The Brain's Sense of Movement - Alan Berthoz

PERSPECTIVES IN COGNITIVE NEUROSCIENCE
Stephen M. Kosslyn
general editor
THE BRAIN’S SENSE OF MOVEMENT
Alain Berthoz
Translated by Giselle Weiss
HARVARD UNIVERSITY PRESS

Cambridge, Massachusetts
London, England
Copyright © 2000 by the President and Fellows of Harvard College
all rights reserved
Printed in the United States of America
First published in France as Le Sens du Mouvement
Copyright © Editions Odile Jacob, 1997
Published with the assistance of the French Ministry of Culture–CNL
Pages 327–328 constitute an extension of the copyright page
Library of Congress Cataloging-in-Publication Data
Berthoz, A.
[Sens du Mouvement. English]
The brain’s sense of movement / Alain Berthoz ; translated by Giselle Weiss.
p. cm. — (Perspectives in cognitive neuroscience)
Includes bibliographical references and index.
ISBN 0-674-80109-1 (cloth)
ISBN 0-674-00980-0 (paper)
1. Motion perception (Vision) 2. Orientation (Physiology)
3. Proprioception. 4. Brain. 5. Neuropsychology.
I. Title. II. Series.
QP493 .B47 2000
612.8⬘2—dc21 00-023758
FOREWORD TO THE AMERICAN EDITION
This book was published in 1997. Science moves fast, and many new ideas and
observations have emerged in the intervening time. Yet the ideas proposed in
this book have, if anything, been reinforced by recent knowledge about the
brain. I did not think it useful to revise all the references, but I have included a
number of papers published by our laboratory or close collaborators since
1997. I have also included a number of reviews from 1998 and 1999 in which
the reader will find more general recent references.
ACKNOWLEDGMENTS
I would like to thank Odile Jacob, who agreed to publish the French edition of
this book at the suggestion of Jean-Pierre Changeux and who saw it through
production with the combination of keen intuition and professionalism that is
her trademark.
Mountain climbers need a guide. Writing a book is a little like climbing a
mountain. The guide knows the way to the top, avoiding loose rocks and haz-
ardous trails. He knows the moods of the mountain and its traditions. He
keeps everyone on course, makes detours when needed, sets the pace, and
transforms the climb step by step into a rewarding human adventure. My edi-
torial guide for this book was Gérard Jorland. The breadth of his knowledge as
a philosopher and historian, his generosity, his trust, and his critical comments
helped me, encouraged me, and touched me.
Maya, you accompanied this book. In form as well as in substance, you
helped me to make it not only readable but understandable.
The work of our laboratory would not have been possible without the
support of the Centre National de la Recherche Scientifique (CNRS), the Cen-
tre National d’Etudes Spatiales (CNES), and the Collège de France, which
managed simultaneously to have faith in us, to rigorously evaluate our work,
and to tolerate the gambles that are necessary for new ideas to emerge.
I would also like to thank the people who willingly read parts of the
manuscript, my friends and colleagues Pierre Buisseret, Valérie Cornilleau-
Pérès, Jacques Droulez, Jean-René Duhamel, Werner Graf, Alexej Grantyn,
Isabelle Israël, Joseph McIntyre, Edmund Rolls, Jean-Michel Roy, Jean-Jacques
Slotine, Yves Trotter, Pierre-Paul Vidal, Paolo Viviani, Sidney Wiener, as well
as our entire laboratory team, who carried out the experiments described
here.
I am grateful to the journal La Recherche for permission to reproduce a
portion of the text and illustrations that appeared in a special issue on the
brain in 1996.
I am also grateful to Solange Fanjat de Saint Font for editorial help and for
assembling the bibliography. Her expertise was critical in bringing this difficult
project to completion. And last but not least, I thank Frédéric Lacloche, whose
computer graphics skill I admire. He adapted the illustrations.
Finally, I am most grateful to Giselle Weiss who translated this book in
English with remarkable speed, exceptional insight into what I wanted to say,
and a wide and deep culture, which has earned my admiration. My gratitude
also extends to the team at Harvard University Press for the thorough profes-
sionalism with which they handled this project.
The translator is deeply indebted to Greer Ilene Gilman and Barbara Breasted
Whitesides for editorial and research assistance.
viii •
ACKNOWLEDGMENTS
CONTENTS
introduction 
 Perception Is Simulated Action 

The Motor Theory of Perception 
The Concept of Acceptor of the Results of Action 
Bernstein’s Comparator 
Memory Predicts the Consequences of Action 
Mental Nodes 
Mirror Neurons 
Simulation, Emulation, or Representation? 
 The Sense of Movement: A Sixth Sense? 

Proprioception 
The Vestibulary System: An Inertial Center? 
The Functions of the Vestibular System 
Seeing Movement 
 Building Coherence 
How Vision Detects Movement 
Visual Movement and Vestibular Receptors 
Am I in my Bed or Hanging from the Ceiling? 
The Coherence between Seeing and Hearing 
The Problem of the Coherence and Unity of Perception 
Autism: The Disintegration of Coherence? 
 Frames of Reference 
Personal Space and Extrapersonal Space 
Egocentric and Allocentric Frames of Reference 
Natural Frames of Reference 
Selecting Frames of Reference 
 A Memory for Predicting 

Topographic Memory or Topokinetic Memory? 
The Neural Basis of Spatial Memory: The Role of the Hippocampus 
 Natural Movement 

Pioneers 
The Problem of Number of Degrees of Freedom 
The Invention of the Eye 
The Form of a Drawing Is Produced by the Law of Maximal Smoothness 
 Synergies and Strategies 

Vestibular Axon Branching and Gaze Stabilization 
The Baby Fish that Wanted to Swim Flat on Its Stomach 
The Neural Bases for Encoding Movement of the Arms 
Coordination of Synergies 
 Capture 
The Toad’s Decision 
The Art of Braking 
What If Newton Had Wanted to Catch the Apple? 
 The Look that Investigates the World 

Gaze Orientation 
“Go Where I’m Looking,” not “Look Where I’m Going” 
Eye-to-Eye Contact 
Gaze and Emotion 
The Neural Basis of Gaze-Orienting Reactions 
 Visual Exploration 

The Brain Is a Fiery Steed 
A Model of Perception-Action Relationships 
x •
CONTENTS
Imagined Movement and Actual Movement 
Dynamic Memory and Predictive Control of Movements 
Was Piaget Right? 
 Balance 
A Physiology of Reaction 
How to Make the University of Edinburgh Oscillate 
Toward a Projective Physiology 
 Adaptation 
Adaptation and Substitution 
The Rheumatologist and the Ophthalmologist 
The Role of Activity in Compensating for and Preventing Disorientation 
 The Disoriented Brain: Illusions Are Solutions 

Illusion: The Best Possible Hypothesis 
Illusions Caused by Acceleration and Gravity 
Illusions of Movement of the Limbs 
Space and Motion Sickness 
A Few Other Illusions 
 Architects Have Forgotten the Pleasure of Movement 
Conclusion: Toward a Tolerant Perception 
Notes 269
Works Cited 
Credits 
Index 
CONTENTS •
xi
INTRODUCTION
In the beginning was the Deed.

—Faust
Irish storytellers always begin this way: “It was not in my time, not in your
time, not in anybody’s time.” The story I am about to tell also belongs to no
time because it relates to the greatest mystery of all time: the brain.
This book is a reflection on how the brain works, based on the idea that
the brain is used to predict the future, to anticipate the consequences of action
(its own or that of others), and to save time. To this end, very diverse biologi-
cal mechanisms developed over the course of evolution: the architecture of
the skeleton, the subtle properties of sensory receptors, and the marvelous
complexity of the central nervous system. These mechanisms endowed the
brain with internal models of the world and of the body—not just any mod-
els, but models that reflect the laws of nature, the Umwelt of each species, to
use Uexküll’s term, and that ensure the survival of each animal.1 The brain is
not a reactive machine; it is a proactive machine that investigates the world.
To become a ski champion, it is not enough for the skier to continuously pro-
cess sensory cues and correct his trajectory; he must go over the run in his
mind, anticipate its stages and the state of his sensory receptors, foresee possi-
ble solutions to every error, take chances and make decisions before he makes
a move.
Our ability to understand the brain and cognition is of course limited, but
it is not for lack of trying. First we must consider the question of how the
brain relates to the mind. At a recent meeting on the subject of cognition and
geometry that brought together mathematicians, physicists, physiologists, and
philosophers, one participant, a philosopher, suddenly lost his temper, ex-
claiming angrily: “I don’t even want to hear what physiologists have to say, be-
cause brains are just bloody things that I see at the local butcher shop!” This
outburst would be a suitable epigraph for an anthology on militant dualism. It
shows that we still have a long way to go to persuade even the most eminent
thinkers that cognition is a property arising from the amazing complexity of
the brain.
I think that we are partially responsible for prejudices like those the philos-
opher and mathematician Châtelet exposes when he writes:
Typically it seems as though the work of science is limited to maintain-
ing knowledge, ignoring the contributions of those who further that
knowledge and disseminate it. But these are the very people who rescue
science from the burden of endless accumulation and stratification, the
fatuity of established ideas, the convenience of expediency, and, finally,
the temptation to seek refuge in rules. They demonstrate how urgently
we need to construct a genuine theory of the way information is pro-
cessed and to learn about learning. Such reasoning is of course the an-
tithesis of a neural barbarism that is stubbornly dedicated to hunting
down the neuronal basis of thought and that confuses the activity of
learning with the fruits of learning, that is, knowledge. Schelling saw
this more clearly: he knew that, in any event, thought was not encapsu-
lated in a brain, that it could be anywhere . . . outside . . . in the morn-
ing dew.2
We may never succeed in combating these attitudes if we do not describe

the complexity of the brain. Now, complexity can be an excuse for not explain-
ing anything, and I know that this book may well appear very complex. But I
am convinced that the image often portrayed of the brain is too simplistic and
especially too static to allow a glimpse of its mechanisms. It is not enough to
say that there are millions of neurons. Anyone who believes in the soul will
conclude the need for a higher principle to organize this intricacy. The dy-
namic, flexible, and adaptable character of biological mechanisms must be
demonstrated. And physiology would appear to be the right discipline to do
this, because it brings together discoveries in anatomy and in cellular biology,
mathematical and physical models, and experiments in cognitive psychology
to frame its explanations.
The most essential properties of human thought and sensibility are dy-
namic processes—ever changing, ever adapting relationships among the brain,
2 •
the body, and the environment. “Panta chōrei, ouden menei,” said Heraclitus of
Ephesus. “Everything flows and nothing stays.” Thought and sensibility are
nothing more than states of cerebral activity induced by certain relationships
among the physical world, the body, the hormonal and neuronal brain, and its
memory of thousands of years of culture.
The anger and incredulity that this idea evokes are rooted in our often
oversimple vision of what the brain actually is. Like Leonardo da Vinci, we
once thought that it was composed of cavities, then, like Descartes, that it was
full of animal spirits. Today, it is known to be populated with little creatures
called neurons. But some are not convinced that these neurons are really the
basis of the subtle capacities that produce music and mathematics. At the
dawn of this century, when the heart turned out to be just a pump, people had
to make the best of it. Poets kept on singing about love nonetheless.
I must show the complexity, but I must explain it in simple terms. And
there is the challenge. Reader, we must meet each other halfway. I will try to
be simple, but you must seek complexity in what I write; and in the shortcuts I
take, accept that behind each tree hides a forest. The great misunderstanding
between physics and biology stems from the fact that the first is good at de-
scribing reality with simple formulas, whereas the second is wary of formulas
and aims instead to build a genuine theory of dynamic complexity.
Why, given this climate, write a book about movement? First, because I
think that the most refined cognitive abilities of the brain are a product of the
need to carry out many difficult tasks. The species that passed the test of natu-
ral selection are those that figured out how to save a few milliseconds in cap-
turing prey and anticipating the actions of predators, those whose brains were
able to simulate the elements of the environment and choose the best way
home, those able to memorize great quantities of information from past expe-
rience and use them in the heat of action. Relationships between perception
and action are the model of choice for studying the functions of the nervous
system. Unlike language, they lend themselves to analysis of human and ani-
mal behavior as well as to exploration of the neural mechanisms that underlie
them, across the multitude of species that evolution has produced.
In fact, the main problem posed by the command and control of move-
ments is that of inertia and the considerable forces that oppose these move-
ments in the water, in the air, and on land. In general we only have one chance
to survive—a single shot, as it were—but one that commits our muscles and
our bodily mass to movement. To catch prey that is moving at thirty-six kilo-
meters per hour, that is, ten meters per second, a predator must anticipate its
position in less than one hundred milliseconds and head for where the prey
INTRODUCTION •
3
will be in a moment’s time. It must also prepare the gesture of capture as well
as that needed by the muscles to compensate for the weight of the prey and
overcome its resistance. To anticipate the behavior of the prey, the predator
must make a guess, take a gamble; it must construct a theory of mind by pre-
dicting what dodges the prey might attempt in that context. These extremely
fast and essentially dynamic processes, and everything that goes on during
them, happen in a few dozen milliseconds. The brain is above all a biological
machine for moving quickly while anticipating.
In the chapters that follow I will return to all these aspects. They have
been neglected by researchers whose study of cognitive function gave priority
to language and reasoning. And yet, even poets, those guardians of the lan-
guage, have recognized the importance of action. Thus, in Goethe’s drama,
Faust reflects:
It is written: “In the beginning was the Word!”
Already I have to stop! Who’ll help me on?
It’s impossible to put such trust in the Word!
I must translate some other way
If I am truly enlightened by the spirit.
It is written: “In the beginning was the Thought!”
He hesitates again, then continues:

It should be: “In the beginning was the Power!”
Yet even as I write it down,
Already something warns me not to keep it.
The spirit helps me! All at once I see the answer
And write confidently: “In the beginning was the Deed!”3
Despite Faust’s warning that “in the beginning was the Deed,” many read only
the first line of Goethe’s poem and still believe that “in the beginning was the
word.” Through an understandable fascination, it was actually believed for
twenty years that understanding language, an attribute specific to humans,
would enable us to understand cognitive function. This belief, reinforced by
the influence of several groups in the eastern United States who have pro-
moted the idea that the brain is a computer, led to a symbolic and computa-
tional conceptualization of the nervous system.
The American functionalist school and some of its European adherents
defend a cognitive psychology that maintains in principle that the higher func-
tions of the brain must be studied without any reference whatsoever to their
neural underpinnings. These functions may be emergent or dissociated, but in
4 •
the end they are viewed merely as superstructure. We have probably not heard
the last of the dualist tendencies.
This risk was brilliantly exposed by Kant:
Misled . . . by such a proof of the power of reason, the demand for the
extension of knowledge recognises no limits. The light dove, cleaving
the air in her free flight, and feeling its resistance, might imagine that its
flight would be still easier in empty space. It was thus that Plato left the
world of the senses, as setting too narrow limits to . . . the understand-
ing, and ventured out beyond it on the wings of the ideas, in the empty
space of the pure understanding. He did not observe that with all his ef-
forts he made no advance—meeting no resistance that might, as it
were, serve as a support upon which he could take a stand, to which
he could apply his powers, and so set his understanding in motion. It
is, indeed, the common fate of human reason to complete its specula-
tive structures as speedily as may be, and only afterwards to enquire
whether the foundations are reliable.4
Plato forgot the body. This book is an apology for the body.
I will proceed in the following way. First I will lead you on a walk through
the land of the senses. In addition to the receptors that make up the usual five
senses—vision, smell, hearing, touch, and taste—we must identify several oth-
ers, those of the muscles, joints, and inner ear. In fact, we do not have only
five senses, but eight or nine. Is there any point in listing them?
Not at all, because the brain does not process sensory cues independently.
Each time it commits to an action, it makes assumptions about the state of
certain receptors as the action unfolds. The ski champion cannot constantly be
checking the state of all his sensory receptors; he mentally simulates the
course of his run down the slope, and it is only from time to time, intermit-
tently, that his brain checks to see whether the state of certain sensory recep-
tors is in accordance with its prediction of the angle of the knees, the distance
from the ski poles, and so on. These groupings of receptors are called configu-
rations, and it appears that the brain checks configurations of specific recep-
tors as it plans movement.
With that in mind, I will define all the sensory receptors that enable us to
analyze movement in space. These receptors are collectively responsible for
what is called the sense of movement, or kinesthesia. Kinesthesia is the result
of cooperation among several sensors, and it requires the brain to coherently
reconstruct movement in the body and in the environment. When this coher-
ence cannot be achieved, perceptual and motor disturbances result, as well as
INTRODUCTION •
5
illusions, which are actually solutions the brain devises to deal with discrepan-
cies between sensory information and its internal preperceptions.
Next I will examine how the brain uses memory to predict the conse-
quences of action. An especially interesting example is that of episodic mem-
ory and working memory. Various mechanisms enable the brain to preserve
the traces of recent events, to combine motor and sensory signals, and to rep-
resent the required process for accomplishing a move or attaining a goal.
Having completed this ramble through the land of the senses and the
mechanisms of perception, I will ask the reader to make a special effort to
consider a fundamental aspect of the relationships between perception and ac-
tion: the mechanical properties of bodily masses. Indeed, it is impossible to
understand anything about how the brain works without conceding that its
main problem is to put mass into motion. Instead of saying “mass,” one might
say “inertial moment,” or the considerable, complex forces that arise as soon
as a mass is in motion, like the Coriolis forces created by three-dimensional
movements of the head, the resultant of angular accelerations in several
planes. The Russian physiologist Bernstein drew attention to the fact that in
animals and in man, the limbs have enormous “degrees of freedom,” and con-
sequently nature has had to find tricks to simplify the work of the brain. Study
of the geometry of movements helps to grasp the natural solutions devised by
the nervous system to solve this problem. Skeletal anatomy can be explained
in this context; moreover, prewiring motor synergies is another way of simpli-
fying neurocomputation, and simple kinematic relationships connect the ge-
ometry of movement and dynamics.
After examining these aspects of perception and movement, I will offer a
few concrete examples of motor organization in connection with locomotion,
gaze orientation, and control of balance. I will consider how neural mecha-
nisms and internal models enable prediction, how preselection of sensory
messages is carried out, and especially what the fundamental role of synaptic
inhibition and simultaneous parallel and hierarchical processing is. I will exam-
ine how the same structures can be activated in both executed and imagined
movement and, in the case of damage or of sensory conflict, how the brain in-
vents new solutions to restore functional plasticity.
In short, I present a theory one step ahead of those that consider the brain
to be a simple proactive or representational organ. I suggest that the brain is a
biological simulator that predicts by drawing on memory and making assump-
tions. Flight simulators neither predict nor invent. The brain has a need to cre-
ate; it is an inventive simulator that forecasts future events. It also acts to emu-
late reality.
6 •
I finish with a tirade against architects who have abandoned the brain’s in-
stincts and preferences in favor of their new idols: uniformity, profitability, and
the tyranny of the right angle. They have forsaken all sense of finesse for a
geometric approach of unparalleled mediocrity.
My scientific tool will be physiology. It is necessarily multidisciplinary, be-
cause it involves both structure and function. In the compelling words of
Bergson, whose seductive dualism will be challenged several times in these
pages,
We shall find that wherever it is a question of this idea[,] Claude Ber-

nard attacks those who refuse to see in physiology a special science dis-
tinct from physics and chemistry. The qualities, or rather the disposi-
tions of the mind, which make the physiologist are not the same,
according to him, as those which make the chemist and physicist. He is
not a physiologist who has not the organizing sense, that is to say, the
sense of that special coordination of the parts to the whole characteris-
tic of the vital phenomenon . . . Claude Bernard defends physiology
both against those who believe the physiological fact to be too elusive
to lend itself to experimentation and against those who, while judging
it to be accessible to our experiments, would not distinguish these ex-
periments from those of physics or chemistry. To the first group he an-
swers that the physiological fact is governed by an absolute determin-
ism and that physiology is consequently a rigorous science; to the
second, that physiology has its proper laws and proper methods, distinct
from those of physics and chemistry, and that physiology is in conse-
quence an independent science.5
A final word: I will not talk much about emotion in this book, except in
the chapter where I briefly examine the problem of decision making in the
context of a theory of Damasio. And yet, there is no perception of space or
movement, no vertigo or loss of balance, no caress given or received, no
sound heard or uttered, no gesture of capture or grasping that is not accompa-
nied by emotion or induced by it. But it is necessary to construct a physiology
of relationships between movement and emotion, as Ribot suggested:
What are called agreeable or painful states only constitute the super-
ficial part of the life of feeling, of which the deep element consists in
tendencies, appetites, needs, desires, translated into movements. Most
classical treatises (and even some others) say that sensibility is the fac-
ulty of experiencing pleasure and pain. I should say, using the same ter-
INTRODUCTION •
7
minology, that sensibility is the faculty of tending or desiring, and conse-
quently of experiencing pleasure and pain. There is nothing mysterious
in the tendency; it is a movement or an arrest of movement in the na-
scent stage. I employ this word “tendency” as synonymous with needs,
appetites, instincts, inclinations, desires; it is the generic term of which
the others are varieties; it has the advantage over them of embracing at
the same time both the psychological and physiological aspects of the
phenomenon.6
This book blends analytical considerations of the components of percep-

tion and action with synthetic concepts borrowed from experimental and cog-
nitive psychology. One of the major developments of the last ten years has
been the reconciliation of psychology and the neurosciences, which made it
possible to revive study of cerebral function according to complex paradigms.7
Despite the rear guard battle being waged by some to keep the soul and the
body apart, a fruitful, irreversible collaboration has taken root among psychia-
trists, neuropsychologists, philosophers, psychologists, and neurobiologists. I
hope to add my modest contribution to it. Pardon the philosophical quota-
tions, for they are intended neither as passing references nor as pedantry. We
all make implicit assumptions, and these citations are one way of clarifying
them—perhaps inappropriate, but it gives me pleasure. And is not pleasure the
stimulus for all knowledge?
8 •
1
P E R C E P T I O N I S S I M U L AT E D A C T I O N
Our sensations are purely passive, while all our perceptions or ideas are born
out of an active principle which judges.
—J.-J. Rousseau
The Motor Theory of Perception
A major theme of this book is that perception is more than just the interpreta-
tion of sensory messages. Perception is constrained by action; it is an internal
simulation of action. It is judgment and decision making, and it is anticipation
of the consequences of action. This is not an entirely new concept, as a
glimpse at some of the ideas that prefigured it will show. There are so many
theories of perception that I will not try to cover them all here. I wish simply
to provide a taste of what is in store.
A history of the motor theory of perception was recently published by
Viviani, who recalls that it was very fashionable before 1940.1 One of the first
modern works to signal the importance of movement in perception was that
of Lotze in 1852.2 His theory affirmed that spatial organization of visual sensa-
tions results from their integration with a muscular sense. The idea that the
information that triggers a motor command is used by the brain to recognize
movement was proposed by Helmholtz.3 For him, motor control operates by
comparing sensations with predictions based on the motor command. In 1890,
William James also described a neural circuit that anticipates the sensory con-
sequences of movement (Figure 1.1).4 A simple way of conceptualizing the
perception of movement, he proposed, would be to assume that a sensory cell
Motor neuron
M
S
K
“idea of movement”
Figure 1.1. William James’s concept of an anticipatory neuronal pathway.
S is excited and activates a motor neuron M that induces a muscle contraction.

A kinesthetic cell K detects the movement and modifies the motor neuron M.
James imagined an additional circuit: a collateral axon of cell S activates the
kinesthetic cell K at the same time that the motor neuron M is activated. K is
thus activated even before it receives information about the muscle move-
ment, which allows it to anticipate the consequences of movement.
In France, the ideas of Janet were very close to those of the pioneers of
the motor theory of perception. He established a hierarchy of actions. The
most elementary are reflex actions, followed by actions that are “perceptual,
social, simple intellectual; and at the level of verbal and volitional action,
thoughtful, rational, experimental, progressive.”5 He wrote: “Reflex action is
impulsive action as opposed to the considered action that characterizes per-
ceptual behaviors.”6 In other words, perception is interrupted action, and espe-
cially goal-oriented action. “In all these perceptual actions, the starting point is
a complex object, like prey or its burrow, and the action leads to a transforma-
tion and use of the object: the action adapts to the object, rather than simply
following from superficial stimuli.”7 But Janet went further and suggested that
perceptual action was predictive: “The action that is triggered by the initial
stimulus adapts not only to this stimulus but to all the others that the object
induces successively; it adapts to stimuli that do not yet exist, but that will only
occur later owing to the action itself. This adaptation to an aggregate of fu-
ture and potential stimuli is characteristic of perceptual behaviors.”8
Thus the dissociation between perception and action must be discarded.
Perception is simulated action. A further example, also borrowed from Janet, il-
lustrates this point in a way that is very close to my own interpretation:
“When we perceive an object, an armchair, for example, we do not see our-
selves as acting at that instant, because we are still standing, unmoving, per-
ceiving the armchair. This is an illusion. In reality, we already have within
10 •
us the action we associate with the armchair, which we call a perceptual
schema—here, the action of sitting in a particular way in the armchair.”9
Merleau-Ponty put it very well: “Vision is the brain’s way of touching.”10
Research on the physiological bases of the relationship between percep-
tion and action nevertheless remained limited until just recently. Viviani at-
tributes the post-1940 eclipse of the motor theory of perception to the ap-
pearance of the analytical neurophysiology of Sherrington, the influence of
Gestalt, and even to the constructivism of Piaget. The compartmentalization
of disciplines whose efforts should converge, such as biomechanics, experi-
mental and cognitive psychology, psychophysics, functional neurobiology, and
so on, were for a long time another obstacle that today the cognitive sciences
are attempting to overcome. After 1950, the motor theory of perception
was revived. For example, consider the work of Lashley, Gibson, the Teuber
school, and more particularly the work of Held and Hein on the role of action
(not just activity) in the development of the visual system. Or consider the
work of the psychologist Johansson in Scandinavia and that of Fessard and
Piéron in France—Piéron maintaining that perception is awareness of external
objects and events that give rise to sensations. In the 1970s the research groups
of Imbert and Jeannerod made major discoveries concerning the development
and functioning of the visual system and its relationship to the vestibular sys-
tem, the control of ocular movements and posture.
The Concept of Acceptor of the Results of Action
Though underrated in the West, the Russian Anokhin’s seminal ideas made
him a trailblazer.11 In his day, Pavlov’s theories and experiments on condi-
tioned reflexes, which profoundly influenced our century, were predominant.
But Anokhin found Pavlov’s definition of reflexes too limiting. He also criti-
cized Descartes for having sidestepped the question of the relevance of the
reflex response to the higher brain and thereby influenced future studies of
the complex adaptive actions of animals and man for years to come. Anokhin
introduced a key idea to the theory; that at the end of the reflex arc there is a
reflex action. Although Pavlov’s idea was already old, Western physiologists
ignored subsequent contributions of Russian physiologists for fifty years, ex-
cept to plagiarize their ideas without citing them—intellectual petty thievery
that the isolation of the Russians helped to disguise. Western physiologists
used the word “response” to refer to the effect of a stimulus in a reflex arc,
whereas the Soviet literature emphasized and still emphasizes the concept of
PERCEPTION IS SIMULATED ACTION •

11
reflex action. The crucial difference is that if every reflex requires a complete
action, this involves the entire organism and consequently its faculties of in-
vention, creativity, and adaptation.
Anokhin provided no experimental proof, but he constructed a theory he
called “acceptor of the results of action” that is worthy of comment. He be-
gan with the following line of questioning: If the result of reflex activity is an
action, doesn’t execution of this action need to be approved one way or an-
other by some configuration of sensory afferent information? And doesn’t exe-
cution of the action also need to be compared with a predicted configuration?
Say that we wish to pick up a cup from a table cluttered with dishes but
that, just when we are about to grasp the cup, we are distracted and pick up a
pitcher in its place. As we all know from personal experience, in general we
correct an error like this one immediately. But what physiological mechanism
permits us to notice our error and to correct it? The appearance of the pitcher
and the gripping of its handle as well as the appearance of the cup and the
gripping of its handle are just an aggregate of afferent signals that differ by a
few constituents. Why is it, then, that the latter afferentation is precisely the
one we choose to sanction our action?
According to Anokhin, our move with the pitcher was initially satisfactory
because the set of sensory signals constituting its grasp matched a configura-
tion that was predicted, expected, specified before making the gesture. Mathe-
maticians would say that the sets of afferents corresponding to the cup and
the pitcher contain an intersection sufficient to be allowed by Anokhin’s accep-
tor of action. He states that this ready-made excitatory complex, which pre-
cedes the reflex action, must in some way be an afferent control apparatus that
determines how closely the return afferentation from the central nervous sys-
tem corresponds to it.
Thus, Anokhin developed a concept equivalent to what is now called an
internal model of a group of preselected elements. He was careful not to use
the word “representation.” He referred to Pavlov’s earlier observation that
the chemical composition of saliva from a conditioned dog is related precisely
to the kind of food used for reinforcement and thus to the character of sali-
vation.
Anokhin next turned to the neural basis of this concept. He used the ex-
pression “acceptor of the results of action” to refer to a cortical system spe-
cialized in the analysis of complex afferents (sensory information) resulting
from reflex action. This analyzer determines how the afferent inputs it re-
ceives relate to the planned action as a function of the past experience of the
12 •
animal. Noting that he could as well have called his apparatus “the acceptor of
the afferent results of a completed reflex action,” Anokhin explained that he
had chosen the word “acceptor,” from the Latin acceptare, because it combined
two ideas: accept and approve. Consequently, he incorporated a clear role for
the acceptor of the result of action in decision processes. For example, if a
person sitting in the living room decides for one reason or another to go into
the dining room, at the precise moment of the decision, the set of afferents
from all the stimulatory cues he has received in the dining room in the past
(acceptor of action) is reproduced in his cerebral cortex. If, after the person
has entered the dining room, the cues coincide perfectly with what the accep-
tor of the result of action predicts, the person then moves on to the next ele-
ment of intended behavior. But if the acceptor of the result of action detects
an error, an incongruity with what is predicted, the brain produces an orienta-
tion reaction, as described in the Soviet literature—that is, it reacts by analyz-
ing new events.
Bernstein’s Comparator
Another master of modern physiology who had an even more profound influ-
ence on our generation was the Russian physiologist Bernstein. Persecuted for
his overly original ideas in the time of Pavlovian hegemony, he studied natural
movement and inferred general rules of cerebral functioning from it.12 To
avoid a naïve linear description of the regulation of the coordination of move-
ment as a succession of phases—prediction, preparation, execution, and con-
trol—Bernstein proposed a circular schema (Figure 1.2) that introduced the
concept of the action-perception cycle. The basic element is a “comparator”
that establishes the “required value.”
This required value fulfils at least three different functions, all equally im-
portant. The first is detecting an error between accomplished movement and
predicted movement that triggers a correction (from a cybernetic vantage).
The second is recognizing that an action has been accomplished, which allows
progression to the next action in sequence. “This aspect of function,” says
Bernstein, “is mainly reminiscent of what Anokhin has termed ‘sanctional
afferentation.’” The third function is the adaptation itself. Indeed, when action
encounters surprises, it is impossible, or irrelevant, for corrective impulses to
reestablish the initial plan of action. In this case, the receptor of information
acts as an initiator (and not as a regulator) of adaptive changes in the program
being executed. It does this by introducing either small technical changes into

13
5
ENCODER
CORRECTIONS ∆w
PROGRAM
6 4
REGULATOR Sw COMPARATOR
2 “Internally controlled element”

PROGRAMMER
Iw
ENERGY
Iw
1
3
EFFECTOR
RECEPTOR
(muscle)
Sensory
receptor
INFORMATION
Locus of
activity
Figure 1.2. Schematic diagram showing the cerebral organization for control of
movement proposed by Bernstein.
14 •
the movement or taking an adjacent trajectory. And it does it until the pro-
gram has been completely reorganized, even down to the repertoire of con-
secutive elements and the staging of motor action—in other words, it adopts a
new tactical approach to the task.
I think that the highest cognitive functions are the result of an evolution-
ary thrust toward developing this ability to reorganize action according to un-
foreseen events. This ability requires developing a memory of the past, the
faculty for predicting and simulating the future, and the metafaculty, in a way,
to mobilize all these capabilities rapidly, because they must integrate with a
perception-action cycle that sometimes lasts only a tenth or twentieth of a
second.
For Bernstein, these corrective processes depend heavily on what he calls
the comparator. This neural device occupies a strategic position between the
information supplied by the receptors and the elements that will effect the
necessary corrections or reorganization. It does not function between two
successive or simultaneous receptors to compare two distinct events, but be-
tween flowing, continuous reception and an internal guide.
An important property of the comparator is its capacity to detect varia-
tions in sensory information owing to the central nervous system’s use of
fresh traces. Our bodies, says Bernstein, have no receptive apparatus capable
of perceiving velocity directly. This task is resolved in the central nervous sys-
tem by the comparator. It instantaneously compares the cues about the posi-
tion of the moving organ with the fresh trace of its position approximately 0.1
second before. The brain thus recognizes two positions with a certain interval
of time between them. So it can easily reconstruct the velocity, which is dis-
placement (the difference from one position to the other) divided by time.
Bernstein saw that the control of movement is not continuous but dis-
crete, under the control of the comparator. He arrived at this conclusion from
contemporary work that linked motor action with brain waves having a fre-
quency of 8 to 14 cycles per second—that is, α waves measured by electro-
encephalogram.13 This is, at least in part, the frequency of rhythmic oscilla-
tions of excitability for the principal elements of the reflex circuitry of our
motor apparatus. Bernstein regarded the intervals between each cycle at the α
rhythm as units of an internal physiological clock, or pacemaker.
Bernstein’s vision was prophetic, since it was not until 1990 that Llinás
suggested that movement is subtended by neural activity oscillating at 10 Hz
(ten oscillations per second).14 This suggestion remains viable, as it is possible
to observe synchronizations of 10 Hz in the neck muscles of an alert cat.
Movement may therefore be subtended by coupled oscillators that function at
15
this frequency, as Bernstein foresaw. The work of Sokolov and his school in
Russia on the hippocampus,15 that of Rougeul-Buser in Paris, and of olfaction
and sleep specialists has also demonstrated the importance of these central
rhythms during various tasks that require alertness or organized movements.16
It seems clear now that several oscillatory processes with frequencies whose
values are 8 to 12 Hz, 16 Hz, 40 Hz, 70 to 90 Hz, and so on subtend the inter-
nal workings of perception and movement. Llinás summed up his ideas by
saying that “we think at 40 Hz and move at 10 Hz,” asserting laconically that
the minimum time to process mental data is of the order of 25 milliseconds,
and the minimum time for a motor control operation around 100 millisec-
onds. He proposed a theory that explains how the brain uses internal circuits
that oscillate forty times per second (40 Hz) to develop multisensory percep-
tion and guarantee its coherence (Figure 1.3).
Bernstein completes his analysis of the microfunction of the perception-
action cycle with more general considerations of the role of anticipation. He
observes that in fact a wide range of movements anticipate or extrapolate by
analyzing fresh traces of the past. I will offer many examples of these neural
memories, within the cell itself or in complex networks, which maintain sen-
sory elements for a certain time, the new idea being that they allow prediction
of the future. If I am not mistaken, Husserl’s concept of simultaneity of
“protention” and “retention” is quite close to this idea.
A final basic conceptual contribution by Bernstein concerns the very prin-
ciple of motor control. Suppose that I wish to move my arm from one posi-
tion to another: the motor command an engineer might use would employ a
potentiometer to measure the angle, then construct a movement command
that controls the motors, which in this case are the muscles. A force com-
mand, or rather a coupling command (force multiplied by the lever arm),
would be the controlling variable, and the resulting displacement the variable de-
tected by the muscle receptors that are the biological equivalent of the potenti-
ometer.
Bernstein reasoned that the brain controls another variable, the point of
equilibrium between the two muscles that are in opposition around a joint—
the biceps and triceps in the arm, for example. He started from the idea that
when the forces exerted by these two muscles—which depend on the activa-
tion of motor neurons and their mechanical properties—are equal, the arm
will be in a given position. To define a position in space or to maintain a cer-
tain relationship among muscular forces is equivalent. Thus, space is perhaps
not coded explicitly, but the trajectory of movement results from this dynamic
equilibrium. Movement is simply a progressive shifting of postures.
16 •
A B
Layer I
+ Bursting
+ 40Hz
+
CORTEX IV
+
−
Layer V −
+
Interneuron
VI 40Hz +
P
Reticular + +
nucleus + +
THALAMUS − −
+ +
Specific nuclei Intralaminar
nonspecific nuclei
Figure 1.3. Thalamo-cortical circuits accounting for perceptual coherence

through synchronization of neuronal oscillators. (A): On the left, sensory in-
formation received from vision, proprioception, and so on converge on specific
thalamic nuclei. The neurons in these nuclei project to the cerebral cortex (to
the pyramidal neurons) (P) in layer VI. Activated by various interneurons that
oscillate at 40 Hz, the pyramidal neurons of the cortex project in turn (two
projections, one inhibitory, one excitatory) to the thalamus and thus constitute
a circuit swept by activity that oscillates at 40 Hz. To the right, a second circuit
comprises the intralaminar nonspecific thalamic nuclei that project to the
more superficial layers of the cortex and toward the reticular nuclei. The py-
ramidal neurons activated at 40 Hz also project back to the thalamus. (B):
Sketch showing the neurons of the intralaminar nucleus of the thalamus,
which projects in a starlike formation toward the cortex. Note that these neu-
rons are arranged in a circular ring. The oscillation at 40 Hz is reproduced in
the intralaminar nucleus along this ring. Consequently, the neurons of the ce-
rebral cortex are activated sequentially. The other nuclei of the thalamus are
indicated by two shaded regions.
Memory Predicts the Consequences of Action
A concept proposed a long time ago to link perception, action, and memory is
that of the motor schema. According to Schmidt, movement structures, which
he calls “schemas,” are stored in the brain.17 These schemas are not sensory or
motor elements, but memorized relationships—topological links, mathemati-
cians would say—between several sensory or motor components of action
(like the position of limbs, the state of a target in space, and so on). Schmidt
17
Initial conditions
Desired outcome
Recognition
Past
schema
schema
Recall
Past Past sensory

response
actual consequences
specifications
outcomes
(Controls (Estimates
rapid consequences)
movements)
Expected
Response
sensory
specifications
consequences
Figure 1.4. The brain anticipates the consequences of movement from past ac-
tion to prepare and initiate a movement. Reading from top to bottom, antici-
pated movements begin with a set of initial conditions and a plan of action that
will lead to the desired outputs. These are compared with the results of past ac-
tions recalled by two types of memory: left, memory of motor commands, and
right, memory of sensory data associated with past movements and their ef-
fects on the environment. It is therefore possible to recall the expected sensory
consequences, or the messages detected by the receptors during and after ac-
tion over the course of movement. Control of movement involves estimating
discrepancies between anticipated and actual sensory data.
proposes the following definition of a schema (Figure 1.4). When a person

executes a move to reach a goal, he stores up four types of elements: the ini-
tial conditions, provided by the senses; the program for the motor com-
mand, which Schmidt calls “specification of past responses”; the sensory con-
sequences of the movement; and other consequences of the movement, like
outcome. The schema is not the set of these data, but their relationships.
What is interesting about this theoretical work is that it links prediction of
the consequences of action with the memory of past consequences, with-
out reducing movement either to a simple chain of packaged reflexes or
to the simple execution of a centralized motor program. Schmidt’s model
18 •
also shows how relationships between the elements of the schema can be
modified.
This concept turns up again in the work of Neisser, who insisted on the
inseparable character of what he calls the “perception-action cycle.”18 In this
cycle, exploration of the visual world by a subject is directed by defined antici-
patory schemas, like blueprints, for perceptual action—perception is thus con-
ceived as active.
Mental Nodes
The idea that the brain does not simply detect the physical parameters that ex-
cite the senses is a very old idea. One interesting (unfortunately not very well
known) expression of it was offered by MacKay in several theoretical essays
dating from 1970 to 1980. His theory, called “mental nodes,” aimed at integrat-
ing interactions between perception and action. Others before him had tried
this approach: “Not all theoretical thinking in psychology has adopted the as-
sumption that components for perception and action are completely separate
or unshared. In particular, Lashley proposed that speech comprehension and
production make use of common components and mechanisms because ‘the
processes of comprehension and production of speech have too much in com-
mon to depend on wholly different mechanisms.’ . . . I conclude that percep-
tion and production share some components but not others.”19 But, as he ob-
served at the time, the dominance of research on perception isolated from
action derived as well from the philosophical view of researchers that “ac-
tion [was] functionally, temporally, and evaluatively subordinate to perception;
functionally subordinate because they considered perception the sole means
by which knowledge is acquired (empiricism); temporally subordinate because
they considered perception a necessary precursor to action (paleobehavior-
ism); and evaluatively subordinate because they viewed the contemplative life
as superior to a life of action (see Plato).”20
According to MacKay, two major mechanisms underlie most mental oper-
ations in sensorimotor systems. The first processes information from the sen-
sory receptors that the system forwards to a comparator where it is subjected
to criteria, such as a desired position. Error signals for checking corrections
are sent from the comparator—Bernstein’s classical loop—but MacKay adds a
second circuit based on sensory information. Sensory data in this second cir-
cuit are not processed by the nervous system to extract velocity, force, vol-
ume, dilation, and so on; they are interpreted as attributes, that is, configura-
tions of pertinent information that have a category-specific meaning. These
19
data are directed to an operator that organizes or decides or chooses or finally
executes a projective feedforward under the control of the supervisor. This op-
erator recognizes pertinent cues for the task and evaluates the sensory context
to anticipate the movement; it controls motor activity.
What is missing in this model is direct action by the organizer on the re-
ceptors; that is, an influence of active perception on sensation, although exten-
sive proof exists for such action as I have indicated here, and as I will show
again later. In particular, what is missing is a conception of the brain as a rep-
ertoire of sensorimotor schemas that are just so many possible actions and
that organize perception even before sensory stimuli are processed. I will de-
velop this idea further, but it is worth examining a clear example at the outset.
Mirror Neurons
Support for the idea that the brain contains schemas in its neuronal organiza-
tion that constitute veritable behavioral actions was recently provided by
Rizzolatti.21 He discovered neurons in area F5 of monkeys, later termed “mir-
ror neurons,” which fire each time the animal makes a particular gesture, for
example, lifting a peanut to its mouth, turning a handle, and so on. But the
same neurons also fire when the monkey sees the experimenter make that
same gesture. In other words, these neurons are activated both when the ani-
mal makes the gesture and when it sees the gesture being made. From that it
can readily be deduced that the network of neurons, which includes this par-
ticular neuron, codes for a schema of the behavioral repertoire of the monkey.
Do we make that deduction? I will discuss it; but whatever the outcome of the
debate, this discovery argues for the existence of a repertoire of preperceptions
linked to a repertoire of actions. Consequently, the brain can simulate actions
to predict their consequences and to choose the most appropriate among
them.
A word in passing about the term “encoding.” I will use this term to desig-
nate the message contained in the firing of neurons whose properties I will de-
scribe. This usage is defensible even though it makes a major assumption,
which is that the temporal or frequency characteristics of the discharge contain
signals relevant to the functioning of the nervous system. The frequency of
neural discharge is not the only code possible. The coincidence between the
firing of several neurons or the temporal organization of their firing, the aver-
age activity of a population of neurons, synaptic modifications at the molecu-
lar level, are just so many different codes that we still do not understand. Sub-
stantial research efforts are ongoing to understand the respective contributions
20 •
of temporal encoding and frequency encoding. I can only touch on this ques-
tion, though it is fundamental.
Simulation, Emulation, or Representation?
A final word about the philosophical framework of this book. I will rarely use
the word “representation,” and when I do, I will use it reluctantly. Indeed, I
think that it is very trendy and very convenient—for hiding our ignorance. Un-
der the guise of seemingly centralist theories, it very often conceals subtle
forms of dualism that I reject. In a recent review of theories of perception, in
particular that of Helmholtz, Bouveresse articulated the dangers inherent in
the concept of representation:
Theories that describe perception as the construction of internal repre-

sentations of the external world commit themselves at the outset to a
path that may perhaps lead somewhere, but not where they claim to go.
If we assume that our concept of mental representation is inspired by
the idea of a material image for which the object itself is lacking, and
that this image we have is distinct from the object itself, it is unlikely
that we can thereby achieve a satisfactory theory of how objects are
perceived.22
Bouveresse suspects Helmholtz of dualist tendencies:
If the transformation of sensory excitation into psychic representations

of objects in the external world is the result of neurological processes,
which remain entirely below the threshold of consciousness, it seems
there are good reasons for thinking that in the final analysis it is no less
physiological than such activity that results in the production of sensa-
tions. The risk in dividing things up the way Helmholtz does is that of
returning to an implicit form of dualism that accounts for both the raw
data coming from stimulation of our sensory receptors by external ob-
jects and the activity that results in the nerves, like a sort of material,
which the psychic apparatus (in plain language, the soul) must then pro-
cess and transform according to its own particular principles or meth-
ods. It is probably this risk that Mach denounces when he makes the
point that “everything psychical is physically based (fundiert), and not
just one part of the psyche. Consequently, explanation solely by means
of judgment and inferences is unacceptable. The apparatus for hearing
does not imply that the soul proceeds independently, apart from hear-

21
ing. Cerebral phenomena are in effect themselves the psychic life.” And
Mach concludes that “the eye must not make a detour through the in-
tellect.” At any rate, it must not do so if the word “intellect” refers here
to things that are supposed to occur in the soul, in the traditional sense
of the word, rather than in the human body.23
In choosing the word “simulation,” can I avoid these traps? I am certainly

not happy with it, but I prefer it to “representation.” For me, the brain is a
simulator in the sense of flight simulator and not in the sense of computer
simulation. Simulation means the whole of an action being orchestrated in the
brain by internal models of physical reality that are not mathematical opera-
tors but real neurons whose properties of form, resistance, oscillation, and
amplification are part of the physical world, in tune with the external world. A
few years ago, we constructed the scheme in Figure 1.5, suggesting that the
brain processes movement according to two modes. One, conservative, func-
tions continuously like a servo system; the other, projective, simulates move-
ment by predicting its consequences and choosing the best strategy.
When I talk about simulation, I mean the set of operations carried out by
a simulator; that is, a real machine that possesses at least some portion of the
properties of physical reality. As recently as twenty years ago, Gurfinkel had
the same idea. He rejected computer simulation of all the physical constraints
on walking as impossible; a model, however limited, “incarnated” or rather
embodied in a robot would do it better. So he constructed a spider with the
help of the department of robotics at the University of Moscow. It is in this
sense that the word “simulation” is better. The word “representation” is too
contaminated by the idea of visual image. “Now that I perceive the thing itself
and not a representation, I would only add that this thing is an extension of
my gaze and, in general, of my exploring,” wrote Merleau-Ponty.24
The development of models of neural networks out of silicone compo-
nents called VLSI (Very Large-Scale Integration)25 relates to the same idea, its
technological objective notwithstanding: biological models must be embodied
in physical reality just as the brain is incarnated in the body. This does not nec-
essarily lead to physicalist theories of the nervous system, or at least this drift
should be avoided. The notion of simulation is important, to avoid too nar-
rowly identifying the brain with a computer that calculates. It also must be
possible to distinguish, as Jeannerod proposes, the semantic aspect from the
pragmatic aspect of the control of movement.26 If the neurons of the colli-
culus allow a cat to catch a mouse by anticipating its future position, it is be-
cause the neurons are sensitive to the velocity of movement: they do not calcu-
22 •
Output Operation based Input
switch on internal maps switch
Projective
process
Motor
command Sensory
input
Operators on Discrepancy Conservative

Correction
state variables between actual process
vector state and
desired state
Figure 1.5. The brain activates two parallel modes over the course of an action.
The first, which is the oldest, is conservative. It tends to maintain certain vari-
ables within limits defined by the intended action. It functions in closed circuits
that couple the sensory receptors to motor commands either via reflexes or
through central structures that specify preconfigured motor programs in a rep-
ertoire of interconnected synergies. It also uses motor primitives—gestures
with recognizably distinctive features organized in specialized groups of neu-
rons in the brain (the basal ganglia, the motor and premotor cortex, the cere-
bellum, and so on). In this mode, the brain functions like a controller to correct
for mechanical variables such as velocity and force. The second mode of con-
trol evolved more recently. It is projective, using internal maps to simulate
movement without carrying it out. These maps are not the same as geographic
maps. Although some (like those of the colliculus) are genuine environmental
maps, the concept of map here means that the groups of neurons in a particu-
lar region of the brain have a structured set of properties analogous to the ex-
ternal world. If the input and output switches (that is, the sensory information
and motor activity) cease to operate, this mode can still function as a simulator,
as in a dream.
late a velocity. This property allows the cat to survive without having to do
calculations. Roboticists also began by making robots that simulated by calcu-
lating, but these robots are slow. The new generation of robots is endowed
with properties of adaptation and prediction. These robots use synthetic work
parameters that simplify calculations and that permit internal simulation of
processes for finding new solutions. I will come back to these ideas in Chap-
ter 8.
23
The fact that perception and simulation of action are embedded in each
other has been demonstrated recently in an experiment concerning mental ro-
tation. Imagine that you wish to compare two identical or different but similar
objects that are tilted with respect to each other by a certain angle. If you are
asked to decide whether they are different or identical, it will take you a cer-
tain time to answer (which can extend to seconds). The value of this mental
processing time is proportional to the angle between the two objects. This fact
has led psychologists to suggest that the brain performs a mental rotation of
the objects to compare them more easily. We have shown that if, during the
mental time of comparison, the subject is asked to perform a movement with
his hand in either the same or a different direction than the mental rotation, a
change in the time to respond occurs.27
Now that I have provided a general idea of the field, why not tackle the
first problem that arises: How do we get from a physiology of sensation to a
physiology of perception that acts, judges, and decides on the basis of action
and of the future consequences of action?
24 •
2
T H E S E N S E O F M O V E M E N T: A S I X T H S E N S E ?
Praised be our senses. Magically, they make manifest for us various vibrations
as light, color, and heat—not just sound; they capture the forces of chemical at-
traction as flavors or odors. In short, for our experience of all the enchanting
splendor and invigorating freshness of the physical world, we are indebted to
the senses and the symbols that serve as their intermediaries, through which
we gain news of these things . . . Think of the months, perhaps the years of
work that it would cost a physicist to define the color tones of a landscape seen
only one time, which our eye apprehends with a single look and is immediately
ready to exchange for another image.
—H. von Helmholtz
In the first chapter, I claimed that the senses are not passive receptors. Now I
will show that anticipation really is an essential characteristic of their func-
tioning, and that even at this level certain mechanisms allow internal simula-
tion of movement. I will pay particular attention to senses that are not very
well known: muscular proprioception and the vestibular system.
Indeed, to the five traditional senses—touch, sight, hearing, taste, smell—
we must add the sense of movement, or kinesthesia. Its characteristic feature
is that it makes use of many receptors, but remarkably it has been forgotten in
the count of the senses. We seem to be stuck at Aristotle’s assertion: “In the
Psychology we have given a general account of the objects corresponding to
the particular sense-organs, to wit colour, sound, smell, flavour, and touch.”1
By what twist did language suppress the sense most important to survival? A
plausible explanation is that it is not identified by consciousness, and its recep-
tors are concealed. It seems normal to us to recognize the movement of our
Vestibular receptors
Visual receptors Semicircular
canals
Otoliths
Cochlea
Muscle receptors
Cutaneous receptors
Joint receptors
Figure 2.1. Sensory receptors contribute to the sense of movement. The recep-
tors located in the muscles (neuromuscular spindles, Golgi tendon organs) and
the joints (joint receptors) all detect movements of the limbs. This is
proprioception. Cutaneous receptors detect skin pressure and friction from
contact of the limbs with each other and the external world. Vision detects
shifting images of the visual world on the retina, the position of objects in
space, shape, color, and so on. Movement of the head in space is detected from
a combination of visual and proprioceptive cues, but it is also detected indepen-
dently by the vestibular receptors. These receptors are situated in the inner ear
near the cochlea and comprise three semi-circular canals (horizontal, anterior
vertical, and posterior vertical) in three perpendicular planes and the otoliths
(the utricle, in the plane of the horizontal canal, and the saccule, in the plane of
the anterior vertical canal).
arm or vertical direction, but nothing indicates that we have receptors for
stretch and force in our muscles, for rotation in our joints, for pressure and
friction in our skin, and that in each inner ear we have five receptors (the
utricle, the saccule, and the three semicircular canals) that specifically detect
movements of the head (Figure 2.1).
Proprioception
The Sense of Position and of Velocity

In all the muscles of our body, there are very specialized little fibers that run
parallel to the muscle fibers themselves. These are the neuromuscular spin-
dles. They are sensory receptors that detect muscle stretch.
Let us conduct an experiment. If a small vibration is applied to a muscle
(for example, the biceps), around 50 to 100 Hz, oscillating at an amplitude of
only about a few hundreds of microns, we will observe two phenomena.2
First, if the arm is free to move, we will notice a reflex contraction of the vi-
brated muscle. Now, if we immobilize the arm on the edge of a table and
close our eyes as the biceps is vibrated, we will experience an illusion of un-
controllable displacement of the arm, characterized by two different percep-
tions: change of position in space and velocity of displacement. The arm is per-
ceived at the same time to be somewhere else and in motion. This illusion of
movement is often accompanied by remarkable muscle activity: the opposing
muscle (called the antagonist) to the vibrated muscle is now activated. In
other words, the brain activates the muscle, perceived to be in motion, as if it
were the perception (and not the sensation triggered by the receptors) that
leads to the contraction.3
What are the mechanisms of these two phenomena, the reflex contraction
and the conscious perception of illusory movement? Sherrington elucidated
the first one: the contraction is due to a very short reflex, called myotatic
reflex, that connects neuromuscular spindles and agonist muscle.4 When a
muscle is stretched, the spindles and their fibers are stretched as well, and they
emit discharges proportional to the length and velocity of muscle stretch in
the sensory fibers. The structure of these spindles is extremely refined and
enormously diverse: some are more tonic, others more dynamic.5 The dis-
charge makes its way to the medulla, where it induces the activity of the mus-
cle via very short neural relays. This reflex makes it possible to resist the ten-
sion exerted on the limb. The spindles detect not only the length of the
muscles but also the first derivative of length, which is velocity. In other
A SIXTH SENSE? •
27
words, the spindle discharge is maximal before elongation is complete because
velocity increases more rapidly than length. This anticipation allows compen-
sation for the delays between α motor neuron discharge and muscle contrac-
tion. A muscle actually contracts very slowly. It attains its maximum force
about 80 milliseconds after a neural command. Eighty milliseconds is a very
long time if you are trying to get away from a predator. In the language of
servo systems, the phase lead of the spindles compensates for the low-pass
filter characteristics of the muscles. The result is that when a muscle is
stretched, the time it takes to contract exactly compensates the time it takes to
stretch owing to a dynamic anticipation made possible by the properties of
spindles. Typically, the more a receptor returns a high-order derivative of the
mechanical variable it detects, the greater the likelihood the brain will antici-
pate this variable itself at some future time. Practically all the sensory recep-
tors detect the derivatives of the variables that specifically activate them. Evo-
lution obviously selected receptors capable of predicting the future.
I have explained the first phenomenon, contraction induced by vibration:
this contraction excites the sensory endings of the spindles just as natural
stretching would; it simulates stretching and thus induces a myotatic reflex,
which explains the contraction.
What is the mechanism behind the illusion of displacement of the arm?
This illusion, too, is due to stimulation of the spindles by the vibration. The
cerebral cortex, which receives information about the state of the spindles, in-
terprets this activity as elongation of the vibrated muscle. It thus works out a
perception of displacement and activates the muscle that corresponds to this
percept. This allows the cerebral cortex to situate the motor response in the
global behavioral context of the actions of the subject. When the same vibra-
tion is applied to the same muscle in different contexts, for example, when the
subject is seated or standing up, leaning on the vibrated limb or free to move,
completely different illusions result.6 The brain assigns a status to sensory in-
formation based on its assessment of the general state of the body. We are
very far from a simple potentiometer.
The spindles are not passive. The brain can modify their sensitivity. They
are accessible to motor intention. Indeed, they are endowed with small mus-
cles at their extremities that are innervated by motor neurons of the medulla
called γ motor neurons, themselves under the control of volition and motor
intention via corticospinal pathways. Activation of these γ motor neurons en-
ables the brain to simulate muscle contraction or stretch. Their discharge or
their silence induces activity or rest in the neuromuscular spindles as if the
movement had actually taken place, despite the absence of any externally im-
28 •
posed movement of the muscle or movement induced by the α motor neu-
rons that normally spark muscular contraction.
The brain thus uses γ motor neurons as a tool for modulating sensory in-
formation at its source, to adapt it to the requirements of movement, and to
simulate movement. During locomotion, the brain induces modulation of the
dynamic sensitivity of the spindles.7 I studied this property because, being
an ergonomist at the time, I had observed that a seated person subjected to
low-frequency (4 Hz) mechanical vibrations experienced significantly ampli-
fied body oscillations. I wanted to understand the origin of this resonance.
Wishing to apply variable forces to human subjects, I had the idea—I think for
the first time—of using a motor run on magnetic powder that made it possi-
ble to generate variable forces.8 This technology has since been replaced by
torque motors. A person’s arm or head, or an animal’s paw, was attached to
the machine to study the nerve responses to the application of sudden or os-
cillating forces. This machine allowed me and my colleagues to describe, in ac-
cordance with Viviani, the dynamics of the cervical column in humans and to
show that tilting movements of the head employ two centers of rotation in
different ways according to the velocity of the movement. One day we used
the machine to establish the modulation of dynamic γ commands in the
course of animal locomotion. We applied sinusoidal stimulation to a muscle
tendon during locomotion and recorded the activity of the sensory fibers in
the neuromuscular spindles of the same muscle. The goal was not to study
the response to the stimulus, but to test, by stretching, the state of the recep-
tors. We noted at the time that spindle discharge in response to stretching en-
tailed a phasic component that varied with each walking cycle, evidence of
modulation by dynamic γ motor neurons. Since then, a body of research has
confirmed this modulatory sensitivity of the sensory receptors. At each phase
of the step, the brain adjusts the sensitivity of its sensors.
The first example of a mechanism of anticipation is thus modulation of dy-
namic properties of the neuromuscular spindles according to the movement
required by the task and the context. It is worth noting that this modulation
could, in theory, allow the brain to simulate a movement without executing it.
Indeed, if the spindles are activated directly by the central nervous system,
through contraction of the γ fibers, they induce, as in the case of a vibration, a
percept of movement.
The Sense of Touch

Among the least studied of the senses and yet one of the most important is
tactile sensitivity. Our skin contains numerous receptors that are sensitive to
A SIXTH SENSE? •
29
different aspects of contact with the external world. Some—Meissner and
Pacinian corpuscles, for example—detect pressure and are endowed with
more or less phasic properties: some are sensitive to rapid variation in pres-
sure, others to its prolonged duration, still others, attached to hairs, are sensi-
tive to friction and stroking, and are activated when the hairs bend. There are
receptors that detect heat and cold and constitute a class of thermoreceptors.
And others transmit painful stimuli, called nociceptors. How rich and how ex-
tremely subtle the sense of touch is! In this case as well as in that of the other
sensory apparatus, nature has segregated the variables that the different recep-
tors detect and has made some capable of anticipation, since they detect rapid
variations (derivatives) in pressure or velocity of an object sliding over the
skin.
The distribution of these receptors on the skin is very uneven. They are
concentrated on the parts of the body that are the most involved in tactile per-
ception (the mouth, hands, genitals, thighs, soles of the feet, and so on). Their
representation in the cerebral cortex is also uneven and reflects these func-
tional differences, which vary according to species. Every region of the skin
activates particular neurons of the somatosensory cortex: in other words, the
neurons in such regions have receptor fields, as vision does. For a long time
these receptor fields were thought to be fixed and unchanging. However, re-
cent research shows that when a finger is accidentally severed, or when a small
section of a rat’s whiskers is anesthetized, the cortical projections of the tac-
tile sensors are very rapidly reorganized. This reorganization affects not only
the cortex. Simultaneous recording of the neurons of the sensory pathways
that carry information from the rat’s whiskers shows that reorganization af-
fects the thalamus as well as the cortex. It also depends on the use made, for
example, of one finger as opposed to another.9 Moreover, brain imaging has
shown that after amputation of a limb, the cortical representations of the
parts of the limb are reorganized.10
Similarly, the transmission of tactile information to the brain is subject to
an intense process of selection. Tactile transmission is facilitated when the rat
pays attention to which part of its body is being stroked. So-called presynaptic
inhibitory mechanisms can modulate or block tactile information even before
it enters the medulla, and other centralized mechanisms perform this filtering
at the level of the first neural relays in the medulla or in the brainstem.
Tactile receptors contribute to the perception of force, but they cannot
distinguish between the force that I am exerting on the weight I am holding in
my hand and the force the weight exerts owing to its mass and movement,
since these two forces are equal if the weight does not slip out of my hand.
30 •
Other centralized information like the sense of effort or recognition of the
motor command I send to the muscles (what is called corollary discharge of
the motor command) resolves this ambiguity.
The Sense of Effort: Does the Brain Use Composite Variables

to Control Movement?
Some receptors detect the effort exerted by muscle on a joint: these are the
Golgi tendon organs, encapsulated receptors located at the junction of muscle
and tendon. They fire when force is applied to the extremity of the muscle,
even at constant duration. They seem thus to be sensors of force or of varia-
tion in force. But does that mean that force is really perceived and detected di-
rectly in this way? It has not been proved. What these Golgi receptors transmit
to the medulla are signals linked to changes in force.11 The latter is then recon-
structed in the medulla by the combination of this information with that con-
cerning the length and velocity of muscle stretch. According to the theory of
nonlinear control of robotic movement developed by Slotine, composite vari-
ables (a mixture of variables and their successive derivatives) can be used by
the nervous system.12 The advantage of these composite variables is that non-
linear problems of control are made linear and considerably simplified. I and
my colleagues have proposed the idea that composite variables are also used
by the brain.13
It is not yet clear how force is represented in the firing rate of the neurons
of the motor cortex. It is doubtful that the so-called pyramidal cortical neu-
rons, which control muscle contractions, each encode the force provided by
individual muscles. The neural basis of the sense of effort remains to be dis-
covered. It appears that to detect effort the brain uses the motor command it-
self. Indeed, a person whose arm is numb (because the sensory receptors have
been anesthetized) can still produce an effort of a given magnitude. However,
the debate between the two theories—peripheralist, which stresses the role of
sensory information, and centralist, which stresses the role of internal mes-
sages—remains unresolved. The respective roles of these two components in
perceiving force probably depend on the task and the context in which the
movement occurs. A recent experiment elegantly demonstrated the central
nature of the sense of effort: A subject was asked to accomplish a movement
while walking; then he had to stand still and imagine the movement.14 It takes
the same amount of time to mentally execute a motor task as to actually exe-
cute it. This “isochrony” is used by coaches in training champions. If the sub-
ject is asked to take the same walk while imagining that he has a weight on his
shoulders, the length of the mental journey is modified. Moreover, this virtual
A SIXTH SENSE? •
31
weight leads to changes in cardiac rate and several other vegetative indices, as
if the effort really were greater.15 Imagining the effort detected by the recep-
tors thus has the same consequences as its reality.
Do the Golgi receptors, like the neuromuscular spindles, have an efferent
control that allows the central nervous system to regulate them through antic-
ipation? So far, it would appear not. On the other hand, as soon as they enter
the medulla, their axons are subjected to control by a presynaptic inhibitory
mechanism that makes it possible to block transmission of neural messages in
the first relays. Via the γ motor neurons, the brain can regulate the stiffness of
muscles and thus the way in which they will respond to a force applied to
them.
The Vestibulary System: An Inertial Center?
The brain is thus capable of recognizing movements of body segments, select-

ing and modulating information supplied by the receptors at the source. But
proprioceptive receptors can themselves only detect relative movements of
body masses. They are inadequate for complex movements—locomotion, run-
ning, jumping—where the brain must recognize absolute movements of the
head and body in space. To do this, nature has, since the earliest stages of evo-
lution, used inertial cues, so called because they detect the forces of inertia.
These are the vestibular receptors.
The Discovery: Scarpa, Flourens
The Italian anatomist Scarpa deserves the credit for having described, in 1789,
the anatomy of the three semicircular canals and the two otoliths located in
the inner ear.16 Flourens subsequently discovered the function of these recep-
tors in 1832.17 A student of Cuvier at the Museum of Natural History, then
professor at the Collège de France, Flourens studied lesions. His approach to
the brain ran counter to the most advanced ideas of his era: he disagreed with
Gall’s localizationalist phrenology and rejected Darwin’s theory of evolution.
But he was a good experimenter, attentive to the unexpected. One day he in-
troduced a lesion in the inner ear of a pigeon, expecting to get an auditory dis-
turbance; but the animal lost its balance, went around in circles, and fled to
the dark corners of the room. Flourens very rightly inferred that the ear does
not function only to hear, and he thought, this time incorrectly, that he had
discovered a “moderating apparatus of cerebral activity.”
Fifty years passed before the laws that govern the directional sensitivity of
the semicircular canals were formulated. Around 1910, these receptors at-
32 •
tracted the attention of the physicists Mach and Helmholtz, polymaths at a
time when that was not so unusual. They understood that the canals detect
angular acceleration and the otoliths linear acceleration, and that together
they constitute a genuine gravito-inertial receptor system.
The semicircular canals are essentially inertial receptors. They function
without a base and, consequently, are as advantageous to the bird flying as to
the lion capturing its prey or the monkey jumping from branch to branch. But
this ostensibly advantageous characteristic masks a weakness. Indeed, in 1911
Einstein showed that an inertial sensor cannot distinguish between gravity and
an acceleration of the same magnitude (9.81 meters per second) due to a
movement. The first physicists who became interested in the vestibular recep-
tors discovered the ambiguity of vestibular signals, which confuse head tilt, ac-
celeration, and deceleration. Visual inputs are necessary to resolve this ambi-
guity, which happens to account for many airplane accidents and otherwise
unexplained falls (see Chapter 13).
A Basic Euclidean Frame of Reference

The semicircular canals are rings of a sort filled with a viscous liquid—the
endolymph. At the end of each canal is a bulge, the cupula, that contains the
hair cells that constitute the vestibular receptor. These hair cells are sensitive
to the pressure that the endolymph exerts on them. I will not delve into the
details of the mechanism by which this pressure is transformed into nerve dis-
charge, but it is important to understand which mechanical variables produce
the variations in pressure that activate the receptors: these are angular acceler-
ations of the head—in other words, changes in velocity. Thus, these receptors
detect the second derivative of angular displacement; some vestibular recep-
tors are even sensitive to the third derivative of movement, or jerk (expressed
in degrees per second per second).
When you turn your head sharply, the endolymph lags behind the head
and the membrane of the canal that contains it. This lag is due to the forces of
inertia, defined by the basic relationship of dynamics: to put a body into
movement takes a force proportional to the product of its mass times acceler-
ation. These forces are connected to the acceleration of the head. The greater
the acceleration, the greater the movement of the endolymph in the tube
of the canal. The semicircular canal is thus a sensor for angular acceleration of
the head.
It also detects braking, which is another kind of acceleration (negative).
Try the experiment yourself. Stand up with your eyes closed, and turn around
four or five times at a constant speed. Then, stop yourself abruptly by tapping
A SIXTH SENSE? •
33
hard with one foot so that the stop is clean. Keep your head still, and you will
distinctly perceive an illusion of rotation in the opposite sense. This sensation
derives from the fact that the cupula of the horizontal semicircular canal,
which is subjected to the forces of inertia during a sudden deceleration, devi-
ates slightly and takes several seconds (around 10) before returning to its posi-
tion of rest. If you lay your fingers lightly on your closed eyelids, you can even
feel the movement of your eyes caused by the vestibular stimulation. This
movement is called nystagmus.
The sensitivity of these receptors is exquisite—they detect accelerations of
a few tenths of degrees per second per second—and depends on the frequency
of movement. Two words about this concept. When we walk, our heads oscil-
late at a frequency (number of oscillations per second) of around once per
second, or a frequency of 1 Hz. When we run, the frequency increases. When
we are standing still, waiting for the bus, our heads oscillate very slowly, at
around one oscillation every 10 seconds; a frequency of 0.1 Hz. This terminol-
ogy can also be used in the case of simple movements. For example, if you
turn your head once to the left, this single rotation can be regarded as one cy-
cle of a periodic movement. The rate of oscillation of this periodic movement
can be determined by doing a Fourier analysis of the movement signal. Low
frequencies correspond to slow movements and high frequencies to rapid
movements. Each movement contains a certain number of frequencies that
determine what is called the band of frequencies. The considerable advantage
of this description is that it permits the description of the dynamic properties
of movement using only a few numbers, the receptors that detect the proper-
ties, and the motor commands.
The geometry of the semicircular canals is adapted to each species.18 The
fluid that transmits the forces of inertia to the sensory cells of the canal—the
endolymph—has remained unchanged throughout the millions of years of
evolution. Consequently, the diameter of the canal has increased only slightly
compared with the size of animals. Why? The geometry of the semicircular
canal is adapted to the rapidity of the movements of the species, that is, to the
frequency of physical movements (between 0.1 and 1 Hz). Indeed, in going
from the mouse to the elephant, it was necessary to attenuate the frequencies
to which the receptor is sensitive, since the elephant moves a lot more slowly
in walking than the mouse. Geometry solved a mechanical problem over the
course of evolution. What extraordinary adaptation!
Thus, the semicircular canals are sensitive to dynamic aspects of head
movement: acceleration and its derivative, jerk. The analysis of feedback and
34 •
control systems, introduced in neurophysiology by Hartline, Terzuolo, and
others in the 1960s, provided a quantitative method for evaluating receptor dy-
namics: a nerve fiber was recorded while the animal was rotated on a turnta-
ble that was subjected to sinusoidal movement. Varying the frequency of oscil-
lation (the number of oscillations per second) made it possible to obtain the
response of the receptor to movements of differing rapidity. It just required
measuring the amplitude of the modulation of neural discharge and the tem-
poral relation between discharge and movement (phase).
Consider the case of rotations in the horizontal plane, a movement you
make hundreds of times each day in turning your head. When an animal is
subjected to sinusoidal horizontal rotations alternating clockwise and counter-
clockwise and whose frequency of repetition goes from one cycle every 100
seconds (0.01 Hz) to five per second (5 Hz), the neurons of the vestibular re-
ceptors fire with a frequency that also varies in a sinusoidal manner. The am-
plitude of their modulation is weak for the slowest movements (0.01 Hz), then
increases with frequency, remains stable between 0.1 and 1 Hz (the frequency
of everyday movements), then increases again beyond that. The delay with
which the neurons fire in relation to the displacement of the head varies with
the frequency of oscillation. They are in phase with acceleration for slow
movements, which is normal since the semicircular canals are receptors for an-
gular acceleration; then they are in phase with the velocity of the head for the
frequency of natural movements (between 0.1 and 1 Hz) and with very rapid
movements, like those made when falling down; and they come back into
phase with acceleration, which makes possible the rapid reactions needed for
getting back up.
Generally, the signals from these receptors enable anticipation of the fu-
ture position of the head owing to their sensitivity to derivatives (jerk, acceler-
ation, and velocity) of movements of the head. These receptors not only
transmit signals to the brain, they are also subjected to centrifugal control
from the central nervous system that modulates their properties. As a matter
of fact, efferent neurons project from the brainstem to the receptors of the
sensory cell. The exact mechanism of centrifugal control is not yet completely
understood. Nor is it the only one: interactions between receptor cells enable
retrocontrol of sensory messages; that is, rapid and subtle adjustment of the
sensitivity of the cell to the origin of the sensory signal.19 The sensory mes-
sage is thus processed by the brain at its source. The same two major princi-
ples already described for the muscle receptors also apply to the canals: detec-
tion of derivatives of movement that enable anticipation, and modulation of
A SIXTH SENSE? •
35
the message at the source by the central nervous system. One of the basic
functions of the vestibular system is thus to detect movements of the head in
a Euclidean reference system.
Very early in evolution the canals appeared in certain animals: the lamprey,
one of the oldest among the fishes, has two sets. Mammals have three of them
on each side of their heads. The canals are situated in three approximately
perpendicular planes, though not the ones you might guess merely by observ-
ing animals from outside. One of the planes corresponds to the horizontal
plane of the head. To locate it in humans, just look at a person from the side
and plot a line from the meatus of the ear to the external edge of the eye. The
horizontal canal is in a plane that overhangs this line by 20 degrees (see Fig-
ure 2.4).
The two remaining planes of the canals are at 45 degrees with respect to
the frontal and saggital planes of the body.20 Why is this anatomical detail im-
portant? Because the three planes so constituted form a basic egocentric frame
of reference with respect to which, as I will discuss later, our entire perception
of movement in space is organized. The geometry of the canals dictates the
organization of the cerebral analysis of visual movement and perhaps also
other movements. It might yet prove to be the basis of Euclidean geometry.
Poincaré was interested in this question. In his book La valeur de la science
(The Value of Science), he commented on the theories that pitted Cyon, on
the one hand, against Mach and Delage, on the other. According to Cyon,
“The three pairs of canals would have as sole function to tell us that space has
three dimensions. Japanese mice have only two pairs of canals; they believe, it
would seem, that space has only two dimensions . . . The lampreys, having
only one pair of canals, believe that space has only one dimension, but their
manifestations are less turbulent.”21 Poincaré objected to this view and main-
tained, along with Mach and Delage, that the semicircular canals “contribute,
therefore, to inform us of the movements that we have executed, and that on
the same ground as the muscular sensations.”22 For Poincaré, “the sense-or-
gans are designed to tell us of changes which happen in the exterior world. We
could not understand why the Creator should have given us organs destined
to cry without cease: ‘Remember that space has three dimensions,’ since the
number of these three dimensions is not subject to change.”23
Elsewhere Poincaré asks an important question: Why are Euclidean move-
ments perceived as actual movements? He says, “Our mind adapted itself to
the conditions of the external world through natural selection; it adopted the
geometry most advantageous or, to put it another way, most convenient to the
species. Geometry is not true; it is advantageous.”24 “One geometry cannot be
36 •
more true than another; it can only be more convenient. Now, Euclidean geom-
etry is, and will remain, the most convenient: 1st, because it is the simplest,
and it is not so only because of our mental habits or because of the kind of di-
rect intuition that we have of Euclidean space; . . . 2nd, because it sufficiently
agrees with the properties of natural solids.”25
What are the relationships between geometric space and representative
space? Are their properties similar? “It is often said,” Poincaré writes, “that the
images we form of external objects are localised in space, and even that they
can only be formed on this condition. It is also said that this space, which thus
serves as a kind of framework ready prepared for our sensations and represen-
tations, is identical with the space of the geometers, having all the properties
of that space.”26 However, he argues, geometric space is continuous, infinite,
three-dimensional, homogeneous (all its points are identical), isotropic (all the
straight lines that pass through the same point are identical). In contrast, rep-
resentative space (which is primarily visual space) is two-dimensional (retinal
space); it becomes three-dimensional owing to convergence and accommoda-
tion; it is not homogeneous, because the part of the retina that is most sensi-
tive to shape, the fovea, is nonhomogeneous with the periphery. It is thus not
isotropic.
Poincaré concludes from this that representative space is only an image of
geometrical space distorted by the rules that govern how our perceptive appa-
ratus works. “Thus we do not represent to ourselves external bodies in geomet-
rical space, but we reason about these bodies as if they were situated in geo-
metrical space.”27 Modern psychophysics confirms this intuition.
Poincaré introduces a fundamental notion that is essential to my own the-
ory: “To localize an object simply means to represent to oneself the move-
ments that would be necessary to reach it. I will explain myself. It is not a
question of representing the movements themselves in space, but solely of
representing to oneself the muscular sensations which accompany these
movements and which do not presuppose the preexistence of the notion of
space.”28 But then, if we cannot represent geometrical space for ourselves, and
if our representation of space is in fact only what I call a simulation, that is,
the movements that we need to make to travel it, the question becomes one
of knowing how the idea of geometric space could even have come about. We
must, says Poincaré, look at the way we evaluate how change affects an object.
An object can undergo a change of state or of position. In either case, the
change is signaled “by a modification in an aggregate of impressions.”29 By this
Poincaré means something very close to what I call a configuration of states
of sensory receptors. Inasmuch as position can be corrected, the brain distin-
A SIXTH SENSE? •
37
guishes change of state and change of position: “We could restore the primi-
tive aggregate of impressions by making movements which would confront us
with the movable object in the same relative situation . . . [A] motionless be-
ing,” he adds, “could never have acquired [the idea of space], because, not be-
ing able to correct by his movements the effects of the change of position of
external objects, he would have had no reason to distinguish them from
changes of state.”30 Sight and touch would not be able to give us a sense of
space without the “muscular sense.”
These processes require that objects be rigid. If there were no solid bodies
in nature, there would be no geometry. For Poincaré, geometry is thus the de-
scription of phenomena called displacements, that is, external changes that
can be compensated for by changes in our own bodies detectable by the ves-
tibular receptors.
The Distinction between Tilt and Acceleration

The second type of receptor in the vestibular system—the otolith—is also an
accelerometer that detects linear translations (it is even found in the spiny lob-
ster, where it is called a statocyst). In addition, it detects the direction of grav-
ity, and thus tilting of the head. Gravity is a particular acceleration, indiscern-
ible, as Einstein demonstrated in 1911, from a linear acceleration (Figure 2.2).
When the head tilts, a component of gravity acts on the receptor and thus
supplies an indirect measure of tilt of the head with respect to earth’s field of
gravity. Animals obviously discovered early in the course of evolution that
gravity is a very useful frame of reference, present at any point on the earth
and constant during any movement. Even before crawling out of the water
and having to construct mechanisms to compensate for the mechanical effects
of gravity—“this sticky force,” in the words of an astronaut returning from a
space trip and a little sad at finding himself glued to Earth—marine animals
used gravity as a reference.
In insects and batrachians, the otoliths also pick up mechanical vibrations,
which permits the animal to detect the presence of a predator and to initiate
flight reactions. So in these animals, the otoliths are a sort of ear that is sensi-
tive to very low-frequency vibrations. In fish, birds, and mammals, the otoliths
have reached a high degree of refinement and simultaneously detect both lin-
ear acceleration and static tilt of the head. They are composed of two cavi-
ties—the utricle and the saccule—that contain crystals floating in the same
fluid (the endolymph) as that of the semicircular canals (Figure 2.3; see also
Figure 2.2). When the head is displaced in translation, force is exerted on the
crystals owing to the linear acceleration of the head. This force induces a tiny
38 •
Otoliths
Hair
cells
Vestibular nerve
Gravity
Gravity
Figure 2.2. The otolithic receptors. The upper part of the figure shows schemat-
ically the mass of crystals floating in the endolymph, which together constitute
the otoliths. The endolymph is not shown. The hairs of the sensory neurons
are situated beneath the crystal mass. The lower part of each drawing depicts
the base of the receptor (called the macula). Ionic changes triggered by move-
ment of the hairs result in discharge of the sensory cells that travels along the
vestibular nerve to the brain, which is thereby informed about the magnitude
of acceleration of the head. In this case, gravity has no effect on the receptor,
because it is perpendicular to it. If the receptor is tilted, the gravitational com-
ponent is in the plane of the macula. This receptor indicates both acceleration
and tilt of the head. The receptor shown here is the utricle, which detects hori-
zontal accelerations of the head; another receptor, called the saccule, is located
perpendicular to it and detects accelerations and gravitational force in the verti-
cal plane.
A SIXTH SENSE? •
39
A
Acceleration
Acceleration
Inertial force
Re
su
lta
nt
Gravity
Gravitational
component in
the plane of
the otoliths
Gravity
Figure 2.3. Ambiguity results when the head is simultaneously accelerated and
tilted. (A): If the head accelerates forward, a backward inertial force is exerted
on the crystals of the utricle. The combination of this force and the downward
gravitational force produces a resultant tilting force. The brain has the illusion
that this resultant is gravity. And because it uses gravity as a fixed reference
with respect to the earth, the brain deduces that the head is tilted. (B): When
the head is motionless but tilted backward, a gravitational component in the
plane of the receptor gives the impression of a forward acceleration. (The size
of the receptor has been considerably enlarged.)
40 •
displacement of the crystals, which leads to activation of the vestibular nerves.
The same activation is produced by tilting the head. In this last case, tilting in-
duces a modification of the magnitude of the component of gravity acting on
the crystals. They are thus both accelerometers and “tiltometers.” One could
simply say that the saccule (the sensor of vibrations in insects and batrachia),
whose plane is close to that of the vertical anterior semicircular canal, detects
vertical linear accelerations of the head (when you jump up and down with
your feet together, for example), and that the utricle, whose plane is close to
that of the horizontal semicircular canal, detects horizontal linear accelera-
tions of the head (when your car brakes or accelerates, for example).
These two kinds of vestibular receptors, the canals and the otoliths, detect
movement of the head in space without needing a reference point. They con-
stitute a genuine inertial system like those mounted on board airplanes. More-
over, they use gravity as a reference to supply the brain with information
about the static tilt of the head. In microgravity—in space stations—these re-
ceptors can thus always detect angular or linear accelerations of the head, but
can no longer rely on gravity for detecting angle of tilt. The brain has to rein-
terpret the information transmitted by these receptors to be able to compare
them with sensations from other proprioceptive and visual receptors.
Finally, let me reiterate a point that is often poorly understood: vestibular
receptors are not only sensors of movement. They also signal immobility. In-
deed, these sensors have a basic discharge whose lack of variation is inter-
preted by the brain as immobility. They are essential for assessing what is
called the subjective vertical (see Chapter 4).
Although they make it possible to recognize head movements, vestibular
receptors alone are not enough, because of the ambiguous information they
provide. For example, they cannot distinguish between acceleration of the
head in one direction and braking in the other direction. Tactile or visual in-
formation helps to resolve this ambiguity. This clearly shows how perception
requires cooperation among the senses.
Another ambiguity that has already been mentioned is the distinction be-
tween the angle of tilt of the head and linear acceleration. This is known as
the problem of gravito-inertial differentiation, and I will refer to it again in Chap-
ter 13. To this problem nature found a solution, called frequential filtering. It
consists in separating the two components of movement at the level of the
first sensory relays, in the so-called vestibular nuclei and perhaps even at the
level of the fibers that connect the receptors to these first relays. In fact, in the
vestibular nuclei there are two kinds of neurons. Some respond especially to
slow variations in acceleration, but are indifferent to rapid movements. The
A SIXTH SENSE? •
41
solution is an elegant one, because detecting the angle of tilt of the head in-
volves slow neurons, and detecting acceleration involves rapid neurons. Oth-
ers are highly activated by very rapid movements of the head and only slightly
by slow movements. Differentiation of gravito-inertial forces is resolved by
segregation at the very level of capture of sensory information.
This very early capacity for segregation is a basic feature of sensory ana-
lyzers. I will discuss it further in Chapter 3.
A Fundamental Dissociation: Rotation and Translation

One of the most puzzling questions raised by the study of the properties of
the vestibular system is that of dissociation between detection of rotations by
the semicircular canals and detection of translations by the otoliths. Does the
geometrical and kinematic distinction between the concepts of rotation and
translation have a biological basis? It is tempting to think so.
More generally, the nervous system, of which the senses are an extension,
actually begins to segregate information from the initial inputs. Detection of
angle, length, and force is dissociated at the level of the muscle and joint re-
ceptors. In the case of the vestibular system, rotation and translation are disso-
ciated at the level of the receptors. Next, the first central sensory relays begin
filtering out inertial movements and static tilt of the head, as I discussed ear-
lier with respect to the function of otoliths and their perceptual ambiguities.
Segregation also happens very early in the visual system (see Figure 3.1). For
example, already at the level of these first relays—the lateral geniculate nu-
cleus—a definite segregation occurs between two pathways. The first is the so-
called magnocellular pathway, whose neurons are insensitive to color and sen-
sitive to movement, with high sensibility to contrast and a weak capacity for
distinguishing visual elements that are close together (low spatial resolu-
tion)—in other words, favorable for transmitting information about move-
ment. The second pathway is the so-called parvocellular layer, whose neurons
are sensitive to color, slow to respond, and have weak sensitivity to contrast
and a high spatial resolution. This segregation is enhanced at the next level of
the primary visual cortex. Visual information arriving via the deep layers,
which are different for movement and color, is transmitted to the superficial
layers where aggregates of neurons sensitive to shape and color are found,
whereas information about movement is transmitted to the next cortical relay
in the mediotemporal (MT) lobe, also called V5. This segregation is main-
tained in different forms in the visual areas V2, V3, and V4, the last of which,
for example, is specialized for processing color. Later on I will show that infor-
42 •
mation about visual movement borrows yet another pathway called the acces-
sory optic pathway, where geometric segregation happens early on.
It is as if the senses were specialized organs for detecting variables perti-
nent to the survival of each species, nature having put in place very periph-
eral mechanisms for selecting these variables and dissociating them. Once
the brain has these bits of information, it reconstructs them into more com-
plex messages and encodes them for adaptation to various central operations
needed for perception and action. Gibson stressed the idea that the senses are
analyzers that do not merely detect the physical variables likely to stimulate
their endings, but that in a way they have integrated into their functioning the
elements of nature important for the animal’s repertoire of actions. The early
segregation of sensory information by the senses and their first analytical neu-
ral networks supports this idea and does seem to suggest that the transduction
of physical variables (light, sound, pressure, etc.) responds to preexisting ques-
tions that the nervous system asks the world.
The Functions of the Vestibular System
Postural Stabilization
The vestibular system ensures postural stability. If we trip over a root, our
head accelerates, and the canals and the otoliths are activated, exciting the so-
called vestibulospinal reflex pathways that initiate the reactions needed to re-
adjust posture: lifting the head and the rest of the body by means of proprio-
ceptive reflexes whose receptors are located in the neck. I will not describe this
function, known since the work of Magnus31 and Rademaker32 at the begin-
ning of the century and studied by the Italian school of neurophysiology dur-
ing the 1950s, the Japanese school of Fukuda, and by Roberts at Edinburgh.
What I will stress is the anticipatory properties of these reflexes. The anticipa-
tory power of the vestibular receptors comes from the fact that they detect ac-
celeration. Now, acceleration is maximal from the beginning of a fall or loss
of balance. The reflexes for getting back up are thus very rapid since they are
initiated as soon as the perturbation begins. The conjunction of the discharge
of the tactile receptors of the spindles sensitive to the velocity of stretching—
for they are sensitive to variations in pressure—and the vestibular receptors
sensitive to acceleration supplies the nervous system with preliminary infor-
mation about the nature of loss of balance. The muscle contractions that lift
the body are the result of this anticipation, and the swiftness of the postural
A SIXTH SENSE? •
43
adjustment is further facilitated by virtue of the brain’s repertoire of ready-
made postural reactions. These stereotypical reactions are triggered by spe-
cially configured sensory signals that set in motion what are called synergies
(see Chapter 7). The vestibular receptors are initiators, not just detectors. The
information they convey is a sign, not just a signal.
Gaze Stabilization
Perceptual stabilization is the second major vestibular function. Indeed, if the
world appears stable when we move, it is first and foremost owing to reflexes
of vestibular origin that stabilize the image of the world on the retina. These
reflexes connect the vestibular receptors to the muscles of the eye. The anat-
omy of neural connections is such that moving the head in one direction leads
to moving the eye in the other, which has the effect of suppressing, or dimin-
ishing, the shifting of images on the retina (retinal slip).
You can very easily see for yourself. Just look at a point on the wall in front
of you. Close your eyes, turn your head while imagining the point on the
wall, and open your eyes. You will notice that your gaze has remained on the
point that you memorized. Your eye made a movement in a direction contrary
to that of your head, and of the same amplitude. Its origin is vestibular: it is
produced, in the horizontal plane, by the semicircular canals. But how do I
know, you say, that the muscle receptors in the neck have not detected the
movement of my head? To answer your question, you have only to sit in a
rolling chair and ask someone to repeat the experiment by turning you to one
side or the other. This time, as your eyes are closed during rotation, only the
semicircular canals detect the rotation (the tactile receptors alone are not suf-
ficient to detect a rotation angle). You will see that your gaze has remained
fixed on the same point on the wall. Thus the semicircular canals are able to
do the job by themselves. (However, the receptors in the neck do help during
natural rotations of the head, cooperating with the vestibular receptors.)
But, you say now, how do I know that it is a reflex? Since I am introducing
the memory of a target on a wall, it is very likely that I am also involving my
cerebral cortex and the structures that are implicated in spatial memory. You
are correct. In fact, the vestibulo-ocular reflex is nothing like a simple reflex;
its amplitude is regulated by the cortex, and I will show in Chapter 3 that ves-
tibular information is transmitted to the cortex and that the zones that receive
it project in their turn to the vestibular nuclei.
Around 1930, Lorente de Nó discovered a network of neurons in the
brainstem that connects the semicircular canals and the muscles of the eyes.33
He suggested that the reflex is subtended by an arc of three neurons (Fig-
44 •
Cerebellum
(flocculus)
Vestibular nuclei
Semicircular
canals
Motoneuron
Figure 2.4. The vestibular reflex stabilizes retinal images during head move-
ments. What is called the vestibulo-ocular reflex is the product of the network
of neurons that join the vestibular receptors to the muscles of the eyes. Vestib-
ular information is relayed to the cerebellum via the Purkinje neurons, inhibi-
tory neurons that coordinate and modulate the reflex (black circle). The part of
the cerebellum involved in rotations is the flocculus.
ure 2.4): a peripheral neuron that connects the semicircular canals to the ves-
tibular nuclei of the brainstem; an intermediary neuron, called the second-or-
der vestibular neuron, whose cellular body is in the vestibular nucleus and
which projects to the motor nucleus called abducens (motor nucleus VI); and
finally a neuron that causes the muscles of the eyes to contract (motor neu-
ron). A second reflex pathway includes a mechanism for generating the rapid
A SIXTH SENSE? •
45
phase in the form of a cascade of neurons situated in the reticular formation.
Around 1965 Szentagothai established the exquisite correspondence between
each canal and each of the three pairs of eye muscles.34 The basic anatomy of
a reflex was thus pieced back together.
It was Eccles’s perfection of the methods of intracellular recording that
furthered the understanding of the synaptic organization of the reflex.35 This
organization turns out to be both very simple and very subtle. A network of
excitatory and inhibitory synapses underlies it and ensures a very rigid kind of
“push-pull” functioning. Moreover, indirect pathways alter signals of move-
ment and produce vestibular nystagmus. Finally, circuits that traverse the cere-
bellum exert feedforward control on the reflex. The discovery that I made,
with Baker, of the role of the prepositus hypoglossi nucleus as a supplementary
reticular relay for control pathways further complicates the organization of
this reflex, which is composed of a series of parallel pathways that interact
one with the other.36 But this is not the place to describe this remarkable
mechanism.
Just when the complexity of neural organization could have trapped elec-
trophysiologists in a profusion of details, systems theory emerged with several
welcome simplifying concepts.
The Concept of the Neural Integrator

