Vision and Attention, 1st Edition

Visit the link below to download the full version of this book:

https://medipdf.com/product/vision-and-attention-1st-edition/

Click Download Now

Michael Jenkin Laurence Harris
Department of Computer Science Department of Psychology
York University York University
4700 Keele Street 4700 Keele Street
Toronto, Ontario M3J lP3 Toronto, Ontario M3J lP3
Canada Canada
[email protected] [email protected]

Cover illustration: The cover art shows frames taken from the film Gorilla Thump, which can be
found on the CD.
Cover and CD iIlustration © 1999, Daniel 1. Simons. Used with permission.

Library of Congress Cataloging-in-Publication Data

Vision and attention / edited by Michael Jenkin, Laurence Harris.
p. cm.
Includes bibliographical references and index.
ISBN 978-1-4684-9520-1
1. Attention-Congresses. 2. Visual perception-Congresses. 1. Jenkin, Michael
(Michael Richard McLean), 1959- II. Harris, Laurence (Laurence Roy), 1953-
QP405 .V574 2000
155.l4-dc2 1 00-059474

Printed on acid-frec paper.

© 2001 Springer Seienee+Business Media New York: except:
© Chapter 9, R. Rensink; © Chapter 10, D. Simons; © Chapter 14, K. O'Regan; and © AII sections
of the Galactic Web movie, 1. Howard.
Softcover reprint of the hardcover 1st edition 200 1
Originally published by Springer-Verlag New York, Ine. in 2001

This work consists of a printed book and a CD-ROM packaged with the book, both of which are
protected by federal copyright law and international treaty. The book may not be translated or
eopied in whole or in part without the written permission of the publisher Springer Seienee+
Business Media, LLC except for brief excerpts in connection with reviews or scholarly analysis.
For copyright information regarding the CD-ROM, please consult the printed information
packaged with the CD-ROM in the back of this publication, and which is also stored as a "readme"
file on the CD-ROM. Use of the work in connection with any form of information storage
and retrieval, electronic adaptation, computer software, or by similar or dissimi-
Iar methodology now known or hereafter developed other than those expressly granted in the CD-
ROM copyright notice and disclaimer information is forbidden.
The use of general descriptive names, trade names, trademarks, etc., in this publication, even if the
former are not especially identified, is not to be taken as a sign that such names, as understood by
the Trade Marks and Merchandise Marks Act, may accordingly be used freely by anyone. Where
those designations appear in this book and Springer-Verlag was aware of a trademark claim, the
designations follow the capitalization style used by the manufacturer.

Production managed by Jenny Wolkowicki; manufacturing supervised by Erica Bresler.

Photocomposed pages prepared from the editors' files .
.\ddiliunal maleriallulhis huuk can he downluaded from hllp:llexlras.sJlringer.emn.

9 8 7 6 5 432 1
ISBN 978-1-4684-9520-1 ISBN 978-0-387-21591-4 (eBook)
DOI 10.1007/978-0-387-21591-4
To HLM and CMO.
Preface
This book is based on a conference on Vision and Attention, the fifth conference
of the York Centre for Vision Research organized by I. P. Howard, D. M. Regan
and B. J. Rogers in June 1999 and sponsored by the Centre for Vision Research,
the Departments of Psychology and Computer Science, of York University.
The CD-ROM that accompanies this book contains colour imagery and video
clips associated with various chapters and the conference itself. The CD-ROM
is presented in HlML format and is viewable with any standard browser (eg.
Netscape Navigator or Microsoft's Internet Explorer). To view the videos on the
CD, you will need Quicktime, which is available free from Apple. To view the CD,
point your browser at the file index.htm on the CD. Please note that some of the
larger movies require a machine with t 28Mbytes of memory in order to play them
correctly.
The cover art shows frames taken from the film Gorilla Thump, which can
be found on the CD and which is Copyright 1999, Daniel 1. Simons. Used with
permission.
The York Vision Conference, and this book, would not have been possible with-
out the advice and support of Ian P. Howard, David Martin Regan and the Human
Performance in an Aerospace Environment Theme of the Centre for Research in
Earth and Space Technology (CRESTech). Behind any successful endeavour is the
person who really runs things, and none of this would have been possible without
Teresa Manini.

