Kinds of Minds Toward An Understanding of Consciousness - Daniel C Dennett

KINDS OF MINDS
KINDS OF MINDS
Toward an Understanding of Consciousness
D A N I E L C. D E N N E T T
• BasicBooks
A Division of HarperCollinsPublishers
The Science Masters Series is a global publish
ing venture consisting of original science books
written by leading scientists and published by
a worldwide team of twenty-six publishers
assembled by John Brockman. The series was
conceived by Anthony Cheetham of Orion Pub
lishers and John Brockman of Brockman Inc., a
New York literary agency, and developed in
coordination with BasicBooks.
The Science Masters name and marks are

owned by and licensed to the publisher by
Brockman Inc.
Copyright © 1996 by Daniel Dennett.
Published by BasicBooks
A Division of HarperCollinsPublishers, Inc.
All rights reserved. Printed in the United States

of America. No part of this book may be used or
reproduced in any manner whatsoever without
written permission except in the case of brief
quotations embodied in critical articles and
reviews. For information address Basic Books,
10 East 53rd Street, New York, NY 10022-5299.
Designed by Joan Greenfield
FIRST EDITION
ISBN 0-465-07350-6
96 97 98 99 oo •:•/HC 10 9 8 7 6 5 4 3 2 1
CO N T E N T S
Preface vii
What Kinds of Minds Are There? 1
Knowing Your Own Mind 1
We Mind-Havers, We Minders 3
Words and Minds 8
The Problem of In comm unica tive Minds 12
2 Intentionality: The Intentional Systems Approach 19
Simple Beginnings : Th e Birth of Agency 19
Adopting th e In ten tional Stance 27
Th e Misguided Goal of Propositional Precision 41
Original and Derived In ten tionality 50
3 The Body and Its Minds 57
From Sensitivity to Sen tience? 57
The Media and th e Messages 65
"My Body Has a Mind of Its Own!" 73
v
vi CONTENTS
4 How Intentionality Came into Focus 81
Th e Tower of Generate-and- Test 81
The Search for Sen tien ce: A Progress Report 93
From Phototaxis to Metaphysics 98
s The Creation of Thinking 119
Un thinking Natural Psychologists 119
Making Things to Think With 134
Talking to Ourselves 147
6 Our Minds and Other Minds 153
Our Consciousn ess, Their Minds 153
Pain and S uffering: Wha t Matters 161
Further Reading 169
B ibliography 1 75
Index 1 80
P R E F A CE
I am a philosopher, not a scientist, and we philosophers are

better at questions than answers . I haven't begun by insult
ing myself and my discipline , in spite of first appearances.
Finding better questions to ask, and breaking old habits and
traditions of asking , is a very difficult p art of the grand
human project of understanding ourselves and our worl d .
Philosophers c a n make a fine contribution to this investiga
tion , exploiting their professionally honed talents as ques
tion critics, provided they keep an open mind and restrain
themselves from trying to answer all the questions from
" obvious" first principles . There are many ways of asking
questions about different kinds of minds, and my way-the
way I will introduce in this book-changes almost daily, get
ting refined and enlarged, corrected and revised, as I learn of
new discoveries , new theories, new problems. I will intro
duce the set of fundamental assumptions that hold my way
together and give it a stable and recognizable pattern , but the
most exciting parts of this way are at the changeable fringes
of the pattern , where the action is. The main point of this
book is to present the questions I'm asking righ t n ow-and
some of them will probably lead nowhere , so let the reader
beware. But my way of asking questions has a pretty good
track record over the years , evolving quite smoothly to incor
porate new discoveries, some of which were provoked by
vii
viii PREFACE
my earlier questions. Other philosophers have offered rival