The way the vestibulo-ocular reflex operates involves a paradox that is worth
analyzing to understand how the nervous system manages to handle speed
and precision at the same time. As noted earlier, the vestibular receptors are
accelerometers (measuring degrees per second per second) that detect move-
ments very rapidly, since certain sensory cells are even sensitive to jerk, which
is a derivative of acceleration. But the command that displaces the eye in its
socket necessarily produces a rotation, specified as degrees of angular mea-
sure; so although the brain gains from an early, dynamic piece of information,
it must transform it by delaying it to control the position of the eye with re-
spect to the environment: it must be subjected to low-pass filtering, or integra-
tion. An initial filtering, or first integration (going from acceleration to veloc-
ity), is carried out at the level of the receptors themselves. In fact, they are
endowed with viscoelastic properties that delay the signals in a variable man-
ner according to the rapidity of the movement. But a second integration must
take place in the brain to go from velocity to position.
A part of integration is certainly carried out at the level of the vestibular
nuclei, for the response to rotation is delayed at this level. If, instead of being
46 •
subjected to sinusoidal stimulation, the head is briefly rotated, the fibers of the
vestibular receptors respond very quickly, and their response diminishes spon-
taneously after about 5 to 12 seconds once the head has stopped moving. The
neurons of the vestibular nuclei do not stop firing until about 20 to 25 seconds
afterward. You can feel this effect by closing your eyes and turning your head
very rapidly, and concentrating on the impression of rotation that follows: you
will continue to feel it for about 20 seconds. But a supplementary integration
intervenes between the vestibular nuclei and the movement of the eyes.
The research on this central neuronal integrator has greatly fascinated
physiologists. To avoid confusion, let me make clear that I am not talking
about the concept of integration as Sherrington used it, in the sense of a com-
plex combination of multiple signals to “integrate their messages,” but rather
in the literal mathematical sense of progression from the derivative of a vari-
able of movement to its integral; that is, of acceleration to velocity or of ve-
locity to position. The theory of servo systems, which was developed mainly
during World War II to control radar and which flourished in the 1950s,
was very useful for expressing these dynamic transformations in quantitative
terms. Engineers use concepts borrowed from the analysis of servo systems to
describe these dynamic transformations.
At least four mechanisms have been proposed to explain this integration.
Lorente de Nó was the first to suggest that simply returning discharge from a
neuron to itself in a so-called re-excitation or recurring excitation loop sufficed
to maintain activity and produce integration. The same effect could be pro-
duced by an extended long loop action involving other groups of neurons. A
third hypothesis was the intervention of inhibitory mechanisms that could
produce the same effect (lateral inhibition). Finally, a cascade of successive
partial integrations in a nucleus close to the vestibular nucleus, the prepositus
hypoglossi nucleus, could lead to this progression from velocity to a signal of
position. The question has not been resolved, although Baker and his col-
leagues have identified precise zones of the brainstem in the fish where these
networks of neurons that enable integration are probably located. It is also
possible that integration is due to so-called intrinsic properties of the same
neurons. For example, certain neurons can respond to a burst of activity with
a continuous discharge. This mechanism involves ion channels in the neuronal
membrane that depend on the concentration of calcium ions. Settling on the
right mechanism is a matter of time. It will require new methods that are only
now being developed and that will permit manipulation of the properties of
membranes in the intact brain.
A SIXTH SENSE? •
47
The Problem of Geometry: Is the Brain a Tensor?
How does the brain handle problems of geometry? Control of a reflex as sim-
ple as the vestibulo-ocular reflex does not consist merely in resolving relational
problems among acceleration, velocity, and position—in other words, prob-
lems of dynamics. The six muscles of the eye and the thirty muscles of the
head, situated in very varied orientations, also must be controlled by the three
semicircular canals and the otoliths, which are located in three planes. Re-
stating the problem will better indicate its complexity. Imagine that you are
sitting in a boat and a wave comes and rocks the boat; your head will move in
rotation with the boat. If you want to stay upright, you will have to make a
movement of straightening your body, equal to and in a contrary sense from
that of the boat. Your semicircular canals (ignoring the otoliths for now) de-
tect three projections of this rotation in three perpendicular planes. Thus you
have three values that represent the projection of the true rotation in a system
of special coordinates whose axes are at right angles. Put simply, the rotation
is encoded as covariant coordinates. The numerical values of these coordi-
nates constitute a vector that can be represented by an arrow whose magni-
tude and direction are measures of the intensity of the sensation detected by
the receptor.
A system of coordinates encodes the movement that the eyes, head, and
even the trunk must make to compensate for this rotation due to the motion
of the boat. These systems of coordinates are the directions of tension of the
muscles of these different parts of the body. The brain thus has to transform
the sensory information, encoded in covariant coordinates, into motor com-
mands expressed in the systems of coordinates of the muscles and their planes
of action.
Pellionisz and Llinás maintained that certain portions of the brain trans-
form sensory coordinates into motor coordinates.37 They observed that as
there are in general more dimensions in muscular frames of reference than in
sensory frames of reference, in principle the problem of transformation has
many solutions. One of their solutions was the idea that perhaps the brain
functions like a tensor. A tensor is a group of mathematical operators called
matrices that carry out transformations between vectors. They have proper-
ties that are irrelevant to the discussion here. P. Churchland’s book Neuro-
philosophy contains a summary of this theory.
Pellionisz and Llinás applied their theory to two examples: the vestibulo-
ocular reflex, which stabilizes gaze, and the vestibulo-nuchal reflex, which lifts
the head when the boat rocks.38 In the case of the vestibulo-ocular reflex, they
48 •
assumed that rotations of the head are encoded covariantly by the vestibular
receptors and that several successive transformations precede the motor com-
mand that turns the eye in the coordinates contravariant to the six muscles of
the eyes. In another model, they suggested that the cerebellum was the struc-
ture that carries out the most important transformations.39 I will try to explain
broadly the basic idea of these models. When the motor command is sent to
the motor neurons of the neck muscles, it is distorted by the geometrical
properties of the components of the forces (contravariant projection) and by
the mechanical properties of the head and neck. To obtain a movement of the
head that is in perfect opposition to that of the boat such that the vestibular
receptors can detect it, one solution, invented by roboticists, is that the com-
mand signal itself be already distorted in an inverse way before it is sent to the
muscles. The simplest image I have been able to come up with to represent
this process is that of a casting mold, which is in a way the inverse of the ob-
ject it is intended to shape. In fact, we use this strategy in many movements;
for example, if we need to lift an object that is hanging at the end of a rubber
band, we make a much bigger movement, because we are anticipating the
elasticity of the rubber band. The advantage of the internal model that ac-
complishes the inverse is precisely that we do not have to produce a different
motor command: it is automatically transformed. This inverse transformation
is assumed to take place in the cerebellum.40
To sum up: The movement of the boat is detected by the vestibular recep-
tors in three planes. It is thus represented by three vectors. This information is
sent to a nerve center (in this case, the cerebellum), which transforms these
signals to supply them with properties that will, in advance, compensate for
the distortions the signal will undergo in the movements coordinated by the
muscles, as a function of the mechanics of the limbs. The signal is then sent to
the muscles; it undergoes contravariant transformations of the coordinates,
and the head is stabilized.
Even if the brain is not a tensor, and if today the subscribers to this theory
are not legion, the fact remains that the theory of tensors has forced a genera-
tion of physiologists to take very seriously the question of geometry and the
solutions that nature appears to have found to simplify neurocomputation.41 In
addition, this theory contains an important idea, that of the internal model,
which can be found today in the most progressive theories. I will introduce
other examples in Chapter 8.
I would like to make a general observation, central to the premise of this
book. The examples taken from the theory of servo systems and of tensors
have in common that they consider reflexes and sensorimotor systems to be
A SIXTH SENSE? •
49
continuous chains of transformations that proceed from sensation to motor
command. Although this approach incorporates the idea of an internal model
and, thus, an important anticipatory mechanism, it gives short shrift to the in-
fluence of action on sensory processing. It is thus very far from a projective
conceptualization of the nervous system.
Seeing Movement
The study of the role of vision in perception keeps an enormous number of

physiologists, psychologists, and mathematicians busy doing research on cere-
bral sensory mechanisms. I will not attempt to review this work here. A hege-
mony of research and ideas concerning cortical vision has become established
in the neurosciences that has overshadowed the importance of subcortical vi-
sion, which, however, preceded cortical vision by several millions of years
with refinements whose subtlety we are nowhere near to understanding. This
hegemony of research on cortical vision is particularly troubling because it
leads to isolation of its functioning from that of other senses. Moreover, as-
sumptions about how the visual cortex is organized—for example, its architec-
ture in columns—have totally obscured the importance of the transversal or-
ganization of cortical functions for reasons that have to do as much with
sociology and science as with the rigor of the data.
This dictatorship of vision had several causes: First, the idea that vision is
the most highly developed sense in primates and in humans and that, together
with language, it is what makes humans distinctive. Furthermore, the manipu-
lation of visual stimuli is still very simple in comparison with manipulating
stimuli to the other senses; so much so that it has easily allowed implementa-
tion of the dominant stimulus–response paradigm in neurobiology. In addi-
tion, vision is a fascinating model of sensory activity, because psychophysical
and neural research can be pursued in parallel.
Finally, cortical vision induces a conscious perception that is obviously
connected to the organ, whereas it is impossible to introspectively connect
conscious perceptions or actions resulting from the subcortical visual mecha-
nisms of the vestibular system and proprioception to the receptors that pro-
duce them. For example, if I am able to catch a tennis ball in midair practically
without having to think about it, it is thanks to subcortical mechanisms of
movement detection and visual prediction of the trajectory, which have noth-
ing conscious about them.
This ignorance is at the source of the surprise provoked by the discov-
ery of what was called blind vision. It took several years of intellectual bat-
50 •
tle before Weiskrantz could get people to accept the existence of the so-called
residual vision that persists after destruction of the visual cerebral cortex.42
One unhappy consequence of this resistance is the extremely conservative at-
titude of educational institutions toward visually impaired children. I describe
below an extraordinary experiment in teaching speed sports to visually im-
paired children taken out of school and for whom sports are normally pro-
hibited.
Visually Impaired Children Playing Basketball

One day in the 1980s, I was visited by Messrs. Chaumiène, a professor of phys-
ical education from a high school for visually impaired children in Montgeron,
and Leguern, a researcher at the French national institute of sports. They
asked me to take a few minutes to look at a videocassette of a basketball game
between children, and then asked me the following question: “One of the
teams is made up of children who see normally. The other is made up of am-
blyopic children who demonstrate less than one-tenth of normal vision in clas-
sical ophthalmological tests, read in Braille, and do not attend normal educa-
tional institutions. For safety reasons, administrative regulations forbid these
children to participate in sports. Which is the handicapped team?”
It was impossible for me to tell! Using an enriched environment, these two
instructors had managed to teach visually impaired children to play ball, to
fence, and so on. The faster the movement, the more the children seemed to
succeed. We worked together for several years to try to understand this re-
markable achievement. For me, it was an opportunity to discover many as-
pects of this disability. For instance, the ophthalmological files for most of
these children were incomplete. We had to organize bus trips to the hospital
in Créteil to get tests done. Then, thanks to the sympathetic understanding of
the school administrators, whose help was indispensable, we began studies of
how these children performed visuomotor tasks that we hoped would also in-
terest other researchers in the field.
This fascinating experience convinced me that ophthalmological tests ex-
amine only a tiny fraction of visual function and, in any case, totally ignore
perception of movement. This is still true today. Although millions of people
wear progressive lenses that modify the apparent velocity of visual motion in
a nonlinear manner on the surface of the lens, no test worthy of the name ex-
amines dynamic vision, which is the most important for coordinating gestures
and perceiving three-dimensional shapes, as research on the contribution of
movement to perception of curvature shows. We are working on this ques-
tion in collaboration with glass manufacturers.
A SIXTH SENSE? •
51
Vection: Perceiving Self-Motion
Perception of self-motion is among the major functions of vision. Remark-
ably, this function of vision was completely ignored for close to fifty years de-
spite the pioneering work of eminent scientists like the physicist Mach.43
Moreover, anatomists labeled it an accessory system, and it was not until 1975
that someone worked out its properties. Areas that process information about
optic flow and that appeared later in evolution (like MT and MST), have only
recently been studied in the monkey, following the pioneering work of Wurtz.
We have all felt the compelling illusion of moving forward when, seated
on a stationary train, we observe the train on the next track start to pull away;
or when, looking from a bridge at the river below, we get the feeling that the
bridge is moving forward. Mach called this illusion of bodily self-motion, in-
duced by a visual displacement, vection. At the beginning of this century, he
constructed several machines intended to elicit it.
For example, he rotated rugs on a revolving circular platform to study the
illusion of rotation of the body induced in subjects exposed to this stimula-
tion, which today is called optokinetic. He deduced from his findings that in
fact what was involved was a basic mechanism of detection of bodily self-mo-
tion. But the problem languished until 1970, when a psychologist, Lee, at-
tacked it with renewed interest and opened a period fertile with discovery.
Lee’s original experiment consisted in hanging from the ceiling of an audito-
rium of his university a wooden box about 3 meters on a side.44 One side of
the box and the bottom were left open. Lee caused this hanging chamber to
oscillate slowly. A subject standing still within the box had the undeniable illu-
sion that the box was stationary and that the University of Edinburgh was
rocking! Again, the situation was one of vection such as Mach had described.
Lee did not stop there. A student of Gibson, he attributed this effect to
what he called the “proprioceptive function” of vision. He described it qualita-
tively and showed, in particular, that during the illusion, the subject’s body
began to oscillate and thus that the perception was accompanied by active
postural readjustments. He also showed that two-year-old children could be
thrown off balance by the effect and fall flat on their back.
Between 1970 and 1975, three groups generated the first quantitative de-
scriptions of vection and the motor effects associated with it. First, in Jung’s
laboratory in Fribourg in Switzerland, Dichgans and Brandt studied circular
vection in a horizontal plane (obtained by rotating a cylinder around the sub-
ject);45 then Young’s group in Cambridge, Massachusetts, in collaboration with
Dichgans, studied vection induced by the rotation of scenes in the frontal
52 •
plane, and linear vection induced by flight simulators; finally, in my laboratory
in Paris, we investigated the properties of linear vection that we manipulated
at the same time as vestibular perception by placing subjects on a trolley,
shown in Figure 2.5.46
It turned out that the intensity of vection is proportional to several vari-
ables of movement in the visual world. The first is surface: the greater the sur-
face of the visual field in motion, the stronger the intensity of vection, though
a very small surface field is also likely to induce vection, as will sometimes
even a simple point of light.
This idea is in line with my suggestion that perception is a function not so
much of the intensity of a stimulus as of the agreement between the stimulus
and an assumption the brain makes. This can be tested on an airplane before
takeoff. Shortly before the plane starts to move, a few minutes go by during
which everyone is anticipating a slight acceleration. It only takes a luggage
trolley or other vehicle passing by in front of a window, setting in motion a
tiny part of the visual field, for the movement of the vehicle to be interpreted
by the brain as that of the airplane. Anticipation is crucial to the perception of
self-motion. A similar kind of anticipation, which involves not vision but the
otolithic system, can happen in an elevator when, thinking you have reached
the floor you want, you perceive an illusory deceleration.
Another important variable for determining the intensity of vection is the
velocity of scenic motion. Below a certain threshold, vection is imperceptible;
above that threshold, the intensity of sensation increases with velocity up to a
maximum beyond which vection disappears abruptly if velocity increases. In
front of a scene that moves by very quickly, we have the impression that we
are motionless. This perceptual inversion is absolutely spectacular on the high-
way. It is very familiar to drivers of fast cars and auto-racing champions.
Above 200 kilometers per hour, instead of feeling that they are gaining on the
cars in front of them, they suddenly have the extraordinary conviction that the
cars in front are approaching them. I don’t recommend that you try this exper-
iment! These drivers have lost vection, that is, the sensation of self-motion,
and have the illusion that they are motionless before a world that is hurtling
toward them.
This perceptual inversion is very dramatic. But why does it happen when
the visual stimulus is so marked? Perhaps the velocity of the visual scene is too
great compared with the range of velocities of bodily self-motion. The brain
can no longer interpret the retinal slip as the result of a natural displacement.
The question is one of a genuine perceptual decision, not disappearance of
perception.
A SIXTH SENSE? •
53
s
i m
lw
Z
X Vt
Y
Figure 2.5. A trolley for studying interactions between vestibular and visual de-
tection of translational movements. The subject is seated on a stool placed on
the trolley (t), which is driven forward or backward by translational movements
that are controlled by a computer. A projector (represented by a light bulb) dis-
plays a reel of transparent film (r) with various shapes (squares, points, and so
on). Theses images are projected by a set of mirrors (m) onto a screen (s) lo-
cated above the subject. The mirrors create virtual images of the screen on
each side of the subject, who sees them through lateral windows (lw) in such a
way that he has the illusion of moving in an optical tunnel. This visual move-
ment makes him feel that he is advancing or moving backward in a direction
contrary to the movement of the visual scene (called vection). The vestibular
system can be stimulated by movement of the trolley, so that the interaction
between visual and vestibular information in the perception of movement may
be studied.
54 •
The spatial frequency of the scene, that is, the number of elements per
unit of surface, also has an influence on the intensity of vection, which is
greater when you go through a village that has many visual elements per unit
of surface than when you cross the desert. Road accidents are probably caused
by disparities between the information supplied from both sides of the road;
night, especially, gives a driver erroneous impressions of changing directions
due to the difference in vection induced by each retina.
The final element is the distance of the scene in motion and, particularly,
its position as backdrop or forestage. The most distant part of the environ-
ment is what determines the intensity and direction of vection. It is yet further
proof of the extraordinary assessment of the perceptual context carried out
by the brain.
If a subject is exposed to visual movement for a long time, vection dimin-
ishes in intensity. In the example of the highway, driving for several hours in-
duces underestimation of vection and, thus, underestimation of the sensation
of velocity. When drivers leave the highway, they have a tendency to drive too
fast, which is why speed ramps exist. Adaptation to stimulus is a familiar phe-
nomenon in psychophysics. Disappearance of vection can be interpreted as
the suppression of a repetitive signal that has nothing more to contribute to
the ongoing action. If that were true, the question would appear to be one of
an interesting mechanism that would be fairly easy to test. But it seems to me
that, on the contrary, one must assume that an active neural mechanism
opposes the stimulus, the brain creating an internal movement that blocks
the stimulus supplied by the external movement. When the stimulus is sup-
pressed, all that is left is this active neural construction, which was counterbal-
ancing it. At issue here is not a simple mental object, but a mental movement.
An illusion that probably involves similar mechanisms is called “the waterfall.”
If you look at a waterfall for a long time and then cast your eye on the sur-
rounding landscape, you perceive apparent movement resulting from the ad-
aptation of your brain to the visual scene in motion. I will return to the mean-
ing of these illusions in Chapter 13.
Another especially clear example of this type of mechanism is provided by
what is called consecutive optokinetic nystagmus. This odd-sounding expres-
sion describes a very simple phenomenon. Suppose that you are on a bus,
looking out the window. Your eye follows the visual scene, then turns back.
That is optokinetic nystagmus. Now suppose that the journey lasts several
minutes and that you are suddenly plunged into darkness. Instead of stopping,
your eye will continue to move. This consecutive optokinetic nystagmus is the
result of a still mysterious mechanism: it is as if a neural memory of move-
A SIXTH SENSE? •
55
ment had “charged” (as you charge a capacitor in electronics) a signal con-
nected to the velocity of the eye, which runs down in the dark. It is likely that
this perceptual memory is of the same nature as that involved in a long sea
voyage. Indeed, sailors know that several days after their return, they will still
feel as though they are moving, even though they are on terra firma. Here,
again, the brain has constructed a dynamic internal response to the motion of
the sea, probably using visual and proprioceptive vestibular information. Dur-
ing the voyage, this internal neural activity—which predicts the disruptive ef-
fects of the oscillations caused by the waves—opposes the sensory stimuli in-
duced by pitching and rolling. Hence sailors are able to live at sea without
feeling that they are moving and are able to coordinate their own movements.
The decision to interpret a movement in the visual world as a movement
of the body is thus the result of multiple central neural processes, and yet it
seems so obvious. How does the brain succeed in giving perceptual unity to
all these variables and all their values? It is thanks to special mechanisms for
building coherence, like those that make it possible to perceive a single visual
object after it is analyzed in the first relays of visual pathways. I will return to
these mechanisms later on.
56 •
3
BUILDING COHERENCE
Sensory receptor function has a predictive quality. Receptors can detect the de-
rivatives (that is, velocity, acceleration, changes in force and pressure) of the
physical variables that stimulate them. Detecting changes in a variable allows
the receptors to predict the value of that variable at a future time. In addition,
modulation of receptor sensitivity preselects receptors that specify future
movement by endowing them with properties that anticipate the nature of
the movement, tonic or dynamic. Such modulation allows simulation of the
movement without having to carry it out. In fact, simple activation of γ fibers
simulates a displacement of the arm by triggering spindle activity, and this
spindle activity produces the perceptual illusion of movement, as vibration ex-
periments demonstrate. Finally, filtering of messages by mechanisms such as
presynaptic inhibition increases the efficiency of preselection even more.
In this chapter, I will show how the combination of signals from the differ-
ent senses—the essentially multisensory character of perception—and the
brain’s use of endogenous (that is, produced by the subject) signals further in-
creases the predictive power of the brain.1 Indeed, a true physiology of per-
ception must abandon the idea of splitting up sensory functions and instead
approach them by way of their multisensory character. Aristotle himself won-
dered about this question:
We perceive things as a whole, rather than what some may refer to as a

continuity of their parts. Yet we can say that things do not always ap-
pear to us as they are; and that is why the size of the sun as we see it is
not its true size. But let us return to our earlier question, whether we
can perceive several things simultaneously, that is, in a single part of the
soul, in an indivisible moment. It has been proved that the soul per-
ceives all sensations with one and the same faculty, which collects the
information from all the senses. Yet this faculty, though numerically
one, differs in its accounts: it is the same soul, but differently disposed.2
The significance of this message was neglected until very recently. For fifty
years, the entire field of the physiology of perception had isolated the senses.
Consequently, laboratories studying vision proliferated, even though as early
as 1876 Ferrier was writing: “Without these labyrinthine impressions, optic
and tactile impressions are of themselves unable to excite the harmonious ac-
tivity of the centres of equilibration.”3 The same is true for hearing, since vi-
sion can modify auditory perception in a decisive way.4 Ventriloquists provide
the best example of this phenomenon: they create the illusion that their words
are coming from their dummy. The mere fact of not moving their lips and
working those of the dummy displaces the perception of the source of the
sound. I will describe the neurophysiological discoveries that may explain this
effect. Similarly, the art of entertaining demonstrates that the flavor of a dish
depends no less on the way it looks than how it smells. The same agreeable
smell at the beginning of a meal may become disagreeable by the end, owing
to a property of the brain called allesthesia—taste is a source of pleasure when
a person is hungry and of repulsion when the person is sated.
The importance of combining sensations does not appear to have inter-
ested physiologists before 1960, although a hundred years ago Sechenov tied it
to how the image of one’s own body is constructed during active exploration:
“The child often sees a toy in the hands of its mother and in its own hands.
The first of these sensations is simple; the second one is complex, including
also tactile and muscular elements. This happens many thousand times. The
two sensations become separated from each other, and the child’s own hand
appears in its consciousness with a tinge of self-consciousness.”5
In work with animals, pioneers such as Held and Hein showed the impor-
tance of activity in the development of visual functions and the coordination
of sensory systems.6 In the 1960s, Gibson suggested the basis for a new experi-
mental test of perception. “We shall have to conceive the external senses in a
new way, as active rather than passive, as systems rather than channels, and as
interrelated rather than mutually exclusive. If they function to pick up infor-
mation, not simply to arouse sensations, this function should be denoted by a
different term. They will here be called perceptual systems.”7 One compelling
explanation for the multisensory nature of perception is that every animal is
58 •
only interested in certain aspects of reality. It selects configurations of features
that are useful to it as well as the relationships between the properties of ob-
jects. For example, as Turvey says, the height of the steps in a staircase ex-
pressed in centimeters is less important than the relationship between the
height of the steps and the height to which we are able to lift our feet.8 This
explains why the steps in the staircase of a French château are not very high,
since horses had to climb them. Similarly, the height of the stairs at the opera
has to be adjusted to accommodate the step of women in evening gowns.
Gibson called these relationships that define the constants useful to a par-
ticular species “affordances.”9 They are a measure of the feasibility of carrying
out actions important for the species in question, taking into account its
sensorimotor repertoire. Sometimes opposite properties are interesting to dif-
ferent animals. Thus the degree of firmness (which can be measured as the re-
lationship between pressure and displacement) of the ground allows humans
to walk, whereas its friability permits the earthworm to move about. A single
object such as a slipper can serve as a warm refuge for a person’s foot, oral
stimulation for a puppy, and a scrumptious meal for a larva. The geometrical
constants essential for humans have ceased to be essential for other species.
What are the neural mechanisms underlying the multisensorial and ac-
tion-related character of perception? When microelectrodes were invented in
the 1950s, it became possible to stimulate different sensory receptors with
electrical impulses and to record neuronal activity in anesthetized animals. It
turned out that multisensory convergences occur at all levels of the nervous
system.
They occur first of all in the medulla, where the most important inter-
neurons—that is, neurons that link sensory and motor neurons—(like inter-
neuron Ia) are the site of convergences from muscular spindle receptors, tac-
tile receptors and the Golgi tendon organs. In the cerebellum, also the center
of intense multisensory interactions, each Purkinje cell in the cerebellar ver-
mis receives visual and proprioceptive information, just as the neurons of the
flocculus receive vestibular information, and so on. The superior colliculus is
an important structure for orienting movements. It receives visual and pro-
prioceptive information and is another focus of convergence between audi-
tory and visual signals. Convergences also occur in the so-called sensory
thalamus and in many cortical areas. Multisensory convergence is thus the
rule. What is not known is what fundamental role it plays. It was not until
the 1970s that recordings of neurons in alert animals shed some light on this
question. The last five years, however, have yielded an exceptional harvest of
BUILDING COHERENCE •
59
data on this subject. An example from our own work will show how vision
and the vestibular system work together to reconstruct movements of the
head in space.
How Vision Detects Movement
An engineer with two sensors at his disposal for detecting movements of the
head—the retina and the semicircular canals—runs into a problem of geome-
try. Indeed, the semicircular canals detect movement of the head in their three
perpendicular planes, so information is encoded in three dimensions; that is,
in a Euclidean frame of reference. The retina encodes retinal slip on a spheri-
cal, or two-dimensional, surface. It receives information of great complexity.
To better understand the geometrical problem posed by the fusion of infor-
mation about movements of visual and vestibular origin requires a brief dis-
cussion of the concept of optic flow.
When people move about in the real world, the image of the environment
on their retinas is displaced and becomes distorted in a very complex way. This
distortion of the image on the retina during displacement is called optic flow.
It can be represented by a velocity vector at any point on the retina that repre-
sents the direction and velocity from the point of the visual environment that
is projected onto the retina.
Gibson, who worked for the American army, looked into the problem of
airplane landings and described the form of these fields of vectors, which de-
pend both on the geometry of the environment and the movement of the air-
plane. He observed that optic flow is very straightforward during a rotation
because all the points are animated by an identical angular velocity. In con-
trast, during a forward translation—for example, moving forward while walk-
ing, or landing an airplane—optic flow takes the form of a complex expansion
of the field of vectors, which appear to start at a vanishing point at the edge
of this perspective: the “focus of expansion.” During lateral translations, as
when we are looking out the window of a train, optic flow takes on a more
subtle complexity: all along the track the visual world moves in a sense con-
trary to the movement of the train, whereas the distant landscape appears to
move in the same direction as the train.
How does the brain achieve this fusion of three-dimensional messages
supplied by the canals and by optic flow? The last twenty years have provided
new data on the neural mechanisms of the contribution of vision to the per-
ception of self-motion (vection), how moving objects are sensed, and the
“blindsight” of movement. I will summarize these findings here.
60 •
Movement
stereo Color
Retinal ganglion cells
MT ? V4
Color
V2
Stereo Form
Lateral geniculate nucleus
LGN Parvocellular layer Form

Color-selective
Slow Color
Stereo Form
Low contrast sensitivity
High resolution
Form
Color
Stereo Form
1
Form
Magnocellular layer Color Color 2 and 3
Color-blind
Form
Fast
High contrast Movement Movement 4A
Low resolution
α
4C
β
5
6
Primary visual cortex
V1
Figure 3.1. Segregation of inputs for identifying shape, color, motion, and depth
in the first relays of the visual system.
Movement is detected mainly in the retinal periphery. Three kinds of gan-

glionic cells can be distinguished in the retina. In primates, two of these are
sensitive to movement: first, the cells that activate both movement and strong
contrasts between light and darkness and flickering; second, neurons sensitive
to slow movements. Segregation of information about shape, color, depth,
and movement, which is carried out on the retina, is maintained in the first
central relay in the thalamus, the lateral geniculate nucleus. This nucleus is di-
vided into two parts, the magnocellular zone and the parvocellular zone (Fig-
ure 3.1). From there the information is directed either toward the superior
colliculus (a very old structure, called the optic tectum in birds) or toward the
primary areas of the visual cortex—V1, V2, and V3 (area V4 mostly receives
information about color). Segregation of featural elements of color, shape,
and depth still occurs at this level.
Two areas of the cerebral cortex are involved more particularly in process-
ing information about visual movement. They are called MT (middle tempo-
ral area) and MST (medial superior temporal area).10 The MT receives im-
pulses from the visual cortex, and its neurons are directly influenced by
movement on the retina. The activity of the MT is transmitted to the MST,
61
where the visual data are combined with signals of movement from the eyes
and vestibular signals; in other words, extraretinal signals. The neurons in this
area also solve an important problem for the visual perception of movement:
they are able to “decide” whether an object is moving in the environment or
whether it is stationary. Shadlen and Newsome recently investigated this “per-
ceptual decision.”11 Moreover, positional signals of the eyes and head compen-
sate for alterations in optic flow that occur when you turn your head while
walking; they also keep the perception of the heading direction constant.
Modern brain-imaging techniques bring daily findings of the existence of
other zones of the cortex that are involved in processing visual data on move-
ment. For instance, it has just been discovered that the sensitivity of the neu-
rons of the visual cortex is modified by the behavioral context, that is, the ac-
tion in which the subject is involved.12
Ocular Pursuit Predicts the Movement of Targets

Say you are on a bus and you are looking out the window. Your eye follows
the landscape. The movement of your eye is the result of cooperation be-
tween two mechanisms. The first is ocular pursuit. This continuous tracking
of a moving target is possible only for animals that possess a fovea, the mon-
key and man; it does not exist in fish, rabbits, or even cats, whose eyes can
only skip from one visual space to another in motions called saccades. Ocular
pursuit is not very rapid. Move your finger in front of you with a back and
forth motion. You will observe that when your movement goes faster than
once per second (1 Hz, or about 60 degrees per second), you can no longer fol-
low your finger. But the brain of primates and man has compensated for this
slowness by endowing pursuit with great predictive capability: after a few back
and forths, the movement of your eye anticipates the movement of your
finger. Similarly, if you suddenly stop moving your finger, your eye will con-
tinue to follow your absent finger for an instant, which suggests an anticipa-
tory neural activity. The pursuit also continues when the light is briefly turned
off or when the target disappears behind an obstacle. This ability to follow a
target behind an obstacle appears in infants during the first months of life.
Ocular pursuit is produced along a pathway (Figure 3.2) that transmits the
movements of the target to the MT and MST areas, then to the frontal cortex
in an area called the frontal eye field (FEF). In the MT, neurons are assigned a
particular direction of the movement of the target. Each neuron fires when
the target moves in a given direction. In the MST the convergence of sensory
signals and internal messages signaling movements of the eyes or head (extra-
retinal) endows the neurons of this area with a special property: they can fire
62 •
V3a
Parietal
LIP cortex
V1
7a
Occipital cortex V2 7b
STS
MT
Temporal MS
T
cortex
LG
N NO
T
FEF Frontal
cortex
DL
PN
DM
PN
Figure 3.2. The main centers involved in processing information about visual
movement. (1): The accessory optic system. In this system, visual movement is
broken down into three planes (one horizontal and two vertical) that are
aligned with those of the semicircular canals. Shown here is the nucleus of the
optic tract (NOT), which encodes horizontal movement of the visual world,
that is, shifting images on the retina that are in the plane of the horizontal
semicircular canal (when the eye is centered in its socket). (2): Pathways for vi-
sual pursuit of objects in motion. In the monkey and in humans, the image of
the object in motion is transmitted to the lateral geniculate body (LGB) by reti-
nal neurons sensitive to movement, and from there to the visual cortex (cortical
areas V1, V2, and V3). Messages about movement are next transmitted to the
parietal cortical areas (7a, 7b, LIP, and so on), the areas of the mediotemporal
cortex (MT and MTS), and the frontal oculomotor field (FOF). From the cere-
bral cortex, control and pursuit signals are sent to the dorsolateral and
dorsomedial pontine nuclei (DLPN and DMPN).
63
during pursuit of a target even if the target disappears for a brief moment.
Messages about eye movement, which are transmitted from the centers of the
brainstem and control eye muscle contractions, activate these neurons for a
brief instant after the light disappears and thus ensure that a person’s gaze will
be in the right spot when the target reappears. This convergence of external
and internal signals also enables us to make tracking movements toward tar-
gets that are actually motionless but that we perceive to be moving, as when a
row of targets is illuminated one after another and the brain creates a percept
of movement.13
Tracking signals next activate the centers of the brainstem (the pontine
nuclei) and of the cerebellum, linking the vestibular nuclei and other pre-
motor nuclei that project to the motor neurons of the eye muscles.
Visual Movement and Vestibular Receptors
The second mechanism that enables us to follow the landscape from the bus is
a very old reflex action, the optokinetic reflex, present even in the fish and the
frog, who have no fovea. Actually, all sighted animals possess this mechanism,
which allows them to follow the movements of the visual environment. It also
contributes to maintenance of posture. If you tilt the aquarium in which it is
swimming, a fish will tilt its body to align it with whichever surface of the
aquarium is lit.
The nerve pathway that processes these signals thus also influences the
centers that control posture. Anatomists discovered it a long time ago and
called it the accessory optic pathway for want of understanding its function.
What is remarkable is that all along this pathway neurons respond to visual
movements in preferred directions aligned along the planes of the semicircular
canals of the vestibular system.14
This alignment is also characteristic of the neurons of the retinal gan-
glia, which fire preferentially when the visual movement is in the plane of the
semicircular canals. This geometrical segregation becomes more pronounced
in the second relay. The neurons sensitive to movement in the plane of the
horizontal canal are united in the so-called nucleus of the optic tract (NOT)
following a very precise orientation, the temporonasal direction. Another nu-
cleus, called the medial terminal nucleus (MTN), contains the neurons sen-
sitive to downward visual movement, and yet another, the dorsal terminal nu-
cleus (DTN), contains those sensitive to upward visual movement. Other
groups of cells belonging to the same accessory optic system, situated in a re-
64 •
lay portion of the ventral tegmental area, encode combinations of rotational
and linear components of optic flow.
Consequently, just as information is broken down into its main compo-
nents (color, movement, contrast, texture, and so on) in the primary cortical
visual projections (V1–V4), the geometric properties of movement repre-
sented in optic flow are also segregated. This early segregation in the sequence
of neural processing is further refined in the next relay, the dorsal motor nu-
cleus of the inferior olive, a brainstem structure that projects to the cerebel-
lum, where some of the visuovestibular convergence takes place. This dorsal
motor nucleus contains three groups of neurons that correspond exactly to
the three planes of the canals. The next relay, the cerebellum, also contains
neurons, located in the flocculus, whose preferential planes of activation are in
those of the canals.
The existence of separate components organized according to the planes
of the semicircular canals clearly facilitates the convergence, that is, the fu-
sion, of these inputs, whose origin is visual, and those from the vestibular re-
ceptors, with which the visual messages converge.
The geometrical correspondence between the visual system and the semi-
circular canals extends to the motor system. Indeed, despite the migration of
the eyes from a lateral to a frontal position, the plane of action of the three
pairs of extraocular muscles, which enable rotation of the eyeball in all spe-
cies, has remained approximately parallel to the planes of the semicircular
canals.15
The demonstration of an organization in three planes was made using ten-
sor analysis and the theories of Pellionisz and Llinás. They proposed that the
muscles be represented by their eigenvectors, that is, their own virtual vectors
that represent the direction and amplitude of the force exerted by each mus-
cle. These vectors are classical mathematical quantities, but are very easy for
nonmathematicians to grasp. They have the advantage of being accessible to
experimentation: in animals or in humans, all that is needed is to establish
points of attachment of the muscles and to reconstruct the skeletal anatomy
of this motor apparatus, and then to record the activity of these muscles dur-
ing activation by tilting the head, which initiates the so-called vestibulo-nuchal
reflex. The eigenvectors of the neck muscles of a cat are situated in the plane
of the canals. The geometry of the semicircular canals is thus basic to the per-
ception and control of movement.
When a monkey jumps from branch to branch, when a diver leaps from
his diving board, or a gymnast grabs the high bar or jumps on a trampoline
65
H
0.1 m
Figure 3.3. A video camera connected to a computer recorded movements of an

athlete’s body during trampoline jumping. The computer traced the lines of
motion of markers placed on his head (H), trunk (T), legs, and feet. Jumping
was broken down into three main phases: First (left), the body goes up, the
head is stabilized in rotation, and the brain is able to use information from vi-
sion and the vestibular system. Second, the brain launches a very rapid rota-
tional movement (greater than 600 degrees per second) during which vision is
useless because images on the retina are moving too quickly. The brain relies
only on vestibular cues. Finally, the rotation ends, and the body lands vertically.
During this phase, the brain uses vision, the vestibular sensors, and
proprioception—as well as internal models of the effects of gravity—to plan a
soft landing.
(Figure 3.3), the brain has to reconstruct the movement of the head in space
very rapidly and with the most exquisite precision. How does it do it using vi-
sual and vestibular information, and proprioceptive information from the
neck?
From the first central synapse, the velocity of the head in space is, in fact,
detected by combining the information supplied by the vestibular accelerome-
ters and the visual tachometer. This discovery was made by recording the neu-
rons of the vestibular nuclei in the fish and, later, in the monkey.16 These nu-
clei are located in the brainstem and constitute the first central relay of
vestibular information; that is, the fibers of the semicircular canals project di-
66 •
rectly to them. They are called vestibular nuclei (see Figure 2.4) for purely ana-
tomical reasons.
The famous experiment of Dichgans and his co-workers is a good illustra-
tion of the way vision and the vestibular receptors cooperate in detecting
movement of the head. Dichgans and his group put their fish on a turntable
and subjected them to three separate experiments. First, a rotation in the dark,
broken down into three stages: acceleration, then constant velocity, finally de-
celeration. We can replicate this stimulus by making a few turns with our eyes
closed. The form of the stimulus is a trapezium of velocity: one or two inter-
vals of acceleration separated by constant velocity. The vestibular receptors
are activated only by the acceleration and do not fire until the animal begins
to turn or when it brakes. The neurons of the vestibular nuclei respond to
such stimulation by increasing their discharge followed by a slow diminution
after 10 to 20 seconds. Consequently, the brain receives no information about
the rotation of the head during the period of rotation at constant velocity. It is
the same for us if we twirl around for some time: when the effect of the tran-
sitory activation of the canals wears off, we can no longer perceive any move-
ment if all we have is the vestibular information.
In the second experiment, the animal was immobilized and the visual
world rotated around it, using a rotating cylinder. The researchers observed
that these neurons, called vestibular, are also activated by visual movement.
This activation is due to transmission along the pathways of the accessory op-
tic system. The activity increases very slowly at first, then holds steady during
the rotation, and finally diminishes slowly, the animal immobile the whole
time.
The third experiment involved rotating the animal in light. Combining vi-
sual and vestibular stimuli by imitating a natural rotation, the neurons of the
vestibular nuclei, like the velocity of rotation of the head, fire in a trapezoidal
pattern. In other words, the angular velocity of the head is perfectly repro-
duced from the first neural relays by means of the combination of dynamic
properties of the visual and vestibular receptors. This discovery was repeated
in several species (frog, rabbit, cat, monkey): it is thus very likely true for hu-
mans as well.
We are aware of the angular velocity of the head thanks to the sensitivity
to rotation of both the visual and vestibular systems. Vision alone is not suf-
ficient—it is too slow. The contribution of the vestibular receptors is to inform
the brain about the movement under way by virtue of their having detected
derivatives of displacement (velocity and acceleration). The term “dynamic
complementarity” is certainly apropos.
67
Moreover, the neurons of the vestibular nuclei do not only receive visual
and vestibular information: they are also sensitive to activation of the muscle
receptors in the neck. Very short synaptic pathways connect them to the
neuromuscular spindles (numerous in the muscles of the neck), which detect
the length and velocity of muscle stretch. When the head turns, the brain is
informed of the rotation by this combination of visual, vestibular, and pro-
prioceptive information. From the very beginning—from the first synapse of
the central nervous system—the detection of movement is multisensory.
The Direction of Gaze Modiﬁes the Vestibular

Perception of Movement
Among the various neurons contained in the vestibular nuclei, the neuron
called the second-order vestibular neuron is very interesting. Actually, it is a
sensory neuron, because it receives visual, vestibular, and proprioceptive infor-
mation. But it is also a motor command neuron, because it projects directly to
the ocular motor neurons, and in some cases also to the neck. Reflex theories
explain it as a simple relay of sensory information in a sensorimotor loop. We
recorded the activity of its axons during spontaneous ocular movements and
were very surprised to record activity connected to the discharge of the ocular
motor neurons.17 What we believed to be a sensory neuron turned out to be
activated by a motor discharge! A similar phenomenon is observable in the
first visual relays.18
Some time later, I invented a technique that makes it possible to record the
activity of a neuron under physiological conditions, to analyze its entire struc-
ture, and to determine the target structures of the axon.19 We were able to
penetrate the neuronal axon (whose diameter is around 5 microns) thanks to a
very fine glass microelectrode whose tip measured 1 micron in diameter. We
presented targets to the animal to induce spontaneous movements of its eyes
(saccades), and we rotated the animal in the dark to stimulate the semicircular
canals and induce a nystagmus. The power of this method derives from the in-
jection of a tracer, horseradish peroxidase, which makes it possible to recon-
struct the connectivity of the neuron.
We observed that these vestibular neurons, which normally respond to ro-
tation of the head, and thus to its angular velocity, by increasing their firing
rate, exhibit a basic activity that changes completely as a function of the direc-
tion where the animal is looking. A corollary signal, “a copy of the motor
command” (Figures 3.4 and 3.5), is thus present as early as the first sensory re-
lay.20
What is the role of this positional signal of the eye in vestibular neurons?
68 •
Motor
command
Motor
M
center
Efferent
Perception P
copy
R E
Receptor Effector
Figure 3.4. The principle of efferent copy according to von Holst and
Mittelstaedt. Sensory signals are sent to the neural centers where perception is
derived. A motor command travels in a loop from these centers to the effectors
(the muscles). But a copy of the motor command is also sent to the perceptual
centers, where it alters sensory information according to the ongoing action.
The brain is thus able to anticipate the results of movement without having to
wait for information from the sensors.
The question has not yet been answered. For some, it is a simple change of co-
ordinates; for others, it has to do with the coordination of the eyes and the
head. In all these cases, the data show that the senses are not isolated. Action
influences perception at its source.
Am I in my Bed or Hanging from the Ceiling?
I have described the nerve pathways that transmit the movement detected
by vision to the cerebral cortex. I have also analyzed the vestibular detection
of head movement and the mechanisms by which vision is enhanced by
the vestibulo-ocular reflex. But vestibular information does not remain at the
level of the first sensory relays. It is also transmitted to the cerebral cortex,
where it contributes to many functions: conscious perception of orientation,
gaze movements, control of posture, coordination of gestures. It also plays a
role in constructing a coherent perception of the relationships between the
body and space. This vestibular contribution to the highest cognitive functions
was neglected until recently. Yet, a few pioneers such as Penfield had consid-
ered it.
Penfield’s subjects reported that they felt vertigo or rotating sensations,
characteristic of the vestibular system, during electrical stimulation of a tem-
69
Motor nucleus
of the eye
Eye VI
movement
Vestibular
Vision Vestibular nuclei
neuron
Vestibular
receptors
Medulla
Figure 3.5. Second-order vestibular neurons are influenced by the direction of
gaze. A second-order vestibular neuron transmits movements of the head de-
tected by the semicircular canals to the muscles of the eyes. One can see the
axonal branching to various centers of the brain that process this information.
This neuron is the first sensory relay for vestibular information, but it receives
signals from vision and receptors in the muscles and joints as well. It also re-
ceives copies of the motor commands for eye movements.
poroparietal region.21 It also has been known for a long time that illusions of
dissociation of the body are induced during the aura—that is, alteration of
consciousness—that often precedes seizures in patients suffering from tempo-
ral lobe epilepsy.22 During an epileptic fit, these patients sometimes have the
feeling that they are simultaneously lying in their bed and floating at the level
of the ceiling, as if their body were split. The area of the brain responsible for
these illusions is located deep within the lateral sulcus, around the superior
temporal gyrus, at the caudal portion of the insula and along the retroinsular
region, near the auditory area.
Messages of vestibular origin contribute to the most cognitive aspects of
the body schema and of spatial representation.23 In humans, the existence of
this vestibular cortical area was recently confirmed by brain-imaging meth-
ods.24 Patients with lesions of the parieto-insular zone present with devia-
tions of the vertical subjective: the world appears to them to be leaning to
70 •
the side opposite the lesion. The effect is so strong that they take skewed pho-
tographs.25
The Neural Basis of Vestibular Inﬂuences on

Cortical Function
The parieto-insular vestibular cortex (PIVC) was first identified in the mon-
key.26 It is the principal center for relaying vestibular information to the rest of
the cerebral cortex, and is located in the temporal portion of the lateral
sulcus. Like the vestibular nuclei, this vestibular cortical area is multisensory.
There is no known direct route for visual inputs to the PIVC, but there are sev-
eral candidate pathways. Its role in the multisensory representation of move-
ment of the head in space was confirmed by recordings of the activity of neu-
rons in the cortex of a monkey during vestibular stimulation in the dark.27
Two-thirds of the neurons of this area respond to vestibular stimulation. The
other third respond mainly to somatosensory stimulation of the neck and
shoulders. Almost all of the neurons that respond to vestibular stimulation are
also activated by movements of the visual environment and by somatosensory
stimulation.
The characteristic feature of PIVC neurons is their sensitivity to angular
rotations in the various planes and to visual movements in the direction oppo-
site to that of the head. In other words, they respond in a way that is perfectly
adapted to detecting movements of the head in space, since when we make a
rotation while the head is turning to the right, the visual world is turning to
the left. These experimental findings suggest that the PIVC neurons encode
movement of the head in space based on vestibular and somatosensory visual
information, the latter mostly likely transmitted by the receptors in the neck
and the torso.
One might speculate that one of the processes accomplished in this part
of the cortex is encoding of movement of the head in planes more numerous
than those of the vestibular nuclei. Optokinetic and vestibular coding of the
movement of the head (through the accessory optic system) is organized ac-
cording to the geometry of the planes of the semicircular canals. This geo-
metric selection is probably very useful in maintaining a connection between
these two signals and combining them at the level of the brainstem. However,
at the level of the cerebral cortex, the representation of movement of the
head is in all likelihood combined with signals from temporal visual areas like
the MT and MST, which mainly process information about visual movement.
Strict separation in three planes is thus insufficient. What is required is a multi-
dimensional representation of the movement of the head.
71
According to Grüsser, the cortical areas involved in processing form an in-
ner vestibular circuit that processes information about movements of the head
in space and maintains continuously activated neural representations useful
for simulating movements internally and predicting their consequences.28
Neurophysiology has confirmed the functional importance of vestibular
influences on the cortical areas.29 I will describe next two examples of the in-
fluence of these vestibular messages on vision.
Why Does the Obelisk Look Vertical Even

Though my Head Is Tilted?
The neurons of the primary visual areas are activated by the border between
light and dark, the orientation of the border, and the direction of its move-
ment. They can be activated, for example, by showing an animal bars of light.
At birth, most cats and monkeys respond to orientations of these bars of light
in all directions, whereas only a few show a preference for horizontal and ver-
tical orientation. Consequently, selection based on geometry occurs here also
from the first sensory relays.
This directional selectivity is determined very early during a critical period
that in the cat extends from 3 to 10 weeks.30 The active movement of the eyes
detected by the receptors of the ocular muscles influences the neurons of the
visual cortex and leads to a preference in these cells for given lines of orienta-
tion. In this way, projection of the proprioceptors of the eye (and perhaps of
the neck) on the neurons of the visual cortex contributes to their direction-
ality. It suggests as well that the muscles of the eye constitute a frame of refer-
ence with the same capacity as the semicircular canals. Because the planes of
action of the eye muscles are close to those of the canals, it is difficult to favor
one or the other of these two hypotheses. Perhaps embryology will be able to
answer this question.
Proprioception is not the only factor that influences the directionality of
the neurons of the visual cortex. Indeed, a recent discovery shows a vestibular
influence on supposedly purely visual areas, like area V2: 40 percent of the
neurons of this area (which is one of the primary visual areas) are sensitive to
the contours of visual shapes, but their activity is influenced by the tilt of the
head.31 The preferential direction of the contour to which each neuron is sen-
sitive goes in a direction opposite to that of the head with respect to the verti-
cal. This vestibular influence makes it possible to keep the perceived orienta-
tion of the contour constant. When I tilt my head and the Obelisk at the Place
de la Concorde still looks vertical to me, it is perhaps thanks to this mecha-
nism. It is not a reflex that makes my eye turn (moreover, the torsional
72 •
vestibulo-ocular reflex is almost inactive), but a mechanism of mental rotation
of visual representations.
The Vestibular System and Face Recognition

Everyone knows that it is extremely difficult to recognize a face tilted at 90 de-
grees or, especially, upside down, even though the image of the face on the
retina remains quite clear. The reader also knows that it is absolutely impossi-
ble to look at a slightly crooked picture. So gravity may exercise a constraint
on the perception of visual forms.
The role of gravity in the perception of visual forms was studied in our
laboratory in the particular case of the perception of symmetries aboard space
flights that put humans in a situation of microgravity. On earth, when a sub-
ject is asked to say whether a shape is symmetrical or not, he does it more
rapidly if the axis of symmetry is horizontal or vertical. When the axis is diag-
onal, the subject takes a little more time. In particular, the symmetry is espe-
cially easy to identify when its axis is vertical. But when the astronauts of the
space station MIR were subjected in space to a recognition test of symmetrical
shapes (polygons, for example), they no longer showed any preference for the
vertical. This same lack of preference for the vertical obtains when you are ly-
ing on your back and the gravitational vertical is no longer aligned with that
of the axis of symmetry of the visual object.
Otolithic detection of gravity thus plays an important role in our visual
perception of the environment.
A Neural Compass?
Sit in a revolving chair. Look at an object in front of you, close your eyes, then
make one or two turns. Even with your eyes closed, you will find you can still
point to the object. A neural mechanism of vestibulary origin assesses the di-
rection of the head in the horizontal plane, memorizes it, and updates it dur-
ing the movement. One of the most important discoveries of recent years was
that of the existence of neurons whose discharge is connected to the orienta-
tion of the head in the horizontal plane. These neurons have been identified in
the thalamus of the rat.32 The thalamus transmits sensory information to the
cortex and the postsubiculum, the nucleus of the hippocampal formation,
which, as I will explain later in detail, plays a role in spatial memory.
These neurons signal the direction of the head no matter where an animal
is in a room, a property that is called local invariance. Thus they constitute an
internal compass. It is a sophisticated compass, because it indicates a wide
range of directions. For example, one of these neurons, silent when the rat
73
moves its head in many directions, suddenly starts to fire when the animal
turns toward the north, no matter where the animal is in the room. These
neurons are, however, very dependent on the local visual environment. The
animal has to be familiar with the room it is in for the directions preferred by
each neuron to be well stabilized. When the animal’s location in the room is
changed, its directional neurons are reorganized. But what is really remarkable
is that when the animal is familiar with a location, the neurons do not lose
their sensitivity to a given direction in the dark. It follows that these neurons
participate in spatial memory.
Vestibular information about head orientation is transmitted from the me-
dial vestibular nucleus to the dorsolateral thalamus and from there to the
hippocampal formation. It is also known that the hippocampus contains an-
other category of neurons, the place cells. These cells, independent of the di-
rection of the head, discharge when the animal is in a given location. It is thus
possible that memory of the object, in the experiment above, is due to vestibu-
lar information, which can update representations of space in the hippocam-
pus. I will come back to this topic in Chapter 5.
Neglect
Constructing a coherent perception of the relationships between the body and
space thus depends on both hierarchical and parallel mechanisms that com-
bine sensory cues and signals connected to action. So it is not surprising that
lesions or deficits in certain circuits induce a breakup of this coherence. One
example, which I will mention very briefly, is spatial neglect.
Patients who have a lesion of the right parietal cortex often present with
impaired space perception. Although they are looking at space in its entirety,
they can only perceive the right side, and they neglect the left side. Hence the
name of this illness: hemispatial neglect. If you ask these patients to draw a
flower, they will only draw the petals on the right side; if you ask them to
draw a clock, they will only draw the hours from 12 to 6; if you ask them to
paint objects, they will only paint one half of each object; finally, they will
only eat what is on one side of a dish of strawberries, and so on. These pa-
tients do not have all these symptoms at once. In fact, it is possible to distin-
guish several kinds of neglect: for example, personal, a person who neglects
his left arm; extrapersonal, a person who neglects to draw objects on the left
side of the room; or representational, a person who cannot picture the left
side of his living room, of his apartment, or of his city. It is also possible to dis-
tinguish perceptual from motor neglect. Motor neglect consists in not using a
limb even though its motor function is intact. Several authors have made the
74 •
remarkable observation that the simple injection of cold water into the left
ear—that is, introduction of a vestibular caloric stimulus—causes some of the
symptoms of neglect to disappear temporarily.33 This effect is not due to the
deviation of the eye induced by the stimulus.
Patients with spatial neglect sometimes have associated neurological defi-
cits, such as hemianopia (loss of vision for one half of the visual field) and
hemianesthesia (anesthesia of one half of the body).34 These deficits, appar-
ently primary—affecting primary structures in the sensory areas of the cor-
tex—also sometimes disappear after injection of cold water; in other words,
following a vestibular stimulus.35 Almost complete—albeit temporary—remis-
sion of hemianesthesia following damage to the right cerebral cortex is also
obtained as well as improvement of other impairments of the perception of
the body.36 For example, certain patients no longer feel it when someone
touches their hand. Bottini and co-workers found that vestibular stimulation
restores perception of touch in a transitory manner.37 Brain activity in such a
patient was recorded with the aid of positron emission tomography. In normal
subjects, tactile and vestibular signals share projections to the neurons of the
putamen, insula, somatosensory areas of the cortex, premotor cortex, and the
supramarginal gyrus. Some of these areas that were intact in this patient were
activated by the vestibular stimulus at the same time that sensation reap-
peared. Thus, “the somatosensory primary cortex has no priority for tactile
perceptual consciousness; rather, multiple sensory representations of the body
exist in the brain and any of these can contribute to perception.”38
Vestibular stimulation has a spectacular effect on self-awareness of body
parts. In particular, it offers relief from somatoparaphrenia. Patients with this
illness deny possession of their own limbs.39 For example, one patient denied
that her left arm was her own.40 She said: “This is my mother’s arm.” We in-
jected water into her ear (caloric stimulation of the vestibular receptors), and
this woman reclaimed possession of her arm. Unfortunately, the effect was
only temporary.
These observations suggest that spatial neglect has a single cause con-
nected to the representation of the body schema, even though it has multiple
manifestations. The efficacy of the vestibular stimulation could be due to a ba-
sic imbalance induced by the lesion and compensated for by an imbalance in
the opposite direction produced by the stimulus through the ascending vestib-
ular pathways described above, as well as through those that correspond to
the “directional neurons of the head.” If this assumption is correct, the fact
that the vibrations of the neck, the transcutaneous stimulation of the tactile
receptors, optokinetic stimulus, and wearing prism goggles also produce a
75
remission of spatial neglect is not surprising, given the multimodal conver-
gences toward all the centers of the vestibular pathways.41 I emphasized above
that several structures formerly called vestibular are in fact multimodal and
are involved in reconstructing orientation and movement of the head in space.
These considerations are important because they provide a biological basis
for so-called out-of-body parapsychic phenomena. Quacks who may profit
from such bizarre perceptions to exploit the credulity and pocketbooks of
people by selling these illusions as supernatural phenomena should be de-
nounced.
Distance Perception
The perception of distance is another example of the multisensory nature of
perception. But, here again, the combining messages originate in part from
copies of the motor commands. In fact, the perception of distance is the result
of visual messages and signals from the convergence of the two eyes. Action is
used as information about the world.
Lorenz suggested that fish evaluate the distance of objects from the bot-
tom of the sea using a combination of visual information about disparity (that
is, the difference between the two images of the object supplied by the two
eyes) and to convergence of the two eyes on the object.42 The idea that the
perception of distance is due to a multisensory interaction was also formu-
lated by Poincaré: “However, sight enables us to appreciate distance, and
therefore to perceive a third dimension. But everyone knows that this percep-
tion of the third dimension reduces to a sense of the effort of accommodation
which must be made, and to a sense of the convergence of the two eyes, that
must take place in order to perceive an object distinctly.”43
Perception of distance is actually due to a combination of visual and mo-
tor inputs.44 In the primary visual area V1, certain neurons are activated by the
disparate images of the same object on the two eyes, with some connection to
the distance of the object. By presenting the animal with two patterned planar
images called “Julesz stereograms”—duplicating the two perspectives the two
eyes will have of the object—the apparent distance of the patterns can be
made to vary by varying only the disparity (or gap). Three kinds of neurons
from area V1 are activated respectively when the image is located behind or in
front of the point on which the eye is focused.
This sensitivity to retinal disparity can be modified by a nonvisual (extra-
retinal) signal. In fact, if the subject puts on prism goggles that bring the target
closer without changing the disparity, the two eyes converge. A neuron acti-
vated by a visual target 20 centimeters from the eye is activated again when
76 •
the target is 80 centimeters from the eye because of the prisms. This experi-
ment shows that a signal connected to the convergence of the two eyes modu-
lates the response of these neurons. The distance can also be ascertained by
measuring how much the eyes cross! The exact origin of these signals of con-
vergence is unknown. Two possibilities are the ocular muscle receptors or an
efferent copy of the motor command for convergence; in other words, a mo-
tor signal or a combination of the two. Many structures of the visual cortex—
V1, V2, MST, LIP (lateral intraparietal), PO (parieto-occipital), and so on—re-
ceive information about the position of the eyes in this way.
The idea that variables such as distance are constructed by unconscious in-
ferences, as Helmholtz suggested, is not accepted by all the philosophers. Even
Merleau-Ponty, whose ideas have been validated by experimentation, was con-
fused on this point: “How is it possible,” he writes, “to assert that perception
of distance is a construct of the apparent size of objects, of the disparity be-
tween the retinal images, of the accommodation of the lens, of the conver-
gence of the eyes, and that the perception of depth is a construct of the differ-
ence between the image supplied by the right eye and that supplied by the left,
when, if we hold to phenomena, none of these ‘signs’ is clearly presented to
our consciousness? How can we reason in the absence of premises?”45
The Coherence between Seeing and Hearing
What of the cooperation between two other sensory modalities, sight and
hearing? Everyone knows that you can catch an object more easily if, at the
same time you see it, you hear it coming. This cooperation brings into play a
special structure of the brain, the superior colliculus. Essentially a biological
machine that recognizes moving objects and identifies their distinctive features
using multisensory cues, the superior colliculus holds the key to how the brain
directs multisensory fusion and extraction of pertinent signals. It controls ori-
entation and avoidance reactions, among other behaviors, and it is a splendid
example of a sensory and motor structure that directs the execution and cor-
rection of movements carried out by multiple effectors like the eyes, the head,
the trunk, and the limbs.
It is impossible to catch moving prey without focusing in front of where
the prey is and very rapidly correcting the trajectory in the event of error. The
colliculus is thus involved in anticipation and motor prediction. Its presence
also explains how the brain assembles motor commands in a manner suf-
ficiently general for the movement to be accomplished by a variety of effect-
ors (tongue, hand or paw, foot, and so on) with differing dynamic properties.
77
The colliculus (from the Latin collis, hill), called the optic tectum in non-
mammalian vertebrates, is composed of two symmetrical parts (hence the ex-
pression “the colliculi”). There is a superior colliculus and an inferior col-
liculus. I will consider the superior colliculus here. Lesions of this organ cause
impaired reactions of orientation and of catching a target. A damaged animal
can catch a mouse, but less rapidly, which indicates its ecological importance.
The superior colliculus is composed of seven layers of nerve fibers and neu-
rons, and its complexity is formidable. There are, however, three principal lay-
ers, called, prosaically, superficial, intermediary, and deep. These neurons re-
ceive information from more than twenty cerebral structures. The afferents of
the superior colliculus are mainly visuosensory, proprioceptive (also called so-
matic), and auditory. But it receives many cortical projections as well as pro-
jections from structures such as the basal ganglia, the cerebellum, and so on.
Other afferents, whose functions are poorly understood, come from structures
such as the raphe nucleus and the locus ceruleus. As Stein and Meredith sum
up, “Some of these latter inputs may play a role in varying the likelihood of a
given movement depending on the prior experience and immediate needs of
the organism.”46
First I will address the visual responses. If you move a point of light in
front of an animal, the neurons of the colliculus, like those of the visual cor-
tex, will respond with a discharge of action potentials. The response to the vi-
sual stimulus is selective for a particular region of visual space (this is the idea
of a receptor field). The receptor fields of these neurons are expansive. They
are bordered by an inhibitory region: if a target is present simultaneously in
the center of the receptor field and at the periphery, the response from the
center is reduced. The size of the effective stimulus is, in general, less than
that of the receptor field. Some neurons are particularly sensitive to move-
ment in certain directions. Finally, repeated exposure induces a habituation
that results in extinction of the discharge. The colliculus participates in detect-
ing the unusual. If the animal turns its attention to the visual target, the sen-
sory response is increased. This amplification of the response to a stimulus
adds weight to arguments in favor of the active character of perception.
Amplification also takes place just before an ocular saccade toward the target;
further proof that intended movement modulates the sensitivity of the first
sensory relays.
In the deep layers of the colliculus are neurons that project to the brain-
stem and the medulla where they induce ocular movements toward the target
(saccades) (Figure 3.6). A crossed projection is involved in orienting move-
ments of the head, the eyes, the trunk, and perhaps the limbs. Another, un-
78 •
Tecto-reticulo-spinal
Superior
neuron (TRSN)
colliculus
Electrodes for injection
and recording
Abducens Vestibular and

nucleus prepositus
nuclei
2
1
toward
the spinal
cord
5 mm Reticulospinal
neuron
Figure 3.6. The neurons of the superior colliculus project to the motor centers
of the eyes and the head. The colliculus and the motor centers of the
brainstem interact to control movements and posture during orientation of the
head toward an object. The tecto-reticulo-spinal neuron (TRSN) connects the
colliculus to the brainstem. It is active when an animal directs its gaze to a vi-
sual target. It projects to neurons of the brainstem that activate the motor cen-
ters of the eyes, head, and perhaps the arms and trunk.
crossed, initiates recoiling, or flight when a predator is approaching. Other

neurons project to the thalamus and probably inform the cortex, via the
thalamus, of ongoing movements. Still others, projecting to the contralateral
colliculus, undoubtedly ensure the coordination of the two sides and play a
role in reactions to approaching objects. These neurons are arranged in such a
way that their location in the deeper layers of the colliculus is in the same reg-
ister as the retinal representation of the higher layers.
In other words, if a target (a bird, for example), appears in one’s visual
field, the population of neurons of the colliculus that will be activated in the
superficial layers is situated exactly above the neurons of the deep layers that
have to be activated to make a combined movement of the eyes and head to-
ward the bird. To put it simply, the colliculus contains two sorts of maps: sen-
sory maps (in the superficial layers) and motor maps (in the intermediary and
deep layers) in the same conical section.
Many of the neurons in the deep and superficial layers of the colliculus re-
spond to multisensory stimuli. There are, for instance, three sensory maps
that share the same neurons: a visual map (retinotopic), a map of auditory
space (audiotopic), and a map representing the parts of the body (somato-
topic). These maps are particular to each species. For example, in the rat,
whiskers play a fundamental role in exploring space and in trapping; they are
represented extensively in the colliculus. In primates that have a fovea, the cen-
tral region of the retina is represented in numerous neurons. The receptor
79
fields of each of these modalities are different, and it is important to beware
of taking too simplistic a view of these three neuronal representations of sen-
sory space. The essential point is that these three domains of perception share
a common sensoritopy: retinal space.47
This representational coherence is very important; it goes a long way to-
ward simplifying the coordination of receptor space to enable a quick, single
orientation, directed toward a goal. At the same time, I will show that these
maps are the center for multiple phenomena of selection and of dynamic re-
adjustment that, provided they are investigated, will one day help to explain
the predictive character of orienting movements.
The colliculus is an interesting structure for understanding how the brain
handles the problem of spatial and temporal coherence of messages supplied
by the different senses. It is obvious, in fact, that if you suddenly hear the
sound of a door lightly closing on your right and you turn your head quickly
to direct your gaze toward the door, you have to coordinate at least three
spaces: vision, that is, the retinal space on which the image of the room you
are in is projected; hearing, that is, the auditory space detected by your two
ears; and the space of your body, which has to turn to look in the direction of
the door that has shut. These adjustments of retinal, auditory, and somatic
space are accomplished in the superior colliculus. Let us first look at an exam-
ple of adjustment between auditory space and vision.
Although auditory signals are received by another structure, the inferior
colliculus, the neurons of the superior colliculus are also activated by sound.
These neurons do not have exactly the same properties as those of the pri-
mary auditory pathways. The latter are sensitive to narrow bands of fre-
quency and constitute what is called a tonotopic map; that is, they gradually
allow the animal to recognize sounds emitted by others of its kind, as is the
case with color recognition via visual pathways.
At any rate, no representation of auditory space in the classical hearing
pathways can be found; it is encoded in the superior colliculus. Its neurons are
relatively insensitive to pure sounds. They prefer complex sounds like those
produced by clapping the hands or whistling—in short, natural noises like
those prey or predators might make. They are very sensitive to novel sounds.
Most of them are binaural: they respond to stimulation of both ears. They
construct large receptor fields based on this binaural detection, taking into ac-
count the (very short!) time between the arrival of sounds in both ears.
Remember that reciprocal reinforcement exists between visual and audi-
tory inputs at the level of each neuron where these two modalities converge.
How can this cooperation be demonstrated? A neuron that responds with a
80 •
strong discharge to a bird singing cannot respond to either of the two aspects
of the stimulus—the image of the bird or the song—presented alone.
A fundamental property of multisensory integration in the colliculus is
this: the receptor field is the pertinent reference for multisensory integration,
not the external space. Merging of receptor signals occurs within the space of
the receptor fields and not by centrally reconstructing the external Cartesian
space. Here there are rules that will be encountered again later with respect to
the neurons in the putamen and area 6 of the cerebral cortex that are acti-
vated by both visual and tactile stimuli.48
The auditory map constructed in this way can be shifted by movements of
the eyes or ears or by wearing prism goggles that shift the visual space with re-
spect to that of the head. What explains this flexibility and this influence of
motor signals on neural maps?
Say that a bird appears at the periphery of your visual field. Because the
position of the bird in space is given on the retina (in retinal coordinates), its
position on the retina will depend on the direction of your gaze. Try this ex-
periment. Look in front of you, eyes straight ahead. If the bird appears at 45
degrees to the right and begins to sing or to make a little chirp of fright, the
two maps—visual and auditory—will coincide, and the same population of
neurons of the colliculus will be activated and will produce the same orienting
movement of the eyes and the head. Now, if you keep your head facing front
but shift your eyes, the two maps may not coincide. The influence of the di-
rection of gaze on the colliculus recalibrates the two maps without the brain
having to make complicated calculations.
Other interactions of a spatial nature modify the neurons of the colliculus.
The response to two stimuli from the same place is favorable to the selection
of objects of interest by the configurations of stimuli that characterize them.
Another mechanism that allows selection of interesting targets is the depres-
sion that accompanies the simultaneous presentation of two stimuli that are
not located in the same place. In fact, each multisensory neuron of the
colliculus has a zone at the periphery of its receptor field that inhibits re-
sponse. This property is very widespread throughout the nervous system and
in the visual cortex: it is believed to function in spatial selection.
Another problem the brain has had to resolve to enable the fusion of
multisensory information is that of time shift. Different lengths of time are ac-
tually needed for sensory signals to reach the colliculus. For example, a sound
introduced to the ear takes around 13 milliseconds to get there; a touch
around 25 milliseconds; a visual stimulus around 40 to 150 milliseconds, ac-
cording to the alertness of the animal.
81
But it gets even more complicated. If a bird sings in a tree 50 meters away
from a cat, the sound transmitted through the air at 330 kilometers per second
will take about a tenth of a second to arrive at the colliculus of the feline,
whereas light, which moves at 300,000 kilometers per second, will arrive at the
retina almost instantaneously.
The solution adopted by the nervous system is one called “temporal win-
dows.” It is extremely simple and extremely flexible. Electrophysiological re-
cordings show that a light stimulus induces a discharge that can be maintained
for more than 100 milliseconds; if a sound arrives after several hundred milli-
seconds, the amplification of the response can still occur. The neuronal net-
work of the colliculus thus develops a memory that maintains the sensitivity
of the multimodal neurons during a certain time, hence the name temporal
windows.
Temporal windows would explain in large part the success of predators.
Stein and Meredith note:
The presence of these features hardly seems coincidental. Quite the
contrary, they are likely to have developed and been maintained during
evolution because: (1) the presence of wide temporal windows gives
critical flexibility in detecting and responding to minimal, albeit impor-
tant stimuli at different distances from the animal . . . and (2) the ap-
pearance of maximal depression when spatially disparate stimuli are in
close temporal proximity is a means of focusing attention on the stron-
gest, and presumably the most important, stimulus in the presence of
potential distractors.49
But the nervous system has other mechanisms for resolving the prob-
lem of time shift. The cerebellum receives sensory information from a wide
range of sources, in particular the proprioceptors of the limbs. It is important
for this structure to be able to compare the messages from the receptors of
the feet, arms, and neck, for example, because it is involved in coordinating
movements, especially via the precise temporal organization of gestures and
relationships between posture and movement. The solution used by the ner-
vous system may be to vary the velocity of conduction of the nerve fibers so
that all the signals arrive at the same time. The signals from the receptors of
the feet, located more than a meter from the cerebellum, are transmitted
more rapidly than those of the receptors of the neck located a few dozen
centimeters away! But the problem of time shifts is so important that the
brain has probably found numerous different solutions that should interest
roboticists.
82 •
The Visible and the Tangible
Many philosophers have emphasized the cooperation between vision and the
sense of touch. Merleau-Ponty wrote:
We must get used to the idea that everything visible emerges from what
is tangible; that every tactile being is destined in some way to be seen;
and that there is encroachment and overlap not just between touched
and touching but also between the tangible and the visible embedded in
it, just as, conversely, the tangible itself is not devoid of visibility, that is,
not without visual existence. Since the same body sees and touches, vis-
ible and tangible belong to the same world. A wonder too seldom no-
ticed is that any movement of my eyes—better yet, any displacement of
my body—has its place in the same visible universe, that through these
movements I examine and explore, just as, conversely, seeing occurs
somewhere in tactile space.50
Sartre, too, stressed the reciprocal analogies between vision and touch: “1.
A series of kinaesthetic (or tactile) impressions can function as analogue for a
series of visual impressions . . . 2. A movement (given as a kinaesthetic series)
can function as analogue for the trajectory that the moving body describes or
is assumed to describe, which means that a kinaesthetic series can function as
analogical substitute of a visual form.”51
The neuropsychologist De Ajuriaguerra, known for his rigorous analysis
of the relationships between mothers and infants, focused on the complex na-
ture of perception in his published work and in his courses at the Collège de
France. “Perception can be divorced neither from ocular motor function,” he
wrote,
nor from cognition, nor from the affective life of the subject. Perception
confirms the world of objects: there can be no cognizance of things and
beings without exploration: tactile, for the blind; tactile and visual, for
the sighted. Although it has been proposed that the two domains (optic
and haptic) are controlled by different laws, studies of blind subjects
who have recovered their sight show that they do experience some kind
of transfer between the two modalities.52
Seeing with the Skin

The Braille alphabet gives congenitally blind people access to reading. To the
extent that it allows them to construct a tactile representation of a page of
83
text, it is an example of sensory substitution and equivalence between touch
and vision.
This functional equivalence between the visible and the tangible was bril-
liantly demonstrated by the experiments of Bach-y-Rita on the use of vibra-
tions as a visual substitute for the blind.53 He perfected a technique of stimu-
lating the skin with a matrix of small vibrators activated by an image on a
television camera. The boundaries of tactile perception had been known since
Mountcastle’s work on vibrational sensitivity, comparing the thresholds of ac-
tivation of cutaneous receptors in humans and in monkeys. The curve of sen-
sitivity to cutaneous vibrations has a bell shape with a maximum around 80 to
100 Hz. Actually, the bell shape conceals two curves that correspond to the
two types of cutaneous receptors activated by the vibrations: the Meissner
corpuscles, which have a maximum sensitivity around 40 Hz, and the Pacinian
corpuscles, which are especially sensitive around 80 to 100 Hz. These delicate
vibrations applied to the skin were of very weak intensity; they acted mainly
on the cutaneous receptors and induced no muscular reflexes or illusions of
displacement of the vibrated limb.
In this way, a tactile vibratory image was created on various parts of the
body (hand, abdomen, back, and so on). Remarkably, the perceptions induced
by these tactile images all had the normal properties of visual perception such
as parallax, perspective, enlargement (as when an object approaches the body,
for example), and depth perception. Tactile and visual perception were shown
to be equivalent. If a ball was thrown in front of the subject, even if the image
of the ball was projected by the vibrators attached to the subject’s back, he
still perceived it as being thrown in front of him. Thanks to this tactile matrix
device, a blind subject could play a simplified version of table tennis and thus
predict the trajectories of moving targets in space.
When a person using this television camera trains with a matrix placed on
the abdomen, the training persists when the camera is placed on the forehead
or on the hand. The reason for this equivalence is that the image is not per-
ceived to be skin-deep but in the external space, as one perceives the ground at
the end of the walking stick and not in the palm of the hand that holds the
handle and where the force feedback is felt. The image is projected into space
by the brain. What a wonderful illustration of what I mean by the need to
construct a projective theory of perception!
This was the technique used to help congenitally blind people to read: a
little vibrator placed at the end of the finger projected letters onto the skin.
However, the method ultimately had some limitations and did not replace the
Braille tactile alphabet. There is probably something in the active character of
84 •
Braille that cannot be replaced by techniques that passively transmit sensory
information. Indeed, although the subject can manipulate the television cam-
era to explore the visual space, he probably does not produce the combination
of active movements and sensations made possible by the Braille method. I
will return to the importance of active exploration in perception.
These data suggest that tactile inputs have access to the centers of the
brain that process visual information. Is there presently proof of this? Record-
ings of cerebral activity in early blind people have shown that the first central
sensory relays that process visual information in the cerebral cortex are acti-
vated by tactile stimulation when these subjects are trying to imagine an ob-
ject.54
I have discussed how the visible and the tangible are combined in perceiv-
ing shapes and movement. These two modalities are also involved in the co-
herent perception of one’s own body and in establishing multiple frames of
reference connected to different parts of the body. The neurons in several
zones of the brain—the putamen, the frontal cortical area 6, and the parietal
area 7b—respond when a target appears in certain parts of the visual field and
are also sensitive to touch. By this I mean that they fire when certain places on
the skin are stroked or brushed; they are bimodal.55 These bimodal neurons
have fascinating properties. For example, if a neuron fires when a monkey’s
cheek is touched, the same neuron will also fire when the hand is brought
close to the cheek without touching it. The visual spatial receptor field is an-
chored to the tactile receptor field (Figure 3.7). Thus the activity of the neu-
ron can be triggered by placing the hand or an object within a space of about
10 centimeters surrounding the cheek. This discovery has several important
implications.
First, it confirms the equivalence between visual and tactile stimulation. It
also explains why, when I bring my hand to my cheek, I feel my hand on my
cheek even before it actually touches me. This convergence, this bimodality,
actually entails something more profound than simple equivalence. In this
case, the visual perception is not only the analog of tactile perception, as in
the experiments of Bach-y-Rita—it anticipates it. Mere proximity becomes
contact through anticipation of the area of the body that will be touched.
Gross described a remarkable cell that has a tactile receptor field extending
from the eyebrows and a visual receptor field within a cone whose summit is
the eyebrow and that projects in front of the subject, in the direction of his
gaze. This neuron is activated by any object that approaches the eye; in a way,
it extends the skin’s zone of receptivity. Vision is indeed, as Merleau-Ponty
said, “the brain’s way of touching.”
85
Area of visual space
that activates the
same neuron
Area of skin on the

cheek that activates
the neuron
Figure 3.7. Vision and touch: regions of the skin and visual space for which a bi-
modal neuron of the putamen is activated in the monkey. When the finger ad-
vances in the visual receptor field of the neuron, the neuron discharges, and
discharges again when the finger touches the skin in the tactile receptor field.
When the finger gets close to the skin but does not touch it, the monkey will
have the impression of contact even before contact is made.
But Gross’s discovery had an even more profound impact on our under-
standing of multisensory functioning and mechanisms of interaction. Take
the properties of the neurons of the arm that are sensitive to stroking. In the
putamen, bimodal neurons have a tactile receptor field that is anchored to the
visual receptor field: if a monkey’s arm is placed on a table before it, the tactile
receptor field being, for example, on the wrist, the visual receptor field will be
within a cone connecting the gaze to the wrist or in a space surrounding the
wrist. If then the monkey’s arm is moved, its visual receptor field will follow
to stay in the same spatial register as its arm. This spatial updating is probably
a response to proprioceptive signals; it can be induced by copies of motor
commands.
To Be Touched or to Touch? That Is the Question

I have repeatedly emphasized that the multisensory character of perception
includes signals that derive not from the senses but from the intention to
move. The active character of perception is evidenced by this profound influ-
ence of the intentional character of gesture. Here is an example.
Close your eyes, and touch your left hand with your right hand. You will
feel two very different sensations that either overlap or cancel each other out.
Merleau-Ponty was absolutely fascinated by this observation:56
86 •
Husserl57 refers to an experiment of touching. When I touch my left
hand with my right hand, my touching hand grasps my touched hand
like an object. But suddenly I notice that my left hand is beginning to
feel. The relationships are reversed. We experience an overlap between
the contribution of the left hand and that of the right hand, and a rever-
sal of their function. As a physical object the hand always remains what
it is, and yet it is different according to whether it is touched or touching.
Merleau-Ponty often comes back to this example, which for him remains one
of the great mysteries of perception. I have put the final part of the last sen-
tence in italics, because it suggests very clearly how the idea of perception is
different in these two cases, one where the limb is touched passively, and the
other where it is touched by a motion the subject makes himself. This long
philosophical introduction is just to say that evidence now exists of a differ-
ence in activity of the neurons of the brain between these two cases.
Tactile perception is not only connected with vision, it is also influenced
by the active character of visual attention. In one zone of the superior tem-
poral sulcus (STS) in the monkey, neurons are activated when the hand of
the monkey is lightly stroked with a stick; but if the monkey grabs the stick
himself, the same neurons stop firing. The active movement suppresses the
transmission of tactile information. Perrett attributes this disappearance of
response to the predictive character of the contact, and observes that the neu-
rons of this area are very sensitive to unpredictable tactile stimuli.58 In con-
trast, when the movement of the animal induces the contact, a still mysterious
mechanism of inhibition suppresses the activity. This observation corresponds
very well with the subjective impression that we have of a very reduced sensa-
tion of contact when we touch ourselves.
When the monkey’s active exploration leads it to encounter repeatedly

an object of particular texture and compliance at a particular loca-
tion, the object’s tactile properties can be said to become “expected.”
As shown above, tactile feedback, which results from the monkey’s
reaching and contacting such expected objects, fails to activate somato-
sensory cells in the STS. If, however, a novel object of different texture/
compliance is introduced at the same location, tactile feedback does re-
sult in activation of STS cells.59
This selectivity also holds true for vision: the cells of the STS sensitive to
the entry of visual objects into the visual field of the monkey lose their sensi-
87
tivity when the arm of the animal comes into the field. These effects are not
due to the simple fact of attention. They are actually very selective inhibitions
of an aggregate of multisensory properties (texture, visual aspect, space) of a
part of the body of a monkey or a known object. Perception is selection and
anticipation. Time will reveal their mechanisms.
This selection by anticipation is not an isolated case: there are many ac-
counts of reduced effects of sensory stimuli due to self-activity of the animal.
For example, the Purkinje cells in the cerebellar cortex—an organ important
for the coordination of gestures—respond much less to cutaneous stimuli
when animals make endogenous movements than when experimenters touch
them. Generally, the effect of a sensory stimulus is reduced during endoge-
nous movements. However, when the task is new, or when a stimulus is unex-
pected, sensitivity is restored. In a recent review, Prochazska concluded: “We
have seen many examples of task and context-dependent sensory transmission
in widely ranging species and widely ranging motor behaviours. From all this
it seems safe to conclude that anticipatory gain control of sensory transmis-
sion is indeed a fundamental strategy of motor systems.”60
Force Feedback and Visuomotor Functions

Cooperation between vision and the different senses extends to other pro-
prioceptive functions—to the sense of effort, for example. To study such coop-
eration, in this case the contribution of haptic information to the visual per-
ception of the properties of objects, my team and I built a force-feedback
joystick. But I am getting ahead of myself. If I lay a pair of glasses or a pen on
my desk, I can, thanks to my vision, glean quite a bit of information about the
volume and the geometry of the object lying there. I can even infer certain dy-
namic properties; for instance, the number of degrees of freedom, the object’s
weight, its elasticity. Now, if I close my eyes and simply take the object be-
tween two fingers (try it with a pair of glasses or any complex object nearby),
the perception of the force the object is exerting on my fingers given me by
my haptic sense—that is, the combination of proprioceptive information from
receptors in the skin, muscles, tendons, and joints—allows me to confirm
many of the dynamic properties of the object: its mass, its inertia, the number
of degrees of freedom, its viscosity, and so on. I use the term “force feedback”
to describe the nervous system’s use of force as information.
Here is an example that explains which mechanisms underlie this multi-
sensory cooperation. Suppose that I want to follow a rigid surface with my
finger, so as to position a peg in a hole. My brain can adopt any of three strate-
gies for following this surface.61 First, it can ignore any forces and carry out a
88 •
simple visual tracking of the contour of the surface. Second, it can apply a
continuous force to the surface and detect that my finger loses contact when
the force becomes nil. To keep my finger in contact with the surface, I just
have to control the intensity of the force detected by touch, as with a servo-
control. Finally, my brain can adopt a third strategy: it can form an internal
model of the shape of the object to follow (here a surface) and program a
movement to follow a trajectory that passes slightly inside the surface. At the
same time, the muscles are controlled so that my fingertip acts like a supple
spring (in robotics terms, I might say that the strategy is compliant). The dif-
ference between the programmed position of my fingertip (inside the surface)
and the actual position of my fingertip (on the rigid surface) will cause the vir-
tual spring to compress, resulting in a slight pressure of my fingertip on the
surface. In addition, the spring will expand or contract to absorb moderate dif-
ferences between the actual surface and the internal model, thus maintaining
contact with the surface throughout the movement.
These conditions make clear the respective roles of vision and force feed-
back in each of these strategies: in the first, vision is essential; in the second, it
is scarcely involved; in the third, it serves to construct an internal model of the
trajectory that makes it possible to anticipate and control the consequences of
the movement.
I want to stress the essentially predictive character of the third strategy.
The fact of having constructed an internal model of the trajectory allows my
brain not to have to wait until my finger has run the entire course to correct
errors but, on the contrary, to anticipate certain properties of the trajectory
and to make assumptions. Perception thus becomes a comparison between an
expected state and an actual state. This is the heart of my thesis.
How can this type of process be studied? Suppose, first, that this surface
appears on a computer screen and that the object can be made to move with
the aid of a joystick. I can ask the subject to follow the object with the joy-
stick. This task is similar to that of playing video games at a bar. But I can also
introduce a motor between the joystick and the apparatus that measures its
movement. Each time the object touches the surface on the screen, the motor
exerts a force on the joystick to give the impression of resistance. This impres-
sion will be all the stronger the more the subject leans on the joystick. In this
way I can simulate the force feedback that the real surface exerts on a subject’s
hand when he actually carries out the task. This device shows that the brain
uses a mixture of the second and third strategies. Consequently, vision and the
haptic sense cooperate in a task of contour-following by constructing an inter-
nal model of the surface.
89
The Problem of the Coherence and Unity of Perception
Models of multisensory interaction are interesting because they enable us to

make predictions about the way sensory receptors combine their signals. But
they do not do anything to solve the fundamental problem, which is that of
the unity of perception, or coherence. How the perceptual unity of a visual
object can be achieved by temporal synchronization of the discharge of neu-
rons that respond to different properties of the object was discussed in Chap-
ter 1. But constructing coherence means solving many problems, because sen-
sory inputs have properties that separate them and make their fusion difficult.
Sensory inputs are ambiguous. The problem is one of the difference be-
tween acceleration in one direction and braking in the other in the canals and
otoliths, of Coriolis accelerations in the canals, and of speed in vection. Sen-
sory inputs do not occur in the same systems of coordinates. The canals are in
a Euclidean frame of reference connected to the head; vision occurs in the ret-
inal space; proprioception of the eyes, neck, and limbs occurs in the space of
the muscles, so sensory space is, in this case, utterly linked to motor space;
hearing takes place in the space of frequencies. I have offered one example of
a solution: the similarity of the planes of detection for visual and vestibular in-
formation.
Sensory inputs are staggered in time. Try to move your finger and your
foot in synchrony. Wait! Don’t start right away. You will notice that your first
attempt will not be synchronous. It will take you several tries to synchronize
your movements. In fact, signals from the muscle receptors of the foot arrive
at the cerebellum with a delay over those of the fingers that can be consider-
able when you think that nerve impulses circulate in the medulla at velocities
of 10 to 80 meters per second. Thus in humans it takes 10 to 100 milliseconds
for contact of the foot on the ground to reach the cerebellum and two times
less long for signals from the finger. Ten times less for signals from the tongue!
It is thought that the brain anticipates the movement of the foot or retards the
information from the finger. I have shown several solutions to this in neural
operations in the colliculus.
Sensory inputs do not cover the same range of velocities. The vestibular
receptors are very fast and detect accelerations; vision is slower. The receptors
of the muscular spindles have widely varying response times, which is also
true for the tactile receptors, whose bandwidths are different from those of
vision, whereas very tight perceptual relationships exist between sight and
touch.
90 •
Finally, sensory inputs are often fuzzy. They produce innumerable random
fluctuations that are unrelated to the magnitude detected. This noise is due to
the chemical and mechanical properties, among others, of transduction, but
also to neural noise introduced into central relays.
So the problem of coherence is not only a problem of geometry or of
dynamics. It assumes active central mechanisms that permit resolution of am-
biguities, catching up with or anticipating differential delays between recep-
tors, unifying space via clever biological mechanisms that are more than just
changes of coordinates, and so on. A genuine theory of coherence has yet to
be constructed.
I will start from the supposition that constructing coherence is not only an
effect of convergence, but also the product of a central activity that depends
on the knowledge of a priori mechanisms that each species possesses. “I real-
ised,” wrote Marcel Proust,
that it is not only the physical world that differs from the aspect in
which we see it; that all reality is perhaps equally dissimilar from what
we believe ourselves to be directly perceiving and which we compose
with the aid of ideas that do not reveal themselves but are none the less
efficacious, just as the trees, the sun and the sky would not be the same
as what we see if they were apprehended by creatures having eyes dif-
ferently constituted from ours, or else endowed for that purpose with
organs other than eyes which would furnish equivalents of trees and
sky and sun, though not visual ones.62
Perception is an interpretation; its coherence is a construction whose rules

depend on endogenous factors and on the actions that we plan. The difficulty
in building a theory of coherence is that there is most likely not one single co-
herent theory for all of perception. There are often several ways of arranging
sensory data to construct a coherence. This range of possibilities is probably a
key to the way illusions are manufactured.
Many mathematical models have been proposed to account for multi-
sensory interactions, but those that emphasize the problem of coherence are
rare. In general, these models merely execute a combination of sensory sig-
nals with more or less predictive operators.
Droulez and Darlot proposed a model of multisensory fusion that directly
addresses the problem of coherence.63 Their central idea is that the brain is not
interested in the physical variables of classical mechanics, or which stimuli are
specific to which receptors, in and of themselves. The brain seeks to recon-
91
struct variables relevant to the behavior and action of the organism. To this
end, it uses the fact that messages about movement, supplied by the array of
sensory receptors, are in general redundant. In other words, the same move-
ment variable can be calculated or estimated by several combinations of re-
ceptors, in addition to any specialized receptors.
The angular velocity of the head in space during movement provides a
good example. It is an important variable because it is a derivative of displace-
ment. One can thus integrate (in the mathematical sense) and obtain a dis-
placement that the brain will use either to activate stabilizing reflexes or to es-
timate trajectories or even to coordinate the movements of the eyes, head,
hands, and so on.
According to the basic assumption of the model of Droulez and Darlot,
the brain estimates this velocity of the head in at least two ways. The first, di-
rect, is detection by a specialized receptor, the semicircular canals, which de-
tect angular acceleration of the head in its plane, called the horizontal plane.
But this specialized receptor transmits a velocity to the brain, because its
viscoelastic properties delay the signals. This is precisely what an integrating
filter does. The second, indirect, way that the brain estimates the head’s veloc-
ity is a multisensory convergence of other variables—for example, the velocity
of retinal slip and that of the eye in its socket. Adding these all together gives
an estimation of the angular velocity of the head.
The main measure of this model is thus the discrepancy between a specific
sensory input and a prediction of this variable from combinations of other
signals. It was first suggested in 1989 to account for the fusion of the sen-
sory messages supplied by the semicircular canals, the otoliths, and vision.
Droulez and Cornilleau-Pérès wrote:
The fact that individual sensors provide reliable information only in a
limited (and sometimes variable) working range, in both spatial and
temporal domains, and the fact that the degradation of the measure by
noise can depend on the context (see for instance the analysis of optic
flow) and the existence of multiple interpretations suggest that sensory
signals cannot be processed as a set of measures but as a series of con-
straints upon the internal estimates. We suggest that sensory signals are
not used to directly estimate the relevant variables but to estimate the mis-
match between internal estimates and measures.64
For example, it really does seem that in vision the brain employs at least
two constraints: continuity and rigidity. Indeed, the objects that surround us
are in general locally continuous (that is, they are not divided). The painters
92 •
called pointillists provided a vision of nature that only makes sense because it
is filtered by the continuity of perception. Moreover, objects are generally
rigid. A body of evidence shows that the brain makes these assumptions of
continuity and rigidity, which helps in constructing coherence. The brain also
favors symmetry and thus tends to impose symmetry on the world it per-
ceives.
But the neural basis of coherence is still not well understood. In all likeli-
hood, coherence is due, in part, to the simultaneity of neural activities in sev-
eral parts of the brain.65 Many recent theories and experimental results in fact
suggest that establishing relations among features of perception that have
been analyzed separately in several areas of the brain might also be achieved
by temporal synchronization; that is, simultaneous activation of several
groups of neurons. This temporal encoding most likely takes a variety of
forms that have yet to be discovered.
Autism: The Disintegration of Coherence?
It seems to me that the disintegration—the fragmentation—of coherence is

the source of numerous symptoms of several major illnesses that have as
much to do with psychiatry as with neurology. A few reflections on autism
will illustrate this thesis. Autism is a developmental disorder that strikes very
young children. It is not known what causes it. The main symptoms of autism
are, first, a qualitative alteration in social interaction: autistic children do not
look at the people speaking to them; when they are put in a group, their gaze
does not follow the activities of the members of the group. Their behavior
isolates them. Next, they begin to exhibit impaired communication: they
hardly speak, do not return a smile or respond to facial expressions of emo-
tion, and so on. Finally, they engage in stereotypical behaviors that they repeat
endlessly, totally divorced from any social context or even any connection with
the activity in which they are engaged. They suffer from a perseveration syn-
drome. Autistic children show disturbed initiation of action and, more gener-
ally, disturbed intention. Like patients with Parkinson’s, they freeze in initiat-
ing certain actions. By “freeze” I mean that they are perfectly capable of
making a move, but that they decide not to begin it.
They display what appear to be nonmotor disturbances of anticipation of
the hand, since if they suddenly commit their hands to action, muscle tone re-
turns. They exhibit visual and auditory disturbances. Some of these children
can be classed in subgroups that present particular symptoms to a greater or
lesser degree.
93
It is possible to associate these symptoms with several types of dysfunc-
tion: initiation of walking to disturbances of the basal ganglia, the absence of
posture and of anticipatory synergy to disturbances similar to those described
by Babinski in patients with lesions of the cerebellum, disturbances of motor
intention, and perseveration with prefrontal and frontal deficits. But it is also
possible that these disturbances have a common origin. These children are in-
capable of relocating local aspects of their action in a global context.
In The Enigma of Autism, Uta Frith gives several examples of the disintegra-
tion of coherence that are helpful in explaining what I mean by coherence.
She cites the example of little Jerry:
According to Jerry, his childhood experience could be summarized as

consisting of two predominant experiential states: confusion and terror.
The recurrent theme that ran through all of Jerry’s recollections was
that of living in a frightening world presenting painful stimuli that
could not be mastered . . . Nothing seemed constant; everything was
unpredictable and strange. Animate beings were a particular problem
. . . I would like to suggest that the repetitions and sensations that were
such poignant experiences in Jerry’s life (and also in the lives of many
other autistic people) are two sides of the same coin: part of the same
underlying problem. This problem is what I have already labelled lack
of central coherence. Let us consider what it would be like if percep-
tion reflected a fragmented rather than a coherent world.66
Autistic children develop what Uta Frith calls a “local coherence.” She cites
another case; that of Elly, a child who had constructed a system of telling time
based on the shadow of her body cast by the sun.
(For instance, Elly paid close attention to shadows because they were
for some reason important and meaningful to her, and very relevant
to her moods. When she travelled to a different time zone, she was
alarmed that her shadow at 6 pm was not where it would have been at
home. She could not relax until her mother had explained to her that
6 pm on her watch meant it was only 5 pm at the new place.) This exam-
ple suggests that Elly had a limited but strongly coherent scheme about
the position of the sun at a certain time and the length of her shadow. It
really mattered to her when there were unexpected discrepancies. The
scheme had to be kept coherent. This insistence on sameness is a type
of local coherence. It is not at all like central coherence.67
94 •
I think this example is very interesting because Elly had devised a method but
had great difficulty placing it in a broader context—in thinking about changes
in time zones, for example.
Interestingly, studies in cognitive psychology on the problem of decision
making show that our cognitive apparatus makes many mistakes, due in large
part to our inability to depart from stereotyped schemas characterized by Uta
Frith’s local coherence. We have trouble revising according to context. The
perseveration of autistic children, their immutability, is perhaps only an ex-
treme version of many widespread behaviors.
But the most interesting aspect of Uta Frith’s theory is that she goes even
further in the analysis of the importance of coherence: coherence, she says, is
necessary not only to construct a perception of the body or of its relationship
with the environment, but also to elaborate what is called a theory of mind.
This notion was invented by psychologists to refer to the fact that we attribute
thoughts to others, that we have an idea, a theory of what they have in their
minds, of their intentions, and so on. Premack showed that this ability to at-
tribute thoughts to others shows up in nonhuman primates, and he gives sev-
eral examples of it. And actually anybody who has a dog will have noticed this
faculty in the animal. Uta Frith says with respect to the autistic child: “We
have hypothesized that they [autistic children] do not have the basic propen-
sity to pull together vast amounts of information from events, objects, people
and behaviour. Even if they had the cognitive prerequisites that enable them
to mentalize, they would only form ‘small’ theories about mental states, but
not a comprehensive theory of mind. Autistic children are behaviourists. They
do not expect people to be kind or to be cruel. They take behaviour as it is.”68
She is right in saying that if autistic children have no coherent representa-
tion of the world, they cannot construct a theory of mind for others and thus
communicate. It is certainly not possible to make any internal assumptions
about the intentions of others if one has not succeeded in giving coherence to
one’s perception of relationships between oneself and the environment, and
with all the information it contains.
Uta Frith completed her analysis by observing in autistic children the im-
pairment of what she calls “planning functions” (working memory, control of
impulses), which are normally a function of the frontal lobe of the cerebral
cortex. There is a paradox here, for although the impairments of autistic chil-
dren imply a deficit of the frontal lobe, patients presenting with this deficit are
not autistic. It is possible that the question is one of differences connected to
stage of development.
95
In any case, how is it possible to imagine that children can coherently eval-
uate the people they see if they cannot evaluate relationships between their
own bodies and the environment? How is it thinkable that children who per-
ceive only a fragmented universe would have any desire to talk? How can be-
ings whose brains are the center of multiple incongruities have even the slight-
est desire to communicate with a world with which they cannot identify?
Finally, I would like to suggest a reason why autistic children have dif-
ficulty in perceiving the “global” features of the world and are confined to “lo-
cal” events. My hypothesis is that they may not be able to mentally manipulate
“allocentric” relations in their perception of space and remain stuck at “ego-
centric” coding of space, as discussed in Chapter 5.
96 •
4
FRAMES OF REFERENCE
The ethologist Golani’s description of the exchange of looks between two

dogs about to chew out each other’s throats will serve as an excellent in-
troduction to the problem of frames of reference. Golani used the system of
notation invented by the Israeli choreographer Eshkol to plot the movements
of dancers in three different but coexisting frames of reference: a so-called
body frame of reference that I will call egocentric; an environmental frame
of reference, also egocentric, but accounting for movement with respect to
the external framework of the space where the dance is taking place; and an
interactive frame of reference based on the connection of two partners that I
will call relative because it is defined by the relationship between the two
dancers.1
Golani based his account of the extraordinarily rigid connection arising
from the fixed gaze of two hostile dogs on the third frame of reference. He
showed that a line of attention as unmalleable as a steel rod is established be-
tween the eyes of the two animals; the dominant dog has only to tilt his head
as a first step, before ever touching, to fell the other. Any deviation means
death for one of them. In fact, the distance between their two heads is fixed,
and their relative position has only to vary several centimeters to facilitate a fa-
tal aggression. This finding would have been impossible without recognizing
that the frame of reference on which both dogs base their movements must be
relative; they ignore their environment just as the moon and the earth main-
tain a connection only through their mutual attraction.
This chapter explores one question: Which frames of reference does the
brain use to organize perception and action?
Personal Space and Extrapersonal Space
The concept of a frame of reference is tied to that of space. Our actions un-
fold in a space organized, according to Grüsser, into “personal” space, “extra-
personal” space, and “far” space.2 Each of these spaces is itself organized into
several subspaces explorable by different mechanisms that provide distinct
frames of reference.
The first experimental proof that the distinction between personal and
extrapersonal space had a neural basis was supplied by Hyvarinen3 and con-
firmed by Mountcastle and his students.4 Neurons situated in the parietal cor-
tex of the monkey fire each time a person enters its grasping space with a
grape. The sight of the grape (monkeys are fond of them) outside of this
space is not enough to trigger the activity of these neurons.
Personal space consists of self-space (egocentered). It is perceived by the
internal senses and is in principle located within the limits of a person’s own
body. But there are some important distinctions. First, the body can be per-
ceived as an external object, especially by vision. The hand that I see at the
end of my arm is not necessarily my own (in Chapter 2 we described the case
of a patient who denied possession of her arm). Thus, attributing an element
perceived by the body to itself requires a perceptual decision. The dissocia-
tions and denials of ownership of elements of the body that manifest in cer-
tain illnesses are all too familiar. Moreover, a perception of the body can occur
even in the absence of parts of the body. The case of phantom limbs, for in-
stance, reveals the existence of mental representations of the body—that is, in-
ternal models of the body independent of its presence. Patients whose arms
have been severed feel their arms perfectly and even feel the movement of
their hands. Sometimes the presence of the whole body is perceived as ambig-
uous, and the subject may experience a phenomenon called autoscopy.
Grasping space is itself divisible into intraoral and perioral space—espe-
cially important during the first six months of life—as well as into other local
spaces connected to specific activities related to grasping. It is marked out by
visual, tactile, and olfactory cues. Indeed, the perception of the body can be
extended by a tool. For example, you feel the tip of your hand as if it were at
the end of the pencil you are holding: the body extends into the tool. Contact
with the ground is felt at the end of stilts, the driver of a car feels the wheels
on the ground, and it is well known that pilots feel the wheels of the airplane
on landing as if they were their own feet. Haptic sensations play a decisive role
in these perceptions. This property of the body—integrating with the physical
elements it has grasped—is very important, for it often determines the frame
98 •
of reference. What is remarkable about this extension is that the object is per-
ceived where it is supposed to be in the extrapersonal space and not at the
point of contact of the instrument with the body. The brain is able to con-
struct a spatially correct extension.
This ability to extend the body and to localize the point of contact is prob-
ably acquired very early, during multiple movements the baby makes in carry-
ing things to its mouth. If the object is long, the baby’s brain learns to corre-
late the tactile sensations of the hand holding the object with the sensations of
the mouth. The essential element of this connection seems, to me, to be the
rigidity of the object, which introduces covariations of pressure between the
two zones (hand and mouth) costimulated by the object. The assumed rigidity
of objects is an important property that turns up again in the visual perception
of their curvature and shape. In this case, it is said that the brain formulates a
rigidity hypothesis. The brain makes assumptions about the world based on
which it constructs internal models of reality. In the case of extension through
a tool, over the course of development a simple correlation between two parts
of the body could become the simulation of a correlation between a part of
the body (the one that is holding the tool) and a point in space.
Egocentric and Allocentric Frames of Reference
The concept of reference frames excited the imagination of many researchers,

who devised different reference systems, just as geographers devised several
systems of projection to make maps of the world. Which frames of reference
does the brain use, and does it use them in the mathematical sense of the con-
cept?
It is possible to represent the position of objects in several ways. Suppose
that you wish to describe the relationships between two objects in a room (a
chair and a table) and your body. One way of encoding these relationships is
to relate everything to yourself, to estimate the distance and the angle of each
of these objects in relation to your body. This polar type of encoding is typi-
cally egocentric. The position of the two objects can also be encoded in Carte-
sian coordinates, using the two axes formed by the frontal and sagittal planes
of your body, say, or the perpendicular planes of the vertical semicircular ca-
nals, and so on. The frame of reference is still egocentric.
A second means of encoding spatial relationships is to use the relation-
ships between the objects themselves or relate them to a frame of reference
external to your body. Take as a reference point the door of the room and this
time evaluate the position of the chair and the table with respect to the door.
FRAMES OF REFERENCE •
99
There is no reference to your body. This encoding is called allocentric, cen-
tered elsewhere than on your body. The distinctive feature of this encoding is
that it moves the basic point of reference away from the center into the exter-
nal space. Often the term “exocentric” is used in place of “allocentric”; these
two terms have similar meanings.
The difference between these two types of frames of reference is impor-
tant. It seems that most animals are capable of egocentric encoding, but only
primates and humans are genuinely capable of allocentric encoding. The
power of the latter is that it enables mental manipulation as well as manipula-
tion of the relationships between objects without having to continuously re-
late them to the body. The ability to think about the distance between the vil-
lage bell tower and the police station, between the synagogue and the pastry
shop, to wonder whether one is larger than the other is what is distinctive
about the human brain and leads to speculation about geometric relation-
ships. Moreover, allocentric encoding is constant with respect to a person’s
own movement; thus it is well suited to internal mental simulation of dis-
placements. It appears rather late in children, who first relate space to their
own bodies.
One particular example of an egocentric frame of reference is our percep-
tion of the midsagittal planes of our bodies: that is, we perceive very easily
whether things are to our left or to our right. This perception is impaired in
patients with right parietal lobe lesions who exhibit a syndrome called “spatial
neglect.” The midsagittal plane shifts at the same time that they ignore the left
part of space, or the left part of their plate when they eat, or the left part of
their body. Using functional brain imaging in normal healthy people, my col-
leagues and I have recently shown that the perception of this midsagittal plane
involves activity in a pair of areas in the parietal and the frontal cortex; the
same areas involved in the neglect syndrome when they are lesioned.5 Other
data from Galati and his colleagues, soon to be published in Experimental Brain
Research, suggest that the allocentric frame of reference comprises in addition
an occipitotemporal system involving the parahippocampus, whose role in the
processing of spatial information has also recently been demonstrated.
Natural Frames of Reference
Gravity
The semicircular canals of the vestibular organ define a basic Euclidean frame
of reference that may be at the root of our geometric perception of space. By
100 •
its very structure, it provides a reference frame only for movements of the
body. It is an egocentric system.
Nature bequeathed us another frame of reference connected to external
space: gravity. From our perspective, this omnipresent force has several very
important features. First, it does not vary in magnitude or direction with re-
spect to a plane tangential to the ideal surface of the earth—it is a constant of
terrestrial space. Second, it can be detected by specialized receptors, the oto-
liths. Finally, it constitutes a reference point external to the body and conse-
quently, as Paillard says, “an external plumb line” related to bodily movements
in a frame of reference he calls “geocentric.”6
To demonstrate the importance of this second frame of reference, I will
examine the bodily motion of a person who is jumping in a complicated way,
or running, or doing moguls down a ski slope; or the way storks fly, lions run,
and so on. My examination will reveal that during what André Thomas calls
“surplus equilibrium” (l’équilibre de luxe) with regard to movements in sports
or in dance, the head is stabilized in rotation.
I discovered that this property also applies to humans during daily locomo-
tor movements such as running. I took some photographs that Muybridge had
made of people in motion,7 and I drew on the photographs the line that con-
nects the external canthus (the corner) of the eye to the meatus of the ear
(Figure 4.1). This line is approximately parallel to one of the planes of the
semicircular canals. Making a montage of the photographs superimposing the
meatus of the ear on all of them revealed the head to be remarkably stable
during rotation. Using video cameras connected to a computer, we showed
that during walking, rotation of the head is stabilized about positions deter-
mined by the direction of gaze. In other words, we are not unlike birds when
they fly or gazelles and ostriches when they run. If you look at them on film,
you will see that their heads remain perfectly stable in relation to the vertical.
Most likely, the otoliths estimate how much the head is tilted with respect to
gravity.8 The direction of gaze determines the plane of stabilization.
It is as if the brain creates a stabilized platform to coordinate movements
of the limbs. In these complex movements, the feet rarely touch the ground,
so much so that the earth’s surface cannot serve as a reference point. The
brain uses the gravity detected by the vestibular system to stabilize the head
and create a mobile platform as a frame of reference. The advantage of this
solution is that because they do not undergo rotation, the visual and vestibular
receptors are freed from the problem of gravito-inertial differentiation that I
discussed in Chapter 2. They can cooperate better to detect translations based
on the information supplied by optic flow. Engineers who must control the
101
Figure 4.1. In these drawings of a man running, the head is stabilized to control
posture and coordination of movements. Photographs by Muybridge were su-
perimposed at a single point of the head, the auditory meatus. While running,
the head is stabilized in rotation at an angle that depends on the direction of
gaze. This stabilization relies on the vestibular system, which detects the angle
of the head in relation to gravity. The straight line connects the canthus of the
eye with the meatus of the ear. It indicates roughly the plane of the horizontal
semicircular canal.
movement of satellites in space adopt the same principle. They attach a small
platform to the body of the satellite and stabilize its position with reference to
the stars.
Gravity intervenes in the organization of movements at specific times in
development. When its influence on young rats is altered, locomotion is sig-
nificantly delayed.9 Thus there is a period critical to motor function, around
ten days after birth, during which the nervous system needs gravity as a refer-
ence for organizing coordination of movements. This experiment supplies
additional evidence for the existence of critical periods such as that already
demonstrated in the development of the visual system.10 In this latter case,
proprioceptive information is important for specifying the directional sensitiv-
ity of the neurons of the visual cortex.
The Vertical Subjective as a Multisensory Construction