York University Michael Jenkin

October 2000 Laurence Harris
Contents
Preface vii

Contributors xv

1 Vision and Attention 1

Laurence R. Harris and Michael Jenkin

1.1 What Is Attention? . . . . . . . . . . . . . . . . . . . . . . . . I

1.1.1 Should "attention" be regarded as a discrete behaviour? . 3
1.2 Selective Visual Attention . . . . . . . . . . . . . . . . . . . . 3
1.2.1 What is selected? . . . . . . . . . . . . . . . . . . . .. 6
1.2.2 How is selection achieved? How much salience is due to
the sensory input itself and how much to higher processes? 8
1.2.3 What is the connection between selection and attention? . 10
1.2A Mechanisms of selective attention . . . . . . . . . . . .. 11
1.3 Parsing Attention. Is Attention Central to the Act of Seeing or is
it Merely a Servant Carrying its Master to the Right Place? 12
IA Directing Attention 13
1.5 Conclusions......................... 14

2 Shifts of Attention and Saccades Are Very Similar. Are They Causally
Linked? 19
Sally McFadden and Josh Wallman

2.1 Spatial Attributes of Attention 19

2.2 Coordinate Space of Focal Attention 21
2.3 Overt and Covert Orientation . . . . 22
2A Top-Down versus Bottom-Up Attentional Control 22
2.5 Shifting Attention . . . . . . . . . . . . . . . . . 23
2.6 Coupling Between Saccadic Eye Movements and Attentional Shifts 24
2.7 Adaptive Control of Saccadic Eye Movements. 27
2.8 Nature of the Error Signal. . . . . . . . . . . . . . . . . . . . ., 27
2.9 Are Shifts of Attention also Adaptable? . . . . . . . . . . . . . , 28
2.10 Might Attention Provide an Error Signal to Saccade-Gain Adap-
tation? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 31
x Contents

3 Contrast Gain, Area Summation and Temporal Tuning in Primate

Visual Cortex 41
Michael J. Hawken, Robert M. Shapley, Michael P. Sceniak, Dario L.
Ringach and Elizabeth N. Johnson

3.1 Introduction...... 41
3.2 Gain Control . . . . . . . . . . . 42
3.3 Contrast-Gain Control . . . . . . 43
3.4 Beyond the Classical Receptive Field. 46
3.5 Area Summation and Contrast . . . 47
3.6 Temporal Tuning and Contrast . . . 53
3.7 Temporal Tuning and Contrast in V I 53
3.8 Discussion.............. 57

4 Global Processes in Form Vision and Their Relationship to Spatial

Attention 63
Frances WilkinsonandHughR. Wilson

4.1 Introduction to the Ventral Visual Pathway . 63

4.2 Components of Intermediate Form Analysis 64
4.3 Changing Views ofV4 . . . . . . . . . . . 65
4.4 Evidence for Global Orientation Pooling in Human Vision 66
4.5 Neural Model for Configural Units . . . . . . . . . . . . . 68
4.6 Configural Units and Receptive Field Size . . . . . . . . . 69
4.7 Evidence Pointing to Configural Units in V4 in the Human Brain 71
4.8 Application of V4 Model Units to Faces 72
4.9 Selective Attention . . . 73
4.10 Summary and Overview . . . . . . . . 77

5 Visual Attention: The Active Vision Perspective 83

JohnM. Findlay and lain D. Gilchrist

5.1 Introduction. 83
5.2 Active Vision . . . 85
5.3 Reading...... 87
5.4 Scenes and Objects 91
5.5 Search . . . . . . . 93
5.6 Rethinking Covert Attention 97
5.7 Conclusion . . . . . . . . . 98

6 Complexity, Vision, and Attention 105

John K. Tsotsos

6.1 What Is Computational Complexity? 105

6.1.1 Some basic definitions . . . 106
6.1.2 Dealing with NP-completeness . 107
6.1.3 Vision and NP-completeness . . 108
6.2 Can Perception Be Modeled Computationally? 109
6.3 Visual Search . . . . . . . . . . . . . . . . . . III
 Contents xi