ways of asking the questions about minds , but the most influ
ential of these ways , in spite of their initial attractiveness ,
lead to self-contradictions, quandaries, or blank walls of mys
tery, as I will demonstrate. So it is with confidence that I rec
ommend my current candidates for the good questions .
Our minds are complex fabrics, woven from many differ
ent strands and incorporating many different designs. Some
of these elements are as old as life itself, and others are as
new as today's technology. Our minds are just like the minds
o f other animals in many respects and utterly unlike them in
others . An evolutionary p ersp ective can help us see how and
why these elements of minds came to take on the shapes
they have , but no single straight run through time, " from
microbes to man , " will reveal the moment of arrival of each
new thread. So in what follows I have had to weave back
and forth between simple and complex minds , reaching
back again and again for themes that must be added, until
eventually we arrive at something that is recognizably a
human mind. Then we can look back, one more time , to sur
vey the differences encountered and assess some of their
implications.
E arly drafts of this book were presented as the Agnes
Cuming Lectures at University College, Dublin, and in my
p ublic l ectures as Erskine Fellow at Canterbury University,
Christchurch, New Zealand , in May and June of 1 9 9 5 . I want
to thank the faculty and students at those institutions , whose
constructive discussions helped make the final draft almost
unrecognizably different , and (I trust) better. I also want to
thank Marc Hauser, Alva Noe, Wei Cui , Shannon Densmore ,
Tom Schuman , Pascal Buckley, Jerry Lyons , Sara Lippincott ,
and my students in " Language and Mind" at Tufts, who read
and vigorously criticized the p enultimate draft.
Tufts University
December 2 0 , 1 995
CH A P T E R I
W H AT K I N D S O F M I N D S A R E T H E R E ?
K N OW I N G YO U R OW N M I N D
Can we ever really know what is going on in someone else's

mind? Can a woman ever know what it is like to be a man?
What experiences does a baby have during childbirth? What
experiences , if any, does a fetus have in its mother's womb?
And what of nonhuman minds? What do horses think about?
Why aren 't vultures nauseated by the rotting carcasses they
eat? When a fish has a hook sticking through its lip , does it
hurt the fish as much as it would hurt you , if you had a hook
sticking through your lip? Can spi ders think , or are they just
tiny robots , mindlessly making their elegant webs? For that
matter, why couldn't a robot-if it was fancy enough-be
conscious? There are robots that can move around and
manipulate things almost as adeptly as spiders; could a more
complicated robot feel pain , and worry about its future , the
way a person can? Or is there some unbridgeable chasm sep
arating the robots (and maybe the spiders and insects and
other " clever" but mindless creatures) from those animals
that have minds? Could it be that all animals except human
beings are really mindless robots? Rene Descartes notoriously
2 K I N DS O F M I N DS
maintained this in the seventeenth century. Might he have

been dead wrong? Could it be that all animals , and even
p l ants-and even bacteria-have minds?
Or, to swing to the other extreme , are we so sure that all
human beings have minds? Maybe (to take the most extreme
case of all) you're the only mind in the universe; maybe
everything else, including the apparent author of this book,
is a mere mindless machine. This strange idea first occurred
to me when I was a young child, and perhaps it did to you as
well. Roughly a third of my students claim that they, too ,
invented it on their own and mulled it over when they were
children. They are often amused to learn that it's such a
common philosophical hypothesis that it has a name-solip
sism (from Latin for " myself alone" ) . Nobody ever takes
solipsism seriously for long , as far as we know, but it does
raise an imp ortant challenge : if we know that solipsism is
silly-if we know that there are other minds-how do we
know?
What kinds of minds are there? And how do we know?
The first question is about what exists-about on tology, in
philosophical p arlance; the second question is about our
knowledge-about epistemology. The goal of this book is not
to answer these two questions once and for all , but rather to
show why these questions have to be answered together.
Philosophers often warn against confusing ontological ques
tions with epistemological questions. What exists is one
thing, they say, and what we can know about it is something
else. There may be things that are comp letely unknowable to
us , so we must be careful not to treat the limits of our knowl
e dge as sure guides to the limits of what there is. I agree that
this is good general advice , but I will argue that we already
know enough about minds to know that one of the things
that makes them different from everything else in the uni
verse is the way we know about them. For instance , you
know you have a mind and you know you have a brain , but
W H A T K I N D S O F M I N D S A R E T H E R E? 3
these are different kinds of knowledge . You know you have a

brain the way you know you have a spleen: by hearsay.
You've never seen your spleen or your brain (I would bet) ,
but since the textbooks tell you that all normal human
beings have one of each, you conclude that you almost cer
tainly have one of each as well. You are more intimately
acquainted with your mind-so intimately that you might
even say that you are your mind. (That's what Descartes
said: he said he was a mind, a res cogitans, or thinking
thing . ) A book or a teacher might tell you what a mind i s , but
you wouldn't have to take anybody's word for the claim that
you had one. If it occurred to you to wonder whether you
were normal and had a mind as other people d o , you would
immediately realize, as Descartes pointed out, that your very
wondering this wonder demonstrated beyond all doubt that
you did indeed have a mind.
This suggests that each of us knows exactly one mind
from the inside, and no two of us know the same mind from
the inside. No other kind of thing is known about in that
way. And yet this whol e discussion so far has been con
ducted in terms of how we know-you and I . It presupposes
that solipsism is false. The more we-we-reflect on this
presupposition, the more unavoidable it appears . There
couldn't be just one mind-or at least not just one mind like
our minds .
W E M I N D- H AV E R S , W E M I N D E R S
If we want to consider the question of whether nonhuman

