Thematic Group 1

EUROPEAN RESEARCH IN MATHEMATICS EDUCATION III

THE NATURE OF MATHEMATICS AS VIEWED

FROM COGNITIVE SCIENCE
Laurie D. Edwards
ST. Marys College of California
[email protected]
Abstract: The theory of embodied cognition proposes that the source of many
ideas, including mathematical ones, lies in the common physical experience of
human beings. Through mechanisms such as conceptual mappings and
metaphors we are able to construct new ideas and understandings that are more
abstract, in the sense of being less tied to physical experience. The purpose of
this paper is to review work on the theory of embodied mathematics and to
describe an application of the theory to understanding how transformation
geometry is learned.
My goal in this paper is to characterize one perspective on the nature of mathematics,
drawing primarily from recent cognitive science and in particular the work of George
Lakoff and Rafael Nez (Lakoff & Nez, 2000), and to offer an example of the
application of this perspective in my own research. A theory of how human beings
think and understand mathematically should be able to account for both the correct
and consistent mathematical ideas of professional mathematicians, and for the variety
of ideas that diverge from this body of shared knowledge. In addition, unless
mathematics education wants to set mathematics aside as a special case, such a theory
should draw on conceptual primitives and mechanisms that are not unique to
mathematical thinking. This paper presents a perspective that addresses the nature of
mathematics by drawing from work by a number of researchers in contemporary
cognitive science, including Lakoff and Nez (2002), Johnson (1987), Varela,
Thompson and Rosch (1991), Dehaene (1997), Fauconnier and Turner (2002) and
others.
The theory of embodied mathematics
The theory of embodied mathematics states in essence that:
(1) mathematics emerges through the interaction of the mind with the world,
and
(2) that each of these elements (the mind and the world) offers constraints on
what kind of mathematics human beings are able to be constructed.
A specific aspect of the theory describes a conceptual mechanism that has been used
to understand thought and language in areas outside of mathematics, that of
conceptual metaphor. As Lakoff and Nez state,
L. D. Edwards

Thematic Group 1

EUROPEAN RESEARCH IN MATHEMATICS EDUCATION III

(3) "A large number of the most basic, as well as the most sophisticated,
mathematical ideas are metaphorical in nature" (Lakoff and Nez, 2000, p.
364).
What Lakoff and Nez refer to as the theory of mind-based mathematics does not
deny the objectivity of the material world; in fact, it accords to this world a
constraining influence on the kind of mathematics that humans are able to construct.
The world behaves consistently, fairly reliably, predictably - when you put a
collection of two objects together with another collection of two objects, you are
reliably going to end up with a collection of four objects, not one of five or three.
Human infants behave as if they have some awareness of this kind of predictable
behavior, though in their case, this awareness extends only to the very smallest of
collections, namely, single objects combined with single objects (cf., Dehaene, 1997).
Evolutionarily, it seems to have been important for humans to be able to subitize, to
keep track of small collections of objects, and this provides part of what the mind
brings to the picture. There is a small amount of innate or virtually innate arithmetic
that humans share with a few other kinds of animals, for example, chimps and parrots
(Dehaene, 1997). But this starting point is very restricted, and one of the questions
addressed by the theory of embodied mathematics has to do with identifying the
conceptual mechanisms and building blocks that allow humans to go from this
starting point to develop the incredible richness of contemporary and historical
mathematics.
Conceptual mechanisms and primitives
In addressing the nature of mathematical thought, Lakoff and Nez offer a set of
mechanisms and cognitive primitives drawn from existing work in cognitive
linguistics and embodied cognition. These primitives and mechanisms include
prototypes, image schemas, aspectual concepts, conceptual metaphor, conceptual
blends, and metonymy (for details, see Lakoff, 1987 and Lakoff & Nez, 2000).
Conceptual metaphor is an important (but not the only) mechanism proposed to
account for how mathematical understandings are connected to the world and to each
other. Conceptual metaphor is an unconscious mapping between a well-understood
source domain to a less-well-understood target domain, one which carries with it the
inferential structure of the source domain, thus allowing an understanding of the
target domain to be constructed. A non-mathematical example might be the way in
which research is conceptualized as a process of physical construction: we speak of
someone's work as having a firm foundation, or as extending previous work, even
when there is clearly no actual process of physical building going on. Within
mathematics, a simple example of conceptual metaphor would be the "arithmetic as
object collection" metaphor, where our common, embodied experience of grouping or
collecting objects serves as the source domain for constructing the arithmetic of
natural numbers.