6.3.1 Definition.. 112

6.3.2 Theory ... 112
6.3.3 Implications 113
6.4 Complexity Level Analysis of Vision . 115
6.5 The Selective-Tuning Model of Visual Attention . 116
6.6 Conclusions.................... 123

7 Motion-Disparity Interaction and the Scaling of Stereoscopic Dispar-

~ m
Michael S. Landy and Eli Brenner

7.1 Cue Combination in Depth Perception . . . . . 130

7.2 Depth Scaling . . . . . . . . . . . . . . . . . . 131
7.2.1 Failures of depth constancy with stereo 132
7.2.2 Distance scaling of size, shape, and depth 132
7.3 Stereomotion Interaction for Depth Scaling 133
7.3.1 Why combine stereo and motion? 134
7.3.2 Evidence with a single object . . . 134
7.3.3 Two neighboring objects . . . . . . 136
7.3.4 Two objects and alternative computations 142
7.3.5 Two objects at unequal distances. 143
7.4 Summary . . . . . . . . . . . . . . . . . . . . . 147

8 Signal Detection and Attention in Systems Governed By Multiplicative

Noise 151
Christopher W. Tyler

8.1 Introduction......................... 151

8.2 Signal Detection Theory for Ideal and Non-ideal Observers 152
8.2.1 Overview of ideal observer analysis . . . . . . . . 152
8.2.2 The concept of probability summation . . . . . . . 154
8.2.3 Attentional summation in 2AFC experiments derives from
the difference distribution . . . . . . . . . . . . . . . . . 155
8.2.4 2AFC attentional summation with uncertainty within a
fixed attention window 158
8.3 Distraction Theory . . . . . . . . . . . . . . . . . . . . . . . . . 158
8.4 Effects of Multiplicative Noise . . . . . . . . . . . . . . . . . . . 160
8.4.1 Multiplicative noise makes the psychometric function shal-
lower . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
8.4.2 Dramatic probability summation with multiplicative noise 160
8.4.3 Suprabehavioral neural sensitivity and its implications for
attentional selection . . . . . . . . . . . . . . . . . . . . 161
8.4.4 Fully multiplicative noise introduces psychometric satu-
ration . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
xii Contents

9 Change Blindness: Implications for the Nature of Visual Attention 169

RonaldA. Rensink

9.1 Visual Attention: Role in Scene Perception. 170

9.1.1 Change blindness . . . 170
9.1.2 Coherence theory . . . 171
9.1.3 Virtual representation. 174
9.2 Visual Attention: Mechanisms 177
9.2.1 Methodology . . . . . 177
9.2.2 Experimental results: Capacity . 179
9.2.3 Implications for attentional mechanisms. 181
9.3 Concluding Remarks . . . . . . . . . . . . . . . 185

10 The Role of Expectations in Change Detection and Attentional Cap-

ture 189
Daniel J. Simons and Stephen R. Mitroff
10.1 Change Blindness . ....... 190
10.1.1 The intentional approach 191
10.1.2 The incidental approach 192
10.1.3 Summary ..... . . 196
10.2 Attentional Capture . .... . . 197
10.2.1 The intentional approach 198
10.2.2 The incidental approach 199
10.2.3 Implications .203
10.3 Conclusions . . . . . ...... .204

11 Attention, Eye Movements, and Neurons: Linking Physiology and Be-

havior 209
Narcisse P. Bichot

11.1 Introduction ...... .209

11.2 Attention and Saccades .211
11.3 Frontal Eye Field ... · 213
11.4 Bottom-Up Influences on Visual Selection .214
11.5 Top-Down Influences on Visual Selection · 217
11.6 Conclusions . . . . . . . . . . . . . . . . .225

12 Vision and Action in Virtual Environments: Modern Psychophysics in

Spatial Cognition Research 233
HeinrichH. BiilthoffandHendrikA. H. C. van Veen

12.1 Introduction . . . . . . · 233

12.2 Biological Cybernetics .235
12.3 Enabling Technologies .239
12.4 Stimulus Control ... · 241
12.5 Stimulus Relevance . . .245
12.6 Spatial Cognition in VEs .246
 Contents xiii