animals have mind s , we have to start by asking whether they
have minds in some regards like ours , since these are the
only minds we know anything about-at this point. (Try ask
ing yourself whether nonhuman animals have flurbs. You
4 KINDS OF MINDS
can 't even know what the question i s , if you don't know
what a flurb is supposed to be. Whatever else a mind i s , it is
supposed to be something like our minds; otherwise we
wouldn't call it a mind. ) S o our minds , the only minds we
know from the outset , are the standard with which we must
begin. Without this agreement, we'll just be fooling our
selves , talking rubbish without knowing it.
When I address you , I include us both in the class of
mind-havers . This unavoidable starting p oint creates, or
acknowle dges , an in-group , a class of privileged characters ,
set off against everything else in the universe. This is almost
too obvious to notice , so deeply enshrined is it in our think
ing and talking , but I must dwell on it. When there 's a we,
you are not alone; solipsism is false; there 's company pres
ent. This comes out p articularly clearly if we consider some
curious variations :
" We left Houston at dawn, headin ' down the road-just

me and my truck. "
Strange . If this fellow thinks his truck is such a worthy com

p anion that it deserves shelter under the umbrella of " we , "
h e must b e very lonely. Either that, o r his truck must have
been customized in ways that would be the envy of roboti
cists everywhere . In contrast , " we-just me and my dog "
doesn' t startle us at all, but " we-just me and my oyster " is
hard to take seriously. In other words , we 're pretty sure that
dogs have minds , and we're dubious that oysters do .
Membership in the class of things that have minds pro
vides an all-important guarantee : the guarantee of a certain
sort of moral standing. Only mind-havers can care; only
mind-havers can mind what happens. If I do something to
you that you don't want me to do , this has moral signifi
cance. It matters , because it matters to you. It may not matter
much , or your interests may be overridden for all sorts of
reasons , or (if I'm punishing you j ustly for a misdeed of

yours) the fact that you care may actually count in favor of
my deed. In any event, your caring automatically counts for
something in the moral equation. If flowers have minds ,
then what we do to flowers can matter to them , and not j ust
to those who care about what happens to flowers . If nobody
care s, then it doesn't matter what happens to flowers .
There are some who would disagree ; they would insist
that the flowers had some moral standing even if nothing
with a mind knew of or cared about their existence . Their
beauty, for instance , no matter how unappreciated , is a good
thing in itself, and hence should not be destroyed , other
things being equal. This is not the view that the beauty of
these flowers matters to God, for instance , or that it migh t
matter to some being whose presence is undetectable by us.
It is the view that the beauty matters , even th ough it matters
to no one-not to the flowers themselves and not to God or
anybody else. I remain unpersuaded, but rather than dismiss
this view outright I will note that it is controversial and not
widely shared . In contrast , it takes no special p leading at all
to get most people to agree that something with a mind has
interests that matter. That's why people are so concerned,
morally, about the question of what has a mind: any pro
posed adj ustment in the boundary of the class of mind
havers has maj or ethical significance.
We might make mistakes. We might endow mindless
things with minds , or we might ignore a mindful thing in
our midst. These mistakes would not be equal. To overat
tribute minds-to " make friends with" your houseplants or
lie awake at night worrying about the welfare of the com
puter asleep on your desk-is, at worst, a silly error of
credulity. To underattribute minds-to disregard or discount
or deny the experience, the suffering and j oy, the thwarted
ambitions and frustrated desires of a mind-having p erson or
animal-would be a terribl e sin. After all , how would yo u
feel if you were treated as an inanimate object? (Notice how