One of the clearest examples of the essentially multisensory character of per-
ception is that of perception of the vertical. All of us can close our eyes and
accurately indicate the direction of the terrestrial vertical; that is, the direction
of gravity. On the other hand, if we are in a lit room tilted in relation to the
102 •
true vertical, we might make a misjudgment due to our perception of the ver-
tical based on the visual cues in the room—what we call the visual vertical.
Thus vision contributes to our perception of the terrestrial vertical and may
override vestibular cues.
Several laboratories have demonstrated the influence of vision on percep-
tion of the vertical by constructing tilted rooms where the subject not only
perceives the room’s vertical as the true vertical and thus neglects in part the
information from the otoliths, but also unconsciously tilts his body to align it
with the visual vertical. If a subject is placed in a darkened room and if the
only thing illuminated is a rod tilted toward him, he will perceive the vertical
as intermediate between the gravitational vertical and the visual vertical (the
Aubert effect).11 Paradoxically, subjects are able to perceive the vertical tilt
more than the tilt of the rod (the Müller effect)12 when the angles are sig-
nificant. In consequence, perception of the vertical is the result of a multi-
sensory compromise. The brain receives messages from the visual vestibular
sensors of vision and proprioception and compares them with the intrinsic ref-
erences of the axial direction of the body.
Patients who have suffered vestibular lesions use static information from
the visual environment to work out a vertical. Astronauts also use visual cues
from the space station to compensate for the absence of gravity. The following
experiment, carried out by our laboratory in conjunction with the French na-
tional center for space research over the course of several space flights, estab-
lished the role of vision in the reorganization of sensory control of posture in
the absence of gravity.13 We asked an astronaut to maintain an upright pos-
ture. On earth, he could do this in an instant, but trying to do it aboard a space
vessel makes no sense because the vertical is no longer signaled by gravity. So
we asked him to hold himself erect, perpendicular to the floor of the station.
We used an experimental trick, a sort of shoebox lit from within and attached
to his head, covering his eyes. He thus had no visual information about the po-
sition of his body in space. Moreover, given that his body had mass but no
weight, the forces exerted on his feet had changed, such that he could no
longer rely on tactile pressure to evaluate the position of his body in relation
to the floor.
The first day of the flight, the subject leaned forward markedly but re-
sponded that he was perfectly upright and perpendicular with respect to the
space station. At the same time, the tonicity of the muscles that maintain the
upright position—the extensors and flexors of the ankles—was redistributed.
Instead of the tonic activation of the extensors that on earth keep us from fall-
ing forward, the flexors were tonically activated. The combined information
103
from all the proprioceptors was consequently insufficient to detect the angle
of the astronaut’s body: this angle was not calibrated, the body schema was
distorted, and tonic motor activity was asymmetrical. The subject remained
unaware of the distortion. Since he could maintain a perpendicular posture
without difficulty when the box was removed and his view of the station was
restored, vision was essential to this recalibration of proprioception. A few
days later, the same subject, placed in the same conditions, was able to hold
his posture perpendicular to the station. His brain had come up with new
mechanisms for calibrating proprioception.
The brain seeks to work out a reference; it extracts from the physical
world a relevant variable that simplifies the neural processing of sensory infor-
mation and that guides action. Action is tied to a frame of reference.
Receptors for Gravity in the Stomach

I have described how the brain uses a basic egocentric frame of reference, the
vestibular system, that enables it to refer to the vertical plumb line. But it
seems that it also makes use of another more mysterious reference point
called the idiotropic vector. This is an innate perception of the longitudinal
axis of the body in humans.
A subject is placed in an armchair that tilts laterally in relation to the ter-
restrial vertical. With his eyes closed, he is asked to indicate what he perceives
to be the direction of the terrestrial vertical. He is observed to be slightly tilted
with respect to the true direction of gravity. The direction he indicates is inter-
mediate between the axis of his body and that of the true vertical, showing
that the perceived vertical (often called the subjective vertical) is the resultant
of two verticals, that of gravity detected by the otoliths, and of the axis of the
body or idiotropic vector.
Mittelstaedt, who studied the mechanisms underlying the perception of
the gravitational vertical, argued for adding another sense to the existing
list.14 Indeed, he discovered that receptors located in the stomach also detect
gravity.15
Ferrier had already noted this possibility:
We have thus far considered the influence of tactile, optic and labyrin-
thine impressions on the functions of equilibration and co-ordination,
and it has been shown that the influence of each is capable of experi-
mental demonstration. Though these are apparently the main factors in
the general synaesthesis, the possible participation of other afferent fac-
104 •
tors in the general result is not absolutely excluded . . . But there appear
to me grounds for attributing some influence to visceral impressions. It
is well known that cats and other members of the family Felidae, in-
cluding animals which possess in a marked degree the faculty of equili-
bration, have in their mesentery relatively large numbers of Pacinian
corpuscles, which are specially adapted for transmitting pressure stimuli
to the sensory or afferent centres . . . It would seem, therefore, not im-
probable that the viscera are in relation with the centres of equilibra-
tion, and that they mutually affect each other. This is supported by the
phenomenon of a distressing form of dyspepsia, characterised by sud-
den attacks of giddiness, described by Trousseau under the name of ver-
tigo a stomacho laeso.16
Mittelstaedt recently concluded that this physiological system for somatic

detection of gravity might be located in the kidneys or in the vascular system.
Only additional experimentation can resolve this issue.
Next I will consider more generally the multisensory character of percep-
tion. The contribution of so many sensory receptors to the perception of the
vertical raises a theoretical question. There are two conflicting ways of ex-
plaining a single percept such as the subjective vertical. The first regards it as
the resultant of all the verticals signaled by the various sensory systems. This
is the accepted approach. Another holds that there is one or several brain
structures organized in such a way as to constitute an internal model of the
bodily vertical, or rather the axis of the body, and that this internal organiza-
tion, this “body schema” according to Gurfinkel’s idea,17 is modulated or al-
tered according to the configuration of receptors. Although you might con-
clude the two ideas are similar, they do not at all imply the same neural
organization. One proceeds from the periphery to the center (bottom-up orga-
nization), the other from the center to the periphery (top-down organization).
The Rotisserie Experiment

Vision is not the only sense that cooperates with the vestibular receptors in de-
termining the orientation and movement of the body in space. Touch plays a
fundamental role that is worth examining, for it reveals the richness of multi-
sensory interactions.
To begin with, skin plays a role in detecting asymmetry in the control
of posture. Rademaker established this function of cutaneous receptors. He
showed that when a dog—or a horse—is lying on its side, reflexes of vestibular
105
origin (righting reflexes) automatically trigger the coordination of motor ac-
tivity required for the animal to get back up. He also discovered that simply
pressing a board held parallel to the surface of the ground to the exposed
flank of the prone animal inhibited the righting reflex.18
Interpreting this observation is straightforward. Righting reflexes are trig-
gered by asymmetric impulses of vestibular origin and by the visual detection
of body tilt in relation to the visual vertical. But when a board is pressed to the
flank of the animal, the tactile information supplied by the pressure of the
ground on one side and of the board on the other gives the brain symmetrical
information. Apparently, the brain’s confidence in this tactile information (en-
gineers would call it gain, neurophysiologists synaptic weight) is sufficient to
block the righting reflex. This old experiment argues for my hypothesis that
the brain uses configurations of receptors to work out a perception and initi-
ate action.
Experiments in space have contributed further proof.19 On the ground, a
subject was positioned before a disk rotating in a vertical plane, which caused
a circular vection.20 A very slight tactile stimulation of his shoulder, which trig-
gered a tactile asymmetry, markedly increased the discrepancy between the
true vertical and the perceived vertical. In space, aboard the American space
shuttle, floating and holding on only to a bite bar with his teeth, the astronaut
felt a considerably stronger vection and had the impression that the vertical he
perceived was turning along with him. The absence of gravity and the lack of
contradiction between visual and tactile information increased the credibility,
for the brain, of bodily rotation.
Here, now, is the rotisserie experiment. A subject lying on a bed can dis-
cern which way is up without difficulty. Lackner had the idea of placing a sub-
ject in a machine developed during the 1950s by American air force research-
ers interested in how the otolithic organs of the vestibular system detect the
vertical.21 This machine looks like a rotisserie. Subjects are laid in it horizon-
tally, as if they were brochettes or chickens, and turned at a constant velocity
around a rotational axis perpendicular to terrestrial gravity.
When they are in the light, with their eyes open, subjects have no trouble
perceiving the direction of their rotation. If the light is switched off, they have
only vestibular and tactile information to rely on for detecting the axis of their
rotation. The semicircular canals are stimulated by the initial acceleration and
the otoliths by the sweep of the components of gravity in the planes of the re-
spective maculae of the utricle and the saccule. Then, because the velocity of
the rotation is constant, and the angular acceleration therefore nil, the re-
106 •
sponse of the semicircular canals subsides after about 20 seconds. Only detec-
tion by the otoliths of the angle in relation to gravity persists. Consequently,
during rotation at constant velocity, only the otoliths and the skin contribute
to the perception of the direction of the body with respect to the vertical.
In general, the subject correctly perceives the axis of rotation in the dark.
But Lackner demonstrated that just manipulating tactile cues is sufficient to
completely alter the perception of orientation of the body. If the subject’s feet
are suddenly pressed, he has the impression of tipping up and rotating in a
vertical position. If his buttocks are pressed, he feels as though he is sitting and
spinning in a chair. If his head is pressed, he experiences conical motion about
his head at the apex. In other words, despite the sweep of pressure on the side
of the body, the brain treats the local point of pressure as a point of reference
that determines the perceived center of rotation. In this case, the tactile cues
determine the frame of reference in which the rotation occurs.
Sometimes the postural context affects how the brain interprets tactile
cues. An experiment of Gurfinkel’s is a nice illustration of this property. Lay
your hand flat, palm up, on a table and ask someone to draw a “p” on your
palm with a pencil. Now put your hand behind your back and ask the person
to repeat the drawing. You will note that in the second case you perceive the
letter as a “b” and not as a “p.” In other words, the same sensation yields a dif-
ferent perception according to the orientation of the hand in relation to the
body.
The extremely discontinuous character of these alterations in perception
shows that the brain carries out genuine perceptual decisions based on the state
of several receptors, which we call a configuration of receptors.
Multiple Frames of Reference Connected to the Limbs?

Is it really necessary to have a single reference point for planning movement?
The points of reference used by the brain are flexible, and depend on move-
ment. When you lift your arm and point your finger at a target, the brain
simplifies the command in a remarkable way by monitoring only two kine-
matic variables: elevation and azimuth.22 The other aspects of movement are
controlled by rules of phase relations among the parts of the limbs. When you
move an object on a table, the angle made by your upper arm and forearm,
and by your arm and trunk covary: a single command suffices to execute the
two movements. This way of controlling movement assumes that the brain
uses a frame of reference. It might use, for example, the spatial vertical in rela-
tion to which the polar coordinates of elevation and azimuth are calculated.
107
Principal sulcus
PROPRIOCEPTION
Memory
space
VESTIBULAR
RECEPTORS
Frontal eye
field
Posterior
parietal cortex VISION
(7a, 7b, LIP, VIP,...)
Oculomotor
space
Ventral
premotor Superior
cortex colliculus
Parahippocampal
Body-part-centered Putamen
cortex Oculomotor
visuomotor space
space
Entorhinal
Body-part-centered cortex
visuomotor
space
Hippocampus
Environmental space
Figure 4.2. The brain uses multiple frames of reference. The information sup-
plied by the sensory receptors (proprioception, vestibular receptors, vision)
converges in the parietal cortex, where it is integrated with many other signals
about movements and planned actions. Actions are subsequently encoded in
varied frames of reference that correspond to multiple spaces relative to the
body or the environment, even to internal memory space.
108 •
But during hand-pointing tasks the brain also uses a point of rotation lo-
cated on the shoulder. In this way it constructs a local reference point to sim-
plify neurocomputation. The brain chooses a frame of reference connected to
the limb that is executing the movement. Yet again, the advantage is that it re-
duces the number of variables that need to be monitored (engineers say de-
grees of freedom), limiting the need for neurocomputation to the single part
of the body involved in the movement.
The idea that the brain can choose multiple frames of reference depending
on the task and the context is shown by the neurons discovered by Graziano
and Gross that I described in Chapter 3. Recall that they found bimodal neu-
rons in the putamen and parietal cortex whose visual and tactile receptor
fields remain anchored in space during the course of a movement. One inter-
pretation of these findings is that the spatial encoding of movement and the
position of the limbs is not experienced with respect to a single reference
frame, as the egocentric frame of reference might suggest. Instead, there are
many frames of reference, each related to a part of the body (eyes, limbs,
trunk, and so on) (Figure 4.2). Action appears to be organized in modules
based on a repertoire of specific types of actions (gaze, reaching, locomotion),
which are themselves organized by specific basal ganglia-thalamo-cortical
loops.
The convergence of sensory information is the result of integrating data
from the senses important for each limb. For example, neurons encode the po-
sition of the head based on convergence of visual cues and other cues from
the muscles of the neck that detect movement of the head on the trunk,
as well as cues from the vestibular receptors, which also detect movement of
the head.
Selecting Frames of Reference
All of these findings suggest that the brain uses not one but multiple frames of
reference according to the task at hand together with essential or available sen-
sory cues.
Perrett and co-workers showed that in the monkey, the neurons of the
temporal cortex are involved in recognizing faces.23 The retinal image of the
face is first broken down into fragments corresponding to the principal chan-
nels of the primary visual pathways that dissociate color, shape, movement,
and so on; then other neurons reconstruct the facial features. Certain of these
neurons are activated by the eyes, others by the hair, still others by the nose,
109
and so on. These features then converge on neurons that respond to faces, but
which are also sensitive to behavior, such as the direction of gaze. This ability
to detect the direction of gaze is all-important to the monkey; it probably
helps to identify the intention of another of its kind: is it friend or foe?
In this chain of processing, the meaning of a facial image is gradually re-
constructed.24 In fact, there are neurons that detect whether the face is turned
sideways or facing front, and so on. At the next level, the face is situated in the
context of the body, and neurons fire when the face is associated with a body
facing front or turned sideways, and so on. Finally, other neurons fire when
the face is familiar: a cognitive level is reached that brings memory into play.
Up to that point, even in context, the face is processed in a frame of reference
centered on the observer. The multiplicity of neurons that respond to the
same face, whether it is viewed from different sides or from a different angle,
enable the brain to configure the activity of neurons that will respond to a
given face whatever the angle from which it is seen. This flexibility implies a
level of abstraction that is independent of the frame of reference in which the
visage is perceived.
It is also possible to imagine that frames of reference are constructed in
connection with action—for example, linked to an object, a goal, and so on.
The multiplicity of possible representations enables the brain to construct ad
hoc frames of reference. This hypothesis has yet to be verified, but will be ex-
traordinarily fruitful if it turns out to be correct. It would explain in part our
ability to do geometry, which requires us to mentally change our perspective
as we view tangible objects and the environment. For Kosslyn and his collabo-
rators, categorical spatial relations (front, back, above, and so on), which main-
tain constancy of certain relationships between objects or parts of the body,
are encoded by different mechanisms of metric spatial relations that specify
distances.25
Relative Frames of Reference

Frames of reference are not only useful for establishing relationships between
the body and space. Arbib formulated the idea of “opposition space” between
the thumb and the other fingers in grasping tasks (Figure 4.3).26 He rejects the
idea of a single basic frame of reference to which all other frames of reference
refer. He writes: “The task of perception is not to reconstruct a Euclidean model
of the surroundings. Perception is action-oriented, combining current stimuli
and stored knowledge to determine a course of action appropriate to the task
at hand.”27 Building on the findings of Jeannerod, who demonstrated that the
hand preshapes to the size and orientation of the object it is about to grasp at
110 •
Visual input
Recognition Position
criteria
Size Orientation
Activation of
visual search
Visual
Visual location location
of target
Size Orientation
Target recognition recognition
location
Orientation
Visual and Size Visual and
tactile tactile
Activation of input input
reaching
Ballistic Finger Hand

Reaching and movement adjustment rotation
grasping of target
Hand
reaching
Adjustment Grasping
Figure 4.3. How to grasp an object. The brain analyzes visual inputs with re-
spect to three properties: position, size, and orientation. Similarly, neural pro-
cesses analyze movement of the hand close to an object, adjustment of the
fingers to the size of the object, and a rotation of the pincer motion depending
on the orientation of the object.
the instant the movement is begun,28 Arbib suggests that the frame of refer-
ence used by the brain is relative, particularly the relationships between the
fingers (opposition space) that are about to grasp the object: “There is no one
absolute space represented in one place in the brain, only a coupling of sen-
sory and motor spaces in such a way as to yield movement to achieve some
goal.”29
A point of reference can thus be constructed on the basis of relationships
between parts of the body. It can also be constructed actively. For example, I
said earlier that the head is used as an inertial platform that is stabilized during
complex movements and especially when there is no reference to the ground.
In this case, the brain uses gravity to stabilize the head, which can then serve
as a point of reference for coordinating the limbs. I have emphasized that a
movement’s point of reference depends on the task.30 If I am holding a full
111
glass of champagne while walking or bending over, I necessarily have to adjust
its movement to the direction of gravity so I do not spill it. But if I am reading
a book while walking, the objective of my movements is to minimize the rela-
tive movement between the page and my retina, and thus I adjust the move-
ment of my hand to that of my head.
It also must be considered that the brain constructs a series of frames of
reference for each phase of the same movement.31 This is illustrated by the
trampoline jumping shown in Figure 3.3. A new physiology is needed; one
that will investigate these rapid changes, these swings from one frame of refer-
ence to another. This flexibility is so basic that it must be programmed into
the very structure of the nervous system.
Explicit and Implicit Frames of Reference: The Theory of the

Equilibrium Point
Little is known about the way the nervous system processes frames of refer-
ence, but the ideas derived from Bernstein’s so-called equilibrium point model
deserve mention. Feldman,32 in Moscow, recently took up this theory again
and gave it a formal expression that was in turn elaborated by several Western
groups, including Bizzi’s at MIT. This theory is interesting because it is at the
heart of the question of anticipation and of the way the brain implicitly or ex-
plicitly simulates external reality.
Feldman’s model emerged from investigations of the functioning of mo-
tor neurons (the neurons that control muscle contraction). It is well known
that the discharge of a motor neuron is connected to its membrane potential.
Normally, this membrane is hyperpolarized (negative potential inside with re-
spect to the outside). A central command or the influence of sensory recep-
tors can depolarize the neuron and bring it to its firing threshold. Thus, the
motor neuron produces action potentials that lead to muscle contraction.
Classical models of control of movement (called “α models”) assume that
a central command directly regulates discharge of motor neurons. But Feld-
man reasons that the brain regulates an indirect variable, the firing threshold of
the motor neuron. This could be defined as the value of the membrane potential
at which the neuron fires. Why is this interesting? Because it also enables us to
control the position of the arm. How? If my arm is in a given position, nor-
mally the neuromuscular spindles fire and activate the muscle, which con-
tracts reflexively. But if I change the firing threshold so that, for this length of
muscle, the motor neuron—even if it is activated by the sensory receptors—is
no longer activated, my arm does not move. The λ variable that controls the
112 •
threshold is thus equivalent to a position, an angle of the arm. Feldman says
that it has a spatial dimension, and that the position is reached by regulating
the λ variables of the two antagonist muscles. According to this theory, there
is no need for space to be represented explicitly, for it is implicitly contained in
the regulation of the threshold. This fits the theories of Viviani and Flash on
the morphogenetic and creative properties of geometric trajectories and the
law of minimum jerk (Chapter 6).
Feldman shows that although his hypothesis is based on neuronal func-
tioning, it has an impact on the general problem of the frames of reference ac-
cording to which movements are organized.
These theoretical results can be understood in terms of physics. Body
movement is defined as a change in position in relation to another ob-
ject, frame of reference, or system of coordinates. But the Galilean
principle of the relativity of movement is implicit in the concept of
frame of reference: movement can also be induced by a displacement
of the frame of reference. The λ threshold can be considered as the
point of origin of a frame of reference for the mobilization of mo-
tor neurons. In displacing λ, the level of control specifies a new refer-
ence point and produces movement. In this way, displacements of the
positional frame of reference underlie the control of movement even
though activation of motor neurons and production of forces are a con-
sequence of this process.33
The scientific community is actually divided on the validity of the λ

model. However, the model demonstrates the plausibility of theories about
the control of movement that reject the idea that space is explicitly repre-
sented in the brain, even if it is consciously perceived that way.
Ideas about frames of reference have evolved from a monolithic conceptu-
alization of a single point of reference to the notion of flexible frames of ref-
erence connected to the task at hand. An illustration of this evolution can be
found in the remarkable fountain sculpted by Bernini—the personification of
the four rivers—in the Piazza Navona in Rome. The sculptor shifted the foun-
tain completely off center in relation to its central column by creating extraor-
dinarily varied frames of reference for the action of his figures. The frames of
reference refer to a universe that is relative for each of the characters.
So the problem is more general and very basic: it is the nature of the ego.
Indeed, if there is no single egocentric frame of reference but instead multiple
representations of the body, it becomes necessary to construct a theory that
113
explains the unity of perception of the body. An explanation of how these di-
verse local neuronal subsystems are integrated in a body schema is needed.
The situation of this science is a bit like the global political situation of the
1990s. In the wake of ideologies aimed at imposing a single model, there has
been a veritable explosion of nationalisms. There is no single frame of refer-
ence, but rather a fascination for the multiplicity of cultures. How will we
manage to put together from this a world that can account for both individual
differences and what might be termed the unity of man? This is the same
mammoth challenge facing the neurosciences.
114 •
5
A M E M O RY F O R P R E D I C T I N G
Perceived in its nascent state, a movement is always movement that is going

somewhere, which is absurd for the physicist, who defines the movement not
by its objective, but rather by what brought it about. The movement as it is
perceived, however, proceeds from its point of arrival to its point of departure.
—M. Merleau-Ponty
Perception is essentially multisensory: it uses multiple, labile frames of refer-

ence adapted to the task at hand. It is predictive; receptors detect derivatives,
and the brain contains a library of prototype shapes of faces, objects, and per-
haps movements and synergies. Nature has devised simplifying laws among
the geometric, kinematic, and dynamic properties of natural movements. But
the predictive nature of perception is also—perhaps especially—due to mem-
ory. For memory is used primarily to predict the consequences of future ac-
tion by recalling those of past action.1
What are the recent findings on memory from psychology, neurophysiol-
ogy, and neuropsychology that will best identify memory’s role in the relation-
ship between perception and action? Modern neuropsychology distinguishes
several types of memory that are implemented in different parts of the brain
(Figure 5.1). Memory can be declarative, implicit, working, episodic, proce-
dural, short-term, long-term, iconic, topographic, spatial, semantic, lexical,
motor, and so on. Neuropsychologists and neurophysiologists have made re-
markable strides in understanding the neural organization of multiple facets
of memory. Memory is a basic property of the brain that is expressed at all
levels of cerebral functioning. Even muscular fibers have a memory; some
have so-called catch properties that allow them to remain contracted for sev-
MEMORY
DECLARATIVE NONDECLARATIVE
(EXPLICIT) (IMPLICIT)
NONASSOCIATIVE
LEARNING
FACTS EVENTS “PRIMING”
SKILLS AND SIMPLE

HABITS CLASSICAL
CONDITIONING
Figure 5.1. Classification of the components of memory. Declarative (explicit)
memory is memory of facts and events, and it is a function of the temporal
lobe of the cerebral cortex. Nondeclarative (implicit) memory involves a reper-
toire of skills independent of the temporal lobe. Nonassociative learning con-
cerns memory that suppresses sensitivity to a stimulus when it is repeated (ha-
bituation) or, on the contrary, is maintained and increased. “Priming” refers to
the change in response observed when a subject is supplied a clue (which he
holds in memory for a brief moment) to what the stimulus will be just before it
is applied. This schematic diagram does not distinguish between short-term
and long-term memory, nor does it include the concept of working memory
developed by Baddeley.
eral moments in certain cases. The motor neurons of the medulla possess
what are called plateau potentials that maintain the depolarization of their
membranes. Other mechanisms at the cellular level, such as long-term poten-
tiation, which is most likely responsible for properties of memorization in the
hippocampus and the cortex, or long-term depression, discovered in the cere-
bellum by Ito and probably involved in plasticity, or even the slow potentials
due to calcium ions, are the elementary manifestations of memory, controlled
by molecular mechanisms among which the NMDA (N-methyl-d-aspartate)
synaptic receptor plays a major role. The phenomenon of the neural integra-
tor, which we discussed earlier with respect to the vestibulo-ocular reflex and
which transforms velocity signals into positional signals, is a specific case of
memory arising from the intrinsic properties of neurons or resulting from re-
verberating circuits in which a collateral signal is returned (positive feedback).
Other mechanisms call into play activities supported by oscillations of internal
neural circuits.
The discovery of the neural basis of these memories will in fact be one of
the major issues for neurobiology in the next century. Here I will consider just
one particular form of memory—spatial memory—which has not been exten-
sively studied by neuropsychologists but which nevertheless constitutes an im-
116 •
portant model for understanding the relation between perception and move-
ment.
Spatial memory is essential in representing space. Paillard defined its role
very clearly:2
A basic assumption of our argument is that a sensorimotor mode of pro-

cessing spatial information coexists with a representational mode and
that both modes are generating and storing their own mapping of space
. . . The sensorimotor mode concerns mainly that part of the physical
world to which the organism is attuned by virtue of its basic sensori-
motor apparatus. Local sensorimotor instruments entertain direct dia-
logues with that world and thus contribute to the continuous updating
of a body-centred mapping of extra-corpor[e]al space where objects are
located and to which actions are directed . . . The representational mode
derives from neural activities which explore and consult internal repre-
sentations of the physical environment, that are embodied in memory
stores. They include mental representations of local maps, spatial rela-
tionships of routes relative to landmarks, relative positions between
objects, and the position of the body itself in relation to its stationary
environmental frame. The question arises as to whether the two pro-
cessing modes operate in parallel, each using its own neutral [sic] cir-
cuitry and generating its own mapping of space in two fundamentally
different ways.3
Topographic Memory or Topokinetic Memory?
Navigational Memory
Suppose that I ask you to recall the trip you make to get to work. You will
have to remember both the way, that is, the topographic aspects of the route,
and the movements you make walking or driving.
Topographic memory allows us to locate a place and to find our way back
to it. In a sense, it is a form of procedural memory, because it involves a suc-
cession of places or local scenes and movements (turning left, right, and so
on). But it can also be a mental survey of the places that represent the trip as
on a geographic map. It is believed that this kind of memory is responsible for
anamorphoses, or distorted images. For example, if a subject is asked to draw
A MEMORY FOR PREDICTING •

117
a map of Paris with his eyes closed, he has a tendency to make the center large
and the periphery small.
Among these navigational abilities, there is one that shows up in all species
of animals, from the fly to humans, and that is dead reckoning or, more gener-
ally, the ability to find one’s way back. Darwin was interested in this function
of navigation. He wrote:
With regard to the question of the means by which animals find their
way home from a long distance, a striking account, in relation to man,
will be found in the English translation of the Expedition to North Sibe-
ria, by Von Wrangell . . . He there describes the wonderful manner in
which the natives kept a true course towards a particular spot, whilst
passing for a long distance through hummocky ice, with incessant
changes of direction, and with no guide in the heavens or on the frozen
sea. He states (but I quote only from memory of many years standing)
that he, an experienced surveyor, and using a compass, failed to do that
which these savages easily effected. Yet no one will suppose that they
possessed any special sense which is quite absent in us. We must bear in
mind that neither a compass, nor the north star, nor any other such
sign, suffices to guide a man to a particular spot through an intricate
country, or through hummocky ice, when many deviations from a
straight course are inevitable, unless the deviations are allowed for, or a
sort of “dead reckoning” is kept. All men are able to do this in a greater
or lesser degree, and the natives of Siberia apparently to a wonderful
extent, though probably in an unconscious manner. This is effected
chiefly, no doubt, by eyesight, but partly, perhaps, by the sense of mus-
cular movement, in the same manner as a man with his eyes blinded
can proceed (and some men much better than others) for a short dis-
tance in a nearly straight line, or turn at right angles, or back again. The
manner in which the sense of direction is sometimes suddenly disar-
ranged in very old and feeble persons and the feeling of strong distress
which, as I know, has been experienced by persons when they have sud-
denly found out that they have been proceeding in a wholly unexpected
and wrong direction, leads to the suspicion that some part of the brain
is specialised for the function of direction.4
Darwin’s text was prophetic. As I noted in Chapter 3, neurons have an ac-

tivity connected to the orientation of the head in space. Since Darwin’s time, a
large number of publications have been devoted to mechanisms of navigation.
118 •
However, the literature is less than clear on what we mean by the term “navi-
gation,” and it is worth trying to clarify it here.
By navigation we mean the ability to find one’s way at sea and, by exten-
sion, on land and in the air. Navigation has been studied in connection with
the flight of pigeons and bees, reproduction in salmon, the migration of Es-
kimo peoples in the Antarctic and, more recently, the voyages of Micronesian
sailors, who traveled to far islands without instruments, using only cognitive
maps and techniques for updating their position in relation to virtual land-
marks.5 It is important to make a distinction between far space that can be per-
ceived directly (in a stadium or in coastal navigation, for example) and very
great distances, which involve their own particular representation (as for air-
planes and migrating birds). The ability to find one’s way home or to memo-
rize a route is not unique to humans; crabs return to the sea, bees to the hive,
and desert ants to their nest using the sun to find their way. Each species has
devised individual solutions to the same problem.
Inertial Navigation
Darwin’s account in 1873 aroused very interesting reactions at the time. To ex-
plain this navigational facility, some people proposed a mechanical analogy
very close to the inertial navigation systems used on modern airplanes.6 Per-
haps the vestibular apparatus of animals served as an inertial detection system
for navigating. Or perhaps the vestibular apparatus enabled the brain to di-
rectly detect the Coriolis forces and to use this information in navigating. It
was even proposed that the Coriolis effect could be estimated by detecting the
movement of blood in the vascular bed, a very controversial idea.
The assumption that the vestibular system plays an important role in navi-
gating short distances was taken up by Beritoff, an Armenian physicist who
also went by the name of Beritashvili. He elucidated these processes in dogs,
cats, infants, and adult humans. He suggested that spatial orientation is the ap-
titude of organisms to locate the position of objects and to relate this position
to themselves and to other objects.7 From these perceptions, organisms form
images of the distribution of objects in the world around them. These images
initiate and guide movement oriented toward the objects, even if the objects
are not seen or perceived by other sensory organs.
Beritoff conducted experiments with dogs. He blindfolded them before
taking them into the laboratory from one side and leading them along a spe-
cific path to a food source located on the other side. Then he led them back to
the point of departure before moving them to other parts of the room. He
discovered that even blindfolded, the dogs had no trouble finding the part of
119
the room where the food was located, whatever their starting point. He inves-
tigated the possible role of all of the sensory cues (auditory, olfactory, tactile)
but found performance of the task to be impaired only after bilateral laby-
rinthectomy. Moreover, the animals found their way to the food just as well
when they were carried to it during training as when they walked to it. These
results seem to rule out kinesthesia (the set of cues supplied by the proprio-
ceptors of the joint muscles and locomotor commands). Though he realized
that the cerebral mechanisms called into play were extremely complex, Beri-
toff was the first to demonstrate a possible role of the vestibular system in
dead reckoning.
A precise return to a point of departure in the dark and deterioration of
performance following vestibular lesions have been observed in the rat.8 Sup-
plementary proof of this contribution from the vestibular system came from
work on batrachians and desert mice9 as well as the golden hamster.10 In this
animal, a lesion of the vestibular system causes deficits in navigational tasks
that are reversible following recovery of vestibular function.11 However, de-
tecting linear movement using vestibular cues has been the subject of a major
debate, for homing experiments with rodents suggest that these animals do
not correct linear displacements or rotations as easily. There seems to be a dis-
sociation between integration of linear and angular trajectories.12
Rotational Memory
The brain thus possesses mechanisms for memorizing displacements. Does
the vestibular system really contribute to these mechanisms, as Beritoff sug-
gested? One analytical approach separates out the components of rotational
and translational displacements and asks the question: Can the brain memo-
rize rotatory displacements and translations and reproduce them using the
eyes, head, or whole-body movements? This is the question we sought to an-
swer. I will describe these experiments in detail, because they led us to a fun-
damental concept: that vestibular memory is memory of movement, not of
position.
One elegant experiment, called vestibular memory-contingent saccades,
demonstrates the ability of the human brain to accurately detect the angle of
a rotation from the acceleration signaled by the vestibular receptors.13 The
subject is seated in a revolving chair in front of two visual targets. One target
is fixed in relation to the earth, the other in relation to his head. He is asked to
look at the first target and to memorize its location in space. Then all the
lights are switched off, except for the light on the target mounted on his head,
which turns with him and keeps his eyes from moving during the rotation.
120 •
The subject thus has only vestibular cues about the displacement of his body
to rely on, since he is in the dark and can make no movement of his eyes
(which could otherwise inform his brain about his displacement by means of
extraocular proprioception and motor discharge).
When the revolving chair is brought to a standstill, the illuminated target
mounted on the subject’s head is turned off, and he is now completely in the
dark. Several seconds are allowed to pass for the vestibular effects to diminish,
after which the subject is asked to recall the position of the earth-fixed target
and to make an ocular saccade toward that spot. In other words, he is asked to
displace his gaze from the place where the chair has stopped to the point of
departure he memorized. This task amounts to having the subject make a vi-
sual oculomotor return to his point of departure. After this memory saccade,
the light is turned back on. If the subject’s gaze is on the target, his eye does
not move; if his gaze is slightly deviated, he will execute a little correcting
saccade that makes it very easy to estimate the error.
Normal subjects can make saccades toward memorized targets after rota-
tion in the dark with extraordinary accuracy.14 The brain is thus capable of es-
timating displacement of the head solely from information supplied by the
horizontal semicircular canals. The subject is asked to wait 4 or 5 minutes
in total darkness after the chair has stopped moving before executing the
saccade. Despite this delay—which requires that the subject memorize the
spatial information about the angle of rotation of the body, that the brain re-
construct this information, and that it activate the cortical centers that pro-
duce the saccade—the result is always excellent. In other words, vestibular in-
formation about displacements of the body can be stored in spatial memory.
This is true for all orientations in space, which is only natural, since the vestib-
ular system did not develop solely to execute horizontal movements but to en-
able complex movements, such as jumping from tree to tree.
Next a subject is placed on a turntable and rotated in total darkness. We
ask him to reproduce the angular displacement in the opposite direction,
which he can do using a lever that controls the movement of the table. A
healthy subject can return to his point of departure with great precision.15
Which areas of the brain are involved in this ability to estimate rotations?
Applying this test to patients with lesions of the cerebral cortex gives a partial
answer.16 Serious impairment has been observed in patients with damage to
the supplementary oculomotor field, the prefrontal cortex, and the vestibular
cortex. These findings suggest first and foremost that the prefrontal cortex is
involved not only in visual memory but also in vestibular memory. This is no
surprise, since in my opinion memory of space is memory of movement in
121
space: thus it is essentially multisensory. Moreover, it seems to me perfectly
natural to find deficits in patients with lesions of the vestibular cortex, given
its importance in reconstructing movements of the head in space and its im-
portance for transmitting to the rest of the cortex information about displace-
ments of vestibular and tactile origin (see Chapter 3).
Translational Memory
Man’s aptitude for using otolithic vestibular cues to estimate translations in
the course of linear movements was recently established. Recall that the oto-
liths are sensors of linear acceleration in their planes. We placed the subject on
a trolley for translational movement.17 This sort of experimental sled was con-
structed in our laboratory at the Ecole Nationale Supérieure des Arts et
Métiers during the 1970s. It used a linear motor technology initially conceived
to control the movement of high-speed trains, but eventually abandoned ow-
ing to the dissipation of heat that occurs with this mode of propulsion. The
trolley made displacements of several meters with accelerations of up to
about 1 meter per second per second. The subject was seated transversally, his
displacement thus occurring in the frontal plane (that is, the plane of the two
ears). Before beginning the move, the subject was shown a visual target fixed
with respect to the earth—in this case a clothespin hanging on a line. Then the
light was switched off, and the trolley was moved 60 centimeters to 1 meter.
The subject’s eyes did not move during the translation. Once the trolley had
stopped, the light was switched back on, the effect of the accelerations on the
otoliths was given a few seconds to dissipate, and the subject was asked—as
for the rotations—to make a saccade to where he remembered the clothespin
to be. The movements of his eyes adapted perfectly, whether his head was
fixed or free. Additional support for these findings was provided by experi-
ments made with the SLED (Figure 5.2) installed at the time in a glass shell at
the Cité des Sciences in Paris. The subject was displaced in an anteroposterior
direction, facing the movement. He was asked to signal his crossings under
targets positioned all along the linear track.
Later, we adapted a mobile robot to study perception of displacements.18
The subject was seated on the robot, in total darkness, and was subjected to
noise to obliterate any acoustical cues resulting from his displacement. The ro-
bot followed a rectilinear trajectory whose profile was either triangular (accel-
eration, then deceleration), trapezoidal (acceleration, constant velocity, then
deceleration), or square (sudden acceleration, constant velocity, then sudden
deceleration). Once the passive displacement—1 to 10 meters—was com-
122 •
Figure 5.2. The SLED in the Spacelab laboratory. This space sled was used in
the American space shuttle to study the influence of microgravity on the func-
tioning of the otolithic receptors. It consists of a runway approximately 4 me-
ters long on which a trolley moves by means of wires connected to a motor.
The subject is placed in any of three positions (front, lying on his back, or cross-
wise) during translations. A digital camera, constructed by LETI (part of the
French Atomic Energy Authority), records movement of one eye, and a minia-
ture television screen projects visual scenes to the other eye, which stimulates
the optokinetic reflex. Eye movements are also recorded with electro-
oculography. The subject uses a lever to indicate which way he thinks he is
moving. A Pelletier effect system allows injection of hot air into the subject’s
ears to study the effect of microgravity on the vestibulo-ocular reflex. This ap-
paratus was constructed by the European Space Agency. Our laboratory partici-
pated in its use with the French National Center for Space Research.
pleted, the subject was asked to reproduce the displacement, from memory,
with the aid of a lever that controlled the velocity of the robot. All normal
subjects executed this task accurately.
How does the brain go about this reproduction? Does it calculate the dis-
placement in meters and memorize numeric values, or does it perhaps store
other variables? We assume that it does not calculate displacements but in-
stead reproduces a dynamic velocity profile. It simulates the movement car-
ried out. Recall the words of Poincaré: “To localize an object simply means to
represent to oneself the movements that would be necessary to reach it.” In-

123
deed, when different velocity profiles were imposed (triangular, trapezoidal,
square), the subject tended to reproduce these profiles and not a standard
profile that allowed him to optimally detect the displacement.
Which cues does the brain select among the available sensory data? A
paraplegic subject managed a perfect performance despite the lesion to his spi-
nal medulla that left only his head sensitive to muscular and tactile sensations.
For him, the vestibular information seemed to be determinative. However,
subjects presenting with bilateral severance of the vestibular nerves were not
able to reproduce the velocity profile as accurately, though they succeeded in
carrying out the task all the same, probably by using other combinations of in-
formation about or estimation of the displacement. The brain devises many
ways of solving the same problem.
So the brain memorizes movement, not just places. Paillard speaks of ve-
locity space and positional space. Reproducing a distance requires comparing
an internal simulation of a memorized movement and information from the
senses. The brain is a comparator that detects discrepancies between its own
predictions based on the past and the information it selects about the world
according to its purpose.
Locomotor Memory
Thus the role of the vestibular system in the memory of passively experienced
displacements becomes apparent. What happens to this memory during self-
displacement while walking? In this case, the information supplied by proprio-
ception in the legs—length of step, signals of the motor commands—is very
important for evaluating distances.
Anyone can do the following experiment, credited to Thomson. Look at a
point on the ground about a dozen meters away; close your eyes and walk to-
ward the point. You will see that you arrive at your destination pretty exactly.
Another test is to look at a target point far away on the ground and walk there
with your eyes closed but in a roundabout way. You will again observe that
you reach your goal with good accuracy. You can also return to your point of
departure with your eyes closed (what ethologists call dead reckoning). Finally,
you can repeat your same route several times, always with your eyes closed,
which means that the information has been stored in your spatial memory.
How do we explain this? There are only three possibilities: the distance
covered may be calculated from proprioceptive data supplied by the muscles
and joints of the legs; it may be derived from the memory of the motor com-
mands for each step; and finally, it may be derived from inertial cues of vestib-
ular origin. For the last of these possibilities, the brain would have to calculate
124 •
the distance covered based on vestibular information about acceleration, a
process that is called path integration. Poincaré had suspected this possibility:
“Knowing the acceleration of rotational movements of the head at each in-
stant, we can deduce through unconscious integration the final orientation of
the head with respect to its initial orientation.”19 According to this theory, the
brain calculates a distance covered. But we suggest that it dynamically updates
spatial representations.20 I have proposed the term “topokinetic memory” or
“topokinesthetic memory” to indicate this dynamic type of spatial memory.
I return to Thomson’s experiment described above. There are three ways
of interpreting his results: Perhaps the brain has a memory like a map (in the
geographical sense) of the environment and calculates its position on the map,
or perhaps the distance is encoded in “action units,” to use the expression fa-
vored among Gibson’s psychologist disciples; that is, the number of steps,
without there having to be a cartographic representation of the route in the
brain. Perhaps a memory of movement (rotations, translations, and their com-
binations) is retained and detected by the sensory receptors.
When the target is situated 3 or 4 meters away, people who no longer have
the function of their vestibular receptors still carry out the task, making only a
few mistakes in selecting a path. Recently we reproduced this experiment by
asking subjects to reach a target spot over a triangular path with their eyes
closed. Each side of the triangle had a length of several meters. In every case,
subjects who had no labyrinth, or who had unilateral labyrinthic lesions, dem-
onstrated significantly impaired performance. So it is plausible that the aggre-
gate of sensory messages contributes to this internal updating of spatial repre-
sentations. Recent work by Reiser, Loomis, and Tresky in the United States, as
well as by Thynus-Blanc in Marseille, shows that there are several possible
mental strategies for updating.
The memory of displacements is probably a genuine dynamic memory
that, when invoked, induces internal simulation of the path. The brain does
not merely compare sensory information with memorized information, it also
calls into play anticipatory mechanisms.
Recent observations confirm the anticipatory character of cerebral activity
during navigational tasks. For example, 1 or 2 seconds before every bend, the
gaze of a driver on a mountain road becomes fixed on the tangent inside the
bend.21 A calculation based on these data shows that the direction of this point
in relation to that of the car makes it possible to predict the curvature of the
road beyond the bend. In other words, the gaze of the driver is positioned on a
point such that the information supplied by the optic flow allows him to pre-
dict the curvature of the trajectory. Thus the brain does not merely know the
125
curvature of displacement of the car at a given moment based on visual and
vestibular cues: it seeks to predict the curvature at a future time.
Similarly, when we turn a street corner, our gaze anticipates the rotation
of our body. This anticipation appears in the infant in the course of its devel-
opment, and I think it must be absent in patients with lesions of the parts of
the brain that participate in anticipation.
It thus seems reasonable to suppose that in carrying out a navigational
task, the brain in a way follows an internal representation, a model of the tra-
jectory that anticipates the path, and not the other way around. This principle
could be very useful in robotics. Instead of building machines that use so-
called sensory receptors to guide their movements, we recently suggested to a
team of roboticists that they control movements using gaze to anticipate the
trajectory. This principle of guided navigation turns out to be much more in-
teresting than that based on passive detection of displacement. Here, as in
other tasks, the brain prefers a “go where you look” strategy, in which the ob-
ject guiding a control strategy is placed close to the center of vision. I think
that these anticipatory mechanisms reflect the fact that the brain carries out
tasks of guided movement based on mental paths that it constructs and that
enable it to make predictions. Navigation is no more than the execution of an
internal plan based on past experience, and the senses are used, as in the case
of the ski champion I mentioned earlier, to ensure that the plan unfolds and to
make corrections.
The Neural Basis of Spatial Memory:

The Role of the Hippocampus
What are the neural mechanisms of navigational memory for re-finding one’s
way in a city? The beginning of an answer to such a question may be found by
turning first to neuropsychology to discover which lesions of the brain cause
impairment of navigation or what is called place memory or topographic
memory.
Although several neurologists since J. Hughlings Jackson have noted dis-
turbances in locating one’s position and navigating in familiar places, it was
Milner who established the basic role of the hippocampus in spatial memory22
(Figure 5.3) based on study over many years of an epileptic patient whose two
hippocampi had been removed.23 Generally, studies of lesions introduced in
the rat and the monkey and sustained by humans following vascular injury or
surgical excision suggest that the hippocampal region plays a determinative
role in spatial memory. However, a recent reassessment of these lesions and
126 •
ENTORHINAL
CORTEX
DENTATE
GYRUS
GRANULAR
CELLS
CA1
GIANT
PYRAMIDAL NEURONS
CA2 C4
CA3
Figure 5.3. Anatomy of the hippocampus.
their effects indicated that the deficits induced might also be due to a small
neighboring region that could have been damaged at the same time as the hip-
pocampus—the entorhinal cortex. Nevertheless, several different conceptual-
izations compete with one another. They are worth summarizing briefly, be-
cause they explain why I maintain that the brain uses memory to predict the
consequences of action.
Short-term Memory
In humans and in animals, damage to the hippocampus and neighboring re-
gions of the temporal lobe impairs memory; so any plausible explanation of
this phenomenon must be compatible with the physiological properties of the
hippocampus. Marr was one of the first to propose that the hippocampus
plays a role in short-term memory.24 He was interested, among other things, in
its possible role in the genesis of dreams. He posited that during paradoxical
sleep (during the dream phase), the hippocampus fires stored memories at the
parietal cortex. Later, it was hypothesized that the hippocampus contributes
127
to medium-term memory. Mishkin suggested that the hippocampus is in-
volved in “recognition memory.”25 He demonstrated this in a delayed recogni-
tion task used largely to test short-term and working memory: the monkey
had to match the arrangement of objects in front of it before the delay with a
new arrangement after the delay.
Spatial Map or Memory of Spatial View?

Another theory is that of O’Keefe and Nadel: the hippocampus is a “cognitive
map” used by animals to update spatial maps of the environment based on
“place cells” and to calculate the twists and turns needed to reach the destina-
tion.26 This theory is based on the observation that if a rat moves freely in a
test box, neurons of the hippocampus called pyramidal cells fire each time the
animal passes by a particular spot in the box, whatever its orientation. Record-
ing the action potentials that represent the neuronal activity reveals an in-
crease in frequency each time the animal passes by or visits certain locations in
the box. These neurons are called place cells. They cover a spatial domain that
varies but that does not go beyond a radius of several dozen centimeters. The
group of these place cells constitutes a spatial cognitive map.
The hippocampus thus stores spatial cues in allocentric coordinates, that
is, geometric space independent of the animal, in contrast to the egocentric
coordinates of the vestibular system. But perhaps this interpretation is insuf-
ficient.
In the monkey, it appears that space is encoded differently in the hippo-
campus. The neurons do not encode the spot where the animal is in the room
but an area of space where the neuron is activated by a cue relevant for the an-
imal. The activity of place neurons does no more than reveal an area of space
where the neurons are involved in a wide variety of behaviors.27 They are in-
fluenced by several sensory modalities (visual, auditory, olfactory, and so on).
The neurons memorize conjunctions of these modalities, as well as their com-
binations in relation to gaze movements and perhaps elements of the reper-
toire of actions (turning, walking, and so on) in which the animal is involved
at a given instant.28 It is also possible that certain neurons are activated by ob-
jects important for the animal (such as a banana for a monkey).
When a monkey is given a task to perform; for example, to recognize
whether a bowl placed on top of a wall is the one that contains food, certain
cells behave like spatial view neurons; they respond to a particular view of the
surrounding space. Similarly, other neurons are activated in connection with
the direction of gaze of the animal toward a particular point in space. The de-
velopment of the cerebral cortex in primates may have minimized egocentric
128 •
encoding of place neurons (“where I am”) in favor of allocentric encoding of
topographical relations between landmarks in the external space and those of
the place where the subject is in relation to these landmarks. For example, in
an apartment, instead of encoding the bathroom, kitchen, and bedrooms in
relation to you when you are seated at the dining-room table, what would be
encoded would be the relationships between these rooms (the kitchen is next
to the bathroom). This distinction is important, because it facilitates mental
processes involving external space. Here is another example: In a village, each
significant landmark (the train station, the post office, the park, the church,
the synagogue) can be represented in relation to you if you are seated outside
a café and you imagine the way you would have to take to go to the post
office. But to find the best route, it is also very useful to be able to imagine the
distances between each of these places and to calculate their relationships.
Understanding this perceptual decentering requires that the function of
the hippocampal neurons be studied in new experimental situations. The hip-
pocampus is sensitive not only to visual, auditory, and olfactory cues. Vestibu-
lar cues also can alter the activity of the hippocampal neurons. In the rat and
monkey, these neurons are activated when the animal is passively transported
to a place that it knows. But their activation is not connected just to the dis-
placement of the body; these neurons are activated only when the animal is
moved along certain walls, for example, or when it is moved toward a door or
turns a certain corner. The same movement of the body in another location
of the enclosure evokes no neuronal activity. So things become progressively
more abstract as successive levels are crossed from the vestibular system to the
hippocampus.29
In the vestibular nuclei, information about bodily self-motion is encoded
in combination with visual and proprioceptive cues, and is influenced by the
direction of gaze. But here encoding is exclusively egocentric (independent of
the environment). In the parietoinsular cortex described above, movements of
the head are situated in the more general context of movements of the body
and the body schema. It is only at the level of the hippocampus that move-
ment of the body is situated in allocentric space, and especially, that it is re-
lated to the particular use the animal might make of this or that part of its
space. The brain labels its perceptions according to its intentions and goals,
and the hippocampus appears to play a significant role in this process.
A major function of the hippocampus is thus to detect and memorize si-
multaneous combinations of multiple sensory messages. The hippocampus is
critical in associating together simultaneous perceptual cues. This memoriza-
tion of independent but simultaneous perceptual or behavioral events, which
129
Rolls calls “an episode,” is thus a configurational memory of events and the
temporal sequence in which these events occurred.30 The hippocampus mem-
orizes episodes in connection with other structures, the parietal cortex or the
temporal cortex, where these episodes are stored over the long term. It also
plays a role in association with the inferotemporal cortex in visual memory
and identification of shapes.31 For example, if one day we see two people at
the same time, and then see one of them a month later, the memory of the
other person comes to us. Or, if it is a couple, we might say: “Hey, isn’t that
the guy I saw with that girl a month ago?” Along with other structures in the
temporal lobe, like the inferior temporal (IT) area, which is important in iden-
tifying visual objects, the hippocampus may participate in this associative vi-
sual memory.
But an episode is not only made up of images, sounds, and smells. The
hippocampus does more than remember combinations of sensory informa-
tion and detect new sensory signals. It is also involved in remembering and
identifying combinations—what are called configurations—of perceptions and
actions. And this is where the hippocampus probably plays a major role in or-
ganizing sequences of actions, such as a series of saccades or perhaps the steps
in learning how to tie a shoe or knot a tie.
Proust’s Madeleine
For me, one of the most fascinating aspects of Rolls’s theory is that he identi-
fied, in the neuronal structure of the hippocampal network, properties that
enable it to recall an episode or a combination of sensations given only a por-
tion of the information initially memorized (Figure 5.4). This property is the
result of the autoassociative structure of the networks of neurons of the hip-
pocampus, owing to connections of hippocampal neurons that project to
themselves or to neighboring neurons and endow the structure with a ca-
pacity for memory and for recovering memorized information (recall). This
mechanism is used in certain machines to reconstruct an image from fragmen-
tary clues, and it enables the use of past episodes as models for what an ongo-
ing action might produce. In other words, to predict the consequences of an
ongoing action, the memorized action is recalled, even if its context is differ-
ent, to serve as a model for anticipating the consequences of action and possi-
bly to alter the action. What is still not known is whether the hippocampus is
mainly involved in storing new events or in recalling past ones. The right
prefrontal cortex as well as the medial parietal cortex are also involved in re-
calling memories.
Rolls gives the following example: One day in Oxford he is crossing High
130 •
A B
Temporal Parietal Prefrontal

F
neocortex S
D
Neocortex
Parahippocampic Perirhinal
gyrus cortex
PHG S
D
PHG
Entorhinal 2 3
cortex
5,6
pp
Entorhinal
Dentate
gyrus Subiculum
DG DG
CA3 CA1
Hippocampus Subiculum
CA1
CA3
Nucleus Mammillary
accumbens body
Anterior nuclei
of the thalamus
Figure 5.4. Relations between the hippocampus and other parts of the cerebral
cortex involved in spatial memory and planning action. (A): The main centers
of the brain. (B): Axonal connections to neurons. (Lower right): A characteristic
feature of the neurons of the CA1 and CA3 layers of the hippocampus is the
projection of their axons back to the same cell or to other adjoining cells. This
arrangement gives the network so-called autoassociative properties that allow it
to memorize combinations of signals and to recall the entire combination even
if some of the information the neurons memorized at the outset is missing.
Street when a bicyclist runs over his foot. The visual cues (the street, the bicy-
cle arriving at the periphery of his visual field), proprioceptive cues (the bicy-
cle on his foot), the motor action (the aggregate of information associated
with the action of placing his foot in the street), and the auditory cues (the cry
of the cyclist) constitute the configuration of sensory cues about this episode.
One week later, he is just about to step out into another street. The situation
contains enough similar cues for him to recall the bicycle episode. He sus-
pends his gesture for an instant, fortunately, for a bus is passing by at great
131
speed, and his hesitation saves him from being seriously injured. Memory of
the past allows him to predict the consequences of his action.
This mechanism might also explain the brain’s faculty for “filling in.” By
this we mean its ability to reconstruct episodes, shapes, words, and gestures
from a few elements among a configuration of signs. For example, perhaps
you do not always realize that a word is missing a letter, or an image some de-
tail; the brain supplies the missing information. Phenomenologists, Husserl in
particular, became interested in this capacity, which they studied in connection
with caricature: how are we able to recognize a face in a simple caricature, or
a man dancing, as Johansson showed, with only five or six points in motion?
The episode of the madeleine, which Proust describes at length in Remem-
brance of Things Past, is a nice example of filling in. He has only to smell the
madeleine dunked in a cup of tea to be able to recall the entire memory.
It is a general property of the brain, and among the most remarkable, not
to need every bit of information to identify a memory, even a very complex
one. Shepard, whose theories are discussed in detail in Chapter 8, proposes the
idea of resonance between a repertoire of internal representations (preper-
ceptions) and clues, even incomplete, from the environment, and here he parts
ways with Gibson: “Instead of saying that an organism picks up the invariant
affordances that are wholly present in the sensory arrays, I propose that as a
result of biological evolution and individual learning, the organism is, at any
given moment, tuned to resonate to the incoming patterns that correspond to
the invariants that are significant for it.”32 A resonator can react to a signal only
very slightly different from the one to which it is tuned, provided it has some
relationship—for example, a harmonic—with the tuning frequency. Shepard
uses this capacity of internal systems to resonate with stimuli that resemble
normal stimuli to explain certain properties of perceptual filling in; that is, the
capacity of the brain to continue to see the external world even if some por-
tion of the required information is absent.
Oscillations that Make It Possible to Predict Trajectories

A recent development in hippocampal model theory attempts to explain how
the brain can predict a trajectory during displacements in space. When a rat is
active, its hippocampal neurons fire in bursts that repeat about eight or twelve
times per second. When the global activity of the hippocampus is recorded,
that is, the electrical sum of all the activities of the neurons, the result is an os-
cillating wave called θ rhythm. No one knows yet what this wave does, but we
do know that when a rat moves in an enclosure, the discharge of the pyrami-
dal neurons of its hippocampus is connected to θ rhythm. According to
132 •
O’Keefe and Recce, there is a relationship between the instant of firing of the
neuron in each cycle of θ rhythm (the phase) and the position of the animal in
space.33
But a second rhythm in the hippocampus of the rat that has an oscillating
frequency of around 40 Hz also exists in parts of the cortex involved in pro-
cessing visual, olfactory, and auditory messages. This is called γ oscillation.
The two oscillations are superimposed in the hippocampus.
For each oscillation of θ rhythm, there are around seven to nine oscilla-
tions of the γ type; hence Lisman’s hypothesis: Each θ cycle contains a series
of seven to nine small packets of information—seven to nine episodic memo-
ries that will be replayed in this way six times per second.34
He takes the example of memory for telephone numbers. We cannot
readily memorize more than about seven numbers. In Lisman’s model, each
figure corresponds to the simultaneous activation of a certain number of neu-
rons from a population of hippocampal cells. With each γ oscillation, one of
the figures is activated. Thus we can store a set of seven figures, but no more.
What makes it possible to memorize them is that the brain perpetually replays
the seven figures in the form of seven γ oscillations, eight or twelve times per
second. In other words, external reality is broken down into packets of infor-
mation contained in the simultaneous activation of subpopulations of neu-
rons that are themselves activated every 25 milliseconds. Series of seven pack-
ets are repeated around every 125 milliseconds.
How does this mechanism enable prediction of a memorized trajectory?
Lisman starts with the assumption that if the seven packets store seven loca-
tions (identified by the configuration of synapses activated by the different sig-
nals that constitute the episode) in a room—that is, the configurations of sen-
sory information that correspond to the seven positions of the animal in the
room—when the animal passes by the first of the positions, the automatic cy-
cle of the sequence of seven positions is played at the γ frequency and thus
predicts the course of anticipated positions.
It is not possible to describe this complex mechanism here, but what is es-
sential is that the succession of the animal’s positions in space will be replayed
by the brain at least seven or eight times per second. In other words, seven or
eight times per second, the animal will experience all the different sensory
configurations that correspond to the positions he will occupy if he correctly
negotiates the trajectory he has learned. I can think of no better example of
what I mean when I say that perception uses memory to predict the future
consequences of action. This interplay of past and future repeats on a time
scale of several dozen milliseconds in the brain. It is also possible to imagine
133
that slower mechanisms exist. These ideas are of course only models, but they
are becoming increasingly plausible, even though the presence of θ rhythm
and rhythm at 40 Hz in the hippocampus of the monkey have not yet been
firmly established.
How Was Prehistoric Man Able to Draw So Well?

Recent findings on the projective mechanisms of filling in inspired me to a
new theory to explain the extraordinary quality of the designs found in prehis-
toric caves. Although convention requires a scientist to support his conjectures
with rigorous experimentation, I cannot provide any proof of my idea. But
from time to time it is useful to listen to one’s imagination.
Here, as I see it, is the cognitive mechanism that enabled Lascaux man,
working in half-light, to draw the marvelous animals whose shape so accu-
rately conveys their anatomy and movements. In the cave, Lascaux man’s
torch projected an evocative play of light and shadow onto the walls. You have
only to spend a few minutes in a dimly lit section of these caves to see how the
exceptional variety of shapes made by the limestone can stimulate the imagi-
nation. While a flashlight adds movement, candlelight considerably enhances
the apparent reality of the forms.
Trying to recognize an animal in the shape of a stone or a rock or in the
branches of a dead tree is a common experience. Country traditions are re-
plete with rocks named after animals. It is also common to seem to recognize
a familiar face in an anonymous crowd. Likewise, I think that prehistoric man,
strolling about these caves, perceived on the walls the shapes of well-known
animals.
The animals that Lascaux man saw or perceived were the recollections of
animals in action that he had in his memory. He projected onto the walls the
shapes of these familiar animals just as we project onto passersby the features
of a face that we know and say, “She looks like so-and-so.”
But the essential point of my theory is that the drawings of Lascaux are
perfect because the painter was tracing, in some way, the image of the animal
stored in his memory and projected onto the wall, just as one draws an image
on a screen where a slide has been projected. The motion of the torch in
the cave must have added considerable force to these impressions. Modern
psychophysics tells us that perception of the curvature of a planar object can
be caused by motion. This extrapolation of three-dimensional structure from
simple motion has been studied in our laboratory and is probably made in the
parietal cortex in association with the temporal cortex.
134 •
The perceptual theory of Empedocles is a naive form of the idea of a pro-
jective brain that I defend in this book. The brain projects the image of the an-
imal on the wall of the cave; Lascaux man sees it there, external to him. Recall
that Michotte wrote:
The role of stimuli is not, as was believed for a long time, to give rise to
“sensations” that are then combined, linked one to the other, and even
altered by certain psychic processes under the predominant influence of
acquired experience. Their role seems, on the contrary, to boil down to
the simple initiation of endogenous constructive processes that obey
the proper laws of organization, largely autonomous and independent
of experience, and that lead directly to constructing the world of phe-
nomena.35
It was enough for the shapes on the wall to suggest a part of the form of a
buffalo; for example, to enable the brain of the painter to mentally reconstruct
the entire body. I have shown in connection to my discussion of the hippo-
campus that neural networks of the autoassociative type permit construction
of an internal representation of an episode from just a few clues.
At the same time, the person looking at the animals on the wall of the
cave was probably filled with wonder at the supernatural existence, as it were,
of the animal he normally saw in nature. Perhaps this wonder even gave rise
to quasi-religious notions of a kingdom of the beyond where these creatures
lived and where one could interact with them in sanctuaries. Be that as it may,
prehistoric man was not drawing out of any mysterious symbolic impulse; he
only fixed on the walls the shapes seen there. He must have been amazed by
what he saw, even though his brain was recreating these shapes from various
cues, just as I am present as a spectator at my own lectures, an impression pro-
duced by a brain whose expression I listen to with astonishment.
The example of cave paintings has a very general significance. Artistic cre-
ations seem to me to be projections of internal simulations of the world. I re-
cently saw a fifteen-year-old sculptor crafting a horse, or maybe it was a dog,
out of clay, without the aid of any model or scale, without even looking at
what he was doing, essentially by feeling it. You could clearly see that the form
emerging from the hands of this young prodigy was not even in his imagina-
tion; it was in the tips of his fingers, just as the pilot feels the landing strip un-
der the wheels of the airplane as if it were his own feet, or as we feel the tip of
a pencil on paper like the end of our finger. “Vision is the brain’s way of
touching,” said Merleau-Ponty. He wrote:

135
The animals painted on the walls of Lascaux are not there in the same
way as the fissures and limestone formations. But they are not elsewhere.
Pushed forward here, held back there, held up by the wall’s mass they
use so adroitly, they spread around the wall without ever breaking from
their elusive moorings in it. I would be at great pains to say where is the
painting I am looking at. For I do not look at it as I do at a thing; I do
not fix it in its place. My gaze wanders in it as in the halos of Being. It
is more accurate to say that I see according to it, or with it, than that I
see it.36
Perception is not representation: it is simulated action, projected onto the

world. Painting is not a set of visual stimuli, but a perceptual action of the
painter who has translated, through his gesture on a limited medium, a code
that evokes the scene he perceives, not the scene represented. The painting
moves us because it reproduces the miracle of the images of Lascaux in re-
verse. I look at the canvas in the place of the painter who has projected onto it
his mental activity. A genius is someone who helps me to perceive things the
way he does.
136 •
6
N AT U R A L M O V E M E N T
The most direct, and in a sense the most important, problem which our con-
scious knowledge of nature should enable us to solve is the anticipation of fu-
ture events, so that we may arrange our present affairs in accordance with such
anticipations.
—H. Hertz
The problems the brain has to solve are mainly problems of mechanics. Mod-
ern philosophy seems to have forgotten this, so captivated as it is by language
and so convinced that the higher functions of the brain have to do with formal
logic or can be explained by analogy to computers. But if the body is to be re-
habilitated in modern neurobiology, the rules that underlie its movements
have to be rediscovered. These rules1 are intuitively understood by sculptors,
who are able to render the movements of the body and their relationship to
emotions, as are actors in Asian theater.2 These actors demonstrate that pos-
ture is the first expression of movement. In other words, it is intended or sug-
gested movement, the dynamic form of which Bernstein called “readiness to
move.” They also demonstrate that the kinematics of movement conveys
meaning, and that the trajectory of a finger, the displacement of the head, the
swaying of the body must respond to laws that are at the crossroads of me-
chanics and neurology. Moreover, they confirm that natural movement is a
source of pleasure.
It might seem surprising that I open this chapter by speaking of pleasure.
Yet pleasure is a necessary element of perception and cognition. And the
source of this pleasure is in movement. The proof of it is the delight taken in
an elegant dance step, a nicely formed letter, a ball well thrown, or the joy got-
NATURAL MOVEMENT •
137
ten from certain movements that are possible only in particular situations. As I
reported earlier, an astronaut once confessed to me his sadness at returning to
“this sticky earth”—so happy was he to be freed from the constraints of grav-
ity. He made me think of the poet Ronsard, a great master in the art of plea-
sure: “So let this muddy hide rot out, / Whose lot both destiny and fortune
gambled over. / Be spirit only. Let the body be.”3
One of the greatest pleasures of my own brain is to fancy that I am
floating like a glider, free of gravity. I have extended dreams in which I fly over
the mountains and along rivers. I particularly like to launch myself from the
top of a hill and drift slowly, close to the meadows, teasing the goats, abandon-
ing myself to the delights of effortless forward motion, impervious to me-
chanical forces. By day I also derive extreme pleasure in feeling vection, which
is the illusion of advancing that you experience watching a river flow or
watching clouds go by when you are lying on the sand.
Over the course of evolution, we learned how to make use of mechanics.
In terms an author of the Enlightenment might have used, this book is a
paean to the mechanics of the body in complex beings and the brain’s accom-
modation to it. In this chapter, I will show how the brain managed to conquer
mechanics.
It did it first by simplifying the problems, a common device in mathemat-
ics. You reduce a complex problem to a simpler one that you know how to
solve. The work of the roboticist Slotine is a good example of this approach.
Slotine developed a set of concepts from using robots that he taught to play
tennis with remarkable dexterity.4 In this way he demonstrated that making
them work at what he called “composite variables” simplified calculations con-
siderably and increased the robots’ capacity for prediction and adaptation. The
robots responded much more quickly and even solved problems for which
they lacked sufficient data. The principle of composite variables itself is sim-
ple: instead of asking a robot to control position, or velocity, or acceleration
separately, one asks it to work on a variable s that is a combination of all these
variables and whose movement is defined by what is called a Lyapunov equa-
tion. This equation is chosen in an ad hoc manner. The remarkable advantage
of this straightforward transformation is that nonlinear problems involving
successive derivatives of the nth order (for example, velocity, acceleration,
jerk, and so on) are replaced by linear problems (problems that can be resolved
by much simpler, so-called first-order equations). Of course, actual implemen-
tation of these techniques is very sophisticated and does not concern us, but
the idea is important because it developed over the course of evolution.
In the nervous system simplification was achieved by creating internal
138 •
models of physical reality that enabled simulation of movement and that con-
strain perception. This dual effect is difficult to dissociate because its two as-
pects—perception and action—are so entangled. Neuropsychology, which
grapples with dissociations in patients who have lesions, is in this respect indis-
pensable, but it cannot explain the mechanisms. Thus new theoretical and ex-
perimental tools have to be constructed. I will provide a few examples of re-
cent efforts.
A neuroethology of natural movement also will have to be constructed. It
should be one that clarifies the relationships between movement and the emo-
tions movement arouses or expresses. Darwin showed how posture expresses
the emotions of the animal or human subject (see Chapter 11).5
Unfortunately, the neurophysiology of motor systems remained for a long
time a neurophysiology of connections between nerve centers. Structure and
function were not always associated. Theory was dominated by the stimulus-
response paradigm of Pavlov and the cybernetic paradigm of Wiener. Few al-
ternatives were entertained. The physiology of reflexes was predominant
from the beginning of the century, reinforced by the discovery of numerous
tools for analytical study of the nervous system—tools like electrical stimula-
tion of nerves. Generations of neurophysiologists (a group to which I belong)
duly measured reactions and responses resulting from the application of easily
manipulable physical stimuli such as light, force, stretch, sound, and so on;
they also constructed an analytical neurobiology. How is it possible to get
from a physiology of reaction to a physiology of action, from analytical
neurobiology to holistic neurobiology? To do so we have to study natural
movement and abandon formulaic reductionism.
Pioneers
Aristotle, in De animalium motu, then Borelli, and finally the brothers Weber
were the first to attempt to adapt the laws of bodily mechanics to the move-
ment of animals. Then, in the second half of the nineteenth century, Marey
was able to generate the first real descriptions of natural movement using
chronography. At the inauguration of the chair in the natural history of com-
plex organisms at the Collège de France, where he succeeded Flourens, Marey
declared:
When we stretch our fingers and we think about the sequence of ac-
tions that had to occur . . . , at the start we find the action of volition, a
psychic action, then the transmission of this volition, a nerve action,
139
then the contraction of the muscle, a muscular action, and finally the
movement of the organ, a mechanical action. In what order should we
study these events? A philosopher of the past, a Spinoziste, would not
have hesitated: follow the logical path; introduce the facts in the very
order of their appearance. This is precisely the approach that our con-
temporary school rejects. The physiologists of today rethink the order
of events by beginning with the crudest and the most visible, and work-
ing up progressively to the most refined and obscure.
Marey undertook to capture movement at the center of gravity with cine-

matography, calculating the work done and then the impulsive force, from
which he deduced the amount of movement.6 He used a dynamograph to in-
fer the activity of the muscles. The limitation of his work, like that of his suc-
cessor, Jules Amar, was in failing to go beyond this descriptive stage and come
up with hypotheses about cerebral mechanisms. Bernstein did similar work
first in Russia, then in semisecrecy in the Soviet Union between 1910 and 1950,
owing to the official support of Pavlovian theory.
Today, rapid digital cameras, connected to computers capable of simulta-
neously capturing images and recording neural activity, enable us to envisage a
physiology of movement that benefits from the power of kinematic analysis
(see Figure 3.3). Why is this important? The measurements of the movement
of the planets, painstakingly carried out by Tycho Brahé, made it possible for
Kepler to infer their laws. Similarly, kinematic observation of natural move-
ments reveals algorithms that the brain uses to control movement.
For example, these methods reveal that all natural movements (writing,
drawing, complex movements, and so on) are organized in discrete segments.7
Write the letter “a” on a piece of paper in different sizes. Then sketch this
same letter “a” with your finger in space, or with your foot in the sand or on
the living-room rug. These movements will all have approximately the same
distribution of velocity and acceleration, in accordance with the principle of
motor equivalence. Their trajectory, to all appearances continuous, is in reality
composed of units along which tangential velocity is constant and propor-
tional to the cube root of the curvature of the trajectory. This relationship il-
lustrates a law called the two-thirds power law.
The anticipatory nature of the control of movement is also made evident
by the simple analysis of bodily movements. Try, for example, to bend over to
pick up something from the floor. Just before this flexion, an automatic syn-
ergy, discovered and studied by the neurologist Babinski, moves the body very
slightly backward. Without this displacement, the projection of the body’s
140 •
center of gravity in front of your feet would cause you to fall down. Patients
presenting with lesions of the cerebellum or disturbances of the basal ganglia,
like those with Parkinson’s, no longer display this anticipatory movement.
The analysis of movement thus brings to light solutions devised over the
course of evolution to anticipate the consequences of action and to simplify
the control of movement. I will take a look at some of these simplifications.
The Problem of Number of Degrees of Freedom
Bernstein was the first to formulate the question of how many degrees of
freedom need to be controlled. “The first clear biomechanical distinction be-
tween the motor apparatus in man and the higher animals and any artificial
self-controlling devices . . . lies in the enormous number (which often reaches
three figures) of degrees of freedom which it can attain . . . Because of this there
is no direct relationship between the degree of activity of muscles, their ten-
sions, their lengths, or the speed of change in length.” Bernstein gives the ex-
ample of a ship at sea and an automobile, and continues:
I draw my second example, for comparison, from the field of normal
human motor co-ordination . . . Fasten the handle end of a ski-stick in
front of the buckle of a subject’s belt. Attach a weight of 1–2 kg to the
far end and on the right and left sides of the wheel attach a length of
rubber tubing long enough to allow the ends to be held in the subject’s
left and right hands. Instruct the subject, turning the stick point for-
wards, to stand before a vertical board on which a large circle, square or
other simple figure has been drawn, and to try, manipulating the ski-
stick only by pulling on the rubber tubing, to follow the contours of the
figure with the point of the ski-stick. The stick here represents one seg-
ment of an extremity with two degrees of freedom; the tubing is analo-
gous to two antagonistic muscles introducing a further two degrees of
freedom into the system. This experiment (which is very useful for
demonstrations in an auditorium) makes clear to all who attempt it just
how difficult and complicated it is to control systems which require the
co-ordination of four degrees of freedom.
He emphasizes that the basic problem is one of coordination: “The co-ordina-

tion of a movement is the process of mastering redundant degrees of freedom
of the moving organ, in other words, making it a controllable system.”8
The hundreds of degrees of freedom (Figure 6.1) that characterize the an-
atomical and dynamic organization of the skeleton would have made control
141
a8
a7 a5
a6
a4
a3
a2
a1
a0
a8
a6
a7 a5
a4
a3
a2
a1
a0
Figure 6.1. Two ways of holding a glass. These two natural postures result from
controlling the number of degrees of freedom of the joints. Defining a posture
simply requires defining the relationships between angles rather than the spatial
position of each limb. This relative method simplifies control.
of movement impossible if, over the course of evolution, methods had not
been devised for decreasing both the number of degrees of mechanical free-
dom by means of the geometric organization of the skeleton and the number
of degrees of freedom the brain has to control. Roboticists, who to this day
have still not found a way to build machines matching the complexity of the
least little insect, know the extent to which any computer is quickly saturated,
both in terms of its ability to make calculations and in speed, by the few de-
grees of freedom of the robots they construct. What are the tricks nature has
found to reduce the number of degrees of freedom?
142 •
Skeletal Geometry
If you were to take a walk through the new gallery of the Museum of Natural
History in Paris to see the display of animal skeletons patiently constructed by
the systematic mind of French zoology in the last century, what might strike
you is both the enormous diversity and a certain invariance of the collection.
Despite the considerable efforts of zoology, the laws that govern the organiza-
tion of the skeleton—its geometry—are not always related to the constraints
of movement, and yet the purpose of this geometry is to reduce the complex-
ity in controlling movement. Several recent findings have nevertheless at-
tempted to link skeletal geometry to motor functions and to predatory behav-
ior. Coppens and the paleoanthropologists thereby demonstrated that the
skeletal anatomy of certain forerunners of modern man, like the famous
Lucy, resembles both tree-dwelling quadrupeds and bipeds that walk upright.
In man, nonhuman primates, and other mammals, the anatomy of the
cervical column was dictated by the need to reduce the number of degrees of
freedom. Watch a stork flying, or a lion or gazelle or ostrich running. They
keep their head held horizontally, such that the plane of the horizontal semi-
circular canals is perpendicular to gravity (see Figure 4.1). This constancy of
position in relation to gravity makes the head a stabilized platform, which
considerably simplifies processing of vestibular and visual information, as well
as their coordination.
Indeed, suppression of the rotation of the head restricts the optic flow to
alterations of the visual field connected to translations. These cues comple-
ment those supplied by the vestibular receptors of the otoliths which signal
linear accelerations, but cannot distinguish translation or tilt of the head. Sta-
bilization of the head thus simplifies the fusion of visual and vestibular infor-
mation. Of course, it also reduces retinal slip and allows the vestibulo-ocular
reflex to ensure the stabilization of the visual world on the retina.
Maintaining the head in this posture is achieved, in part, through the ar-
rangement of particular cervical vertebrae.9 In fact, curvature of the cervical
column, of which the swan is a good example but which is a feature of all
birds and all mammals, facilitates lifting of the neck, which places the head on
a pivot that has the double advantage of positioning the semicircular canals
horizontally perpendicular to gravity and creating a preferred plane of ro-
tation.
A second property of the skeleton that simplifies biomechanics is limits on
possible movements. Once more, movement of the head provides a good ex-
143
ample. The anatomy of the cervical vertebrae allows only certain well-defined
movements, familiar to specialists in kinesthesiology and rheumatologists. It is
easy to make very quick orienting movements of the cervical column, which
constitutes a pivot. Horizontal rotations are facilitated by a specialized muscu-
lar system, linked to horizontal movements of the eye. This architecture is
what allows the sparrow you see on the lawn to turn its head in the horizontal
plane with dizzying speed.
Look quickly straight up toward the ceiling. You will have difficulty doing
it. Your center of rotation is located at the level of the first cervical vertebra.
All the other vertebra are anatomically locked, which consequently forces a
single center of rotation. You will note, moreover, that it is relatively easier to
make an oblique movement upward and left, or upward and right, in the plane
of the vertical semicircular canals.
Now bend your head quickly downward to look at the floor. This time,
the cervical vertebrae are locked in such a way that the center of rotation is lo-
cated at the junction of the cervical vertebrae and the thoracic vertebrae. On
the other hand, the movement appears simpler because gravity aids the fall of
your head.
In other words, nature used vertebral anatomy to reduce a very complex
system to several basic movements, which simplifies its control. This descrip-
tion of anatomical constraints could be extended to the movements of the
arms, legs, and so on. Biomechanics imposes geometric solutions that have
been optimized over the course of evolution. For that matter, it is fascinating
to note that the brain recognizes, so to speak, this organization of the possible
movements of the skeleton.
The skeletal architecture of animals limits their movements in such a way
as to considerably reduce the number of degrees of freedom. It is represented
in internal circuits that allow mental simulation of movement. A new science
is needed to understand this dynamic architecture.
Kinematic Phase Constraints

If you draw a figure eight in the air, the angles between the various parts of
your arm will probably seem very different to you. But they are not. The rela-
tionships between them are very precise. A second trick devised by nature to
simplify the number of degrees of freedom was to create kinematic con-
straints on the different parts.10 When the angle of the arm increases in rela-
tion to the body, the angle of the arm in relation to the forearm decreases by
an equal amount: they are in phasic opposition. This kinematic constraint uses
144 •
a single parameter, varying only the relation of amplitude of the two angles,
which considerably simplifies control. In this way, the brain controls global
variables (elevation and azimuth of the movement from the end of the finger)
and not the local variables (the multiplicity of the angles that make up the seg-
ments of the limbs themselves).11
Muscular Geometry
Yet another trick discovered by nature to reduce the number of degrees of
freedom is to play upon the arrangement of the muscles. Muscular geometry
complements the simplifications contributed by skeletal geometry. For exam-
ple, most of the muscles are located between two joints (like the knee and the
hip), but others connect distant joints (like the foot and the hip): these are
called biarticular muscles, and people have been trying for a long time to deci-
pher their role. The following example makes it easy to understand why: Take
a mug of beer and set it on the table. Now, pick it up and bring it to your
mouth. The interplay of angles made by your arm, your forearm, and your
shoulder is such that had we only monoarticular muscles, some would do neg-
ative work, which is not very economical energetically. The fact that we have
biarticular muscles enables us to recover this negative work and to reconstruct
it in a positive form at one of the joints.12 Of course, it takes a bit more mathe-
matics to demonstrate than can be shown here. The point is that nature found
an elegant way to resolve a problem of mechanics using geometry, just as car-
penters do in solving problems of static distribution of weight.
Relationships between Kinematics and Geometry

Relationships or covariations between geometry and kinematics are another
way of reducing the number of degrees of freedom. Take, for instance, the
mechanical system formed by the hand, the forearm, and the arm. This sys-
tem has a geometry, a shape. In general, the geometric characteristics of a me-
chanical system cannot impose a law of movement a priori along a given tra-
jectory. The trajectory depends mainly on the temporal organization of the
motor commands.
Mathematicians tell us that if a principled relationship exists between geo-
metric and kinematic variables—for example, the radius of curvature of a tra-
jectory (a geometric feature) and the tangential velocity along the trajectory (a
kinematic feature)—“which is invariant for a class of movements generated by
one controlling system, this relationship must be the reflection of a general
rule that the system follows in planning the forces. Ultimately, any consistent
145
pattern of covariation between quantities related to geometry and kinematics
is likely to provide a clue for understanding the logic of the controller.”13 Two
covariations that meet these criteria have been discovered.
Two-thirds power law. When you draw an ellipse on a sheet of paper with a
natural movement, you may have the impression that the speed with which
you draw is totally independent of the shape. Well, that is not the case. There
is an extraordinarily precise relationship between the curvature of the form
you are drawing and the tangential velocity along the curve. If you try to draw
an ellipse, you will see that the speed of your pencil increases at the parts
where the curves are greatest. The movement of the pencil along the ellipse
can be described in at least three ways: first, describing the motion in time of
the Cartesian coordinates x(t) and y(t) of the tip of the pencil; second, using
equations to describe the shape of this trajectory and the law of movement
along the trajectory; third, describing the movement by specifying two param-
eters: radius of curvature R(s) of the trajectory and tangential velocity V(s) of
a point along the trajectory (s represents the curvilinear coordinate of move-
ment). The trajectory has been shown to be completely defined by these two
parameters.
By measuring these parameters, it is possible to demonstrate the relation-
ship between curvature and the velocity of movements of the hand. In 1983
Viviani and his co-workers discovered a simple relationship between curvature
(C = 1/R) and angular velocity (A = V/R) : A = KC 2/3, which they called the
two-thirds power law.
This law held only for a certain class of movements. Subsequent investiga-
tions of the law in relation to three-dimensional movements resulted in a re-
formulation that involves a greater variety of natural movements than draw-
ing; it even holds for points of inflection.14 The two-thirds power law gradually
becomes operative in infants as motor function matures. It connects the radius
of curvature at any point of the trajectory to the tangential velocity V, as well
as angular velocity. It is expressed by the following equation:
β
 R( s) 
V ( s) = K( s)  
 1+ αR( s) 
Factor K depends on the length of the trajectory and not on its shape.
When β = 1/3, α = 0, and K is constant, the new formulation is equivalent to
the original equation. Is this law connected to the mechanical properties of
the limbs, or does it reflect the general principles of motor control by the
146 •
brain? It appears to be due to neural mechanisms and not to geometric or me-
chanical constraints connected to the muscles and limbs. Natural movements
are often complicated. They can be broken down into parts, and this law then
applies to each separate part.
The principle of isochrony. This second principle establishes that the velocity
of movement from one point to the other tends to increase with the distance
between the points. If you have two objects before you on a table, the princi-
ple of isochrony predicts that you will cover the distance between one object
and the other more quickly if the distance is greater. Go ahead and do the ex-
periment. It is also true for movements of the eyes: the velocity of a saccade
(see Chapter 10) increases with its amplitude. A saccade of 10 to 15 degrees is
made at around 30 degrees per second; a saccade of 50 degrees is made at
more than 500 degrees per second. According to this principle, which holds for
all natural movements, the duration of a movement is relatively independent
of its linear extent, hence the term “isochrony” (fixed duration).
Recently, Viviani and Flash noted that modulation of the average velocity
along a trajectory depends on two factors: the length of the trajectory and the
distribution of the curvature along this trajectory.15 These relationships dis-
close a more fundamental property that involves the prediction of movement.
“In both cases, the fact that velocity is modulated by a global geometrical
quantity (the linear extent of the path), even before the trajectory is fully exe-
cuted, suggests that an estimate of this quantity is available to the motor con-
trol system as part of the internal representation of the intended move-
ment.”16 This theory assumes that the motor control system is equipped with
a spatial blueprint even before the movement begins, which fits with the as-
sumptions of motor program theory; it also assumes that the geometry of the
planned trajectory limits the kinematic and temporal aspects of action.
The Invention of the Eye
Kinematic constraints do not only involve movements of the limbs. An ex-

traordinary property of the movements of the eye testifies to the work done
by nature to simplify the number of degrees of freedom. The eye is one of the
most remarkable inventions of evolution. First it was a simple pivoting retina,
such as crustaceans have; then, in fish, this little orbit also became an organ
that could quickly direct gaze in three dimensions toward prey or predator in
just several dozen milliseconds.
I maintain that in the fish the eye is not only a retina that pivots but a gen-
147
uine organ of anticipated simulation of movement. The fish must maneuver a
complex and heavy body in an environment that allows it only a single move.
Catching prey or escaping from a predator affords the fish no room for error.
Displacement of gaze by a simplified mechanical organ (six muscles operat-
ing in pairs produce rotations of the eye in three perpendicular planes) allows
the fish to confirm that a movement it is planning to execute is the correct
one.
What a fantastic invention this little biological machine is! With a single
movement of the eye, the brain can visually pinpoint prey and check whether
its gesture is adapted to the reality of the external world. The movement of
the eye simulates the one the predator must make to capture its prey. Thus the
eye is more than just a pivoting retina: It is a very simple physical model of the
action needed to secure a meal. If I am a predator, all I have to do to figure out
how much I need to move my body to capture my prey is to duplicate with
my body the movement specified by my eye. Nature first had to simplify the
number of degrees of freedom of the problem and then solve it with a very
simple instrument.
Next, nature invented the principle of relative control of movement. It is
actually not necessary for the fish to calculate its position in space (and it
would have a hard time doing it in the water); its brain has only to minimize
the distance from its present position to its goal. Borrowing the language of
cybernetics, all the brain has to do is to minimize motor errors.
In primates, this internal simulation of a change in direction of gaze was
refined to an even higher degree by the invention of the displacement of at-
tention. Look straight ahead and turn your gaze quickly to the right. You will
notice that just before your eye turns, you feel your gaze move forward. In
Chapter 10 I discuss briefly the anticipatory displacement of the receptor field
made by the neurons of the parietal cortex before the saccade of the eye and
the mechanisms that enable internal simulation of gaze movements.
Transforming 3D Movements into 2D

Biology devised remarkable ways of enhancing the efficiency of this anticipa-
tory action of the eye and of simplifying neural processes.
Consider the eyeball. It consists of an orbit moved by six muscles. In the
fish, these muscles work in pairs and guarantee the horizontal, vertical, and
torsional movements of rotation that you observe in an aquarium. Over the
course of evolution human eyes migrated in their sockets and moved from a
lateral position to a frontal position. At the same time, the fovea developed
(the area of the retina that enables fine spatial resolution and highly sophisti-
148 •
cated color vision). The fovea was a late find in evolution. The rabbit has only
a horizontal furrow of greater visual resolution, and in the cat a retinal area of
approximately 10 degrees (the area centralis) has a higher concentration of
retinal cells. Only primates and man have a real fovea.
In man, horizontal eye movements are produced by the pair of horizontal
muscles, but vertical movements require the cooperation of the vertical mus-
cles and of torsion, and ocular torsion has very limited amplitude. The work
of the physiologist Hering, in 1879, gave rise to this question: How does the
brain encode rotations that enable the eye to go from one position to another?
Suppose that you are looking at a point on the wall in front of you. Displace
your gaze by an oblique saccade to the right and upward. From a geometrical
point of view, this rotation of the eye can be broken down into three rota-
tions, horizontal, vertical, and torsional (Fick coordinates), or vertical, hori-
zontal, and torsional (Helmholtz coordinates).
The problem is that this change in the sequence of rotations does not re-
turn the eye to the same position. The axis of gaze is appropriately oriented
toward the target, but the torsion observed at the end of the two movements
is not the same. Indeed, the rotations do not constitute what mathematicians
call a commutative set.
Figure 6.2 illustrates this noncommutativity with dice. Take a die and turn
it over a few times, making the same rotations but in different order. You will
Figure 6.2. A game of dice demonstrates the noncommutativity of rotations.