L. D. Edwards

Thematic Group 1

EUROPEAN RESEARCH IN MATHEMATICS EDUCATION III

It should be clarified that not all conceptual mappings draw from direct
physical experience, or have to do with the manipulation of objects. In fact, only the
most basic level of mathematics is constructed from metaphors that link to physical
experience. The metaphors utilized at this level are called "grounding metaphors."
The majority of the concepts and processes of mathematics, according to the theory,
are constructed through mappings between and among existing mathematical
domains. An example of this kind of linking metaphor will be given later, when space
is conceptualized in terms of sets of points. A third type of metaphor, not based on
physical experience, is the redefinitional metaphor, defined as "metaphors that
impose a technical understanding replacing ordinary concepts" (Lakoff & Nez,
2000, p. 150).
Conceptual metaphor is only one of the building blocks proposed in the theory
of embodied mathematics. Conceptual blends are another. Blends consist of
mappings that draw from more than one source domain or element to allow the
construction of a target domain which is isomorphic to neither of the sources, but
which draws from the inferential structure of each. An example would be the
"numbers as points on a line" blend, where drawing from previously-constructed
understandings of both numbers and lines, new entities, "number-points" are created
that have characteristics of both.
An embodied perspective on the learning of transformation geometry
The theory of embodied cognition can provide a framework for understanding how
both children and adults learn initial concepts within the domain of transformation
geometry. In a series of studies beginning in 1989, I worked with eleven-to-fifteen
year old students using a computer environment I designed for exploring
transformation geometry (see Figure 1). Later, with Rina Zazkis, I investigated adult
undergraduates' learning of the same subject. In the first, most extensive study, I
introduced three euclidean transformations, translation, rotation and reflection, to a
whole class of sixth-grade (eleven- to twelve-year-old) students, utilizing sheets of
overhead transparencies to illustrate them and to elicit the students' own description
of these motions. Then, I worked with twelve of these students for a period of five
weeks, videotaping their interactions with the transformation geometry microworld
(Edwards, 1991, 1992). A similar study was carried out two years later with a group
of ten high school students (fifteen-year-olds). In this study, the students were
introduced to the transformations in pairs, rather than in a whole group, using
drawings on a sheet of paper (Edwards, 1997). Finally, Rina Zazkis and I examined
the responses of fourteen college undergraduates, first to a paper-and-pencil task
requiring them to predict the outcome of several transformations of the plane, and
then, through videotapes of their use of a modified version of the microworld
(Edwards & Zazkis, 1993). Although there were methodological differences across
these studies, and some differences in specific results, overall, the responses of all of
the participants, whether middle-school, high-school or adult, were remarkably
L. D. Edwards

Thematic Group 1

EUROPEAN RESEARCH IN MATHEMATICS EDUCATION III

similar. In particular, the students all seemed to have had the same initial expectations
of how the transformations would work, and they made the same kinds of errors. This
raises the question: why is there such consistency in how students, of various ages,
learn transformation geometry? My initial expectation was that older students and
adults would be less subject to "misconceptions" about geometric transformations and
better able to carry out independent investigations, due to the more advanced state of
development of their mathematical thinking. Yet this was not the case. The question
of why students of various ages respond in a similar way to a "new" mathematical
topic can, I believe, be productively investigated utilizing the theory of embodied
mathematics. To do so will require examining the domain of transformation geometry
from the perspective of contemporary mathematicians, and from the perspective of a
learner who is meeting the domain, in a formal sense, for the first time.