this rhetorical question appeals to our shared status as mind
havers . )
I n fact , both errors could have serious moral conse
quences. If we overattributed minds (if, for instance, we got
it into our heads that since bacteria had minds , we couldn 't
justify killing them) , this might lead us to sacrifice the inter
ests of many legitimate inte rest-holders-our friends , our
pets , ourselves-for nothing of genuine moral importance .
The abortion debate hinges on just such a quandary; some
think it's obvious that a ten-week-old fetus has a mind, and
others think it's obvious that it does not. If it does not , then
the path is open to argue that it has no more interests than ,
say, a gangrenous leg or an abscessed tooth-it can be
destroyed to save the life (or just to suit the interests) of the
mind-haver of which it is a part . If it does already have a
mind , then , whatever we decide, we obviously have to con
sider its interests along with the interests of its temporary
host. In between these extreme p ositions lies the real
quandary : the fetus will soon develop a mind if left undis
turbed, so when do we start counting its prospective inter
ests? The relevance of mind-having to the question of moral
standing is especially c lear in these cases, since if the fetus
in question is known to be anencephalic (lacking a brain) ,
this dramatically changes the issue for most people. Not for
all. (I am not attempting to settle these moral issues here , but
just to show how a common moral opinion amplifies our
interest in these questions way beyond normal curiosity. )
The dictates of morality and scientific method pull in
opp osite directions here . The ethical course is to err on the
side of overattribution , just to be safe. The scientific course
is to put the burden of proof on the attribution. As a scien
tist, you can't just declare, for instance , that the presence of
glutamate molecules (a basic neurotransmitter involved in
signaling between nerve cells) amounts to the presence o f

mind; you have to prove it, against a background in which
the " null hypothesis" is that mind is not pre sent. (Innocen t
.
un til proven guilty is the null hypothesis in our criminal
law. ) There is substantial disagreement among scientists
about which species have what sorts o f mind, but even those
scientists who are the most ardent champions of conscious
ness in animals accept this burden of proof-and think they
can meet it, by devising and confirming theories that show
which animals are conscious. But no such theories are yet
confirmed, and in the meantime we can appreciate the dis
comfort of those who see this agnosti c , wait-and-see p olicy
as jeopardizing the moral status of creatures they are sure are
conscious.
Suppose the question before us were not about the minds
of pigeons or bats but about the minds of left-handed people
or people with red hair. We would be deeply offended to b e
t o l d that it ha d yet to be proved that this category of living
thing had the wherewithal for entry into the privileged class
of mind-havers . Many people are similarly outraged by the
demand for proof of mind-having in nonhuman species , but
if they're honest with themselves they will grant that they,
too , see the need for such proof in the case of, say, j e llyfish
or amoebas or daisies ; so we agree on the principle, and
they 're just taking umbrage at its application to creatures so
very much like us. We can allay their misgivings somewhat
by agreeing that we shm�ld err well on the side of inclusive
ness in all our policies , until the facts are in; stil l , the price
you must pay for scientific confirmation o f your favorite
hypothesis about animal minds is the risk of scientific dis
confirmation.
8 KINDS OF MINDS
WO R D S A N D M I N D S
It is beyond serious dispute , however, that you and I each

have a mind. How do I know you have a mind? Because any
body who can understand my words is automatically
addressed by my pronoun " you ," and only things with
minds can understand. There are computer-driven devices
that can read books for the blind: they convert a page of visi
ble text into a stream of audible words , but they don 't under
stand the words they read and hence are not addressed by
any " you" they encounter; it passes right through them and
addresses whoever listens to-and understands-the stream
of spoken words . That's how I know that you , gentle
reader/listener, have a mind. S o do I. Take my word for it.
In fact that's what we routinely do: we take each other 's
words as settling beyond any reasonable doubt the question
of whether we each have minds. Why should words be so
convincing? B ecause they are' such powerful resolvers of
doubts and ambiguities. You see somebody coming toward
you , scowling and waving an ax. You wonder, What 's his
problem? Is he going to attack me? Is he mistaking me for
somebody else? Ask him. Perhaps he will confirm your
worst fears , or perhaps he will tell you he has given up try
ing to unlock his car (which you 're standing in front of) and
has returned with his ax to break the window. You may not
believe him when he says it's his car, not somebody else's ,
but further conversation-if you decide not to run away-is
bound to resolve your doubts and clarify the situation in
ways that would be all but impossible if you and he were
unable to communicate verbally. Suppose you try asking
him , but it turns out that he doesn't speak your language.
Perhaps you will then both resort to gestures and miming.
These techniques , used with ingenuity, will take you far, but
they're a poor substitute for language-just reflect on how
W H A T K I N D S O F M I N D S A R E T H E R E1 9
eagerly you would both seek to confirm your hard-won

