill acquisition, 69, 102–104, 215
Russell, D.M., 12 Skurvydas, A., 216
Rymer, W.Z., 229 Slater, M., 81
Slifkin, A.B., 12
S Slobounov, S.M., 257, 259, 260
Sabes, P.N., 83 Slope Walking, 41–46, 51
Sahrmann, S.A., 231 Sluiter, J.K., 209
Said, C.M., 259 Smith, J.L., 23, 42, 46, 47, 241
Saimpont, A., 84 Soechting, J.F., 55
Sainburg, R.L., 102 Soeter, M., 173
Sale, D.G., 211 Sogaard, K., 210
Sanders, A.F., 148 Søndergaard, K.H., 218
Sataloff, R.T., 181 Sosnoff, J.J., 256
Sattin, R., 270 Soyuer, F., 253, 254
Index 285
Spaepen, A.J., 208 Tin, C., 83

Spain, R.I., 256 Tinetti, M.E., 42
Sparks, D.L., 60 Ting, L.H., 6, 9, 104
Spence, J.T., 165 Tobin, J.L., 184
Spence, K.W., 165 Todorov, E., 7, 93, 126
Srinivasan, D., 215 Tomasino, B., 82
Sripanidkulchai, K., 172 Training, 103, 135, 151, 157, 181–183, 189,
Srivastava, U.C., 47 190, 195, 208, 269, 270
Stability, 6, 8, 12, 19, 95, 105, 114, 118, 132, Treatment, 164, 170, 173, 174, 180, 200, 201
216, 251, 256, 257, 261, 262, 270 Tresch, M.C., 6
Startle, 62 Trilk, J.L., 184
Steele, C.J., 162 Tsang, W., 270
Steinmetz, A., 164 Tsuchiya, N, 196
Stellman JM., 219 Tufillaro, N.B., 112, 114
Steptoe, A., 165 Turvey, M.T., 28, 42
Stergiou, N., 215, 216 Tuthill, C., 50
Sternad, D., 93, 108, 109, 116
Stochastic Resonance, 110 U
Stock, S, 219 Ulrich, J.A., 201
Stokes, M.J., 207 Uncontrolled manifold hypothesis, 6
Straub, R.H., 201 Ungerleider, L.G., 55
Strick, P.L., 229 Upper limb, 56, 57, 59, 191, 206, 209, 212,
Stroke-Spasticity, 237, 253 213, 242, 244
Stuart, D.G., 206
V
Stuphorn, V., 61
Vaillancourt, D.E., 266
Sub-cortical, 55, 58, 62, 65, 66, 241
Vallbo, A.B., 17, 22
Sundelin, G., 208
Valls-Sole, J., 62
Sunnerhagen, K.S., 229
van den Hoorn, W., 218
Superior colliculus, 60, 61, 63, 64
van der Wees, P.J., 197
Suter, E, 198
van der Windt, D.A., 209
Suurkula J., 209
van Dieen, J.H., 209, 211, 216, 218
Suzuki, I., 43
van Elk, M., 85
Svendsen, J.H., 221
Van Emmerik, R.E.A., 253–255, 257, 259
Svirskis, G., 34
Van Kordelaar, J., 244
Swanson, R.L., 196 van Vugt, F.T., 174
Swift, E.J., 103 Van Wegen, E.E.H., 258, 259
Synergy, 5–10, 91, 241, 244 Van Zuylen, E.J., 232
Szeto, G.P., 164 Vaquero, L., 162
Vargas, C.D., 86
T Variability, 91, 140, 215, 216
Tai-Chi intervention, 270, 271 Vartanian, O., 153
Takala, E.P., 208 Veiersted, K.B., 209
Tarakci, E., 269 Verrel, J., 94
Taylor, W.J., 173 Vestibular System, 47, 48, 268
Ter Haar Romeny, B.M., 232 Virtual tasks, 83, 105, 109, 115, 119, 134, 135
ter Horst, A.C., 73 Visser, B., 206, 209, 218
Theberge, N., 219 Visuomotor, 55, 56, 58, 59, 63, 65, 116
Thilmann, A.F., 235 Viswanathan, S., 73, 75
Thorn, S., 209, 212 Voerman, G.E., 209
Thoroughman, K.A., 102 Voet, D., 196
Threshold control, 198, 229, 240 Voland, E., 152
Ticini, L.F., 157 Vollestad, N.K., 207
Time-to-contact, 258, 259, 262, 263, 265 von Holst, E., 43, 45
286 Index
von Holst, H., 17, 18, 28 Wolpert, D.M., 18, 28