Figure 1: Transformations in the computer microworld

Transformation geometry from the perspective of contemporary mathematics

From the point of view of contemporary mathematicians, geometric transformations
are mappings of the plane, and a "geometry" is "a study of the properties (expressed
by postulate, definitions, and theorems) that are left invariant under a group of
transformations" (Meserve, 1955, p. 21). This definition is in contrast to the notion of
geometry current before the end of the 19th century. Prior to Felix Klein's Erlangen
program, geometry was seen as the investigation of points, lines, planes, and threedimensional space, utilizing the tools of deductive logic and formal proof. The latter
elements, logic and proof, remain central to all of mathematics, including
transformation geometry. However, the Erlangen program, which changed the focus
of geometry from specific objects with concrete, visualizable referents (points, lines,
L. D. Edwards

Thematic Group 1

EUROPEAN RESEARCH IN MATHEMATICS EDUCATION III

planes, etc.), to notions of invariance, group theory, and mappings, is an example of

what Lakoff and Nez call "the discretization program" which has permeated
mathematics since the late 19th century. This evolution within the discipline
established a new conception of space (and of the plane), one that differs
substantially from how space is conceptualized outside of formal mathematics.
Briefly, prior to the 19th century, and outside of formal mathematics today, space (and
planes, and lines) are conceptualized as naturally continuous, and as the background
settings within which objects are located. Space, planes, and lines exist independently
from any objects that are located in or on them, for example, geometric figures.
Points are located at particular places in space, or on a plane or line (Lakoff & Nez,
2000, p. 260); however, points do not exist independently of the places where they
are located.
The discretization program was based on the metaphorical reconceptualization
of numbers as points, in which discrete numbers are associated with discrete points
on a line (and vice versa). In conjunction with the matching of numbers with discrete
symbols, this made possible the progressive arithmetization of mathematical activity
involving space or the plane, as in analytic geometry or calculus. The discretization
program is based on a central metaphor, as follows (ibid., p. 263):
A Space is a Set of Points
Source Domain
A Set with Elements

Target Domain
Naturally-Continuous Space
with Point-Locations

A set
example, a line,

-> An n-dimensional spacefor

a plane, a 3-dimensional space

Elements are members of the set

-> Points are locations in the space

Members exist independently of

space

-> Point-locations are inherent to the

the sets they are members of

they are located in

Two set members are distinct if

they are different entities

-> Two point-locations are distinct if

they are different locations

Relations among members of the set -> Properties of the space

Under this metaphor, the objects of traditional geometry are radically
reconceptualized. As stated in Lakoff and Nez, "There is nothing inherently spatial
about a 'space.' What are called 'points' are just elements of the set of any sort...Like
any members of sets, points exist independently of any sets they are in. Spaces,
L. D. Edwards

Thematic Group 1

EUROPEAN RESEARCH IN MATHEMATICS EDUCATION III

planes and linesbeing setsdo not exist independently of the points that constitute
them...A geometrical figure, like a circle or a triangle, is a subset of the points in a
space, with certain relations among the points" (ibid., p. 263).
The transformation geometry microworld was built based on this "modern"
understanding of geometry. The transformations are intended to be understood as
mappings of the whole plane. The geometric figures that appear in on the computer
screen are simply specific subsets of points of the plane that serve to illustrate,
graphically, the effects of a given transformation. When the middle school students
were first shown the transformations, they were asked to describe what was the same
and what was different before and after each transformation. Thus, the notion of
invariance was introduced explicitly into the context of the activity. A task used near
the end of the study with the middle and high-school students required them to find
the symmetries of a set of geometric figures and sort them into groups having the
same symmetries. In general, the goal was to involve the students in working with
geometric transformations the way that a mathematician would, as mappings of the
plane, rather than as simple motions of geometric figures.
Transformation geometry from the learner's perspective
Although the microworld may have been designed with a contemporary mathematical
perspective on geometry in mind, those who use the software as a learning
environment may interpret it in a quite different way. That is, the software tool is a
different instrument to the designer and to the learner, who have different
expectations about how it will work and different interpretations of what they see on
the computer screen (Lagrange, Artigue, Laborde & Trouche, 2001). The learners
who worked with the transformation geometry microworld, whether young people or
adults, brought to it not the mathematician's commitment to the power of formalism,
but instead, a set of intuitions honed through activity and experience in the physical
world. In other words, their understanding of geometry was an embodied one. This is
supported by examining the nature of their interpretations and "errors" in the
microworld.
The first design of the microworld included only the most general version of
three euclidean transformations, translation, rotation and reflection. To use the
REFLECT command, for example, you need to specify the mirror line "over which"
the plane is reflected; to use ROTATE, you need to enter the location of a center
point and an angle amount for the rotation. When the microworld was pilot-tested,
however, many of the students had difficulty with these two commands, and so
simpler, "local" versions were implemented, versions that worked only with the block
letter "L" shape that was the default figure shown on the screen. Specifically, a FLIP
command was created that always reflected across the long "back" of the "L," and a
PIVOT command was created that always turned around the bottom left corner of the
L-shape. These specialized commands were introduced as easier versions of
L. D. Edwards