12.7 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . 246

13 Selective Feature-Based Attention Directed to a Pair of Lines: Psy-

chophysical Evidence and a Psychophysical Model 253
David Regan and Radha P. Kohly

13.1 Does the Visual System Contain Long-Distance Comparators with

Orthogonal Orientation Difference and Mean-Orientation Labels? 261
13.2 Does the Visual System Contain Long Distance Comparators Whose
Outputs Carry Orthogonal Mean-Location and Separation Labels? 266
13.3 Does the Visual System Contain Long-Distance Comparator Mech-
anisms Whose Outputs Carry Orthogonal Orientation Difference,
Mean Orientation, Mean Location and Separation Labels? . . . . 269
13.4 How Do Discrimination Thresholds for Orientation Difference,
Mean Orientation, Separation, and Relative Mean Location Vary
as a Function of Contrast? . . . . . . . . . . . . . . . . . . . . . 270
13.5 Attentional Implications and a Psychophysical Model . . . . . . . 270
13.5.1 Long-distance comparators whose outputs signal orthog-
onally four stimulus attributes . . . . . . . . . . . . . . . 270
13.5.2 Possible role of long-distance comparators in other psy-
chophysical findings . . . . . . . . . . . . . . . . . . . . 273
13.5.3 Possible role of long-distance comparators in everyday
vision . . . . . . . . . . .273
13.5.4 Attentional implications .276

14 Thoughts on Change Blindness 281

J. Kevin O'Regan

14.1 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . .. .281

14.2 Thoughts on Normal Viewing: Where and What. . . . . . .. . 282
14.2.1 Comments on the central/marginal interest distinction . 282
14.2.2 Locations, objects, or aspects? . . . . . . . . . . .. . 283
14.2.3 Looking without seeing . . . . . . . . . . . . . . . . 284
14.3 The "Where" and "What" Components of Change Detection . 284
14.3.1 Transients tell "where" . . . . . . . . . . . . . . .. . 284
14.3.2 Memory tells "what" . . . . . . . . . . . . . . . .. . 285
14.3.3 A counterintuitive prediction for change blindness in nor-
mal viewing. . . . . . . . . . . . . . . . . . . . . 285
14.4 Thoughts on Disruptions . . . . . . . . . . . . . . . . . .. . 286
14.4.1 Transients as masks and transients as distractors. .. . 286
14.4.2 Measuring diversion with the mudsplash experiment . 287
14.4.3 The (critical?) number of diversions . 287
14.4.4 Proximity of the transient .. . 288
14.4.5 More questions on diversions . . . . 288
14.4.6 A transient pop-out task? . . . . . . . 289
14.4.7 Does local masking interfere with the "what" component? 289
xiv Contents

14.4.8 Prediction for very slow changes . . . . . . . . . . . . . . 290

14.4.9 Estimating the "what" component of change detection us-
ing the masking rectangle experiment . . . . . . . 290
14.5 OtherIssues Concerning the Theory . . . . . . . . . . . . 291
14.5.1 A prediction for the moment of change detection . 292
14.5.2 A prediction: seeing illusory appearances . 292
14.5.3 A special role for layout? . . . . . . . . . . . . . . 293
14.5.4 Implicit knowledge of changes? . . . . . . . . . . 294
14.5.5 Other frameworks for explaining change blindness . 295
14.5.6 Relation to early literature on partial report . 295
14.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . 296

Author Index 303

Subject Index 313

Contributors
Narcisse P. Bichot, Vanderbilt Vision Research Center, Department of Psycholo-
gy, Vanderbilt University, Nashville, TN 37240, USA.
[email protected]

Eli Brenner, Vakgroep Fysiologie, Erasmus Universiteit, Postbus 1738,3000 DR

Rotterdam, The Netherlands.
[email protected]

Heinrich H. BiilthotT, Max-Planck-Institute for Biological Cybernetics,

Spemannstra;3e 38, 72076 Tiibingen, Germany.
[email protected]