Reading left to right: Position a die with the number 1 uppermost. Turning it
first around a vertical axis, then around a horizontal axis will result in a position
of the die with a 2 uppermost. Now, start from the same original position and
invert the order of the rotations. You will end up with a 4 uppermost. Thus, the
order of rotations about an axis changes the result: they are not commutative.
149
see that the final position of the die is not the same. The rotations are not
commutative. The same is true for the eye. If no simplification had occurred,
the images of the same part of the environment would project to the retina in
different ways. The result would be perceptual confusion and major difficulty
in constructing a coherent image of the visual environment based on local im-
ages.
How did nature resolve this problem? Listing17 derived a law, initially pro-
posed by Donders,18 who formalized the biological solution: Every movement
of the eye can be represented by a vector whose axis is perpendicular to the
plane of rotation and the chosen amplitude in a particular relationship with
the angle of rotation. Is it possible to provide a straightforward illustration of
this vector? Simply imagine a top: the vector of rotation of the top is the han-
dle you use to make it turn, and that is perpendicular to its plane of rotation.
Now, suppose that we make the length of this axis proportional to the velocity
of rotation. It will be quite short when the top turns slowly and very long
when it turns very fast. This use of vector of rotation is a very elegant way to
describe the movement of the top in three-dimensional space. All that is nec-
essary is to define the direction of the vector and its amplitude. The three-di-
mensional problem is then reduced to two dimensions.
Listing’s law states a remarkable property: all the vectors of rotation of
the eye are in the same plane, the frontal plane of the head, called Listing’s
plane. All the saccades you make in reading this page have vectors of rotation
in this plane. When Listing’s law is satisfied, there is no torsion of the eye,
which constitutes experimental proof of the law. Many laboratories have now
confirmed the validity of Listing’s law for eye movements.
It was believed that command of eye movements could be expressed using
mathematical rules that govern these rotational vectors. The representation of
movement by rotational vectors presents another advantage: it is possible to
characterize a vector by its constituents in space with the aid of a mathemati-
cal construct called a quaternion. A sequence of rotations can be represented
by calculating the product of their scalar constituents. It was thus suggested
that nature had discovered this property of quaternions and that the brain
uses it to encode movements in the neurons. However, this theory has yet to
be confirmed.
The Law of Two-thirds Power and Ocular Pursuit

Listing’s law may not be the only mechanism for simplifying the control of
eye movements. Following a moving target (a passing car, for instance) calls
150 •
into play the neuronal subsystem of ocular pursuit (see Figure 3.2). This slow
movement of the eye appeared late in the course of evolution and followed
the appearance of the fovea, which, as I mentioned, exists only in primates
and in man. You can see how it works very simply by performing the follow-
ing experiment. Hold a finger out in front of you and fully extend your arm.
Move your finger slowly from left to right and from right to left, ever more
quickly. You will notice that your gaze follows your finger for slow move-
ments, whose cyclic frequency is about once per second (1 Hz). But as you in-
crease the speed, your gaze will start to lag behind your finger, and you will
get to the point where you cannot follow it anymore. In fact, the tracking sys-
tem is a slow system (up to 100 degrees per second)—unlike the saccade,
which is extraordinarily rapid (up to 800 degrees per second)—probably be-
cause it involves complex pathways. Ocular pursuit follows the law of two-
thirds power (this is also true of manual pursuit),19 but it remains a slow move-
ment. Nature’s way around this slowness is prediction.
Oculomotor pursuit is essentially predictive. Indeed, it is not only the tar-
get itself that is tracked, but an internal simulation, an internal model, of the
trajectory predicted by the target.20 If you move your finger in front of you
like a metronome, one oscillation per second, you will observe that after
about two or three oscillations your eye is ahead of the finger it is tracking.
Your brain is anticipating the trajectory.
The Form of a Drawing Is Produced by the Law of

Maximal Smoothness
Most theories about the control of movement assume that a trajectory is
shaped by a motor program that defines a kind of image, a sort of drawing of
movement. But one can also imagine that the shape results from the simple
laws of mechanics, just as the trajectory of a shell can be explained principally
by its velocity leaving the cannon, the force of gravity, its mass, and the laws
of Newtonian mechanics.
It has been suggested that the motor system follows a principle of mini-
mum energy that leads to movement produced with minimum jerk. Thus we
can define a cost function (CF) proportional to the mean square of the jerk,
jerk being defined as the derivative of acceleration.21
According to the principle of minimum jerk, shape results from properties
of morphogenesis (creation of shapes) that are determined by the value of
this single variable. The principle is very satisfying; it predicts the trajectory
151
A
Figure 6.3. Natural movement and the minimum-jerk model. Three subjects
were asked to draw three different shapes (left: A, B, C). To the right are traces
of the trajectories predicted by the minimum-jerk model.
and the velocity based on the total duration of movement, and the position,
velocity, and acceleration at points where it starts, ends, and through which it
must pass. Experiments have shown that natural movements (drawing a figure
eight, a spiral, and so on) made by subjects could be modeled very precisely
using the law of minimum jerk (Figure 6.3). However, this model does not
take into account the global isochrony of movement. One compromise is thus
to search among the theories that take into account global isochrony and rela-
tionships between geometry and velocity. The synthesis proposed by Viviani
and Flash assumes that the brain first constructs an abstract representation of
the intended movement, encoded as a sequence of boundary conditions, or
velocities and positions of departure and arrival. These conditions are then
transformed into trajectories by the laws that constrain movement (minimum
jerk, point of equilibrium, and so on).
Current theories of motor function thus adopt two completely opposite
points of view. One starts with a geometric representation of the trajectory
152 •
and assumes that the brain organizes a sequence of basic actions to follow it;
the other suggests that the brain ignores the shape of the trajectory, which re-
sults quite simply from high-order constraints such as the stability of paired
oscillators operating in tandem or the minimum jerk of a movement. Taking
it one step further, some would say the body follows the form, whereas for
others, the form follows from the functioning of the body.
153
7
S Y N E R G I E S A N D S T R AT E G I E S
In olden times, children might be given a splendid Harlequin doll made from
cardboard as a present. Its back concealed a network of strings, connected in
such a way that a single string made the arms and legs flex at the same time.
This invention is a legacy of Chinese shadow puppets and of Asian puppet
theater, and it illustrates a basic form of synergy. The principle involves coor-
dinating the limbs so that a single command sets in motion a network of com-
bined actions that constitute this synergy. What remarkable simplicity! Indeed,
in puppet theater, different gestures are created by manipulating various com-
binations of strings. Musical automata relied on the same principle.
The word synergy comes from syn (together) and ergos (work). This con-
cept was proposed by Bernstein to support the idea that, since the nervous
system cannot control all degrees of freedom, evolution selected a repertoire
of simple and complex movements, which are called natural movements.
These involve groups of muscles working together. I have already mentioned
the constraints the skeleton imposes on the variety of movements possible at
each joint. Moreover, this repertoire is not very extensive. Watch a dancer, and
see just how meager the motor repertoire of the human body is. What is rec-
ognized as the genius of choreography and the expressive richness of dance is
the combination in time and space of the elements of movement and the in-
terplay of partners.
Motor synergies are the basis of movement. The remarkable flexibility of
neural networks makes it possible to manipulate these synergies to produce
what I call strategies: the selection either of a particularly well adapted syn-
ergy, or of a sequence of synergies that constitute a complex movement ori-
ented toward a goal. In fact, movements are organized in sequences of syner-
gies on which behaviors are based, as Lorenz demonstrated superbly in his
experiments with greylag geese. How synergies are chosen depends on the
goal of action.1
The challenge is to construct with ethologists what I call a neuroethol-
ogy of motor function, and to work with molecular biologists to ascertain
whether some of these synergies are genetically determined by knocking out
specific parts of the genetic code in the rat or the mouse.
I will first describe the contribution of anatomy, especially the branching
of axons, to the construction of synergies; then we will analyze data from re-
search on neurons of the motor cortex with reference to problems discussed
at the end of the preceding chapter on the laws governing movement of the
arms.2
The reader would be well advised to resist the temptation to skip this pas-
sage on anatomy. The digression is worthwhile. You do not need to under-
stand the architecture of a flower to enjoy it, nor do you need to analyze how
the branches of an oak are arranged to benefit from the shade it provides. But
anyone who seeks the principles of physiological functioning must find them
first in anatomy. My objective is not to review the wealth of anatomical facts
accumulated in the aftermath of the pioneering work of Ramón y Cajal, ow-
ing to the diversity of techniques anatomists have at their disposal. Rather, I
will discuss how the branching of axons determines motor synergies and how
it links motor function and perception, in short, how anatomy comes to the
aid of physiology.
Vestibular Axon Branching and Gaze Stabilization
Look straight ahead and turn your head upward while keeping your gaze fixed
on an object. Your gaze will remain stable. The vestibulo-ocular reflex, de-
scribed in Chapter 2, contributes to this stabilization; vision is not necessary.
Indeed, try the following experiment: Look straight ahead, letting your gaze
rest on an object. Close your eyes and think of the object. Lift your head while
fixating behind your closed lids on the object you memorized. Open your
eyes. You will see that your gaze has remained where it was. In the dark your
eyes moved in a direction contrary to that of your head as a result of the
vestibulo-ocular reflex. Several muscles in each eye had to be jointly activated
to make this compensatory movement. This is a very simple example of mo-
tor synergy.
SYNERGIES AND STRATEGIES •

155
Actually, this reflex is complex. It is not the simple movement of a Harle-
quin doll. Try another experiment: Again, look straight ahead and close your
eyes. But this time, concentrate mentally on the end of your nose. With your
eyes closed, lift your head, fixating, in the dark, on the mental image of your
nose. When you open your eyes, you will see that this time you raised your
gaze. The vestibulo-ocular reflex was blocked by a signal related to the goal.
So this reflex is both stereotypical and flexible. Its stereotypical character is
due to your anatomy, which composes a synergy. Its flexibility is due to a wide
range of mechanisms that involve the cerebellum, the cerebral cortex, interac-
tions with other elements of the oculomotor repertoire, and so on. As with
the Harlequin doll, organization into synergies simplifies control by obviating
the need to keep track of all the muscles separately. Each synergy constitutes a
unit of movement. But first I will explain how these synergies are organized.
In the fish the problem of the relationship between vestibular receptors
and the muscles of the eyes is simple. The eyes are lateral, and three pairs of
extraocular muscles produce, respectively, horizontal movements (directly lat-
eral and medial), vertical movements (straight up and down), and torsional
movements (inferior and superior oblique). An arc of three neurons connects
the vestibular receptors and the muscles of the eyes. However, in the course
of evolution, in higher organisms the eyes migrated and became frontal at the
same time that the fovea appeared. A vertical eye movement in the monkey, as
in man, requires an activation synergy of the vertical muscles and the oblique
muscles. This synergy implies that synaptic activation of these different mus-
cles is very precisely distributed. This distribution is achieved through the con-
nectivity of each second-order vestibular neuron. So-called collateral axon
branches innervate the muscles that participate in this synergy. Figure 3.5
shows an example of these connections in the cat.3 Axonal connectivity and ef-
ficient synapsing of the axonal endings in each motor nucleus of a second-or-
der vestibular neuron thus encode the geometry of the movement by syner-
gistic activation of several muscles.
When someone jostles you and your vestibular reflex simultaneously
rights your head and your eyes to stabilize your gaze, the synergy is much
more complex, since your neck has about thirty muscles whose insertions are
extremely varied. This coordination of the movements of the eyes and the
head is also facilitated by axonal connectivity. Some second-order vestibular
neurons project both to the motor nuclei of the eyes and to those of the mus-
cles of the neck, which are located in the caudal-most end of the medulla.
Thus, it seems plausible that this axonal distribution underlies the eye-head
synergy that facilitates movement in various directions.4
156 •
Remarkably, the extent of the axonal projections of these neurons is not
limited to the motor nuclei. Collateral axons seek out very precise targets that
are required for the more complex aspects of this synergy, as well as for its
perceptual and behavioral context.
First, axonal collaterals project to the vestibular nuclei of the other side of
the brainstem to keep the eyes and head from moving in the opposite direc-
tion. This is called reciprocal innervation. Next, other collaterals project to
neurons whose target is the cerebellum, which plays an essential role in con-
trolling the plasticity of reflexes. Collaterals branch off to areas of the reticu-
lar formation where the cell bodies of the reticulospinal neurons play a role in
controlling posture; this connection ensures coordination between gaze and
posture. Finally, other collaterals project to interneurons, which feed back to
the colliculus, the cerebral cortex, and different parts of the nervous system
associated with perceptual stabilization, like the thalamus and the vestibular
cortex. These organs influence cortical processing of the perception of move-
ment.
A copy of the motor command is sent to multiple centers,5 contributing
to the various mechanisms involved in gaze stabilization. The question really
is one of central anticipation, since the nervous system does not wait for sen-
sory cues to reorganize its functioning in this reflex activity. Note also that this
so-called reflex is a complex entity integrated in a behavior, arising from the
most basic level of the three-neuron arc. Axonal branching allows a single
neuron to activate different levels in the synergy and to participate simulta-
neously in the motor command and in perception. The notion of simple
sensorimotor assemblies with an entry and an exit disappears, to be replaced
by a concept of a neuronal character that is both extremely specific to the ner-
vous system and completely integrated with a set of behaviors and percep-
tions.
Interpreting the anatomy of the neurons can sometimes lead to varying
conclusions depending on the implicit and explicit theoretical conceptualiza-
tions of the observer. Thus, at the beginning of the century, the neurons of
the reticular formation were considered to be hairy neurons that projected ev-
erywhere. Following on work by Morruzzi and Magoun, these neurons were
attributed diffuse functions of activation or inhibition. This was certainly true
for the ascending neurons of the dopaminergic and serotinergic systems, for
example. But now we know that their axons and collateral axons have exqui-
sitely precise targets that correspond to differentiated functions (motor, recip-
rocal innervation, corrective, perceptual, and so on). Accordingly, synergies
have to be understood as genuine perceptuomotor actions.
157
The Baby Fish that Wanted to Swim Flat on Its Stomach
But there is more. Axonal organization and the structure of connections are
alterable, at least at certain stages of development. I have emphasized that
connectivity controls the geometry of movement. For example, in the flatfish
the brain can use anatomy to change this geometry. Here is the story of the
baby flatfish that wanted to swim flat on its belly.
When it is born, the baby flatfish, like all other fish, swims vertically up-
right with its eyes on each side of its body (Figure 7.1a). A distinctive fea-
ture of this fish is that it undergoes an extraordinary metamorphosis when
it is young. It suddenly flips onto its side and starts moving around flat at
the bottom of the sea (Figure 7.1b). One of its eyes is thus on the under-
dorsal
A
horizontal
semicircular canal
eye axis
ventral
B
eye axis
ventral dorsal
horizontal
semicircular canal
Figure 7.1. Eyes change position in the flatfish. When it is born, the flatfish has
eyes on each side of its head and swims upright like all other fish (A). Soon af-
ter, the fish flips onto one side and swims around flat at the bottom of the sea.
As it develops, one of its eyes migrates to the other side of its head. (B): Be-
cause the vestibular receptors do not change position, the vestibulo-ocular
reflex no longer functions correctly; in response, the anatomy of the vestibular
neurons is reorganized.
158 •
A B
oculomotor
nuclei
vestibular
nuclei
Figure 7.2. Reconstruction of the second-order vestibular neurons (involved in

the horizontal vestibular reflex) in the flatfish after metamorphosis. (A): An ex-
citatory neuron; (B): an inhibitory neuron. At first these neurons, which receive
signals from the horizontal semicircular canals, activate the vertical and oblique
muscles. The projections of their axons are subsequently reorganized, resolving
the problem of the new geometric relationship between the horizontal canal
and the eyes after metamorphosis.
side of its body. To compensate for this disadvantage, the lower eye migrates
to the other side of the fish’s head. The two eyes end up right next to each
other.
This daring transformation of the position of the visual sensor risks com-
pletely changing the functioning of the vestibulo-ocular reflex that stabilizes
the image of the world on the retina. But in fact, the semicircular canals and
the otoliths do not move in the skull. When the animal turns, the precise con-
nections that I have described displace the eye, which has migrated in a direc-
tion that no longer corresponds at all to the correct geometry. A remarkable
reorganization of the axons of the vestibular neurons then follows, which
can be seen by injecting these neurons with a tracer (horseradish peroxidase)
that makes it possible to reconstruct the anatomy of the neuron (Figure 7.2).6
Following the transformation, the anatomy of the neurons is completely
changed: the projections are reorganized and now ensure gaze stabilization
with the new arrangement of the eyes.
159
Thus axonal anatomy does not determine only the geometry of the eye. It
facilitates adaptation to changes in the body. This biological solution to the
problem is extraordinarily elegant because it spares the animal complicated
neural calculations to reorganize its perceptual and motor space.
The Neural Bases for Encoding Movement of the Arms
If the neurons that underlie local reflexes are organized in a way that produces
synergies, how is the cortical control of movements organized? Do the neu-
rons of the cerebral cortex that direct our movements control the muscles one
by one, or do they use the same principle as the vestibulo-ocular reflex, that is,
does each neuron drive a specific synergy? Consider the case of a part of the
cortex up to now considered simple: the motor cortex located at the apex of
the skull in the central gyrus, also called Brodmann’s area 4. The early work
concerning neural activity in the motor cortex was done by the American
physiologist Evarts in 1968. It was first assumed that each projection neuron
of the motor cortex (large pyramidal cells that project to the medulla via the
corticospinal pathway) activates a muscle or possibly a group of antagonist
muscles at the joint.7 Further studies contradicted this approach. The axons of
the corticospinal neurons, like the vestibular and collicular neurons, actually
branch out systematically into various levels of the medulla. Activation of any
one of these neurons simultaneously activates several groups of muscles
throughout the body; this contributes to a precise movement (for example,
leaning forward); that is, to a synergy. In other words, once again, anatomy
underlies synergy.
The connection between the pyramidal neurons of the cortex and the
various muscles of a motor synergy is, moreover, specific to a function and
not to the muscles that constitute their target. This functional specificity is il-
lustrated by the fact that the neurons of the motor cortex can be active during
precision gripping, as when you are picking a strawberry, but the neuron is si-
lent if the same muscles are activated with great force, in what we call power
gripping.
How these neurons control force is still a mystery. Each pyramidal cell
maintains a variable relation with the force exerted by the finger that excludes
a simple causal relationship. Certain neurons of the motor cortex are activated
only when the frequency of the movement is increased, which suggests that
these neurons control stiffness (the connection between a force exerted and
the resulting displacement) of a given joint.
160 •
The Theory of Vectorial Encoding by a
Population of Neurons
A major surprise awaited researchers once they turned to arm movements
more complex than simple grasping. They expected that many types of dis-
charges would be associated with local variables of particular joints. Instead,
they discovered that the cortex was involved in the overall movement of each
limb, such as the complete trajectory of the finger. The activity of the popula-
tion of these neurons is associated with the direction of the finger in space
(this is also true in varying degrees for other cortical areas).8 If a monkey is
asked to point its finger toward targets located in various directions in front of
it, specific neurons are activated for a particular direction of the pointing
movement. In other words, certain neurons fire more when the movement is
made, for example, toward two o’clock on the face of a clock. But these pref-
erential directions are somewhat indistinct, and some neurons fire for move-
ment in all directions. If the activities of a population of neurons (about a
hundred) are simultaneously recorded during the same movement toward two
o’clock on the clock face, some neurons will fire frequently because the ges-
ture is made close to their preferred direction; others will fire less because
their preferred direction is farther off. Taking the average of all these dis-
charges will reveal that the resultant vector of the population is very precisely
oriented in the direction of the movement of the finger toward the target.
Neurons of the Motor Cortex and Mental Rotation

The motor cortex also participates in anticipating movement via a signal that
changes over time. Suppose that I ask you to point to a target in front of you.
Then, suddenly, just before you start toward the target, I indicate another
point to the right of it. Your brain has to reorient the gesture; it must make a
mental rotation in the new direction.
When a monkey is asked to make a movement toward a target, but the
target is moved just before the monkey begins his movement, the animal must
redefine the direction of its movement. The vector representing the activity of
the population of neurons in the motor cortex also changes direction during
the preparatory period, as if it were participating in a genuine mental rotation
of the planned movement.9 This change in direction of the vector signals par-
ticipation of the motor cortex in anticipating and reorganizing movement, not
just in its execution. The motor cortex does not only control force, or even
only synergies; it is also involved in a complex way in the spatial planning of

161
trajectories. But the debate on the true mechanisms of control of trajectories
is still open. As I explained in Chapter 6, defining an optimal trajectory may
require no more than specifying points of arrival and departure, and adhering
to the law of maximal smoothness (minimum jerk). The role of the cortex
may thus be to define trajectories by modulating the constraints imposed by
this law. Yet again, movement probably results from a reciprocal adjustment of
neural commands and biomechanical properties.
I would like to reflect a few moments longer on the profound significance
of this concept of synergy. If a neuron of the motor cortex controls a motor
synergy (for example, a pincer motion of the fingers), this neuron can also be
used by the brain during an internal simulation of the movement. Suppose I
ask you to imagine that you are plucking a strawberry. Look at your fingers—
they have not moved. Yet the brain can still fire the neuron that causes this
synergy, blocking its execution via an inhibitory mechanism. (I will discuss
later how these mechanisms are organized, taking gaze as an example.) An
axonal collateral of this neuron can transmit a copy of its activity to other
structures in the brain. The idea of plucking the strawberry is thus realized
when the neuron fires. This discharge does not represent the synergy; it is the
synergy.
Coordination of Synergies
The organization of movement is thus based on a repertoire of synergies that

make up a multitude of possible actions. But it is not enough to have a library
of movements that are easily activated and mutually compatible because they
use the same frames of reference or are bound by the same geometrical prin-
ciples. They still must be selected.
Locomotion is one of the best examples of this type of mechanism.
Horses must move from trotting to galloping, from walking to jumping. How
are such synergies selected? How are strategies organized? There are so many
mechanisms of selection that to describe them is beyond the scope of this
book. The most interesting modern concept is that of the functional flexibility
of networks. For example, in the case of locomotion, a network in the medulla
(described in detail for swimming in the lamprey by Grillner and his team) or-
ganizes a locomotor rhythm. Alternative modes of swimming are selected at
centers located at critical spots in the network, which reorganize the move-
ment through neurons in these centers with the aid of neurotransmitters.
Some connections are inhibitory and others are excitatory. A basic mechanism
of selection thus facilitates reorganization of the same set of neurons. This
162 •
Figure 7.3. This movement is a synergy. It consists of a rotation of the head that
orients gaze to one side, while the arm on that side is extended and the arm on
the other side is flexed. The leg on the same side as the gaze is extended, and
the other leg is flexed. This synergy is present in the baby; it is an integral fea-
ture of many gestures, like that of an athlete catching a ball or a swordsman
fencing.
type of functional flexibility has also been described by Moulins and co-work-
ers in invertebrate animals such as the snail. A network of neurons produces
rhythmic contraction of the stomach muscles of this animal and is also impli-
cated in an additional motor activity. The change in neural configuration be-
tween these two modes of functioning is assured by specialized neurons, inde-
pendent of the network. Moulins and Clarac showed that several synergies
can share the same network of neurons.
Thus, genetically determined local synergies—different kinds of locomo-
tion, ocular motor systems (saccades, vestibulo-ocular reflexes, and so on),
sexual behaviors, postures (Figure 7.3), and so forth—make up the sensori-
motor repertoire of each species. These synergies are organized as behavioral
strategies guided by global mechanisms. In higher animals and in man, these
strategies can be anticipated, selected, and internally simulated before being
executed, using the same neural structures as those of the action itself. I will
show examples of these processes in gaze orientation and postural control.
163
The cerebellum probably plays a fundamental role in organizing synergies.
The scope of this book prevents me from examining the role of this structure
in the coordination of gestures, for I would also have to compare its function
with that of the basal ganglia, and so on. I will merely take note of the follow-
ing: It is well known that stimulation of muscular, tactile, and joint receptors
activates the Purkinje cells of the cerebellar cortex. These sensory projections
are distributed over microzones, each receiving projections from sensors lo-
cated on the tongue, neck, arms, hands, feet, and so on. For several years it
was surprising to find that the same part of the body projects to several
microzones. Why, for example, are several projections of the head required?
It was discovered that the projections corresponding to the parts of the
body involved in a precise motor action are grouped together. In other words,
they compose a synergy.10 For example, projections for whiskers, mouth, and
front paws are found side by side; they correspond to the action of cleaning.
Here again anatomy reflects activity, an element of the motor repertoire orga-
nizing actions and not isolated movements. “In the beginning was the Deed,”
says Faust. The choice of a strategy for action is thus simplified: all that is
needed is to sequentially activate these sets of contiguous neurons that con-
trol the repertoire of actions. Similarly, the very same sets of neurons can be
used to internally simulate movement because these sets are the neural mir-
ror of action. It is known that cortico-ponto-cerebello-thalamo-cortical loops
exist, within which internal simulation of movement can occur completely in-
dependent of its actual execution. These loops contain . . . here I do well to
hesitate, like Faust. “Representations” is too vague; “models” is modern but
probably vague as well; “images” is too visual; “schemas” is the term perhaps
most common in the literature; “kinesthetic series” would make Husserl
happy.
There is still a long way to go before it is understood how synergies and
strategies are assembled in the various parts of the brain that control move-
ment. How the rigidity of their repertoire is compensated by the plasticity of
the rearrangements, as revealed by recent findings from neurobiology, is still
unclear. A tantalizing scientific adventure awaits.
164 •
8
CAPTURE
Perception is active exploration; it is a question put to the world, a wager, pre-

selection—it is also capture. Consider gaze. Visual perception is possible only
by actively exploring the environment through gazing, through the changes in
perspective enabled by movements of the eyes that I call stationary locomo-
tion. Each gaze cast constitutes capture, especially if the object of regard is in
motion. This capture is anticipatory, predictive. For example, try to read this
page aloud. You will be struck by the coexistence of two coordinated ac-
tions: your voice, which articulates the text, and a silent reading preceding it.
Bernstein said that we have two texts in our heads: the one we read aloud and
the one that we look at in advance to prepare our reading. This phenomenon
is well known to musicians, who are always reading the score one or two mea-
sures ahead.
According to Bernstein, planning a motor action, however it is encoded by
the nervous system, necessarily involves recognizing patterns of what will be
but does not yet exist.1 Planning requires exploration of the future. Just as the
brain constructs an image of the real external world, it must be capable of
planning in advance.
What happens when an object is in motion? Frogs manage to catch flies
and cats mice only because they make use of anticipatory mechanisms. Ob-
serve a tennis player. To initiate coordinated activation of his muscles and to
assume a posture that will enable him to hit the ball, his brain needs informa-
tion about the ball’s velocity. A certain amount of time (several hundred milli-
seconds) passes between the start of movement and the instant in which the
racket is in the desired position. During this time, the ball will have traveled
several meters. Under these conditions, the player is perpetually condemned
to miss it. This is the problem beginners have. The movement of the racket in
the direction of the future position of the ball has to be programmed. To do
that, the direction of the ball and its trajectory must be predicted.
This ability to predict the trajectory of objects in motion plays a funda-
mental role in the survival of species, even the most primitive. At a very early
stage of evolution, the nervous system developed the means for detecting po-
tential predators or prey and predicting the direction of their movements. The
lizard thrusts its tongue toward the future position of the fly it hopes to catch.
At the beginning of this book, I proposed that in the course of evolution
the brain was invented essentially to facilitate capture of prey or escape from
predators; in other words, to allow decisions to be made almost instanta-
neously based on perception. Before I look at several striking examples of this
kind of anticipation in detail, I would like to describe the neural mechanism
that enables a toad to decide if a moving shape is prey or predator.
The Toad’s Decision
When a hungry toad sees something moving, it has to decide whether the
thing is an earthworm (prey) or an eagle (predator). In the first case, it must
spring up and capture it; in the second, it must hide. If you have the time to sit
by a stagnant pond, you will notice that toads, despite their nonchalant ap-
pearance, are capable of making these decisions in a flash. The naïve psychol-
ogy of representation would claim that the toad recognizes the worm and the
eagle as such, and keeps central representations of them that it classifies as
prey and predator, such that identification of the object induces a reaction of
fight or flight.
Actually, the mechanism is much simpler and remarkably efficient. Specific
neurons are responsible for the behavior.2 They are located in the optic tec-
tum, a brain structure in the frog that is sensitive to visual cues from the retina
indicating the position and movement of objects. This structure is the ances-
tor of the superior colliculus in mammals and primates. The neurons of the
optic tectum are sensitive to a particular configuration of visual stimuli: longi-
tudinal movement of any elongated object. This is obviously a characteristic
feature of earthworms, so a lure such as a little rectangle of paper will also
serve, provided that it moves in the direction of its length. This assembly of
neurons in the optic tectum is responsible for orientation and capture.
Other neurons located in the posterior portion of the thalamus—an im-
portant neural center for multisensory convergence as well as a center for re-
laying sensory information to the cerebral cortex—are sensitive to another vi-
166 •
sual shape, which Ewert called “antiworm” because, unlike the worm, this
configuration corresponds to a large target, like an eagle or a large animal.
These neurons are connected to motor assemblies that induce avoidance be-
havior. An essential property of these systems is this: each inhibits the other.
The two behaviors of capture and flight are thus mutually exclusive.
Of course, one could say that the earthworm is represented at the central
level by neurons sensitive to the movement of an elongated object. However,
a genuine physiological interpretation of the brain would emphasize the fact
that this neural system is very simple. Receptors and first-order neural pro-
cessing detect the relevant variables directly. A few of the properties of the
neural networks—classification (the ability to recognize two sensory configu-
rations), neuronal connectivity, and a clever combination of elements of syn-
aptic inhibition and excitation—are sufficient to work out the mechanisms
that sensitize the animal to what I call expected sensory configurations, or
neural hypotheses. Schemas as Schmidt understands them guide the selection
of sensory configurations recognized by the system.3 Compared with more
highly evolved animals, the toad lacks the flexibility or capability to recognize
a fake worm. You can catch frogs using a bit of red cloth, but you cannot catch
a cat with a stuffed mouse.
The classification of perceptions (for example, of prey or predator) is de-
termined by the repertoire of possible actions (in this case, capture or flight).
Hence a further illumination of my proposition, perception is simulation of ac-
tion. But there is more. We capture first of all with our gaze. To study how
mechanisms of gaze control anticipate targets in motion, we constructed a vi-
sual game in the laboratory like the ones at a country fair where you have to
shoot at a line of ducks or pop multicolored balloons that an air current blows
around at random. Despite their simplicity, these games are difficult, which is
how the fairgrounds people manage to stay in business.
In our experiment, the subject first has to fixate on a spot in the center of
a screen on which, from top to bottom, a second spot appears that falls sud-
denly, like a ball. An auditory signal set off at various moments during the fall
instructs the subject to catch the falling target with an ocular saccade. During
the 150-odd milliseconds during which the eye remains fixated, the brain esti-
mates the velocity of the target, which moves toward the periphery of the eye.
You can do this experiment yourself by focusing on a point in front of you and
holding an object in your right hand above the horizontal. Let the object fall,
and then try to catch it with your gaze during its fall.
We observed that at the instant the auditory signal sounds, the eye moves
not to where the target is but further on in its trajectory. The brain guesses—
CAPTURE •
167
predicts—the future position of the target. How are perception and action
connected to be able to program a saccade toward the future position of the
target? We still do not know.
The Art of Braking
The toad thus exhibits two kinds of behavior that depend on prediction: cap-
ture and evasion. A second avoidance behavior is collision prevention. When
you are in an automobile, when do you brake to avoid running into a stalled
truck in the middle of the road? Lee’s hypothesis is that the brain uses an optic
variable, which he calls “time τ-to-collision.”4 In fact, Lee observed that if the
speed of the automobile is constant, τ is equal to the relation between the ob-
served diameter of the car in front and which you are approaching, and the
speed with which this diameter expands.
Time-to-collision is thus defined as the function of a derived distance.
Speed equals distance divided by time. If distance (the observed diameter of
the car on the retina) is divided by speed (the speed at which this diameter ex-
pands), the distance units cancel out, and pure time in seconds for small angles
is the result.
If movement proceeds at constant velocity, time-to-collision can thus be
evaluated by the speed with which the retinal image of the object expands. It
is a genuine optic variable. This expansion is detected by specialized neural el-
ements. In the monkey, neurons sensitive to this expansion of the image were
found in the parietal cortex, which as it happens was predicted by the Dutch
physicist Koenderink, an optic flow theorist.
What is new about Lee’s theory is that he estimates time without ever cal-
culating distance. Time τ to contact is given solely by retinal cues. I use the
word “given” to reiterate an expression of Husserl. It is a good example of
what can be given to perception in contrast with what has to be calculated, of
what is better explained by a biological, not computational, theory of the brain.
This does not mean that these processes cannot be represented using mathe-
matical methods or that the processes that take place in the nervous system
are not equivalent to mathematical operations. But a model remains a formal-
ization of a biological reality, accounting for some aspects of it.
This model was originally developed to describe an approaching obstacle,
but it obviously holds for estimating the time-to-collision (here, time-to-con-
tact) with a ball thrown toward someone (or an eagle diving toward a toad!).
If the motion is accelerated—a falling ball, for example—the time-to-contact
168 •
V
Z
Figure 8.1. Lee’s (1976) theory about the perception of time-to-collision during
a movement. Without having to estimate the distance, the driver of a car can
directly work out time-to-collision with a stalled truck based on optic flow. As
the car approaches the truck, the image of the truck expands on the retina at a
velocity proportional to that of the proximity of the two vehicles. If velocity V
is constant, time-to-collision is equal to the apparent diameter of the image of
the truck on the retina divided by the velocity of expansion of this image.
These two pieces of information are supplied by the retina and require no
knowledge of the distance.
depends on the current velocity of the target and its acceleration. Lee pro-
posed an extension of his theory to take this case into account. He assumes
that the brain uses what he calls the “τ margin.”
Returning to the definition Lee gives, suppose that I am driving a car at ve-
locity V (for example, 110 kilometers per hour) and that a truck suddenly ap-
pears in front of me (Figure 8.1). Given my velocity V, the time-to-collision, if
I do not brake, has a certain value τ. Now I must ask what margin of time (τm)
I need to brake and not run into the truck. If I need a certain time τ to begin
braking, and if the time required to avoid a collision is τm, the distance to the
collision is (τm − t)V, and the constant deceleration to avoid the collision is V/
2(τm − t). Lee’s calculations show that if the driver needs 2 seconds to begin
braking, it would be reasonable to use a margin τm of 4 seconds for speeds up
to 110 kilometers per hour, subject to the car’s ability to decelerate at about 4
meters per second per second. However, this deceleration is so high that it is
uncomfortable, and if one wishes to decelerate at less than 2.5 meters per sec-
ond per second, τm must be approximately 8 seconds.
Thus the question is whether the brain uses these optic variables, which
are directly measurable on the retina. Two experiments suggest that the brain
uses either τ or the margin value τm—that is, optic variables—and not the
physical variables that constitute the angles, velocities of targets, and their ac-
celerations.
CAPTURE •
169
Plummeting Gannets
Picture a gannet diving from different heights to the sea to catch fish. It begins
by adopting a posture that enables it to effectively battle air resistance while
still allowing it to direct its flight. But if it were to reach the water in this posi-
tion, wings spread wide, it could get hurt. So just before entering the water,
the gannet folds its wings and assumes another aerodynamic—or, rather,
aquadynamic—posture. How does it know the precise moment to assume this
posture?
Studies of films suggest that whatever the altitude above the sea from
which it begins its descent, the bird begins to fold its wings using the τ margin
given by the expanding retinal image of the surface of the water.5 The higher
the bird starts, the higher the altitude at which it reaches this value. The films
show that the higher the altitude at which it begins its descent, the sooner the
gannet begins to streamline. This strategy allows the bird a greater distance,
as it is falling, to ready itself for contact with the water, and it has two advan-
tages: it is simple, because it uses only one optic indicator on the retina, and it
is useful, because it reduces the risk of injury.
Catching a Ball
Does this theory apply to catching a ball? Early studies on catching objects in
motion borrowed their ideas from information theory. But investigation of
natural movement quickly revealed to experimenters the existence of anticipa-
tory mechanisms whose very nature defied description in terms of informa-
tion processing. For example, well before contact, the shape of the hand
adapts to the apparent shape of an object; the muscular activity of the arm is
always initiated at the same time (before contact), no matter what the velocity
of the target. The brain seems to extract information about the ball’s move-
ment from sampling the retinal slip during a segment of the trajectory and
processes this information. Geometrical features, such as asymmetries, can be
important for predicting impact. Although this apparent distortion of the ob-
ject may indicate a change of trajectory, that is certainly not the case when the
object moves at a precise distance from the body. In an attempt to resolve
these difficulties, several research groups applied Lee’s theory to games such
as baseball, tennis, and volleyball.6
Originally, the concept of τ margin was applied only to the case in which
the ball was aimed at a person, as in baseball. But what about when the ball
goes quite far into the so-called extrapersonal space, as is the case in tennis?
170 •
Studies reveal that here, too, the movement of the racket is remarkably con-
stant and independent of the approach velocity. Thus the beginning of the
movement may in fact be adjusted by the value of the τ margin. But some-
times the movement begins after a relatively flexible delay, and a connection
between perception and action enables a person to adjust the stroke or catch
according to the trajectory of the ball.
Now imagine that a volleyball player is leaping up to reach a ball in free
fall, thrown very high by the opposing team. Lee studied this case: He asked
the player to bend his knees, then to jump right up to the ball and strike it. He
measured the angles of the elbow and the knees to evaluate the temporal
course of the movement and connected that to the height from which the ball
fell. The results demonstrated that the simplest relationships between the ac-
tion of reaching the ball and the magnitude of the fall were again obtained us-
ing the τ margin and not the actual time-to-contact.
If the brain does extract dynamic information from the appearance of the
image in motion, it is helpful to know how much time is needed to gauge
these variables and to estimate the distance, or the time, to contact. For catch-
ing a ball, numbers of 240 to 300 milliseconds have been posited. To catch a
ball travelling at a speed of 10 meters per second, the athlete has a temporal
window of 50 milliseconds during which his hand must remain closed. If he
closes it too early, he will miss the ball; if he closes it too late, the ball will
bounce off his hand. A solution roboticists like Slotine have found is to assign
the hand a speed equal to that of the ball and to have a robot hand catch the
ball in this moving frame of reference.
An experiment was conducted to determine whether subjects do use the τ
margin to catch a ball. This experiment works best with a rubber ball whose
size changes as it approaches the subject (it deflates). If the τ variable triggers
the closing of the hand, the subject must close his hand later, when the ball
deflates, for its relative size diminishes and its speed of expansion is less. This
hypothesis has since been confirmed. The distance of the ball from the body
can also be perceived in units of object size. The brain operates not by seeking
the most perfect predictive information but rather the most useful informa-
tion for regulating the action. In this case, the information does not specify
“when to be where” but “how to be in the right place at the right time” with-
out worrying about where that place will be.7
Even if these ideas are disproved in the future, they are interesting because
they suggest that the brain does not construct Cartesian space or topographic
space, but units of space connected to action.
CAPTURE •
171
The Secrets of the Graceful Gesture
When we are trying to get close to an object to grasp it, we have to curb our
movement to reach the object at zero velocity relative to it. This is also true
for birds in flight who wish to alight on a tree branch or on the ground. Lee
recently became interested in the gentle deceleration that makes it possible to
reach a target without perturbing it thanks to these graceful postures. This
achievement amounts to maintaining a constant deceleration of the moving
body until it reaches zero velocity at a precise location. Pigeons landing on
their feeder, who have to slow down their flight to land gently; acrobats exe-
cuting perilous jumps, who have to control their landing so they do not frac-
ture their ankles; a person carrying a strawberry to his mouth, who wants to
grasp the fruit delicately—anyone or anything executing a gentle braking
must evaluate time-to-contact.8 Time will judge the correctness of this the-
ory. There are other alternatives, most recently ideas inspired by the theories
of advanced dynamics systems for the control of movement of Kelso9 and
Schöner10 (see the end of this chapter).
What If Newton Had Wanted to Catch the Apple?
The theories of Lee and his successors give short shrift to gravity. Does the
brain use this basic acceleration, which acts on all moving bodies, to predict
their trajectory? Have we internalized certain properties of the gravitational
field? For example, what if instead of simply having watched the apple fall
Newton had tried to catch it? How would his brain have resolved the problem?
Imagine the following experiment: You have waited impatiently for au-
tumn, and now here you are, seated beneath an apple tree in Normandy, antic-
ipating the gust of wind that will knock an apple from the tree. You hold out
your hand, palm toward the sky. The wind rises and the fruit falls straight into
your hand. This problem is simple compared with our tennis problem—you
have only to place your hand within the vertical trajectory of the apple. You
do not even have to brake gently, just adjust the dynamic parameters of your
muscles to adapt the stiffness of your hand and arm so the apple does not
bounce. It will, however, be necessary to predict the time-to-contact and the
force of the impact. How does the brain do it?
An experiment provided the answer.11 A seated subject, forearm positioned
horizontally and hand turned toward the ceiling, was asked to track and catch
a falling ball. The trick of this experiment was to place a motor at the level of
the elbow that subjected the forearm to small, very short rotations that
172 •
stretched the biceps. The subject responded to these impulsive forces by a
reflex contraction. This made it possible to test the amplitude of the reflex,
stiffness of the muscles, and so on before, during, and just after the ball
dropped. The results showed that catching a ball is preceded by an anticipa-
tory muscular activity that prepares for impact and is followed by a reflex ac-
tivity induced by the impact. If Lee was right and if the subject’s brain uses a τ
margin, the longer the duration of the fall, the earlier the anticipation of con-
tact and initiation of the muscular reaction must be. The start of anticipatory
activity of the muscles is fixed, but subjects seem to base their response,
whether anticipation or reflex, on a prediction of the dynamic properties of
the ball under the influence of gravity.
Thus the brain anticipates the dynamic properties of the ball based on sev-
eral kinds of cues: its velocity (detected by vision), its mass (estimated based
on past experience and remembered relationships among its angle, mass, and
inertia), and the influence of gravity on both limb and ball. Here we encoun-
ter the role of memory. Lacquanti concludes:
This then suggests that . . . an internal model of the dynamic interac-
tion that is expected to occur at impact is built based on a priori knowl-
edge and available on-line information (such as visual information on
the velocity of the ball and the limb). The response of this model to the
actual perturbation is compared with the response of the limb to pro-
duce an error signal. This error is subsequently used to calibrate the pa-
rameters of the neural controller of the plant [the bones and muscles]
and to update the internal model. Thus, if the model does not accu-
rately predict the desired performance, possibly because of a faulty esti-
mate of the properties of impulsive impact, kinesthetic and cutaneous
information obtained with the first trial can be used to correct the esti-
mate.12
So this model includes an internalization (whose mechanisms are still mys-

terious) of the effect of gravity on moving limbs and objects. Newton’s laws
are identified in the course of development and written in the neural circuits
that make up the internal model. This idea confirms the possibility of an in-
ternal simulation of movement without execution, since our nervous system
would contain models both of our limbs and of the effect of gravity on our
movements. It also explains why we dream of actions that seem to unfold in a
perfectly Newtonian universe even though no force opposes them, except, of
course, for dream flight, which shows that our dreaming brains do not simu-
late all the effects of gravity.
CAPTURE •
173
We proposed that NASA carry out this experiment in space. This was ac-
complished in April, 1998, during a flight of the space shuttle Neurolab. We
questioned what strategy the nervous system chooses when movement is not
accelerated by gravity but has a constant velocity.
The idea of internalization of gravity in the nervous system is not new.
In the 1950s, the Soviet school had already assumed that the brain internal-
izes properties of the physical world in the structure and activity of its neu-
rons. I will come back to this point. Perception and action are thus connected
through these mysterious internal models of the properties of the limbs and
objects in the physical world. The consequences of movement can be simu-
lated and predicted by the brain using these internal models.
Because it is worthwhile to delve more deeply into this idea of internaliza-
tion, I will suggest yet another example.
The Problem of the Waiter

The problem of the waiter can be formulated in the following way: when the
waiter arrives with a tray full of bottles and lifts one to give it to a customer,
he reduces the weight supported by his arm by about 1 kilogram. Yet his tray
remains perfectly stable. But if a customer takes the bottle without warning
him, his tray becomes unsteady.
The fact that the tray does not move when the waiter removes the bottle
means that he has an internal model that enables him to predict the effect of
the unballasting on his arm. Before removing the bottle, he decreases the
tonicity of his muscles. This anticipation probably requires a motor memory.
In this case the frame of reference used by the brain is the horizontality of the
tray in relation to gravity. This ability to predict appears in infants approxi-
mately 1 year old.
The concept of the internal model even turns up in cybernetics. In fact,
engineers were the first to discover that performance of servo systems im-
proves significantly if instead of waiting for the sensors to detect errors and
make corrections (which takes time and leads to oscillations), a calculator is in-
cluded in the circuit that estimates the initial state of the sensors and their
variation (or mathematical derivative) based on their current value.
It was an engineer, Kalman, who developed the predictive filter that bears
his name.13 The Kalman filter is the optimal filter, the best-known predictor of
the state of a linear system, and so it is used in many automatic guidance sys-
tems. Young, from MIT, was among the first to use it in a formal model of
postural control based on the idea that cues from the sensory receptors (vi-
sion, proprioception, inertial vestibular sensors) are compared with estimates
174 •
made by a Kalman filter. The motor command for correcting posture during a
perturbation is thus not the simple response of a servomechanism or a mea-
sure of the discrepancy between the state of the system and the desired state.
It is dictated by comparison of the state of the sensors with the prediction
made by the Kalman filter.
Is the brain equipped with Kalman filters? We have no idea.
The Shortest Path from One Point to Another

Has the nervous system internalized the physical properties of the world?
Many great thinkers have embraced the idea of a cerebral image (in modern
terms, an internal model) of the world. Among them is von Uexküll, who had
a profound influence on modern physiologists of the corollary discharge
school like von Holst and Mittelstaedt. Von Uexküll defined the concept of
Umwelt, or environment.14 According to Merleau-Ponty, “for von Uexküll,
Umwelt signals the difference between the world as such (Welt) and the world
as the realm in which this or that living thing exists. It is a transitional reality
between the world as it exists for an absolute observer and a purely subjective
domain.”15 The sea urchin is an example of an animal that is no more than
reflexes and a repertoire of independent synergies; the sea urchin, says von
Uexküll, is a “republic of reflexes” that exists in an Umwelt that represents
things that are often dangerous but to which it is so well adapted that it truly
lives as if there were only one world. Its nervous system contains no image of
the world. On the other hand, for higher animals, Umwelt is not closure but
openness. The world is possessed by the animal. The external world is distilled
by the animal which, differentiating sensory data, is able to respond to them
with fine motor actions; and these differentiated actions are possible only be-
cause the nervous system is equipped like a replica of the external world
(Gegenwelt), like a copy. It is a mirror of the world (Weltspiegel). The Gegenwelt
is itself divided into Merkwelt, which depends on how the sensory organs are
arranged, the world of perception, and the Wirkwelt, which is the world of ac-
tion. Today it is clear the sea urchin is more than a republic of reflexes, but
von Uexküll’s ideas have the benefit of setting the debate.
The idea that internalization of the properties of the physical world con-
strains perception, that we perceive the external world through the laws of the
physical world integrated within the functioning of neural networks, is also
supported by psychologists. Shepard, for example, writes:
I believe the external constraints that have been most invariant through-
out evolution have become most deeply internalized, as in the case of
CAPTURE •
175
the circadian rhythm. Such constraints may be extremely general and
abstract: The world is spatially three-dimensional, locally Euclidean,
and isotropic except for a gravitationally conferred unique upright di-
rection, and it is temporally one-dimensional and isotropic except for a
thermodynamically conferred unique forward direction.16
Several properties are subject to internalization by the nervous system, in

particular the laws of geometry and kinematics, which control relationships
between movements of rigid objects or various parts of nonrigid objects. For
example, despite the countless ways of superimposing two identical objects lo-
cated separately in space, there is always a simple rotation that can do it. This
optimal path property was established between 1763 and 1830 by Italian
(Mozzi and Giorgini) and French (Chasles) mathematicians. Suppose that we
wish to move an object in space from point A to point B by combining a ro-
tation and a translation. This is the movement a helicopter makes when it
takes off. It is also the twisting movement of a screw. According to Chasles’s
theorem, there is only one axis in space such that the object can be moved
from A to B by both a rotation around the axis and a simultaneous translation
along it. Shepard maintains that the brain has internalized these kinds of prop-
erties.
If the brain does know these laws, the proof should be in the way it splits
up perception of movement into rotations and translations. The vestibular sys-
tem actually seems to do this for movements of the head, as evidenced by the
semicircular canals, which detect rotations, and the otoliths, which detect
translations (see Chapter 2). In the monkey, this geometric segregation is
found in the analysis of optic flow by the neural networks of the cortex: in the
visual cortical areas of the superior medial temporal sulcus some neurons are
sensitive to a combination of rotation and expansion of visual objects, but oth-
ers analyze the components of rotation and translation separately. In the ac-
cessory optic system, geometric segregation of visual movement is carried out
very early, as I discussed in Chapter 3.
Shepard also considered the case of Chasles’s planar theory: for two posi-
tions, A and B, of an object in a plane, there is always a single axis such that
the object can be rotated from A to B (imagine a top with points A and B indi-
cated on the circumference of the circle the top makes, for which the handle
would be the axis). To confirm that the brain really has internalized this prop-
erty, Shepard showed a subject two identical polygons in a row, each inscribed
within a circle, the second having undergone a rotation relative to the first.
The succession of the two images gives the impression either that the polygon
176 •
is moving or that it is spinning. The polygon gives an illusion of movement
that follows Chasles’s theory to the letter.
Several other experimenters have demonstrated that perceptual illusions
of movement appear to respond to the constraints of kinematics. For Shepard,
when the brain predicts a trajectory, it does not take into account specific ana-
tomical characteristics; it predicts the trajectory according to a hierarchy of
simple presuppositions. The suppositions at the top of the hierarchy assume
the rigidity of the moving object. However, when the brain has to work
quickly, it takes shortcuts and can no longer stick to this rigidity hypothesis. If
different trajectories are possible in a three-dimensional space, the brain will
choose geodesic curves—the shortest on the surface in question, just like air-
planes going around the earth from Paris to New York or from Paris to Tokyo.
What is the neural basis of this internalization? Shepard has no answer. He
does, however, propose that neural mechanisms enable genuine simulations in
the internal circuits, and in this he differs with Gibson, who denies any auton-
omous simulation activity that is not connected to reality.
The fundamental laws of geometry and kinematics are not the only ones
to be internalized. If when we are looking at a series of photographs of a per-
son jumping, like the photographs by Marey or Muybridge, we are shown an-
other photograph and are asked whether it belongs to the series of jumping
movements, we can identify it all the more quickly if it pictures a stage that
corresponds to the continuation of the natural (predictable) trajectory of the
jump. Similarly, shown two photographs, we can very quickly recognize the
movement of a person’s arm if the motion required to move from one pos-
ture to the other is physiologically possible. Here the principles of the motor
theory of perception, introduced in the first chapter, again come into play.
Proof is accumulating that the laws of motion constrain the perception of
movement. Recently, Viviani showed that if a subject is shown a point of light
moving on a screen in an elliptical trajectory, for the subject to have the im-
pression that the point of light is moving at a constant velocity, in reality it
must move with a velocity that varies with the curvature of the ellipse, as the
law of two-thirds power predicts. Thus the point has to move as the finger
would naturally move. The laws of natural movement constrain perception!17
Internal Models and the Inverse Problem

I have discussed the concept of internal models extensively. Before going any
further, I must explain the so-called inverse problem. Say that you want to
make a drawing, and that the neurons of your motor cortex fire, activating the
motor neurons of your arm so they bring about the movement required to
CAPTURE •
177
draw a circle. This neural command activates muscles that have a particular
geometry, stiffness, and viscosity that will activate the arm, which has an iner-
tia, leverage mechanisms, and so on. The motor command is thus mediated
by the biomechanical properties of the limb, and the movement runs the risk
of being completely distorted. For the motor command to be faithfully trans-
mitted, it would have to be distorted in advance so that it could somehow mir-
ror the distortions the limb will impose on it. This is called the inverse prob-
lem, because the question is one of subjecting the motor command to an
inverse transformation of what it is about to undergo. Roboticists have de-
vised many ways of making this inverse calculation, because it is a problem
that comes up with all robots: it is easy to program a circle into the computer
and to send commands for drawing a circle to the motors in the robot’s arm.
But if this arm is elastic, subject to frictional forces, heavy, and so on, such
properties have to be represented in a model, and the inverse of the command
signal has to be applied before it is sent to the motors.
To plan a defined trajectory, it is thus necessary to account for this filtering
of the limb in advance. One solution is to have an internal model in the ner-
vous system, before the final motor command, that modifies motor signals ac-
cording to a transformation exactly the inverse of the modification imposed
by the arm, that is, the controlled object. When the movement is executed,
the internal (inverse) model and the arm are linked. The result would be the
following sequence: “desired trajectory–inverse properties of the arm–proper-
ties of the arm.” And what you would have at the end would be the desired
trajectory undisturbed by the filtering of the arm.
Recently, Kawato proposed an additional concept, also used in robotics: a
model of object control.18 This feedforward model is used as a substitute for a
limb or object controlled by the brain. Setting a desired trajectory, an inverse
model, and an internal model in a series results in a simulated virtual trajec-
tory, which it is possible to correct without having to execute the action in the
external world.
The different operations of these models can be represented schematically,
ranging from the least to the most predictive:
Initial situation
Planned trajectory → motor command → controlled object → distorted trajec-
tory
Addition of an inverse model
Planned trajectory → motor command → inverse model → modified motor
command → controlled object → ideal trajectory
178 •
Addition of a feedforward model
Planned trajectory → motor command → inverse model → modified motor
command → model of the object → virtual trajectory → comparison with the
planned trajectory → error → internal correction of error → modified motor
command → controlled object → ideal trajectory
For the brain, the advantage of working with these kinds of internal mod-
els is that if one takes into account the properties of neurons, processing time
is on the order of 30 milliseconds—much less than a return using proprio-
ception or vision (on the order of 100 to 200 milliseconds). This mechanism is
also capable of learning and adapting. In fact, it is possible to correct these in-
ternal models by sending them error signals detected by the senses or by the
internal neural circuits that monitor relationships between command and exe-
cution.
More generally, how is it possible to imagine that the networks of neurons
in the brain can be adapted to the conditions of the physical world? This pro-
cess must be implemented in the baby. Walton and Llinás made an important
observation.19 They noticed that baby rats tremble when they are with their
mother in the first days of life. This trembling is the result of electrical oscilla-
tory properties intrinsic to the motor neurons of the medulla that involve all
the limbs. Movement, or the trajectory of the limb, results from the filtering
of these neural oscillations by the mechanical properties of the body. The rat’s
sensory receptors detect these movements and relay the information back to
the medulla. This return oscillation can be used to specify the difference be-
tween central activity and actual movement and to calibrate the neural net-
works by constructing an internal model. The implications of this idea have
not been studied, and how these internal models might be constructed is not
well understood.
Children’s first movements and play thus function both as training for mo-
tor programs and also as a way to construct these internal models. Their sig-
nificance and especially the variability of the skills that children acquire de-
pend on the internal models that they have managed to construct.
Dynamic Theories
This theory of internal models of mechanical properties of the body is one
way of accounting for the capacity of the brain to simulate relations with the
environment that it can use in anticipating. But it is not the only theory, and
its working assumptions (for example, that the brain contains a representation
of the trajectory of the arm) are controversial. For several years, another
CAPTURE •
179
school of thought has taken a different tack, that of the theory of dynamic
systems, which holds that the behavior of the nervous system is integrated
with nonlinear interactions of oscillators.
The main objective of this theory is to define a paradigm for studying the
neural basis of movement. It seeks to substitute a paradigm of paired oscilla-
tors for the stimulus-response paradigm. When these oscillators function in-
teractively, they have a tendency to revert to a stable state. This tendency to re-
vert to a preferred mode of oscillation when there is a deviation from its
functioning is considered a reflection of internal properties of the system.
The dynamic systems approach refers to this property.20 It considers diver-
sion of the system from a stable state to see how it reverts to that state or how
it reorganizes itself to reach another stable state. Stable does not mean mo-
tionless. Thus, the rhythm of walking is a (dynamically) stable state. Tap on
the table with one finger of each hand at the same time. You will see that you
have a natural tendency to adopt a particular rhythm of tapping; you establish
a dynamically stable state. Now try to tap a bit more quickly with the right
hand: you are upsetting this dynamically stable pattern. Moreover, it is known
that a particular region of the brain—the supplementary motor area—is im-
mediately activated when you make such a change.
Actually, dynamic theory assumes that the variables controlled by the ner-
vous system are not only intrinsic variables of the elements of the body
(length of the muscles, muscular force, angles of joints), but also variables
connected to the task and to the kinematics of movement. These last two are
the most important; the nervous system organizes motor activity by control-
ling these so-called coordination variables.
The biological relevance of this approach remains to be demonstrated, but
it is supported by many arguments. In fact, internal neural circuits do exist,
like those that connect the cerebral cortex, the basal ganglia, and the thalamus
(which projects in turn to the cerebral cortex), or that connect the cerebral
cortex to the nuclei of the brainstem (called pontine nuclei), which themselves
project to the cerebellum and, from there, to the cortex. Internal oscillation
generators cause rhythmic activity that sweeps through these circuits (see
Chapter 1). For example, the cortico-striato-thalamo-cortical circuit is swept
by an activity oscillating at 40 Hz, other motor circuits involving the inferior
olive are swept by bursts of activity at 10 Hz, those involving the hippocampus
and the limbic system by activities of 6 to 9 Hz, and so on (see Chapter 5).
180 •
9
T H E L O O K T H AT I N V E S T I G AT E S T H E W O R L D
I have demonstrated that the critical issue is not so much a matter of catalog-
ing sensors, but what questions the curious brain asks the world based on the
assumptions it develops and tasks it proposes to carry out. The brain not only
formulates hypotheses and tests them using the sensory receptors. Evolution
made the brain a machine for predicting, not just a machine to take account
of situations. It also made it an organ for detecting, predicting, and interpret-
ing movement, for there is no action without movement. Gaze orientation was
one of the first functions that required the development of a brain that could
predict, a brain that was curious, and a brain that could simulate action.
Gaze Orientation
Orientation toward a sensory source is a behavior that turns up in the sim-

plest organisms. Ethologists call it taxis, and distinguish phototaxis, heliotaxis,
thermotaxis, and so on. It often involves locomotion, flight, or swimming to-
ward the source of attraction. In insects such as the fly, movements of orienta-
tion result from very subtle but relatively rigid automatic mechanisms. In the
toad, as I mentioned in Chapter 8, movements of orientation are associated
with capturing prey (an earthworm, for example) or with evading capture.
Lorenz studied the repertoire of orienting movements in mammals and
birds. He emphasized that some of the synergies involved in these movements
are completely independent of sensory inputs and “physiologically they [mo-
tor patterns] are distinguished by the fact that their fixed pattern of move-
ments, contrary to what one would expect, is not activated by a succession of
reflexes but by processes which take place in the central nervous system itself,
without receptors being involved.”1
But let us consider gaze orientation. Gaze orientation is projection, ques-
tioning, prediction. It is stationary locomotion. It is the model of choice for
understanding the physiology of perception. Chapter 2 described the mecha-
nisms that stabilize the visual world on the retina owing to the vestibulo-ocu-
lar reflex. And Chapter 3 explored how the analysis of movement in the visual
world through the pathways of the accessory optic system (responsible for
optokinetic nystagmus) is carried out via geometric segregation of retinal slip
in three planes that are the same as those of the semicircular canals. Also in
Chapter 3, I wrote of the late appearance in the course of evolution of the
mechanism by which an image is maintained on the fovea.
The idea that gaze is a projection of the brain on the world is not new.2
Even in Babylon, gaze was either male, and projected, or female, and receptive
to light. In Grecian times, Empedocles (Figure 9.1) tried to explain sight as fol-
lows: the interior is fire, around which are water, earth, and air, through which
fire passes by virtue of its subtlety, like the light in lanterns. The pores of fire
and water are arranged in staggered rows; through the first, white objects are
perceived, and through the second, black objects. According to the so-called
theory of extramission, this internal fire produces light that in turn is reflected
by the object and returns to the eyes (perhaps the origin of the expression “a
burning gaze”?).
Plato proposed his own theory of interaction. Visual rays emanate from
the organism and lead to interaction with ambient light to form a cone of vi-
sion whose vertex is in the eye and whose base is at the object. This cone
makes the object vibrate and transmits this vibration to the eye. This signal ac-
tivates the cognitive components of the soul that are located in the brain.
Later, Alahazen (965–1038) proposed another interactive theory of gaze: effer-
Empedocles (~490–430 b.c.)
Figure 9.1. The projective gaze. According to Empedocles’ theory of

extramission, a light projected by the eye grasps an object, which reflects the
light back to the eye.
182 •
ent visual signals produced by the brain (spiritus visibilis) at the level of the op-
tic chiasm penetrate the eye and interact with the visual waves produced by
the objects. This interaction is then sent back to the brain, where it combines
with that of the other eye to produce a perception.
The art of the Renaissance proceeds largely from the discovery of perspec-
tive. In contrast to the stilted view of earlier paintings in the East as well as in
the West, gaze came to life and became a way of looking at the world from
various vantages. It estimated proportions and guided architectural relation-
ships.
The theory of corollary discharge is a modern version of a theory of in-
teraction (see Figure 3.4). This theory is often attributed to von Holst and
Mittelstaedt,3 but it can be traced to Bell and to Purkinje (1823–1825). Bell and
Purkinje observed that external objects seem to move when the eye is made to
move, whereas they remain constant during active gazing movements (for
more on this subject, see Chapter 10).
A copy of the motor command is sent to the center of perception when
the movement is actively triggered by the brain. This copy makes it possible to
update representations of the object and to compensate the displacement of
its image by a virtual image that remains constant. Within the framework of
cybernetic theories of servo control, which have dominated thinking about
gaze mechanisms for 20 years, the copy of the motor command corresponds
to the concept of feedforward as opposed to feedback.
The capacity for actively exploring space via orienting movements, pro-
duced not in response to environmental stimuli but according to a person’s
intention, is important. Lorenz even suggested that the organization of a pro-
gression, that is, a locomotor trajectory—based on constructing a representa-
tion of space and requiring an associated sequence of orienting movements—
is the most highly evolved form of orientation reaction.
This idea turns up again in the development of gaze in infants. When
seated in a highchair, a baby first directs his gaze using his own body as a refer-
ence point. He turns his head and eyes about the axis of his body, an egocen-
tric frame of reference. But once he begins to walk, he can no longer explore
the world with his eyes in a fixed reference system and must develop a new
strategy for keeping objects constant in space. Before 18 months, he fixates on
a landmark (for example, a corner of a table) to which he anchors his progres-
sion around the room. This strategy is probably the first sign that he is con-
structing allocentric relationships (locating an object in the environment). If
his anchor point is hidden from him, the infant reverts to an egocentric strat-
egy (related solely to his own body). At 18 months the infant adopts a third
THE LOOK THAT INVESTIGATES THE WORLD •
183
Figure 9.2. The man is climbing the stairs while keeping his head stabilized—his
head does not turn in the vertical plane. The angle of his head is dictated by the
direction of gaze. The straight line drawn on the head joins the meatus of the
ear to the external corner of the eye socket. It indicates roughly the plane of
the horizontal semicircular canal. The direction of gaze is lower than this line.
(After photographs by Muybridge)
type of functioning. He no longer fixes his gaze, but is able to mentally update
his position in the room while he is walking, employing a mechanism of men-
tal rotation and translation.
In adults, gaze is not directed at random during locomotion, as shown in
the images I drew based on photographs by Muybridge (Figure 9.2). The gaze
of a person mounting a staircase or running, or of people carrying out a task,
is oriented toward a precise goal. It can be readily stabilized in space to enable
processing of optic flow and combining of information about movement (see
184 •
Chapter 4). It can be anchored to an object, even an object someone is carry-
ing. Enormous functional flexibility succeeds egocentric organization in the
infant.
“Go Where I’m Looking,” Not “Look Where I’m Going”
Although we are freed from the constraint of having to fix our gaze, we con-
tinue to guide locomotion, or, rather, our direction indicates anticipation of
the trajectory. You can do the experiment yourself. Clear a space of around 3
meters by 3 meters in your living room and draw a circle on the floor with
chalk. Place yourself anywhere in the circle, and look at the circle. Then close
your eyes, and make two complete tours trying to stay within the circle.
You can do the same experiment by drawing a triangle or square 3 meters
on a side, which obliges you to walk in a straight line and then to turn corners.
During these different tasks both in the light and in the dark, a person’s gaze
turns toward the interior of the concavity about 300 milliseconds in advance
of the feet and the trunk. In other words, we head toward the place we are
looking, and not the contrary. The brain does not develop a simple motor pro-
gram that enables it to proceed along a circular trajectory. It follows an inter-
nal model of the trajectory just as you pursue a moving target using your
eyes or your hand. In other words, we mentally simulate the trajectory, and
we compare the actual movement executed by our feet with the predicted
movement.4
Eye-to-Eye Contact
Gaze orientation not only enables us to fix on a target selected from the envi-
ronment, it is also required to construct a coherent representation of space in
primates and in man, whose vision is fragmented owing to development of
the fovea. Analyzing the evolution of vision in vertebrates, Gibson reminds us
that fish have a fisheye, simulated today by camera lenses, which can take in
wide surroundings with a single glance. In the course of evolution, the ap-
pearance of foveal vision was accompanied by migration of the eyes from a
lateral to a frontal position, which enabled perception of depth, tracking, and
so on. But the disadvantage of frontal vision was that animals lost their pan-
oramic vision and had to reconstruct the visual environment based on a series
of restricted views. Orientation thus was not used solely to guide the mouth
or the body, but to explore the visual world. Coordination of exploratory
movements ensures the acquisition of successive images that, according to
185
Gibson’s example, is essential to a coherent representation of the entirety of
the room in which one happens to be.
But gaze movements do not just serve for capture or for coordinating
points of view; they also govern relationships with others. One of the func-
tions of gaze is data gathering, but it also plays a fundamental role in inter-
actional equilibrium, for example, the exchange between a mother and her
child. I will come back to this example in relating the development of ocular
movements in the infant. This exchange has fascinated the research groups of
De Ajuriaguerra, Minkowski, and Lésine, and we are indebted to Saint Anne
Dargassies for one of the first reliable descriptions of it. De Ajuriaguerra calls
it “eye-to-eye contact.” It has long been known to so dominate the act of eat-
ing that infants are sometimes totally distracted from suckling. This eye con-
tact may be among the innate triggers of maternal response. But it is not
enough to be able to direct gaze; one must also be able to turn away from the
target that attracted it.
From studies of precommunication or communication modalities be-
tween mother and infant, with reference to gaze reciprocity, we find a
whole set of ambiguous problems in their mutual understanding: the
eye that touches and that encompasses, that takes and rejects, that hears
or that listens, that licks and devours . . . the infant can follow the other
through ocular scanning without being able to leave her; fascination
gives way to intrusion. For this the two gazes must encounter each
other. And thus vision becomes sorcery . . . The infant feels more com-
fortable once this fixation becomes reversible, movable, and relative;
and when, in the love-hate battle that is created over the course of life,
exchanges that safeguard control and intentionality are established. This
happens as soon as the infant is able to break away from the magnetic
hold of another’s gaze by turning his head and closing his eyelids.5
It is now known that development of the frontal cortex enables this evasion,
because patients presenting with lesions of these regions persevere in their
fixation.
The eye of the other is not always benevolent. All Mediterranean civiliza-
tions are familiar with the “evil eye.” In North Africa and in the Near East, it is
experienced as an ambivalent organ, both receptor for the physical world and
transmitter of a living force. It flows out like water from a spring for which
the eye is named in Arabic. First fascination, then desire, it brings on misfor-
tune. As early as the third century, Heliodorus of Emesa had summed up the
essentials of these Mediterranean beliefs: “After that, it is enough for anyone
186 •
to look at a beautiful object with an envious eye to communicate pernicious
qualities to the surrounding air; his breath spreads, full of bitterness, over his
neighbour, and, thanks to its subtlety, penetrates to the bones, even to the very
marrow. Thus it is that, frequently, this species of envy which has been called
by the name of ‘fascination,’ affects the health of those whom it attacks.”6 So
the glance exchanged may also be one of horror; the orientation reaction may
turn into an avoidance reaction.
This exchange also mediates the social status of the individual in a group.
Bourdieu describes the role of gaze restraint on the habitus of the Kabyles in
this way:
The virile man who goes right to the point, bluntly, is also one who,
ignoring looks, words, gestures, contortions, faces up to and looks
straight in the face of whomever he is about to welcome or toward
whomever he is heading; always on the alert, because always under
threat, he lets pass nothing of what goes on around him, a glance lost in
the air or glued to the ground being the mark of an irresponsible man
who has nothing to fear because he is of no consequence within his
group. In contrast, a properly brought up woman, one who does noth-
ing out of turn “neither with her head, nor her hands, nor her feet,” is
expected to go about slightly bowed, eyes lowered, to avoid any gesture,
any unwarranted movement of her body, head, or arms, looking no-
where but the place where she will put her foot, especially if it happens
that she has to pass in front of a group of men . . . In short, truly femi-
nine virtue, that is, “lah’ia,” modesty, restraint, reserve—orients [italics
mine] the entire female body downward, toward the earth, the interior,
the house, whereas male excellence—the “nif ”—is confirmed in move-
ment upward, outside, toward other men.7
In this exaggerated interpretation of the orientation reaction, the stimulus

is no longer simply a sensory configuration. As Bourdieu makes clear in his
definitions,
in practice stimuli do not exist solely as objective truth; they operate

only as conditional and conventional triggers, dependent on encounter-
ing agents that are conditioned to recognize them. The habitus can only
produce the response written objectively in its “formula” insofar as it
brings to the situation its effectiveness as a stimulus by constructing the
situation according to its principles, that is, in causing it to exist as a rel-
evant issue in relation to a particular way of questioning reality.

187
To my mind, this text is fundamental. The sociologist, just like the physi-
ologist, concludes that the brain does not just passively experience a set of
sensory events in the surrounding world. Quite the opposite, it questions
events based on its presuppositions. A genuine physiology of action would be
based on this principle.
Gaze and Emotion
To conclude this overview, which is intended to show the complexity of the

orientation reaction, I must mention the relationships between gaze-orienting
movements and emotion. More may be needed than a biochemistry of pas-
sions. Facial expression is a powerful indicator of emotions. The essential role
of gaze, reinforced by general body posture (see Figure 11.3), was illustrated
wonderfully by Darwin in The Expression of Emotions in Man and Animals.8 The
physiology of orienting movements is thus also a physiology of expression. In
other words, it is projection and not only reception. When I direct my gaze to-
ward a target, I am trying at the same time to act on that target. Movement is
an integral part of perception.
Theater provides excellent examples of this intimate link between the
body and its movements, perception, and the expression of emotions. In the
West, especially in France, attempts were made to dissociate textual theater
from the theater of expression through a Cartesian intellectualism that sepa-
rated the intelligent mind from the emotional body and language from ac-
tion. However, great directors like Kantor, Grotovski, and Mnouchkine dis-
covered the essential role of movement and of the body in the actor’s acting,
in his ability to make language and ideas come alive in a direct way. One of
the remarkable examples of the theatrical use of gaze is that of kathakali,
which has influenced many contemporary directors. One of the rules that
masters of this choreographic art teach their students could not be more rele-
vant for us: where the hands go to represent an action, the eyes must fol-
low; where the eyes go, the intellect must follow; and the action represented
by the hands must give rise to a specific feeling that is reflected on the face of
the actor.
The repertoire of an actor of kathakali consists of a complex alphabet of
gestures that are composed of facial expressions, gesticulations, and pre-estab-
lished bodily kinetics. Thus, there are nine movements of the head, eleven
ways of looking, six movements of the eyebrows, four positions of the neck,
and sixty-four movements of the limbs. The gestures of the hands and the
fingers have a narrative function based on fixed shapes, the mudras (in Sanskrit,
188 •
“symbolic gestures”). Thus kathakali plays on two registers: the face, which
expresses the emotion of the character in the situations he finds himself, and
the hands and the body, which communicate the narrative aspect of the epi-
sodes.
Gaze plays an important role in many ritual activities. Here is an example
practiced by the Chakyars, a community in Kerala famous for its excellent ac-
tors. On the very first day, as soon as the brand-new moon appears, the stu-
dent sits down to train his eyes upon it. His eyes are anointed with clarified
butter. He moves his eyes over the moon continuously until the heavenly body
disappears. The first day, this exercise lasts about an hour, the time it takes the
moon to cross the sky. The second day, the student sits down at the same time
and applies himself to the same sort of exercise, this time for twice as long,
because that is the period between the appearance and disappearance of the
moon. The third day proceeds the same way. Thus he continues to exercise his
eyes every night, always increasing the length of the exercise. On the fifteenth
day, at the full moon, the student is seated from 6 o’clock in the evening to
6 o’clock in the morning, moving his eyes up and down, left and right, in a cir-
cle and diagonally, from one quadrant to the other, without interruption. Only
at dawn does he stop. The butter has a refreshing effect on the continually ro-
tating iris. This system is known as nilavirikkuka, literally, “seated by the light
of the moon.”
The actors train at least one hour per day for eight years. Orientational
training toward an object is as follows: the eyes are wide open, and the head
turns while observing, as though the eyes were moving the head. Suddenly,
the head stops abruptly, and the eye fixes on an object that is not yet its final
goal. The head remains in this stationary position, while the eyes move (slowly
or rapidly according to intention) toward and then reach the predetermined
goal. Only then does movement of the head follow; and with this movement
completed, the face assumes a particular expression (hatred, contempt, joy,
and so on).
But gaze is not only capture, an instrument of coherence, or an expression
of emotion. It can also be absent, as it is, for example, in autistic children. Uta
Frith describes Peter, an autistic child, as follows: “At last, Peter started to
speak. But language did not open the doors to communication as everyone
had hoped would surely happen. Strangely, he often echoed what other people
said. Peter was quite indifferent to make-believe play or simple group activities
. . . Often the family felt as if there were an invisible wall preventing them
from making proper contact with Peter . . . Most of the time it seemed as if he
was looking through people, not at them.”9
189
The Neural Basis of Gaze-Orienting Reactions
The historical definition of the orientation reaction must thus include pro-
cesses much more general than simple movement of the eyes toward a target.
Pavlov provided one of the first physiological descriptions of the orientation
reaction.10 According to him, the appearance of a new stimulus immediately
evokes a searching reflex; the animal trains all its relevant sensory receptors to-
ward the source of the perturbation: pricking up its ears, directing its gaze to-
ward the source, and sniffing the air. For Sokolov, who devoted many studies
to the neural basis of the orientation reflex, the question is one of a general-
ized state of alertness that is not specific to any sensory modality.11 In the
Soviet literature, the orientation reaction is thus more preparation than execu-
tion; moreover, it involves structures, like the hippocampus, that process infor-
mation at the highest level of the brain and that are particularly sensitive to
novelty.
First off, the orientation reaction involves suspension of any ongoing
movement through an increase in muscular tonicity,12 which follows from
Bernstein’s sense of the word, regarding posture as a readiness to move. It also
involves a lowering of sensory thresholds, probably connected to the nature
of the source and to the selective redistribution of tonicity, to prepare to
move. This lowering of thresholds for certain inputs preselects the senses that
will be used for orienting movements. Two kinds of simultaneous processes
occur next: one is inhibition of cortical resting activity, indicated by character-
istic slow waves in recordings of electrical potentials on the surface of the
skull. (Indeed, researchers recently demonstrated the importance of cortical
and thalamic rhythms associated with states of alert attentiveness and empha-
sized alteration of these rhythms during transitions between waiting states
and orienting movements in the cat and the monkey [see Chapter 1].)13
At the same time as these changes occur in the oscillatory activity of the
brain, and perhaps sequentially, activity increases in numerous centers (for ex-
ample, the cerebral cortex and the colliculus) activated by the hippocampus—
that detector of novelty that compares configurations of actual stimuli with
stored configurations. The excitability of other specialized receptors (for ex-
ample, vision and hearing) is also increased. Moreover, alterations occur in vis-
ceral functions and in functions of the vegetative nervous system.
In this way the entire organism prepares itself for orientation. This re-
sponse to the orienting stimulus disappears with repetition. To explain this ex-
tinction, Sokolov introduced a basic concept in his “neuronal model of stimu-
lus.” That concept continues to play a role in many theories under the names
190 •
“internal model,” “central estimation,” “prediction,” “intrinsic hypothesis,”
and so on. “This neuronal model of stimulus can be conceptualized as a ma-
trix of potentiated synapses that encode the system with reference to proper-
ties of repetitively applied stimuli.” This model registers not only the basic
properties of stimuli, but also their recurrence and co-occurrence.
The orientation reflex thus involves not only short sensorimotor loops,
but also a connection between the neocortex (a basic mechanism for analyzing
signals) and the hippocampus, which detects novelty. The reflex results in a
mismatch—today it would be called an error—between the neuronal model
of stimulus and the external signal. The core idea of this theory is that the ori-
entation reflex must not be considered simply as a motor reaction that directs
gaze or attention toward a target, but as a mechanism that establishes a transi-
tion from one state of the organism to the other.