Thematic Group 1

EUROPEAN RESEARCH IN MATHEMATICS EDUCATION III

REFLECT and ROTATE, and they made it possible for the students to begin working
with the microworld. However, the learning activities created in the microworld
required that the students come to understand the more general commands.
In learning about the general transformations, a number of students exhibited
what I, at that time, referred to as a conceptual "bug" regarding rotation. Instead of
seeing rotation as mapping all the points of the plane around any center point on the
plane, these students expected that a ROTATE command would slide the shape to the
given center point, and then pivot it around it (see Figure 2). In other words, they had
a hard time seeing rotation as occurring "at a distance" from the object. This
"misconception" cropped up among approximately a third of the students in the first
study with middle-schoolers, and among a smaller number of high school and adult
students. In contrast to other mistakes made when using the microworld, the students
did not find it easy to correct their own understanding by further exploration, and I
generally needed to step in and explain how the ROTATE command actually worked.

Figure 2: The rotate "bug"

I first thought about the rotate "bug" simply as a misconception, something

idiosyncratic occurring with a few specific students that could be overcome through
better instruction or through redesigning the microworld. Later, I thought that the
source of the "misconception" was social. That is, the conventions that
mathematicians had developed for this concept, derived from the values of coherence
and consistency, were not obvious to the students, but that once these conventions
were explained, the students would have no problem with rotation.
However, after finding that the "rotation bug" showed up in sixth graders, high
school students, and adults, and that it was not easy to teach the "correct" version of
the transformation, it seemed that something more fundamental was going on. These

L. D. Edwards

Thematic Group 1

EUROPEAN RESEARCH IN MATHEMATICS EDUCATION III

results from the studies of transformation geometry could be accounted for within an
embodied perspective.
Transformations: motions or mappings?
Data on how students interacted with the computer microworld indicate that the
conceptualization of transformation geometry of contemporary mathematicians and
that held by learners are quite different. The students' responses to the computer
microworld grow out of an embodied understanding of space (and the plane) as
naturally continuous. When they look at the graphical window of the microworld,
they see a continuous (but normally "invisible") plane that provides a background, on
which geometric figures are located. The commands (the geometric transformations)
are used to move the geometric figures around on top of this plane. By contrast, for
the mathematician, the plane consists of an infinite number of discrete points, and
transformations are simply mappings of those points that preserve some properties
and change others (euclidean transformations preserve size and shape; affine
transformations preserve area, and so on). From the point of view of the discipline of
mathematics as it has evolved since the discretization program, there are no motions
involved in transformations of the plane. These two conceptualizations are contrasted
in the table below:
Learner's Transformations

Mathematician's Transformations

The plane is an empty, invisible The plane is a set of points

background
Geometric figures are subsets of points of
Geometric figures sit on the plane
the plane
Transformations are physical motions

Transformations are mappings of all the

of geometric figures on top of the points
plane
of the plane
The "objects" of geometry are points, The "objects" of geometry are groups of
lines,
Transformations.
circles, triangles, etc.
In fact, in introducing the computer microworld, I would emphasize that
transformations were mappings of the whole plane, and that the "L" shape shown on
the screen was simply a subset of those points, shown as dark dots in order to be able
to easily see the effects of a given transformation, Yet this explanation did not
prevent students from consistently talking about transformations as movements, nor
from interpreting the general mapping ROTATE as a sequence of simpler, local
motions, with rotation always occurring around a point that's a part of the object.