John M. Findlay, Centre for Vision and Visual Cognition, Department of

Psychology, University of Durham, South Road, Durham, DHI 3LE, England.
[email protected]

lain D. Gilchrist, Department of Experimental Psychology,

University of Bristol, 8 Woodland Road, Bristol, BS8 I TN, England.
I.D [email protected]

Laurence R. Harris, Department of Psychology, York University,

4700 Keele Street, Torotno, Ontario, M3J IP3, Canada.
[email protected]

Michael J. Hawken, Center for Neural Science, New York University,

4 Washington Place, New York, NY 10003.
[email protected]

Michael Jenkin, Department of Computer Science, York University,

4700 Keele Street, Toronto, Ontario, M3J IP3, Canada.
[email protected]

Elizabeth N. Johnson, Center for Neural Science, New York University,

4 Washington Place, New York, NY 10003, USA.
[email protected]
XVI Contributors

Radha P. Kohly, Department of Biology, York University, BSB, Room 375,

4700 Keele Street, Toronto, Ontario, M3J IP3, Canada.
[email protected]

Michael S. Landy, Department of Psychology and Center for Neural Science,

New York University, 6 Washington Place, New York, NY 100m, USA.
[email protected]

Sally McFadden, Department of Psychology, University of Newcastle,

Newcastle, NSW, Australia.
[email protected]

Stephen R. Mitroff, Department of Psychology, Harvard University,

33 Kirkland Street, #820, Cambridge, MA, 02138, USA.
[email protected]

J. Kevin O'Regan, Laboratoire de Psychologie Experimentale, CNRS, EHESS,

EPHE, Universite; Rene Descartes, 71, avenue Edouard Vaillant, 92774
Boulogne-Billancourt Cedex, France.
[email protected]

David Regan, Department of Biology and Psychology, York University,

4700 Keele Street, Toronto, Ontario, M3J IP3, Canada.
dregan @yorku.ca

Ronald A. Rensink, Cambridge Basic Research, Nissan Research &

Development, Inc., 4 Cambridge Center, Cambridge, MA 02142-1494, USA.
[email protected]

Dario L. Ringach, Center for Neural Science, New York University,

4 Washington Place, New York, NY 10003, USA.
dario@uc\a.edu

Michael P. Sceniak, Center for Neural Science, New York University,

4 Washington Place, New York, NY 100m, USA.
[email protected]

Robert M. Shapley, Center for Neural Science, New York University,

4 Washington Place, New York, NY 100m, USA.
[email protected]

Daniel J. Simons, Department of Psychology, Harvard University,

33 Kirkland Street, #820, Cambridge, MA, 02138, USA.
[email protected]
 Contributors xvii

John K. Tsotsos, Department of Computer Science and Centre for Vision

Research, York University, 4700 Keele Street, Toronto, Ontario, M3J I P3, Canada.
[email protected]

Christopher W. Tyler, Smith-Kettlewell Eye Research Institute, at California

Pacific Medical Center, 2318 Fillmore Street, San Francisco, CA, 94115, USA.
[email protected]

Hendrik A. H. C. van Veen, TNO Human Factors Research Institute,

P.O. Box 23, 3769 ZG Soesterberg, The Netherlands.
van [email protected]

Josh Wallman, Department of Biology, City College, City University of New

York, New York, NY, USA.
wallman @sci.ccny.cuny.edu

Frances Wilkinson, Department of Psychology, Atkinson College,

York University, 4700 Keele Street, Toronto, Ontairo, Canada, M3J IP3.
[email protected]

Hugh R. Wilson, Department of Biology, York University, 4700 Keele Street,

Toronto, Ontario, Canada, M3J 1P3.
[email protected]
1

Vision and Attention

Laurence R. Harris
Michael Jenkin

The term "visual attention" embraces many aspects of vision. It refers to processes
that find, pull out and may possibly even help to define, features in the visual
environment. All these processes take the form of interactions between the observer
and the environment: attention is drawn by some aspects of the visual scene but
the observer is critical in defining which aspects are selected.
Although this book is entitled Visual Attention, none of the processes of "visual"
attention are exclusively visual: they are neither driven exclusively by visual inputs
nor do they operate exclusively on retinal information. In this introductory chapter,
we outline some of the problems of coming to grips with the ephemeral concept
of "visual attention."