191
10
V I S U A L E X P L O R AT I O N
Read this line and answer the following question: Does this sentence contain
the letter w? To answer, you have to scan the line with rapid movements of
your eyes, which are called saccades. These gaze movements are guided here
by a high-level cognitive task; they are not simple orientation reflexes toward a
spot of light. Now, look at the word below and focus your gaze on the letter N
in the middle. Without moving your eyes, see if you can find the letter Z in
the word:
anemelectrobackuppedalocutwiNdshowershednailprotectedcycle
You will note that it is possible to examine the letters of the word, at
least those that are close to N, thanks to displacements of attention.1 These
displacements of attention are in reality ocular saccades whose mechanisms
are in part the same as those for exploratory saccades, but whose execution
is inhibited. Is it possible to understand the neural mechanisms of these
gaze movements? I will proceed by detailing the neural basis of ocular sac-
cades.
But before going into the details of how saccades are organized, I will first
examine the role of inhibition.
The Brain Is a Fiery Steed
The brain is like a spirited horse that inhibition handles much the way a horse-
man handles his reins. Refinement of sensorimotor function is in part con-
nected to the appearance, over the course of evolution, of subtle inhibitory
mechanisms, the source of new motor competencies. It is easy to think that
the brain contains only excitatory neurons. Rarely is any other role accorded
to inhibition than that of suppression. This is borne out by the current expres-
sion “to be inhibited” in the negative sense of not being able to do something.
Yet neural inhibition is actually one of the basic mechanisms of production of
movement and its flexibility, and probably the main mechanism of sensori-
motor training. It is also the source of perceptual mechanisms of filtering and
selection and plays a decisive role in certain cognitive functions such as deci-
sion making.
MacKay assigns inhibition the leading role in his theory of mental nodes.2
Autoinhibition is necessary to ensure that centers of integration, called mental
nodes, are not superactivated by the echo of signals coming from nodes lo-
cated downstream in the network that controls action, or by sensory mes-
sages. So stuttering, for example, corresponds to an inability to inhibit such an
echo, leading to repetition or disruptions in language.
Inhibition takes extraordinarily varied forms in the nervous system. Local
inhibitory neurons were the first to be described. For example, the motor neu-
rons of the medulla (which activate the muscles) are inhibited by small local
neurons, called Renshaw neurons, that are activated by the motor neurons
themselves, whose recurring inhibition is their mode of regulation. Also in the
medulla, inhibitory interneurons called Ia inhibitory neurons combine infor-
mation from several sensory receptors with commands from the brain. They
play a fundamental role in organizing movements. In the cerebellum, inhibi-
tory neurons called Golgi neurons or basket neurons control the internal ac-
tivity of the organ. In the thalamus, inhibitory neurons contribute to the pro-
duction of oscillations, and so on. The full list would be long, but it would be
worth drawing up.
One of the first functional roles attributed to inhibition was reciprocal
innervation, that is, braking or blockage of antagonist muscle during a ges-
ture. This mechanism is very simple. When you want to lift your arm, you
have to activate the motor neurons of the biceps, but you also have to inhibit
the activity of the antagonist muscle, the triceps. Reciprocal innervation is as-
sured throughout the nervous system by inhibitory neurons. These neurons
are also found in mechanisms of gaze control. For example, they are found in
the vestibulo-ocular reflex: projection neurons called inhibitory second-order
vestibular neurons induce reciprocal innervation of the muscles of the eyes
(see Chapter 2 and Figure 2.4).
Ito’s discovery of the inhibitory nature of the Purkinje cell in the cerebel-
lum was a major finding.3 Eccles and his students had made a detailed study of
this remarkable cell, which is the sole neuronal exit from the cerebellum. This
VISUAL EXPLORATION •
193
neuron (whose morphology was revealed by the work of Sotelo in Paris,
among others) projects to several structures important in regulating and coor-
dinating movement. Moreover, it is now thought that the cerebellum also
plays a role in certain cognitive activities.4 The fact that the main projection
neuron of an organ as important as the cerebellum is inhibitory demonstrates
the importance of a form of neural activity that researchers of the time
thought could do no more than brake, or simply reverse, the direction of a
command or a signal.
Another form of inhibition probably plays a fundamental role in the selec-
tion of sensory messages that accompany the planning of action: inhibitory
neurons, located in the spinal medulla, inhibit sensory fibers just before they
synapse on the target neurons of the medulla. This presynaptic inhibition is a
gating mechanism for sensory input or modulation of its amplitude. Thus
there are at least two major mechanisms for controlling messages from the
proprioceptive receptors: γ and β commands on the spindles (see Chapter 2)
and presynaptic inhibition. The brain has many ways of selecting sensory
messages.
Cellular mechanisms of inhibition are also extraordinarily complex.
Twenty years ago, this inhibition was thought to result from regulating trans-
port of chloride ion across the neuronal membrane. The inhibitory neuron
produced a hyperpolarization of the neuronal membrane, which distanced it
from its firing potential and, consequently, prevented it from firing. It is now
known that numerous neurotransmitters are responsible for synaptic inhibi-
tion (GABA, glycine, and so on), that several can be present in the endings of a
single neuron, that inhibitory action can take diverse forms via blockage of
numerous ion channels, that two successive inhibitions can be expressed by an
excitation, that a combination of inhibition and disinhibition can have a multi-
plier effect or be used simply to introduce a delay, and so on. Molecular biol-
ogy is currently completely overhauling our ideas about the mechanisms of
inhibition. The work of Korn and his group in Paris also highlighted the pow-
erful role of inhibition in modulating the excitability and plasticity of neural
activity.5
It is impossible to broach this immense domain in just a few pages without
falling into one of two traps, either a surfeit of details or neural tourism. But I
will look at one example: the control of gaze, more particularly, the ocular
saccade, to show how inhibition operates as a mechanism of selection and in
what Piaget called “internalization” of sensorimotor behavior, and in motor
imagination.6
194 •
A Model of Perception-Action Relationships
A saccade is a change in the direction of gaze when the eye turns. It is rapid
(20 to 150 milliseconds) and attains angular velocities of 800 degrees per sec-
ond. It is the fastest movement humans are capable of producing. Each sac-
cade consists of a rotation, followed sometimes by a slight shift to adjust the
axis of gaze on the target, and sometimes other small correctional saccades if
the gaze has not reached its objective.
Look at a point straight ahead of you and make a big saccade toward an
object in the room. You will perceive neither movement nor change of posi-
tion in the room, even though its image has shifted on your retina. This per-
ceptual invariance is due to a mechanism called saccadic inhibition, which
blocks the neural messages from the retina during the saccade and prevents
you from being aware of the shift, seen but not perceived by the retina, be-
cause it remains blocked at the first visual relays. Moreover, a mechanism of
perceptual shift maintains the positional stability of the environment.
I have made the following observation, which I invite you to confirm. If I
focus on an object in the room for about 20 seconds, and then make a big
saccade to focus my gaze on another object in the room, at the start of the
first saccade, I actually have the impression that the world is moving slightly.
But if I repeat it several times, my perception of the world is remarkably sta-
ble. I think that in the latter case, I anticipate the displacement of my gaze,
which spares me the slight impression of shifting. I will examine the mecha-
nisms at the level of the parietal cortex that may account for this stabilizing
anticipation.
The results of the following experiment reproduce the observations of
pioneers such as Purkinje, Bell, Helmholtz, and numerous other physicists and
biologists who were all fascinated by these phenomena. Press laterally, rather
hard, on your eyes to displace them. Don’t be afraid—eyes aren’t fragile! If
you look straight ahead of you, you will see that during this passive move-
ment of your eyes, the room seems to move in front of you. In other words,
when your retinal image of the world is displaced, you have a perception of
displacement of the world. In contrast, when you are actively gazing, the
world looks stable despite retinal slip. Feedforward neural processes associated
with the production of movement thus ensure perceptual stabilization. Many
mechanisms have been suggested to explain this phenomenon. The most plau-
sible is the use of an efference copy or corollary discharge (see Chapter 3 and
Figure 3.4).7
195
These examples show what an interesting model the saccade is for study-
ing the most highly developed mechanisms of active perception. But it gets
even more interesting. Yarbus, a Russian psychologist, was the first (followed
by Stark at Berkeley, Jeannerod in France, and the neurologists Lhermitte and
Chain of the Salpêtrière Hospital) to show that the oculomotor path (the se-
quence of saccades) followed to explore a face is completely different depend-
ing on what the observer is thinking: whether she thinks that the individual is
rich, sad, or well-coifed, that his ears are protruding, and so on. Similarly, in
front of a landscape, the oculomotor path taken depends on what a person is
seeking. To explore a face or an environmental scene requires complex cogni-
tive decisions, so it is easy to see why the saccade is an interesting model for
studying motor selection and decision processes. These paths are altered in pa-
tients presenting with lesions of the brain.
Finally, the production of saccades, like all our actions, requires access to
memory. Visual memory is the first type: I can ask you to look at an image
that follows a precise oculomotor path with a defined rhythm, then, in dark-
ness, to repeat the same path.8 Vestibular memory is another: I can ask you to
focus on an object in front of you on the table, then to turn in your chair in
the dark and not move your eyes; then I can ask you to make a saccade toward
the memorized position of the object. In both cases, the saccades will be per-
fectly executed, which demonstrates the saccadic system’s access to spatial
memory.9 I touched on these mechanisms in Chapter 5.
Next, I will consider the principal stages in producing a saccade. I will not
go into particulars, but will roughly describe the general organization, as a
physiologist does, even though it means losing a little of the fine grain of the
analysis. I will start at the end—the motor centers of the brainstem—and
move little by little toward the cortex. This method, from the bottom up, is
contrary to the apparently more natural approach of examining the successive
stages of image analysis. But I want to show that saccades are indeed executed
hierarchically, which enables gating at different levels by inhibitory mecha-
nisms, and that the same neural structures are brought into play in imagined
movement of the eye as well as in actual movement.10 The brain does not only
carry out sensorimotor transformations; at several levels, motor commands in-
fluence how sensory data are processed. So, starting from the motor com-
mand comes back to thinking of action as an essential element of neural func-
tioning and enables us to study how it organizes perception instead of simply
trying to find out how perception defines action. A saccade is a decision to act,
not a response to a stimulus.11
196 •
Saccadic Generators in the Brainstem
Figures 10.1 and 10.4 summarize some of the neural operations involved in
the control of ocular saccades. My account is, of course, not exhaustive. I will
review the saccadic system briefly and then discuss the individual elements. I
use the word “neuron” to characterize each type of neuron, but what is in-
volved is populations; that is, groups of diffusely or centrally localized neu-
rons, or nuclei.
Let us begin with the last stage of execution: the ocular saccade is actually
produced by the firing of motor neurons, called motoneurons (Mn), which ac-
tivate motor fibers in the muscles of the eyes. These discharge with a very par-
ticular modulation of frequency made up of two successive parts.
First, a very intense burst of phasic activity (that is, a series of action po-
tentials of high frequency and short duration) penetrates the viscous eye. This
allows the saccade to be executed in a few dozen milliseconds. The premotor
neuron, which produces the phasic activity that moves the eye toward its new
position, is called the excitatory burst neuron (EBN) and is located in the
brainstem for horizontal movements (EBNH) and in the reticular formation
of the mesencephalon for vertical movements (EBNV). There are thus two
separate centers for the two types of movement. This is the same geometric
segregation I have mentioned in other chapters. The discharge of EBNs pre-
cedes the start of the saccade by around 10 to 15 milliseconds and projects di-
rectly to the ocular motor neurons. Their immediate frequency is connected
(depending on one’s interpretation) either to motor error or to the velocity of
the eye. Generating an accurate copy of the displacement of the eye simply re-
quires integrating the activity of these neurons (that is, adding up the number
of action potentials). These phasic neurons also project to the local inhibitory
neurons (called interneurons) situated in the vestibular nuclei. The functional
significance of this projection is important: it blocks the vestibulo-ocular reflex
during orienting movements. If not for this inhibition, the eye and head could
not be directed toward a target, because the vestibular-ocular reflex tends to
displace the eye in a sense contrary to that of the head (see Figure 2.4).12 This
is an example of the way in which inhibition achieves automatic coordination
between two motor subsystems that have opposing functions.
Second, a sustained tonic discharge is necessary to maintain the position
of the eye in its socket. The more the eye moves, the higher the tonic fre-
quency: this is the signal of movement that is sent to the motor neurons. The
two saccadic generators that receive a dynamic signal from the brain trans-
197
E
Coordination
∆E of responses SMA SEF
Short-term
memory
5 4 PM FEF Pre Fr
PO 7A LIP VIP +
Str
V4 V3 MT +
GPe
V2 Thal STN
V1 Suppression − + Disfacilitation
SNpr
LGN
Temporal Pul
and spatial
selection
− +
Fix
Dynamic SC
remapping − ?
TRSN P
Motor error −
and desired
eye velocity Cajal
V Eye
Mn
+
EBN LR
Corollary V Ph E Mn
discharge +
H
−−
EBN
H Proprioception
− IBN + ?
RSN
Coupling of
head and eyes
RSN
Nuchal command
NECK
Figure 10.1. Saccadic movement of the eye (lower right) is produced by phasic
excitatory neurons located in the pontine and mesencephalic reticular forma-
tion. The position of the eye is controlled by integrators. Copies of signals
about movement of the eye are sent to the colliculus and to the cortex (arrows
at right) and cause changes in the maps where position is encoded and where
the saccade is initiated. The saccadic generator can be inhibited at a first level in
the brainstem by the pauser neurons (P) and at other levels by a cascade of inhi-
bitions.
form it into continuous activity owing to a local neural mechanism called the
integrator. There is assumed to be one integrator for each of the two types of
movement, horizontal and vertical. These are indicated by the symbol ∫ in
Figure 10.1.
As with the integrator of the vestibulo-ocular reflex, the mechanisms at
work are still not well understood. They may involve properties of the neural
membrane as well as those of neural networks. It is known that if a lesion is
introduced to a small area close to the vestibular nuclei, the eye no longer
maintains its horizontal position after a saccade. By analogy with electronic
circuits, it is said that the integrator leaks. This appears to be a local mecha-
nism. The final position of the eye is thus developed very late in the process,
probably using internal models of the eye. You will see why this is interesting
for imagining movement.
Inhibition Blocks Saccades

This premotor neural circuit, each neuron of which is well defined, is under
the permanent control of a remarkable first-order inhibitory system. Neurons
that fire continuously, whose cellular body is located in the median region of
the reticular formation, inhibit EBNs and normally prevent saccades from oc-
curring. For the saccade to take place, these neurons themselves must cease
firing, or pause, to induce disinhibition of EBNs. Indeed, every time a saccade
is made, these neurons pause briefly. Thus they are called pause (P) or omni-
pause neurons.13 These pausers are a mechanism of temporal selection; they
control when the saccade will occur and for how long. Up to this point, the
signals from the brain are only those of intended movement. Removal of inhibi-
tion enables them to crystallize into orienting and gaze movements.
To remove the inhibition induced by the pause neurons, these neurons
themselves must be inhibited. This inhibition is the job of neurons located in
other structures of the brain that follow two parallel pathways, one descend-
ing directly from the cerebral cortex (prefrontal cortex, frontal oculomotor
fields), the other most probably descending from the frontal (so-called rostral)
pole of the superior colliculus.14
But saccades are not only controlled at the level of the brainstem, close to
the ultimate motor command neurons. Execution of saccades is also blocked
at a higher level, in the superior colliculus (SC). I explained in Chapter 3 that
the colliculus is an important neural structure connecting visual and auditory
space. The neurons of its superficial layers form a sensory map of visual space
in retinal coordinates. The intermediate and deep layers contain tecto-reticulo-
spinal neurons (TRSNs) that project to the brainstem and the medulla (see
199
Figure 3.6). These cells participate in producing orienting movements of the
eyes and the head toward visual or auditory targets.
The precise role of TRSNs in generating saccades is still unknown, but it is
clear that at least some of them fire in a burst of activity that precedes and ac-
companies the saccade.15 They project directly to the EBNs, or alternatively to
other intermediate neurons located in the brainstem, as well as to reticulo-spi-
nal neurons involved in producing the saccade.
The firing rate of each TRSN determines the dynamics of the saccade,16
and the connectivity of its axon determines the saccade’s direction. Horizontal
and vertical saccades are produced by TRSNs that project respectively to the
neurons of the horizontal and vertical motor nuclei; an oblique saccade is pro-
duced by neurons whose axons project to both. Thus the motor synergy re-
quired to produce a saccade is determined in part by anatomy, as the exam-
ple of vestibular neurons in Chapter 7 demonstrates. Geometry is encoded by
anatomy.
The colliculus constitutes the second level of potential blocking of execu-
tion of the saccade because the neurons of its intermediate layers are un-
der the inhibitory control of another formation—the substantia nigra pars
reticulata (SNpr)—which is part of the basal ganglia at the center of the
brain.17 It has recently been established that projection of the substantia nigra
to the superior colliculus is organized in a very precise way: some of its neu-
rons inhibit other very specific populations of neurons. Each collicular neuron
that projects to the saccadic generators induces a movement of the eye in a
precise direction and thus orients gaze toward a defined location in space. This
mechanism implies genuine spatial selection, since it is possible to inhibit sac-
cades made in a precise direction in space. Another group of neurons, located
in the rostral part of the superior colliculus, fire when the eye fixates. Thus, an
independent mechanism exists for fixing the eye in its socket.
What might be called a spatiotemporal decision is thus worked out at the
level of the superior colliculus. It enables displacement of visual attention
from one area of space to another, since, depending on which neural popula-
tion of the colliculus is disinhibited, a person’s gaze will fall on this or that
area of space. This attentional mechanism constitutes motor preparation, as
the expression “motor theory of attention,” originated by Rizzolatti, empha-
sizes.18 This attentional orientation is tremendously complicated; it involves
many multisensory processes and is probably influenced by mechanisms at the
level of the cerebral cortex and even the hippocampal formation.
The third level of inhibitory action is also located in the basal ganglia. The
200 •
neurons of the substantia nigra that inhibit the collicular neurons19 cannot sus-
pend their activity and trigger the saccade unless they are themselves inhibited
by the neurons of the caudate nucleus in the striatum.20 These neurons fire
during voluntary saccades and are also active during saccades toward memo-
rized targets. They are most likely involved in cognitive control of saccadic ini-
tiation.
But the process of saccadic selection is even more complex and subtle. Re-
cent findings reveal that, concurrently with the cascade of inhibitions involv-
ing the caudate nucleus and the substantia nigra, a second cascade of inhibi-
tion and excitation accompanies and reinforces selection. This pathway, whose
target is the colliculus and whose terminus is the subthalamic nucleus (STN),
is made up of an inhibitory neuron, located in the striatum and projecting to
the globus pallidus externus (GPe), which projects to another inhibitory neu-
ron that in turn projects to the subthalamic nucleus. The target neuron in the
subthalamic nucleus is excitatory and projects to an inhibitory neuron in the
substantia nigra that suppresses activity in the colliculus, just as the neurons
of the first pathway do. The STN also projects to the colliculus.
What is important is that the basal ganglia are not just components of
simple command pathways: they form part of the internal circuits that con-
nect the cortex, the basal ganglia, and the thalamus (Th), and that feed back to
the cortex. In Chapter 1 I mentioned the possible role of these circuits and the
sustained oscillations that traverse them in the development of mental repre-
sentations. Hikosaka recently asserted that signals connected to a reward can
influence them (Figure 10.2).21 According to Hikosaka, the basal ganglia play a
coordinating role among the different cortical areas involved in developing
motor strategies. These areas are functionally connected in the course of
training based on reward. The limbic system signals the basal ganglia as to
whether the chosen movement will lead to a reward. This method of selection
is thus even more complex and subtle than a simple gating mechanism at the
level of the brainstem or spatiotemporal selection at the level of the colliculus
(Figure 10.3).
The pleasure taken in looking at something, or the fear of what might be
seen, or how interesting an object is—all contribute to the selection of sac-
cades by these parallel cascades of inhibition and excitation.
It is not possible to describe here the totality of structures that contribute
to the development of saccades. For example, it is clear that the cerebellum
plays an essential role in regulating motor commands that modulate the am-
plitude, velocity, and duration of saccades. Lesions of the cerebellum cause
201
Brain 1
Motor
program Motor
outputs
Basal ganglia
Reward
Sensory inputs
Brain 2
Motor
program Motor
outputs
Prediction
of Basal ganglia
reward
Internal model
Reward
Sensory inputs
Figure 10.2. Evolution of the basal ganglia. In primitive animals (Brain 1), the
basal ganglia activate the elements of a genetically programmed repertoire of
movements. The efficiency and success of the movements are enhanced by in-
centives that adjust these behaviors under similar conditions. In higher animals
and in humans (Brain 2), the behavior is more complex. A new mechanism
(which probably involves the prefrontal cortex) makes predictions based on an
internal neural model that simulates the different possible movements and eval-
uates their chances of success. The consequences of the resulting behavior are
compared with this prediction and may modify the internal model.
202 •
Associative cortex
Limbic system
Cerebellum Motor Basal

Behavioral
memory ganglia
set
Motor cortex
Learned movements Innate movements

Fingers-hand Eye-head
Locomotion
Chewing
Vocalization
Figure 10.3. The basal ganglia coordinate interactions among cortical areas.
When first learning new movements, the different cortical centers act indepen-
dently. Their activity converges in the basal ganglia, which are sensitive to their
interaction. When a movement has been rewarded, the limbic system pro-
motes the successful combination in the neuronal synapses of the basal gan-
glia. As the new combination of actions produced by the cortical centers is re-
inforced, a new behavior is learned.
many oculomotor signs. The reticular neurons that control saccades are mod-
ulated by the cerebellum, itself involved in regulatory circuits that originate in
and return to the cerebral cortex. Remember that the motor exit of the cere-
bellum is inhibitory (the Purkinje cells are inhibitory). Once again, inhibition
is involved in the control of movement.
Thus, concurrently with the hierarchical cascade of inhibitions of the
pausers and basal ganglia, there exists a supplementary regulatory mechanism
based on inhibition. Moreover, subdivisions of the cerebellum play different
roles in controlling the movement of eyes. The vermis (the median portion of
the cerebellum), which dates far back in evolution, probably regulates auto-
matic movements, and the lateral portions (the neocerebellum), which are
connected to the frontal cortex, probably regulate more complex gaze move-
ments, though this is pure speculation. If, as recent theories suggest, the cere-
bellum contains internal models of motor apparatuses and also facilitates the
temporal coordination of movements, this structure, together with the cere-
bral cortex, is crucial in preparing gaze movements.
203
Cortical Control of Saccades
To examine the contribution of the cerebral cortex in controlling saccades, I
will start with the retina. Two major pathways transmit information from the
retina to the colliculus: one goes directly from the retina to the superficial lay-
ers of the colliculus; the other projects the image of visual targets toward the
visual cortex through the intermediary of a relay called the lateral geniculate
nucleus (LGN) (Figures 10.1 and 10.4). Interestingly, neural activity connected
to the saccade itself has been found in this first visual relay. As with the vestib-
ular nuclei, the motor command modulates sensory inputs. In other words,
action modifies perception at its source. The brain is not constructed of simple
systems that transform sensory signals into motor commands, but of closed
circuits.
This reentrant information—these copies of the motor command that
represent eye movements—is perhaps used as a substitute for sensory inputs.
The movement of the eye is equivalent to that of a target, since moving my
eye 20 degrees to the right amounts to the same thing as moving the target 20
degrees to the left.22 Information on the position of the target is next transmit-
ted to primary visual area V1, from whence it activates the neurons of areas
V2, V3, then the parietal cortex, a very important relay in developing saccades.
The parietal cortex is an area of the brain that is essential in representing
space and relationships between the body and the external world.23 Although
from the early part of the century the parietal cortex was conceded a basic
role in constructing the body schema, opinion is still divided between theories
that further define that role as directing visual attention and others that see it
as planning movements. The parietal cortex is part of the dorsal system of vi-
sual information processing that is concerned with localizing objects, their
movements, and the relationships between objects and intended action. The
parietal cortex is not only involved in localizing objects (knowing where they
are) and their relationship to the body; it also contributes to preparing actions
(knowing how to grasp objects and handle them). Lesions of the right parietal
cortex cause hemispatial neglect (see Chapter 3).
Neuronal recordings reveal the involvement of the parietal cortex in visuo-
spatial behavior. Some authors posit that a change of coordinates occurs in the
parietal cortex that reconstructs the position of the target in space by sum-
ming its position in relation to the eye with that of the eye in relation to the
head and of the head in relation to the body, and so on.
The parietal cortex is divided into specialized zones connected to particu-
lar areas in the frontal cortex. Some are more specifically involved in preparing
204 •
V3a Parietal cortex
LIP
V1 L
7a
Occipital cortex V2
7b
CN
SN
-
LG
-
N
-
SC Tha
l
-
SEF FEF
-
Frontal cortex
-
PFC
Prefrontal cortex
EBN
Figure 10.4. The principal production pathways of the ocular saccade. The first goes di-
rectly to the superficial layers of the superior colliculus (SC), where it is transferred to the
deep layers whose neurons project to those that produce the saccade (excitatory burst neu-
rons, EBN). A second pathway traverses the lateral geniculate nucleus (LGN) in the
thalamus and projects to the visual areas of the occipital cortex (V1, V2, V3). From there,
the information is transmitted to the parietal cortex, where it is combined with informa-
tion concerning the movements of the head and body in space and with internal signals
(memory, general context of action, and so on). The centers of the parietal cortex them-
selves project to the supplementary eye field (SEF), which coordinates sequences of sac-
cades, and to the frontal eye field (FEF), located in humans in the precentral gyrus. The
precentral gyrus projects to the SC and also directly to the brainstem. The prefrontal cor-
tex (PFC) plays a role in the most cognitive aspects of the production of saccades, and a
mechanism containing a cascade of inhibitions facilitates their selection. It involves the
caudate nucleus (CN) in the striatum and the substantia nigra pars reticulata (SNpr), which
projects to the colliculus.
205
gaze movements, in particular saccades or ocular pursuit. The intensity of
their neuronal responses depends on the direction of gaze.24 So a signal of de-
tection of the position of the eye in its socket from the depths of the brain—
from the neural integrator of the brainstem—is added to the activity of the
neurons caused by vision. The position of the target in space is also encoded
by other visual areas.25 Recent findings have shown that this area is the center
of multisensory convergence where visual, vestibular, and proprioceptive mes-
sages are combined. The neurons of these areas contribute to preparing and
directing the saccade; they fire just before its initiation. Some say that the pari-
etal cortex forms and maintains spatial coordinates that enable localization of
objects and determine their position in space,26 and that the prefrontal cortex
(PFC) uses this knowledge to direct a response; others say that movements are
encoded in coordinates connected to the head, in particular to the eye itself,
and do not require any internal reconstruction of the position of the target in
spatial coordinates.27 Nevertheless, the question of the frame of reference in
which this system functions has yet to be resolved.
The lack of activity of certain areas of the parietal cortex during voluntary
saccades in the dark shows that they are mainly involved in visuospatial pro-
cessing. However, using positron emission tomography (PET), we have dem-
onstrated for the first time that they are also activated (together with frontal
areas) in humans during tasks of memorized saccades carried out in the dark.28
Successfully performing this task requires accessing the spatial memory that
recorded the position of the targets, and recalling their order; thus it means
producing a mental image of their positions to make a sequence of saccades.
Patients with lesions of the parietal cortex make less accurate, delayed sac-
cades.
Without summarizing the properties of the parietal cortex, which may
also be involved in the memory of movement and in spatial memory,29 in deci-
sions connected to the perception of movement,30 and so on, I will once again
consider the essential role of anticipation and prediction in the organization
of the brain. Indeed, I have shown from the outset that the senses detect the
derivatives of sensory signals, that in this way they can anticipate future states.
They can do this because motor signals modify sensory inputs, which alters
perception based on action, and inhibition selects messages based on intended
movement. But there is something even more remarkable: the nervous system
can ensure the stability of the perceived world during movement by updating
visual frames of reference in advance.
When the eye is stationary, visual objects are picked up by various neurons
of the visual system according to where the objects are in space. It is said that
206 •
these neurons have receptor fields (see also Chapter 3). These fields are small
for the first relays of the visual system (area V1), but become increasingly
larger over the course of successive stages of visual processing (areas V2, MT,
MST, and so on). Up to now the question was thought to be one of the rigid
properties of the neurons of the visual cortex, with fixed coordinates in the
retinal space. The localization in space of the receptor field of a neuron of the
parietal cortex is abruptly displaced before the saccade, in the direction of the
gaze movement about to be made.31 In other words, before the saccade that
leads the gaze from one point in space to another, the brain displaces the re-
ceptor field as if it wished to anticipate the consequences of movement.
The advantage of this transitory anticipation is that the retinal representa-
tion of the world remains stable during movement. In other words, the brain
places itself in the final reference system, adopting the vantage it will have on
the visual world even before making the movement.
More than 20 years ago, an experiment done in our laboratory predicted
these findings, which were obtained recently in the monkey.32 You can try this
experiment yourself. Look at the upper left corner of this page and make a
single saccade toward the right lower corner. If you hold the book in front of
you, the saccade will have an amplitude of about 40 degrees. Your eye will
move about 400 degrees per second and thus cannot make use of vision be-
cause of the considerable shifting of the image on the retina. We observed
that subjects could grasp words, although their eyes always moved faster than
200 degrees per second (which we recorded precisely using a magnetic field
technique that I introduced in France) even when they slowed down in the
middle of the page. We concluded from this that an active process of displace-
ment of the receptor fields enabled the brain to compensate for the image’s
shifting on the retina via an anticipatory mechanism connected to the in-
tended movement. In the 1950s, Verguiless and Zinchenko in Moscow spoke
of the “functional fovea.”
This is why I keep saying that the boundaries between sensation and mo-
tor function are disappearing. For the same reason, an internal copy of a
movement command signal can be used as a signal given by the senses, and is
itself influenced by movement from the first central neurons and, even, at the
periphery.
It does appear that the information needed to produce simple saccades to-
ward a visual target are transmitted from the parietal cortex to the frontal eye
field (FEF), which projects directly to the colliculus.33 Ferrier,34 then Penfield,35
showed that electrically stimulating the frontal eye field produces eye move-
ments. In the monkey, the frontal eye field contains several types of neu-
207
rons that fire in connection with saccades, ocular fixation, pursuit, and so on.
In humans, its activation during saccades was demonstrated through sev-
eral pioneering efforts.36 Its localization in the precentral gyrus is now well es-
tablished.37 Generally, this area is activated during all voluntary saccadic move-
ments.
The importance of this area is not due only to the fact that it projects di-
rectly to the colliculus. It is part of an internal circuit that I have mentioned
several times as being one of the neural bases important for the internal simu-
lation of movement. In fact, a circuit connects this cortical area with the basal
ganglia and the thalamus.
Another area important for anticipation and prediction in developing
movements is the supplementary motor area (SMA), located in the median
portion of the frontal lobe, in front of the actual motor areas. Its location was
studied by recording electrical (EEG) and magnetic (MEG)38 activity and by
PET.39 It is also important to remember that it is part of the circuits that con-
nect certain structures of the basal ganglia and the thalamus. Its function is
thus associated with theirs, and even with that of the cerebellum. It also func-
tions in connection with the parietal, prefrontal, and cingulate cortex.
The neurons of this area are arranged in a sort of map of the body as in
the motor cortex (somatotopy). For example, in humans and in monkeys, fa-
cial movements are obtained by stimulating the anterior portion. This area
may command groups of muscles and complex motor synergies. It is involved
in global actions, since it initiates, for example, activity of the muscles of the
left hand only when the movement mobilizes both hands. It also comes into
play when the subject must make a choice, or decide whether to make a move
(go–no go). Moreover, it is activated during sequences of movements.
In humans, lesions of the SMA cause alternating movement deficits of the
limbs, mutism, disordered language, and impairments in gestural sequencing,
bimanual coordination, memorized movements, and especially endogenously
produced movements (movements that are freely constructed by the brain
even in the absence of external stimulation). These findings have been con-
firmed by EEG recordings taken on the surface of the human skull. In the area
of the cortex that corresponds to the SMA, a slow electric potential (readiness
potential, discovered by Kornhuber and Deecke), precedes a simple finger
movement by about a second and a half. This area is strongly activated when
the subject is asked to do a bimanual tapping task with a different rhythm for
each hand, as in playing a piano piece.40 It scarcely registers in simpler move-
ments, which mainly activate the motor cortex.
It has been suggested that the motor cortex contributes primarily to en-
208 •
dogenous movements, whereas the premotor cortex contributes to move-
ments caused by sensory signals. This distinction could be profound and may
go far back in evolution. In the mammalian family tree, the SMA has its origin
in the hippocampus and a limbic root in the anterior cingulate cortex. The
premotor cortex probably had its origin in the piriform cortex and its root in
the insula. Subcortical inputs from the SMA mostly come from the basal gan-
glia via the thalamus, whereas the premotor cortex receives its inputs from the
cerebellum. The SMA connects to several areas of the cortex involved in per-
ception of space and of the body. This distinction indicates a possible special-
ization of these areas in endogenous movements for the SMA and exogenous
movements for the premotor cortex.
Whereas the posterior portion of the SMA is clearly involved in coordinat-
ing movements, the anterior portion contributes to more cognitive aspects.
The anterior portion (pre-SMA) is connected with the prefrontal cortex and
receives inputs from area 46 (anterior premotor cortex) and the anterior
cingulum involved in intended movement and attention, and perhaps in train-
ing sequences of new movements (playing a Bach fugue, or dancing choreo-
graphed steps). Electrical or magnetic stimulation of the pre-SMA arouses an
intention to move; subjects say, “I feel like moving.” Recent findings obtained
by PET suggest that the anterior portion of the pre-SMA is activated during
imagined movement of the fingers, and the posterior portion during the ac-
tual movement.
There appears to be a further subdivision of the SMA itself unique to gaze
movements. Stimulating this area produces ocular saccades that instead of
having a fixed direction, all converge toward a single point.41 In the monkey,
this area is called the supplemental eye field (SEF). Patients with lesions of the
SEF present with problems in executing sequences of two or three saccades
toward memorized visual targets,42 as well as saccades toward a memorized
target after a rotation; that is, after vestibular stimulation.
The SMA is also connected to the basal ganglia and to the thalamus via a
neural circuit that is probably involved in establishing new movements and in
assessing the rewards or outcome of an ongoing action (see Figures 10.2 and
10.3). A special circuit for eye movements exists among the basal ganglia-
thalamo-cortical loops specifically concerned with different motor functions.
Under the frontal cortex, the motor and parietal cortex are connected to a
region of the cortex called the cingulate cortex, whose anterior portion is in-
volved in attention and in perceptual decision making. It is associated with the
prefrontal cortex and perhaps also is involved in emotion. The median portion
has more to do with motor function, and its neurons probably contribute to
209
organizing movement of the hands, eyes, and head. In humans it is activated
during saccades43 and during endogenous (that is, produced by the subject)
movements of the fingers even in the dark, and thus contributes to creating
movements based on memory. The posterior portion is involved in perceptual
reorganization after a movement, or even following corrective control. In fact,
the neurons of the posterior cingulate cortex fire after ocular saccades, and
their activity depends on the characteristics of the preceding saccade.44 Thus
this region may contribute to integrating the consequences of action in a sen-
sory or motor context.
I have examined the role of the brainstem as an ultimate controller, the
colliculus as a visual-oculomotor interface, the parietal cortex as a predictor
and spatio-temporal transformer, and the frontal areas as preparatory centers.
I will now consider the role of the prefrontal cortex in the higher cognitive op-
erations that control gaze orientation. Findings from neurophysiology show
that in the monkey, neurons of the prefrontal cortex (PFC) are involved in orga-
nizing ocular saccades that are simple but are carried out with a delay.45 They
are also involved in training for sequences of movements or actions that obey
a rule,46 or require either memorizing a context or deciding among several
movements, which necessitates selective attention combined with memory.47
This function of representational memory and of work is thought to be one
of the main functions of the prefrontal cortex.48
Patients with lesions of this portion of the cortex have trouble refusing re-
quests—controlling themselves, one might say. They also have trouble making
decisions and especially changing the rules of conduct once they have decided
to act according to a certain rule or have established a plan of action. These
patients are often deficient in performing tasks that require making saccades
toward visual targets they must memorize49 or when they are expected to
make a saccade that corresponds to a rotation in the dark; that is, one based
on vestibular information (see the protocol for vestibular saccades described in
Chapter 5).50
They also fail at so-called antisaccadic tasks, which consist in making a
saccade in a sense contrary to that of the illuminated target.51 (Try it yourself:
Ask a friend to strike a match on your right and force yourself to direct your
gaze with a single saccade in a direction opposite to the match; for example, to
the left if the match is on your right.) Moreover, in the monkey and in hu-
mans, prefrontal regions are involved in the command not to execute a move
(no go).52 Thus, there are probably mechanisms for selection and control of
gaze orientation at this level that could be identified by imaging. Connections
210 •
of the prefrontal cortex with the hippocampus, the parietal cortex, the SMA—
all the areas involved in the voluntary production of saccades—suggest that a
cortical system is involved in the construction of exploratory paths taken
when we look at a visual scene, but probably also in spatial memory in gen-
eral.53 In any case, the prefrontal cortex surely contributes to the mental simu-
lation of movements and displacements.54
So the prefrontal cortex is involved in processes of decision in the course
of “voluntary” actions,55 and more specifically when we alter our gaze based
on affective or social constraints. Impairment of gaze organization observed in
persons with schizophrenia (for example, distraction and deterioration of the
ability to make predictions) are perhaps attributable in part to dysfunctions of
the prefrontal cortex.56 Understanding these mechanisms, which are funda-
mental to the development of thought and reasoning, will require new experi-
ments using brain imaging in humans and neural recordings in animals.
Imagined Movement and Actual Movement
Now I will present the arguments supporting the idea that inhibition plays an
essential role in internal simulation of action. But first I will summarize. First
to appear were mechanisms that block saccades by means of pauser neurons;
that is, a blocking mechanism of the collicular orienting reflex to prevent the
animal from making a saccade the instant a visual target appeared. Then inhi-
bition enabled blocking of the vestibulo-ocular reflex. Both mechanisms were
required for a saccade to occur.
Next came inhibitory control by the substantia nigra of discharges of the
neurons of the colliculus. This control was more subtle, because it allowed
genuine selection of the part of the visual field toward which a saccade would
be made. Moreover, inhibitory cascades in the striatum probably enabled
working out the decision to make a saccade based on many other criteria im-
portant for the animal. At this stage of evolution, it was already possible to
simulate saccades without having to make them. Thus, displacements of at-
tention are saccades that are blocked at the lower premotor levels. I have pro-
posed a theory of the hierarchical organization of inhibitory mechanisms to
explain the common mechanisms of executed and imagined or simulated
movements, and to account for the possibility of selectively blocking the exe-
cution of movement at several levels of the sensory motor system. This
theory suggests that there is not a simple dichotomy between imagined and
executed movements but a hierarchical superposition of different levels of
211
possible blockage. This theory extends the motor theory of perception advo-
cated for eye movements by Rizzolatti and colleagues.57
Development of the cerebellum probably enabled modulation as well as a
choice of coordination between the saccade and other motor systems. Finally,
the entire neural machinery of the cortex was put into place. When the
colliculus is inhibited and the pausers are active, the set of cortical structures
can prepare and simulate oculomotor paths: no saccade is observed. Inhibitory
gating of execution thus permits the brain to imagine displacements of gaze
without carrying them out. You can see for yourself that you do indeed have
this option. Focus on a point in front of you and displace your internal gaze—
your attention, as some would say, your “functional fovea,” as Verguiless
would say. You will have the undeniable impression that your gaze is displaced
from one point of the room to the other.
The same neural structures thus ought to be used for imagined move-
ments and for actual movements, since it is possible to gate execution at differ-
ent levels without suppressing the functioning of internal circuits in which
gaze movements are developed and simulated. To check this assumption, we
compared the cortical areas activated during actual saccades and imagined sac-
cades.58 We measured ocular movements to make sure that the amplitude of
the saccade remained insignificant during the imagined movement. The sub-
ject was asked to focus on a point of light in front of him and, first, to make
voluntary horizontal saccades, then, with the target obscured by the dark, to
make imagined saccades similar to the actual ones. Under these conditions,
the oculomotor field, the SMA, and the cingulate cortex were activated during
the imagined movements just as they had been during the actual saccades.
So perhaps the motor theory of attention is true. When you look at the
long word that appears at the beginning of this chapter, focusing on the N in
the middle and looking to see if there is a Z, you bring into play the same cir-
cuits as those that are active during an actual saccade. But there may be alter-
native interpretations. Only experimentation will tell. I mentioned earlier that
there were possibly differences in the roles of the pre-SMA and the SMA.
Dynamic Memory and Predictive Control of Movements
Although the structures involved in orienting gaze are beginning to be under-

stood, their description in terms of general algorithms that describe the pro-
cesses involved remains controversial. In particular, it is obvious that the firing
rate of many premotor neurons is connected, to different degrees, to accelera-
tion, velocity, or the angular position of the eye. Moreover, the problem has
212 •
been restricted to one dimension and to small movements to ensure the valid-
ity of a linear approximation of the processes. Finally, and more generally, the
question of neural mechanisms that underlie the control of movement as sim-
ple as the displacement of the arm from one point to another has not been re-
solved: Which is controlled, the final position of the arm, or the balance of
the contractions of agonist and antagonist muscles?
The solution to these various problems extracts from them a general con-
cept that makes it possible to account for observed neural activity and at
the same time to explain the control of orienting movements. It derives from
a model conceptualized by Droulez called dynamic memory, mentioned in
Chapter 5. This model is too complex to describe briefly. But I can give some
idea of its essential properties.
The novelty of this model is twofold. First, it shows that the brain does
not need to calculate spatial coordinates of the targets toward which saccades
are made. Automatic mechanisms that use the motor command itself update
the neural maps on which the image of the target is projected. Moreover, the
saccade can be predicted even if the target disappears, owing to a mechanism
of neural memory that maintains neural activity after the disappearance of
the target.
It also predicts the results of a critical experiment, during which the au-
thors diverted the course of a saccade unbeknownst to the animal by means
of central electrical stimulation.59 Despite this perturbation, the eye reached
the target, which suggests that the brain encodes movement in spatiotopic co-
ordinates (connected to space). We have shown that dynamic memory makes
it possible to perform such manipulations without reconstructing a represen-
tation of the target’s location in space in the brain.
Dynamic memory has another property. Its output can be used to make a
move using different effectors. Indeed, orienting movements or movements of
capture can be achieved with the eye only, or with the head, the trunk, or even
the whole body. The model predicts that the orienting movement can be
specified in terms of velocity in a rather general way, before it is executed by
the eye, head, or a limb, owing to local integrators that determine changes in
position of the effector based on its biomechanical characteristics. This prop-
erty is very general, since it is easy to write a letter using a finger, hand, or
foot. Very few current models include this property of motor equivalence.
Our attempt was, of course, only a first step, but its formalism complements
the empirical study of sensorimotor processes because it makes experimental
prediction possible. In particular, it assumes that the brain is not a Cartesian
machine that reconstructs the world in spatial coordinates. There is no little
213
geographic map inside the brain that details places; instead, the brain notes
discrepancies and updates the activity of its own internal maps in the course
of movement. The model of dynamic memory is also a model of dynamic
updating.
Was Piaget Right?
It is interesting to compare the facts and theories explained in this chapter

with the theory proposed by Piaget in 1949.60 “The entire evolution of intelli-
gence in the infant is characterized by the internalization of effective actions
into represented actions and reversible internal processes.” He notes that be-
tween years 2 and 7 mental images remain strictly dependent on movements
and preserve a motor component. The first mechanism that he considers is
straightforward inhibition of action. He cites Bain: “To think is to keep from
moving.” I have shown, in the case of the ocular saccade, how inhibition can
block execution of action. This first level, which enables internalization of ac-
tion, is in fact present in the neural basis of gaze.
“A second stage,” says Piaget, “is reorganization of motor function at
around 7 years. Until then, after each representation, the child acts on objects
and perceives the results of his action. This is a semi-internalization.” Between
7 and 12 years, the question is no longer one of connections between afferents
and efferents, but between associative pathways. Piaget predicts the appear-
ance of combined mechanisms of orientation and of inhibition (which he also
calls “channeling” and “blocking”). In this description, I see a possible analogy
with the high-level mechanisms that enable selection of collicular neurons
by the inhibitory cascade of the substantia nigra and the caudate nucleus;
the difference between imagined movement and actual movement is thus not
a dichotomy. A gaze movement, when executed, also entails a component of
internal simulation, as our scheme from 1985 (see Figure 1.5) suggests. Imag-
ined movement can bring into play several levels of the cascade of hierarchical
mechanisms. There is not just one kind of imagined movement; there are sev-
eral nested degrees of imagination and of execution.
These findings confirm the fact, discussed in the first chapter, that to per-
ceive an object is to imagine the actions its use will involve. Similarly, listening
to music has the same effect as singing it or playing it. In 1983 I had a series of
talks with Melvill-Jones, who pointed out that as a member of a choir he had
learned hymns simply by listening to them. Following these talks, and after
having observed my daughter Rebecca learn the violin by the Suzuki method,
which enables a child to play pieces just by listening to them on cassettes and
214 •
singing them—without having to figure out the notes—I joined a team of psy-
chologists in a study of how violinists are trained. I proposed the idea that
singing provides the brain with a general dynamic template of the musical
melody, which can then be transferred to the coordination system of the hand.
This work is in progress, but preliminary results do indicate a greater facility
in going from singing to violin playing than the reverse.
215
11
BALANCE
What a proud look the young child displays to his admiring relatives the first
time he stands on his own: in an instant he has summed up several million
years of evolution, definitively freeing his hands for the use of tools and feats
of manual dexterity and art. Suddenly, escaping the caves of early infancy,
head in the stars, Caliban turned Ariel due solely to the play of muscles that
control posture, he triumphs over gravity, which up to now held him to the
floor and obliged him to sit down to touch objects and beings. What mecha-
nisms underlie this achievement?
The discovery by Sherrington and his school of the mechanism that causes
a muscle to contract when it is stretched led to a view of balance as the link-
ing up of a multitude of local reflexes: when the body bends under the effect
of gravity, muscle receptors as well as those of the joints, vestibular system,
and even vision detect this bending and cause a muscle contraction that rights
it. In the West, this conceptualization flourished until the 1950s, concurrently
with the spectacular growth in techniques of analytical neurophysiology that
described links between sensory receptors and muscles. Each time neuro-
physiologists stimulated a sensory ending, they obtained a muscular contrac-
tion with a more or less prolonged latency, which reinforced the idea of pos-
tural organization based on servomechanisms that detect errors through the
sensory receptors.
In reality, control of posture and its coordination with movement also
bring into play very different mechanisms of these reflex loops. Remarkable
biological solutions simplify this eminently complex problem. Mechanisms of
anticipation and internal simulation were devised and make maintaining bal-
ance a superb model for studying reflexive and cognitive cerebral functions.
Yet my teacher and friend, the neurologist Pierre Rondot, used to say that
as late as 1970 the postural manifestations of the major neurological ill-
nesses were still interpreted by imperial fiat: in the consulting rooms of the
Salpêtrière Hospital, the Mecca of French neurology, the “prince,” that is, the
senior consultant, explained the signs of this or that disturbance of balance
based on the wealth of findings established by the French neurologists Du-
chenne de Boulogne, Déjerine, Thomas,1 and so on. This qualitative semiology
is still viable, for it is remarkably effective. It would be interesting if today, in
cooperation with clinicians, a quantitative semiology of disturbances of pos-
ture and coordination of posture and movement were to be established.
A Physiology of Reaction
Engineers were the first to be interested in the quantitative study of posture

and balance. They applied methods of analysis derived from servo systems de-
veloped during the Second World War by specialists in radar and electrical
motors. A new science of automatics was born that provided a basic theoreti-
cal framework for reflex theories of postural control.
This approach is good for so-called linear systems—those whose output
fluctuations (for example, tilting of the body) are connected, in a straightfor-
ward way, to disturbances of input (for example, displacement of a movable
platform on which a subject is positioned). Relationships between inputs and
outputs are expressed in the form of equations called transfer functions. This
is how gain (the relation between the magnitude of variation of input and that
of output) and phase (the delay in variation of output compared with that of
input) are defined. In Chapter 2 I discussed how these techniques are used to
describe the anticipatory property of the myotatic reflex, which compensates
the slowness of muscle reaction by phasic anticipation. Similarly, some 30
years ago the English physiologist Merton developed the concept of servo-as-
sistance to model the control exerted by γ motor fibers on the neuromuscular
spindles.
The use of movable platforms was also a powerful tool in understanding
the mechanisms of balance under natural conditions. First studied was the
contribution of proprioception to the control of balance. For example, selec-
tive anesthesia of the nerves that transmit tactile information from the feet
helped to demonstrate that a dog placed on a movable platform had difficulty
resisting perturbations of balance. Thus it was possible to study posture from
a functional point of view using established methods. The path lay open to a
new physiology of balance.
BALANCE •
217
How to Make the University of Edinburgh Oscillate
Researchers then examined the contribution of vision to balance. Although its

role was certain, its mechanisms were unknown. A decisive step forward was
taken the day Lee hung a large box from the ceiling of the amphitheater of
the University of Edinburgh.2 The box measured 3 meters on a side, and was
open at the bottom and on one side. He placed a subject upright within the
box and gently oscillated it like a large pendulum, front to back. Instead of
perceiving the box moving around him, the subject perceived it as unmoving
and the amphitheater to be oscillating. In the present of a relative movement
between the box and the ground, the subject’s brain concluded that the
ground was moving, not the room represented by the box. He made a genu-
ine perceptual decision among the possible interpretations.
Lee observed a second phenomenon: simultaneous with the subject’s per-
ception of self-motion in relation to the visual world, judged to be non-
moving, the subject’s body began to oscillate from front to back with the box.
These oscillations could have been the result either of a direct effect of vision
on muscle tone, or of a reinterpretation of the subject’s perception of his
body’s vertical in space. Lee opted for the first possibility and suggested that
vision plays a proprioceptive role.
These results were subsequently confirmed to be rotatory movements of
the visual world.3 Subjects were placed before a disk spinning in the frontal
plane, at the center of which was a small line that could be tilted in relation to
the vertical. You can easily try this experiment yourself with a bicycle wheel
placed at the height of your eyes. Turn it with your hand or with a small mo-
tor. If you stand upright in front of the wheel, you will perceive a rotation of
your body in a sense contrary to the movement. This circular vection in the
frontal plane is mitigated somewhat because the otoliths and the tactile mes-
sages from the soles of your feet contradict the visual messages. You can re-
move this contradiction by lying down on your back and positioning the
wheel horizontally: now the illusion that your body is rotating in the horizon-
tal plane around your head becomes more intense.
When you are upright, the slight vection is accompanied by a very strong
alteration of the vertical subjective. This alteration is made obvious in the fol-
lowing way: the subject is upright in front of the revolving disk whose center
is level with the subject’s eyes. At the center is a smaller disk 20 centimeters in
diameter, on which a line is painted. This disk is stationary, but the subject can
turn it manually using a button. This enables the subject to tilt the line at will.
He is asked to place the line in such a way that it is parallel with the vertical he
218 •
perceives. Recall that he sees only the revolving disk and has no view of the
environment; he must thus rely on an internal representation of the perceived
vertical, which is called subjective.4 In general, subjects tilt the line 5 to 20 de-
grees. This readjustment of the perceived vertical is accompanied by tilting
the head and the body in the same direction.
The visual stimulus thus simultaneously induces an illusion of movement,
a perception of change in the direction of the vertical, and a postural readjust-
ment. This association of perceptions and motor effects is, to my mind, one of
the most convincing indications of the intricacy of perceptual and motor
mechanisms as well as of the fundamental role of perception as the source of
postural reorganization. Posture is controlled by perception and not by local
reflexes. Vision does not act only as a proprioceptor; it influences the percep-
tion of the vertical which, in turn, induces postural changes, as proposed long
ago by Gurfinkel in Moscow.
The influence of vision on the control of posture is decisive in infants of 2
to 3 years. They fall down when the walls of the box are displaced only a few
centimeters.5 If a seated baby is placed in front of a disk revolving around the
axis of gaze, the baby leans to the side when the disk turns. This effect in-
creases up to the age of 2 years, providing evidence of the fundamental role
of vision in the development of postural control. It diminishes gradually, dis-
appearing around age 16.
I think that this diminution of visual dependence is connected to the grad-
ual installation of complex methods of postural control based on internal
models that make vision less dependent on the sensory environment. But vi-
sion reassumes its role if the other sensors controlling posture are impaired.
For example, patients presenting with accidental or surgical lesions of the ves-
tibular system are dependent on vision for control of posture.6 It is not always
possible to detect the disturbance simply by asking them to close their eyes;
sometimes they have to be subjected to a situation called stabilized vision. Illu-
minated boxes are placed over their heads so that vision is activated, but the
vertical is not signaled. Using this same trick, we discovered that in the ab-
sence of gravity on board space stations, astronauts recover their visual depen-
dence during the first moments of flight when they have to recalibrate mecha-
nisms of postural control.7
In the late 1970s, to better study the postural readjustments induced by
the movement of visual scenes, we equipped subjects with a portable mirror
device,8 derived from the one that enabled me to study vection (Figure 11.1).9
These mirrors reflected visual scenes moving from front to back or from back
to front, projected onto the ceiling. In this optic tunnel, the subject had the il-
BALANCE •
219
s
Vi S
s s
Go S1
G'
PROJECTOR x θ
RO O R1
mgk
θ
Figure 11.1. Device for studying the influence of visual movement on the con-
trol of posture. The subject is upright. Visual scenes in motion are projected
onto a screen, S. The scrolling velocity of the scenes (Vi) is controlled by a com-
puter. Mirrors create virtual screen images (s). In this way, the subject is in a vir-
tual tunnel and has the impression of moving forward or backward according
to the direction of visual movement. The tilting of the body that results is mea-
sured by potentiometers located at his ankles.
lusion of moving forward or backward in a sense contrary to the motion of

the scene. It was a primitive virtual reality machine. The illusion was accom-
panied by general tilting of the body in the same direction as the motion of
the scene. Training of the body by vision and by changing the perception of
the vertical occurs at very low frequencies—less than one oscillation per sec-
ond—of visual movement and depends as much on the velocity of movement
of the scene as on its spatial composition. Training is reinforced if the visual
scene includes objects and contours.
The following anecdote shows how a negative result can lead to a discov-
ery. One day, a subject proved to be totally insensitive to all visual stimulation.
Her case was a real challenge, and I struggled to understand the reason for her
resistance. I asked her what she did for a living, and she told me she was a
flutist. I hypothesized that she had developed a visual inhibition so she could
220 •
play in an orchestra and not be distracted by the movements of the other in-
strumentalists. To suppress this mechanism, I asked her to count backwards
from 100.10 And, as had other subjects, the flutist immediately responded to vi-
sual stimulation. It was then clear that for her, training to play the flute en-
tailed selection of sensory information that made her in some sense blind or
less sensitive to visual perturbations.11 It may also be that her resistance to vi-
sual perturbation was a true gender difference. Experiments from our lab
show that when subjected to visuo-vestibular conflict, women recalibrate their
perception of body rotation less than men.12
This observation is instructive for several reasons: first, postural mecha-
nisms depend on the general context of the action; second, people resolve the
problem of sensorimotor integration in their own ways. Captives of statistics,
researchers sometimes try too hard to define an average behavior, a typical
subject, and standard means and deviations in normal populations. Psycho-
physiology of sensorimotor integration must now move in new directions, in-
dicated in France by the differential psychology of Reuchlin.13 Researchers
need to take into account both general rules of behavior and individual solu-
tions. All the observations that have been made concerning postural strategies
in subjects with vestibular lesions, in astronauts during and after space flights,
and so on lead to a challenge. The challenge consists of constructing a physiol-
ogy of individual perceptuomotor strategies and of the common repertoire of
behaviors that these strategies draw on. The functional flexibility that makes
possible the choices and decisions characteristic of the brain of primates and
man must be fully explored. I will come back to this point in the chapter on
adaptation.
Toward a Projective Physiology
An emphasis on functional flexibility also suggests a top-down model of regu-

lation of postural mechanisms of selection and anticipation as opposed to the
accepted bottom-up model of reflexes. For me, a chance encounter opened
new prospects.
One day, I was presenting the results of these experiments at an interna-
tional physiology conference in Paris. I was standing in front of my poster.
Opposite me a young American, Lewis Nashner, was also presenting a poster.
He wore splendid cowboy boots, was abroad for the first time, and spoke not
a word of French. His poster claimed that proprioception contributes to rapid
postural reactions during a perturbation on a moving platform. However,
BALANCE •
221
when his subjects closed their eyes, he observed no difference in the amplitude
of their fall and concluded that vision made no contribution. In contrast, my
results indicated that vision played a very important role.
We had in common the fact that no one was the least interested in our re-
sults. So we began to talk. I felt compelled to criticize both his premise and his
conclusions. In fact, I explained to him that when a person closes his eyes, the
brain no longer has access to visual cues and thus reorganizes other cues that
it uses instead. I argued that if vision were preserved by inhibiting visual cues
about velocity, the brain, expecting this information, would make a mistake,
and a major perturbation would result. I made him a friendly challenge and in-
vited him to come and try the experiment, which, one year later, he did.
Together we constructed a novel device, placing the subject upright on a
platform that could be abruptly displaced.14 With respect to the paradigm of
postural perturbation, our device was nothing new. The trick consisted in sur-
rounding the subject’s head with a box made of light polystyrene that slid
from front to back. This box could be fixed in space or suddenly joined to the
movement of the head by an electronic command: the perturbation was trans-
mitted, the subject fell forward, and, right at that moment, an electronic de-
vice initiated the coupling between the subject’s head and the box. In this way
the visual world was stabilized in a transitory manner. The brain perceived no
movement by vision. We had fooled it. Only the proprioceptors of the mus-
cles and the joints, along with the vestibular sensors, signaled the movement.
The result was spectacular: the muscular reaction of the legs was cancel-
led for a short interval. Lacking visual information, the brain did not trigger a
compensatory reaction. This experiment convinced me that when vision is ef-
fective, the brain waits for a cue from this preferred sensor, together with oth-
ers, to initiate preprogrammed reactions of balancing. These signals require a
specific configuration and combination to trigger the reaction. Vision is both a
sensor of velocity and an analyzer that initiates responses when its stimulation
by a movement is conjointly signaled by several other sensors.
The multisensory control of balance is thus not due to a simple string of
responses to stimuli. It involves comparison of the state of the sensors with
prediction. Now a good many contradictory findings became clear. All the au-
thors before us who had claimed that vision played no role in rapid postural
reactions had asked their subjects to close their eyes. Comparing the eyes-
closed performance with the eyes-open performance revealed no differences.
My interpretation was that when the subject closes his eyes, the brain actively
reorganizes the configurations of sensors on which it bases its decisions and
222 •
expectations. Hence the new idea that the brain chooses sensory inputs ac-
cording to context.
This theory was reinforced by the so-called broom experiment proposed
by Vidal, one of my collaborators. He positioned a broom handle horizontally
before a subject. This broom handle constituted a stationary rod on a plat-
form, enabling the subject to stop himself with his arms. During a postural
perturbation, instead of activating the muscles of his legs, the subject reacted
exclusively with his arms, which caused a complete reorganization of his
postural reactions. This experiment was also conducted and published by
Cordo and Nashner. They called this flexible functional state “set,” or state of
preparation. It remains to reconcile the concept with the definition given by
Bernstein of the posture he called “readiness to move.”
The Set Concept

The term Einstellung (set) was introduced around 1905 at Würzburg by Külpe,
Ach, and Watt.15 It was adopted first of all to describe the internal state of sub-
jects awaiting instruction to concentrate their attention on this or that aspect
of an object. For example, if a subject was asked to describe the number of
objects present in a room, he had trouble remembering their color several mo-
ments later, which gave rise to the idea of a perceptual filter.
The set concept is tightly linked to the problem of relationships between
posture and movement. Sherrington considered these to be two distinct mech-
anisms and proposed that posture accompanied movement like a shadow.16
Later, movement was considered to be scarcely more than a succession of pos-
tures. This idea turns up in various recent formulations of the so-called point
of equilibrium theory, which holds that movement is due to a gradual change
in static balance between the forces exerted by the muscles located around a
joint, called antagonist muscles (for example, in the arm, the biceps bends the
forearm, and the triceps stretches it out).17 According to Sherrington, this shift
is induced by a central command that regulates the relation between the
forces of the muscles, which produces the transition from one position of bal-
ance to another. For some, a movement intended to reach a target—a teleoki-
netic movement—can only occur when it is supported by an initial so-called
ereismatic posture.18 It was also suggested that there are two submechanisms
of postural control, one that steadies the reference point of the movement,
the other the distribution of muscle tone in the moving limbs.
The set concept was assigned a wide range of meanings, as various con-
texts for set in the literature attest: mental, neural, voluntary, unconscious,
BALANCE •
223
postural, organic, preparatory, task-oriented (Aufgabe), situational, goal-ori-
ented, temporary, permanent, reactive, perceptual, expectational, anticipatory,
intentional, and so on. It has also been defined as a preparatory state for re-
ceiving a stimulus that has not yet arrived.19 Discussions of the set concept
were accompanied by numerous studies of events associated with impending
movement.20
Concurrently with this research carried on in the West, a very original
program of research was developing in the Soviet Union, which at the time
was closed to scientific communication. The idea of a preparatory influ-
ence on perception and control of movement had already been suggested by
Sechenov: “When a stimulus is expected, the activity of any other mechanism
interferes with the phenomenon, and curbs and delays the reflex.”21
In the 1960s, Gurfinkel and his colleague Alexejev in Moscow established
the importance of anticipation, instruction, and a person’s mental state in the
control of posture. Indeed, if you say to someone, “Wait for the bus. It will be
here in two hours,” the gain of this reflex is much less than if you say to him,
“Watch out! I’m going to kick you!” This observation might appear trivial. But
the idea that rigid reflexes underlay the organization of motor function was so
anchored in people’s minds, when in fact they are modulated by a global plan
of action, that it led to a genuine revolution.
How to Pick Up a Load without Falling Down

Posture thus both supports and prepares for movement. But the predictive
brain has devised something even better. It maintains a repertoire of postures
specific to each animal. In Chapter 7 I defined the concept of synergy. Balance
is guaranteed by a repertoire of synergies that are organized into strategies.
Let me explain. When the bus brakes abruptly, you can only bend your body
two ways: either by rotating around your ankles and keeping your body rigid
like a reverse pendulum (ankle synergy), or by flexing your trunk at the pelvis,
leaving your legs rigid (trunk synergy).22 A third strategy would be to take a
step forward. These synergies involve sequential activation and inhibition of
the muscles of the legs, hips, and trunk whose temporal organization is very
precise. Thus time is as important in organizing a synergy as the distribution
of activities.
The goal of some of these synergies is to avoid loss of balance caused by
ordinary body movements. Indeed, any number of operations can lead to loss
of balance: rising to your tiptoes, lifting your leg, bending over to pick some-
thing up, and so on. Try standing on a beam or an unstable part of the floor:
simply lifting your arm is enough to shift the center of gravity forward and
224 •
cause you to fall.23 To avoid these disruptions, the brain automatically trig-
gers a slight tilting of the body backwards prior to the main gesture of lifting
the arm.
Now try standing on your left leg and lifting your right leg to the side as
high as you can, as dancers do. The first time you do it, your torso and head
have a tendency to move slightly to the right before your leg goes up. This
slight adjustment enables the center of gravity to shift directly below your
standing foot to maintain balance.24 Training entails a reorganization of the
gesture and the posture that accompanies it: you no longer incline to the right,
and the movements of the body are redistributed around a perfectly vertical
head and torso. With Pozzo (see Chapter 4), I showed that the head is used as
a stabilized frame of reference for coordinating movements of the limbs.
Try one last exercise: Bend over to pick something up. Position yourself a
few millimeters away from a wall without touching it with your back. You will
notice that just before you bend over, your buttocks touch the wall. A move-
ment of the body backward anticipates the bending forward. It displaces the
center of pressure that your body exerts on the floor toward the back and in
this way compensates, in advance, for the body’s forward displacement. So
balance is not maintained by detecting an error and correcting it; it is main-
tained by anticipating the consequences of action. This synergy is called the
Babinski synergy.25 It also comes into play when you lean backward. Babinski
pointed out that it is absent in patients with lesions of the cerebellum. Thus
the cerebellum contributes to this motor anticipation.
Finally, a very common but interesting phenomenon is the initiating
movement of a step, which is also preceded by anticipatory postural activity.
Suppose that you lift your foot to take a step forward. You cannot do it with-
out lifting the 30 or 40 kilograms of body weight that are bearing on that foot.
To relieve the weight and initiate the step, your center of gravity has to be
shifted forward by a burst of activity in the flexor muscle of the ankle, the an-
terior tibialis. The brain accomplishes this anticipation by producing a shift in
the center of gravity just before you lift your foot. This anticipatory motor
synergy is disturbed in patients with Parkinson’s, who have a dopamine deficit
in the basal ganglia, the nuclei at the center of brain that are essential in orga-
nizing movements.26 Initiating activity of the anterior tibialis is very weak in
these patients, and they can no longer coordinate their movements. The basal
ganglia are thus indispensable in anticipatory shifts of the center of gravity as-
sociated with the initiation of movement. Together with the supplemental
motor area of the frontal cortex,27 they play a fundamental role in coordinat-
ing posture and movement. These anticipatory postural synergies, which are
BALANCE •
225
Control of posture Control of movement
Control
of posture
(feedforward)
Reference points Orientation Stability
Multisensory
inputs
Vision
Local
Vestibular Head feedback
receptors Body schema
(geometry,
kinetics, − Neural
networks Local
Proprioception verticality, Trunk
for control feedback
frame of
reference) + of posture
Cutaneous
receptors Local
Limbs feedback
Graviceptors
Control of posture
(feedback)
Figure 11.2. Coordination between posture and movement. At left, postural

control is achieved based on frames of reference that indicate the desired orien-
tation of the body or limbs and the degree of stability required. These data are
sent to the parts of the brain that contain the body schema and are used to up-
date it and to detect discrepancies between the assumed posture and the value
required for the task. The sensory data not only trigger postural readjustments
by the reflexes, they also contribute to changes in the body schema, which is
the principal determinant. Making a move (right) results in anticipatory adjust-
ment of posture (feedforward).
also found in animals,28 are involved in manual tasks requiring coordination

between the position of the arm and movement (for example, in lifting a
load).29
The repertoire of natural motor synergies—all preceded by slight move-
ments of anticipation that enable posture to be reorganized in advance, as it
were, and not, as reflex theory predicted, to correct the difference between the
ideal position and the desired position—is the best proof that instead of sim-
ply being the result of passive servomechanisms, posture is programmed into
general planning for action (Figure 11.2).
This is also confirmed by the observation that when acrobats walk on
their hands, using their arms rather than their legs to maintain their balance,
the postural regulatory activities are immediately transferred to the muscles
of their arms.30 You might say that the control of balance obeys the law of
motor equivalence. Motor equivalence refers to a simple and remarkable
property of the brain that enables the same movement to be made using very
226 •
different effectors. For example, I can write the letter A with my hand, or my
foot, or my mouth. I can even draw a letter A while walking on the beach.
This property fascinates neurobiologists specializing in motor function and is
considered to be proof that the brain encodes a motor structure (morpho-
kinesis) very generally, which then enables it to express the structure or to exe-
cute it using very different combinations of muscles. This property is also
valid for maintaining balance. The transfer of postural programs from the legs
to the arms demonstrates, once again, the global rather than local character
of mechanisms of balance and posture.
Posture is thus not a passive state of reactions initiated by reflexes. It is a
state of preparation to move based on internal simulation of predicted se-
quences of movement and general goals of action.
The Concept of Body Schema

What are the higher-level mechanisms of control of balance and posture? A
clue to this mystery lies in a concept that is a little woolly but turns out to be
surprisingly useful: that of the body schema. The neurologists Head and
Holmes were among the first to suggest that the cerebral cortex contains what
they called “a schema of the body” that directs posture and the coordination
of movements.31 According to them, the brain contains an internal model of
the relative sizes of body segments, their relationships, and their positions.
The body schema might also be the source of our perception of the body.32
They situated the body schema in the parietal cortex, which, as Sherrington
foresaw, assigned the brain a top-down rather than bottom-up control of
posture.
This idea also shows up in the writings of Adrian, another of the great
British physiologists: “In this way the organism carries in its head not only a
map of external events but a small-scale model of external reality and of its
own possible actions.”33 It is the word “possible” that is so unusual for the
time: the body schema was not conceived as a representation of the body but as
a blueprint of possible actions.
The Phantom Arm

Phantom arm syndrome is further proof of the reality of the body schema.
People whose arm is severed either accidentally or surgically continue to sense
their hand as if it were still there. They feel it move and experience aches and
pains that existed before the amputation.
A situation similar to amputation is that of ischemia, that is, anesthesia of
BALANCE •
227
the arm. Subjects with this condition also perceive a phantom arm, whose po-
sition does not coincide necessarily with that of the real arm. Gurfinkel con-
ducted the following experiment based on this condition.
Rest your right arm on the table and angle it about 45 degrees to your
right. Close your eyes and imagine that your arm is anesthetized and that
you perceive a phantom arm at 45 degrees to the left, that is, to the other side.
Lay a pencil on the table in front of you and move it to the left: it represents
your phantom arm (you will have to open your eyes briefly to arrange the
pencil). Now bring your left hand, that is, the one that is not anesthetized, be-
fore you, resting your finger on the table, right in front of you. If you have fol-
lowed the instructions correctly, your real right arm is to the right, the finger
of your left hand is in the center, and the phantom arm (the pencil) is off to
the left.
Gurfinkel asked his subjects to close their eyes and to displace their anes-
thetized arm toward the finger of the left hand; in your case, you will have to
move your right arm toward the center. The subjects who perceived the phan-
tom arm to the left made a move of the real arm toward the right; that is, this
movement was dictated by the position of the phantom arm in space. This ex-
periment shows that planning a movement is determined by the central repre-
sentation of the position of the body in space and not by its real position as it
is detected by proprioception.
The importance of the central representation of movement of the body
and its domination by peripheral sensory factors is also illustrated by illusions
of movement induced by the vibrations of muscles I described earlier.34 For
example, if the nape of the neck is vibrated, subjects perceive an illusory
movement of the head. The postural reactions of the legs caused by this vibra-
tion of the nape of the neck are consistent with the illusion of movement of
the head and not with the local motor reflexes caused by the vibration.
You can do another experiment that tests the same principle. Sit comfort-
ably in an armchair. Turn your head to the right and keep it there, eyes closed.
During the first few minutes, you will have the impression that your head is
turned to the right, and an electromyographic recording of the activity of
your legs would show a very slight asymmetry of muscle tone that reflects
your postural asymmetry. After 8 or 10 minutes, if you have the patience to
keep your head fixed to the right, you will experience the illusion that your
head is gently returning to the center and, after another 10 minutes or so, you
will have the distinct impression that your head has resumed its natural posi-
tion, facing front. At the same time, the muscular activity of your legs will
228 •
also regain symmetry. Here again, muscular activity follows the illusory per-
ception of the position of the head. It is as if you have a phantom head that is
looking straight ahead, even though your real head is turned to the right, and
postural regulation is following the phantom head and not your real head.
Gurfinkel and his collaborators also used hypnosis to show the importance
of the internal representation of the body. Hypnotized subjects were per-
suaded that their head was turned to one side, even though it was perfectly
straight. At the same time that the illusory percept was established, muscular
contractions were observed in the legs, of the kind that generally accompany
a real rotation of the head. The percept directed the postural activity.
Similar results were obtained using vestibular stimulation. It is well known
in neurology that pathways descending from the vestibular sensors and flow-
ing through the vestibular nuclei ensure postural reactions that keep the body
upright, stable, and in balance. They also initiate compensatory reflexes dur-
ing tilting of the body. These are the so-called vestibulospinal reflexes. In prin-
ciple, they are well-identified and depend on rigid neural networks that con-
nect each semicircular canal with the postural muscles and those of the nape
of the neck. It is thus reasonable to assume that in humans, stimulation of the
labyrinth by application of a constant current (galvanic stimulation) placed
close to the ear would induce highly stereotyped postural reactions. Yet these
reactions also change completely based on the illusory position of the head.
It follows that the internal representation of the body in space is capable
of altering the organization and direction of reflexes, which means that the
whole definition of reflex itself must be reconsidered.35
Internal Models and Body Schema

Does the brain contain one or several body schemas? A long time ago, Russian
physiologists of the Moscow school demonstrated that electrical pulse stimu-
lation of the spinal cord of a frog, severed at the level of the neck, caused the
frog’s foot to move to very precise positions that varied depending on where
the foot was positioned at the outset.36 This suggests that a neural mechanism
located in the spinal cord has available a model of the limb that enables auto-
matic readjustment of the set of muscular synergies necessary to reach a goal.
This internal model is an elementary form of body schema, for it includes
both the geometry and dynamics of the limbs.
The second example is that of the cerebellum. It is currently being sug-
gested that the cerebellum is an inverse model or a proactive model used by
the brain to evaluate the consequences of movement and to create predicted
BALANCE •
229
trajectories even before they are executed.37 In this sense, the cerebellum
might thus contain unique elements of a body schema or a specific form of
body schema.
A third example is found in the thalamus, whose multimodal neurons inte-
grate sensory messages with motor activity.38 In fact, the grouping that com-
prises the thalamus, cortex, and basal ganglia, which are interconnected in the
form of circuits and traversed by sustained rhythmic activity (see Figure 1.3),
might also constitute an internal representation of the body and its possible
movements.
Finally, an even higher level is probably located in the parietal cortex,
where the body schema as such is situated in the context of external space,
whose properties are analyzed by vision. Head and others after him situated
the body schema in the parietal cortex probably because it contains the only
level of generality capable of directing motor behavior in an integrated way
toward an external goal. It is most likely at this level that a global body schema
is to be found, whereas in the medulla, the schema is local.
In all probability there is not one body schema but multiple body schemas,
each adapted to a particular function, just as there are multiple representations
of the body. This idea of multiple representations of the body was taken up
again recently in connection with patients with impaired body recognition, or
autotopoagnosia. This deficit is often associated with others such as aphasia,
apraxia, disorders of manual grasping, and hemispatial neglect. However,
autotopoagnosia is a specific impairment of body recognition. Some patients
cannot indicate a part of their body verbally but can move it if the examiner
touches it.39 The same subjects have trouble referring to parts of an object ex-
ternal to their body. The impairment is perhaps not unique to body function
but a general function of designating the parts of a whole. However, patients
may have specific difficulties in indicating a “discrete image of the body.”40
A recent observation also argues for multiple representations of the body.41
A patient whose cerebral lesions were diffuse could neither localize the parts
of her own body nor, moreover, those of other people, although she could
name them. In other words, she could not localize them spatially, but she
could identify them verbally. The authors of this observation suggest that
there are at least four types of processes that enable one to represent one’s
body. The first processes semantic and lexical messages about the parts of the
body; the second, specific visuospatial representations of a person’s own body
and also objects in the environment, that is, the location of parts in relation to
each other (for instance, the nose is in the middle of the face, the ears are on
each side of the head) and their boundaries; the third is a body frame of reference
230 •
based on a body schema; and the final process involves movements themselves,
which organize perception of the body.
Consequently, it is futile to seek a single biological reality underlying the
concept of body schema. Gross reached similar conclusions concerning the
frames of reference the brain uses to integrate sensory information (see Chap-
ter 4). There are as many sensorimotor spaces as there are limbs. Given the ex-
treme complexity of the construction of a single representation of the body,
especially when it is in motion, how is this coherence assured? This is an im-
portant but still unanswered question.
Brain-Imaging the Body Schema

Recent findings obtained by brain imaging using PET elucidate the neural
bases of the body schema.42 Bonda and co-workers recently conducted the fol-
lowing experiment. Subjects were shown photographs of hands in a variety of
postures in different orientations and were asked whether they were looking
at the right hand or the left hand. This experimental protocol derives from
cognitive psychology experiments on mental rotation, which showed that sub-
jects respond more quickly the more natural a posture appears. If you try the
experiment, for instance, by asking someone to show you his hands in various
positions while hiding behind a screen, as in a puppet theater, you will be
aware that you are making a mental rotation of your own hand to align it
with the hand on display. Performing this task thus involves reference to the
body schema.
The areas of the brain activated during this task are the following: the
anterodorsal portion of the insula (which is not the retroinsular area, where
vestibular activities occur), the superior parietal cortex, especially the left, at
the point of the anterior parietal sulcus, and the motor cortex. Mesulam main-
tained that area 25, the insula, the cingulate cortex, the prefrontal cortex, and
area 5 constitute a set of structures that contribute to the body schema.43
Mishkin also suggested that the insula is involved in representations of the
body and that there is cooperation between the superior parietal cortex and
the anterior insula to produce what might be called a mental kinematics of
the body.
I think that understanding the mechanisms of control of balance and coor-
dination between posture and movement will in future require intense co-
operation among mathematicians, physicists, neurobiologists, and neuropsy-
chologists. The time when posture was conceptualized as nested reflexes is
gone. The time when the idea of a body schema can be expressed in terms of
precise neural functioning is not yet come.
BALANCE •
231
Dog approaching another dog Dog in a humble and affectionate
with hostile intentions. frame of mind.
Figure 11.3. To the left is an aggressive dog; to the right, an affectionate dog
waiting to be petted. Note that each posture both expresses an emotion and
prepares for action: the posture of the aggressive dog is adapted to fighting,
that of the submissive dog to being petted. This is also the position taken by a
female in heat ready to receive the male, and it has been shown recently how
this posture involves activation of a special synergy of the back muscles.
Relationships between posture and emotions also have to be taken into ac-
count. Darwin showed how different postural synergies are used to express
feelings. The images of the aggressive dog and the affectionate dog in Figure
11.3 show how these expressive postures are also a preparation for action—in
the first case, preparation to aggress, and in the second to be submissive and to
be petted.
232 •
12
A D A P TAT I O N
One of the most important properties of the nervous system is its ability to
adapt. This term encompasses several mechanisms that complement one an-
other.
The first mechanism is diminution of activity over time. For example, tac-
tile receptors stop firing when pressure is maintained, which enables them to
avoid overloading the nervous system with perpetual signals. When you are
seated, you forget that your chair is supporting you, which is not a particularly
useful bit of information while you concentrate on reading this book! Simi-
larly, the membrane of motor neurons is endowed with adaptive properties
that under the control of numerous neuromediators act to diminish their dis-
charge in reaction to a sustained stimulus. The visual system is also equipped
with mechanisms that suppress the transmission of stimuli when they are pro-
longed, and so on.
A second mechanism regulates the sensitivity of the sensory receptors. For
example, in Chapter 2, I mentioned that the γ motor neurons control the ac-
tivity of the neuromuscular spindles, thus modulating the intensity of sensory
messages from the muscles and regulating the influence of proprioception on
the motor commands for walking and balance. Another example is mecha-
nisms of accommodation, which regulate the optic properties of the eye at a
distance from objects, just as other mechanisms regulate the amplitude of the
vestibulo-ocular reflex according to the distance from the object of regard. If I
want to look at my nose, the reflex has to be cancelled; if I wish to keep my
gaze fixed on a tree in the distance, my eye must make a displacement equal
to and in the opposite direction from that of my head.
A third mechanism involves changing strategies. For example, when I am
on a bus, I can respond to a slight braking with a simple contraction of my leg
muscles; but if the braking is abrupt, I respond by moving my foot forward. If
the target I am following with my eyes is moving too rapidly for ocular pur-
suit, I follow it with saccades. When an astronaut no longer has gravity to rely
on as a reference, he employs his own body as a frame of reference and uses
vision to anchor his moves in the space station, and so on.
The major physiological theories on adaptation all take into account this
diversity of solutions and considered phenomena of plasticity at the cellular
level, as well as higher-level mechanisms like functional substitution that in-
volve reorganization of behavior as a whole.
I will now examine a few examples to give an idea of the state of the art
of research on this subject and to show how the brain’s predictive functions
come into play in adaptation. Though this chapter barely touches on this im-
mense domain, it will at least contribute some additional information. The
brain possesses mechanisms enabling it to choose solutions that anticipate the
results of action, and that activity is a necessary tool for testing these assump-
tions. Adaptation to disturbances of our senses, to sensorimotor conflicts, or
to lesions triggers strategies that bring to light the underlying processes, which
alter local mechanisms through global direction of the organism.
Adaptation and Substitution
The example that I propose is a familiar one. If you wear glasses, you will have
noticed that when you first put on a pair that magnify slightly, you need some
time—perhaps several days—to get used to this new vision. Hunters know
that it is practically impossible to walk while looking through binoculars, even
if they only magnify twofold. And if you are a lover of opera and theater, it
does not take you long to figure out that if you do not want to lose sight of
the actors while looking at them through opera glasses, you have to keep your
head still. Why? Very simple—because the amplitude of the vestibulo-ocular
reflex (see Chapters 2 and 4) that stabilizes the image of the world on the ret-
ina is, normally, exactly equal to that of the movement of the head. When an
error is introduced by glasses or binoculars, the brain has to automatically ad-
just the reflex to the new speed of visual motion. It detects the error, then im-
plements neural mechanisms that adapt the reflex to this new situation. Two
extreme conditions allow experimenters to probe the limits of this mecha-
nism.
The first consists in placing a subject, human or animal, on a revolving
chair inside an illuminated cylinder that moves with the subject. Under these
234 •
circumstances, the visual world is stable, and the reflex must be cancelled. In
the other experiment, subjects are asked to wear prism goggles called Dove
prisms that reverse the direction of visual movement. When the head turns to
the right, the visual world, instead of turning to the left as it would under nat-
ural conditions, turns in the same direction as the head. The brain not only
has to adjust the amplitude of the movement of the eye but also to reverse the
direction of movement. The entire anatomy of the vestibular neurons whose
remarkable arrangement I described in Chapters 2 and 4 is useless in this case.
The creation of millions of years of evolution has to be undone. How does
the brain do it?
At the beginning of this century, psychologists and ethologists showed
that chickens could quite easily adapt and peck at their grain or earthworms
while wearing prism goggles. In humans, wearing prisms induces profound
transformations that are adaptively reversible. Gonshor and Melvill-Jones con-
ducted a critical experiment that revealed this surprising adaptability of the
central nervous system.1 They had a subject wear prism goggles for three
weeks and tested his vestibulo-ocular reflex every day. The tests were con-
ducted in the dark, without allowing the subject any visual cues regarding the
actual movement of his head. After five or six days, the reflex was reversed. In
other words, an adaptive mechanism had totally altered the functioning of this
remarkable wiring.
A good dozen laboratories all over the world have been working for 30
years to understand the mechanism of this adaptation, which has become a
model for the neurobiology of sensorimotor plasticity. An experimental fact
very soon helped to focus the theories: a lesion of a small area of the cerebel-
lum—the flocculus—prevented this adaptation in animals (see Figure 2.4).
Why the flocculus? The answer came from anatomy and neurophysiology:
projections of the vestibular receptors reach the Purkinje cells of the flocculus
via the mossy fibers of the cerebellum and are combined with projections
from the accessory optic system described in Chapter 3. The flocculus is thus
the center for a visuo-vestibular convergence. What critical event happens at
this level to determine adaptation? What happens is that the flocculus projects
to the neurons of the vestibular nuclei, which are involved in the reflex.2 This
is an inhibitory projection, so it is able to modulate the amplitude of the
reflex. You can experiment with this modulation yourself. Hold a finger in
front of you and oscillate your head horizontally from one side to the other,
moving your finger with your head so that your finger follows your head’s
movement exactly. Under these conditions, you will notice that your gaze is
fixed in the socket. There is no longer a vestibulo-ocular reflex. Patients with
ADAPTATION •
235
lesions of the cerebellum lose this visual inhibition of the vestibulo-ocular
reflex.
In other words, the flocculus is part of a reflex control loop modulated by
stimuli that are both visual and vestibular. Ito subsequently suggested the so-
called flocculus hypothesis.3 He proposed that a synaptic mechanism detects
the difference between visual and vestibular inputs at the level of the Purkinje
cells and, based on this error signal, alters their influence on the reflex.
Ito later refined his hypothesis based on a mechanism he called “long-term
depression,” caused by the coincidence of two different synaptic events (those
associated with the climbing fibers and the mossy fibers, which are the two
main groups of neurons that convey visual and vestibular information to the
cerebellum) that initiate this alteration. Very fine molecular mechanisms un-
derlie this adaptive capacity, which Ito used as a general model of motor train-
ing. But the model was controversial. Some saw motor training as a more
complex function, not localized to a single synapse or a single long-term sup-
pressive mechanism. Others situated the critical synapse for adaptive changes
not at the level of the cerebellum but at the level of the brainstem, where spe-
cialized neurons ensure its regulation: according to this conceptualization, the
cerebellum is also implicated in alterations, but as controller.4 Adaptation thus
operates on at least two levels. Whatever their differences, these theories as-
sume that adaptation is the product of neurons situated within the networks
that control the reflex.
But there is another way of thinking about adaptation that leads to a com-
pletely different theory: with Melvill-Jones, I proposed that instead of revers-
ing the vestibulo-ocular reflex, the brain replaces it with a reverse pseudo-
reflex using another element of the oculomotor repertoire, such as saccades
or pursuit.
To prove this theory, we repeated the original experiment together, asking
subjects to wear Dove prism goggles. After putting the goggles on, the sub-
jects succeeded within several hours, and sometimes immediately, in produc-
ing an adaptive movement composed of a sequence of saccades that gradually
merged to approximate a vestibular reflex.5 The brain had suppressed the
vestibulo-ocular reflex via the mechanism described above involving the floc-
culus, instead using the saccadic system to construct a new adaptive move-
ment.
We hypothesized that this pseudo-reflex was guided not by visual targets
but by an internal simulation of the movement of targets; that is, by a percept
that directs adaptation from top to bottom just as the body schema controls
posture. A high-level mechanism, calling into play the cortical structures in-
236 •
volved in the representation of space and motor imagery, appeared to be re-
sponsible for this adaptation. We reasoned that the physical presence of visual
targets was not necessary to achieve adaptation, and that simply imagining the
target might be enough. Consequently, we asked subjects to sit in a revolving
chair that oscillated and to imagine that they were focusing on a target at-
tached to their nose. The conflict they experienced had to do with the fact that
their vestibular system was telling their brain that it was turning, thus causing
a vestibulo-ocular reflex that inhibited the very intention of fixing their regard
on the target attached to their nose. After several hours of this exercise, we
observed that the vestibulo-ocular reflex had changed. A simple internal men-
tal effort—subjects keeping their gaze focused in front of them during a rota-
tion of the head in the dark—was enough to cause adaptation.6 Here, as for
postural control, adaptation is caused by a percept of movement worked out
in the areas of the cerebral cortex involved in the representation of space and
not by low-level reflexes. We have recently shown that there is a great differ-
ence between men and women in their adaptation to a visuo-vestibular con-
flict. We have exposed both normal and anxious subjects (the anxiety level was
judged by standard psychiatric tests) to a conflict by turning their bodies by a
certain amount and turning the visual world in a virtual reality helmet by a
different amount. We tested their vestibular perceptions before and after the
exposure to the conflict and observed that men adapt their vestibular percep-
tion more than do women under these circumstances (or women remain
more independent). Also, anxious men adapt even more. We are performing
these studies on anxiety because I believe that part of the mechanism underly-
ing spatial anxiety is related to the difficulty of building a coherent perception
of the relation between the body and space and difficulties in coping with in-
congruous sensory inputs.7
The brain thus employs various mechanisms to adapt. Some are low-level
and promote neural plasticity; others are high-level and, especially in humans,
rely on remarkable capacities for functional substitution, based on internal
simulation of the signals to be controlled. In other words, the brain is not a
servo system for regulating variables; it is a composer that orchestrates the in-
struments at its disposal. The hierarchy of nerve centers developed over the
course of evolution enables the brain to find new solutions when adaptation is
required. Low-level mechanisms then stabilize the new solutions. In fact, brain
imaging has revealed that during motor training, the cerebral cortex is used at
the beginning of the training and gradually becomes silent; the activity is then
transferred to the basal nuclei and to the cerebellum, which leaves the cortex
free to confront new problems and find new solutions to them.
ADAPTATION •
237
For all these reasons, inter- and intraindividual differences must regain the
central status they have lost in the behavioral sciences and neurosciences. If
the brain does have this ability to choose solutions, we must vary our efforts
to find one or several solutions, rather than focusing on typical behavior, as we
have done. A diversity of strategies better clarifies the mechanisms at work
than does an average profile. However, it must also be recognized that the
number of possible solutions is not infinite. It is constrained by the genetic
legacy of sensorimotor subsystems. To what extent it is possible to escape
simple selection to achieve genuine invention is a fine subject, but it is a mat-
ter for another discussion.
The Rheumatologist and the Ophthalmologist
Once upon a time, in Lisbon, there lived a rheumatologist whose patients

were suffering from back pains. He noticed that they had ocular disturbances
too. In the same city, an ophthalmologist was also attending to patients with
back pains. These two doctors shared a fondness for wine, so much so that
they frequented the same wine-tasting establishments specializing in fine Por-
tuguese wines. During one oenological meeting, they were discussing their
overlap in presenting syndromes and agreed to exchange patients. Over the
years, they came to the conclusion that some of their patients were suffering
from the same illness. Indeed, if these patients were asked to close their eyes
and to position their arms horizontally, they placed their arms at different
heights; if they were asked to close their eyes and to position their feet one
next to the other, exactly parallel, the doctors noted that the patients main-
tained a considerable gap between their feet; if the patients were asked to
walk in a straight line, the doctors observed that the patients staggered.
Ophthalmological tests revealed significant asymmetries in the perception
these patients had of the position of objects in three-dimensional space. Gen-
erally, they had a distorted perception of their own body and its spatial local-
ization.
Da Cunha, the rheumatologist, identified what he called a “syndrome of
disorientation” and hypothesized that this syndrome was caused by a distor-
tion of the body schema, which might explain the constellation of postural
oculomotor symptoms, and possibly even gastric or cephalic symptoms. He
achieved spectacular remissions by a combination of having patients wear
prism goggles, under ophthalmological supervision, and of motor reeducation
aimed at making subjects aware of the position of their own bodies; for exam-
ple, by making them move in front of a mirror. Electroencephalography con-
238 •
firmed his theory, revealing asymmetric activity of the parietal cortex in cer-
tain patients. The asymmetry disappeared with reeducation.
Such a theory would have been inadmissible even a few years ago, and
these doctors would have been charged with quackery, so rigid was the con-
ception of the brain. The possibility that the cortical structures involved in
sensorimotor function are accessible to training was pretty much unthinkable.
Recently, however, neuronal recordings in the monkey and in humans using
brain-imaging techniques have shown that neurons from the somatosensory
areas of the cerebral cortex can reorganize the maps where these parts of the
body are represented.8
A similar discovery was made recently by Rousié, a physician from Lille
who is attached to the maxillofacial surgery clinic of a university known for
expertise in oculography, anatomy (especially of the vestibular system), and
neurology. This physician noticed that many patients who came to consult for
facial asymmetries also presented symptoms that appeared to indicate a more
general etiology than simple malformation of the bones of the face or jaw. To-
gether with several specialists, we designed a complete battery of postural,
vestibular, ophthalmological, and oculomotor tests, and proposed that a basic
vestibular asymmetry was the source of all the disturbances. Dr. Rousié or-
dered radiological examinations, which did indeed show a functional asymme-
try of the vestibular receptors in these patients.9 These vestibular asymmetries
caused a perturbation of the body schema that was compensated for by other
postural or oculomotor asymmetries. The local symptoms are influenced by
this global perturbation of a patient’s perception of his body.10
The Role of Activity in Compensating for and

Preventing Disorientation
It is well known that following a unilateral operation on a vestibular organ, pa-
tients are often afflicted, albeit transiently, by postural disturbances, abnormal
ocular movements, vertigo, and general spatial disorientation. When they are
walking along the street, they are often confused, and do not know where
they are. These symptoms generally disappear after several months in a pro-
cess called compensation. The nature of these compensations is still not well
understood, and they are as likely to involve low-level local mechanisms affect-
ing the structures of the brainstem as higher-level mechanisms of cortical or
limbic structures associated with spatial orientation. It may be that the distur-
bances caused by these vestibular asymmetries do not just affect the vestibular
nuclei but also affect the structures of the cortex and the hippocampus that re-
ADAPTATION •
239
construct movements of the body in space. In certain patients, these distur-
bances never completely disappear.
How do we reeducate these patients? Several practitioners tried to subject
them to optokinetic stimulation while standing upright, projecting points of
light into a room in a continuous stream. Surprisingly, exposure to visual
scenes in motion diminished the symptoms in a fair number of cases. Indeed,
the result raises many questions, for it only takes a few intermittent and brief
exposures to decrease a postural susceptibility to a visual stimulus. Whereas
the posture of patients subjected to this stimulus for the first time is highly
perturbed, after a month they can stand without difficulty. Moreover, they no
longer feel disoriented walking around in a natural environment. We mea-
sured eye movements caused by the visual movement and discovered that they
revert to symmetry after reeducation.11
I think that no assumption of local action on the neural mechanisms of
each subsystem resulting from this brief stimulation can explain this abso-
lutely spectacular therapeutic effect. A global theory is needed. When subjects
are standing in a room in which the visual world is moving in a complex way
(if the points of light are distributed over all the walls and moving in various
directions, for example), they are completely disoriented. They are in an un-
natural situation that obliges their brains to find a solution to the conflict be-
tween vision, which tells them that their bodies are falling in several directions
simultaneously, and the vestibular system and proprioception, which tell them
that their bodies are motionless.
One solution is to reduce retinal slip; that is, to force the optokinetic nys-
tagmus in a way that the eye constantly follows the visual movement, with no
resulting visual perception of bodily motion. In this case, because the image
of the world is perfectly stabilized on the retina, the body only has to worry
about the information supplied by all the other sensory receptors to regulate
posture. This is probably the strategy used by the brain, since the amplitude
of the nystagmus is controlled and, at the end of training, the eye manages to
follow the scene very well. This specific training of the optokinetic nystagmus
might also plausibly be used by subjects when they move in a total visual envi-
ronment, such as in the street or in natural surroundings, and would thus help
in stabilizing the image of the world on their retinas. At issue is an effect on
one of the subsystems responsible for postural control and spatial orientation.
A second strategy might be to de-emphasize the importance attached to vision
and narrow the focus and concentration of the subjects, as the flutist I men-
tioned earlier did (see Chapter 11).
Owing to the work of Igarashi and Lacour, it is well known that motor ac-
240 •
tivity is fundamental in compensating for disturbances associated with vestibu-
lar processes. It is probably what enables the nervous system to find ways of
adapting and to discover new solutions. A supporting argument was provided
by a recent space flight. An astronaut flying for the first time was particularly
sensitive to motion sickness in space. He was determined not to be sick in
space and trained his brain—for example, using the revolving armchair that
obliges astronauts to tilt their heads, a reliably nausea-inducing stimulus—to
identify the conflict but to disregard it. Carrying out this cognitive distancing
from an elementary mechanism is a very interesting process of adaptation,
and one that has not been much investigated. It is believed that it is a product
of mental manipulations in space. Several astronauts have stressed that after
several days in flight, they could quickly change what they perceived to be the
floor and the ceiling. This flexibility, this ability to decide frames of reference
seems to accompany the disappearance of symptoms of motion sickness in
space. One astronaut who does not get sick told us that she “decided to believe
what [she] saw” and not to believe what the disoriented part of her body was
telling her. This is why it is necessary to construct what I call a cognitive the-
ory of adaptation to sensory conflicts.
The intervention of high-level cognitive mechanisms in adaptation to sen-
sory conflicts is really a problem of frames of reference. As I discussed earlier,
alterations in the body schema can cause serious disturbances. Similarly, an
abrupt change in frame of reference can lead to difficulties. During a recent
space flight, an astronaut reported to us that she was losing track of every-
thing she set down in the space station, even though she was able to handle
objects without difficulty. The trouble she had remembering where things
were supports the idea of a dissociation between the brain’s use of allocentric
encoding of the identification of objects and of their place in the environment
and an egocentric encoding of handling. Several studies in microgravity show
that in the absence of a gravitational reference point, astronauts tend to revert
to an egocentric frame of reference. All manipulations and perceptual actions
(to use Janet’s terminology) in this frame of reference are thus intact, even
when those that require allocentric encoding have deteriorated. To recall
where an object is in the space station requires constructing an allocentric rep-
resentation of the space station; but to recall it for the purpose of picking it
up, the two systems of encoding have to be integrated.
The ability to move from one frame of reference to another and to com-
bine these frames of reference are probably among the most important as-
pects of adaptation to microgravity as well the basic elements of reeducation
following lesions of the vestibular system.
ADAPTATION •
241
13
THE DISORIENTED BRAIN: ILLUSIONS ARE SOLUTIONS
Here we put our finger on the mistake of those who maintain that perception
springs from what is properly called the sensory vibration, and not from a sort
of question addressed to motor activity.
—H. Bergson
In this chapter, I maintain that perceptual illusions are solutions devised by the
brain to deal with sensory messages that are ambiguous, or that contradict ei-
ther each other or the internal assumptions that the brain makes about the ex-
ternal world. An illusion is generally held to be a sensory error, as in this text
by Sully, written in 1883:
We see, then, that in spite of obvious differences in the form, the pro-
cess in all kinds of immediate cognition is fundamentally identical. It is
essentially a bringing together of elements, whether similar or dissimi-
lar and associated by a link of contiguity, and a viewing of these as con-
nected parts of a whole; it is a process of synthesis. And illusion, in all
its forms, is bad grouping or carelessly performed synthesis. This holds
good even for the simplest kinds of error in which a presented element
is wrongly classed; and it holds good for those more conspicuous errors
of perception, memory, expectation, and compound belief, in which
representations connect themselves in an order not perfectly answering
to the objective order.1
I suggest that illusion is not an error or a bad solution, but rather the best
possible hypothesis.
Illusion: The Best Possible Hypothesis
What do I mean by “hypothesis”? In the Introduction, I put forward the idea

that the brain is a generator of hypotheses. This idea is not exactly new.
Indeed, Lashley wrote: “There are many indications that . . . in the discrimina-
tion box, responses to position, to alternation, or to cues from the experi-
menter’s movements usually precede the reaction to light and represent at-
tempted solutions that are within the rat’s customary range of activity.”2
Krechevsky had the following things to say, also on the subject of rats:
To the animal any new situation is not a confused, meaningless con-

glomeration of sensory impressions to which he makes confused,
meaningless uncoordinated and unrelated responses. The animal is not
altogether a victim of his immediate environment in the sense that each
specific reaction is the result of a specific, momentarily-acting stimu-
lus. He brings to each new situation a whole history of experiences.
These experiences the animal is ready to apply . . . Such responses,
“false solutions,” “early systematic attempts,” etc., we have dubbed
with the dubious name of “hypotheses.” The use of the term “hypothe-
sis” to describe such behavior is a confession of failure. We have been
unsuccessful in finding any one term which might adequately describe
such behavior yet not carry with it the stigma of being fantastically an-
thropomorphic . . . “Hypothesis” carries with it all the mentalistic im-
plications that have become associated with such words as “reasoning,”
“consciousness” and many others. “Hypothesis” as we wish to use it in
this instance, however, need not and does not do so.3
Krechevsky continues, specifying that what he means by “hypothesis” is a situ-

ation in which the animal exhibits multifaceted behavior: it is systematic, goal-
oriented, to a degree abstract, and not entirely dependent on the immediate
environment, either to be undertaken or to be carried out. Finally, he adds:
“A hypothesis is a person’s interpretation of data and not a phenomenon that
derives from the data themselves.” Hypothesis is thus inference, to use
Helmholtz’s terminology.
Here are a few examples. The published literature includes an immense
quantity of articles and reports about visual illusions, but most of these are
static. In line with the theme of this book, I will focus on illusions connected
to spatial orientation and the sense of movement.
THE DISORIENTED BRAIN •

243
Illusions Caused by Acceleration and Gravity
Some visual illusions are connected to vestibular perceptions of bodily orien-

tation.
The Moon Illusion

On a night when the moon is full, it seems larger at the horizon than at its ze-
nith. This is a perceptual illusion. If you are not convinced, wait for a time
when the moon is near the horizon and try the following simple experiment,
which requires no special equipment. If you are at the beach, it is very easy to
do; if you are in the city, it is better to go up to the roof of the building where
you live. In either case, the exercise is the same: Turn your back to the moon
and look at it from between your legs by bending forward and looking at it up-
side down. You will be amazed to see that it will seem to have gotten smaller. I
will leave you to your amazement. What might explain this illusion? It is possi-
ble that the otolithic receptors play an essential role in this perceptual asym-
metry. Indeed, the apparent size of the moon, such as it is perceived, is not a
datum linked exclusively to its objective size on the retina, which can change,
for example, according to the density of the atmosphere. It is commonly
known that the diameter of the moon is slightly greater on foggy days, when
it is nearer to the horizon, than when it is at its zenith. It is likely that a basic
perceptual asymmetry related to our perception of top and bottom has some-
thing to do with this illusion. But it has also been proposed that the illusion is
connected to the non-Euclidean character of distant visual space.
So the position of our body in space can cause changes in the apparent
size of objects as the result of visual and gravitational factors. In this case, the
illusion is not a solution, it is a sensory error. As Poincaré said, representative
space is not identical to geometrical space.
Dynamic Illusions
Another illusion—this one a little more difficult to demonstrate—concerns the
change in the apparent position of a point of light in space. To do the experi-
ment, you will have to stand on an enclosed rotating circular platform or a
centrifuge and look outward, turning your back to the center of rotation. Un-
der these conditions, the vestibular receptors are subjected to a centrifugal ac-
celeration that goes from your ears to your nose. Now, if the lights are turned
off and you are told to follow a little point of light that moves along with you
in the compartment, you will suddenly have the impression that when the
platform or centrifuge starts up, the light is moving up or down according to
244 •
Ayz +z
Az = 1g
+α
GIF
−y
2
Ay = ω r
Gravito-inertial
force
ω (GIF)
Figure 13.1. Illusion of body tilt on a centrifuge. When the centrifuge turns
with an angular velocity ω, the acceleration of gravity Az (directed upward and
leading to a downward gravitational force) and acceleration due to rotation Ay
(directed toward the center and creating a centrifugal force) combine to pro-
duce a gravito-inertial acceleration Ayz tilted at angle α. This acceleration
causes a resultant gravito-inertial force that the brain interprets as the true ver-
tical.
the direction of the acceleration. If you are shown an illuminated horizontal

bar, you will see it go up or go down as well. As I mentioned in Chapter 3, this
effect of linear acceleration on visual perception is probably connected to the
existence of vestibular projections on the visual cortex. The stability of the
perceived world thus depends on the vestibular sensors and their detection of
gravity. On the centrifuge, the apparent direction of gravity is changed, be-
cause the gravitational vector, which is vertical, is added to the centrifugal
force: the sum of the forces of acceleration is a vector that is tilted in relation
to the axis of the body. Subjects who are seated facing the centrifuge, as in Fig-
ure 13.1, have the impression that the vertical is tilted. If the subjects are in the
dark, they may have the illusion that their bodies are tilted, because they as-
sume that the vertical really is vertical. The illusion is both a solution and a
problem.
245
Another way of experiencing an illusion whose source is otolithic is to
stand on a turntable whose axis of rotation is tilted between 20 and 30 de-
grees.4 The table is then rotated at a constant velocity. After 20 to 30 seconds,
stimulation of the semicircular canals ceases, leaving only stimulation of the
otoliths by the gravitational vector, whose projection to these receptor ele-
ments turns with the rotation. You would then perceive yourself propelled by
a strange conical movement. The perceived movement corresponds precisely
to what physiological examination of the otolithic receptors predicts. The
brain thus constructs a perception in accordance with the aggregate of sen-
sory inputs.
The same would be true if you were to sit on a revolving chair, like a bar-
ber’s chair, in total darkness except for a tiny visual target in front of you, at-
tached to your head. When the chair is halted, you will see the target turn to
the left or the right in the horizontal plane. This illusion is due to the fact that
after the stool stops moving, the semicircular canals remain activated, and an
illusory movement is perceived by the vestibular system. I have shown that the
vestibular signals also activate visual and multisensory structures such as the
parietoinsular cortex. But neither the other proprioceptors (muscular, tactile)
nor the other sensors signal movement of the body. When faced with this acti-
vation of multisensory cortical structures, the brain chooses—decides—that
the perceived movement is due to the visual world moving.
In reality, then, these illusions are solutions that the brain has found to in-
terpret a relationship between a perceived change in the vertical subjective and
maintenance of a visual stimulus.
The Cemetery Illusion and Other Illusions Connected to