Voon, V. 169 Wood, D.E., 231
Vora, A.J., 200 Woolf, A.D., 205
Woollacott, M.H., 9
W Woolsey, C.N., 241
Wahlstedt, K., 206 Wu, Y.-H., 13, 102
Walsh, L.D., 23, 24 Wurdeman, S.R., 256
Wan, C.Y., 162
Wang, D., 38 Y
Wanke, E.M., 184 Yang, J.-F., 6
Warburg, A., 154 Yarrow, K., 150
Ward, A.B., 241 Yeldan, I., 269
Waters-Metenier, S., 102 Yen, J.T., 94
Weaver, W., 266 Yerkes, R.M., 165
Weber, E.H., 22 Yin, T.C., 148
Weinshenker, B.G., 243 Yoshie, M., 165
Welford, A.T., 148
Werner, W., 61 Z
Westad, C., 218 Zacks, J.M., 85
Westgaard, R.H., 209, 211, 212 Zagoraiou, L., 34
Westling, G., 9, 118 Zaidel, E., 59
Williams, J., 73 Zatorre, R.J., 153, 161
Wilmink, R.J., 49 Zatsiorsky, V.M., 6, 200
Wilson, G.D., 163, 165 Zehr, E.P., 241
Wilson, M., 163, 165 Zeki, S., 153, 154
Windholz, H., 166 Zelaznik, H.N., 55
Wing, A.M., 118 Zennaro, D., 209, 211, 212
Winkel, J. 209, 211 Zhang, L.Q., 231
Winter, D.A., 251, 267 Zhou, T., 8, 9, 11
Wise, S.P., 56, 59 Zhu, H., 34

(Advances in Experimental Medicine and Biology 826) Mindy F. Levin (Eds.) - Progress in Motor Control - Skill Learning, Performance, Health, and Injury-Springer-Verlag New York (2014)

Uploaded by

Copyright:

Available Formats

(Advances in Experimental Medicine and Biology 826) Mindy F. Levin (Eds.) - Progress in Motor Control - Skill Learning, Performance, Health, and Injury-Springer-Verlag New York (2014)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

(Advances in Experimental Medicine and Biology 826) Mindy F. Levin (Eds.) - Progress in Motor Control - Skill Learning, Performance, Health, and Injury-Springer-Verlag New York (2014)

Uploaded by

Copyright:

Available Formats

Advances in Experimental Medicine and Biology

2013 Impact Factor: 2.012

More information about this series at http://www.springer.com/series/5584

Progress in Motor Control

ISSN 0065-2598 ISSN 2214-8019 (electronic)

Library of Congress Control Number: 2014947961

© Springer Science+Business Media New York 2014

Printed on acid-free paper

Springer is part of Springer Science+Business Media (www.springer.com)

as well as those linked to psychological mechanisms are invoked as contributing to

Montreal Mindy F. Levin

1 Motor Control: On the Way to Physics of Living Systems . . . . . . . . . . 1

2 Motor Control and Position Sense: Action-Perception Coupling . . . . 17

3 Reconfiguration of the Electrical Properties of Motoneurons

4 The Regulation of Limb Stiffness in the Context of Locomotor

5 Subcortical Visuomotor Control of Human Limb Movement . . . . . . . 55

6 Rethinking the Role of Motor Simulation in Perceptual Decisions . . . 69

7 Use of the Uncontrolled Manifold (UCM) Approach to Understand

8 Acquisition of Novel and Complex Motor Skills: Stable Solutions

9 The Dynamical Analysis of Inter-Trial Fluctuations Near Goal

10 Motor Control in Action: Using Dance to Explore the Intricate

11 Apollo’s Curse: Causes and Cures of Motor Failures in Musicians:

12 Motor Control and the Injured and Healthy Artist . . . . . . . . . . . . . . . . 179

13 Adaptations to Neck/Shoulder Fatigue and Injuries . . . . . . . . . . . . . . . 205

14 Deficits in Spatial Threshold Control of Muscle Activation

15 Enhancing Postural Stability and Adaptability in Multiple

Eckart Altenmüller Institute of Music Physiology and Musicians’ Medicine,

Jinger S. Gottschall Department of Kinesiology, Pennsylvania State University,

T. Richard Nichols School of Applied Physiology and Department of Biomedical

1.1 Motor Control as a Subfield of Physics of Living Systems

© Springer Science+Business Media New York 2014 1

1.2 Equilibrium-point Hypothesis and Control With Referent

Generalization of the control with tonic stretch reflex thresholds to a two-muscle

1.3 Equifinality and its Violations: Mass–Spring Analogy

1.4 Redundancy and Abundance

1.5 Synergies and the Uncontrolled Manifold Hypothesis

Fig. 1.3 An illustration of a

1.6 Feed-forward Control of Synergies: Anticipatory Synergy

Stability of actions is one of the important features of motor control, particularly

1.7 Synergies Within a Hierarchical Control Scheme

The idea of hierarchically organized control of movement is very old. Arguably,

Fig. 1.6 A perturbation

1.8 Equifinality in a Redundant System

1.9 What Now? Let Me Tell You What Now

Anatol G. Feldman, Nabil Ilmane, Samir Sangani and Helli Raptis

A. G. Feldman () · N. Ilmane · H. Raptis

© Springer Science+Business Media New York 2014 17

2.2 Alternative Views of Motor Control

The resetting mechanism represents an empirically based solution to the posture-

2.2.1 Alternative Views on the Role of Central Control Processes

2.2.2 Explanations of Position Sense Based on the PS Rule

The PS rule provides an explanation of PS in different motor tasks as well as of

2.3 Testing Alternative Hypotheses on PS

The prediction of the effort-based hypothesis of PS was tested by asking subjects

Fig. 2.2 The sense of position a 6

Reproduced position (F^)

40 Initial -40 -20 20 40°

Experiments were also conducted to test specific predictions of PS rule. According

2.4 Kinesthetic Illusions