1.1 What Is Attention?

Attention implies allocating resources, perceptual or cognitive, to some things at
the expense of not allocating them to something else. This definition implies a limit
to the resources of an individual such that they cannot attend to everything at once,
all the time. In one sense this is obvious in that the senses already provide a filter.
The visual system, for example, does not tell us about what is happening behind
us or in the infrared part of the spectrum. But attention refers to selection from the
array of information that is arriving at the brain and is potentially available.
The term "attention" suffers from the fact that it is a word in both common and
scientific usage and the common and scientific meanings only loosely overlap.
Furthermore, both common-parlance and scientific "attention" each have several
different meanings. The common usage has an implication of urgency and alertness
well - illustrated by the single-word sentence substitution "ATTENTION!!." In
scientific usage, we do not necessarily want to incorporate this sense of the unusual,
although we cannot avoid the tint.
Although closely related to "active vision" (see Aloimonos et aI., 1988, and
also Harris and Jenkin, 1998), attention describes the requirement that the visual
system attends at least at a computational level (although not necessarily a physical
level) to different events in the visual field. Active systems may require attentional
mechanisms in order to direct their sensors at different salient events, but attentive
systems are not necessarily active in a physical way.
2 Laurence R. Harris and Michael Jenkin

The word "attention" has ancient roots that link it with concepts of general
alertness and conscious receptivity on the one hand and with concentration and
focusing on the other. Like other words imported from common parlance, for
example "stress," "attention" can perhaps only be meaningfully considered if ac-
companied by an adjective. Just as there is no such thing as unqualified "stress"
physiologically, it might be useful to start from the stance that there is no such
thing as unqualified attention. What types of "attention" are there? We suggest four
distinct types:

• Selective attention. At anyone time we seem to attend to and be aware

of only some aspects of the sensory input. What is selected is sometimes
determined by the demands of a particular task, such as concentrating on a
tool while it is being employed. But selective attention is not only activated
by interest in a goal. Although it is obvious that we are selectively attending
while actively searching for those lost keys, selection is a feature central to
the act of seeing anything. It is impossible, with our meagre brain equipment,
to process the whole of the retinal image. Selectively attending to something
implies that the feature being attended to has already been defined, whether
this is a basic attribute such as a colour, a visual direction, or an actual object.

• Parsing attention. Attention might be a part of the process of recognizing

objects and separating them from their backgrounds. It has been suggested
that the act of attending to something is critical to the binding together of
the the various features that define a perceptual object.

• Directing attention. When something happens, a primitive reflex system

instinctively orients us towards it. This is an emergency, interrupting system,
sometimes called attentional capture, which overrides normal behaviour
when something potentially important or dangerous demands immediate
analysis. But directing attention is also a more gentle, omnipresent drive
used for exploring the environment or for maintaining attention on an object
while carrying out a task. There might be a continuum between emergency
interrupt, normal exploration, and maintaining attention or perhaps these
behaviours represent quite different control systems.

• Alertness attention. It seems intuitive that a certain level of arousal is nec-

essary for normal perceptual processes to operate. Perception is normally
associated with a state of being awake and responsive to sensory input: a
state in which behaviours are planned and carried out and in which we are
interacting with the environment. But how much undirected arousal is in fact
necessary for perception? Some perception can occur while in a daydream or
even while asleep (Mack and Rock, 1998). Alertness attention may operate
by modulating other forms of attention.
 1. Vision and Attention 3