Aerial Acrobatics
I have just described two illusions related to changes in the apparent direction
of gravity under conditions of motionlessness (the illusion of the moon) and
acceleration. Under other conditions (for example, airplane flight), accelera-
tions combine in a much more complex way, and this leads to illusions that
can have very serious consequences. In Chapter 2 I mentioned that on take-
off, the pilot of an airplane can have a false impression of bodily orientation,
connected to ambiguous messages from the otoliths (Figure 13.2). In the case
of more complex configurations, a whole catalog of illusions has been de-
scribed—especially by the American military—that have led, and still do, to
dangerous accidents.
For example, the inverse illusion occurs when a plane climbs rapidly and
then gradually regains the horizontal. You can feel minor effects from it in a
246 •
Resultant
Acceleration
Inertia
Gravity
(true vertical) Resultant (perceived vertical)
Figure 13.2. The ambiguity of otolithic messages can cause accidents. On take-
off, an airplane simultaneously accelerates and inclines. When takeoff occurs in
the dark or in fog, the pilot has only vestibular and tactile information to rely
on to detect the pitch of the airplane. The otoliths are stimulated by two accel-
erations: gravity and the forward motion of the airplane. The two component
forces (gravitational and inertial) combine in a resultant force that is perceived
by the brain as gravity (in the absence of any other visual cue to indicate up or
down). Thus, the pilot believes that his airplane is inclining very sharply. He
compensates for this illusory pitching, and the airplane crashes at the end of
the runway. Fortunately, modern instruments provide pilots with a measure of
the true pitch of the plane.
jet plane when it has finished its climb and suddenly levels off. Owing to the
plane’s circular or parabolic trajectory, centrifugal force combines with gravity
to produce a force oriented upwards. Under these conditions, the pilot, who
had the impression during the climb that he was sitting down, his bottom
pointing toward earth, suddenly has the illusion that he is upside down. This
illusion is also experienced in space flights: when gravity disappears, it is as if
there were an acceleration in a direction contrary to normal gravity. The brain
interprets the absence of gravity to be the result of acceleration in the oppo-
site direction, and the illusion of inversion felt by the cosmonauts is very often
associated with serious symptoms of motion sickness.
Another illusion, called the cemetery illusion because it has been the cause
of numerous catastrophes, may affect pilots in airplanes performing acrobatic
247
maneuvers descending in a helix. After several helical turns, the pilot loses the
sensation that he is turning. Indeed, if the angular acceleration is constant, the
sensations from the semicircular canals abate. When the pilot stops his rota-
tion, he receives a vestibular stimulus due to the combined effects of angular
deceleration and gravity that gives him the impression that he is turning in the
opposite direction, which leads to fatal accidents. This illusion, which occurs
in exceptional conditions, also occurs under more normal conditions when,
for example, the pilot makes a simple circle in a plane flying horizontally for a
fairly long while. After a time, the messages from the canals disappear, and at
the instant the plane resumes a straight-line trajectory, the acceleration caused
by straightening out gives the pilot the mistaken impression that he is inclin-
ing in the direction opposite to that of the circular trajectory. He inclines the
airplane further to correct it, which is obviously very dangerous.
Illusions of Movement of the Limbs
Kohnstamm’s Illusion
An illusion that is very interesting because it causes observable movement was
described for the first time by a physiologist, Kohnstamm. The illusion is evi-
dence of a sort of motor memory, and it seized the imagination of neuro-
physiologists. Try the following experiment: Stand upright next to a wall, in
profile, the back of your hand against the wall and the palm of your hand
against your trousers (or skirt). You can also do the same experiment without
standing next to a wall, but leaning the back of your hand on the edge of the
table. Now, applying as much force as you can, push the back of your hand for
2 or 3 minutes against the wall or the table, as if you wanted to push it away
from your body. What you are doing is applying a sustained tonic force involv-
ing the extensor muscles of your arm. At the end of this time, move away
from the wall or the table, and relax. The experiment can be done entirely
with your eyes open, but it is better to do it with your eyes closed. You will
feel your arm lift all by itself, as if it were pulled by an invisible thread, and re-
main elevated for an instant. This illusion tells us something about the basic
mechanisms of postural control. It could in fact be called involuntary post-
contraction.5
The experiment can also be done with the feet, by sitting in a chair and
pressing very hard against a wall or a table. The illusion only results when the
muscle making the contraction is a proximal muscle—one that belongs to the
postural system. Kohnstamm’s phenomenon cannot be induced using the dis-
248 •
tal muscles of the hands or the feet. For instance, if you hang a 3-kilogram
weight from your foot and let the weight drop abruptly, the involuntary con-
traction appears in the quadriceps (thigh muscle). Similarly, if you hang a 3-ki-
logram weight from your hand for 5 minutes, the involuntary contraction
phenomenon is induced in the biceps muscle, not in the hand. Thus it is a mo-
tor reaction of the postural muscles. Why is this reaction the result of an illu-
sion? Why is it the solution to a problem of the nervous system?
The interpretation of this phenomenon is the following: During the phase
when your hand is pressed against the wall, you are exerting a constant force
on it. The brain adapts to this situation in which motionlessness is accompa-
nied by a constant muscular effort. When your hand pulls away from the wall,
the brain continues to apply this force, because the brain reads the force as
motionlessness, up to the point where the stretch receptors in the muscles sig-
nal clear movement.
In other words, while force is being exerted, the brain learns that motion-
lessness is accompanied by motor activity. Once the obstacle is removed, it still
believes this to be true.
Drawing with Vibrations

At the beginning of this book (Chapter 2), I touched on illusions of bodily self-
motion induced by the vibration of tendons, and I considered the interpreta-
tion proposed by several physiologists: activation of muscle receptors causes a
sensation of movement that simulates an actual movement. A recent series of
experiments went beyond this simple interpretation, showing that the con-
figural vibration of the muscles of the hand can cause design illusions with
complex geometric shapes. Little vibrators were applied to pairs of muscles
that normally control hand gestures. For example, take a pencil and draw a cir-
cle, or a square, or a triangle: the activation sequences of the muscles of the
fingers are very different depending on whether you draw one or the other. Vi-
brating the same muscles in order of activation stimulates the sensory re-
ceptors that follow different sequences corresponding to various geometric
shapes.6 Remarkably, a subject thus vibrated, who has not moved his hand,
will have the illusion that he is drawing the corresponding shapes. Subjects are
genuinely convinced that they have moved their fingers and are very surprised
by the sensation, because they did not intend to move them. As in the case of
vection, this nonintentional illusion of shape generation is generally felt as a
very agreeable experience.
These results can be interpreted in the following way: the parameters of
the illusory sensation can be expressed at any moment by a vector whose ori-
249
gin coincides with the extremity of the limb and whose direction, in the case
of a curvilinear shape, is tangential to the trajectory of the drawing (a turning
vector). The radius of curvature of the illusory drawing is determined by
the velocity with which the frequency of the vibration applied to the two
covibrated muscles increases or decreases. For example, when the change in
frequency is stable and the same vibrational profile is applied to all successive
pairs of muscle groups of the wrist, the sensation is that of a circle. When the
change in frequency varies from one pair of muscles to another, the sensation
is that of the shape of an ellipse, and so on. Another important aspect of this
illusion is the fact that subjects recognize real geometrical shapes (triangles, el-
lipses, circles); in other words, shapes in the Gestalt sense and not simply out-
lines of movement.
But, as the authors of the experiment write,
it is also worth mentioning that, in addition to perceiving illusory move-

ments in response to proprioceptive stimulation, the subjects were able
to recognize and to describe accurately the motor forms elicited by the
various vibratory patterns used. This means that the various instanta-
neous kinesthetic sequences must have been memorized and linked up
into identifiable geometrical figures belonging to a repertoire of possi-
ble geometrical shapes. It therefore emerges that muscle proprioception
is able to generate spatiotemporal inputs that may subserve complex
cognitive operations such as those involved in the memorizing and rec-
ognition of motor forms, including symbolic actions performed with
the human hand such as writing or drawing.7
These results are important for us because they confirm the revival of the mo-
tor theory of perception. They reinforce the concept of a prespecification of
sensory inputs during control of movement. Finally, they validate the idea that
illusions are solutions resulting from an endogenous repertoire of motor or
perceptual forms with which configurations of sensory inputs are compared.
Space and Motion Sickness
Everyone has had the misfortune of experiencing the disagreeable symptoms

of motion sickness in a car, on a boat, (infrequently) on a train, and even at the
movies. What mother has not told her child at least once, “Stop jumping
around like that! You’re making me queasy”? Motion sickness is thus con-
250 •
nected to movement, whether visual or corporeal. Astronauts are susceptible
to it, and for most the only relief is medication.
The mechanism of this peculiar bodily reaction is still not known. One in-
teresting hypothesis that so far has not been rejected is called sensory conflict.
It explains motion sickness as a conflict between the sensory messages sup-
plied by the visual and vestibular systems (and possibly the graviceptors of the
abdomen) about movements of the head. More generally, this symptom is
produced by an incongruity between the messages relayed by the sensors of
orientation and those of movement. For example, the visual system of a child
sitting in the back seat of a car or a sailor below deck detects no movement,
whereas the vestibular and proprioceptive systems do detect movement. Simi-
larly, in the case of complex movements of the head that generate so-called
Coriolis accelerations, the brain cannot establish coherence between accelera-
tions detected by the vestibular system and those indicated by vision. In space,
it is the failure of the otoliths to detect tilting of the head owing to the ab-
sence of gravity versus the ability of vision and the other sensors to detect it
that is the cause of sickness, although cardiovascular reorganization may also
contribute.
Thus, difficulty in reconstructing a coherent perception of movement and
orientation is responsible for central nervous system activity leading to succes-
sive stages of motion sickness, from drowsiness to nausea. We still do not
know why this incoherence manifests vegetatively as it does. It has long been
known that dogs whose flocculus (the portion of the cerebellum that inte-
grates visual and vestibular information) has been removed do not get sick. So
it is possible that low-level mechanisms are responsible for motion sickness.
But I would like to propose another hypothesis.
The hippocampus contributes to the memory of configurations of sen-
sory information and others connected to action, and at this level visuo-
vestibular convergences are probably treated on a global level. For example, in
the monkey the neurons of the hippocampus discharge during movements of
translation and rotation, but only when the animal is in a certain region of the
room, or is heading toward a door.8 It is thus possible that motion sickness is
caused by much higher level mechanisms involving the entirety of the internal
system of simulation of global movement of the body in the space con-
structed by the hippocampus, the prefrontal cortex, and the parietal cortex.
Problems in establishing coherence might lead to vegetative effects whose ori-
gin is cortical.9 This hypothesis explains why training plays such an important
role in preventing motion sickness in space and, among other things, why it

251
can be controlled by exercises that mentally manipulate orientation, as I men-
tioned at the end of the previous chapter. It is also possible that the two mech-
anisms exist concurrently, which would explain why this malady is so resistant
to study and why it is so difficult to devise countermeasures.
If you wish to avoid motion sickness, do not read on winding roads or
when the driver brakes abruptly when you are riding in a car. Focus on the dis-
tant landscape so your visual system has the same reference as your vestibular
system. And when you are on a boat, go up to the deck and look at the hori-
zon: this provides your vision with the exact same reference as the vestibular
system. Once on the deck, make active movements to remain stable, for these
motor commands will add their messages to the other sensory information.
And if you are still feeling queasy, you may console yourself by recalling that
Admiral Nelson was seasick when at sea for his entire life!
A Few Other Illusions
The Sailors’ Illusion

There is another illusion that sailors feel after several days at sea. When they
return to port, they have the impression that the ground is undulating in the
same way as the sea. This illusion reflects a remarkably flexible solution that
explains a lot about how the nervous system adapts. When the illusion occurs,
it means that populations of neurons are animated by a rhythmic modulation
in the centers that detect movement (the vestibular nuclei, vestibular cortex,
and so on). How is this memory of movement constructed? It isn’t known for
certain, but it is assumed that at sea the brain establishes a dynamic modula-
tion in phasic opposition (antiphase) to the undulations of the water. Thus,
when the head is accelerated, this endogenous wave causes the activity in the
centers of movement to diminish. The result is a sensation of stability. And, in
fact, when you are on the water for several days, you eventually forget the mo-
tion of the boat. The same phenomenon is produced when you are walking:
after some time, you no longer have the impression that the visual world is
unfolding so rapidly. In contrast, when you stop after an hour of walking, you
may perceive an apparent motion of the environment, a little like the famous
illusion of the waterfall: when you look at a waterfall for some minutes and
then you look round about, you perceive an apparent movement of the moun-
tain. A dynamic memory of movement establishes a dynamic neural profile
that stabilizes the perception and that, I think, enables the brain to free itself
of this perpetual stimulation to be able to think about something else. Illu-
252 •
sions that result from exposure reveal the solutions the brain has devised to
adapt.
Vertigo and Agoraphobia: Illusions without Solutions

Vertigo is an illusion that has no solution. Fear of heights results when pro-
prioception of the legs signals small oscillations to the brain but vision detects
no movement, owing to the remoteness of the landscape. It is also due, in cer-
tain cases, to difficulty in processing visual information. Why are some people
unable to resolve this conflict? No one knows, but I think that my hypothesis
of a high-level mechanism involving the structures of the cortex or of the
limbic system responsible for perception of the body and the environment is
one worth pursuing.
The same is true for agoraphobia and the so-called panic attacks that come
over people when they suddenly find themselves in wide-open spaces. A body
of research indicates a possible disturbance of vestibular origin in these pa-
tients, and we are investigating these mechanisms in the laboratory. But the
idea of a breakdown of the representation of space and difficulty in construct-
ing coherence among multiple corporeal frames of reference and sensory
messages also bears looking into. Panic attacks are a consequence of an inabil-
ity to find a solution, even in the form of an illusion!
Illusions and Hallucinations

Illusion is a solution to an incongruity, to the loss of perceptual coherence.
Hallucination, which is a creation of the brain, is a different story entirely.10
Hallucination is not the result of sensations that the brain fails to integrate
into a coherent perception but of the sudden combination of endogenous
memories of perceptions. In some sense, hallucination is a waking dream, the
autonomous functioning of internal circuits that normally work to simulate
the consequences of action. Somewhere in the brain a purely internal activa-
tion initiates perceptual hypotheses. Like the painter of Lascaux for whom
shadows suggested the images of the animals he hunted and that he projected
onto the cave walls, here a pathological internal activity triggers the hallucina-
tion. Hallucinations are such that they are projected onto the world, exempli-
fying a fundamental property of the brain. Whereas illusion provides the solu-
tion to conflict, hallucination constitutes clear evidence for the central theme
of this book: the anticipatory, projective function of the brain imposes its as-
sumptions and memories on the world, and it reconstructs movement based
on the slightest hint of change.
Dreaming is also essential evidence. Shepard, speculating on the nature
253
of dreams, wrote: “Thus, although J. Gibson (1970) held that perceiving is
an entirely different kind of activity from thinking, imagining, dreaming, or
hallucinating, I like to caricature perception as externally guided hallucination,
and dreaming and hallucination as internally simulated perception. Imagery and
some forms of thinking could also be described as internally simulated percep-
tions, but at more abstract levels of simulation.”11
254 •
14
A R C H I T E C T S H AV E F O R G O T T E N T H E P L E A S U R E O F M O V E M E N T
Perception is inseparable from action. It predicts the future. It was organized

over the course of evolution according to the natural properties of the physi-
cal world and of biological mechanisms. Not so long ago, our natural world
was replaced by an artificial world of cities, trains, and airplanes. Architects
are responsible for constructing an environment that meets our brains’ expec-
tations. During three thousand years of architecture, humans have experi-
mented with many varieties of habitats. I think that most contemporary archi-
tects have betrayed these centuries of research. What they have built and
continue to build is a tragedy for our brain, its emotions, and the pleasure it
takes in movement.
Evolution made the brain particularly sensitive to the basic elements that
constitute perceived nature, our milieu, our Umwelt, as Uexküll says. The first
of these basic elements is regularity. Regularities can be geometric, like crys-
tals or leaves—for instance, the way a leaf stems from a tree branch. They can
be rhythmic, like the undulation of branches in the wind, waves on the sea, or
sounds that form regularities in time. The second of these basic elements is
chance, which disrupts regularity according to its own laws. We like this game
of regularity disrupted by chance.
The third element is movement. Sometimes it demands attention, like the
movement of the sea etched in its pebbles and rocks, and reflected in the
shape of boats; the movement of the wind, which shapes plants, trees, houses,
and the skin, fur, and skeletons of animals who had to adapt to resist it, even
animals so small as insects; the flicker of fire, light and elusive, which suggests
shapes whose very transience delights us. Not only is there movement in the
elements and in animals, there is also human movement: the gestures of har-
vesting, which dictate the height of trees (and perhaps even the choice of cer-
tain fruits); and the gestures of manual capture, which determine the shape of
the objects we choose to grasp, throw, and catch.
These three basic elements—regularity, chance, and movement—are what
make nature appear as it does. They are also the elements that constitute per-
ception. Among all the people who build artificial objects—engineers, couturi-
ers, and so on—it seems that only architects have forgotten movement. De-
signers of cars, airplanes, and bicycles have necessarily integrated movement
into the architecture of the things they create: a Boeing 747, a high-speed
train, a beautiful car, a fast motorboat, or a sailboat. Even vacuum cleaners
have a pleasing shape! Their rounded forms reflect both their status as moving
objects and our concern that they not bump into the corners of tables.
But in the last twenty years architects have forgotten movement. They
build gigantic structures that resemble filing cabinets, and with a few excep-
tions, like Niemeyer, André Bruyère, or Ricardo Porro, use only straight lines.
By ignoring movement, they offer nothing but flat and square houses to
explore. The monotony of their boxes, built on piles and covered with what
looks like bathroom tiling, discourages lingering glances. These buildings
deny our senses the pleasure of shape and movement that curves and mass
would provide.
For painters, use of a vertical and horizontal frame for their canvases pro-
vides a point of reference, probably required by the selectivity of visual neu-
rons for these directions as well as the influence of the gravitational vertical
and the vestibular system. With distinctly different effect, architects trap our
gaze in bleak prisons of lines that instead of fluidly intermingling, intersect at
the one angle guaranteed to produce an impression of a collision, an accident,
a painful shock: the right angle.
I realized this the day a young architect told me that she had to construct
the grid for her project. This rectangular network is a yoke on the imagina-
tion. Of course, had it not been for Thales of Miletus, who invented the grid,
there would be no modern cities. But that is not the point. It has become an
architectural end in itself, as well as the end of architecture.
I actually like the rectangle of a window or a painting. But, oh, right angle,
you are neither in a wave on the ocean nor in the wind, nor in the knowing
coincidence of a sigh and a smile; you are neither in the petal nor in the leaf.
You express the triumph of the most ordinary, lazy, petty geometric mindset
over the sense of finesse, which might save me from despair.
Perception, which is simulated action, needs to find natural or artificial ob-
jects in the environment that imply action. So our brains take pleasure in play-
256 •
ing, in guessing the real and the false, in lying, in laughing and crying, in cap-
turing and fleeing, in predicting the future—in a word, in living. The architects
of the Grande Bibliothèque, of the Opéra-Bastille—magnificent inside, but
from the outside, like a big bathtub—of Beaubourg, now in the process of be-
ing refurbished, tried above all to arrange people and things in an orderly way,
but they never thought about the pleasure of those who would visit or inhabit
these spaces. Thus they condemned us to boredom. I accuse them of the
crime of melancholy, of leading millions of people to despair, of crimes
against the biological brain—its flexibility, its desire for movement and varia-
tion.
My generation was also to blame: we denuded the apartments we bought
of gilt, friezes, and the stucco decorating the walls, and now the ceilings of
the lower middle class dream of the splendors of Versailles. We admired Bau-
haus and read the theories of its masters, Kandinsky and Gropius, simplistic
though they were, who said that architecture, clothing, and objects should be
combinations of three prototypic forms: circle, square, and triangle. Per-
haps they were right after all. The recent work of Sakata, a Japanese neuro-
physiologist and disciple of Merleau-Ponty, demonstrates that neurons in the
parietal cortex are activated in a specific way by shapes such as a cylinder.
These findings might confirm the Bauhaus theory of prototypic shapes on
which perception is based and on which artificial objects or houses must also
be based.
But the brain does not merely store a static combinatorial system of
shapes; it extracts these shapes from movement. The brain can extrapolate the
three-dimensional shape of objects based on the movement of a cluster of
points on a flat screen. Movement gives rise to structure.
Today, some architects have rediscovered the old compass in their desk
drawer and are trying to create an illusion of movement by using semicircles,
a shape that is still acceptable in calculating profit and in generating the Pert
charts that are used to monitor the progress of a building project. But this is
merely a mockery of the golden rules that intelligence and refinement have
taught over the centuries, according to which curves are pleasing to the hu-
man body and mind. There is no more boring shape than a circle unless it
takes the form of a crêpe Suzette, or an ellipse unless it is seen in the context
of other shapes.
In the 1950s, Jean Prouvé, an architect who was ahead of his time,
preached the advantages of building with glass and iron. His slides showed the
hopeless entanglement of cities, a confusion of wires, poles, houses, and
apartment blocks made from many materials. He advocated a return to a pu-
ARCHITECTS HAVE FORGOTTEN THE PLEASURE OF MOVEMENT •
257
rity pleasing to the spirit, to a uniformity and lightness that offered new possi-
bilities owing to the flexibility of these materials and to the prospect of letting
light in. He was, I think, betrayed by money grubbers and victims of priorities
and mass projects, who saw in these materials only the possibility of using
them to make giant Lego constructions.
The architect Nouvel had no better option than to build the cultural cen-
ter of the Boulevard Raspail out of glass. The only thing his generation knows
how to do is to reflect the architecture of Paris. Look at the Grande Biblio-
thèque from the opposite bank. You will not detect any difference between its
boxes, which purport to look like books, and the low-rent apartment buildings
of the XIIIth arrondissement that form the backdrop. The real shape of a book
is what it contains, and what beauty there is in the geometry of the letters!
This mirroring in most buildings is a loud and honest admission of the
failure of modern architecture. In it, you see everything around, the sky as
well as the city of Haussmann.1 After three thousand years of art, architects
have invented a nonarchitecture that reflects the image of previous centuries.
Nevertheless, this architecture chills me, because it interjects a screen be-
tween me and the city, which prevents me from touching it, from palpating it
with my eye. These buildings do not invite my desire, but rather form a wall
off of which my vision bounces. They no longer portray the weight of the
snow sliding off a roof, the breath of the wind that buildings must resist, the
trickling of water, the slow movement of the sun, or the gesture of opening a
window.
Yet further proof of this disdain for subtlety is the disappearance of
wrought iron. Lift your eyes in Paris! See how your gaze delights in following
the infinitely varied swirls of Parisian balconies. These balconies of wrought
iron play an essential double role: they please the eye and break up the regu-
larity of the facades by giving them a rhythm. When you are inside the house,
they provide a transparent screen that divides your view of your neighbor into
endlessly different fragments depending on the room you are in. They frame
faces perceived on the street in a thousand new and pleasing ways. Their
shapes play with the leaves of the sycamores and chestnut trees, the move-
ment of the wind sometimes impelling, sometimes impeding. They prevent
the dispiritedness that is a by-product of living in the same neighborhood for-
ever. These forms and spaces suggest movement; they symbolize the possibil-
ity of escape.
Now look at what architects have given us in recent years. Prison bars!
The appalling repetition that the brain so despises, shape without variation,
258 •
the curse of preemptory predictability. The brain is a biological forecasting
machine. It follows that its pleasure consists in taking gambles. And it can only
gamble on a reality in motion, ever changing. Shape, even motionless shape, is
an opportunity for mind shifts, for imaginative changes of direction, which
criminal architects would prevent us from enjoying.
I recently visited a renovated building that had a staircase going up to the
second floor. It was extended from the building’s original lower staircase. I
compared the two banisters. The more recent one was square, as rigid as an
insensitive person, sad and gloomy, going nowhere; the staircase that belonged
to the old building had that marvelous curvature you see in Parisian banisters,
the legacy of a thousand years of craftsmanship. It showed the way without
hesitation, its subtle helix contrasting with the drumbeat of the steps that sec-
tioned space and promised a difficult, jerky climb. The old banister was shape
and movement; already I felt in the palm of my hand the rounded curves that
embraced me as I got my spinal locomotor generator into gear. The banister
was refinement itself, the lover of the craftsman who had created her. She was
unlike any other, she was calm regularity, she was movement. I loved her.
What can I say about the grillwork on doors? The best example of failure
is that of several schools in Paris that shall remain nameless. For the hundreds
of children who cross its threshold each day, the schoolyard gate is the transi-
tion from family to school, between the city and this neutral and protected
place that the philosopher Alain was so fond of. The gate is the symbol of a
touching, eminently contradictory moment of transition between two worlds:
the place where parents wait, and the passage through which children take
flight at the close of school. It is a symbol both of enclosure and of liberation
from family constraints and the teacher’s clutches. This object must serve as a
border and a crossing, a versatile window on the city and a peephole for catch-
ing a glimpse of the sanctuary of school. The gate must be pleasing to cling
to, to resist, to rest one’s eyes on, to distract one’s gaze while waiting, or to
examine for symbols of a changing world. What have architects made of
it? Prison bars. Total renunciation of the unexpected gesture, the desired
meaning.
It took me a long time to understand why the architect André Bruyère was
so interested in doors, but now it is obvious to me. The door is one of the
most important elements in architecture, experienced as a front for emotions,
fading actions, and the construction of memories.
I understand about priorities. I understand the requirements of economies
of scale and of standardization. I understand that the straight line is the short-
ARCHITECTS HAVE FORGOTTEN THE PLEASURE OF MOVEMENT •

259
est path from one point to another, and that the curve is a luxury that poverty
cannot afford. Nor do I deny the importance of repetition. It is a component
of our body. It is pleasing to the brain.
But look at a face. It engages us because it is organized both with a rigor-
ous geometry and a marvelous suppleness of form. Nothing about it is orna-
mental; everything about it is functional; yet nothing about it is fixed. Every
expression is accompanied by some action, by the movement that shapes it.
But these movements are subtle, like the line of my lips when I imagine gently
crushing a morsel of chocolate cake in a gourmand’s kiss. The shapes are free
and unconstrained; they are at the service of my imagination, of my desire to
act, and neither master nor jailer.
260 •
C O N C L U S I O N : T O WA R D A T O L E R A N T P E R C E P T I O N
Now it is time to reflect upon the distance covered and try to sum up the ideas
presented in this book. To be honest, these ideas are more a work in progress
than a finished theory. Let me suggest some thoughts to begin the discussion.
The Computer Metaphor: A New Dualism
The brain is neither a computer nor an artificial intelligence machine. It is an

original biological structure, the product of the slow and astonishing work of
evolution and, in the case of the human brain, of history and culture. The
brain-computer metaphor is useful as a guide to thinking about the brain, but
it is very dangerous. Indeed, it implicitly assumes a separation between com-
plex cognitive processes (so-called computational processes similar to soft-
ware) and a neural substrate whose molecular biology and neurobiology dis-
close its extraordinary capabilities (still largely unexplored). For many years
this metaphor bolstered the functionalist theory, which was recently revived
by one of its most ardent defenders, the philosopher Hilary Putnam. It is pos-
sible to see in it a modern form of Bergson’s mentalism, which I cannot agree
with when he says, in substance, that the mind is to the brain what the coat is
to the hook that holds it.
Listen to Bergson:
Anyone who approaches, without preconceived ideas and on the firm

ground of facts, the classical problem of the relations of soul and body,
will soon see this problem as centering upon the subject of memory,
and, even more particularly, upon the memory of words: it is from this
quarter, undoubtedly, that will come the light which will illumine the
obscurer parts of the problem . . . Speaking generally, the physical [sic]
state seems to us to be, in most cases, immensely wider than the cere-
bral state. I mean that the brain state indicates only a very small part of
the mental state, that part which is capable of translating itself into
movements of locomotion. Take a complex thought which unrolls it-
self in a chain of abstract reasoning. This thought is accompanied by
images, that are at least nascent. And these images themselves are not
pictured in consciousness without some foreshadowing, in the form of
a sketch or a tendency, of the movements by which these images would
be acted or played in space—would, that is to say, impress particular at-
titudes upon the body, and set free all that they implicitly contain of
spatial movement. Now, of all the thought which is unrolling, this, in
our view, is what the cerebral state indicates at every moment. He who
could penetrate into the interior of a brain and see what happens there,
would probably obtain full details of these sketched-out, or prepared,
movements; there is no proof that he would learn anything else.1
In other words, Bergson accepts the idea that the brain is used in the inter-
nal simulation of movement, but he limits its contribution to this mental
structuring and refuses to grant that it plays any role whatsoever in the high-
est cognitive functions that are the exclusive domain of the mind. Moreover,
in the same passage he says that the whole of psychophysiology provides the
brain with a level of comprehension no greater than that of an observer
watching the comings and goings of actors on stage to understand the play.
Bergson is confident, however, that the brain is a tool for simulating. He
writes: “I should say that the brain is an organ of pantomime, and of panto-
mime only. Its part is to play the life of the mind, and to play also the external
situations to which the mind must adapt itself. The work of the brain is to the
whole of conscious life what the movements of the conductor’s baton are to
the orchestral symphony.”2 The computer metaphor also assumes an implicit
dissociation between the hardware (the neurons of the brain) and software
(the symphony).
The Sense of the Senses
There are more than five senses. Consider, for example, the vestibular sensors,
the proprioceptors of the muscles and joints, and, for certain species, other
senses such as echolocation and magnetic sense. But to give any meaning to
262 •
perception, what is needed most is a complete revision of the word “sense.” I
suggest returning to a classification of the senses that corresponds to percep-
tual function. So to the sense of taste and smell, touch, vision, and hearing,
add, as the vernacular already does, the sense of movement, space, balance, ef-
fort, self, decision, responsibility, initiative, and so on. This idea of the senses
shows the way, determined by the subject, toward a goal. To varying degrees,
separately and together, each of the senses calls upon the receptors, and even
certain properties of each receptor, that it needs. The brain filters the mes-
sages supplied by the senses according to its own plans. The mechanisms of
this selection have yet to be deciphered: we understand only a few forms of se-
lectivity.
In other words, we must completely change the way we study the senses.
We must begin by considering the goal of the organism, so that we may un-
derstand how the brain is informed by the receptors, how it regulates their
sensitivity, combines messages from them, and estimates their value, accord-
ing to an internal simulation of the expected consequences of action.
The relationships between perception and action require study that takes
into account several key features of the brain and their complementarity,
which no one yet has been able to characterize. According to so-called central-
ist theories, action is determined by internal plans of action (my projective
formulation of a schema) whose mechanisms are endogenous and are based
on a repertoire of innate and acquired behaviors characteristic of each species
and each individual. Functions are not confined to a single structure of the
brain but are the result of cooperation among specific structures. This cooper-
ation constitutes pathways for neural activity that is characteristic for each
structure. Many brain functions are not just the product of pathways along
which processing proceeds sequentially, but of closed circuits. Within these
circuits, structures that are activated sequentially, or in a synchronous way
through oscillations, are often connected by what Edelman calls “reentrant
loops.” These functional loops interact at several levels and thus influence
each other in multiple ways, and this interconnectivity is subtended by assem-
blies of very specific neurons. In other words, it is wrong to think that in the
brain everything is connected with everything else.
The subsystems that define a repertoire of sensorimotor behaviors are
subject to variability. The brain has mechanisms for exploiting this variability
and combining elements of the repertoire to form strategies that are specially
adapted to its current purposes. I think that functional flexibility, recently dis-
covered at the level of the cortex, is only a subsidiary manifestation of this
fundamental property. The brain can choose a variety of solutions to the same
CONCLUSION •
263
problem. This ability to choose is at the root of illusions, which are solutions
to problems of perceptions that are ambiguous or incompatible with environ-
mental cues. In contrast, hallucinations, like dreams, are creations that the en-
vironment cannot confirm.
The Problem of Coherence
The central issue is the neural basis of coherence. Indeed, all the findings of
neuropsychology and of modern neurophysiology confirm the multiplicity of
cooperating subsystems, the role of segregation, the modularity of cerebral
operations, and their interrelatedness. There are many frames of reference in
which the body is represented; the body schema may be nothing more than
the synthesis of multiple schemas through as yet unknown mechanisms. Each
sense breaks down perceptible reality into elements that are subsequently re-
constructed and connected. The concept of coherence is thus central. The
theory of temporal encoding asserts that temporal synchronization consti-
tutes de facto coherence through simultaneity.
I think that major illnesses like autism, schizophrenia, and agoraphobia
share disturbances of coherence that take diverse forms according to the prin-
cipal features of each illness.
A Memory for Predicting the Future
Memory is essentially a tool for predicting the future. In this sense, the con-
cept of working memory and its association with the prefrontal cortex is im-
portant, but it accounts for only a portion of memory. Researchers began with
the idea that memory served the brain to recall bygone experiences of youth
in old age. Then it was discovered that this memory could in fact be broken
down into several categories associated with various parts of the brain, and ul-
timately one of these categories is working memory. I think that it is critical to
start with the idea that memory is a mechanism that enables us to predict the
future. The simulating brain uses memories to implement mental processes of
prediction. So the hippocampus—the highest level for encoding messages in
the nervous system, a supramodal level that takes into account the set of rela-
tionships between behavior and the messages encoded in each specialized sys-
tem—together with the prefrontal and parietal cortex might play a central role
in memory as a guide to future action: it reconstructs multisensory episodes
from partial elements, as in the example of Proust’s madeleine. But many
other centers participate in the memory of movement. The parietal cortex is
264 •
probably involved in long-term motor memory and in the mental simulation
of movements;3 the cerebellum, along with the basal ganglia, in manual dex-
terity, and so on.
Perception and Emotion
We are far from understanding the relationships between perception and emo-
tion. Yet we have just entered a century that Malraux said would be either “re-
ligious or not be at all.” If recent expressions of religion betraying only intol-
erance, fanaticism, and hatred are any indication, this century, if it is religious,
will be a century of unprecedented violence. Citizens in every country have
an enormous responsibility. To be able to wander about with a bomb strapped
to one’s back as a young Palestinian fanatic did in March 1995, to murder
thousands of Muslims as the Serbs did, or a million Cambodians as did Pol
Pot, to murder a playmate in his backyard, like two French youths who were
caught up in a game of fantasy role playing, to kill hundreds of one’s own fol-
lowers—to commit such acts the mind must be so constrained by the percep-
tion that fascinates it that it suppresses any alternatives. The functioning of
the brain is corrupted, as it is with hypnosis. Perception and action should be
interesting not only to educators, ergonomists, or physicians, but to all those
who feel a responsibility to fight the hatred of others. My generation believed
that the basis of selfishness was economic. We were not completely wrong,
and the present return to the ruthless struggle between the strong and the
weak and the disappearance of feelings of solidarity can only result in sorrow
and suffering.
But there is more. The human brain operates with powerful mechanisms
of territorialism, of distrust toward those who are different, of selfishness that
precludes the open hand of generosity. The most bloodthirsty dictators are
supported not only by economic forces but by the forces of obscurantism,
which derive their hateful influence from archetypal images that are stiff,
petty, and grotesque, and which blot out consideration for the other and jus-
tify his destruction. In this book I have maintained that the brain projects its
internal perceptions onto the world, that it constructs its perceptions with ref-
erence to actions, much the same way that Lascaux man saw bison on the
walls of his cave. Unfortunately, this leaves the brain with a propensity for re-
treating into pre-established schemas that it then projects onto the world and
onto others. Skilled alchemists of the mind can thus manufacture ready-made
perceptions or caricatures, like those seen in the antisemitic exhibitions orga-
nized under the Vichy regime, or those presently being forced on people
CONCLUSION •
265
around the world to feed religious wars. The same process is probably at work
in sects, where gurus painstakingly brainwash their followers, distorting the
way others are perceived, so that perception of them is tainted by uncondi-
tional adoration of the master. And what is the point of playing this role, if
not to impose a schema for perception that defines the relationships the guru
compels a person to maintain with others? All these various social structures
have a common mechanism for restricting the operation of the brain within a
set of rigid interpretations, using methods somewhat akin to hypnosis. This
perversion of the faculty—the projective nature of perception—must be re-
sisted. It is a scourge graver than the gravest illness, for it does not merely de-
stroy the person; it leads its victims to destroy others.
Anger and hatred often lead us back to these perceptual prototypes, which
we suddenly project onto every object or every person in sight. When we do,
anticipation becomes a prison for perception and a trap for action; the serene
path of reason is abandoned for that of the emotions. Tolerance requires a
generous and benevolent appreciation of differences. Along with the pleasure
of discovering the mechanisms of perception and action, and the obligation to
enhance well-being, human beings also have a duty to understand the perver-
sions of the mind, so that this new century may be one of tolerance. Failing
that, we risk the worst.
266 •
NOTES
WORKS CITED
CREDITS
INDEX
NOTES
Introduction
1. J. von Uexküll, Umwelt und Innenwelt der Tieren und Menschen (Berlin: Springer, 1934).
2. G. Châtelet, Les Enjeux du mobile—Mathématiques, physique, philosophie (Paris: Le Seuil,
1993), p. 39.
3. Johann Wolfgang von Goethe, Faust, trans. Randall Jarrell (New York: Farrar, Straus,
and Giroux, 1959), p. 61.
4. Immanuel Kant, Critique of Pure Reason, trans. Norman Kemp Smith (New York: St.
Martin’s Press, 1961), p. 47.
5. H. Bergson, The Creative Mind, trans. Mabelle L. Andison (New York: Philosophical Li-
brary, 1946), pp. 243–245. I have recently published an anthology of one hundred years
of the “Leçons inaugurales” [inaugural lectures] of the professors at the Collège de
France that indicate this central role of physiology in cognitive sciences in France. See
A. Berthoz, Leçons sur le corps, le cerveau et l’esprit (Paris: Odile Jacob, 1999).
6. T. Ribot, The Psychology of the Emotions, Contemporary Science Series (London: W.
Scott, 1897), p. 2.
7. J.-P. Changeux, L’Homme neuronal (Paris: Fayard, 1983).
1 Perception Is Simulated Action

1. P. Viviani, “Motor-Perceptual Interactions: The Evolution of an Idea,” in M. Imbert, P.
Bertelson, R. Kempson, D. Osherson, H. Schnelle, N. Streitz, A. Thomassen, and P.
Viviani, eds., Cognitive Science in Europe: Issues and Trends (Heidelberg: Springer, 1987),
pp. 11–39.
2. R. H. Lotze, Medizinische Psychologie oder Physiologie der Seele (Leipzig: Weidemann,
1852).
3. H. von Helmholtz, Treatise on Physiological Optics, trans. J. P. C. Southall (New York:
Dover, 1962).
4. William James, The Principles of Psychology (New York: Holt, 1890), vol. 2, pp. 582–585.
5. P. Janet, Les Débuts de l’intelligence (Paris: Flammarion, 1935), p. 31.
6. Ibid., p. 37. Janet talks about “kinetic melodies” to designate reflex action that is in no
way a unique response but a succession of movements in a certain order.
7. Ibid., p. 43.
8. Ibid.
9. Ibid., p. 54. Bergson said, “To recognize a common object is mainly to know how to
use it.” H. Bergson, Matter and Memory, trans. Nancy Margaret Paul and W. Scott
Palmer (New York: Zone Books, 1988), p. 93; and further on, “To follow an arithmeti-
cal addition is to do it over again for ourselves,” p. 116.
10. Or, literally, “Vision is palpation by gaze” [trans.]. M. Merleau-Ponty, Le Visible et
l’Invisible (Paris: Gallimard, 1964), p. 177.
11. P. K. Anokhin, Biology and Neurophysiology of Conditioned Reflexes and Their Role in Adap-
tive Behaviour (Oxford: Pergamon, 1974).
12. Human Motor Actions: Bernstein Reassessed, ed. H. T. A. Whiting, Advances in Psychol-
ogy (Amsterdam: North-Holland, 1984), vol. 17, p. 363.
13. α rhythm is an electrical wave that can be recorded on the cranial surface when a hu-
man subject is at rest; this rhythm ceases in states of attentive alertness and is replaced
by other rhythms induced by other oscillating activities.
14. R. Llinás, “Possible Role of Tremor in the Organisation of the Nervous System,” in
L. J. Findley, R. Capildeo, eds., Movement Disorders, Tremor (New York, Oxford Univer-
sity Press, 1984), pp. 475–477; “The Intrinsic Electrophysiological Properties of Mam-
malian Neurons: Insights into Central Nervous System Function,” Science, 242 (1988):
1654–1664; and “The Noncontinuous Nature of Movement Execution,” in D. R. Hum-
phrey and H. J. Freund, eds., Motor Control: Concepts and Issues (New York: Wiley, 1991),
pp. 223–242.
15. E. N. Sokolov and O. S. Vinogradova, Neuronal Mechanisms of the Orienting Reflex
(Hillsdale, N.J.: Erlbaum, 1975).
16. J. J. Bouyer, M. F. Montaron, J. M. Vahnée, M. Albert, and A. Rougeul, “Anatomical Lo-
calization of Beta Rhythms in Cats,” Neuroscience, 22 (1987): 863–869; M. Chatila, C.
Milleret, P. Buser, and A. Rougeul, “A 10 Hz ‘Alpha-Like’ Rhythm in the Visual Cortex
of the Waking Cat,” Electroencephalography and Clinical Neurophysiology, 83 (1992): 217–
222; C. M. Gray, P. König, A. K. Engel, and W. Singer, “Oscillatory Responses in Cat Vi-
sual Cortex Exhibit Inter-columner Synchronization Which Reflects Global Stimulus
Patterns,” Nature, 338 (1989): 334–337; M. Joliot, U. Ribary, and R. Llinás, “Human Os-
cillatory Brain Activity Near 40 Hz Coexists with Cognitive Temporal Binding,” Pro-
ceedings of the National Academy of Sciences, 91 (1994): 11748–11751.
17. R. A. Schmidt, “A Schema Theory of Discrete Motor Skill Learning,” Psychological Re-
view, 82 (1975): 225–260.
18. U. Neisser, Cognition and Reality (San Francisco: Freeman, 1976).
19. D. G. MacKay, The Organization of Perception and Action (New York: Springer, 1987),
p. 7.
20. Ibid., pp. 1–2.
21. G. Rizzolatti, M. Gartilucci, R. M. Camarda, V. Gallex, G. Luppino, M. Matelli, and L.
Fogassi, “Neurons Related to Reaching-Grasping Arm Movements in the Rostral Part
of Area 6 (Area 6a),” Experimental Brain Research, 82 (1990): 337–350. Rizzolatti and his
co-workers also discovered neurons that discharge in relation to other motor actions
like precision gripping. Other teams, such as Sakata’s [H. Sakata, M. Taira, A. Murata,
and S. Mine, “Neural Mechanisms of Visual Guidance of Hand Action in the Parietal
Cortex of the Monkey,” Cerebral Cortex, 5 (1995): 429–438] in Japan found neurons in
the parietal cortex that encode simple motor behaviors. This article describes a concep-
tual schema for guiding movements of the hand.
22. J. Bouveresse, Langage, perception et réalité (Nîmes: Jacqueline Chambon, 1995), p. 54.
23. Ibid., p. 50.
24. M. Merleau-Ponty, Résumés de cours au Collège de France (Paris: Gallimard, 1968).
25. These are transistors whose properties are a model for control of membrane potential
by ionic currents and that can be used to construct so-called neuromorphic networks.
26. M. Jeannerod, “The Representing Brain: Neural Correlates of Motor Intention and Im-
agery,” Behavioral and Brain Sciences, 17 (1994): 187–202.
270 •
NOTES TO PAGES 13–22
27. M. Wexler, S. Kosslyn, and A. Berthoz, “Motor Processes in Mental Rotation,” Cogni-
tion, 68 (1998): 77–94.
2 The Sense of Movement

1. Aristotle, De Sensu and De Memoria, trans. G. R. T. Ross (New York: Arno, 1973),
p. 55.M
2. The first experiments using vibrations in humans were conducted by Hagbarth in Swe-
den around 1966, then by P. B. C. Matthews in the cat around 1970. Many groups sub-
sequently used vibrations to study mechanisms of perception and control of move-
ment, for example, Gurfinkel, in Moscow; Jeannerod, in Lyon; Roll, in Marseille; and
our laboratory. Vibrations have also been used in experiments in space.
3. The illusion of movement is thus a perceptual solution that is chosen by the brain and
leads to motor consequences. For more on this subject, see Chapter 13.
4. G. S. Sherrington, “Observations on the Sensual Role of the Proprioceptive Nerve Sup-
ply of Extrinsic Ocular Muscles,” Brain, 41 (1918): 332–343.
5. F. Emonet-Dénand, Y. Laporte, P. B. C. Matthews, and J. Petit, “On the Subdivision of
Static and Dynamic Fusimotor Actions on the Primary Ending of the Cat Muscle Spin-
dle,” Journal of Physiology, 261 (1977): 827–861; Y. Laporte, F. Emonet-Dénant, and L.
Jami, “The Skeletofusimotor or β Innervation of Mammalian Muscle Spindles,” Trends
in Neuroscience, 4 (1981): 97–99; J. Celichowski, F. Emonet-Dénand, Y. Laporte, and J.
Petit, “Distribution of Static γ Axons in Cat Peroneus Tertius Spindles Determined by
Exclusively Physiological Criteria,” Journal of Neurophysiology, 71 (1994): 722–732.
6. J. P. Roll, J. C. Gilhodes, R. Roll, and J.-L. Velay, “Contribution of Skeletal and Extra-
ocular Proprioception to Kinaesthetic Representation,” in M. Jeannerod, ed., Atten-
tion and Performance XIII (Hillsdale, N.J.: Erlbaum, 1990): 549–566; J. P. Roll, R. Roll, and
J.-L. Velay, “Proprioception as a Link between Body and Extra-personal Space,” in
J. Paillard, ed., Brain and Space (New York: Oxford University Press, 1991), pp. 112–132.
7. C. Perret and A. Berthoz, “Evidence on Static and Dynamic Fusimotor Actions on the
Spindle Response to Sinusoidal Stretch during Locomotor Activity in the Cat,” Experi-
mental Brain Research, 18 (1973): 178–188.
8. A. Berthoz and S. Metral, “A Torque-Producing Stimulator for the Study of Muscular
Response to Variable Forces,” Proceedings of the International Symposium on Biomechanics
III (Basel: Karger, 1973), pp. 158–164.
9. R. J. Nudo, G. W. Milliken, W. M. Jenkins, and M. M. Merzenich, “Use-Dependent Al-
terations of Movement Representations in Primary Motor Cortex of Adult Squirrel
Monkeys,” Journal of Neuroscience, 16 (1996): 785–807.
10. H. Flor, T. Elbert, S. Knecht, C. Wienbruch, C. Pantev, N. Birbaumer, W. Larbig, and E.
Taub, “Phantom-Limb Pain as a Perceptual Correlate of Cortical Reorganization fol-
lowing Arm Amputation,” Nature, 375 (1995): 482–484.
11. L. Jami, “Golgi Tendon Organs in Mammalian Skeletal Muscle: Functional Properties
on Central Actions,” Physiological Reviews, 72 (1992): 623–666.
12. J.-J. Slotine and W. Li, Applied Nonlinear Control (Englewood Cliffs, N.J.: Prentice Hall,
1991).
13. S. Hanneton, A. Berthoz, J. Droulez, and J.-J. Slotine, “Does the Brain Use Sliding Vari-
ables for the Control of Movements?” Biological Cybernetics, 77 (1997): 381–393.
NOTES TO PAGES 24–31 •

271
14. J. Decéty, M. Jeannerod, and C. Prablanc, “The Timing of Mentally Represented Ac-
tions,” Behavioral and Brain Research, 34 (1989): 35–42.
15. J. Decéty, M. Jeannerod, M. Germain, and J. Pastène, “Vegetative Response during
Imagined Movement Is Proportional to Mental Effort,” Behavioral and Brain Research,
24 (1991): 1–5.
16. Scarpa worked at the University of Pavia.
17. P. Flourens, Recherches expérimentales sur les propriétés et les fonctions du système nerveux
dans les animaux vertébrés (Paris: Crèvot, 1824).
18. M. G. Jones and K. E. Spells, “A Theoretical and Comparative Study of the Functional
Dependence of the Semi-circular Canal upon Its Physical Dimensions,” Proceedings of
the Royal Society of London. Series B: Biological Sciences, 157 (1963): 403–419.
19. A. Sans and E. Scarfone, “Afferent Calyces and Type I Hair Cells during Development.
A New Morphological Hypothesis,” Annals of the New York Academy of Sciences, 781
(1996): 1–12.
20. The frontal plane is the vertical plane that passes through the two ears; the sagittal
plane is the vertical plane that passes through the nose and the back of the skull.
21. H. Poincaré, The Value of Science, trans. George Bruce Halsted (New York: Science
Press, 1907), p. 73.
22. Ibid.
23. Ibid., p. 74.
24. H. Poincaré, Science and Hypothesis (New York: Dover, 1952).
25. Ibid., p. 50.
26. Ibid., p. 51–52.
27. Ibid., p. 57.
28. Poincaré, The Value of Science, p. 47.
29. Poincaré, Science and Hypothesis, p. 58.
30. Ibid., pp. 58–59.
31. A. De Kleijn and T. Magnus, “Ueber die Funktion der Otolithen,” Pflugers Archiv, 186
(1921): 6–81.
32. G. G. J. Rademaker, Réactions labyrinthiques et équilibre (Paris: Masson, 1935).
33. R. Lorente de Nó, “Vestibulo-ocular Reflex Arc,” Archives of Neurology and Psychiatry, 30
(1933): 245–291.
34. J. Szentagothai, “The Elementary Vestibulo-ocular Reflex Arc,” Journal of Neurophysi-
ology, 13 (1950): 395–407.
35. A. Berthoz, R. Baker, and W. Precht, “Labyrinthine Control of Inferior Oblique Moto-
neurons,” Experimental Brain Research, 18 (1973): 225–241; R. Baker and A. Berthoz,
“Organization of Vestibular Nystagmus in the Oblique Oculomotor System,” Journal
of Neurophysiology, 37 (1974): 195–217, and Control of Gaze by Brain Stem Neurons (Am-
sterdam: Elsevier, 1977).
36. R. Baker and A. Berthoz, “Is the Prepositus Hypoglossi Nucleus the Source of Another
Vestibular Ocular Pathway?” Brain Research, 86 (1975): 121–127; R. Baker, M. Gresty,
and A. Berthoz, “Neuronal Activity in the Prepositus Hypoglossi Nucleus Correlated
with Vertical and Horizontal Eye Movements in the Cat,” Brain Research, 101 (1976):
366–371.
272 •
37. A. J. Pellionisz and R. Llinás, “Tensorial Approach to the Geometry of Brain Func-
tion. Cerebellar Coordination via a Metric Tensor,” Neuroscience, 5 (1980): 1761–
1770.MM
38. B. W. Peterson, J. Baker, and A. J. Pellionisz, “Comparison of Spatial Transformation in
Vestibulo-ocular and Vestibulo-spinal Reflexes,” Proceedings of the Symposium on the Rep-
resentation of Three-Dimensional Space in the Vestibular, Oculomotor, and Visual Systems
(The Barany Society: Bologna, 1988). See also the chapter by Pellionisz in A. Berthoz
and G. Melvill-Jones, Adaptative Mechanisms in Gaze Control (Amsterdam: Elsevier,
1985).
39. A. J. Pellionisz and R. Llinás, “Brain Modeling by Tensor Network Theory and Com-
puter Simulation. The Cerebellum: Distributed Processor for Predictive Coordination,”
Neuroscience, 4 (1979): 232–348, “Tensorial Approach,” and “Space-Time Representation
in the Brain. The Cerebellum as a Predictive Space-Time Metric Tensor,” Neuroscience,
7 (1982): 2949–2970.
40. A. J. Pellionisz, “Coordination: A Vector-Matrix Description of Transformations of
Overcomplete CNS Coordinates and a Tensorial Solution using the Moore-Penrose
Generalized Inverse,” Journal of Theoretical Biology, 101 (1984): 353–375.
41. M. Arbib and S. Amari, “Sensorimotor Transformations in the Brain (With a Critique
of the Tensor Theory of Cerebellum),” Journal of Theoretical Biology, 112 (1985): 123–
155.
42. J. L. Barbur, A. J. Harlow, and L. Weiskrantz, “Spatial and Temporal Response Prop-
erties of Residual Vision in a Case of Hemianopia,” Philosophical Transactions of the
Royal Society of London. Series B: Biological Sciences, 343 (1994): 366–371.
43. E. Mach, Grundlinien der Lehre von den Bewegungsempfindungen (Amsterdam: E. J. Bonset,
1967).
44. D. N. Lee and E. Aronson, “Visual Proprioceptive Control of Standing in Human In-
fants,” Perception and Psychophysics, 15 (1974): 529–532; D. N. Lee and J. R. Lishman,
“Visual Proprioceptive Control of Stance,” Journal of Human Movement Studies, 1 (1975):
87–95.
45. J. Dichgans and T. Brandt, “Visual-Vestibular Interaction and Motion and Perception,”
in J. Dichgans and E. Bizzi, eds., Cerebral Control of Eye Movements and Motion Perception
(New York: Karger, 1972), pp. 327–338, and “Visual-Vestibular Interactions: Effects on
Self-Motion Perception and Postural Control,” in H. Heibowitz and H. L. Teuber, eds.,
Handbook of Sensory Physiology (Berlin: Springer, 1978), vol. 5, pp. 755–804; T. Brandt, J.
Dichgans, and E. Koenig, “Differential Effects of Central versus Peripheral Vision on
Egocentric and Exocentric Motion Perception,” Experimental Brain Research, 16 (1973):
746–491.
46. A. Berthoz, B. Pavard, and L. Young, “Perception of Linear Horizontal Self-Motion In-
duced by Peripheral Vision (Linear Vection),” Experimental Brain Research, 23 (1974):
471–489; B. Pavard and A. Berthoz, “Perception du mouvement et orientation spatial
(Revue bibliographique),” Le Travail humain, 2 (1976): 207–226, and “Linear Accelera-
tion Modifies the Perceived Velocity of a Moving Scene,” Perception and Psychophysics, 6
(1977): 529–540.

273
3 Building Coherence
1. A. R. Damasio, “The Brain Binds Entities and Events by Multiregional Activation from
Convergence Zones,” Neural Computation, 1 (1989): 123–132.
2. Aristotle, Psychologie. Opuscules (Paris: Dumont, 1847), p. 17.
3. D. Ferrier, The Functions of the Brain (New York: G. P. Putnam’s Sons, 1876), p. 60.
4. The interactions between visual and auditory perception are described in B. E. Stein
and M. A. Meredith, The Merging of the Senses (Cambridge, Mass.: MIT Press, 1993), and
are currently the subject of renewed interest. But they have been little studied in hu-
mans.
5. I. M. Sechenov, Selected Works (Leningrad: State Publishing House for Biological and
Medical Literature, 1935), p. 305.
6. R. Held and A. Hein, “Movement-Produced Stimulation in the Development of Vi-
sually Guided Behavior,” Journal of Comparative Physiological Psychology, 56 (1963): 872–
876.
7. J. J. Gibson, The Senses Considered as Perceptual Systems (Boston: Houghton Mifflin,
1966).
8. M. T. Turvey and P. N. Kugler, “An Ecological Approach to Perception and Action,” in
Human Motor Actions. Bernstein Reassessed (Amsterdam: Elsevier, 1984).
9. Gibson, Senses Considered, and “The Theory of Affordances,” in R. E. Shaw and J.
Bransford, eds., Perceiving, Acting, and Knowing (Hillsdale, N.J.: Erlbaum, 1977).
10. R. E. Wurtz, C. Duffy, and J. P. Roy, “Motion Processing for Guiding Self-Motion,” in T.
Ono, L. R. Squire, M. Raichle, D. I. Perrett, and M. Fukuda, eds., Brain Mechanisms of
Perception and Memory: From Neuron to Behavior (Oxford: Oxford University Press, 1993),
pp. 141–182; C. Salzman and W. T. Newsome, “Neural Mechanisms for Forming a Per-
ceptual Decision,” Science, 265 (1994): 231–237; D. C. Bradley, M. Maxwell, R. A.
Andersen, M. S. Banks, and K. V. Shenoy, “Mechanisms of Heading Perception in Pri-
mate Visual Cortex,” Science, 273 (1996): 1544–1547.
11. M. N. Shadlen and W. T. Newsome, “Motion Perception: Seeing and Deciding,” Pro-
12. G. A. Orban, P. Dupont, B. De Bruyn, R. Vogels, R. Vanderberghe, and L. Mortelmans,
“A Motion Area in Human Visual Cortex,” Proceedings of the National Academy of Sci-
ences, 92 (1995): 993–997.
13. Compare descriptions in O. J. Grüsser and T. Landis, Visual Agnosias and Related Disor-
ders, vol. 12, Vision and Visual Dysfunction (Basingstoke, UK: Macmillan, 1991).
14. J. Simpson, “The Accessory Optic Systems,” Annual Review of Neuroscience, 7 (1984):
13–14; K. P. Hoffman, “Responses of Single Neurons in the Pretectum of Monkeys to
Visual Stimuli in Three-Dimensional Space,” in B. Cohen and V. Henn, eds., Representa-
tion of Three-Dimensional Space in the Vestibular, Oculomotor, and Visual Systems (New
York: New York Academy of Sciences, 1988), pp. 1–261.
15. J. I. Simpson and W. Graf, “The Selection of Reference Frames by Nature and Its Inves-
tigators,” in Berthoz and Melvill-Jones, Adaptative Mechanisms in Gaze Control, pp. 3–20.
16. J. Dichgans, C. L. Schmidt, and W. Graf, “Visual Input Improves the Speedometer
Function of the Vestibular Nuclei in the Goldfish,” Experimental Brain Research, 18
(1973): 319–322; V. Henn, L. R. Young, and C. Finley, “Vestibular Nucleus Units in
274 •
Alert Monkeys Are Also Influenced by Missing Visual Field,” Brain Research, 71 (1974):
144–149.
17. Baker and Berthoz, “Organization of Vestibular Nystagmus”; A. Berthoz, K. Yoshida,
and P. P. Vidal, “Horizontal Eye Movement Signals in Second-Order Vestibular Nuclei
Neurons in the Alert Cat,” Annals of the New York Academy of Sciences, 374 (1981): 144–
156; A. Berthoz, J. Droulez, P. P. Vidal, and K. Yoshida, “Neural Correlates of Horizon-
tal Vestibulo-ocular Reflex Cancellation during Rapid Eye Movements in the Cat,” Jour-
nal of Physiology, 419 (1989): 717–751.
18. M. Magnin, M. Jeannerod, and P. T. S. Putkonen, “Vestibular and Saccadic Influences
on Dorsal and Ventral Nuclei of the Lateral Geniculate Body,” Experimental Brain Re-
search, 21 (1974): 1–18.
19. K. Yoshida, R. A. McCrea, A. Berthoz, and P. P. Vidal, “Interneurones Inhibiteur de la
saccade oculaire horizontale étudiée chez le chat éveillé à l’aide d’injections intra-
axonique de peroxydase,” Comptes rendus hebdomadaires de l’Académie des sciences (Paris,
série III), 290 (1980): 636–638, “Eye Movement Related Activity of Identified Second-
Order Vestibular Neurons in the Cat,” in A. Fuchs and W. Becker, eds., Progress in
Oculomotor Research (Amsterdam: Elsevier, 1981), pp. 371–378, and “Properties of Im-
mediate Premotor Inhibitory Burst Neurons Controlling Horizontal Rapid Eye Move-
ments in the Cat,” in Fuchs and Becker, Progress in Oculomotor Research, pp. 71–81; R. A.
McCrea, K. Yoshida, C. Evinger, and A. Berthoz, “The Location, Axonal Arboriza-
tion, and Termination Sites of Eye-Movement-Related Secondary Vestibular Neurons
Demonstrated by Intra-axonal HRP Injection in the Alert Cat,” in Fuchs and Becker,
Progress in Oculomotor Research, pp. 379–386; Berthoz et al., “Neural Correlates,”
pp. 717–751.
20. E. von Holst and H. Mittelstaedt, “Das Reafferenzprinzip. Wechselwirkungen zwischen
Zentralnervensystem und Peripherie,” Naturwissenschaften, 37 (1950): 464–476.
21. W. Penfield and E. Boldrey, “Somatic Motor and Sensory Representation in the Cere-
bral Cortex of Man as Studied by Electrical Stimulation,” Brain, 60 (1937): 389–443; W.
Penfield and T. Rasmussen, The Cerebral Cortex of Man. A Clinical Study of Localization
of Function (New York: Macmillan, 1957); W. Penfield, “Vestibular Sensation and the
Cerebral Cortex,” Annales d’otorhinolaryngologie, 66 (1957): 691–698.
22. B. H. Smith, “Vestibular Disturbances in Epilepsy,” Neurology, 10 (1960): 465–469.
23. J. Silberpfennig, “Contributions to the Problem of Eye Movements. III. Disturbance
of Ocular Movements with Pseudo-hemianopsia in Frontal Lobe Tumors,” Confinia
Neurologica, 4 (1941): 1–13.
24. L. Friberg, T. S. Olsen, P. Roland, O. B. Paulson, and N. A. Lassen, “Focal Increase of
Blood Flow in the Cerebral Cortex of Man during Vestibular Stimulation,” Brain, 108
(1985): 609–623; P. Tuohimaa, E. Aantaa, K. Toukoniitty, and P. Mäkelä, “Studies of
Vestibular Cortical Areas with Short Living Oxygen 15 Isotopes,” Otorhinolaryngologica,
45 (1983): 315–321; G. Bottini, R. Sterzi, E. Paulesu, G. Vallar, S. Cappa, F. Erminio,
R. E. Passingham, C. D. Frith, and R. S. J. J. Frackowiack, “Identification of the Central
Vestibular Projections in Man: A Positron Emission Tomography Activation Study,” Ex-
perimental Brain Research, 99 (1994): 164–169; E. Lobel, J. F. Kleine, D. Le Bihan, A.
Leroy-Willig, and A. Berthoz, “Functional MRI of Galvanic Vestibular Stimulation,”
Journal of Neurophysiology, 80 (1998): 2699–2709. Caloric stimulation, a technique com-

275
mon in clinical otorhinolaryngology, is used to activate the vestibular receptors during
brain-imaging tests. Hot or cold water is injected into the ear, and this sudden and lo-
cal change in temperature induces an activity in the vestibular receptors as if they had
been stimulated by a rotation of the head. The mechanism of this activity is still not
well known. Barany, a Nobel laureate who devised many vestibular tests, suggested
that the receptors are excited by convection currents induced in the endolymph by a
temperature gradient. But an experiment conducted on the U.S. space shuttle, to
which our laboratory contributed, contradicted this hypothesis. In fact, in space—in
microgravity—convection disappears, and yet the injection of cold air into the ears of
the astronauts produced an ocular movement (nystagmus) characteristic of activation
of the semicircular canals. Our results do not, however, call into question the utility of
the test.
25. T. Brandt and M. Dieterich, “Skew Deviation with Ocular Torsion: A Vestibular Brain-
stem Sign of Topographic Diagnostic Value,” Annals of Neurology, 33 (1993): 528–534;
M. Dieterich and T. Brandt, “Thalamic Infarctions: Differential Effects on Vestibular
Function in the Roll Plane (35 Patients),” Neurology, 43 (1993): 1732–1740.
26. W. O. Guldin and O. J. Grüsser, “Single Unit Responses in the Vestibular Cortex of
Squirrel Monkeys,” Neuroscience Abstracts, 13 (1987): 1224. For a summary, see A.
Berthoz and P. P. Vidal, Noyaux vestibulaires et vertiges (Paris: Arnette, 1993).
27. O. J. Grüsser, “Cortical Representations of Head Movement in Space and Some
Psychophysical Considerations,” in A. Berthoz, P. P. Vidal, and W. Graf, eds., The Head-
Neck Sensory-Motor System (New York: Oxford University Press, 1991), pp. 497–509; O. J.
Grüsser, M. Pause, and U. Schreiter, “Localisation and Responses of Neurones in the
Parieto-insular Vestibular Cortex of Awake Monkeys (Macaca fasciculais),” Journal of
Physiology, 430 (1990): 537–557, and “Vestibular Neurons in the Parieto-Insular Cortex
of Monkeys (Macaca fascicularis): Visual and Neck Receptor Responses,” Journal of Phys-
iology, 430 (1990): 559–583; S. Akbarian, O. J. Grüsser, and W. O. Guldin, “Corticofugal
Projections to the Vestibular Nuclei in Squirrel Monkeys: Further Evidence of Multiple
Cortical Vestibular Fields,” Journal of Comparative Neurology, 332 (1993): 7270–7281.
28. Grüsser and Landis, “Visual Agnosias.” In the monkey, the six cortical areas implicated
in vestibular activity are areas 7a, 6, 3a, 2v, T3, and PIVC. Bottini et al. (“Central Vestib-
ular Projections in Man”) found equivalent areas in humans. In addition, E. Vitte, A.
Sémont, and A. Berthoz [“Repeated Optokinetic Stimulation in Conditions of Active
Standing Facilitates the Recovery from Vestibular Deficits,” Experimental Brain Research,
102 (1994): 141–148] discovered activity in the hippocampus.
29. K. Kawano, M. Sasaki, and M. Yamashita, “Vestibular Input to Visual Tracking Neu-
rons in the Posterior Parietal Association Cortex of the Monkey,” Neuroscience Letters,
17 (1980): 55–60; H. Sakata, H. Shibutani, and K. Kawano, “Functional Properties of
Visual Tracking Neurons in Posterior Parietal Association Cortex of the Monkey,” Jour-
nal of Neurophysiology, 49 (1983): 1364–1380; P. Thier and R. G. Erickson, “Responses of
Visual-Tracking Neurons from Cortical Area MST-I to Visual, Eye, and Head Motion,”
European Journal of Neuroscience, 4 (1992): 539–553. For a review of the cortical influ-
ences of the vestibular system, see A. Berthoz, “How Does the Cerebral Cortex Pro-
cess and Utilize Vestibular Signals?” in R. W. Baloh and G. M. Halmagyi, eds., Disorders
of the Vestibular System (Oxford: Oxford University Press, 1996), pp. 113–125. Demon-
276 •
stration of vestibular input on hippocampal formation was provided by E. Vitte,
C. Derosier, Y. Caritu, A. Berthoz, D. Hasboun, and D. Soulié, “Activation of the
Hippocampal Formation by Vestibular Stimulation: A Functional Magnetic Resonance
Study,” Experimental Brain Research, 112 (1996): 523–526 and V. V. Gavrilov, S. I. Wiener,
and A. Berthoz, “Whole Body Rotations Enhance Hippocampal Theta Rhythm Slow
Activity in Awake Rats Passively Transportated on a Mobile Robot,” Annals of the New
York Academy of Sciences, 781 (1996): 385–398.
30. P. Buisseret and M. Imbert, “Visual Cortical Cells: Their Developmental Properties
in Normal and Dark Reared Kittens,” Journal of Physiology, 255 (1976): 511–525;
P. Buisseret and E. Gary-Bobo, “Development of Visual Cortical Orientation Specificity
after Dark Rearing: Role of Extra-ocular Proprioception,” Neuroscience Letters, 13
(1979): 259–263.
31. X. M. Sauvan and E. Peterhans, “Neural Integration of Visual Information and Direc-
tion of Gravity in the Prestriate Cortex of the Alert Monkey,” in T. Mergner and F.
Hlavackà, eds., Multisensory Control of Posture (New York: Plenum, 1995), pp. 43–50.
32. J. S. Taube, R. U. Muller, and J. B. Ranck, Jr., “Head-Direction Cells Recorded from the
Postsubiculum in Freely Moving Rats. I. Description and Quantitative Analysis,” Jour-
nal of Neuroscience, 10 (1990): 420–435, and “Head Direction Cells Recorded from the
Postsubiculum in Freely Moving Rats. II. Effects of Environmental Manipulations,”
Journal of Neuroscience, 10 (1990): 436–447. L. L. Chen, L. Lin, E. J. Green, C. A. Barnes,
and B. McNaughton, “Head Direction Cells in the Rat Posterior Cortex. I. Anatomical
Distribution and Behavioural Modulation,” Experimental Brain Research, 101 (1994): 8–
23; L. L. Chen, L. Lin, C. A. Barnes, and B. McNaughton, “Head Direction Cells in the
Rat Posterior Cortex. II. Contributions of Visual and Ideothetic Information to the Di-
rectional Firing,” Experimental Brain Research, 101 (1994): 24–34.
33. A. B. Rubens, “Caloric Stimulation and Unilateral Visual Neglect,” Neurology, 35 (1985):
1019–1024; S. Cappa, R. Sterzi, G. Vallar, and E. Bisiach, “Remission of Hemineglect
and Anosognosia during Vestibular Stimulation,” Neuropsychologia, 25 (1987): 775–782;
G. Vallar, R. Sterzi, G. Bottini, S. Cappa, and M. L. Rusconi, “Temporary Remission of
Left Hemianesthesia after Vestibular Stimulation. A Sensory Neglect Phenomenon,”
Cortex, 26 (1990): 123–131.
34. Many theories have been proposed to explain spatial neglect. For reviews see M.
Jeannerod, Neurophysiological and Neuropsychological Aspects of Spatial Neglect (Amster-
dam: Elsevier, 1987), and G. Rizzolatti and V. Galeze, “Mechanisms and Theories of
Spatial Neglect,” in F. Boller and J. Gravman, eds., Handbook of Neuropsychology (Am-
sterdam: Elsevier, 1988), pp. 289–313.
35. G. Vallar, R. Sterzi, G. Bottini, S. Cappa, and M. L. Rusconi, “Temporary Remission of
Left Hemianesthesia after Vestibular Stimulation. A Sensory Neglect Phenomenon,”
Cortex, 26 (1990): 123–131.
36. Ibid.
37. G. Bottini, E. Paulesu, R. Sterzi, E. Warburton, R. J. Wise, G. Vallar, R. S. Frackowiak,
and C. D. Frith, “Modulation of Conscious Experience by Peripheral Sensory Stimuli,”
Nature, 376 (1995): 778–781.
38. Ibid., p. 781.
39. J. Gertsmann, “Problems of Imperception of Disease and of Impaired Body Territories

277
with Organic Lesions. Relation to Body Scheme and Its Disorders,” Archives of Neurol-
ogy and Psychiatry, 48 (1942): 890–913.
40. E. Bisiach, M. L. Rusconi, and G. Vallar, “Remission of Somatoparaphrenic Delusion
through Vestibular Stimulation,” Neuropsychologia, 29 (1991): 1029–1031.
41. H. O. Karnath, K. Christ, and W. Hartje, “Decrease of Contralateral Neglect by Neck
Muscle Vibration and Spatial Orientation of Trunk Midline,” Brain, 116 (1993): 383–
396; Vallar et al., “Improvement of Left Visuo-spatial Hemineglect,” pp. 73–82.
42. K. Lorenz, L’Envers du miroir (Paris: Flammarion, 1975); J. E. Cutting and P. M. Vishton
[“Perceiving Layout and Knowing Distances: The Integration, Relative Potency, and
Contextual Use of Different Information about Depth,” in Perception of Space and Mo-
tion (San Diego: Academic, 1995), pp. 69–117] provide a nice clarification of this ques-
tion.
43. Poincaré, Science and Hypothesis, p. 53.
44. Y. Trotter, S. Celebrini, B. Stricane, S. Thorpe, and M. Imbert, “Neural Processing of
Stereopsis as a Function of Viewing Distance in Primate Visual Cortical Area V1,” Jour-
nal of Neurophysiology, 76 (1996): 2872–2885.
45. Cited in Bouveresse, Langage, perception et réalité, p. 37.
46. Stein and Meredith, Merging of the Senses, p. 47.
47. “Sensoritopy” refers to the spatial organization of neurons in a structure of the central
nervous system that corresponds to the spatial distribution of the receptors at the pe-
riphery.
48. M. Graziano and C. Gross, “Coding of Visual Space by Premotor Neurons,” Science,
266 (1994): 1054–1057, and “The Representation of Extrapersonal Space: A Possible
Role for Bimodal-Tactile Neurons,” in M. Gazzaniga, ed., The Cognitive Neurosciences
(Cambridge, Mass.: MIT Press, 1994): 1021–1034.
49. Stein and Meredith, Merging of the Senses, p. 137.
50. Merleau-Ponty, Le Visible et l’invisible, p. 107.
51. J.-P. Sartre, The Psychology of Imagination (Westport, Conn.: Greenwood Press, 1978),
p. 115.
52. J. De Ajuriaguerra, Résumés de cours. Collège de France (Paris, 1976).
53. P. Bach-y-Rita, “Vision Substitution by Tactile Image Projection,” Nature, 221 (1989):
963–964.
54. F. Uhl, P. Franzen, G. Lindinger, W. Lang, and L. Deecke, “On the Functionality of the
Visually Deprived Occipital Cortex in Early Blind Persons,” Neuropsychologia, 33 (1991):
256–259.
55. Graziano and Gross, “Coding of Visual Space,” 1054–1057, and “The Representation of
Extrapersonal Space,” 1021–1034. Similar neurons have also been found for the mouth
(Rizzolatti, 1987); moreover, J.-R. Duhamel, C. L. Colby, and M. E. Goldberg [“Con-
gruent Representations of Visual and Somatosensory Space in Single Neurons of Mon-
key Ventral Intraparietal Cortex (area VIP),” in Paillard, Brain and Space, pp. 223–236]
proposed that, in the parietal cortex, these neurons participate in a “supramodal analy-
sis of the environment” by establishing a connection between one representation cen-
tered on the body and another centered on the retina (retinotopic).
56. Merleau-Ponty, La Nature.
57. E. Husserl, Idées directrices pour une phénoménologie et une philosophie phénoménologique
278 •
pure, vol. 2, Recherches phénoménologiques pour la constitution (Paris: Presses Universitaires
de France, 1982), p. 207.
58. A. J. Mistlin and D. I. Perrett, “Visual and Somatosensory Processing in the Macaque
Temporal Cortex: The Role of ‘Expectation,’” Experimental Brain Research, 82 (1990):
437–450.
59. J. K. Hietanen and D. I. Perrett, “A Role of Expectation in Visual and Tactile Processing
within the Temporal Cortex,” in Ono et al., Brain Mechanisms, pp. 83–103; quotation
from p. 89.
60. A. Prochazka, “Sensorimotor Gain Control: A Basic Strategy of Motor Systems?” Prog-
ress in Neurobiology, 33 (1989): 281–307; quotation from p. 301.
61. J. MacIntyre, E. Gurfinkel, M. Lipshits, J. Droulez, and V. Gurfinkel, “Measurement of
Human Force Control During Constrained Arm Motion using a Force-Activated Joy-
stick,” Journal of Neurophysiology, 17 (1995): 1201–1222.
62. Proust, Marcel, Remembrance of Things Past, trans. C. K. Scott Moncrieff and Terence
Kilmartin (London: Chatto and Windus, 1981), p. 64.
63. J. Droulez and C. Darlot, “The Geometric and Dynamic Implications of the Coherence
Constraints in Three-Dimensional Sensorimotor Coordinates,” in M. Jeannerod, ed.,
Attention and Performance XIII (Hillsdale, N.J.: Erlbaum, 1989), pp. 495–526.
64. J. Droulez and V. Cornilleau-Pérès, “Application of the Coherence Scheme to the
Multisensory Fusion Problem,” in A. Berthoz, ed., Multisensory Control of Movement
(Oxford: Oxford University Press, 1993), pp. 485–508; quotation from p. 490.
65. M. Abeles, Local Cortical Circuits (Berlin: Springer, 1982); C. von der Malsburg, “Ner-
vous Structures with Dynamical Links,” Berichte der Bunsengesellschaft für physikalische
Chemie, 89 (1985): 703–710; E. Bienenstock and C. von der Malsburg, “Statistical
Coding and Short Term Synaptic Plasticity: A Scheme for Knowledge Representation
in the Brain,” in E. Bienenstock, F. Fogelman, G. Weisbuch, Disordered Systems and
Biological Organization (Berlin: Springer, 1986): 247–272; Gray et al., “Oscillatory Re-
sponses”; W. Singer, “Search for Coherence: A Basic Principle of Cortical Self-Organi-
zation,” Concepts in Neuroscience, 1 (1990): 1–26. For further references, see G. Buzsáki,
Z. Horváth, R. Urioste, J. Hetke, and K. Wise, “High Frequency Network Oscillations
in the Hippocampus,” Science, 256 (1992): 1025–1027.
66. Frith, Autism: Explaining the Enigma, Cognitive Development (Oxford: Basil Blackwell,
1989), pp. 103–104.
67. Ibid., p. 110.
68. Ibid., p. 166.
4 Frames of Reference
1. N. Eshkol and A. Wachmann, Movement Notation (London: Weidenfeld and Nicolson,
1958).
2. Grüsser and Landis, Visual Agnosias and Related Disorders; Cutting and Vishton, “Per-
ceiving Layout and Knowing Distances.”
3. J. Hyvarinen, The Parietal Cortex of Monkey and Man (Berlin: Springer, 1982).
4. V. H. Mountcastle, J. C. Lynch, A. Georgopoulos, H. Sakata, and C. Acuna, “Poste-
rior Parietal Association Cortex of the Monkey: Command Functions for Operations
within Extrapersonal Space,” Journal of Neurophysiology, 38 (1975): 871–908.

279
5. G. Vallar, E. Lobel, G. Galati, A. Berthoz, L. Pizzamiglio, and D. Le Bihan, “A Fronto-
Parietal System for Computing the Egocentric Spatial Frame of Reference in Hu-
mans,” Experimental Brain Research, 124 (1999): 281–286.
6. J. Paillard, “Les déterminants moteurs de l’organisation spatial,” Cahiers de psychologie,
14 (1971): 261–316; “Le corps et ses langages d’espace,” in E. Jeddi, ed., Le Corps en
psychiatrie (Paris: Masson, 1982), pp. 53–69; “Posture and Locomotion: Old Problems
and New Concepts. Foreword,” in B. Amblard, A. Berthoz, and F. Clarac, eds., Posture
and Gait: Development, Adaptation, and Modulation (Amsterdam: Elsevier, 1988), pp. 5–
12.
7. E. Muybridge, The Human Figure in Motion (New York: Dover, 1957).
8. A. Berthoz and T. Pozzo, “Intermittent Head Stabilisation during Postural and
Locomotory Tasks in Humans,” in Amblard et al., Posture and Gait, pp. 189–198; T.
Pozzo, A. Berthoz, and L. Lefort, “Head Stabilisation during Various Locomotor Tasks
in Humans. I. Normal Subjects,” Experimental Brain Research, 82 (1990): 97–106; M. J.
Dai, I. Curthoys, and G. M. Halmagyi, “A Model of Otolith Stimulation,” Biological Cy-
bernetics, 60 (1989): 185–194.
9. K. D. Walton, D. Lieberman, A. Llinás, M. Begin, and R. Llinás, “Identification of a
Critical Period for Motor Development in Neonatal Rats,” Neuroscience, 51 (1992), 763–
767.
10. P. Buisseret and M. Imbert, “Visual Cortical Cells.”
11. H. Aubert, “Über eine scheinebare Drehung von Objecten bei Neigung des Kopfes
nach rechts oder links,” Virchows Archiv, 20 (1967): 381–393.
12. J. Müller, “Über das Aubert’sche Phänomen,” Zeitschrift für Psychologie der Sinnesorgane,
49 (1916): 109–249.
13. G. Clément, V. S. Gurfinkel, F. Lestienne, M. Lipshits, and K. Popov, “Adaptation of
Postural Control to Weightlessness,” Experimental Brain Research, 57 (1984): 61–72; G.
Clément and F. Lestienne, “Adaptive Modifications of Postural Attitude in Conditions
of Weightlessness,” Experimental Brain Research, 72 (1988): 381–389.
14. H. Mittelstaedt, “The Role of Otoliths in the Perception of the Orientation of Self and
World to the Vertical,” Zoologische Jahrbücher der Physiologie, 95 (1991): 419–425; S.
Glasauer and H. Mittelstaedt, “Determinants of Orientation in Microgravity,” Acta
Astronautica, 27 (1992): 1–9; W. Haustein and H. Mittelstaedt, “Evaluation of Retinal
Orientation and Gaze Direction in the Perception of the Vertical,” Vision Research, 30
(1990): 255–262; H. Mittelstaedt and S. Glasauer, “Illusions of Verticality in Weightless-
ness,” Clinical Investigations, 71 (1993): 732–739.
15. H. Mittelstaedt, “Evidence of Somatic Graviception from New and Classical Investiga-
tions,” Acta Otolaryngologica, 115 (suppl. 520) (1995): 186–187, and “Somatic Gravi-
ception,” Biological Psychology, 42 (1996): 53–74.
16. Ferrier, Functions of the Brain, pp. 63–64.
17. V. S. Gurfinkel, “The Mechanisms of Postural Regulation in Man,” in T. Turpaev, ed.,
Physiology and General Biology Reviews (Chur, Switzerland: Harwood Academic Pub-
lishers, 1994).
18. Rademaker, Réactions labyrinthiques et équilibre.
19. L. Young, M. Shelhamer, and S. Modestino, “M.I.T./Canadian Vestibular Experiments
in Weightlessness,” Experimental Brain Research, 64 (1986): 299–307.
280 •
20. See Chapter 2.
21. J. R. Lackner and M. S. Levine, “Changes in Apparent Body Orientation and Sensory
Localization Induced by Vibration of Postural Muscles: Vibratory Myesthetic Illu-
sions,” Aviation, Space, and Environmental Medicine, 50 (1979): 346–354; J. R. Lackner,
“Some Contributions of Touch, Pressure, and Kinesthesis to Human Spatial Orienta-
tion and Oculomotor Control,” Acta Astronautica, 8 (1981): 825–830, “Some Proprio-
ceptive Influences on the Perceptual Representation of Body Shape and Orientation,”
Brain, 111 (1981): 281–297, and “Orientation and Movement in Unusual Force Environ-
ments,” Psychological Science,” 4 (1993): 134–142.
22. J. F. Soechting and B. Ross, “Psychophysical Determination of Coordinate Representa-
tion of Human Arm Orientation,” Neuroscience, 13 (1984): 595–604; J. F. Soechting and
M. Flanders, “Sensorimotor Representations for Pointing to Targets in Three-Dimen-
sional Space,” Journal of Neurophysiology, 62 (1989): 582–594.
23. D. I. Perrett, E. T. Rolls, and W. Caan, “Visual Neurons Responsive to Faces in the
Monkey Temporal Cortex,” Experimental Brain Research, 47 (1982): 342; D. Perrett, A. J.
Mistlin, A. J. Chitty, P. A. J. Smith, D. D. Potter, R. Broennimann, and M. H. Harries,
“Specialised Face Processing and Hemispheric Asymmetry in Man and Monkey: Evi-
dence from Single Unit and Reaction Time Studies,” Behavioural Brain Research, 29
(1988): 245–258.
24. D. I. Perrett, J. K. Hietanen, M. W. Oram, and P. J. Benson, “Organisation and Function
of Cells Responsive to Faces in the Temporal Cortex,” Philosophical Transactions of the
Royal Society. Series B: Biological Sciences, 335 (1992): 23–30.
25. S. M. Kosslyn, C. F. Chabris, C. J. Marsolek, and O. Koenig, “Categorical versus Coordi-
nate Spatial Representations: Computational Analysis and Computer Simulations,”
Journal of Experimental Psychology: Human Perception and Performance, 18 (1992): 562–577.
26. M. Arbib, T. Iberall, and G. Bingham, “Opposition Space as a Structuring Concept for
the Analysis of Skilled Hand Movements,” Experimental Brain Research, 15 (1986): 158–
173.
27. M. A. Arbib, “Interaction of Multiple Representations of Space in the Brain,” in
Paillard, Brain and Space, pp. 380–403; quotation from p. 380.
28. M. Jeannerod, “A Neurophysiological Model for the Directional Coding of Reaching
Movements,” in Paillard, Brain and Space, pp. 49–69; Goodale et al. [M. A. Goodale,
A. D. Milner, L. S. Jakobson, and D. P. Carey, “Neurological Dissociation between Per-
ceiving Objects and Grasping Them,” Nature, 349 (1991): 154–156] showed that patients
with cervical lesions dissociate between perception and scaling of grasp.
29. Arbib, “Multiple Representations of Space,” p. 385.
30. J. Droulez and A. Berthoz, “Servo-Controlled (Conservative) versus Topological (Pro-
jective) Modes of Sensory Motor Control,” in W. Bles and T. Brandt, eds., Disorders of
Posture and Gait (Amsterdam: Elsevier, 1988), pp. 83–97.
31. A. Berthoz, “Reference Frames for the Perception and Control of Movement,” in
Paillard, Brain and Space, pp. 82–111. The ability to switch frames of reference—to
change perspective—is surely one of the distinctive features of the human brain.
32. M. F. Levin and A. G. Feldman, “The Role of Stretch Reflex Threshold Regulation
in Normal and Impaired Motor Control,” Brain Research, 657 (1994): 23–30; A. G.
Feldman and M. F. Levin, “The Origin and Use of Positional Frames of Reference

281
in Motor Control,” Behavioral and Brain Sciences, 18 (1995): 723–744; E. Bizzi, N.
Accornero, W. Chapple, and N. Hogan, “Posture Control and Trajectory Formation
during Arm Movement,” Journal of Neuroscience, 4 (1984): 2738–2744.
33. Feldman and Levin, “Origin and Use of Positional Frames of Reference,” p. 727.
5 A Memory for Predicting

1. See Chapter 1. Ferrier wrote: “In calling up an idea, or when engaged in the attentive
consideration of some idea or ideas, we are in reality throwing into action, but in an in-
hibited or suppressed manner, the movements with which the sensory factors of
ideation are associated in organic cohesion” (Ferrier, Functions of the Brain, p. 285).
2. This line of argument is close to that of Poincaré, cited in Chapter 2.
3. Paillard, Brain and Space, p. 163.
4. C. Darwin, “Origin of Certain Instincts,” Nature, 179 (1887): 417–418; quotation from
p. 418.
5. T. Gladwin, East Is a Big Bird (Cambridge, Mass.: Harvard University Press, 1970).
6. J. S. Barlow, “Inertial Navigation as a Basis for Animal Navigation,” Journal of Theoreti-
cal Biology, 6 (1964): 76–117.
7. J. S. Beritoff, Neural Mechanisms of Higher Vertebrate Behavior (New York: Brown, 1965).
8. S. Miller, M. Potegal, and L. Abraham, “Vestibular Involvement in a Passive Transport
and Return Task,” Physiological Psychology, 11 (1983): 1–10.
9. H. Mittelstaedt and M. L. Mittelstaedt, “Mechanismen der Orientierung ohne
richtende Aussensreise,” Fortschritte der Zoologie, 21 (1973): 46–58, and “Homing by
Path Integration in a Mammal,” Naturwissenschaften, 67 (1992): 566–567.
10. A. S. Etienne, R. Mowrer, and F. Saucy, “Limitations in the Assessment of Path-De-
pendent Integration,” Behaviour, 106 (1988): 81–111; B. L. Matthews, J. H. Ryu, and C.
Bockaneck, “Vestibular Contribution to Spatial Orientation,” Acta Otolaryngologica, 468
(1989): 149–154.
11. Recent experiments conducted in our laboratory show that patients with vestibular le-
sions have trouble finding their way in the dark.
12. Etienne et al., “Limitations in Assessment.” For a recent review of experiments and
projects concerning the study of the effect of gravity on sensory motor systems, see
A. Berthoz and A. Guell, “Space Neuroscience Research,” Brain Research Reviews,
28 (1998): 1–234. This review was written following a planning symposium for the
international space station being built by the United States. (See also Chapter 9,
note 4.)
13. J. Bloomberg, G. Melvill-Jones, B. Segal, S. McFarlane, and J. Soul, “Vestibular Contin-
gent Voluntary Saccades Based on Cognitive Estimates of Remembered Vestibular In-
formation,” Advances in Oto-Rhino-Laryngology, 40 (1988): 71–75.
14. I. Israël, S. Rivaud, P. Pierrot-Deseilligny, and A. Berthoz, “Depayed VOR: An Assess-
ment of Vestibular Memory for Self-Motion,” in J. Reguin and J. Stelmach, eds., Tuto-
rials in Motor Neuroscience (Dordrecht: Kluwer, 1991), pp. 599–607; I. Israël, S. Rivaud,
A. Berthoz, and C. Pierrot-Deseilligny, “Cortical Control of Vestibular Memory-
Guided Saccades,” Annals of the New York Academy of Sciences, 656 (1992): 472–484.
15. T. Metcalfe and M. Gresty, “Self-controlled Reorienting Movements in Response to Ro-
282 •
tational Displacements in Normal Subjects and Patients with Labyrinthine Diseases,”
in D. L. Tomko, B. Cohen, and F. E. Guedry, eds., Sensing Motion (New York: Annals of
the New York Academy of Sciences, 1992).
16. I. Israël, “Memory-Guided Saccades: What Is Memorized?” Experimental Brain Research,
90 (1992): 221–224; Israël et al., “Cortical Control,” pp. 472–484; C. Pierrot-Deseilligny,
I. Israël, A. Berthoz, S. Rivaud, and B. Gaymard, “Role of the Different Frontal Lobe
Areas in the Control of the Horizontal Component of Memory-Guided Saccades in
Man,” Experimental Brain Research, 95 (1993): 166–171.
17. A. Berthoz, A. Grantyn, and J. Droulez, “Some Collicular Neurons Code Saccadic Eye
Velocity,” Neuroscience Letters, 72 (1987): 289–294; I. Israël and A. Berthoz, “Contribu-
tion of the Otoliths to the Calculation of Linear Displacement,” Journal of Neuro-
physiology, 62 (1989): 247–263; I. Israël, N. Chapuis, S. Glasauer, O. Charade, and A.
Berthoz, “Estimation of Passive Linear Whole Body Displacement in Humans,” Journal
of Neurophysiology, 70 (1993): 1270–1273.
18. A. Berthoz, I. Israël, P. Georges-François, R. Grasso, and T. Tsuzuku, “Spatial Memory
of Body Linear Displacement: What Is Being Stored?” Science, 269 (1995): 95–98. I.
Israël, R. Grasso, P. Georges-François, T. Tzuzuku, and A. Berthoz, “Spatial Memory
and Path Integration Studied by Self-Driven Linear Displacement,” Journal of Neuro-
physiology, 77 (1999): 3180–3192. We have recently found [Y. P. Ivanenko, R. Grasso, I.
Israël, and A. Berthoz, “The Contribution of Otoliths and Semicircular Canals to the
Perception of Two-Dimensional Passive Whole-Body Motion in Humans,” Journal of
Physiology, 502 (1997): 223–233], using a combination of translations and rotations
(semicircular path) during passive transport in robots, that in humans there may be a
dissociation between the processing of information relative to translations and to rota-
tions, as found in the rodents (see note 12 above).
19. Poincaré, The Value of Science, p. 74.
20. J. Droulez and A. Berthoz, “The Dynamic Memory Model and the Final Ooculomotor
and Cephalomotor Integrator,” in H. Shimazu and Y. Shinoda, eds., The Oculomotor Sys-
tem (Tokyo: Japan Scientific Societies Press, 1990), pp. 1–19; A. Berthoz, “Hippocampal
and Parietal Contribution to Topokinetic and Topographic Memory,” in N. Burgess,
K. J. Jeffery, and J. O’Keefe, eds., The Hippocampal and Parietal Foundations of Spatial Cog-
nition (Oxford: Oxford University Press, 1999), pp. 381–399.
21. M. F. Land and D. N. Lee, “Where We Look When We Steer,” Nature, 369 (1994): 742–
744. It is, in fact, easy to show that the curvature of the road is connected to direction t
of the point tangent to the trajectory of the car using the following expression: C = 1/
([d cos t] − 1/d) where the curvature (C) is the inverse of the radius of the bend of the
road and (d) is the lateral distance of the driver in relation to the bend. We recently de-
scribed a similar gaze anticipation during locomotion [R. Grasso, S. Slasauer, U. Takei,
and A. Berthoz, “The Predictive Brain: Anticipatory Control of Head Direction for the
Steering of Locomotion,” NeuroReport, 7 (1996): 1170–1174].
22. B. Milner, “Visual Recognition and Recall after Right Temporal Lobe Excision in Man,”
Neuropsychologia, 6 (1968): 191–209.
23. The hippocampus is a structure that is currently the subject of much interest owing
to its role in memory and especially the acquisition of memories based on events.

283
Among the many recent reviews on its very complex role, see especially T. Ono,
B. L. McNaughton, S. Molotchnidoff, E. T. Rolls, and H. Nishijo, Perception, Memory,
and Emotion: Frontiers in Neuroscience (Cambridge: Pergamon, 1996); for a review on
relations of the hippocampus with the prefrontal and parietal cortex, see L. G.
Ungerleider, “Functional Brain Imaging Studies of Cortical Mechanisms for Memory,”
Science, 270 (1995): 769–775.
24. D. Marr, “Simple Memory: A Theory for Archicortex,” Proceedings of the Royal Society
of London. B: Biological Sciences, 262 (1971): 23–81.
25. M. Mishkin, “A Memory System in the Monkey,” Philosophical Transactions of the Royal
Society. Series B: Biological Sciences, 98 (1982): 85–95.
26. J. O’Keefe and L. Nadel, The Hippocampus as a Cognitive Map (Oxford: Oxford University
Press, 1978).
27. R. Tamura, T. Ono, M. Fukuda, and K. Nakamura, “Recognition of Egocentric and
Allocentric Visual and Auditory Space by Neurons in the Hippocampus of Mon-
keys,” Neuroscience Letters, 109 (1990): 293–298, and “Spatial Responsiveness of Monkey
Hippocampal Neurons to Various Visual and Auditory Stimuli,” Hippocampus, 2 (1992):
307–322; E. T. Rolls and S. M. O’Mara, “Neurophysiological and Theoretical Analysis
of How the Hippocampus Functions in Memory,” in Ono, Brain Mechanisms of Percep-
tion.
28. H. Eichenbaum, S. I. Wiener, M. L. Shapiro, and N. J. Cohen, “The Organization of
Spatial Coding in the Hippocampus: A Study of Neural Ensemble Activity,” Journal of
Neuroscience, 9 (1989): 2764–2775; S. I. Wiener, C. A. Paul, and H. Eichenbaum, “Spatial
and Behavioral Correlates of Hippocampal Neuronal Activity,” Journal of Neuroscience,
9 (1989): 2737–2763.
29. S. M. O’Mara, E. Rolls, A. Berthoz, and R. P. Kesner, “Neurons Responding to Whole
Body Motion in the Primate Hippocampus,” Journal of Neuroscience, 14 (1994): 6511–
6523; S. I. Wiener, V. A. Korshunov, R. Garcia, and A. Berthoz, “Inertial, Substratal, and
Landmark Cue Control of Hippocampal CA1 Place Cell Activity,” European Journal of
Neuroscience, 7 (1995): 2206–2219. For a recent review of the the various models con-
cerning the neural basis of spatial processing during navigation, see O. Trullier, S.
Wiener, A. Berthoz, and J. A. Meyer, “Biologically Based Artificial Navigation Systems:
Reviews and Prospects,” Progress in Neurobiology, 51 (1997): 483–544.
30. A. Treves and E. T. Rolls, “What Determines the Capacity of Autoassociative Mem-
ories in the Brain?” Network, 2 (1991): 371–397, and “Computational Constraints Sug-
gest the Need for Two Distinct Input Systems to the Hippocampal CA3 Network,”
Hippocampus, 2 (1992): 189–199; E. T. Rolls and S. M. O’Mara, “Neurophysiological and
Theoretical Analysis of How the Hippocampus Functions in Memory,” in Ono et al.,
Brain Mechanisms of Perception, pp. 276–297. Rolls and Treves [E. Rolls and A. Treves,
Neural Networks and Brain Function (Oxford: Oxford University Press, 1998)] have re-
cently reviewed the computational hypothesis concerning the hippocampus in the gen-
eral frame of brain networks.
31. Y. Myashita, “Where Visual Perception Meets Memory,” Annual Review of Neuroscience,
16 (1993): 245–263; K. Tanaka, “Inferotemporal Cortex and Object Vision,” Annual Re-
view of Neuroscience, 19 (1996): 109–139.
284 •
32. R. N. Shepard, “Ecological Constraints on Internal Representation: Resonant Kinemat-
ics of Perceiving, Imagining, Thinking, and Dreaming,” Psychological Review, 91 (1984):
417–447; quotation from p. 433.
33. J. O’Keefe and M. Recce, “Phase Relationship between Hippocampal Place Units and
the EEG Theta Rhythm,” Hippocampus, 3 (1993): 317–330.
34. J. E. Lisman and M. A. P. Idiart, “Storage of 7 ± 2 Short-Term Memories in Oscillatory
Subcycles,” Science, 267 (1995): 1512–1515. High-frequency (200 Hz) oscillations in the
hippocampus have also been demonstrated by G. Buzsáki et al., “High Frequency Net-
work Oscillations.”
35. A. Michotte, Causalité, permanence et réalité phénoménales (Paris: Béatrice-Nauwelaerts,
1962).
36. The Essential Writings of Merleau-Ponty, ed. Alden L. Fisher (New York: Harcourt, Brace,
1969), pp. 257–258.
6 Natural Movement
1. The term “rule” is used here instead of “law” to stress the fact that it is not a question
only of the laws of mechanics but also of constraints connected to intrinsic processes
of the nervous system.
2. Ariane Mnouchkine maintains that Oriental theater is the only theater (from a play-
bill).
3. Poems of Pierre de Ronsard, trans. and ed. Nicholas Kilmer (Berkeley: University of Cali-
fornia Press, 1979), p. 201.
4. Slotine and Li, Applied Nonlinear Control.
5. C. Darwin, The Expression of the Emotions in Man and Animals (New York: Philosophical
Library, 1955).
6. E. J. Marey, Le Mouvement (Paris: Masson, 1894); La Machine animale (Paris: Revue EPS,
1993).
7. P. Viviani and C. Terzuolo, “Trajectory Determines Movement Dynamics,” Neurosci-
ence, 72 (1982): 431–437.
8. N. A. Bernstein, “Some Emergent Problems of the Regulation of Motor Acts,” Ques-
tions of Psychology, no. 6 (1957), reprinted in Human Motor Actions: Bernstein Reassessed,
ed. H. T. A. Whiting, Advances in Psychology (Amsterdam: North-Holland, 1984), vol.
17, pp. 354–355.
9. C. De Waele, W. Graf, A. Berthoz, and F. Clarac, “Vestibular Control of Skeletal Ge-
ometry,” in A. Berthoz and F. Clarac, eds., Posture and Gait (Amsterdam: Elsevier, 1988),
pp. 423–432; Berthoz et al., Head-Neck Sensory-Motor System.
10. F. Lacquaniti, J. Soechting, and C. Terzuolo, “Path Constraints on Point to Point Arm
Movements in Three-Dimensional Space,” Neuroscience, 17 (1986): 313–324.
11. Soechting and Ross, “Psychophysical Determination of Coordinate Representation”;
Soechting and Flanders, “Sensorimotor Representations.”
12. C. C. A. M. Gielen and E. J. Van Zuylen, “Coordination of Arm Muscles during
Flexion and Supination: Applications of the Tensor Analysis Approach,” Neuroscience,
17 (1986): 527–539.