L. D. Edwards

Thematic Group 1

EUROPEAN RESEARCH IN MATHEMATICS EDUCATION III

The learner's general understanding of transformations, as well as the rotate

"bug" in particular, both arise from our embodied experience. As human infants, our
first physical actions of turning always occur around a center point located within our
own bodies. Our experience of the space we move through is as continuous. When
we interact with objects on a table (a physical analogue to the mathematician's plane),
they sit on the table, they are not embedded "subsets" of it. Other early experiences
with rotation, including the turning of wheels, phonograph records, and the like, are
also of this "local" nature; the rotation occurs around a center within the object itself.
Even when there is a "turning" around a center point at a distance, for instance, on a
swing, there is always some kind of physical connection between the object and the
center of the rotation. Thus, it is difficult for students to understand a general
mapping of the plane around an arbitrary center point, detached from the geometric
figure that constitutes the object of their attention.
This tension between the mathematician's understanding of transformations as
mappings, implemented in the microworld, and the embodied, natural understanding
of motion that the learners brought to the experience is the source of their
"misconceptions." These "misconceptions" are in actuality, conceptions that are
adaptive and functional outside of the context of formal mathematics.
Conclusion
The purpose of this paper was to briefly review the central elements of the theory of
embodied cognition as applied to understanding mathematics, and to examine a
specific case in which embodiment can clarify a persistent student "misconception."
The idea that mathematics is constructed from conceptually simpler building blocks
seems to me to be relatively uncontroversial within the mathematics education
research community. The controversy, or perhaps, better, the scholarly enterprise,
comes in attempting to build a coherent theory about how this occurs, to discover the
nature of the mechanisms and entities involved. The theory of embodied
mathematics, as demonstrated in the case of transformation geometry, can provide a
coherent and plausible account of how mathematics is built from the conceptual raw
material available to us as human beings with bodies, acting in the physical world.
References
Davis, P. &
Boston.

Hersch, R. The mathematical experience. Cambridge: Birkhuser

Dehaene (1997). The number sense: How the mind creates mathematics. Oxford:
Oxford University Press.
Edwards, L. (2002). The nature of mathematics: A personal journey. In A. Cockburn
& E. Nardi (Eds.), Proceedings of the 26th International Conference for the

L. D. Edwards

Thematic Group 1

EUROPEAN RESEARCH IN MATHEMATICS EDUCATION III

Psychology of Mathematics Education , Vol. I. (pp. 149-153). Norwich, UK:

University of East Anglia.
Edwards, L. D. (1997). Exploring the territory before proof: Students' generalizations
in a computer microworld for transformation geometry. International Journal
of Computers for Mathematical Learning, 2,187-215.
Edwards, L. D. (1992). A comparison of children's learning in two interactive
computer environments. Journal of Mathematical Behavior, 11, (1), 73-82.
Edwards, L. D. (1991). Children's learning in a computer microworld for
transformation geometry. Journal for Research in Mathematics Education, 22
(2), 122-137.
Edwards, L. & Nez, R. (1995). Cognitive science and mathematics education: A
non objectivist view. In D. Carraher & L. Meira (Eds.), Proceedings of the 19th
International Conference for the Psychology of Mathematics Education , Vol. I.
(pp. 79-87). Recife, Brazil: Universidade Federale de Pernambuco.
Edwards, L. D. & Zazkis, R. (1993). Transformation geometry: Naive ideas and
formal embodiments. Journal for Computers in Mathematics and Science
Teaching, 12 (2), 121-145.
English, L. & Goldin, G. (2002). The nature of mathematics as viewed from
mathematics education research (Introduction). In A. Cockburn & E. Nardi
(Eds.), Proceedings of the 26th International Conference for the Psychology of
Mathematics Education , Vol. I. (pp. 141-142). Norwich, UK: University of
East Anglia.
Fauconnier, G. & Turner, M. (2002). The way we think: Conceptual blending and the
mind's hidden complexities. New York: Basic Books.
Johnson, M. (1987). The body in the mind: The bodily basis of meaning, imagination,
and reason. Chicago: University of Chicago Press.
Lagrange, J-P., Artigue, M., Laborde, C. & Trouche L. (2001) A meta study on IC
technologies in education. Towards a multidimensional framework to tackle their
integration. Proceedings of the 25th conference for the Psychology of Mathematics
Education, van den Heuvel Panhuizen, M., 1, 111-123.
Lakoff, G. (1987). Women, fire and dangerous things: What categories reveal about
the mind. Chicago: University of Chicago Press.
Lakoff, G. & Nez, R. (2000). Where mathematics comes from: How the embodied
mind brings mathematics into being. New York: Basic Books.
Meserve, B. (1955). Fundamental concepts of geometry. New York: Dover.
Varela, F., Thompson, E., & Rosch, E. (1991). The embodied mind: Cognitive
science and human experience. Cambridge: MIT Press.

L. D. Edwards

10
10