1.1.1 Should" attention" be regarded as a discrete behaviour?

Should attention be regarded as a discrete behaviour that can occur independently
of other behaviours? Can the act of attending to something be treated analogously
to the discrete act of picking something up between the thumb and forefinger or
directing the eyes to converge on a given point? It is possible for finger control
or auditory localization to occur quite independently of any other behaviour. Can
"attending" be regarded similarly or is it unable to exist independently of other
behaviours such that it cannot be regarded as having an independent existence?
Perhaps attention is just a modulation of other behaviours.
This is a significant question when seeking to understand attention because
if attention can be regarded as a discrete behaviour, then it might be expected
to have special brain mechanisms and perhaps brain sites, devoted to it. If, on
the other hand, it exists only as a modulation of other behaviours, then no such
specialized sites would be expected. Should we seek control systems in the brain
that specifically underlie the allocation of attention in the same way as we expect
to find systems devoted to auditory localization or locomotion control?
This debate often takes place in the context of whether there is an executive
controller. The controller would not be directly involved in any sensory processing
but only in the control or modulation of such target systems when it applied atten-
tion. Many parts of the visual sensory system show clear modulation in response to
attention (e.g., parietal cortex, superior colliculus, even lateral geniculate and stri-
ate cortex - see chapter 11 by Bischot), but these areas also show stimulus-related
activity when attention is not involved. One area, the cingulate cortex however
seems to show activity exclusively when attention is required. Hence, it is a strong
candidate for the location of an executive controller (Badgaiyan and Posner, 1998;
Carter et aI., 1998). When a task that previously required attention becomes auto-
matic through practice, activation of the cingulate cortex is no longer found (Frith
et aI., 1991).
This therefore gives us a model for what attention is and how it is implemented
by the brain. The model is driven by a central executive controller, located at least
partially in the ancient limbic association areas, that either makes or at least ad-
ministers decisions. The decision about what is to be attended is then implemented
by selective modulation of the activity or sensitivity of the sensory area or areas
that process the desired attribute. The modulation is as specific or fine-tuned as re-
quired for the task that the controller is executing. The resulting increased activity
in the processing system then gives that coded attribute a competitive advantage
or even a flag that leads to preferential treatment. Why might selected attributes
need preferential treatment?

1.2 Selective Visual Attention

The amount of information falling on the retina at anyone time is truly vast and
most of it is of no survival value at all (Fig. 1.1). Look at the heavens and all the
4 Laurence R. Harris and Michael Jenkin

FIGURE 1.1. Bookshop scene. The amount of information falling on the retina at anyone
time is enormous. A selective process is required not only to acquire useful information
but in order to be able to see at all. Often the selection will depend on the task at hand. A
person just looking at this photo is unlikely to be able to report on the type of lighting, for
example, unless attention was specifically drawn to that type of information.

visible stars within the eyes' visual fields are imaged on the retina at once. Look
at a tree with its complex and detailed branching pattern, much of which can be
resolved at the same time. Look at this page where all the letters are focused on
the retina. Most of the potentially smothering information arriving at the retina is
lost by the limited resolving power of the retina outside the central few degrees of
the fovea. But even with this blessed filter, there remains an enormous amount of
information that could, theoretically, be extracted. Normally, the gaze only stays
at one point for about 300 msecs (Yarbus, 1967), implying that the information
needed is extracted in this small time before the gaze is shifted again.
Visual attention is an essential component of machine-vision systems. Figure 1.2
shows the Eyes 'n Ears sensor (Kapralos et aI., 2000). This is an omnidirectional
sensor based on the Paracamera. Consisting of both audio and video components,
the Eyes 'n Ears sensor's visual input is generated by a video camera mounted
vertically and directed at a semispherical mirror. The resulting view (shown in
Fig. 1.2b), provides a 3600 view of the environment. This view can be unwarped
to provide a perspective panoramic view (Fig. 1.2c). As in the biological example
of Figure 1.1, it is not practical for a machine to attend to each and every pixel in
the view. Many are uninteresting for the specific task at hand. For the Eyes 'n Ears
sensor, locations in which the image changes are of interest, so the sensor uses an
attention-directing mechanism based on attending at image locations that corre-
spond to image differences. Figure 1.3 shows the system in operation. Figure 1.3c
is a retinotopic difference image between the image frames shown in Figs. 1.3a
1.3c. Brighter locations correspond to more interesting image locations.
At anyone time, the brain just does not need to know very much about the world
- the details of the branching structure of a tree are usually of no use and therefore