285
13. P. Viviani and T. Flash, “Minimum-Jerk, Two-Thirds Power Law, and Isochrony. Con-
verging Approaches to Movement Planning,” Journal of Experimental Psychology (Human
Perception), 21 (1995): 32–53; quotation from p. 34.
14. P. Viviani and R. Schneider, “A Developmental Study of the Relation between Geome-
try and Kinematics of Drawing Movements,” Journal of Experimental Psychology (Human
Perception), 17 (1991): 198–218.
15. Viviani and Flash, “Minimum Jerk,” pp. 32–53.
16. Ibid., p. 35.
17. Listing’s law is explained in several chapters in A. Berthoz, ed., Multisensory Control of
Movement (Oxford: Oxford University Press, 1993).
18. F. C. Donders, “Beitrag zur Lehre von den Bewegungen des menschlichen Auges,” in
Hollandischen Beitragen zu den Anatomischen und Physiologischen Wissenschaften (Amster-
dam, 1847), vol. 1, pp. 104–145.
19. P. Viviani, G. Baud-Bovy, and M. Redolfi, “Perceiving and Tracking Kinaesthetic Stim-
uli: Further Evidence of Motor-Perceptual Interactions,” Journal of Experimental Psy-
chology, 23 (1997): 1232–1252.
20. S. Yasui and L. R. Young, “Perceived Visual Motion as Effective Stimulus to Pursuit Eye
Movement System,” Science, 190 (1975): 906–908.
21. Hence the following formula, taken from Viviani and Flash, “Minimum Jerk,” p. 35:
t2
CF = 1 2 ∫ [(d3 x / dt 3 )2 + (d3 y / dt 3 )2 ] dt
t1
According to this expression, movement would be maximally smooth when the value
of the function is at a minimum. Calculus shows that only one pair of equations pro-
duces this minimum value for a given set of boundary conditions; that is, the condi-
tions that define the beginning and the end of movement. The trajectory, specified by
the horizontal and vertical components of a movement, is therefore expressed as a
fifth-order polynomial function of time. Such functions determine the value of various
parameters all along the trajectory, and the model provides a way to define the tempo-
ral structure of movement. This is how Edelman and Flash demonstrated in 1987 that
a simulation of natural motor activity—for example, writing—could be obtained using
this principle of minimum jerk.
7 Synergies and Strategies

1. V. E. Belen’kii, V. S. Gurfinkel, and Y. I. Pal’tsev, “On Elements of Voluntary Movement
Control,” Biofizika, 12 (1967): 135–141; J. Paillard, “L’intégration sensori-motrice et
idéo-motrice,” in M. Richelle, J. Requin, and M. Robert, eds., Traité de psychologie
expérimentale (Paris: Presses Universitaires de France, 1994), pp. 925–961.
2. We need only retain a few very brief ideas from this field of study.
3. A. Berthoz, J. Droulez, P. P. Vidal, and K. Yoshida. “Neural Correlates of Horizontal
Vestibulo-ocular Reflex Cancellation during Rapid Eye Movements in the Cat,” Journal
of Physiology 419 (1989): 717–751.
4. Y. Shinoda, T. Ohgaki, T. Futami, and Y. Sugiushi, “Vestibular Projections to the Spinal
Cord: The Morphology of Single Vestibulo-Spinal Axons,” in O. Pompeiano and J.
286 •
Allum, eds., Vestibulospinal Control of Posture and Movement, Progress in Brain Research
(Amsterdam: Elsevier, 1988), pp. 17–27.
5. E. von Holst, “Relations between the Central Nervous System and the Peripheral Or-
gans,” Journal of Animal Behaviour, 2 (1954): 89–94; Von Holst and Mittelstaedt, “Das
Reafferenzprinzip.”
6. W. Graf and R. Baker, “Adaptive Changes in the Vestibulo-ocular Reflex of the Flatfish
Are Achieved by Reorganization of Central Nervous Pathways,” Science, 221 (1983):
777–779.
7. E. V. Evarts, “Relation of Pyramidal Tract Activity to Force Exerted during Voluntary
Movements,” Journal of Neurophysiology, 31 (1968): 14–27, and “Role of Motor Cortex in
Voluntary Movements in Primates,” in V. B. Brooks, ed., Handbook of Physiology, sect. 1,
vol. 2, Motor Control (Bethesda, Md.: American Physiological Society, 1981), pp. 1083–
1120.
8. A. P. Georgopoulos, J. F. Kalaska, R. Caminiti, and J. T. Massey, “On the Relations be-
tween the Direction of Two-Dimensional Arm Movements and Cell Discharge in Pri-
mate Motor Cortex,” Journal of Neuroscience, 2 (1982): 1527–1537; A. P. Georgopoulos,
A. B. Schwartz, and R. E. Kettner, “Neuronal Population Coding of Movement Direc-
tion,” Science, 233 (1986): 1416–1429; J. F. Kalaska, R. Caminiti, and A. P. Georgopoulos,
“Cortical Mechanisms Related to the Direction of Two-Dimensional Arm Movements:
Relations in Parietal Area 5 and Comparison with Motor Cortex,” Experimental Brain
Research, 51 (1983): 247–260; A. P. Georgopoulos, “Current Issues in Directional Motor
Control,” Trends in Neurosciences, 18 (1995): 506–510.
9. A. P. Georgopoulos, M. D. Crutcher, and A. B. Schwartz, “Cognitive Spatial Motor Pro-
cesses. III. Motor Cortical Prediction of Movement Direction during an Instructed De-
lay Period,” Experimental Brain Research, 75 (1989): 183–194; A. P. Georgopoulos, J. T.
Lurito, M. Petrides, A. B. Schwartz, and J. T. Massey, “Mental Rotation of the Neuronal
Population Vector,” Science, 243 (1989): 234–236.
10. L. Rispal-Padel, F. Cicirata, and C. Pons, “Cerebellar Nuclear Topography of Simple
and Synergistic Movements in the Alert Baboon (Papio papio),” Experimental Brain Re-
search, 47 (1982): 365–380.
8 Capture
1. N. A. Bernstein, The Coordination and Regulation of Movement (New York: Pergamon
Press, 1967).
2. J. P. Ewert, “Neural Mechanisms of Prey-Catching and Avoidance Behavior in the Toad
(Bufo bufo L.),” Brain Behaviour and Evolution, 3 (1970): 36–56, and “Neuroethology
of Releasing Mechanisms: Prey-Catching in Toads,” Behavioral and Brain Sciences, 10
(1987): 337–405.
3. Schmidt, “A Schema Theory,” Psychological Review, 32 (1975): 225–260.
4. D. N. Lee, “A Theory of Visual Control of Braking Based on Information about Time-
to-Collision,” Perception, 5 (1976): 437–459.
5. D. N. Lee and P. E. Reddish, “Plummeting Gannets: A Paradigm of Ecological Optics,”
Nature, 293 (1985): 293–294; D. N. Lee and D. S. Young, “Visual Timing in Interceptive
Actions,” in D. J. Ingle, M. Jeannerod, and D. N. Lee, eds., Brain Mechanisms and Spatial
Vision (Dordrecht: Martinus Nijhoff, 1996), pp. 1–30.

287
6. For reviews, see R. J. Bootsma and C. E. Peper, “Predictive Visual Information Sources
for the Regulation of Action with Special Emphasis on Catching and Hitting,” in L.
Proteau and D. Elliot, eds., Vision and Motor Control (Amsterdam: Elsevier, 1992),
pp. 285–314; L. Peper, R. J. Bootsma, D. R. Mestre, and F. C. Bakker, “Catching Balls:
How to Get the Hand to the Right Place at the Right Time,” Journal of Experimental
Psychology (Human Perception and Performance), 20 (1994): 591–612; Lee and Young, “Vi-
sual Timing,” pp. 1–36.
7. Peper et al., “Catching Balls,” p. 610.
8. Lee calculated two values: first, the value for the trajectory itself (the relationship be-
tween tangential velocity and the proximity to the vehicle, or time-to-collision); sec-
ond, the visual angle subtended by the trajectory divided by the rate of change of the
angle. Lee notes a connection between these two variables. In other words, he discov-
ered a new relationship between the geometry and kinematics of movement.
9. J. A. S. Kelso, “Phase Transitions and Critical Behaviour in Human Bimanual Coordina-
tion,” American Journal of Physiology, 14 (1984): R1000–1004; J. A. S. Kelso, J. D. Delcolle,
and G. Schöner, “Action-Perception as a Pattern Formation Process,” in M. Jeannerod,
ed., Attention and Performance (Hillsdale, N.J.: Erlbaum, 1990), pp. 139–169; J. J. Bu-
chanan and J. A. S. Kelso, “Posturally Induced Transitions in Rhythmic Multijoint Limb
Movements,” Experimental Brain Research, 94 (1993): 131–142.
10. G. Schöner, “A Dynamic Theory of Coordination of Discrete Movements,” Biological
Cybernetiks, 63 (1990): 257–270.
11. F. Lacquaniti and C. Maioli, “The Role of Preparation in Tuning Anticipatory and
Reflex Responses during Catching,” Journal of Neuroscience, 9 (1989): 134–148.
12. F. Lacquaniti, M. Carrozzo, and N. Borghese, “The Role of Vision in Tuning Anticipa-
tory Motor Responses of the Limbs,” in Berthoz, Multisensory Control of Movement,
pp. 379–390; quotation from p. 390.
13. Like our sensory receptors, measuring instruments are imperfect. The information
they provide has a high signal-to-noise ratio. To increase precision in the measurement
of a physical variable, one simple solution is to make several measurements and to cal-
culate their mean. The mean is an example of a filter: it eliminates much of the noise
inherent in measurements, provided of course that each measurement is made inde-
pendently and that there is no systematic bias. To calculate this mean, you can wait
until you have all your measurements, then calculate their sum and divide it by the
number of measurements. But this approach has several disadvantages. All previous
measurements have to be memorized—there is no way to estimate the result before
the final calculation—and the entire operation has to begin from scratch with each
new measurement. A better method is one called iterative. This approach estimates an
initial value that is adjusted with subsequent measurements by calculating a weighted
mean of the preceding estimated value and the new measurement. Such an iterative
filter continuously provides the best possible estimate of the value of a physical vari-
able, especially if the noise is Gaussian. Kalman filters extend this idea of iterative
filters to problems of optimal estimates of dynamic systems whose conditional vari-
ables evolve over time according to known laws. Each step of their calculation entails
two phases that rely on two predictions. Based on the preceding estimate of the condi-
288 •
tional variables, a law of evolution is applied to predict a new state without taking into
account the new measurements; this first estimate is used to predict the information
supplied by the receptors with the help of an internal model of their functioning. The
spread between the predicted information and the actual measurements facilitates opti-
mal correction of the first estimate.
14. From Merleau-Ponty’s discussion of von Uexküll’s work in his book La Nature, pp. 220,
227. J. von Uexküll, Umwelt und Innenwelt der Tieren und Menschen (Berlin: Springer,
1934); Streifzüge durch die Umwelten von Tieren und Menschen (Berlin: Springer, 1965).
15. Merleau-Ponty, La Nature.
16. R. N. Shepard, “Ecological Constraints,” p. 422.
17. P. Viviani and N. Stucchi, “Biological Movements Look Uniform: Evidence for Motor-
Perceptual Interactions,” Journal of Experimental Psychology (Human Perception), 18
(1992): 603–623.
18. M. Kawato, K. Furukawa, and R. Suzuki, “A Hierarchical Neural Network Model for
Control and Learning of Voluntary Movements,” Biological Cybernetiks, 57 (1987): 169–
185; H. Gomi and M. Kawato, “Adaptive Feedback Control Models of the Vestibulo-
cerebellum and Spinocerebellum,” Biological Cybernetics, 68 (1992): 105–114.
19. Walton et al., “Identification of a Critical Period.”
20. Kelso, “Phase Transitions”; Kelso et al., “Action-Perception”; Schöner, “A Dynamic
Theory”; Buchanan and Kelso, “Posturally Induced Transitions.”
9 The Look that Investigates the World

1. Lorenz, Behind the Mirror, p. 56.
2. The history of the concepts and theories about gaze and vision are superbly reviewed
in Grüsser and Landis, Visual Agnosias and Related Disorders, from which the events re-
ported here have been borrowed.
3. Von Holst and Mittelstaedt, “Das Reafferenzprinzip.”
4. Grasso et al., “The Predictive Brain.” We have recently obtained evidence that there
may be a dissociation between the storage in memory of distance and direction during
the steering of locomotion. The length of the path and the direction may be controlled
by two distinct mechanisms. This suggestion fits with the finding of “head direction
cells,” which code the static direction of the head in space, as discussed in Chapter 5. If
verified, this dissociation between the representations of distance and direction, which
has also been suggested for the control of arm movements, would be another proof of
the brain’s segregation of kinematic variables. R. Grasso, P. Prévost, Y. Ivanenko, and
A. Berthoz, “Eye-Head Coordination for the Steering of Locomotion in Humans:
An Anticipatory Synergy,” Neuroscience Letters, 253 (1998): 115–118; R. Grasso, C.
Assaiante, P. Prévost, and A. Berthoz, “Development of Anticipatory Orienting Strat-
egies during Locomotor Tasks in Children,” Neuroscience and Biobehavioral Review, 22
(1998): 533–539; A. Berthoz, A. Amorim, S. Glasauer, R. Grasso, Y. Takei, and I. Viaud-
Delmon, “Dissociation between Distance and Direction during Locomotor Naviga-
tion,” in R. G. Golledge, ed., Wayfinding Behaviour (Baltimore: John Hopkins University
Press, 1999), pp. 328–348.

289
5. De Ajuriaguerra, Résumés des cours.
6. Heliodorus of Emesa, The Aethiopica, vol. 5. The Athenian Society’s Publications (Ath-
ens: privately printed for the Athenian Society, 1897), p. 158.
7. P. Bourdieu, Le Sens pratique (Paris: Editions de Minuit, 1980), pp. 118–119.
8. Darwin, Expression of the Emotions.
9. Frith, Autism, p. 4.
10. I. P. Pavlov, Les Réflexes conditionnés (Paris: Presses Universitaires de France, 1927).
11. Sokolov and Vinogradova, Neuronal Mechanisms.
12. Ibid., p. 234.
13. Chatila et al., “A 10 Hz ‘Alpha-Like’ Rhythm”; Bouyer et al., “Anatomical Localization
of Beta Rhythms in Cats.”
10 Visual Exploration
1. This word is the English translation of the name of a machine invented by the scientist
Cosinus in Georges Columb [Christophe pseud.], L’Idée fixe du savant Cosinus (Paris:
Colin, 1939).
2. D. G. MacKay, The Organization of Perception and Action (New York: Springer, 1987). For
more on this topic see Bergson’s observations in Matter and Memory.
3. J. C. Eccles, M. Ito, and J. Szentogothai, The Cerebellum as a Neuronal Machine (Berlin:
Springer, 1967); R. Llinás and C. Sotelo, The Cerebellum Revisited (New York: Springer,
1992).
4. J. Decety and D. H. Ingvar, “Brain Structures Participating in Mental Simulation of
Motor Behaviour: A Neuropsychological Interpretation,” Acta Physiologica Scandinavica,
73 (1990): 13–34.
5. H. Korn and D. S. Faber, “Organisation and Cellular Mechanisms Underlying Chemical
Inhibition in a Vertebrate Neuron,” in J.-P. Changeux, ed., Molecular and Cellular Interac-
tions Underlying Higher Brain Functions, Progress in Brain Research (Amsterdam: North
Holland, 1983); D. S. Faber, W. S. Young, P. Legendre, and H. Korn, “Intrinsic Quantal
Variability Due to Stochastic Properties of Receptor-Transmitter Interactions,” Science,
258 (1992): 1494–1498; R. Miles, K. Toth, A. Gulyas, N. Hajos, and T. F. Freund, “Differ-
ences between Somatic and Dendritic Inhibition in the Hippocampus,” Neuron, 16
(1996): 814–823. These researchers showed that inhibiting junctions have several ways
of acting on their neuronal target: controlling the base potential of these cells, oppos-
ing the effect of excitatory inputs when these are not sufficiently intense (that is, “sig-
nificant”), adjusting the rate at which these cells emit signals, and finally in synchroniz-
ing the activity of clusters of neighboring cells, which makes them more “effective.”
Moreover, inhibitory junctions evidence training and memory properties until recently
unsuspected.
6. J. Piaget, The Origin of Intelligence in the Child (London: Routledge and Kegan Paul,
1953).
7. Von Holst and Mittelstaedt, “Das Reafferenzprinzip”; B. Bridgeman, “A Review of the
Role of Efference Copy in Sensory and Oculomotor Control Systems,” Annals of Bio-
medical Engineering, 23 (1995): 409–422.
8. Israël, “Memory-Guided Saccades.”
9. I. Israël, S. Rivaud, P. Pierrot-Desilligny, and A. Berthoz, “Delayed VOR: An Assess-
290 •
ment of Vestibular Memory for Self Motion,” in J. Requin and J. Stelmach, eds., Tuto-
rials in Motor Neuroscience (The Netherlands: Kluwer, 1991), pp. 599–607; I. Israël, S.
Rivaud, B. Gaymard, A. Berthoz, and C. Pierrot-Deseilligny, “Cortical Control of Ves-
tibular-Guided Saccades in Man,” Brain, 118 (1995): 1169–1183.
10. A. Berthoz, “Neural Basis of Decision in Perception and in the Control of Movement,”
in A. R. Damasio, et al., eds., Neurobiology of Decision-Making (Berlin: Springer, 1996),
pp. 83–100; A. Berthoz and L. Petit, “Les mouvements du regard: une affaire de sac-
cades. Un modèle pour l’étude des circuits de la décision et de l’imagination motrice,”
La Recherche, 289 (1996): 58–65.
11. Berthoz, “Neural Basis of Decision,” pp. 83–100.
12. See A. Berthoz, “Coopération et substitution entre le système saccadique et les réflexes
d’origine vestibulaires: faut-il réviser la notion de réflexe?” Revue Neurologique, 145
(1989): 513–526, for a detailed description of saccadic mechanisms in the brainstem.
See also Berthoz and Petit, “Les mouvements du regard,” pp. 58–65, for a detailed de-
scription of this schematic diagram.
13. I. S. Curthoys, C. H. Markham, and N. Furuya, “Direct Projection of Pause Neurons to
Nystagmus-Related Excitatory Burst Neurons in the Cat Pontine Reticular Formation,”
Experimental Neurology, 83 (1984): 414–422.
14. D. P. Munoz and R. H. Wurtz, “Fixation Cells in Monkey Superior Colliculus. I. Char-
acteristics of Cell Discharge,” Journal of Neurophysiology, 70 (1993): 559–575; M. A.
Segraves, “Effects of Frontal Eye Field Stimulation upon Omnipause and Burst Neu-
rons in the Monkey Paramedian Pontine Reticular Formation,” Society of Neuroscience
Abstracts, 18 (1992): 296.10.
15. Berthoz et al., “Saccadic Eye Velocity.”
16. A. Grantyn and R. Grantyn, “Axonal Patterns and Sites of Termination of Cat Supe-
rior Colliculus Neurons Projecting in the Tecto-bulbo-spinal Tract,” Experimental Brain
Research, 46 (1982): 243–256; A. Grantyn, R. Grantyn, V. Robine, and A. Berthoz,
“Electroanatomy of Tectal Efferent Connections Related to Eye Movements in the
Horizontal Plane,” Experimental Brain Research, 37 (1979): 149–172; R. E. Wurtz and
M. E. Goldberg, “Activity of Superior Colliculus in Behaving Monkey. III. Cells Dis-
charging before Eye Movements,” Journal of Neurophysiology, 35 (1972): 575–586; A.
Grantyn and A. Berthoz, “The Role of the Tecto-reticulo-spinal System in Control of
Head Movement,” in G. W. Peterson and F. Richmond, eds., Control of Head Movement
(Oxford: Oxford University Press, 1987), pp. 224–244; E. Oliver, A. Grantyn, M. Chat,
and A. Berthoz, “The Control of Slow Orienting Eye Movements by Tectoreticulo-
spinal Neurons in the Cat: Behavior, Discharge Patterns, and Underlying Connec-
tions,” Experimental Brain Research, 93 (1993): 435–449; A. B. Moschovakis, A. B.
Karabelas, and S. Highstein, “Structure Function Relationship in the Primate Supe-
rior Colliculus. I. Morphological Classification of Efferent Neurons,” Journal of Neuro-
physiology, 60 (1988): 232–262, and “Structure Function Relationship in the Primate
Superior Colliculus. II. Morphological Identity of Presaccadic Neurons,” Journal of
Neurophysiology, 60 (1988): 263–302.
17. A. B. Karabelas and A. K. Moschovakis, “Nigral Inhibitory Termination on Efferent
Neurons of the Superior Colliculus. An Intracellular Horseradish Peroxidase Study in
the Cat,” Journal of Cognitive Neuroscience, 239 (1985): 309–329; J. M. Deniau and G.

291
Chevalier, “Disinhibition as a Basic Process in the Expression of Striatal Functions. The
Striatonigral Influence on the Thalamocortical Cells of the Ventromedial Thalamic
Nucleus,” Brain Research, 334 (1981): 227–233.
18. C. Umilta, C. Mucignat, L. Riggio, C. Barbieri, and G. Rizzolatti, “Programming Shifts
of Spatial Attention,” European Journal of Cognitive Psychology, 6 (1994): 23–41; G.
Rizzolatti, L. Riggio, and B. M. Sheliga, “Space and Selective Attention,” in C. Umilta
and M. Moscovitch, eds., Attention and Performance XV (Hillsdale, N.J.: Erlbaum, 1994),
pp. 232–265.
19. O. Hikosaka and R. H. Wurtz, “Visual and Oculomotor Functions of Monkey Substan-
tia Nigra Pars Reticulata. III. Memory-Contingent Visual and Saccade Responses,” Jour-
20. O. Hikosaka, M. Sakamoto, and N. Miyashita, “Effects of Caudate Nucleus Stimulation
on Substantia Nigra Cell Activity in Monkey,” Experimental Brain Research, 95 (1993):
457–472.
21. O. Hikosaka, “Role of the Basal Ganglia in Motor Learning: A Hypothesis,” in Ono et
al., Brain Mechanisms, pp. 497–513.
22. J. Droulez and A. Berthoz, “The Concept of Dynamic Memory in Sensorimotor Con-
trol,” in Humphrey and Freund, Motor Control, pp. 137–161.
23. Kalaska et al., “Cortical Mechanisms”; M. Jeannerod, “The Posterior Parietal Cortex as
a Spatial Generator,” in D. I. Ingle, J. Jeannerod, and D. Lee, eds., Brain Mechanisms and
Spatial Vision (Dordrecht: Martinus Hijhoff, 1985), pp. 279–298; Andersen, “Visual and
Oculomotor Functions”; Y. Burnod, An Adaptative Network, the Cerebral Cortex (Paris:
Masson, 1988). For a review, see P. Thier and O. Karnath, Parietal Lobe Contribution to
Orientation in 3D Space (Berlin: Springer, 1997).
24. R. A. Andersen and V. B. Mountcastle, “The Influence of the Angle of Gaze upon the
Excitability of the Light Sensitive Neurons of the Posterior Parietal Cortex,” Journal of
Neuroscience, 3 (1983): 532–548.
25. R. A. Anderson, G. K. K. Essick, and R. M. Siegel, “Neurons of Area 7 Activated by
Both Visual Stimuli and Oculomotor Behavior,” Experimental Brain Research, 67 (1987):
316–322; R. A. Andersen, “Visual and Oculomotor Functions of the Posterior Parietal
Cortex,” Annual Review of Neuroscience, 12 (1989): 377–403.
26. H. Sakata and M. Kusunoki, “Organization of Space Perception: Neural Representa-
tion of Three-Dimensional Space in the Posterior Parietal Cortex,” Current Opinion
in Neurobiology, 2 (1992): 170–174; Andersen, “Visual and Oculomotor Functions,”
pp. 377–403; Andersen et al., “Neurons of Area 7,” pp. 316–322; R. A. Andersen, R. M.
Bracewell, S. Barash, J. W. Gnadt, and L. Fogassi, “Eye Position Effects on Visual, Mem-
ory, and Saccade-Related Activity in Area LIP and 7a of the Macaque,” Journal of Neuro-
science, 10 (1990): 1176–1196.
27. C. L. Colby, J.-R. Duhamel, and M. E. Goldberg, “Oculocentric Spatial Representation
in Parietal Cortex,” Cerebral Cortex, 5 (1995): 470–481; J.-R. Duhamel, F. Bremmer, S.
Ben Hamed, and W. Graf, “Spatial Invariance of Visual Receptive Fields in Parietal
Cortex Neurons,” Nature, 389 (1997): 845–848.
28. L. Petit, C. Orssaud, N. Tzourio, F. Crivello, A. Berthoz, and B. Mazoyer, “Functional
Anatomy of a Prelearned Sequence of Horizontal Saccades in Man,” European Journal
of Neuroscience, 16 (1996): 3726–3741.
292 •
29. W. F. McDaniel, D. M. Compton, and S. R. Smith, “Spatial Learning following Poste-
rior Parietal or Hippocampal Lesions,” NeuroReport, 5 (1994): 1713–1717. See also our
recent findings from a mental locomotor task [O. Ghaem, E. Mellet, F. Crivello, N.
Tzourio, B. Mazoyer, A. Berthoz, and M. Denis, “Mental Navigation along Memorized
Routes Activates the Hippocampus, Precuneus, and Insula,” NeuroReport 8 (1997): 739–
744].
30. M. N. Shadlen and W. T. Newsome, “Motion Perception: Seeing and Deciding,” Pro-
31. J.-R. Duhamel, C. L. Colby, and M. E. Goldberg, “The Updating of the Representation
of Visual Space in Parietal Cortex by Intended Eye Movements,” Science, 255 (1992):
90–92.
32. P. Viviani and A. Berthoz, “Voluntary Deceleration and Perceptual Activity during
Oblique Saccades,” in R. Baker and A. Berthoz, eds., Control of Gaze by Brain Stem Neu-
rons (Amsterdam: Elsevier, 1977), pp. 23–28.
33. M. A. Segraves and M. E. Goldberg, “Functional Properties of Corticotectal Neu-
rons in the Monkey’s Frontal Eye Field,” Journal of Neurophysiology, 58 (1987): 1387–
1419.
34. D. Ferrier, Functions of the Brain.
35. Penfield and Boldrey, “Somatic Motor and Sensory Representation”; Penfield and Ras-
mussen, Cerebral Cortex of Man.
36. E. Melamed and B. Larsen, “Saccadic Eye Movements in Humans”; P. T. Fox, J. M. Fox,
M. E. Raichle, and R. M. Burde, “The Role of the Cerebral Cortex in the Generation
of Voluntary Saccades,” Journal of Neurophysiology, 54 (1985): 348–369.
37. L. Petit, C. Orssaud, N. Tzourio, G. Salamon, B. Mazoyer, and A. Berthoz, “PET Study
of Voluntary Saccadic Eye Movements in Humans: Basal Ganglia-Thalamocortical Sys-
tem and Cingulate Cortex Involvement,” Journal of Neurophysiology, 69 (1993): 1009–
1017; Petit et al., “Functional Anatomy,” pp. 3726–3741.
38. L. Lang, D. Cheyne, R. Kristeva, R. Beisteiner, G. Lindinger, and L. Deecke, “Three-Di-
mensional Localization of SMA Activity Preceding Voluntary Movement. A Study of
Electric and Magnetic Fields in a Patient with Infarction of the Right Supplementary
Motor Area,” Experimental Brain Research, 87 (1991): 688–695.
39. M. P. Deiber, R. E. Passingham, J. G. Colebatch, K. J. Friston, P. D. Nixon, and R. S.
Frackowiak, “Cortical Areas and the Selection of Movement: A Study with Positron
Emission Tomography,” Experimental Brain Research, 84 (1991): 393–402.
40. W. Lang, M. Lang, F. Uhl, C. Koska, A. Kornhuber, and L. Deecke, “Negative Cortical
DC Shifts Preceding and Accompanying Simultaneous and Sequential Finger Move-
ments,” Experimental Brain Research, 71 (1988): 579–587; W. Lang, H. Obrig, G.
Lindinger, D. Cheyne, and L. Deecke, “Supplementary Motor Area Activation While
Tapping Bimanually Different Rhythms in Musicians,” Experimental Brain Research, 79
(1990): 504–514.
41. J. Schlag and M. Schlag-Rey, “Evidence for a Supplementary Eye Field,” Journal of
42. B. Gaymard, C. Pierrot-Deseilligny, and S. Rivaud, “Impairment of Sequences of Mem-
ory-Guided Saccades after Supplementary Motor Area Lesions,” Annals of Neurology, 28
(1990): 622–626; B. Gaymard, S. Rivaud, and C. Pierrot-Deseilligny, “Role of the Left

293
and Right Supplementary Motor Areas in Memory-Guided Saccade Sequences,” Annals
of Neurology, 34 (1993): 404–406.
43. Petit et al., “PET Study,” and “Functional Anatomy.”
44. S. Y. Musil, C. R. Olson, and M. E. Goldberg, “Visual and Oculomotor Properties of
Single Neurons in Posterior Cingulate Cortex of Rhesus Monkey,” Society of Neurosci-
ence Abstracts, 16 (1990): 1221.
45. P. S. Goldman-Rakic, “Circuitry of Primate Prefrontal Cortex and Regulation of Be-
havior by Representational Memory,” in V. B. Mountcastle, F. Plkum, and D. Hum-
phrey, eds., Handbook of Physiology. The Nervous System V (Bethesda: American Physio-
logical Society, 1987), pp. 373–417; S. Funahashi, C. Bruck, and P. S. Goldman-Rakic,
“Mnemonic Coding of Visual Space in the Monkey’s Dorsolateral Prefrontal Cortex,”
Journal of Neurophysiology, 61 (1989): 1–19.
46. J. P. Joseph and P. Barone, “Prefrontal Unit Activity during a Delayed Oculomotor Task
in the Monkey,” Experimental Brain Research, 67 (1987): 460–468; A. Diamond and P. S.
Goldman-Rakic, “Comparison of Human Infants and Rhesus Monkeys on Piaget’s AB
Task: Evidence for Dependence on Dorsolateral Prefrontal Cortex,” Experimental Brain
Research, 74 (1989): 24–40; C. D. Frith, K. Friston, P. F. Liddle, and R. S. Frackowiak,
“Willed Action and the Prefrontal Cortex in Man: A Study with PET,” Proceedings of the
Royal Society of London. Series B: Biological Sciences, 244 (1991): 241–246; G. Di Pellegrino
and S. P. Wise, “Visuospatial versus Visuomotor Activity in the Premotor and Pre-
frontal Cortex of a Primate,” Journal of Neuroscience, 13 (1993): 1227–1243; A. Bechara,
A. R. Damasio, H. Damasio, and S. W. Anderson, “Insensitivity to Future Conse-
quences following Damage to Human Prefrontal Cortex,” Cognition, 50 (1994): 7–15;
M. Petrides, “Impairments on Nonspatial Self-Ordered and Externally Ordered Work-
ing Memory Tasks after Lesions of the Mid-dorsal Part of the Lateral Frontal Cortex in
the Monkey,” Journal of Neuroscience, 15 (1995): 359–375.
47. D. Boussaoud and S. P. Wise, “Primate Frontal Cortex: Neuronal Activity following
Attentional versus Intentional Cues,” Experimental Brain Research, 95 (1993): 15–27; Di
Pellegrino and Wise, “Visuospatial versus Visuomotor Activity,” pp. 1227–1243; R. P.
Kesner, “Paired Associate Learning in the Rat: Role of Hippocampus, Medial Pre-
frontal Cortex, and Parietal Cortex,” Psychobiology, 21 (1993): 183–192.
48. Goldman-Rakic, “Circuitry of Primate Prefrontal Cortex”; J. M. Fuster, The Prefrontal
Cortex: Anatomy, Physiology, and Neuropsychology of the Frontal Lobe (New York: Raven
Press, 1996); S. Dehaene and J.-P. Changeux, “A Simple Model of Prefrontal Cortex
Function in Delayed-Response Tasks,” Journal of Cognitive Neuroscience, 1 (1990): 244–
261.
49. Pierrot-Deseilligny et al., “Role of the Different Frontal Lobe Areas.”
50. Israël et al., “Vestibular-Guided Saccades in Man.”
51. D. Guitton, H. A. Buchtel, and R. Douglas, “Frontal Lobe Lesions in Man Cause
Difficulties in Supressing Reflexive Glance and in Generating Goal-Directed Saccades,”
Experimental Brain Research, 58 (1985): 455–472.
52. K. Sasaki and H. Gemba, “Electrical Activity in the Prefrontal Cortex Specific to No-
Go Reaction of Conditioned Hand Movement with Colour Discrimination in the Mon-
key,” Experimental Brain Research, 64 (1986): 603–606.
53. J. P. Aggleton, N. Neave, S. Nagle, and A. Sahgal, “A Comparison of the Effects of Me-
294 •
dial Prefrontal, Cingulate Cortex, and Cingulum Bundle Lesions on Tests of Spatial
Memory: Evidence of a Double Dissociation between Frontal and Cingulum Bundle
Contributions,” Journal of Neuroscience, 15 (1995): 7270–7281; J. Mogensen, T. K.
Pedersen, S. Holm, and L. E. Bang, “Prefrontal Cortical Mediation of Rats’ Place
Learning in a Modified Water Maze,” Brain Research Bulletin, 38 (1995): 425–434.
54. E. Salzmann, “Attention and Memory Trials during Neuronal Recording from the Pri-
mate Pulvinar and Posterior Parietal Cortex (Area PG),” Behavioral Brain Research, 67
(1995): 241–253.
55. Frith et al., “Willed Action.”
56. S. Park and P. S. Holzman, “Association of Working Memory Deficit and Eye Tracking
Dysfunction in Schizophrenia,” Schizophrenia Research, 11 (1993): 55–61.
57. B. M. Sheliga, L. Riggio, and G. Rizzolatti, “Orienting of Attention and Eye Move-
ments,” Experimental Brain Research, 98 (1994): 507–522; A. Berthoz, “The Role of Inhi-
bition in the Hierarchical Gating of Executed and Imagined Movements,” Cognitive
Brain Research, 3 (1996): 101–113.
58. W. Lang, L. Petit, P. Hölliger, U. Pietrzyck, N. Tzourio, B. Mazoyer, and A. Berthoz, “A
Positron Emission Tomography Study of Oculomotor Imagery,” NeuroReport, 5 (1994):
921–924.
59. Droulez and Berthoz, “Concept of Dynamic Memory,” “A Neural Network Model of
Sensoritopic Maps with Predictive Short Term Memory Properties,” Proceedings of the
National Academy of Sciences, 88 (1991): 9653–9657, and “The Dynamic Memory Model
and the Final Oculomotor and Cephalomotor Integrator,” in H. Shimazu and Y.
Shinoda, eds., The Oculomotor System (Tokyo: Japan Scientific Societies Press, 1991),
pp. 1–19; L. E. Mays and D. L. Sparks, “Saccades Are Spatially, Not Retinocentrically,
Coded,” Science, 208 (1980): 1163–1165.
60. J. Piaget, “Le problème neurologique de l’intériorisation des actions en opérations
réversibles,” Archives de psychologie, 32 (1949): 242–258. Ferrier also became interested in
this idea. On page 285 of Functions of the Brain he wrote: “In these sentences, and par-
ticularly the last, Bain seems to me to have clearly indicated the elements of attention,
which I conceive to be a combination of the activity of the motor, and of the inhibi-
tory-motor centres.”
11 Balance
1. A. Thomas, Equilibre et équilibration (Paris: Masson, 1940).
2. Lee and Lishman, “Visual Proprioceptive Control of Stance.”
3. Dichgans and Brandt, “Visual-Vestibular Interaction.”
4. See Chapter 4.
5. Lee and Aronson, “Visual Proprioceptive Control of Standing.”
6. K. F. Hamann, P. P. Vidal, J. M. Sterkers, and A. Berthoz, “A New Test for Postural Dis-
orders: An Application of Visual Stabilization,” Agressologie, 20B (1979): 129–130.
7. Clément et al., “Adaptation of Postural Control.”
8. F. Lestienne, J. Soechting, and A. Berthoz, “Postural Readjustments Induced by Linear
Motion of Visual Scenes,” Experimental Brain Research, 28 (1977): 363–384; J. F.
Soechting and A. Berthoz, “Dynamic Role of Vision in the Control of Posture in
Man,” Experimental Brain Research, 36 (1979): 551–561.

295
9. See Chapter 2.
10. This mental calculation, used in clinical practice during vestibular tests, is assumed to
decrease high-level cognitive factors by occupying the subject with a mental task.
11. Sometimes the expression “posturally blind” is used to refer to persons who are insen-
sitive to fluctuations of the visual environment.
12. I. Viaud-Delmon, Y. P. Ivanenko, A. Berthoz, and R. Jouvent, “Sex, Lies, and Virtual Re-
ality,” Nature, 1 (1998): 15–16.
13. M. Reuchlin, J. Lautrey, C. Marendaz, and T. Ohlman, Cognition: L’universel et
l’individuel (Paris: Presses Universitaires de France, 1989); M. Denis, Image et Cognition
(Paris: Presses Universitaires de France, 1989).
14. L. M. Nashner and A. Berthoz, “Visual Contribution to Rapid Motor Responses during
Postural Control,” Brain Research, 150 (1978): 403–407.
15. For a review, see Prochazka, “Sensorimotor Gain Control.”
16. C. S. Sherrington, The Integrative Action of the Nervous System (New Haven: Yale Univer-
sity Press, 1906).
17. Feldman and Levin, “Positional Frames of Reference”; Bizzi et al., “Posture Control
and Trajectory Formation.” See Chapter 4.
18. W. Hess, “Teleokinetisches und ereismatisches Kraftsystem in der Biomotorik,”
Helvetica Physiologica et Pharmacologica Acta, 1 (1943): C62–C63.
19. Evarts, “Pyramidal Tract Activity”; E. V. Evarts, C. Fromm, J. Kroller, and V. A.
Jennings, “Motor Cortex Control of Finely Graded Forces,” Journal of Neurophysiology,
49 (1983): 1199–1215.
20. Paillard, “Le corps et ses langages d’espace,” and “L’intégration sensori-motrice et
idéo-motrice,” in M. Richelle, J. Requin, and M. Robert, eds., Traité de psychologie
expérimentale (Paris: Presses Universitaires de France, 1994), pp. 925–961; A. Riehle and
J. Requin, “Monkey Primary Motor and Premotor Cortex: Single-Cell Activity Related
to Prior Information about Direction and Extent of an Intended Movement,” Journal of
21. Sechenov, Selected Works.
22. J. Massion, V. Gurfinkel, M. Lipshits, A. Obadia, and K. Popov, “Stratégie et synergie:
deux niveaux de contrôle de l’équilibre au cours du mouvement. Effets de la micro-
gravité,” Comptes rendus hebdomadaires de l’Académie des sciences (Paris, Série III), 314
(1992): 87–92; P. Crenna, C. Frigo, J. Massion, and A. Pedotti, “Forward and Backward
Axial Synergies in Man,” Experimental Brain Research, 65 (1987): 538–548; A. Pedotti, P.
Crenna, A. Deat, C. Frigo, and J. Massion, “Postural Synergies in Axial Movements:
Short and Long-Term Adaptation,” Experimental Brain Research, 74 (1989): 3–10; J.
Massion, F. Viallet, R. Massarino, and R. Khalil, “La région de l’aire motrice supplé-
mentaire est impliquée dans la coordination entre posture et mouvement,” Comptes
rendus hebdomadaires de l’Académie des sciences (Paris, Série III), 308 (1989): 417–423.
23. S. Bouisset and M. Zattara, “A Sequence of Postural Movements Precedes Voluntary
Movement,” Neuroscience Letters, 22 (1989): 263–270, and “Biomechanical Study of
the Programming of Anticipatory Postural Adjustments Associated with Voluntary
Movement,” Journal of Biomechanics, 20 (1987): 735–742; D. Bazalgette, M. Zattara, N.
Bathien, S. Bouisset, and P. Rondot, “Postural Adjustments Associated with Rapid Vol-
296 •
untary Arm Movements in Patients with Parkinson’s Disease,” in M. D. Yahr and K. J.
Bergman, eds., Advances in Neurology, vol. 45 (New York: Raven Press, 1986), pp. 371–
374; Clément et al., “Adaptation of Postural Control.”
24. L. Mouchnino, R. Aurenty, J. Massion, and A. Pedotti, “Is the Trunk a Reference Frame
for Calculating Leg Position?” NeuroReport, 4 (1993): 125–127.
25. J. Babinski, “De l’asynergie cérébelleuse,” Revue de Neurologie, 7 (1989): 806–816; see re-
cent findings on the physiopathology of this synergy in Bouisset and Zattara, “Bio-
mechanical Study”; Clément et al., “Adaptation of Postural Control”; Crenna et al.,
“Forward and Backward.”
26. Bazalgette et al., “Postural Adjustments,” pp. 371–374; F. Viallet, J. Massion, R.
Massarino, and R. Khalil, “Coordination between Posture and Movement in Parkin-
sonism and SMA Lesion,” in L. Deecke, J. C. Eccles, and V. B. Mountcastle, eds., From
Neuron to Action (Berlin: Springer, 1990), pp. 65–70; R. G. Lee, I. Tonolli, F. Viallet, R.
Aurenty, and J. Massion, “Preparatory Postural Adjustment in Parkinsonian Patients
with Postural Instability,” Canadian Journal of Neurological Sciences, 22 (1995): 126–135.
27. Massion et al., “La région de l’aire motrice supplémentaire.”
28. Y. Gahery and J. Massion, “Coordination between Posture and Movement,” in E. V.
Evarts, S. P. Wise, and D. Bousfield, eds., The Motor System in Neurobiology (Amsterdam:
Elsevier Biomedical Press, 1985), pp. 121–125.
29. Y. Paulignan, M. Dufosse, M. Hugon, and J. Massion, “Acquisition of Coordination be-
tween Posture and Movement in a Bimanual Task,” Experimental Brain Research, 77
(1989): 337–348.
30. G. Clément, T. Pozzo, and A. Berthoz, “Contribution of Eye Positioning in the Control
of Upside-Down Vertical Posture,” Experimental Brain Research, 73 (1988): 569–576.
31. H. Head and G. Holmes, “Sensory Disturbances from Cerebral Lesions,” Brain, 34
(1911): 102–244.
32. V. S. Gurfinkel, E. E. Debreva, and Y. S. Levik, “The Role of the Internal Model in the
Perception of Position and Planning of Hand Movements,” Fiziologiia Cheloveka, 12
(1986): 769–776; V. S. Gurfinkel and Y. S. Levick, “Perceptual and Automatic Aspects of
the Postural Body Scheme,” in Paillard, Brain and Space, pp. 174–162.
33. E. D. Adrian, The Physical Background of Perception (Oxford: Clarendon Press, 1947),
pp. 93–94.
34. See Chapter 2.
35. A. Berthoz, “Coopération et substitution entre le système saccadique et les réflexes
d’origine vestibulaires: faut-il réviser la notion de réflèxe?” Revue Neurologique, 145
(1989): 513–526.
36. These experiments were continued by E. Bizzi’s group at MIT.
37. Gomi and Kawato, “Adaptive Feedback Control Models.”
38. J. Massion and L. Rispal-Padel, “Thalamus: fonctions motrices,” Revue Neurologique, 142
(1986): 327–336.
39. M. Poncet, J. F. Pellissier, M. Sebahoun, and C. J. Masser, “A propos d’un cas
d’autotopagnosie secondaire à une lésion pariéto-occipitale de l’hémisphère majeur,”
Encéphale, 60 (1971): 110–123.
40. M. I. Posner, M. J. Nissen, and W. C. Ogden, “Attended and Unattended Processing

297
Modes: The Role of Set for Spatial Location,” in H. L. Pick and I. J. Saltzman, eds.,
Modes of Perceiving and Processing Information (Hillsdale, N.J.: Erlbaum, 1988), pp. 137–
157.
41. A. Sirigu, J. Grafman, K. Bressler, and T. Sunderland, “Multiple Representations Con-
tribute to Body Knowledge Processing,” Brain, 114 (1991): 629–642.
42. E. Bonda, M. Petrides, S. Frey, and A. Evans, “Neural Correlates of Mental Transfor-
mations of the Body-in-Space,” Proceedings of the National Academy of Sciences, 92
(1995): 11180–11184.
43. M. M. Mesulam, “A Cortical Network for Directed Attention and Unilateral Neglect,”
Annals of Neurology, 10 (1981): 309–325.
12 Adaptation
1. A. Gonshor and G. Melvill Jones, “Extreme Vestibulo-ocular Adaptation by Prolonged
Optical Reversal of Vision,” Journal of Physiology, 256 (1976): 381–414.
2. See Figure 2.4.
3. M. Ito, “Cerebellar Control of the Vestibulo-ocular Reflex around the Flocculus Hy-
pothesis,” Annual Review of Neuroscience, 5 (1982): 275–296; The Cerebellum and Neuronal
Control (New York: Raven, 1984).
4. F. A. Miles and S. G. Lisberger, “Plasticity in the Vestibulo-ocular Reflex: A New Hy-
pothesis,” Annual Review of Neuroscience, 4 (1981): 273–299; S. Du Lac, J. L. Raymond,
T. J. Sejnowski, and S. G. Lisberger, “Learning and Memory in the Vestibulo-Ocular
Reflex,” Annual Review of Neuroscience, 18 (1995): 409–441.
5. Berthoz and Melvill-Jones, Adaptative Mechanisms in Gaze Control; G. Melvill-Jones,
D. Guitton, and A. Berthoz, “Changing Patterns of Eye-Head Coordination during
Six Hours of Optically Reversed Vision,” Experimental Brain Research, 69 (1988): 531–
544.
6. G. Melvill-Jones, A. Berthoz, and B. Segal, “Adaptative Modification of the Vestibulo-
ocular Reflex by Mental Effort in Darkness,” Experimental Brain Research, 56 (1984):
149–153.
7. A. Berthoz, “Coopération et substitution entre le système saccadique et les réflexes
(1989): 513–526; Viaud-Delmon et al., “Sex, Lies, and Virtual Reality”; I. Viaud-Del-
mon, Y. Ivanenko, A. Berthoz, and R. Jouvent, “Anxiety and Integration of Visual Ves-
tibular Information Studied with Virtual Reality,” Biological Psychiatry, in press.
8. Nudo et al., “Use-Dependent Alterations.”
9. J.-C. Baudrillard, D. Rousié, H. Foucart, C. Defache, J.-C. Hache, P. Van Tichelen, and
J.-M. Lerais, “IRM des canaux semi-circulaires dans les asymétries cranio-faciales et le
syndrome cranio-mandibulaire,” Journal of Radiology, 76 (1995): 579–585.
10. Ivanenko et al., “The Contribution of Otoliths.”
11. Vitte et al., “Recovery from Vestibular Deficits.”
13 The Disoriented Brain

1. J. Sully, Illusions: A Psychological Study (New York: Da Capo Press, 1982), p. 333.
2. K. S. Lashley, Brain Mechanisms and Intelligence (Chicago: University of Chicago Press,
1929).
298 •
3. I. Krechevsky, “Hypothesis in Rats,” Psychological Reviews, 39 (1932): 517–532; quotation
from p. 528.
4. C. Darlot, P. Denise, J. Droulez, B. Cohen, and A. Berthoz, “Eye Movements Induced
by Off-Vertical Axis Rotation (OVAR) at Small Angles of Tilt,” Experimental Brain Re-
search, 73 (1988): 91–105, and “Motion Perception Induced by Off-Vertical Axis Rota-
tion (OVAR) at Small Angles of Tilt,” Experimental Brain Research, 73 (1988): 106–114.
5. A. Fessard and A. Tournay, “Quelques données et réflexions sur le phénomène de la
post-contraction involontaire,” L’Année psychologique, 50 (1950): 216–235. This illusion
was recently investigated by Lestienne and Gurfinkel.
6. J. P. Roll, J. C. Gilhodes, R. Roll, and F. Harley, “Are Proprioceptive Sensory Inputs
Combined into a Gestalt?” in T. Inoue and J. T. MacClelland, eds., Attention and Perfor-
mance XVI (Cambridge, Mass.: MIT Press, 1995), pp. 290–314; Lackner and Levine,
“Changes in Apparent Body Orientation.”
7. Roll et al., “Are Proprioceptive Sensory Inputs Combined?” pp. 307–308.
8. Rolls and O’Mara, “Neurophysiological and Theoretical Analysis”; O’Mara et al.,
“Whole Body Motion.”
9. The recent findings of Damasio regarding the significance of the prefrontal cortex
in relationships between vegetative and cognitive mechanisms and a rereading of
the work of the Sokolov school on relationships between the hippocampus and the
reticular formation stimulated this speculation.
10. H. Hécaen and J. De Ajuriaguerra, Méconnaissances et hallucinations corporelles:
intégration et désintégration de la somatognosie (Paris: Masson et Cie, 1952).
11. Shepard, “Ecological Constraints,” p. 436.
14 Architects Have Forgotten the Pleasure of Movement
1. Baron Georges Eugène Haussmann (1809–1891) was the French city planner whose re-
design of Paris between 1853 and 1870 resulted in the broad streets, radiating avenues,
and vistas for which the city is famous. [Trans.]
Conclusion
1. Bergson, Matter and Memory, p. 46.
2. H. Bergson, Mind-Energy: Lectures and Essays, trans. H. Wildon Carr (Westport, Conn.:
Greenwood, 1975), p. 58.
3. A. Sirigu, J. R. Duhamel, L. Cohen, B. Pillon, B. Dubois, and Y. Abid, “The Mental Rep-
resentation of Hand Movements after Parietal Cortex Damage,” Science, 273 (1996):
1564–1568.

299
WORKS CITED
Abeles, M. Local Cortical Circuits. Berlin: Springer, 1982.

Adrian, E. D. The Physiological Background of Perception. Oxford: Clarendon Press, 1947.
Aggleton, J. P., N. Neave, S. Nagle, and A. Sahgal. “A Comparison of the Effects of Medial
Prefrontal, Cingulate Cortex, and Cingulum Bundle Lesions on Tests of Spatial Mem-
ory: Evidence of a Double Dissociation between Frontal and Cingulum Bundle Con-
tributions.” Journal of Neuroscience, 15 (1995): 7270–7281.
Akbarian, S., O.-J. Grüsser, and W. O. Guldin. “Corticofugal Projections to the Vestibular
Nuclei in Squirrel Monkeys: Further Evidence of Multiple Cortical Vestibular Fields.”
Journal of Comparative Neurology, 332 (1993): 89–104.
Amblard, B., A. Berthoz, and F. Clarac, eds. Posture and Gait: Development, Adaptation, and
Modulation. Amsterdam: Elsevier, 1988.
Amorim, M. A., S. Glasauer, K. Corpinot, and A. Berthoz. “Updating an Object’s Orienta-
tion and Location during Non-Visual Navigation: A Comparison between Two Pro-
cessing Modes.” Perception and Psychophysics 59 (1997): 404–418.
Andersen, R. A. “Visual and Oculomotor Functions of the Posterior Parietal Cortex.” An-
nual Review of Neuroscience, 12 (1989): 377–403.
Andersen, R. A., R. M. Bracewell, S. Barash, J. W. Gnadt, and L. Fogasse. “Eye Position Ef-
fects on Visual, Memory and Saccade-Related Activity in Area LIP and 7a of the Ma-
caque.” Journal of Neuroscience, 10 (1990): 1176–1196.
Andersen, R. A., G. K. K. Essick, and R. M. Seigel. “Neurons of Area 7 Activated by Both
Visual Stimuli and Oculomotor Behavior.” Experimental Brain Research, 67 (1987): 316–
322.
Andersen, R. A., and V. B. Mountcastle. “The Influence of the Angle of Gaze upon the Ex-
citability of the Light Sensitive Neurons of the Posterior Parietal Cortex.” Journal of
Neuroscience, 3 (1983): 532–548.
Anokhin, P. K. Biology and Neurophysiology of Conditioned Reflexes and Their Role in Adaptive
Behaviour. Oxford: Pergamon Press, 1974.
Arbib, M. A. “Perceptual Structures and Distributed Motor Control.” In Handbook of Physi-
ology, sect. 1, The Nervous System, vol. 2, part 2, ed. V. B. Brooks. Bethesda: American
Physiological Society, 1981, 1449–1480.
———. “Interaction of Multiple Representations of Space in the Brain.” In Brain and Space,
ed. J. Paillard. Oxford: Oxford University Press, 1993, 380–403.
Arbib, M., and S. Amari. “Sensorimotor Transformations in the Brain (with a Critique of
the Tensor Theory of the Cerebellum).” Journal of Theoretical Biology, 112 (1985): 123–
155.
Arbib, M., T. Iberall, and G. Bingham. “Opposition Space as a Structuring Concept for
the Analysis of Skilled Hand Movements.” Experimental Brain Research, 15 (1986): 158–
173.
Aristotle. Psychologie. Opuscules. Paris: Dumont, 1847.
———. De Sensu and De Memoria, trans. G. R. T. Ross. New York: Arno, 1973.
Aubert, H. “Über eine scheinebare Drehung von Objecten bei Neigung des Kopfes nach
rechts oder links.” Virchows Archiv, 20 (1967): 381–393.
Babinski, J. “De l’asynergie cérébelleuse.” Revue de Neurologie, 7 (1899): 806–816.
Bach-y-Rita, P. “Vision Substitution by Tactile Image Projection.” Nature, 221 (1989): 963–
964.
Baker, R., and A. Berthoz. “Organization of Vestibular Nystagmus in the Oblique
Oculomotor System.” Journal of Neurophysiology, 37 (1974): 195–217.
———. “Is the Prepositus Hypoglossi Nucleus the Source of Another Vestibular Ocular
Pathway?” Brain Research, 86 (1975): 121–172.
———. Control of Gaze by Brain Stem Neurons. Amsterdam: Elsevier, 1977.
Baker, R., M. Gresty, and A. Berthoz. “Neuronal Activity in the Prepositus Hypoglossi Nu-
cleus Correlated with Vertical and Horizontal Eye Movements in the Cat.” Brain Re-
search, 101 (1976): 366–371.
Barbur, J. L., A. J. Harlow, and L. Weiskrantz. “Spatial and Temporal Response Properties
of Residual Vision in a Case of Hemianopia.” Philosophical Transactions of the Royal So-
ciety of London. Series B: Biological Sciences, 343 (1994): 157–166.
Barlow, J. S. “Inertial Navigation as a Basis for Animal Navigation.” Journal of Theoretical Bi-
ology, 6 (1964): 76–117.
Baudrillard, J.-C., D. Rousié, H. Foucart, C. Defache, J.-C. Hache, P. Van Tichelen, and J.-
M. Lerais. “IRM des canaux semi-circulaires dans les asymétries cranio-faciales et le
syndrome cranio-mandibulaire.” Journal of Radiology, 76 (1995): 579–585.
Bazalgette, D., M. Zattara, N. Bathien, S. Bouisset, and P. Rondot. “Postural Adjustments
Associated with Rapid Voluntary Arm Movements in Patients with Parkinson’s Dis-
ease.” In Advances in Neurology, vol. 45, ed. M. D. Yahr and K. J. Bergman. New York:
Raven Press, 1986, 371–374.
Bechara, A., A. R. Damasio, H. Damasio, and S. W. Anderson. “Insensitivity to Future
Consequences following Damage to Human Prefrontal Cortex.” Cognition, 50 (1994):
7–15.
Belen’kii, V. E., V. S. Gurfinkel, and Y. I. Pal’tsev. “On Elements of Voluntary Movement
Control.” Biofizika, 12 (1967): 135–141.
Bergson, H. “L’âme et le corps.” In Le Matérialisme actuel. Paris: Flammarion, 1913.
———. The Creative Mind, trans. Mabelle L. Andison. New York: Philosophical Library,
1946.
———. Mind-Energy: Lectures and Essays, trans. H. Wildon Carr. Westport, Conn.: Green-
wood, 1975.
———. Matter and Memory, trans. Nancy Margaret Paul and W. Scott Palmer. New York:
Zone Books, 1988.
Beritoff, J. S. Neural Mechanisms of Higher Vertebrate Behavior. New York: Brown and Co.,
1965.
Bernstein, N. A. “Some Emergent Problems of the Regulation of Motor Acts,” Questions of
Psychology, no. 6 (1957). Reprinted in Human Motor Actions: Bernstein Reassessed, ed.
H. T. A. Whiting. Advances in Psychology, vol. 17. Amsterdam: North-Holland, 1984,
354–355.
———. The Coordination and Regulation of Movement. New York: Pergamon Press,
1967.
WORKS CITED •
301
Bertenthal, B. I., and R. K. Clifton. “Perception and Action.” In Handbook of Child Psychol-
ogy, ed. D. Kuhn and R. Siegler. New York: Wiley, 1996.
Berthoz, A. “Coopération et substitution entre le système saccadique et les réflexes
(1989): 513–526.
———. “Reference Frames for the Perception and Control of Movement.” In Brain and
Space, ed. J. Paillard. Oxford: Oxford University Press, 1991, 82–111.
———. “La coopération des sens et du regard dans la perception du mouvement.” In Le
Corps en jeu, ed. O. Aslan. Paris: CNRS Editions, 1993, 17–25.
———, ed. Multisensory Control of Movement. Oxford: Oxford University Press, 1993.
———. “Neural Basis of Decision in Perception and in Control of Movement.” In
Neurobiology of Decision-Making, ed. A. R. Damasio, H. Damasio, and Y. Christen.
Berlin: Springer, 1996, 83–100.
———. “How Does the Cerebral Cortex Process and Utilize Vestibular Signals?” In Disor-
ders of the Vestibular System, ed. R. W. Baloh and G. M. Halmagyi. Oxford: Oxford Uni-
versity Press, 1996, 113–125.
———. “The Role of Inhibition in the Hierarchical Gating of Executed and Imagined
Movements.” Cognitive Brain Research, 3 (1996): 101–113.
———. “Hippocampal and Parietal Contribution to Topokinetic and Topographic Mem-
ory.” In The Hippocampal and Parietal Foundations of Spatial Cognition, ed. N. Burgess,
K. J. Jeffery, and J. O’Keefe. Oxford: Oxford University Press, 1999, 381–399.
———. Leçons sur le corps, le cerveau et l’esprit. Paris: Odile Jacob, 1999.
Berthoz, A., A. Amorim, S. Glasauer, R. Grasso, Y. Takei, and I. Viaud-Delmon. “Dissocia-
tion between Distance and Direction during Locomotor Navigation.” In Wayfinding
Behaviour, ed. R. G. Golledge. Baltimore: John Hopkins University Press, 1999, 328–
348.
Berthoz, A., R. Baker, and W. Precht. “Labyrinthine Control of Inferior Oblique Moto-
neurons.” Experimental Brain Research, 18 (1973): 225–241.
Berthoz, A., J. Droulez, P. P. Vidal, and K. Yoshida. “Neural Correlates of Horizontal
Vestibulo-ocular Reflex Cancellation during Rapid Eye Movements in the Cat.” Jour-
nal of Physiology, 419 (1989): 717–751.
Berthoz, A., W. Graf, and P. P. Vidal. The Head-Neck Sensory-Motor System. New York: Ox-
ford University Press, 1991.
Berthoz, A., A. Grantyn, and J. Droulez. “Some Collicular Neurons Code Saccadic Eye Ve-
locity.” Neuroscience Letters, 72 (1987): 289–294.
Berthoz, A., and A. Guell. “Space Neuroscience Research.” Brain Research Reviews, 28
(1998): 1–234.
Berthoz, A., I. Israël, P. Georges-François, R. Grasso, and T. Tsuzuku. “Spatial Memory of
Body Linear Displacement: What Is Being Stored?” Science, 269 (1995): 95–98.
Berthoz, A., I. Israël, T. Vieville, and D. S. Zee. “Linear Head Displacement Measured by
the Otoliths Can Be Reproduced through the Saccadic System.” Neuroscience Letters,
82 (1987): 285–290.
Berthoz, A., and G. Melvill-Jones. Adaptative Mechanisms in Gaze Control. Amsterdam:
Elsevier, 1985.
Berthoz, A., and S. Metral. “A Torque-Producing Stimulator for the Study of Muscular Re-
302 •
WORKS CITED
sponse to Variable Forces.” Proceedings of the International Symposium on Biomechanics
III. Basel: Karger, 1973, 158–165.
Berthoz, A., B. Pavard, and L. Young. “Perception of Linear Horizontal Self Motion In-
duced by Peripheral Vision (Linear-Vection).” Experimental Brain Research, 23 (1974):
471–489.
Berthoz, A., and L. Petit. “Les mouvements du regard: une affaire de saccades. Un modèle
pour l’étude des circuits de la décision et de l’imagination motrice.” La Recherche, 289
(1996): 58–65.
Berthoz, A., and T. Pozzo. “Intermittent Head Stabilisation during Postural and
Locomotory Tasks in Humans.” In Posture and Gait: Development, Adaptation, and Mod-
ulation, ed. B. Amblard, A. Berthoz, and F. Clarac. Amsterdam: Elsevier, 1988, 189–
198.
Berthoz, A., and P. P. Vidal. Noyaux Vestibulaires et Vertiges. Paris: Arnette, 1993.
Berthoz, A., P. P. Vidal, and W. Graf, eds. The Head-Neck Sensory-Motor System. New York:
Oxford University Press, 1991.
Berthoz, A., K. Yoshida, and P. P. Vidal. “Horizontal Eye Movement Signals in Second Or-
der Vestibular Nuclei Neurons in the Alert Cat.” Annals of the New York Academy of
Sciences, 374 (1981): 144–156.
Bienenstock, E., and C. von der Malsburg. “Statistical Coding and Short Term Synaptic
Plasticity: A Scheme for Knowledge Representation in the Brain.” In Disordered Sys-
tems and Biological Organization, ed. E. Bienenstock, F. Fogelman, and G. Weisbuch.
Bisiach, E., M. L. Rusconi, and G. Vallar. “Remission of Somatoparaphrenic Delusion
through Vestibular Stimulation.” Neuropsychologia, 29 (1991): 1029–1031.
Bizzi, E., N. Accornero, W. Chapple, and N. Hogan. “Posture Control and Trajectory For-
mation during Arm Movement.” Journal of Neuroscience, 4 (1984): 2738–2744.
Bloomberg, J., G. Melvill-Jones, B. Segal, S. McFarlane, and J. Soul. “Vestibular Contingent
Voluntary Saccades Based on Cognitive Estimates of Remembered Vestibular Infor-
mation.” Advances in Oto-Rhino-Laryngology, 40 (1988): 71–75.
Bonda, E., M. Petrides, S. Frey, and A. Evans. “Neural Correlates of Mental Transforma-
tions of the Body-in-Space.” Proceedings of the National Academy of Sciences, 92 (1995):
11180–11184.
Bootsma, R. J., and C. E. Peper. “Predictive Visual Information Sources for the Regulation
of Action with Special Emphasis on Catching and Hitting.” In Vision and Motor Con-
trol, ed. L. Proteau and D. Elliot. Amsterdam: Elsevier, 1992, 285–314.
Bottini, G., E. Paulesu, R. Sterzi, E. Warburton, R. J. Wise, G. Vallar, R. S. Frackowiak, and
C. D. Frith. “Modulation of Conscious Experience by Peripheral Sensory Stimuli.”
Nature, 376 (1995): 778–781.
Bottini, G., R. Sterzi, E. Paulesu, G. Vallar, S. Cappa, F. Erminio, R. E. Passingham, C. D.
Frith, and R. S. J. J. Frackowiak. “Identification of the Central Vestibular Projec-
tions in Man: A Positron Emission Tomography Activation Study.” Experimental Brain
Resarch, 99 (1994): 164–169.
Bouisset, S., and M. Zattara. “Biomechanical Study of the Programming of Anticipatory
Postural Adjustments Associated with Voluntary Movement.” Journal of Biomechanics,
20 (1987): 735–742.
WORKS CITED •
303
———. “A Sequence of Postural Movements Precedes Voluntary Movement.” Neuroscience
Letters, 22 (1989): 263–270.
Bourdieu, P. Le Sens pratique. Paris: Editions de Minuit, 1980.
Boussaoud, D., and S. P. Wise. “Primate Frontal Cortex: Neuronal Activity following
Attentional versus Intentional Cues.” Experimental Brain Research, 95 (1993): 15–27.
Bouveresse, J. Langage, perception et réalité. Nîmes: Jacqueline Chambon, 1995.
Bouyer, J. J., M. F. Montaron, J. M. Vahnée, M. Albert, and A. Rougeul. “Anatomical Local-
ization of Beta Rhythms in Cats.” Neuroscience, 22 (1987): 863–869.
Bradley, D. C., M. Maxwell, R. A. Andersen, M. S. Banks, and K. V. Shenoy. “Mechanisms
of Heading Perception in Primate Visual Cortex.” Science, 273 (1996): 1544–1547.
Brandt, T., J. Dichgans, and E. Koenig. “Differential Effects of Central versus Peripheral Vi-
sion on Egocentric and Exocentric Motion Perception.” Experimental Brain Research,
16 (1973): 476–491.
Brandt, T., and M. Dieterich. “Skew Deviation with Ocular Torsion: A Vestibular Brain-
stem Sign of Topographic Diagnostic Value.” Annals of Neurology, 33 (1993): 528–534.
Bridgeman, B. “A Review of the Role of Efference Copy in Sensory and Oculomotor Con-
trol Systems.” Annals of Biomedical Engineering, 23 (1995): 409–422.
Buchanan, J. J., and J. A. S. Kelso. “Posturally Induced Transitions in Rhythmic Multijoint
Limb Movements.” Experimental Brain Research, 94 (1993): 131–142.
Buisseret, P., and E. Gary-Bobo. “Development of Visual Cortical Orientation Specificity
after Dark Rearing: Role of Extra-ocular Proprioception.” Neuroscience Letters, 13
(1979): 259–263.
Buisseret, P., and M. Imbert. “Visual Cortical Cells: Their Developmental Properties in
Normal and Dark Reared Kittens.” Journal of Physiology, 255 (1976): 511–525.
Burnod, Y. An Adaptative Network, the Cerebral Cortex. Paris: Masson, 1988.
Buzsáki, G., Z. Horváth, R. Urioste, J. Hetke, and K. Wise. “High Frequency Network Os-
cillations in the Hippocampus.” Science, 256 (1992): 1025–1027.
Buzsáki, G., R. Llinás, W. Singer, A. Berthoz, and Y. Chrísten. Temporal Coding in the Brain.
Berlin: Spring, 1994.
Cajal, S. R. Histologie du système nerveux de l’homme et des vertébrés. Madrid: Raycar, 1972
[original pub. Paris, 1911].
Cappa, S., R. Sterzi, G. Vallar, and E. Bisiach. “Remission of Hemineglect and Anosognosia
during Vestibular Stimulation.” Neuropsychologia, 25 (1994): 722–732.
Celichowski, J., F. Emonet-Denand, Y. Laporte, and J. Petit. “Distribution of Static γ Axons
in Cat Peroneus Tertius Spindles Determined by Exclusively Physiological Criteria.”
Journal of Neurophysiology, 71 (1994): 722–732.
Changeux, J.-P. L’Homme neuronal. Paris: Fayard, 1983.
Châtelet, G. Les Enjeux du mobile—Mathématiques, physique, philosophie. Paris: Le Seuil, 1993.
Chatila, M., C. Milleret, P. Buser, and A. Fougeul. “A 10 Hz ‘Alpha-Like’ Rhythm in the Vi-
sual Cortex of the Waking Cat.” Electroencephalography and Clinical Neurophysiology, 83
(1992): 217–222.
Chen, L. L., L. Lin, E. J. Green, C. A. Barnes, and B. McNaughton. “Head Direction Cells
in the Rat Posterior Cortex. I. Anatomical Distribution and Behavioural Modulation.”
Chen, L. L., L. Lin, C. A. Barnes, and B. McNaughton. “Head Direction Cells in the Rat
304 •
WORKS CITED
Posterior Cortex. II. Contributions of Visual and Ideothetic Information to the Direc-
tional Firing.” Experimental Brain Research, 101 (1994): 24–34.
Clément, G., V. S. Gurfinkel, F. Lestienne, M. Lipshits, and K. Popov. “Adaptation of Pos-
tural Control in Weightlessness.” Experimental Brain Research, 57 (1984): 61–72.
Clément, G., and F. Lestienne. “Adaptive Modifications of Postural Attitude in Conditions
of Weightlessness.” Experimental Brain Research, 72 (1988): 381–389.
Clément, G., T. Pozzo, and A. Berthoz. “Contribution of Eye Positioning in the Control of
Upside-Down Vertical Posture.” Experimental Brain Research, 73 (1988): 569–576.
Colby, C. L., J.-R. Duhamel, and M. E. Goldberg. “Oculocentric Spatial Representation in
Parietal Cortex.” Cerebral Cortex, 5 (1995): 470–481.
Columb, Georges [Christophe pseud]. L’Idée fixe du savant Cosinus. Paris: Armand Colin,
1939.
Crenna, P., C. Frigo, J. Massion, and A. Pedotti. “Forward and Backward Axial Synergies in
Man.” Experimental Brain Research, 65 (1987): 538–548.
Curthoys, I. S., C. H. Markham, and N. Furuya. “Direct Projection of Pause Neurons to
Nystagmus-Related Excitatory Burst Neurons in the Cat Pontine Reticular Forma-
tion.” Experimental Neurology, 83 (1984): 414–422.
Cutting, J. E., and P. M. Vishton. “Perceiving Layout and Knowing Distances: The Integra-
tion, Relative Potency and Contextual Use of Different Information about Depth.” In
Perception of Space and Motion. San Diego: Academic Press, 1995, 69–117.
Dai, M. J., I. Curthoys, and G. M. Halmagyi. “A Model of Otolith Stimulation.” Biological
Cybernetics, 60 (1989): 185–194.
Damasio, A. R. “The Brain Binds Entities and Events by Multiregional Activation from
Convergence Zones.” Neural Computation, 1 (1989): 123–132.
———. Descartes’ Error. New York: Putnam, 1994.
Damasio, A. R., H. Damasio, and Y. Christen. Neurobiology of Decision-Making. Berlin:
Springer, 1996.
Darlot, C., P. Denise, J. Droulez, B. Cohen, and A. Berthoz. “Eye Movements Induced by
Off-Vertical Axis Rotation (OVAR) at Small Angles of Tilt.” Experimental Brain Re-
search, 73 (1988): 91–105.
———. “Motion Perceptions Induced by Off-Vertical Axis Rotation (OVAR) at Small An-
gles of Tilt.” Experimental Brain Research, 73 (1988): 106–114.
Darwin, C. “Origin of Certain Instincts.” Nature, 179 (1887): 417–418.
———. The Expression of the Emotions in Man and Animals. New York: Philosophical Library,
1955.
De Ajuriaguerra, J. Résumés des cours. Paris: Collège de France, 1976.
Decéty, J., and D. H. Ingvar. “Brain Structures Participating in Mental Simulation of Motor
Behaviour: A Neurophysiological Interpretation.” Acta Physiological Scandinavica, 73
(1990): 13–34.
Decéty, J., M. Jeannerod, M. Germain, and J. Pastène. “Vegetative Response during Imag-
ined Movement Is Proportional to Mental Effort.” Behavioural Brain Research, 24
(1991): 1–5.
Decéty, J., M. Jeannerod, and C. Prablanc. “The Timing of Mentally Represented Actions.”
Behavioural Brain Research, 34 (1989): 35–42.
Deecke, L., D. W. F. Schwartz, and J. M. Fredrickson. “Nucleus Ventro-posterior Inferior
WORKS CITED •
305
(VPI) as the Vestibular Thalamic Relay in the Rhesus Monkey. I. Field Potential Inves-
tigation.” Experimental Brain Research, 20 (1974): 88–190.
Dehaene, S., and J.-P. Changeux. “A Simple Model of Prefrontal Cortex Function in De-
layed-Response Tasks.” Journal of Cognitive Neuroscience, 1 (1990): 244–261.
Deiber, M. P., R. E. Passingham, J. G. Colebatch, K. J. Friston, P. D. Nixon, and R. S.
Frackowiak. “Cortical Areas and the Selection of Movement: A Study with Positron
Emission Tomography.” Experimental Brain Research, 84 (1991): 393–402.
De Kleijn, A., and T. Magnus. “Uber die Funktion der Otolithen.” Pflugers Archiv, 186
(1921): 6–81.
Deniau, J. M., and G. Chevalier. “Disinhibition as a Basic Process in the Expression of
Striatal Functions. The Striatonigral Influence on Thalamocortical Cells of the Ven-
tromedial Thalamic Nucleus.” Brain Research, 334 (1981): 227–233.
Denis, M. “Approches différentielles de l’imagerie mentale.” In Cognition: L’Universel et
l’individuel, ed. M. Reuchlin, J. Lautrey, C. Marendaz, and T. Ohlman. Paris: Presses
Universitaires de France, 1989.
———. Image et cognition. Paris: Presses Universitaires de France, 1989.
De Waele, C., W. Graf, A. Berthoz, and F. Clarac. “Vestibular Control of Skeletal Geome-
try.” In Posture and Gait, ed. A. Berthoz and F. Clarac. Amsterdam: Elsevier, 1988, 423–
432.
Diamond, A., and P. S. Goldman-Rakic. “Comparison of Human Infants and Rhesus Mon-
keys on Piaget’s AB Task: Evidence for Dependence on Dorsolateral Prefrontal Cor-
tex.” Experimental Brain Research, 74 (1989): 24–40.
Dichgans, J., and T. Brandt. “Visual-Vestibular Interaction and Motion and Perception.” In
Cerebral Control of Eye Movements and Motion Perception, ed. J. Dichgans and E. Bizzi.
New York: Karger, 1972, 327–338.
———. “Visual-Vestibular Interactions: Effects on Self-Motion Perception and Postural
Control.” In Handbook of Sensory Physiology, vol. 5, ed. H. Leibowitz and H. L. Teuber.
Dichgans, J., C. L. Schmidt, and W. Graf. “Visual Input Improves the Speedometer Func-
tion of the Vestibular Nuclei in the Goldfish.” Experimental Brain Research, 18 (1973):
319–322.
Dieterich, M., and T. Brandt. “Thalamic Infarctions: Differential Effects on Vestibular Func-
tion in the Roll Plane (35 Patients).” Neurology, 43 (1993): 1732–1740.
Di Pellegrino, G., L. Fadiga, L. Fogassi, V. Gallese, and G. Rizzolatti. “Understanding Mo-
tor Events: A Neurophysiological Study.” Experimental Brain Research, 91 (1992): 176–
180.
Di Pellegrino, G., and S. P. Wise. “Visuospatial versus Visuomotor Activity in the Premotor
and Prefrontal Cortex of a Primate.” Journal of Neuroscience, 13 (1993): 1227–1243.
Donders, F. C. “Beitrag zur Lehre von den Bewegungen des menschlichen Auges.” In
Hollandischen Beitragen zu den Anatomischen und Physiologischen Wissenschaften, vol. 1.
Amsterdam, 1847, 104–145.
Droulez, J., and A. Berthoz. “Servo-Controlled (Conservative) versus Topological (Projec-
tive) Modes of Sensory Motor Control.” In Disorders of Posture and Gait, ed. W. Bles
and T. Brandt. Amsterdam: Elsevier, 1988, 83–97.
———. “The Concept of Dynamic Memory in Sensorimotor Control.” In Motor Control:
306 •
WORKS CITED
Concepts and Issues, ed. D. R. Humphrey and J.-J. Freund. Chichester: Wiley, 1990, 137–
161.
———. “The Dynamic Memory Model and the Final Oculomotor and Cephalomotor Inte-
grator. In The Oculomotor System, ed. H. Shimazu and Y. Shinoda. Tokyo: Japan
Scientific Societies Press, 1991, 1–19.
———. “A Neural Network Model of Sensoritopic Maps with Predictive Short-Term Mem-
ory Properties.” Proceedings of the National Academy of Sciences, 88 (1991): 9653–9657.
Droulez, J., A. Berthoz, and P. P. Vidal. “Use and Limits of Visual Vestibular Interaction in
the Control of Posture. Are There Two Modes of Sensorimotor Control?” In Vestibu-
lar and Visual Control on Posture and Locomotor Equilibrium, ed. M. Igarashi and F. Owen
Black. Basel: Karger, 1985, 14–25.
Droulez, J., and V. Cornilleau-Pérès. “Application of the Coherence Scheme to the Multi-
sensory Fusion Problem.” In Multisensory Control of Movement, ed. A. Berthoz. Oxford:
Oxford University Press, 1993, 485–508.
Droulez, J., and C. Darlot. “The Geometric and Dynamic Implications of the Coherence
Constraints in Three-Dimensional Sensorimotor Coordinates.” In Attention and Perfor-
mance XIII, ed. M. Jeannerod. Hillsdale, N.J.: Erlbaum, 1989, 495–526.
Duhamel, J.-R., F. Bremmer, S. Ben Hamed, and W. Graf. “Spatial Invariance of Visual Re-
ceptive Fields in Parietal Cortex Neurons.” Nature, 389 (1997): 845–848.
Duhamel, J.-R., C. L. Colby, and M. E. Goldberg. “Congruent Representations of Visual
and Somatosensory Space in Single Neurons of Monkey Ventral Intraparietal Cortex
(Area VIP).” In Brain and Space, ed. J. Paillard. Oxford: Oxford University Press, 1991,
223–236.
———. “The Updating of the Representation of Visual Space in Parietal Cortex by In-
tended Eye Movements.” Science, 255 (1992): 90–92.
Du Lac, S., J. L. Raymond, T. J. Sejnowski, and S. G. Lisberger. “Learning and Memory in
the Vestibulo-ocular Reflex.” Annual Review of Neuroscience, 18 (1995): 409–441.
Eccles, J. C., M. Ito, and J. Szentogothai. The Cerebellum as a Neuronal Machine. Berlin:
Springer, 1967.
Eckhorn, R., R. Bauer, W. Jordan, M. Brosch, W. Kruse, M. Munk, and H. J. Reitboeck.
“Coherent Oscillations: A Mechanism of Feature Linking in the Visual Cortex?” Bio-
logical Cybernetics, 60 (1988): 121–130.
Eichenbaum, H., S. I. Wiener, M. L. Shapiro, and N. J. Cohen. “The Organization of Spa-
tial Coding in the Hippocampus: A Study of Neural Ensemble Activity.” Journal of
Neuroscience, 9 (1989): 2764–2775.
Emonet-Dénand, F., Y. Laporte, P. B. C. Matthews, and J. Petit. “On the Subdivision of
Static and Dynamic Fusimotor Actions on the Primary Ending of Cat Muscle Spin-
dle.” Journal of Physiology, 261 (1977): 827–861.
Ernst, H. Les Principes de la mécanique (1894). Cited in R. N. Shepard, “Ecological Con-
straints on Internal Representation: Resonant Kinematics of Perceiving, Imagining,
Thinking, and Dreaming.” Psychological Review, 91 (1984): 417–447.
Eshkol, N., and A. Wachmann. Movement Notation. London: Weidenfeld and Nicolson,
1958.
Etienne, A. S., R. Mowrer, and F. Saucy. “Limitations in the Assessment of Path-Dependent
Integration.” Behaviour, 106 (1988): 81–111.
WORKS CITED •
307
Evarts, E. V. “Relation of Pyramidal Tract Activity to Force Exerted during Voluntary
Movements.” Journal of Neurophysiology, 31 (1968): 14–27.
———. “Role of Motor Cortex in Voluntary Movements in Primates.” In Handbook of Phys-
iology, sect. 1, vol. 2, Motor Control, ed. V. B. Brooks. Bethesda: American Physiological
Society, 1981, 1083–1120.
Evarts, E. V., C. Fromm, J. Kroller, and V. A. Jennings. “Motor Cortex Control of Finely
Graded Forces.” Journal of Neurophysiology, 49 (1983): 1199–1215.
Ewert, J. P. “Neural Mechanisms of Prey-Catching and Avoidance Behavior in the Toad
(Bufo bufo L.).” Brain Behaviour and Evolution, 3 (1970): 36–56.
———. “Neuroethology of Releasing Mechanisms: Prey-Catching in Toads.” Behavioral
and Brain Sciences, 10 (1987): 337–405.
Faber, D. S., W. S. Young, P. Legendre, and H. Korn. “Intrinsic Quantal Variability Due to
Stochastic Properties of Receptor-Transmitter Interactions.” Science, 258 (1992): 1494–
1498.
Feldman, A. G., and M. F. Levin. “The Origin and Use of Positional Frames of Reference in
Motor Control.” Behavioral and Brain Sciences, 18 (1995): 723–744.
Ferrier, D. “Experiments on the Brain of Monkeys.” Philosophical Transactions of the Royal
Society of London, 165 (1875): 433–488.
———. The Functions of the Brain. New York: G. P. Putnam’s Sons, 1876.
Fessard, A., and A. Tournay. “Quelques données et réflexions sur le phénomène de la post-
contraction involontaire.” L’Année psychologique (1950): 216–235.
Findley, L. J., R. Capildeo, and A. Tremor, eds. Movement Disorders. London: Macmillan,
1988.
Flor, H., T. Elbert, S. Knecht, C. Wienbruch, C. Pantev, N. Birbaumer, W. Larbig, and E.
Baub. “Phantom-Limb Pain as a Perceptual Correlate of Cortical Reorganization fol-
lowing Arm Amputation.” Nature, 375 (1995): 482–484.
Flourens, P. Recherches expérimentales sur les propriétés et les fonctions du système nerveux dans
les animaux vertébrés. Paris: Crévot, 1824.
Fogassi, L., V. Galese, L. Fadiga, G. Luppino, M. Matelli, and G. Rizzolatti. “Coding of
Peripersonal Space in Inferior Premotor Cortex (Area F4).” Journal of Neurophysiology,
76 (1996): 141–157.
Fox, P. T., J. M. Fox, M. E. Raichle, and R. M. Burde. “The Role of the Cerebral Cortex in
the Generation of Voluntary Saccades.” Journal of Neurophysiology, 54 (1985): 348–369.
Friberg, L., T. S. Olsen, P. Roland, O. B. Paulson, and N. A. Lassen. “Focal Increase of
Blood Flow in the Cerebral Cortex of Man during Vestibular Stimulation.” Brain, 108
(1985): 609–623.
Frith, C. D., K. Friston, P. F. Liddle, and R. S. Frackowiak. “Willed Action and the
Prefrontal Cortex in Man: A Study with PET.” Proceedings of the Royal Society of Lon-
don. Series B: Biological Sciences, 244 (1991): 241–246.
Frith, U. Autism: Explaining the Enigma. Cognitive Development. Oxford: Basil Blackwell,
1989.
Fuchs, A., and W. Becker, eds. Progress in Oculomotor Research. Amsterdam: Elsevier, 1981.
Fukuda, T. Statokinetic Reflexes in Equilibrium and Movement. Tokyo: University of Tokyo
Press, 1983.
308 •
WORKS CITED
Funahashi, S., C. Bruck, and P. S. Goldman-Rakic. “Mnemonic Coding of Visual Space in
the Monkey’s Dorsolateral Prefrontal Cortex.” Journal of Neurophysiology, 61 (1989): 1–
19.
Fuster, J. M. The Prefrontal Cortex: Anatomy, Physiology, and Neuropsychology of the Frontal
Lobe. New York: Raven Press, 1996.
Gahery, Y., and J. Massion. “Coordination between Posture and Movement.” In The Motor
System in Neurobiology, ed. E. V. Evarts, S. P. Wise, and D. Bousfield. Amsterdam:
Elsevier Biomedical Press, 1985, 121–125.
Gavrilov, V. V., S. I. Wiener, and A. Berthoz. “Whole Body Rotations Enhance Hippo-
campal Theta Rhythm, Slow Activity in Awake Rats Passively Transported on a Mo-
bile Robot.” Annals of the New York Academy of Sciences, 781 (1996): 385–398.
Gaymard, B., C. Pierrot-Deseilligny, and S. Rivaud. “Impairment of Sequences of Mem-
ory-Guided Saccades after Supplementary Motor Area Lesions.” Annals of Neurology,
28 (1990): 622–626.
Gaymard, B., S. Rivaud, and C. Pierrot-Deseilligny. “Role of the Left and Right Supple-
mentary Motor Areas in Memory-Guided Saccade Sequences.” Annals of Neurology, 34
(1993): 404–406.
Georgopoulos, A. P. “Current Issues in Directional Motor Control.” Trends in Neurosciences,
18 (1995): 506–510.
Georgopoulos, A. P., M. D. Crutcher, and A. B. Schwartz. “Cognitive Spatial Motor Pro-
cesses. III. Motor Cortical Prediction of Movement Direction during an Instructed
Delay Period.” Experimental Brain Research, 75 (1989): 183–194.
Georgopoulos, A. P., J. F. Kalaska, R. Caminiti, and J. T. Massey. “On the Relations between
the Direction of Two-Dimensional Arm Movements and Cell Discharge in Primate
Motor Cortex.” Journal of Neuroscience, 2 (1982): 1527–1537.
Georgopoulos, A. P., J. T. Lurito, M. Petrides, A. B. Schwartz, and J. T. Massey. “Mental Ro-
tation of the Neuronal Population Vector.” Science, 243 (1989): 234–236.
Georgopoulos, A. P., A. B. Schwartz, and R. E. Kettner. “Neuronal Population Coding of
movement Direction.” Science, 233 (1986): 1416–1429.
Gertsmann, J. “Problems of Imperception of Disease and of Impaired Body Territories
with Organic Lesions. Relation to Body Scheme and Its Disorders.” Archives of Neurol-
ogy and Psychiatry, 48 (1942): 890–913.
Ghaem, O., E. Mellet, F. Grivello, N. Tzourio, B. Mazoyer, A. Berthoz, and M. Denis.
“Mental Navigation along Memorized Routes Activates the Hippocampus,
Precuneus, and Insula.” NeuroReport 8 (1997): 739–744.
Gibson, J. J. The Senses Considered as Perceptual Systems. Boston: Houghton, 1966.
———. “The Theory of Affordances.” In Perceiving, Acting, and Knowing, ed. R. E. Shaw
and J. Bransford. Hillsdale, N.J.: Erlbaum, 1977.
Gielen, S. “Muscle Activation Patterns and Joint-Angle Coordination in Multijoint Move-
ments.” In Multisensory Control of Movement, ed. A. Berthoz. New York: Oxford Uni-
versity Press, 293–312.
Gielen, C. C. A. M., and E. J. Van Zuylen. “Coordination of Arm Muscles during Flexion
and Supination: Applications of the Tensor Analysis Approach.” Neuroscience, 17
(1986): 527–539.
WORKS CITED •
309
Gladwin, T. East Is a Big Bird. Cambridge, Mass.: Harvard University Press, 1970.
Glasauer, S., and H. Mittelstaedt. “Determinants of Orientation in Microgravity.” Acta
Astronautica, 27 (1992): 1–9.
Goethe, Johann Wolfgang von. Faust, trans. Randall Jarrell. New York: Farrar, Straus, and
Giroux, 1959.
Goldman-Rakic, P. “Circuitry of Primate Prefrontal Cortex and Regulation of Behavior by
Representational Memory.” In Handbook of Physiology, vol. 5, The Nervous System, ed.
V. B. Mountcastle, F. Plkum, and D. Humphrey. Bethesda: American Physiological So-
ciety, 1987, 373–417.
Gomi, H., and M. Kawato. “Adaptive Feedback Control Models of the Vestibulocerebellum
and Spinocerebellum.” Biological Cybernetics, 68 (1992): 105–114.
Gonshor, A., and G. Melvill Jones. “Extreme Vestibulo-ocular Adaptation by Prolonged
Optical Reversal of Vision.” Journal of Physiology, 256 (1976): 381–414.
Goodale, M. A., A. D. Milner, L. S. Jakobson, and D. P. Carey. “Neurological Dissociation
between Perceiving Objects and Grasping Them.” Nature, 349 (1991): 154–156.
Graf, W., and R. Baker. “Adaptive Changes of the Vestibulo-Ocular Reflex in Flatfish Are
Achieved by Reorganization of Central Nervous Pathways.” Science, 221 (1983): 777–
779.
Grantyn, A. “How Visual Inputs to the Ponto-Bulbar Reticular Formation Are Used in the
Synthesis of Premotor Signals during Orienting.” In Progress in Brain Research, vol. 80,
ed. J. H. J. Allum and M. Hulliger. Amsterdam: Elsevier, 1989, 159–170.
Grantyn, A., and A. Berthoz. “The Role of the Tecto-reticulo-spinal System in Control of
Head Movement.” In Control of Head Movement, ed. G. W. Peterson and F. Richmond.
Oxford: Oxford University Press, 1987, 224–244.
Grantyn, A., and R. Grantyn. “Axonal Patterns and Sites of Termination of Cat Superior
Colliculus Neurons Projecting in the Tecto-bulbo-spinal Tract.” Experimental Brain Re-
search, 46 (1982): 243–256.
Grantyn, A., R. Grantyn, V. Robine, and A. Berthoz. “Electroanatomy of Tectal Efferent
Connections Related to Eye Movements in the Horizontal Plane.” Experimental Brain
Research, 37 (1979): 149–172.
Grasso, R., C. Assaiante, P. Prévost, and A. Berthoz. “Development of Anticipatory Ori-
enting Strategies during Locomotor Tasks in Children,” Neuroscience and Biobehavioral
Review, 22 (1998): 533–539.
Grasso, R., S. Glasauer, U. Takei, and A. Berthoz. “The Predictive Brain: Anticipatory Con-
trol of Head Direction for the Steering of Locomotion.” NeuroReport, 7 (1996): 1170–
1174.
Grasso, R., P. Prévost, Y. Ivanenko, and A. Berthoz. “Eye-Head Coordination for the
Steering of Locomotion in Humans: An Anticipatory Synergy.” Neuroscience Letters,
253 (1998): 115–118.
Gray, C. M., P. König, A. K. Engel, and W. Singer. “Oscillatory Responses in Cat Visual
Cortex Exhibit Inter-Columnar Synchronization Which Reflects Global Stimulus Pat-
terns.” Nature, 338 (1989): 334–337.
Graziano, M., G. Yap, and C. Gross. “Coding of Visual Space by Premotor Neurons.” Sci-
ence, 266 (1994): 1054–1057.
310 •
WORKS CITED
———. “The Representation of Extrapersonal Space: A Possible Role for Bimodel-Tactile
Neurons.” In The Cognitive Neurosciences, ed. M. Gazzaniga. Cambridge, Mass.: MIT
Press, 1994.
Grüsser, O. J. “Cortical Representation of Head Movement in Space and Some Psycho-
physical Considerations.” In The Head-Neck Sensory-Motor System, ed. A. Berthoz, P. O.
Vidal, and W. Graf. New York: Oxford University Press, 1991.
Grüsser, O. J., and T. Landis. Visual Agnosias and Related Disorders, vol. 12. Vision and Visual
Dysfunction. Basingstoke, UK: Macmillan, 1991.
Grüsser, O. J., M. Pause, and U. Schreiter. “Localisation and Responses of Neurones in the
Parieto-Insular Vestibular Cortex of Awake Monkeys (Macaca fascicularis).” Journal of
Physiology, 430 (1990): 537–557.
———. “Vestibular Neurones in the Parieto-Insular Cortex of Monkeys (Macaca
fascicularis).” Journal of Physiology, 430 (1990): 559–583.
Guitton, G., H. A. Buchtel, and R. Douglas. “Frontal Lobe Lesions in Man Cause Dif-
ficulties in Supressing Reflexive Glance and in Generating Goal-Directed Saccades.”
Guldin, W. O., and O. J. Grüsser. “Single Unit Responses in the Vestibular Cortex of Squir-
rel Monkeys.” Neuroscience Abstracts, 13 (1987): 1224.
Gurfinkel, V. S. “The Mechanisms of Postural Regulation in Man.” In Physiology and General
Biology Reviews, ed. T. Turpaev. Chur, Switzerland: Harwood Academic Publishers,
1994.
Gurfinkel, V. S., E. E. Debreva, and Y. S. Levik. “The Role of the Internal Model in the Per-
ception of Position and Planning of Hand Movements.” Fiziologiia Cheloveka, 12
(1986): 769–776.
Gurfinkel, V. S., and Y. S. Levick. “Perceptual and Automatic Aspects of the Postural Body
Scheme.” In Brain and Space, vol. 12, ed. J. Paillard. Oxford: Oxford University Press,
1986.
Hamann, K. F., P. P. Vidal, J. M. Sterkers, and A. Berthoz. “A New Test for Postural Disor-
ders: An Application of Visual Stabilization.” Agressologie, 20B (1979): 129–130.
Hanneton, S., A. Berthoz, J. Droulez, and J.-J. Slotine. “Does the Brain Use Sliding Vari-
ables for the Control of Movements?” Biological Cybernetics, 77 (1997): 381–393.
Haustein, W., and H. Mittelstaedt. “Evaluation of Retinal Orientation and Gaze Direction
in the Perception of the Vertical.” Vision Research, 30 (1990): 255–262.
Head, H., and G. Holmes. “Sensory Disturbances from Cerebral Lesions.” Brain, 34 (1911):
102–244.
Hécaen, H., and J. de Ajuriaguerra. Méconnaissances et hallucinations corporelles: intégration et
désintégration de la somatognosie. Paris: Masson et Cie, 1952.
Held, R., and A. Hein. “Movement-Produced Stimulation in the Development of Vi-
sually Guided Behavior.” Journal of Comparative Physiological Psychology, 56 (1963): 872–
876.
Heliodorus of Emesa. The Aethiopica, vol. 5. The Athenian Society’s Publications. Athens:
Privately printed for the Athenian Society, 1897.
Helmholtz, H. von. Treatise on Physiological Optics, trans. J. P. C. Sothall. New York: Dover,
1962.
WORKS CITED •
311
Henn, V., L. R. Young, and C. Finley. “Vestibular Nucleus Units in Alert Monkeys Are Also
Influenced by Missing Visual Field.” Brain Research, 71 (1974): 144–149.
Hertz, H. The Principles of Mechanics, trans. D. E. Jones and J. T. Walley. New York: Dover,
1956.
Hess, W. “Teleokinetisches und ereismatisches Kraftsystem in der Biomotorik.” Helvetica
Physiologica et Pharmacologica Acta, 1 (1943): C62–C63.
Hietanen, J. K., and D. Perrett. “A Role of Expectation in Visual and Tactile Processing
within the Temporal Cortex.” In Brain Mechanisms in Perception and Memory: From Neu-
ron to Behavior, ed. T. E. Ono, L. R. Squire, M. Raichle, D. L. Perrett, and M. Fukuda.
Oxford: Oxford University Press, 1993, 83–103.
Hikosaka, O. “Role of the Basal Ganglia in Motor Learning: A Hypothesis.” In Brain Mech-
anisms in Perception and Memory: From Neuron to Behavior, ed. T. E. Ono, L. R. Squire,
M. Raichle, D. L. Perrett, and M. Fukuda. Oxford: Oxford University Press, 1993, 497–
513.
Hikosaka, O., M. Sakamoto, and N. Miyashita. “Effects of Caudate Nucleus Stimulation on
Substantia Nigra Cell Activity in Monkey.” Experimental Brain Research, 95 (1993): 457–
472.
Hikosaka, O., and R. H. Wurtz. “Visual and Oculomotor Functions of Monkey Substantia
Nigra Pars Reticulata. III. Memory-Contingent Visual and Saccade Responses.” Jour-
Hoffman, K. P. “Responses of Single Neurons in the Pretectum of Monkeys to Visual Stim-
uli in Three Dimensional Space.” In Representation of Three-Dimensional Space in the
Vestibular, Oculomotor, and Visual Systems, ed. B. Cohen and V. Henn. New York: New
York Academy of Sciences, 1988, 1–261.
Holst, E. von. “Relations between the Central Nervous System and the Peripheral Or-
gans.” Journal of Animal Behaviour, 2 (1954): 89–94.
Holst, E. von, and H. Mittelstaedt. “Das Reafferenzprinzip. Wechselwirkungen zwischen
Zentralnervensystem und Peripherie.” Naturwissenschaften, 37 (1950): 464–476.
Humphrey, D. R., and H. J. Freund, eds. Motor Control: Concepts and Issues. New York:
Wiley, 1991.
Husserl, E. Idées directrices pour une phénoménologie et une philosophie phénoménologique pure,
vol. 2, Recherches phénoménologiques pour la constitution. Paris: Presses Universitaires de
France, 1982.
Hyvarinen, J. The Parietal Cortex of Monkey and Man. Berlin: Springer, 1982.
Israël, I. “Memory-Guided Saccades: What Is Memorized?” Experimental Brain Research, 90
(1992): 221–224.
Israël, I., and A. Berthoz. “Contribution of the Otoliths to the Calculation of Linear Dis-
placement.” Journal of Neurophysiology, 62 (1989): 247–263.
Israël, I., N. Chapuis, S. Glasauer, O. Charade, and A. Berthoz. “Estimation of Passive Lin-
ear Whole Body Displacement in Humans.” Journal of Neurophysiology, 70 (1993):
1270–1273.
Israël, I., R. Grasso, P. Georges-François, T. Tzuzuku, and A. Berthoz. “Spatial Memory
and Path Integration Studied by Self-Driven Linear Displacement.” Journal of Neuro-
physiology, 77 (1999): 3180–3192.
Israël, I., S. Rivaud, A. Berthoz, and C. Pierrot-Deseilligny. “Cortical Control of Vestibular
312 •
WORKS CITED
Memory-Guided Saccades.” Annals of the New York Academy of Sciences, 656 (1992):
472–484.
Israël, I., S. Rivaud, B. Gaymard, A. Berthoz, and C. Pierrot-Deseilligny. “Cortical Control
of Vestibular-Guided Saccades in Man.” Brain, 118 (1995): 1169–1183.
Israël, I., S. Rivaud, C. Pierrot-Desilligny, and A. Berthoz. “Delayed VOR: An Assessment
of Vestibular Memory for Self Motion.” In Tutorials in Motor Neuroscience, ed. J.
Requin and J. Stelmach. Dordrecht: Kluwer, 1991, 599–607.
Ito, M. “Cerebellar Control of the Vestibulo-ocular Reflex around the Flocculus Hypothe-
sis.” Annual Review of Neuroscience, 5 (1982): 275–296.
———. The Cerebellum and Neuronal Control. New York: Raven, 1984.
Ivanenko, Y., R. Grasso, I. Israël, and A. Berthoz. “The Contribution of Otoliths and Semi-
circular Canals to the Perception of Two-Dimensional Passive Whole-Body Motion in
Humans.” Journal of Physiology, 502 (1997): 223–233.
James, William. The Principles of Psychology, vol. 2. New York: Holt, 1890.
Jami, L. “Golgi Tendon Organs in Mammalian Skeletal Muscle: Functional Properties on
Central Actions.” Physiological Reviews, 72 (1992): 623–666.
Janet, P. Les Débuts de l’intelligence. Paris: Flammarion, 1935.
Jeannerod, M. “The Posterior Parietal Cortex as a Spatial Generator.” In Brain Mechanisms
and Spatial Vision, eds. D. I. Ingle, J. Jeannerod, and D. Lee. Dordrecht: Martinus
Nijhoff, 1985, 279–298.
———. Neurophysiological and Neuropsychological Aspects of Spatial Neglect. Amsterdam:
Elsevier, 1987.
———. The Neural and Behavioral Organisation of Goal-Directed Arm Movements. Oxford:
Clarendon Press, 1988.
———, ed. Attention and Performance XIII. Hillsdale, N.J.: Erlbaum, 1990.
———. “A Neurophysiological Model for the Directional Coding of Reaching Move-
ments.” In Brain and Space, ed. J. Paillard. Oxford: Oxford University Press, 1991.
———. “The Representing Brain: Neural Correlates of Motor Intention and Imagery.” Be-
havioral and Brain Sciences, 17 (1994): 187–202.
Joliot, M., U. Ribary, and R. Llinás. “Human Oscillatory Brain Activity near 40 Hz Coexists
with Cognitive Temporal Binding.” Proceedings of the National Academy of Sciences, 91
(1994): 11748–11751.
Jones, M. G., and K. E. Spells. “A Theoretical and Comparative Study of the Functional De-
pendance of the Semi-Circular Canal upon its Physical Dimensions.” Proceedings of the
Royal Society of London. Series B: Biological Sciences, 157 (1963): 403–419.
Joseph, J. P., and P. Barone. “Prefrontal Unit Activity during a Delayed Oculomotor Task in
the Monkey.” Experimental Brain Research, 67 (1987): 460–468.
Kalaska, J. F., R. Caminiti, and A. P. Georgopoulos. “Cortical Mechanisms Related to the
Direction of Two-Dimensional Arm Movements: Relations in Parietal Area 5 and
Comparison with Motor Cortex.” Experimental Brain Research, 51 (1983): 247–260.
Kant, Immanuel. Critique of Pure Reason, trans. Norman Kemp Smith. New York: St. Mar-
tin’s Press, 1961.
Karabelas, A. B., and A. K. Moschovakis. “Nigral Inhibitory Termination on Efferent Neu-
rons of the Superior Colliculus. An Intracellular Horseradish Peroxidase Study in the
Cat.” Journal of Cognitive Neuroscience, 239 (1985): 309–329.
WORKS CITED •
313
Karnath, H. O., K. Christ, and W. Hartje. “Decrease of Contralateral Neglect by Neck
Muscle Vibration and Spatial Orientation of Trunk Midline.” Brain, 116 (1993): 383–
396.
Kawano, K., M. Sasaki, and M. Yamashita. “Vestibular Input to Visual Tracking Neurons in
the Posterior Parietal Association Cortex of the Monkey.” Neuroscience Letters, 17
(1980): 55–60.
Kawato, M., K. Furukawa, and R. Suzuki. “A Hierarchical Neural Network Model for Con-
trol and Learning of Voluntary Movements.” Biological Cybernetiks, 57 (1987): 169–185.
Kelso, J. A. S. “Phase Transitions and Critical Behaviour in Human Bimanual Coordina-
tion.” American Journal of Physiology, 14 (1984): R1000–1004.
Kelso, J. A. S., J. D. Delcolle, and G. Schöner. “Action-Perception as a Pattern Formation
Process.” In Attention and Performance, ed. M. Jeannerod. Hillsdale, N.J.: Erlbaum,
1990, 139–169.
Kesner, R. P. “Paired Associate Learning in the Rat: Role of Hippocampus, Medial Pre-
frontal Cortex, and Parietal Cortex.” Psychobiology, 21 (1993): 183–192.
Korn, H., and D. S. Faber. “Organisation and Cellular Mechanisms Underlying Chemical
Inhibition in a Vertebrate Neuron.” In Molecular and Cellular Interactions Underlying
Higher Brain Functions. Progress in Brain Research, ed. J.-P. Changeux. Amsterdam:
North Holland, 1983.
Kosslyn, S. M., C. F. Chabris, C. J. Marsolek, R. A. Jacobs, and O. Koenig. “On Computa-
tional Evidence for Different Types of Spatial Relations Encoding: Reply to Cook et
al.” Journal of Experimental Psychology: Human Perception and Performance, 21 (1995):
423–431.
Kosslyn, S. M., C. F. Chabris, C. J. Marsolek, and O. Koenig. “Categorical versus Coordi-
nate Spatial Representations: Computational Analysis and Computer Simulations.”
Journal of Experimental Psychology: Human Perception and Performance, 18 (1992): 562–
577.
Krechevsky, I. “Hypothesis in Rats.” Psychological Reviews, 39 (1932): 517–532.
Lackner, J. R. “Some Contributions of Touch, Pressure, and Kinesthesis to Human Spatial
Orientation and Oculomotor Control.” Acta Astronautica, 8 (1981): 825–830.
———. “Some Proprioceptive Influences on the Perceptual Representation of Body Shape
and Orientation.” Brain, 111 (1981): 281–297.
———. “Orientation and Movement in Unusual Force Environments.” Psychological Sci-
ence, 4 (1993): 134–142.
Lackner, J. R., and M. S. Levine. “Changes in Apparent Body Orientation and Sensory Lo-
calization Induced by Vibration of Postural Muscles: Vibratory Myesthetic Illusions.”
Aviation, Space, and Environmental Medicine, 50 (1979): 346–354.
Lacquaniti, F., M. Carrozzo, and N. Borghese. “The Role of Vision in Tuning Anticipatory
Motor Responses of the Limbs.” In Multisensory Control of Movement, ed. A. Berthoz.
New York: Oxford University Press, 1993, 379–390.
Lacquaniti, F., E. Guigon, L. Bianchi, S. Farraina, and R. Caminiti. “Representing Spatial In-
formation for Limb Movements: Role of Area 5 in the Monkey.” Cerebral Cortex, 5
(1995): 391–409.
Lacquaniti, F., and C. Maioli. “The Role of Preparation in Tuning Anticipatory and Reflex
Responses during Catching.” Journal of Neuroscience, 9 (1989): 134–148.
314 •
WORKS CITED
Lacquaniti, F., J. Soechting, and C. Terzuolo. “Path Constraints on Point to Point Arm
Movements in Three-Dimensional Space.” Neuroscience, 17 (1986): 313–324.
Land, M. F., and D. N. Lee. “Where We Look When We Steer.” Nature, 369 (1994): 742–
744.
Lang, W., D. Cheyne, R. Kristeva, R. Beisteiner, G. Lindinger, and L. Deecke. “Three-Di-
mensional Localization of SMA Activity Preceding Voluntary Movement. A Study of
Electric and Magnetic Fields in a Patient with Infarction of the Right Supplementary
Motor Area.” Experimental Brain Research, 87 (1991): 688–695.
Lang, W., M. Lang, F. Uhl, C. Koska, A. Kornhuber, and L. Deecke. “Negative Cortical
DC Shifts Preceding and Accompanying Simultaneous and Sequential Finger Move-
ments.” Experimental Brain Research, 71 (1988): 579–587.
Lang, W., H. Obrig, G. Lindinger, D. Cheyne, and L. Deecke. “Supplementary Motor Area
Activation While Tapping Bimanually Different Rhythms in Musicians.” Experimental
Brain Research, 79 (1990): 504–514.
Lang, W., L. Petit, P. Hölliger, U. Pietrzyck, N. Tzourio, B. Mazoyer, and A. Berthoz.
“A Positron Emission Tomography Study of Oculomotor Imagery.” NeuroReport, 5
(1994): 921–924.
Laporte, Y., F. Emonet-Dénand, and L. Jami. “The Skeletofusimotor or β Innervation of
Mammalian Muscle Spindles.” Trends in Neuroscience, 4 (1981): 97–99.
Lashley, K. S. Brain Mechanisms and Intelligence. Chicago: University of Chicago Press, 1929.
Lee, D. N. “A Theory of Visual Control of Braking Based on Information about Time-to-
Collision.” Perception, 5 (1976): 437–459.
Lee, D. N., and E. Aronson. “Visual Proprioceptive Control of Standing in Human In-
fants.” Perception and Psychophysics, 14 (1974): 529–532.
Lee, D. N., and J. R. Lishman. “Visual Proprioceptive Control of Stance.” Journal of Human
Movement Studies, 1 (1975): 87–95.
Lee, D. N., and P. E. Reddish. “Plummeting Gannets: A Paradigm of Ecological Optics.”
Nature, 293 (1985): 293–294.
Lee, D. N., and D. S. Young. “Visual Timing in Interceptive Actions.” In Brain Mechanisms
and Spatial Vision, ed. D. J. Ingle, M. Jeannerod, and D. N. Lee. Dordrecht: Martinus
Nijhoff, 1996, 1–30.
Lee, R. G., I. Tonolli, F. Viallet, R. Aurenty, and J. Massion. “Preparatory Postural Adjust-
ment in Parkinsonian Patients with Postural Instability.” Canadian Journal of Neurologi-
cal Sciences, 22 (1995): 126–135.
Lestienne, F., J. Soechting, and A. Berthoz. “Postural Readjustments Induced by Linear
Motion of Visual Scenes.” Experimental Brain Research, 28 (1977): 363–384.
Levin, M. F., and A. G. Feldman. “The Role of Stretch Reflex Threshold Regulation in Nor-
mal and Impaired Motor Control.” Brain Research, 657 (1994): 23–30.
Lisman, J. E., and M. A. P. Idiart. “Storage of Short-Term Memories in Oscillatory Sub-
cycles.” Science, 267 (1995): 1512–1515.
Livingstone, M., and D. Hubel. “Segregation of Form, Color, Movement, and Depth: Anat-
omy, Physiology, and Perception.” Science, 240 (1988): 740–749.
Llinás, R. “Possible Role of Tremor in the Organisation of the Nervous System.” In Move-
ment Disorders, ed. L. J. Findley, R. Capildeo, and A. Tremor. London: Macmillan,
1988, 475–477.
WORKS CITED •
315
———. “The Intrinsic Electrophysiological Properties of Mammalian Neurons: Insights
into Central Nervous System Function.” Science, 242 (1988): 1654–1664.
———. “The Noncontinuous Nature of Movement Execution.” In Motor Control: Concepts
and Issues. Chichester: Wiley, 1991, 223–242.
———. “Oscillations in CNS Neurons: A Possible Role for Cortical Interneurons in the
Generation of 40-Hz Oscillations.” In Induced Rhythms in the Brain, ed. E. Basar and T.
Bullock. Boston: Birkhauser, 1992, 269–283.
Llinás, R., and U. Ribary. “Coherent 40-Hz Oscillation Characterizes Dream State in Hu-
mans.” Proceedings of the National Academy of Sciences, 90 (1993): 2078–2081.
Llinás, R., and C. Sotelo. The Cerebellum Revisited. New York: Springer, 1992.
Lobel, E., J. F. Kleine, D. Le Bihan, A. Leroy-Willig, and A. Berthoz. “Functional MRI of
Galvanic Vestibular Stimulation.” Journal of Neurophysiology, 80 (1998): 2699–2709.
Lorente de Nó, R. “Vestibulo-ocular Reflex Arc.” Archives of Neurology and Psychiatry, 30
(1933): 245–291.
Lorenz, Konrad. Behind the Mirror: A Search for a Natural History of Human Knowledge, trans.
Ronald Taylor. New York: Harcourt Brace Jovanovich, 1977.
Lotze, R. H. Medizinische Psychologie oder Physiologie der Seele. Leipzig: Weidemann, 1852.
Mach, E. Grundlinien der Lehre von den Bewegungsempfindungen. Amsterdam: Bonset, 1967.
MacKay, D. G. The Organization of Perception and Action. New York: Springer, 1987.
Magnin, M., M. Jeannerod, and P. T. S. Putkonen. “Vestibular and Saccadic Influences on
Dorsal and Ventral Nuclei of the Lateral Geniculate Body.” Experimental Brain Re-
search, 21 (1974): 1–18.
Malsburg, C. von der. “Nervous Structures with Dynamical Links.” Berichte der
Bunsengesellschaft für physikalische Chemie, 89 (1985): 703–710.
Marey, E. J. Le Mouvement. Paris: Masson, 1894.
———. La Machine animale. Paris: Revue EPS, 1993.
Marr, D. “Simple Memory: A Theory for Archicortex.” Proceedings of the Royal Society of
London. Series B: Biological Sciences, 262 (1971): 23–81.
Massion, J. “Postural Control System.” Current Opinion in Neurobiology, 4 (1994): 877–887.
Massion, J., V. Gurfinkel, M. Lipshits, A. Obadia, and K. Popov. “Stratégie et synergie: deux
niveaux de contrôle de l’équilibre au cours du mouvement. Effets de la microgravité.”
Comptes rendus hebdomadaires de l’Académie des sciences (Paris, Série III), 314 (1992): 87–
92.
Massion, J., and L. Rispal-Padel. “Thalamus: fonctions motrices.” Revue Neurologique, 142
(1986): 327–336.
Massion, J., F. Viallet, R. Massarino, and R. Khalil. “La région de l’aire motrice supplé-
mentaire est impliquée dans la coordination entre posture et mouvement.” Comptes
rendus hebdomadaires de l’Académie des sciences (Paris, Série III), 308 (1989): 417–423.
Mast, F., S. Kosslyn, and A. Berthoz. “Visual Mental Imagery Interferes with Allocentric
Orientation Judgements.” Neuroreport, 110 (1999): 3549–3553.
Matthews, B. L., J. H. Ryu, and C. Bockaneck. “Vestibular Contribution to Spatial Orienta-
tion.” Acta Orolaryngologica, 468 (1989): 149–154.
Mays, L. E., and D. L. Sparks. “Saccades Are Spatially, Not Retinocentrically, Coded.” Sci-
ence, 208 (1980): 1163–1165.
316 •
WORKS CITED
McCrea, R. A., K. Yoshida, C. Evinger, and A. Berthoz. “The Location, Axonal Arbori-
zation, and Termination Sites of Eye-Movement-Related Secondary Vestibular Neu-
rons Demonstrated by Intra-axonal HRP Injection in the Alert Cat.” In Progress in
Oculomotor Research, ed. A. Fuchs and W. Becker. Amsterdam: Elsevier, 1981, 379–386.
McDaniel, W. F., D. M. Compton, and S. R. Smith. “Spatial Learning following Posterior
Parietal or Hippocampal Lesions.” NeuroReport, 5 (1994): 1713–1717.
McIntyre, J., E. Gurfinkel, M. Lipshits, J. Droulez, and V. Gurfinkel. “Measurement of Hu-
man Force Control during Constrained Arm Motion using a Force-Activated Joy-
stick.” Journal of Neurophysiology, 17 (1995): 1201–1222.
Melamed, E., and B. Larsen. “Cortical Activation during Saccadic Eye Movements in Hu-
mans: Localisation by Cerebral Blood Flow Increases.” Annals of Neurology, 5 (1989):
79–88.
Melvill-Jones, G., A. Berthoz, and B. Segal. “Adaptative Modification of the Vestibulo-ocu-
lar Reflex by Mental Effort in Darkness.” Experimental Brain Research, 56 (1984): 149–
153.
Melvill-Jones, G., D. Guitton, and A. Berthoz. “Changing Patterns of Eye-Head Coordina-
tion during Six Hours of Optically Reversed Vision.” Experimental Brain Research, 69
(1988): 531–544.
Merleau-Ponty, M. Le Visible et l’Invisible. Paris: Gallimard, 1964.
———. The Visible and the Invisible, ed. Claude Lefort, trans. Alphonso Lingis. Northwest-
ern University Studies in Phenomenology and Existential Philosophy. Evanston, Ill.:
Northwestern University Press, 1968.
———. Résumés de cours au Collège de France. Paris: Gallimard, 1968.
———. The Essential Writings of Merleau-Ponty, ed. Alden L. Fisher. New York: Harcourt,
Brace, and World, 1969.
———. La Nature: notes-cours du Collège de France. Paris: Le Seuil, 1995.
Mesulam, M. M. “A Cortical Network for Directed Attention and Unilateral Neglect.” An-
nals of Neurology, 10 (1981): 309–325.
Metcalfe, T., and M. Gresty. “Self-controlled Reorienting Movements in Response to Rota-
tional Displacements in Normal Subjects and Patients with Labyrinthine Diseases.” In
Sensing Motion, ed. D. L. Tomko, B. Cohen, and F. E. Guedry. New York: Annals of
the New York Academy of Sciences, 1992.
Michotte, A. Causalité, permanence et réalité phénoménales. Paris: Béatrice-Nauwelaerts,
1962.M
Miles, F. A., and S. G. Lisberger. “Plasticity in the Vestibulo-ocular Reflex: A New Hypothe-
sis.” Annual Review of Neuroscience, 4 (1981): 273–299.
Miles, R., K. Toth, A. Gulyas, N. Hajos, and T. F. Freund. “Differences between Somatic
and Dendritic Inhibition in the Hippocampus.” Neuron, 16 (1996): 814–823.
Miller, S., M. Potegal, and L. Abraham. “Vestibular Involvement in a Passive Transport and
Return Task.” Physiological Psychology, 11 (1982): 1–10.
Milner, B. “Visual Recognition and Recall after Right Temporal Lobe Excision in Man.”
Neuropsychologia, 6 (1968): 191–209.
Mishkin, M. “A Memory System in the Monkey.” Philosophical Transactions of the Royal Soci-
ety. Series B: Biological Sciences, 98 (1982): 85–95.
WORKS CITED •
317
Mistlin, A. J., and D. I. Perrett. “Visual and Somatosensory Processing in the Macaque
Temporal Cortex: The Role of ‘Expectation.’” Experimental Brain Research, 82 (1990):
437–450.
Mittelstaedt, H. “The Role of Otoliths in the Perception of the Orientation of Self and
World to the Vertical.” Zoologische Jahrbücher der Physiologie, 95 (1991): 419–425.
———. “Evidence of Somatic Graviception from New and Classical Investigations.” Acta
Otolaryngologica, 115 (suppl. 520) (1995): 186–187.
———. “Somatic Graviception.” Biological Psychology, 42 (1996): 53–74.
Mittelstaedt, H., and S. Glasauer. “Illusions of Verticality in Weightlessness.” Clinical Inves-
tigations, 71 (1993): 732–739.
Mittelstaedt, H., and M. L. Mittelstaedt. “Mechanismen der Orientierung ohne richtende
Aussensreise.” Fortschritte der Zoologie, 21 (1973): 46–58.
———. “Homing by Path Integration in a Mammal.” Naturwissenschaften, 67 (1992): 566–
567.
Mogensen, J., T. K. Pedersen, S. Holm, and L. E. Bang. “Prefrontal Cortical Mediation of
Rats’ Place Learning in a Modified Water Maze.” Brain Research Bulletin, 38 (1995):
425–434.
Moschovakis, A. B., A. B. Karabelas, and S. Highstein. “Structure Function Relationship in
the Primate Superior Colliculus. I. Morphological Classification of Efferent Neu-
rons.” Journal of Neurophysiology, 60 (1988): 232–262.
———. “Structure Function Relationship in the Primate Superior Colliculus. II. Morpho-
logical Identity of Presaccadic Neurons.” Journal of Neurophysiology, 60 (1988): 263–
302.
Mouchnino, L., R. Aurenty, J. Massion, and A. Pedotti. “Is the Trunk a Reference Frame for
Calculating Leg Position?” NeuroReport, 4 (1993): 125–127.
Mountcastle, V. H., J. C. Lynch, A. Georgopoulos, H. Sakata, and C. Acuna. “Posterior Pa-
rietal Association Cortex of the Monkey: Command Functions for Operations within
Extrapersonal Space.” Journal of Neurophysiology, 38 (1975): 871–908.
Müller, J. “Über das Aubert’sche Phänomen.” Zeitschrift für Psychologie der Sinnesorgane, 49
(1916): 109–249.
Munoz, D. P., and R. H. Wurtz. “Fixation Cells in Monkey Superior Colliculus. I. Charac-
teristics of Cell Discharge.” Journal of Neurophysiology, 70 (1993): 559–575.
Musil, S. Y., C. R. Olson, and M. E. Goldberg. “Visual and Oculomotor Properties of Single
Neurons in Posterior Cingulate Cortex of Rhesus Monkey.” Society of Neuroscience Ab-
stracts, 16 (1990): 1221.
Muybridge, E. The Human Figure in Motion. New York: Dover, 1957.
Myashita, Y. “Where Visual Perception Meets Memory.” Annual Review of Neuroscience, 16
(1993): 245–263.
Nashner, L. M., and A. Berthoz. “Visual Contribution to Rapid Motor Responses during
Postural Control.” Brain Research, 150 (1978): 403–407.
Neisser, U. Cognition and Reality. San Francisco: W. H. Freeman, 1976.
Nudo, R. J., G. W. Milliken, W. M. Jenkins, and M. M. Merzenich. “Use-Dependent Alter-
ations of Movement Representations in Primary Motor Cortex of Adult Squirrel
Monkeys.” Journal of Neuroscience, 16 (1996): 785–807.
318 •
WORKS CITED
O’Keefe, J., and L. Nadel. The Hippocampus as a Cognitive Map. Oxford: Oxford University
Press, 1978.
O’Keefe, J., and M. Recce. “Phase Relationship between Hippocampal Place Units and the
EEG Theta Rhythm.” Hippocampus, 3 (1993): 317–330.
Olivier, E., A. Grantyn, M. Chat, and A. Berthoz. “The Control of Slow Orienting Eye
Movements by Tectoreticulospinal Neurons in the Cat: Behavior, Discharge Patterns,
and Underlying Connections.” Experimental Brain Research, 93 (1993): 435–449.
O’Mara, S. M., E. Rolls, A. Berthoz, and R. P. Kesner. “Neurons Responding to Whole
Body Motion in the Primate Hippocampus.” Journal of Neuroscience, 14 (1994): 6511–
6523.
Ono, T., B. L. McNaughton, S. Molotchnidoff, E. T. Rolls, and H. Nishigo. Perception, Mem-
ory, and Emotion: Frontiers in Neuroscience. Cambridge: Pergamon, 1996.
Ono, T., L. R. Squire, M. Raichle, D. I. Perrett, and M. Fukuda, eds. Brain Mechanisms of
Perception and Memory: From Neuron to Behavior. Oxford: Oxford University Press, 1993.
Orban, G. A., P. Dupont, B. De Bruyn, R. Vogels, R. Vanderberghe, and L. Mortelmans. “A
Motion Area in Human Visual Cortex.” Proceedings of the National Academy of Sciences,
92 (1995): 993–997.
Paillard, J. “Les déterminants moteurs de l’organisation spatial.” Cahiers de psychologie, 14
(1971): 261–316.
———. “The Multichanneling of Visual Cues and the Organisation of a Visually Guided
Response.” In Tutorials in Motor Behavior, ed. G. E. Stelmach and J. Requin. Amster-
dam: North Holland, 1980, 259–279.
———. “Le corps et ses langages d’espace.” In Le Corps en psychiatrie, ed. E. Jeddi. Paris:
Masson, 1982, 53–69.
———. “Posture and Locomotion: Old Problems and New Concepts. Foreword.” In Pos-
ture and Gait: Development, Adaptation, and Modulation, ed. B. Amblard, A. Berthoz,
and F. Clarac. Amsterdam: Elsevier, 1988, 5–12.
———. Brain and Space. Oxford: Oxford University Press, 1991.
———. “L’intégration sensori-motrice et idéo-motrice.” In Traité de psychologie
expérimentale, ed. M. Richelle, J. Requin, and M. Robert. Paris: Presses Universitaires
de France, 1994, 925–961.
Park, S., and P. S. Holzman. “Association of Working Memory Deficit and Eye Tracking
Dysfunction in Schizophrenia.” Schizophrenia Research, 11 (1993): 55–61.
Paulignan, Y., M. Dufosse, M. Hugon, and J. Massion. “Acquisition of Coordination be-
tween Posture and Movement in a Bimanual Task.” Experimental Brain Research, 77
(1989): 337–348.
Pavard, B., and A. Berthoz. “Perception du mouvement et orientation spatial (Revue
bibliographique).” Le Travail humain, 2 (1976): 207–226.
———. “Linear Acceleration Modifies the Perceived Velocity of a Moving Scene.” Percep-
tion and Psychophysics, 6 (1977): 529–540.
Pavlov, I. P. Les Réflexes conditionnés. Paris: Presses Universitaires de France, 1927.
Pedotti, A., P. Crenna, A. Deat, C. Frigo, and J. Massion. “Postural Synergies in Axial Move-
ments: Short and Long-Term Adaptation.” Experimental Brain Research, 74 (1989): 3–
10.
WORKS CITED •
319
Pellionisz, A. J. “Coordination: A Vector-Matrix Description of Transformations of Over-
complete CNS Coordinates and a Tensorial Solution using the Moore-Penrose Gener-
alized Inverse.” Journal of Theoretical Biology, 101 (1984): 353–375.
Pellionisz, A. J., and R. Llinás. “Brain Modelling by Tensor Network Theory and Com-
puter Simulation. The Cerebellum: Distributed Processor for Predictive Coordina-
tion.” Neuroscience, 4 (1979): 323–348.
———. “Tensorial Approach to the Geometry of Brain Function. Cerebellar Coordination
via a Metric Tensor.” Neuroscience, 5 (1980): 1761–1770.
———. “Space-Time Representation in the Brain. The Cerebellum as a Predictive Space-
Time Metric Tensor.” Neuroscience, 7 (1982): 2949–2970.
Penfield, W. “Vestibular Sensation and the Cerebral Cortex.” Annales d’otorhinolryngologie,
66 (1957): 691–698.
Penfield, W., and E. Boldrey. “Somatic Motor and Sensory Representation in the Cerebral
Cortex of Man as Studied by Electrical Stimulation.” Brain, 60 (1937): 389–443.
Penfield, W., and T. Rasmussen. The Cerebral Cortex of Man: A Clinical Study of Localization
of Function. New York: Macmillan, 1957.
Penfield, W., and L. Roberts. Langage et Mécanismes cérébraux. Paris: Presses Universitaires
de France, 1963.
Peper, L., R. J. Bootsma, D. R. Mestre, and F. C. Bakker. “Catching Balls: How to Get the
Hand to the Right Place at the Right Time.” Journal of Experimental Psychology (Human
Perception and Performance), 20 (1994): 591–612.
Perret, C., and A. Berthoz. “Evidence on Static and Dynamic Fusimotor Actions on the
Spindle Response to Sinusoidal Stretch during Locomotor Activity in the Cat.” Experi-
mental Brain Research, 18 (1973): 178–188.
Perrett, D. I., J. K. Hietanen, M. W. Oram, and P. J. Benson. “Organisation and Function of
Cells Responsive to Faces in the Temporal Cortex.” Philosophical Transactions of the
Royal Society. Series B: Biological Sciences, 335 (1992): 23–30.
Perrett, D., A. J. Mistlin, A. J. Chitty, P. A. J. Smith, D. D. Potter, R. Broennimann, and
M. H. Harries. “Specialised Face Processing and Hemispheric Asymmetry in Man and
Monkey: Evidence from Single Unit and Reaction Time Studies.” Behavioural Brain Re-
search, 29 (1988): 245–258.
Perrett, D. I., E. T. Rolls, and W. Caan. “Visual Neurons Responsive to Faces in the Monkey
Temporal Cortex.” Experimental Brain Research, 47 (1982): 342.
Peterson, B. W., J. Baker, and A. J. Pellionisz. “Comparison of Spatial Transformation in
Vestibulo-ocular and Vestibulo-spinal Reflexes.” Proceedings of the Symposium on the
Representation of Three-Dimensional Space in the Vestibular, Oculomotor, and Visual Sys-
tems. Bologna: Barany Society, 1988.
Petit, L., C. Orssaud, N. Tzourio, F. Crivello, A. Berthoz, and B. Mzaoyer. “Functional
Anatomy of a Prelearned Sequence of Horizontal Saccades in Man.” European Journal
of Neuroscience, 16 (1996): 3726–3741.
Petit, L., C. Orssaud, N. Tzourio, G. Salamon, B. Mazoyer, and A. Berthoz. “PET Study of
Voluntary Saccadic Eye Movements in Humans: Basal Ganglia-Thalamocortical Sys-
tem and Cingulate Cortex Involvement.” Journal of Neurophysiology, 69 (1993): 1009–
1017.
320 •
WORKS CITED
Petrides, M. “Impairments on Nonspatial Self-Ordered and Externally Ordered Working
Memory Tasks after Lesions of the Mid-dorsal Part of the Lateral Frontal Cortex in
the Monkey.” Journal of Neuroscience, 15 (1995): 359–375.
Piaget, J. “Le problème de l’intériorisation des actions en opérations réversibles.” Archives
de psychologie, 32 (1949): 242–258.
———. The Origin of Intelligence in the Child. London: Routledge and Kegan Paul, 1953.
Pierrot-Deseilligny, C., I. Israël, A. Berthoz, S. Rivaud, and B. Gaymard. “Role of the Dif-
ferent Frontal Lobe Areas in the Control of the Horizontal Component of Memory-
Guided Saccades in Man.” Experimental Brain Research, 95 (1993): 166–171.
Poincaré, H. The Value of Science, trans. George Bruce Halsted. New York: Science Press,
1907.
———. Science and Hypothesis. New York: Dover, 1952.
Poncet, M., J. F. Pellissier, M. Sebahoun, and C. J. Masser. “A propos d’un cas
d’autotopagnosie secondaire à une lésion pariéto-occipitale de l’hémisphère majeur.”
Encéphale, 60 (1971): 110–123.
Posner, M. I., M. J. Nissen, and W. C. Ogden. “Attended and Unattended Processing
Modes: The Role of Set for Spatial Location.” In Modes of Perceiving and Processing In-
formation, ed. H. L. Pick and I. J. Saltzman. Hillsdale, N.J.: Erlbaum, 1988, 137–157.
Pozzo, T., A. Berthoz, and L. Lefort. “Head Stabilisation during Various Locomotor Tasks
in Humans. I. Normal Subjects.” Experimental Brain Research, 82 (1990): 97–106.
Prochazka, A. “Sensorimotor Gain Control: A Basic Strategy of Motor Systems?” Progress
in Neurobiology, 33 (1989): 281–307.
Proust, Marcel. Remembrance of Things Past, trans. C. K. Scott Moncrieff and Terence
Kilmartin. London: Chatto and Windus, 1981.
Rademaker, G. G. J. Réactions labyrinthiques et équilibre. Paris: Masson, 1935.
Reuchlin, M., J. Lautrey, C. Marendaz, and T. Ohlman. Cognition: L’universel et l’individuel.
Paris: Presses Universitaires de France, 1989.
Ribot, T. The Psychology of the Emotions. Contemporary Science Series. London: W. Scott,
1897.
Riehle, A., and J. Requin. “Monkey Primary Motor and Premotor Cortex: Single-Cell Ac-
tivity Related to Prior Information about Direction and Extent of an Intended Move-
ment.” Journal of Neurophysiology, 61 (1989): 534–549.
Rispal-Padel, L., F. Cicirata, and C. Pons. “Cerebellar Nuclear Topography of Simple and
Synergistic Movements in the Alert Baboon (Papio papio).” Experimental Brain Re-
search, 47 (1982): 365–380.
Rizzolatti, G., and V. Galeze. “Mechanisms and Theories of Spatial Neglect.” In Handbook
of Neuropsychology, ed. F. Boller and J. Gravman. Amsterdam: Elsevier, 1988, 289–313.
Rizzolatti, G., M. Gartilucci, R. M. Camarda, V. Gallex, G. Luppino, M. Matelli, and L.
Fogassi. “Neurons Related to Reaching-Grasping Arm Movements in the Rostral Part
of Area 6 (Area 6a).” Experimental Brain Research, 82 (1990): 337–350.
Rizzolatti, G., L. Riggio, and B. M. Sheliga. “Space and Selective Attention.” In Attention
and Performance XV, ed. C. Umilta and M. Moscovitch. Hillsdale, N.J.: Erlbaum, 1994,
232–265.
Roll, J. P., J. C. Gilhodes, R. Roll, and F. Harley. “Are Proprioceptive Sensory Inputs Com-
WORKS CITED •
321
bined into a Gestalt?” In Attention and Performance XVI, ed. T. Inoue and J. T.
MacClelland. Cambridge, Mass.: MIT Press, 1995, 290–314.
Roll, J. P., J. C. Gilhodes, R. Roll, and J. L. Velay. “Contribution of Skeletal and Extraocular
Proprioception to Kinaesthetic Representation,” in Attention and Performance XIII, ed.
M. Jeannerod. Hillsdale, N.J.: Erlbaum, 1990, 549–566.
Roll, J. P., R. Roll, and J.-L. Velay. “Proprioception as a Link between Body and Extra-per-
sonal Space.” In Brain and Space, ed. J. Paillard. Oxford: Oxford University Press, 1991,
112–132.
Rolls, E. T. “A Theory of Hippocampal Function.” Hippocampus, 6 (1996): 601–620.
Rolls, E. T., and S. M. O’Mara. “Neurophysiological and Theoretical Analysis of How the
Hippocampus Functions in Memory.” In Brain Mechanisms of Perception: From Neuron
to Behavior, ed. T. Ono. New York: Oxford University Press, 1993, 276–297.
Rolls, E. T., and A. Treves. Neural Networks and Brain Function. New York: Oxford University
Press, 1998.
Ronsard, P. de. Poems of Pierre de Ronsard, trans. and ed. Nicholas Kilmer. Berkeley: Univer-
sity of California Press, 1979.
Rousseau, J.-J. Emile; Or, on Education. New York: Basic Books, 1979.
Rubens, A. B. “Caloric Stimulation and Unilateral Visual Neglect.” Neurology, 35 (1985):
1019–1024.
Sakata, H., and M. Kusunoki. “Organization of Space Perception: Neural Representation
of Three-Dimensional Space in the Posterior Parietal Cortex.” Current Opinion in
Neurobiology, 2 (1992): 170–174.
Sakata, H., H. Shibutani, and K. Kawano. “Functional Properties of Visual Tracking Neu-
rons in Posterior Parietal Association Cortex of the Monkey.” Journal of Neuro-
physiology, 49 (1983): 1364–1380.
Sakata, H., M. Taira, A. Murata, and S. Mine, “Neural Mechanisms of Visual Guidance of
Hand Action in the Parietal Cortex of the Monkey.” Cerebral Cortex, 5 (1995): 429–438.
Salzman, C., and W. T. Newsome. “Neural Mechanisms for Forming a Perceptual Deci-
sion.” Science, 264 (1994): 231–237.
Salzmann, E. “Attention and Memory Trials during Neuronal Recording from the Primate
Pulvinar and Posterior Parietal Cortex (Area PG).” Behavioral Brain Research, 67 (1995):
241–253.
Sans, A., and E. Scarfone. “Afferent Calyces and Type I Hair Cells during Development. A
New Morphological Hypothesis.” Annals of the New York Academy of Sciences, 781
(1996): 1–12.
Sartre, J.-P. The Psychology of Imagination. Westport, Conn.: Greenwood Press, 1978.
Sasaki, K., and H. Gemba. “Electrical Activity in the Prefrontal Cortex Specific to No-Go
Reaction of Conditioned Hand Movement with Colour Discrimination in the Mon-
key.” Experimental Brain Research, 64 (1986): 603–606.
Sauvan, S. M., and E. Peterhans. “Neural Integration of Visual Information and Direction
of Gravity in the Prestriate Cortex of the Alert Monkey.” In Multisensory Control of
Posture. New York: Plenum, 1995, 43–50.
Schlag, J., and M. Schlag-Rey. “Evidence for a Supplementary Eye Field.” Journal of Neuro-
physiology, 57 (1987): 179–200.
322 •
WORKS CITED
Schmidt, R. A. “A Schema Theory of Discrete Motor Skill Learning.” Psychological Review,
32 (1975): 225–260.
Schöner, G. “A Dynamic Theory of Coordination of Discrete Movements.” Biological Cyber-
netiks, 63 (1990): 257–270.
Sechenov, I. M. Selected Works. Leningrad: State Publishing House for Biological and Medi-
cal Literature, 1935.
Segraves, M. A. “Effects of Frontal Eye Field Stimulation upon Omnipause and Burst Neu-
rons in the Monkey Paramedian Pontine Reticular Formation.” Society of Neuroscience
Abstracts, 18 (1992): 296.10.
Segraves, M. A., and M. E. Goldberg. “Functional Properties of Corticotectal Neurons in
the Monkey’s Frontal Eye Field.” Journal of Neurophysiology, 58 (1987): 1387–1419.
Shadlen, M. N., and W. T. Newsome. “Motion Perception: Seeing and Deciding.” Proceed-
ings of the National Academy of Sciences, 93 (1996): 628–633.
Shallice, T. Symptômes et modèles en neuropsychologie. Paris: Presses Universitaires de France,
1993.
Sheliga, B. M., L. Riggio, and G. Rizzolatti. “Orienting of Attention and Eye Movements.”
Shepard, R. N. “Ecological Constraints on Internal Representation: Resonant Kinematics
of Perceiving, Imagining, Thinking, and Dreaming.” Psychological Review, 91 (1984):
417–447.
Sherrington, G. S. The Integrative Action of the Nervous System. New Haven: Yale University
Press, 1906.
———. “Observations on the Sensual Role of the Proprioceptive Nerve Supply of Extrin-
sic Ocular Muscles.” Brain, 41 (1918): 332–343.
Shinoda, Y., T. Ohgaki, T. Futami, and Y. Sugiushi. “Vestibular Projections to the Spinal
Cord: The Morphology of Single Vestibulo-Spinal Axons.” Vestibulospinal Control of
Posture and Movement, ed. O. Pompeiano and J. Allum. Progress in Brain Research.
Amsterdam: Elsevier, 1988, 17–27.
Silberpfennig, J. “Contributions to the Problem of Eye Movements. III. Disturbance of Oc-
ular Movements with Pseudo-hemianopsia in Frontal Lobe Tumors.” Confinia Neuro-
logica, 4 (1941): 1–13.
Simpson, J. “The Accessory Optic Systems.” Annual Review of Neuroscience, 7 (1984): 13–14.
Simpson, J. I., and W. Graf. “The Selection of Reference Frames by Nature and Its In-
vestigators,” in Adaptative Mechanisms in Gaze Control. Facts and Theories. Review of
Oculomotricity Research, vol. 1, ed. A. Berthoz and G. Melvill-Jones. Amsterdam:
Elsevier, 3–20.
Singer, W. “Search for Coherence: A Basic Principle of Cortical Self-Organization.” Con-
cepts in Neuroscience, 1 (1990): 1–26.
Sirigu, A., J. R. Duhamel, L. Cohen, B. Pillon, B. Dubois, and Y. Abid. “The Mental Repre-
sentation of Hand Movements after Parietal Cortex Damage.” Science, 273 (1996):
1564–1568.
Sirigu, A., J. Grafman, K. Bressler, and T. Sunderland. “Multiple Representations Contrib-
ute to Body Knowledge Processing.” Brain, 114 (1991): 629–642.
Slotine, J.-J. and W. Li. Applied Nonlinear Control. Englewood Cliffs, N.J.: Prentice Hall, 1991.
WORKS CITED •
323
Smith, B. H. “Vestibular Disturbances in Epilepsy.” Neurology, 10 (1960): 465–469.
Soechting, J. F., and A. Berthoz. “Dynamic Role of Vision in the Control of Posture in
Man.” Experimental Brain Research, 36 (1979): 551–561.
Soechting, J. F., and M. Flanders. “Sensorimotor Representations for Pointing to Targets in
Three-Dimensional Space.” Journal of Neurophysiology, 62 (1989): 582–594.
Soechting, J. F., and B. Ross. “Psychophysical Determination of Coordinate Representation
of Human Arm Orientation.” Neuroscience, 13 (1984): 595–604.
Sokolov, E. N., and O. S. Vinogradova. Neuronal Mechanisms of the Orienting Reflex.
Hillsdale, N.J.: Erlbaum, 1975.
Squire, L. R., and M. Zola-Morgan. “The Medial Temporal Lobe Memory System.” Science,
253 (1991): 1380–1386.
Stein, B. E., and M. A. Meredith. The Merging of the Sciences. Cambridge, Mass.: MIT Press,
1993.
Sully, J. Les Illusions des sens et de l’esprit. Paris: Bibliothèque scientifique internationale,
1883.
Szentagothai, J. “The Elementary Vestibulo-ocular Reflex Arc.” Journal of Neurophysiology,
13 (1950): 395–407.
Tamura, R., T. Ono, M. Fukuda, and K. Nakamura. “Recognition of Egocentric and
Allocentric Visual and Auditory Space by Neurons in the Hippocampus of Monkeys.”
Neuroscience Letters, 109 (1990): 293–298.
———. “Spatial Responsiveness of Monkey Hippocampal Neurons to Various Visual and
Auditory Stimuli.” Hippocampus, 2 (1992): 307–322.
Tanaka, K. “Inferotemporal Cortex and Object Vision.” Annual Review of Neuroscience, 19
(1996): 109–139.
Taube, J. S., R. U. Muller, and J. B. Ranck Jr. “Head-Direction Cells Recorded from the
Postsubiculum in Freely Moving Rats. I. Description and Quantitative Analysis.” Jour-
nal of Neuroscience, 10 (1990): 410–435.
———. “Head-Direction Cells Recorded from the Postsubiculum in Freely Moving Rats. II.
Effects of Environmental Manipulations.” Journal of Neuroscience, 10 (1990): 436–447.
Thier, P., and R. G. Erickson. “Responses of Visual-Tracking Neurons from Cortical Area
MST-I to Visual, Eye, and Head Motion.” European Journal of Neuroscience, 4 (1992):
539–553.
Thier, P., and O. Karnath. Parietal Lobe Contribution to Orientation in 3D Space. Berlin:
Springer, 1997.
Thomas, A. Equilibre et équilibration. Paris: Masson, 1940.
Treves, A., and E. T. Rolls. “Computational Constraints Suggest the Need for Two Distinct
Input Systems to the Hippocampal CA3 Network.” Hippocampus, 2 (1991): 189–199.
———. “What Determines the Capacity of Autoassociative Memories in the Brain?” Net-
work, 2 (1991): 371–397.
Trotter, Y., S. Celebrini, B. Stricane, S. Thorpe, and M. Imbert. “Neural Processing of
Stereopsis as a Function of Viewing Distance in Primate Visual Cortical Area V1.”
Journal of Neurophysiology 76 (1996): 2872–2875.
Trullier, O., S. Wiener, A. Berthoz, and J. A. Meyer. “Biologically Based Artificial Naviga-
tion Systems: Reviews and Prospects.” Progress in Neurobiology, 51 (1997): 483–544.
Tuohimaa, P., E. Aantaa, K. Toukoniitty, and P. Mäkelä. “Studies of Vestibular Cortical
324 •
WORKS CITED
Areas with Short Living Oxygen 15 Isotopes.” Otorhinolaryngologica, 45 (1983): 315–
321.
Turvey, M. T., and P. N. Kugler. “An Ecological Approach to Perception and Action.” In Hu-
man Motor Actions. Bernstein Reassessed, ed. H. T. A. Whiting. Amsterdam: Elsevier,
1984.
Uhl, F., P. Franzen, G. Lindinger, W. Lang, and L. Deecke. “On the Functionality of the Vi-
sually Deprived Occipital Cortex in Early Blind Persons.” Neuropsychologia, 33 (1994):
256–259.
Umilta, C., C. Mucignat, L. Riggio, C. Barbieri, and G. Rizzolatti. “Programming Shifts of
Spatial Attention.” European Journal of Cognitive Psychology, 6 (1994): 23–41.
Ungerleider, L. G. “Functional Brain Imaging Studies of Cortical Mechanisms for Mem-
ory.” Science, 270 (1995): 769–665.
Vallar, G., E. Lobel, G. Galati, A. Berthoz, L. Pizzamiglio, and D. Le Bihan. “A Fronto-Pari-
etal System for Computing the Egocentric Spatial Frame of Reference in Humans.”
Vallar, G., M. L. Rusconi, S. Barozzi, B. Bernardini, D. Ovadia, C. Papagno, and A.
Cesarani. “Improvement of Left Visuo-Spatial Hemineglect by Left-Sided Trans-
cutaneous Electrical Stimulation.” Neuropsychologia, 33 (1994): 73–82.
Vallar, G., R. Sterzi, G. Bottini, S. Cappa, and M. L. Rusconi. “Temporary Remission of
Left Hemianesthesia after Vestibular Stimulation. A Sensory Neglect Phenomenon.”
Cortex, 26 (1990): 123–131.
Viallet, F., J. Massion, R. Massarino, and R. Khalil. “Coordination between Posture and
Movement in Parkinsonism and SMA Lesion.” In From Neuron to Action, ed. L.
Deecke, J. C. Eccles, and V. B. Mountcastle. Berlin: Springer, 1990, 65–70.
Viaud-Delmon, I., and A. Berthoz. No title. Current Opinion in Neurobiology, in press.
Viaud-Delmon, I., Y. P. Ivanenko, A. Berthoz, and R. Jouvent. “Sex, Lies, and Virtual Real-
ity.” Nature, 1 (1998): 15–16.
———. “Anxiety and Integration of Visual Vestibular Information Studied with Virtual Re-
ality.” Biological Psychiatry, in press.
Vitte, E., C. Derosier, Y. Caritu, A. Berthoz, D. Hasboun, and D. Soulié. “Activation of the
Hippocampal Formation by Vestibular Stimulation: A Functional Magnetic Reso-
nance Study.” Experimental Brain Research, 112 (1996): 523–526.
Vitte, E., A. Sémont, and A. Berthoz. “Repeated Optokinetic Stimulation in Conditions of
Active Standing Facilitates the Recovery from Vestibular Deficits.” Experimental Brain
Research, 102 (1994): 141–148.
Viviani, P. “Motor-Perceptual Interactions: The Evolution of an Idea.” In Cognitive Science
in Europe, ed. M. Imbert, P. Bertelson, R. Kempson, D. Osherson, H. Schnelle, N.
Streitz, A. Thomassen, and P. Viviani. Heidelberg: Springer, 1987, 11–39.
Viviani, P., G. Baud-Bovy, and M. Redolfi. “Perceiving and Tracking Kinaesthetic Stimuli:
Further Evidence of Motor-Perceptual Interactions.” Journal of Experimental Psychol-
ogy, 23 (1997): 1232–1252.
Viviani, P., and A. Berthoz. “Voluntary Deceleration and Perceptual Activity during
Oblique Saccades.” In Control of Gaze by Brain Stem Neurons, ed. R. Baker and A.
Berthoz. Amsterdam: Elsevier, 1977, 23–28.
Viviani, P., and T. Flash. “Minimum-Jerk, Two-Thirds Power Law, and Isochrony. Con-
WORKS CITED •
325
verging Approaches to Movement Planning.” Journal of Experimental Psychology (Hu-
man Perception), 21 (1995): 32–53.
Viviani, P., and R. Schneider. “A Developmental Study of the Relation between Geometry
and Kinematics of Drawing Movements.” Journal of Experimental Psychology (Human
Perception), 17 (1991): 198–218.
Viviani, P., and N. Stucchi. “Biological Movements Look Uniform: Evidence for Motor-Per-
ceptual Interactions.” Journal of Experimental Psychology (Human Perception), 18 (1992):
603–623.
Viviani, P., and C. Terzuolo. “Trajectory Determines Movement Dynamics.” Neuroscience,
72 (1982): 431–437.
von Uexküll, J. Streifzüge durch die Umwelten von Tieren und Menschen. Berlin: Springer, 1965.
———. Umwelt und Innenwelt der Tieren und Menschen. Berlin: Springer, 1934.
Walton, K. D., D. Lieberman, A. Llinás, M. Begin, and R. Llinás. “Identification of a Criti-
cal Period for Motor Development in Neonatal Rats.” Neuroscience, 51 (1992), 763–
767.
Wexler, M., S. Kosslyn, and A. Berthoz. “Motor Processes in Mental Rotation.” Cognition,
68 (1998): 77–94.
Wiener, S. I., V. A. Korshunov, R. Garcia, and A. Berthoz. “Inertial, Substratal, and Land-
mark Cue Control of Hippocampal CA1 Place Cell Activity.” European Journal of Neu-
roscience, 7 (1995): 2206–2219.
Wiener, S. I., C. A. Paul, and H. Eichenbaum. “Spatial and Behavioral Correlates of
Hippocampal Neuronal Activity.” Journal of Neuroscience, 9 (1989): 2737–2763.
Wurtz, R. E., C. Duffy, and J. P. Roy. “Motion Processing for Guiding Self-Motion.” In Brain
Mechanisms of Perception and Memory, ed. T. Ono, L. R. Squire, M. Raichle, D. I.
Perrett, and M. Fukuda. Oxford: Oxford University Press, 1993.
Wurtz, R. E., and M. E. Goldberg. “Activity of Superior Colliculus in Behaving Monkey. III.
Cells Discharging before Eye Movements.” Journal of Neurophysiology, 35 (1972): 575–
586.
Yasui, S., and L. R. Young. “Perceived Visual Motion as Effective Stimulus to Pursuit Eye
Movement System.” Science, 190 (1975): 906–908.
Yoshida, K., R. A. McCrea, A. Berthoz, and P. P. Vidal. “Interneurones inhibiteur de la
saccade oculaire horizontale étudiée chez le chat éveillé à l’aide d’injections intra-
axonique de peroxydase.” Comptes rendus hebdomadaires de l’Académie des sciences (Paris,
Série III), 290, D53 (1980): 636–638.
Yoshida, K., A. Berthoz, P. P. Vidal, and R. A. McCrea. “Eye Movement Related Activity of
Identified Second-Order Vestibular Neurons in the Cat.” In Progress in Oculomotor Re-
search, ed. A. Fuchs and W. Becker. Amsterdam: Elsevier, 1981, 371–378.
Yoshida, K., R. A. McCrea, A. Berthoz, and P. P. Vidal. “Properties of Immediate Premotor
Inhibitory Burst Neurons Controlling Horizontal Rapid Eye Movements in the Cat.”
In Progress in Oculomotor Research, ed. A. Fuchs and W. Becker. Amsterdam: Elsevier,
1981, 71–81.
Young, L., M. Shelhamer, and S. Modestino. “M.I.T./Canadian Vestibular Experiments in
Weightlessness.” Experimental Brain Research, 64 (1986): 299–307.
326 •
WORKS CITED
CREDITS
Figure 1.2. Redrawn from N. A. Bernstein, “Some Emergent Problems of the Regulation of
Motor Acts,” Questions of Psychology 6 (1957).
Figure 1.3. Redrawn from R. Llinás and U. Ribary, “Coherent 40-Hz Oscillation Character-
izes Dream State in Humans.” Proceedings of the National Academy of Sciences, USA 90
(1993): 2078–2081; Fig. 4, p. 2080. Copyright © 1993 National Academy of Sciences,
U.S.A.
Figure 1.4. Redrawn from R. A. Schmidt, “A Schema Theory of Discrete Motor Skill
Learning.” Psychological Review, 32 (1975): 236. Copyright © 1975 by the American Psy-
chological Association. Adapted with permission.
Figure 3.1. Redrawn from M. Livingstone and D. Hubel, “Segregation of Form, Color,
Movement, and Depth: Anatomy, Physiology, and Perception.” Science, 240 (1988): 742.
Copyright © 1988 American Association for the Advancement of Science. Reprinted with
permission.
Figure 3.6. Redrawn from A. Grantyn, “How Visual Inputs to the Ponto-Bulbar Reticular
Formation Are Used in the Synthesis of Premotor Signals during Orienting.” In Progress
in Brain Research, vol. 80, ed. J. H. J. Allum and M. Hulliger. Amsterdam: Elsevier, 1989,
159–170. Reprinted with the permission of the author and Elsevier Science.
Figure 3.7. Adapted from a figure in M. Graziano and C. Gross. Experimental Brain Research
97: 100. Copyright © Springer-Verlag, 1993. Reprinted with the permission of the author
and Springer-Verlag.
Figure 4.2. Redrawn from M. Graziano and C. Gross, “Multiple Pathways for Processing
Visual Space.” In T. Inui and J. L. McClelland (eds.), Attention and Performance XVI. Cam-
bridge, Mass.: MIT Press, p. 199. Reprinted with the permission of the authors and MIT
Press.
Figure 4.3. Redrawn from M. A. Arbib, “Perceptual Structures and Distributed Motor Con-
trol.” In Handbook of Physiology, sec. 1, The Nervous System, vol. 2, pt. 2, ed. V. B. Brooks.
Bethesda: American Physiological Society, 1981, p. 1468. Reprinted with permission of
the American Physiological Society.
Figure 5.1. Redrawn from L. R. Squire and M. Zola-Morgan, “The Medial Temporal Lobe
Memory System.” Science 253 (1991): 1381. Copyright © 1991 American Associaion for
the Advancement of Science. Reprinted with permission.
Figure 5.2. European Space Agency (ESA) documents.
Figure 5.3. Redrawn from S. R. Cajal, Histologie du système nerveux de l’homme et des vertébrés.
Madrid: Raycar, 1972 [Paris, 1911].
Figure 5.4. Redrawn from E. T. Rolls, “A Theory of Hippocampal Function.” Hippocampus, 6
(1996): 601–620. Copyright © 1996. Reprinted by permission of Wiley-Liss, Inc., a subsid-
iary of John Wiley & Sons, Inc.
Figure 6.3. Redrawn from P. Viviani and T. Flash, “Minimum-Jerk, Two-Thirds Power Law,
and Isochrony. Converging Approaches to Movement Planning.” Journal of Experimental
Psychology (Human Perception), 21 (1995): 40. Copyright © 1995 by the American Psycho-
logical Association. Adapted with permission.
Figures 7.1 and 7.2. Redrawn from W. Graf and R. Baker, “Adaptive Changes of the
Vestibulo-Ocular Reflex in Flatfish Are Achieved by Reorganization of Central Nervous
Pathways.” Science, 221 (1983): 777–778. Copyright © 1983 American Association for the
Advancement of Science. Reprinted with permission.
Figure 7.3. Redrawn from T. Fukuda, Statokinetic Reflexes in Equilibrium and Movement. To-
kyo: Igaku-Shoin, 1981. Copyright © 1981 Igaku-Shoin. Adapted with permission.
Figures 10.2 and 10.3. Redrawn from O. Hikosaka, Figs. 30.5 and 30.9 in “Role of the Basal
Ganglia in Motor Learning.” In Brain Mechanisms and Memory: Cells, Systems, and Circuits,
ed. J. L. Gough, N. M. Weinberger, and G. Lynch. New York: Oxford University Press,
1992. Copyright © 1992 by Oxford University Press, Inc. Used with permission of Oxford
University Press, Inc.
Figure 11.1. From F. Lestienne, J. Soechting, and A. Berthoz, “Postural Readjustments In-
duced by Linear Motion of Visual Scenes.” Experimental Brain Research, 28 (1977): 363–
384. Copyright © 1977 Springer-Verlag. Reprinted with permission.
Figure 11.2. Redrawn from J. Massion, “Postural Control System.” Current Opinion in Neuro-
biology, 4 (1994): 877–887; Fig. 1, p. 878. Copyright © 1994 Elsevier Science. Reprinted
with permission.
Figure 11.3. Redrawn from C. Darwin, The Expression of the Emotions in Man and Animals.
New York: Philosophical Library, 1955.
Figure 13.1. Redrawn from LPPA and NASA documents.
328 •
CREDITS
INDEX
Page numbers in italics refer to figures.

Acceleration: detection of, 33–35, 38, 43, 46, Babinski, J., 140, 225
67; illusions caused by, 244–248 Babinski synergy, 225
Acceptor of the result of action, 11–13 Bach-y-Rita, P., 84, 85
Accessory optic pathway, 43, 64 Baker, R., 6, 47
Accessory optic system, 63, 67 Balance: development of, 216; maintenance
Accommodation, 233 of, 224–227; multisensory control of, 6, 43–
Action: acceptor of the result of, 11–13; and 44, 174–175, 222; physiology of, 217; and
vision, 11. See also Movement vision, 218–221. See also Posture
Action-perception cycle, 13–16, 14, 19 Basal ganglia, 200–201, 202, 225, 265
Actions, hierarchy of, 10 Basket (Golgi) neurons, 193
Adaptation: definition of, 233–234; and substi- Bell, 183, 195
tution, 234–238 Bergson, H., 7, 242, 261–262
Adrian, E. D., 227 Beritoff (Beritashvili), J. S., 119–120
Aerial acrobatics, 246–248 Bernard, Claude, 7
Affordances, 59, 132 Bernini, G. L., 113
Agoraphobia, 253, 264 Bernstein, N. A., 6, 13–16, 14, 112, 137, 140,
Allesthesia, 58 141, 154, 165, 190, 223
Allocentric (exocentric) frames of reference, Bimodal neurons, 85, 109
99–100, 128, 129, 241 Bizzi, E., 112
Amar, J., 140 Blind people, “seeing” by, 83–85
Anamorphoses, 117–118 Blind vision, 50–51
Anokhin, P. K., 11–13 Bodily masses, mechanical properties of, 6
Anticipation: in capture, 165–166; in control Bodily self-motion. See Self-motion, percep-
of movement, 140–141; and gaze, 148; of tion of
head movements, 35; in muscle contrac- Body: dissociation of, 70, 98, 139; illusions of
tions, 28, 29; in navigation, 125–126; in oc- movement of, 248–250; impaired percep-
ular pursuit, 62; and postural stability, 43– tion of, 230, 238–239; neglected perception
44; and superior colliculus, 77; in vection, of, 74, 75; phantom perception of, 98, 227–
53 229
Arbib, M. A., 110–111 Body frames of reference, 230–231
Aristotle, 25, 57–58, 139 Body schema, 70, 75, 105, 114, 227, 229–232,
Arm movements, neural bases for, 160– 264
162 Bonda, E., 231
Attention: displacements of, 192, 211; motor Borelli, G. A., 139
theory of, 200, 212 Bottini, G., 75
Aubert effect, 103 Bourdieu, P., 187
Autism, 93–96, 189, 264 Bouveresse, J., 21
Autoinhibition, 193 Brahé, T., 140
Automatics, science of, 217 Braille, 83–85
Autoscopy, 98 Brain: complexity of, 2–3; computer meta-
Autotopoagnosia, 230 phor for, 4, 22, 261–262; description of, 3;
Axon branching, 155–157 and internal models, 177–179, 227,
INDEX •
329
Brain (continued) Composite variables, principle of, 138
229–32; versus mind, 1–2; modulation of Conditioned reflexes, 11
sensory information by, 29, 35–36; predic- Consecutive optokinetic nystagmus, 55–56
tive power of, 57; processing of informa- Constructivism, 11
tion from sensory receptors, 5–6, 263; pro- Continuity, in perception, 92–93
cessing of movement by, 22, 23; as a Contralateral colliculus, 79
tensor, 48–49; use of memory by, 6 Coordination, of movements, 141
Brain imaging, 231–232 Cordo, P., 223
Braking (collision prevention), 168–172 Coriolis accelerations, 251
Brandt, T., 52 Coriolis forces, 6, 119
Broom experiment, 223 Cornilleau-Pérès, V., 92
Bruyère, A., 256, 259 Corollary discharge, theory of, 183
Cortical vision, 50
Capture: calculations to perform, 22–23; per- Cutaneous (skin) receptors, 26, 27, 29–31, 83–
ceptions in, 165–166; and tracking of tar- 86, 105–106. See also Touch
gets, 62–64 Cybernetic paradigm, 139, 174, 183
Caricature, 132 Cyon, E., 36
Catching a ball, 170–171, 172–173
Cemetery illusion, 247–248 Damasio, A. R., 7
Centralist theory, 31, 263 Dargassies, Saint Anne, 186
Cerebellum: and body schema, 229–230; de- Darlot, C., 91–92
velopment of, 212; function of, 64, 65, 164, Darwin, Charles, 32, 118, 139, 188, 232
201–203, 225, 265; inhibitions in, 193–194; Dead reckoning, 118, 120, 124
multisensory convergences in, 59, 82 De Ajuriaguerra, J., 83, 186
Cerebral imagery, 175 De animalium motu (Aristotle), 139
Chakyar community (Kerala), 189 Decision making, 7, 13, 95
Chance, 255 Deecke, L., 208
Châtelet, G., 2 Degrees of freedom, control of number of,
Children: development of movements in, 141–147, 142, 154
179, 214; teaching speed sports to visually Déjerine, J., 217
impaired, 51 Descartes, René, 3, 11
Chronography, 140 Dichgans, J., 52, 67
Churchland, P., 48 Directional sensitivity, 32–33
Cingulate cortex, 209–210 Displacement, 67, 92; of attention, 192, 211;
Cognition, 1–2; and language, 4 and balance, 225; memory of, 120–122,
Cognitive sciences, multidisciplinary nature 125. See also Acceleration; Velocity
of, 11 Distance perception, 76–77
Coherence: and autism, 93–96, 189, 264; Donders, F. C., 150
definition of, 56, 57–59, 94–95, 264; be- Dorsal motor nucleus, 65
tween seeing and hearing, 77–89; and spa- Dorsal terminal nucleus (DTN), 64
tial neglect, 74–76; and unity of percep- Dorsolateral pontine nucleus (DLPN), 63
tion, 90–93, 114. See also Perception Dorsolateral thalamus, 74
Collateral axon branches, 156–157 Dorsomedial pontine nucleus (DMPN), 63
Colliculus. See Contralateral colliculus; Infe- Dove prisms, 235, 236
rior colliculus; Superior colliculus Droulez, J., 91–92, 213
Collision prevention (braking), 168–172 Duchenne de Boulogne, 217
Comparator, 13–16, 14, 19 Dynamic illusions, 244–246
Compensation, 239–241 Dynamic memory, 212–214
330 •
INDEX
Dynamic systems theory, 179–180 Flatfish, 158–160
Dynamic vision, testing of, 51 Flavor. See Taste
Flocculus, 235–236, 251
Eccles, J. C., 46, 193 Flourens, P., 32
Efferent copy, principle of, 69 “Focus of expansion,” 60
Effort, sense of, 31–32, 88 Force, perception of, 30, 31–32
Egocentric frames of reference, 97, 99–100, Force feedback, 88–89
101, 113, 129, 241 Fovea, 62, 79, 148–149, 151, 156, 185
Einstein, Albert, 33, 38 Frames of reference: definition of, 97; selec-
Emotion: and gaze, 188–189; and movement, tion of, 109–114, 241; types of, 99–109,
7–8; and perception, 265–266; and posture, 110–114
139, 188, 232 Frequential filtering, 41
Empedocles, 135, 182 Frith, Uta, 94–95, 189
Encoding, definition of, 20–21 Frontal eye field (FEF), 62, 207–208
Endolymph, 33, 34, 38 Frontal oculomotor field (FOF), 63
Entorhinal cortex, 127 Fukuda, T., 43
Environmental frame of reference, 97 Functional flexibility, 162–163, 221, 263
Epilepsy, and visual illusions, 70 Functionalism, 4–5
Episodic memory, 6
Equilibrium point model, 16, 112–114, 223 γ motor neurons, 28–29, 32, 233
Ereismatic posture, 223 γ oscillation, 133
Eshkol, N., 97 Gannets, plummeting of, 170
Euclidean geometry, of vision, 36–37, 60 Gaze: absent, 189; anticipatory displacement
Evarts, E. V., 160 of, 148; and capture, 165–166; development
“Evil eye,” 186–187 of, 183–185; direction of, 68–69, 70, 81,
Ewert, J. P., 167 101; and emotion, 188–189; neural basis of,
Excitatory burst neuron (EBN), 197; horizon- 190–191; stabilization of, 155–157, 184, 195
tal (EBNH), 197; vertical (EBNV), 197 Gaze orientation, 6, 181–185
Exocentric (allocentric) frames of reference, Gaze reciprocity, 186
99–100, 128, 129, 241 Gaze stability, 44–46, 45
Explicit frames of reference, 112–114 Gender differences, in adaptation to visuo-
Extramission, theory of, 182 vestibular conflict, 221, 237
Extrapersonal space, 98–99, 170 Geocentric frame of reference, 101
Eye, development of, 147–151. See also Gaze; Geometry: internalization of, 176; of mus-
Sight cles, 145; of skeleton, 143–144; of space,
Eye muscles: development of, 149; geometry 100–101; of vision, 36–38, 48–49, 60, 65,
of, 72 71, 72
Eye-to-eye contact, 185–188 Gestalt theory, 11
Gestures, graceful, 172
Face recognition, 73, 109–110 Gibson, J. J., 11, 43, 52, 58, 59, 60, 125, 132,
Facial expressions, 188 177, 185, 254
Feedforward, 183 Globus pallidus externus (GPe), 201
Feelings. See Emotion Goethe, Johann Wolfgang von, 4
Feldman, A. G., 112–113 Golgi (basket) neurons, 193
Ferrier, D., 58, 104, 207 Golgi tendon organs, 26, 31–32
Fessard, A., 11 Gonshor, A., 235
“Filling in” capacity, 132, 134 Gravito-inertial differentiation, problem of,
Flash, T., 113, 147, 152 41–42, 101
INDEX •
331
Gravito-inertial receptor system, 33 Imagined movement, 211–212
Gravity: detection of, 38, 41; as frame of ref- Immobility, sense of, 41
erence, 100–102; illusions caused by, 244– Implicit frames of reference, 112–114
248; internalization of, 173–174; and pre- Inertia, overcoming, 3, 32, 33
diction of movement, 172; receptors in Inertial cues, 32, 33
stomach, 104–105; role in perception, 73 Inertial navigation, 119–120
Graziano, M., 109 Inferior colliculus, 78, 80
Grillner, S., 162 Inferior olive, dorsal motor nucleus of, 65
Gropius, W., 257 Inferior temporal (IT) area, 130
Gross, C., 85–86, 109, 231 Inhibition, of sensations, 87–88, 192–194
Grüsser, O. J., 72, 98 Inhibitory second-order vestibular neurons,
Gurfinkel, V. S., 22, 105, 107, 219, 224, 228, 193
229 Inner ear receptors. See Vestibular (inner ear)
receptors
Hallucinations, 253–254, 264 Interactive frame of reference, 97
Hartline, H. K., 35 Interneurons, 197
Haussmann, Baron G. E., 258 Inverse problem, 177–179
Head: restriction of, 143–144; stabilization of, Involuntary post-contraction, 248–249
101, 102, 111–112, 143, 184; tilt of, 38–42, Ischemia, 227–228
72–73, 92 Isochrony, principle of, 31, 147, 152
Head, H., 227, 230 Ito, M., 116, 193, 236
Hearing, 25, 77–82
Hein, A., 11, 58 Jackson, J. Hughlings, 126
Held, R., 11, 58 James, William, 9–10, 10
Heliodorus of Emesa, 186–187 Janet, P., 10, 241
Helmholtz, H. von, 9, 21, 25, 33, 77, 195, 243 Jeannerod, M., 11, 22, 110, 196
Hemianesthesia, 75 Jerk: definition of, 151; detection of, 33, 34–
Hemianopia, 75 35, 46; minimalization of, 151–152, 152,
Hemispatial neglect, 74 162
Heraclitus of Ephesus, 3 Joint receptors, 26, 27, 31–32
Hertz, H., 137 Julesz stereograms, 76
Hikosaka, O., 201 Jung, C., 52
Hippocampal formation, 73, 74
Hippocampus: anatomy of, 127; function of, Kalman filters, 174–175
126–136, 190, 191, 251, 264 Kandinsky, V., 257
Holmes, G., 227 Kant, Immanuel, 5
Husserl, E., 16, 87, 132, 168 Kathakali, 188–189
Hypothesis, best possible, 243 Kawato, M., 178
Hyvarinen, J., 98 Kelso, J. A. S., 172
Kepler, J., 140
Idiotropic vector, 104–105 Kinematic analysis, 140
Illusions, 6; of agoraphobia, 253; definition Kinematics, internalization of, 176
of, 242, 243; of dissociation of the body, Kinesthesia, 5, 25, 120
70; of epilepsy, 70; and hallucinations, 253– Koenderink, J. J., 168
254, 264; of moon size, 244; and theory of Kohnstamm’s illusion, 248–249
coherence, 91; types of, 244–254; of Korn, H., 194
vection, 52–53; of velocity inversions, 53; Kornhuber, A., 208
of ventriloquism, 58; of vertigo, 69–70, Kosslyn, S. M., 110
253; of waterfalls, 55, 252 Krechevsky, I., 243
332 •
INDEX
Lackner, J. R., 106–107 Milner, B., 126
Lacquanti, F., 173 Mind, 1–2, 95
Lampreys, 36 Minimum jerk, principle of, 151–152, 152,
Language, and cognition, 4 162
Lashley, K. S., 11, 19, 243 Mirror neurons, 20–21
Lateral geniculate body (LGB), 63 Mishkin, M., 128, 231
Lateral geniculate nucleus (LGN), 61, 204 Mittelstaedt, H., 69, 104, 105, 183
Lateral inhibition, 47 Moon illusion, 244
Lee, D. N., 52, 168–169, 172, 218 Motion sickness, 241, 247, 250–252
Lisman, J. E., 133 Motoneurons (MN), 197
Listing, J. B., 150 Motor control, 16
Listing’s plane, 150 Motor equivalence, 226–227
Llinás, R., 15, 16, 17, 48, 65, 178 Motor neurons, firing threshold of, 112–113
Local coherence, 94 Motor prediction: and superior colliculus, 77
Local invariance, 73–74 Motor schema, 17–19
Localization, 37 Motor theory of perception, 9–11, 250
Locomotion, 6, 29, 162 Mountcastle, V. H., 84, 98
Locomotor memory, 124–126 Movement: α models of, 112; in architecture,
Lorente de Nó, R., 44, 47 255–260; control and command of, 3–4, 16;
Lorenz, K., 76, 155, 181, 183 illusions of, 248–250; imagined and actual,
Lotze, R. H., 9 211–212; and inverse problem, 177–179; ki-
Lyapunov equation, 138 nematic phase constraints on, 144–145,
147; pleasure of, 137–138, 255–260; seeing,
Mach, E., 21, 22, 33, 36, 52 26, 33, 50–56, 60–64; and shifting point of
MacKay, D. G., 19, 193 equilibrium, 16. See also Action; Kinesthe-
Magnocellular pathway, 42 sia; Locomotion
Magnocellular zone, 61 MST (medial superior temporal) area, 61–62,
Magnus, T., 43 63, 71
Malraux, A., 265 MT (middle temporal) area, 61–62, 63, 71
Marey, E. J., 139–140, 177 Müller effect, 103
Marr, D., 127 Multisensory convergence, 57–59
Medial terminal nucleus (MTN), 64 Muscle receptors, 26, 27–32
Medulla: and body schema, 230; inhibitions Muscles: contraction of, 27–28, 31, 112–113,
in, 193, 194; multisensory convergences in, 216; equilibrium in muscular forces, 16,
59 223; geometry of, 145; involuntary post-
Meissner corpuscles, 30, 84 contraction of, 248–249; memory in, 115–
Melvill-Jones, G., 235, 236–237 116; and proprioception, 26, 27–32
Memory: brain’s use of, 6, 44; of movement, Muybridge, E., 101, 177, 184
55–56; and prediction, 115–117, 264–265; Myotatic reflex, 27, 217
types of, 115, 196. See also Spatial memory
Mental movement, 55 Nadel, L., 128
Mental nodes, theory of, 19–20, 193 Nashner, Lewis M., 221
Mental rotation, 24, 73, 161–162, 231 Natural frames of reference: connected to
Meredith, M. A., 78, 82 limbs, 107–109; gravity, 100–102; receptors
Merleau-Ponty, M., 11, 22, 77, 83, 85, 96–87, for, 104–107; vertical subjectivity, 102–104
115, 135–136, 175 Natural movements: definition of, 154; orga-
Mesulam, M. M., 231 nization of, 140; pioneering work in, 139–
Michotte, A., 135 141; pleasure of, 137–138
Midsagittal plane, perception of, 100 Natural selection, 3
INDEX •
333
Navigation, definition of, 119, 126 tion of, 33, 38–41, 101, 122. See also
Navigational memory, 117–120 Saccule; Utricle
Neisser, U., 19
Neocerebellum, 203 Pacinian corpuscles, 30, 84, 105
Neocortex, 191 Paillard, J., 101, 117, 124
Nervous system, symbolic and computational Paired oscillations paradigm, 180
conception of, 4 Parahippocampus, function of, 100
Networks, functional flexibility of, 162–163, Parapsychic phenomena, 76
221, 263 Parietal cortex, 63, 204–206, 227, 230, 239,
Neural barbarism, 2 264–265
Neural circuits, anticipatory, 9–10, 10 Parieto-insular vestibular cortex (PIVC), 71
Neural “compass,” 73–74 Parieto-insular zone, lesions of, 70–71
Neural integrators, 46–47, 116, 199 Parvocellular layer, 42–43
Neural memory, 16 Parvocellular zone, 61
Neural “noise,” 91 Path integration, 125
Neuroethology, construction of, 139, 155 Pause (P) neurons, 199, 211
Neuromuscular spindles, 26, 27–29, 68, 233 Pavlov, I. P., 11, 12, 139, 190
Neurons: existence of, 3; recording activity Pellionisz, A. J., 48, 65
of, 68; sensitivity of, 22–23; vectorial en- Penfield, W., 69, 207
coding by population of, 161 Perception: active exploration in, 85, 86–88,
Neurophysiology, 11, 139 110, 136, 165, 167, 204, 255, 263; direc-
Neuropsychology, 139 tional selectivity of, 72; of distance, 76–77;
Neurotransmitters, inhibitions by, 194 and emotion, 265–266; motor theory of, 9–
Newsome, W. T., 62 11, 250; multisensory nature of, 57–59,
Newton, Sir Isaac, 172 102–104, 115; of self-motion, 52–56; unity
Nociceptors, 30 of, 90–93, 114. See also Coherence
Nonlinear control of robotic movement, the- Perception-action cycle. See Action-perception
ory of, 31 cycle
Nucleus of the optic track (NOT), 63, 64 Perceptual actions, 10–11, 19
Nystagmus, 34, 46 Perceptual stability. See Gaze stability
Perceptual systems, definition of, 58
Object control, model of, 178 Peripheralist theory, 31
Obscurantism, 265 Perrett, D. I., 87, 109
Ocular pursuit, 62–64, 150–151 Personal space, 98
Ocular torsion, 149 Phantom limbs, perception of, 98, 227–229
Oculomotor path, 196 Phenomenology, 132
Odor. See Smell Physiological clock (pacemaker), 15
O’Keefe, J., 128, 133 Physiology, 7
Omnipause neurons, 199 Piaget, J., 11, 194, 214
Ia inhibitory neurons, 193 Piéron, H., 11
Opposition space, 110, 111 Place cells, 74, 128
Optic flow, 60, 65; restriction of, 143 Planning functions, 95
Optic tectum, 61, 78, 166 Plato, 5, 19, 182
Optokinetic reflex, 64 Poincaré, H., 36–38, 76, 123, 125, 244
Optokinetic stimulation, 52 Point of equilibrium theory, 16, 112–114, 223
Orientation reaction (reflex), 13, 181, 190– Pontine nuclei, 63, 64
191. See also Gaze Porro, Ricardo, 256
Otoliths: description of, 26, 27, 38, 39; func- Position, sense of, 27–29
334 •
INDEX
Postsubiculum, function of, 73 Retinal ganglia, 64
Posture: in absence of gravity, 103–104, 219; Reuchlin, M., 221
and emotions, 139, 188, 232; as first expres- Ribot, T., 7
sion of movement, 137; as readiness to Righting reflexes, 106
move, 190; stabilization of, 43–44, 64, 105– Rigidity, in perception, 92, 93, 99
106. See also Balance Rizzolatti, G., 20, 200, 212
Power gripping, 160 Robotic movement, theory of nonlinear con-
Pozzo, T., 225 trol of, 31
Precision gripping, 160 Rolls, E. T., 130
Predators. See Capture Rondot, Pierre, 217
Prediction, 57; and dynamic memory, 212– Ronsard, Pierre de, 138
214; memory for, 115–117, 264–265; in oc- Rotational memory, 120–122
ular pursuit, 151; of trajectories, 132–134, Rotations: detection of, 42–43, 46–47, 176. See
151, 185. See also Anticipation; Motor pre- also Mental rotation
diction Rougeul-Buser, A., 16
Prefrontal cortex (PFC), 121, 206, 210–211, Rousié, D., 239
264 Rousseau, J.-J., 9
Prehistoric cave art, 134–136
Preperceptions, 20, 132 Saccades: cortical control of, 204–211; defini-
Prepositus hypoglossi nucleus, 46, 47 tion of, 62, 192, 194, 195–196; inhibition of,
Presynaptic inhibitory mechanisms, 30, 195, 199–203; production of, 197–199
32 Saccadic inhibition, 195, 199–203
Prochazska, A., 88 Saccule: description of, 26, 27, 38; function of,
Proprioception, description of, 26, 27–32 33, 38–41
Proust, Marcel, 91, 132 Sailors’ illusion, 252–253
Prouvé, Jean, 257 Sakata, H., 257
Purkinje, 183, 195 Sartre, J.-P., 83
Putnam, Hilary, 261 Schema, definition of, 17–18, 18, 20
Pyramidal cells, 128 Schizophrenia, 211, 264
Pyramidal cortical neurons, 31, 132–133 Schmidt, R. A., 17–18, 18, 167
Schöner, G., 172
Rademaker, G. G. J., 43, 105 Sechenov, I. M., 58, 224
Ramón y Cajal, S., 155 Second-order vestibular neurons, 68, 70, 159,
Readiness potential, 208, 223 159
Recce, M., 133 Seeing. See Sight
Receptors. See Sensory receptors Segregation, of sensory signals, 42–43
Reciprocal innervation, 193 Selfishness, 265
Re-excitation (recurring excitation loop), Self-motion, perception of, 52–56, 218
47 Semicircular canals: description of, 26, 27, 32–
Reference frames. See Frames of reference 33; function of, 33–38, 44, 66–67, 92; ge-
Reflex actions, 10, 11–13 ometry of, 33–38, 60, 65, 71, 100–101; loca-
Regularity, 255 tion of, 143
Relative frames of reference, 110–112 Semiology, qualitative vs. quantitative, 217
Renshaw neurons, 193 Senses: ambiguity of, 90–91; function of, 25;
Representation, 12, 21–22 number of, 5, 262; revision of meaning of,
Residual vision, 51 263–264. See also Hearing; Sight; Smell;
Response, 11. See also Reflex actions Taste; Touch
Retina: function of, 60, 61 Sensory conflict, 251
INDEX •
335
Sensory receptors: adaptation of, 233–234; Sully, J., 242
configurations of, 5, 37, 107–109, 108, 130; Superior colliculus (SC): description of, 78;
location of, 26, 27; processing information function of, 61, 77–81; and inhibitions,
from, 19–20 199–200; multisensory convergences in, 59,
Sensory thalamus, multisensory convergences 80–81
in, 59 Superior medial temporal sulcus, 176
Sensory thresholds, 190 Superior temporal sulcus (STS), 87
Servo systems, theory of, 47, 49, 174, 183 Supplemental eye field (SEF), 209
Set (Einstellung), 223–224 Supplementary motor area (SMA), 208–209,
Shadlen, M. N., 62 225
Shepard, R. N., 132, 175–177, 253–254 Symmetry: perception of, 73, 228–229, 240;
Sherrington, G. S., 11, 27, 47, 216, 223, 227 preference for, 93
Short-term memory, 127–128 Synergies: coordination of, 162–164; defini-
Sight, 25; and hearing, 77–82; segregation of tion of, 44, 154–155, 164
sensory signals in, 42–43; in sense of move- Synergy, principle of, 154–155, 164, 224
ment, 26, 33, 50–56, 60–64; and touch, 83– Szentagothai, J., 46
89. See also Gaze
Simulation: definition of, 22–24, 37; of mus- Tactile perception. See Touch
cle contractions, 28–29 Targets, tracking of, 62–64
Skeletal geometry, 143–144 Taste, 25
Skin. See Cutaneous (skin) receptors; Touch Tecto-reticulo-spinal neurons (TRSNs), 79,
Slotine, J.-J., 31, 138, 171 199–200
Smell, 25 Temporal encoding, theory of, 264
Sokolov, E. N., 16, 190 Temporal selection, pauses in, 199
Somatoparaphrenia, 75 Temporal “windows,” 82
Somatotopy, 208 Temporonasal direction (orientation), 64
Sound. See Hearing Tensor: brain as, 48–49; definition of, 48
Space: extrapersonal, 98–99, 170; impaired Territorialism, 265
perception of, 74–76; personal, 98 Teuber school, 11
Spatial (cognitive) maps, 128–130 Thalamo-cortical circuitry, 15–16, 17
Spatial memory: definition of, 116–117; neu- Thalamus (Th): and body schema, 230; func-
ral basis of, 73–74, 126–136; types of, 117– tion of, 73, 166–167; inhibitions in, 193
126 Thales of Miletus, 256
Spatial neglect, 74–76, 100 Thermoreceptors, 30
Spatial representation, 70 θ rhythm, 132–133
Spatial selection, 200 Thomas, André, 101, 217
Stabilized vision, 219 Tilt, of head, 38–42, 72–73, 92
Stationary locomotion, 165, 182 Time, and coherence, 90
Stein, B. E., 78, 82 Time shifts, 81–82
Steps, initiation of, 225 Time τ-to-collision, 168–169
Stimulus, neuronal model of, 190–191 Toads, 166–168
Stimulus-response paradigm, 50, 139, 180 Topographic memory, 117–118
Strategies, definition of, 154–155, 224, 263 Topokinetic (topokinesthetic) memory, 125
Subcortical vision, 50 Touch: description of, 25, 29–31; and percep-
Subjective vertical, 41, 104, 218–219 tion of space, 105–107; and sight, 83–89
Substantia nigra pars reticulata (SNpr), 200, Trajectories, prediction of, 132–134
211 Transfer functions, 217
Substitution, and adaptation, 234–238 Translational memory, 122–124
Subthalamic nucleus (STN), 201 Translations, detection of, 42–43, 176
336 •
INDEX
Turvey, M. T., 59 Vestibulospinal reflexes, 43, 229
Two-thirds power law, 140, 146–147, 150–151 Vibrational sensitivity, 84
Vibrations, drawing with, 249–250
Umwelt (environment), 175 Vision. See Gaze; Sight
Utricle: description of, 26, 27, 38, 39; function Visually impaired children, teaching speed
of, 33, 38–41 sports to, 51
Visual memory, 196
Vection, perception of, 52–56, 218 Visual receptors: as necessary for sensing
Velocity: and coherence, 90; and isochrony movement, 26, 33, 50–56
principle, 147; sense of, 27–29, 33, 66–67; Visual vertical, 103
and vection, 53 Viviani, P., 9, 11, 29, 113, 146, 147, 152,
Ventriloquism, 58 177
Vertical, as multisensory construction, 102– VLSI (Very Large-Scale Integration), 22
104 von Holst, E., 69, 183
Vertigo, 69–70, 253 von Uexküll, J., 1, 175, 255
Vertigo a stomacho laeso, 105
Vestibular memory, 196 Waiter, problem of, 174–175
Vestibular memory-contingent saccades, 120– Walton, K. D., 179
121 Waterfall illusion, 55, 252
Vestibular nuclei, 67–68 Weber, E. H., 139
Vestibular (inner ear) receptors: description Weber, W. E., 139
of, 26, 27, 32–43; discovery of, 32; func- Weiskrantz, L., 51
tions of, 43–49; and visual movement, 64– Wiener, N., 139
69 Working memory, 6, 264
Vestibular signals, ambiguity of, 33, 40, 41–42 Wurtz, R. E., 52
Vestibulo-nuchal reflex, 48–49, 65
Vestibulo-ocular reflex, 44–46, 45, 48–49, 69, Yarbus, A. L., 196
155, 159, 234; inhibition of, 193, 197, 211; Young, L., 52, 174
suppression of, 236, 237
INDEX •
337

The Brain's Sense of Movement - Alan Berthoz

Uploaded by

Copyright:

Available Formats

The Brain's Sense of Movement - Alan Berthoz

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

The Brain's Sense of Movement - Alan Berthoz

Uploaded by

Copyright:

Available Formats

PERSPECTIVES IN COGNITIVE NEUROSCIENCE

HARVARD UNIVERSITY PRESS

all rights reserved

Printed in the United States of America

First published in France as Le Sens du Mouvement

Copyright © Editions Odile Jacob, 1997

Published with the assistance of the French Ministry of Culture–CNL

Pages 327–328 constitute an extension of the copyright page

Library of Congress Cataloging-in-Publication Data

 Perception Is Simulated Action 

 The Sense of Movement: A Sixth Sense? 

 A Memory for Predicting 

 Natural Movement 

 Synergies and Strategies 

 The Look that Investigates the World 

 Visual Exploration 

 The Disoriented Brain: Illusions Are Solutions 

 Architects Have Forgotten the Pleasure of Movement 

Conclusion: Toward a Tolerant Perception 

In the beginning was the Deed.

We may never succeed in combating these attitudes if we do not describe

He hesitates again, then continues:

We shall find that wherever it is a question of this idea[,] Claude Ber-

This book blends analytical considerations of the components of percep-

The Motor Theory of Perception

Figure 1.1. William James’s concept of an anticipatory neuronal pathway.

S is excited and activates a motor neuron M that induces a muscle contraction.

The Concept of Acceptor of the Results of Action

PERCEPTION IS SIMULATED ACTION •

PERCEPTION IS SIMULATED ACTION •

2 “Internally controlled element”

Figure 1.3. Thalamo-cortical circuits accounting for perceptual coherence

Memory Predicts the Consequences of Action

Past Past sensory

proposes the following definition of a schema (Figure 1.4). When a person

Simulation, Emulation, or Representation?

Theories that describe perception as the construction of internal repre-

Bouveresse suspects Helmholtz of dualist tendencies:

If the transformation of sensory excitation into psychic representations

PERCEPTION IS SIMULATED ACTION •

In choosing the word “simulation,” can I avoid these traps? I am certainly

Operators on Discrepancy Conservative

The Sense of Position and of Velocity

The Sense of Touch

The Sense of Effort: Does the Brain Use Composite Variables

The Vestibulary System: An Inertial Center?

The brain is thus capable of recognizing movements of body segments, select-

A Basic Euclidean Frame of Reference

The Distinction between Tilt and Acceleration

A Fundamental Dissociation: Rotation and Translation

The Functions of the Vestibular System

The Concept of the Neural Integrator

The study of the role of vision in perception keeps an enormous number of

Visually Impaired Children Playing Basketball

We perceive things as a whole, rather than what some may refer to as a

How Vision Detects Movement

LGN Parvocellular layer Form

Movement is detected mainly in the retinal periphery. Three kinds of gan-

Ocular Pursuit Predicts the Movement of Targets