understanding if a bilingual interpreter were to come along .
A few relayed questions and answers would n o t just allay
any residual uncertainty but would add details that could
not be conveyed in any other way : " When he saw you put
one hand on your chest and push out with your other han d ,
he thought y o u meant that y o u were i l l ; he w a s trying to ask
if you wanted him to take you to a doctor once h e ' d broken
the window and retrieved his -keys. That business with his
fingers in his ears was his attempt to convey a stethoscope. "
Ah , it all falls into place now, thanks to a few words .
People often emphasize the difficulty of accurate and
reliable translation between human languages. Human cul
tures, we are tol d , are too different , too " incommensurab le , "
t o permit the meanings available t o one speaker t o b e p er
fectly shared with another. No doubt translation always falls
somewhat short of perfection, but this may not matter much
in the larger scheme of things . Perfect translation may be
impossible, but good translation is achieved every day-rou
tinely, in fact. Good translation can be obj ectively distin
guished from not-so-good translation and from bad transla
tion, and it permits all human beings , regardless o f rac e ,
culture , age , gender, o r experience , to unite more closely
with one another than individuals of any other species can.
We human beings share a subjective world-and know that
we do-in a way that is entirely beyond the capacities of any
other creatures on the p lanet, because we can talk to one
another. Human beings who don 't (yet) have a language in
which to communicate are the exception, and that's why we
have a particular problem figuring out what it's like to be a
newborn baby or a deaf-mute.
Conversation unites us. We can all know a great deal
about what it 's like to be a Norwegian fisherman or a Niger
ian taxi driver, an eighty-year-old nun or a five-year-old boy
blind from birth , a chess master or a prostitute or a fighter
10 KINDS OF MINDS
pilot. We can know much more about these topics than we

can know about what it's like (if anything) to be a dolphin , a
bat , or even a chimpanzee . No matter how different from one
another we people are , scattered around the globe , we can
explore our differences and communicate about them . No
matter how similar to one another wildebeests are , standing
shoulder to shoulder in a herd , they cannot know much of
anything about their similarities , let alone their differences .
They cannot compare notes . They can have similar experi
ences , side by side , but they really cannot share experiences
the way we do.
Some of you may doubt this . Can 't animals " instinc
tively" understand each other in ways we human beings
cannot fathom? Certainly some authors have said so. Con
sider, for instance, Elizabeth Marshall Thomas , who imag
ines , in Th e Hidden Life of Dogs ( 1 9 9 3 ), that dogs enjoy a
wise understanding of their own ways . One examp l e: " For
reasons known to dogs but not to us , many dog mothers
won't mate with their sons ." (p. 7 6 ) . Their instinctive resis
tance to such inbreeding is not in doubt , but what gives her
the idea that dogs have any more insight into the reasons for
their instincts than we have into ours? There are many
things we feel strongly and instinctively disinclined to do ,
with no inkling about why we feel that way. To suppose
without proof that dogs have more insight into their urges
than we do is to ignore the null hypothesis in an unaccept
able way-if we are asking a scientific question. As we shall
see , very simple organisms may be attuned to their environ
ments and to each other in strikingly apt ways without hav
ing the slightest appreciation of their attunement. We
already know from conversation, however, that people are
typ ically capable of a very high order of understanding of
themselves and others .
Of cours e , we can be fooled. People often emphasize the
W H A T K I N D S O F M I N D S A R E T H E R E? 1 1
difficulty of determining whether a speaker is sincere .

Words , by being the most powerful tools of communication,
are a l so the mos t p owerful t o o l s of deception an d manipula
tion. But while it may be easy to lie , it's almost as easy to
catch a liar-especially when the lies get large and the logis
tical problem of maintaining the structure of falsehood over
whelms the liar. In fantasy, we can conj ure up infinitely
powerful deceivers , but the deceptions that are " possible in
principle" to such an evil demon can be safely ignored in the
real worl d. It would be just too difficult to make up that
much falsehood and maintain it consistently. We kn o w that
people the world over have much the same likes and dis
likes , hopes and fears . We know that they enjoy recollecting
favorite events in their lives . We know that they all have rich
episodes of waking fantasy, in which they rearrange and
revise the details deliberately. We know that they have
obsessions , nightmares , and hallucinations. We know that
they can be reminded by an aroma or a melody of a specific
event in their lives , and that they often talk to themselves
silently, without moving their lips. Long before there was
scientific psychology, long before there was meticulous
observation of and experimentation on human subj ects , this
was all common knowledge . We have known these facts
about people since ancient times , because we have talked it
over with them , at great length . We know nothing compara
ble about the mental lives of any other species , because we
can't talk it over with them. We may think we know, but it
takes scientific investigation to confirm or refute our tradi
tional hunches.
12 KINDS OF MINDS
T H E P RO B L E M O F
I N CO M M U N I CAT I V E M I N D S
It 's very hard to tell what somebody is thinking who won 't
discuss it-or who can ' t , for one reason or another. But we
normally suppose that such incommunicative folks are
indeed thinking-that they do have minds-even if we can't
confirm the details. This much is obvious , if only because
we can readily imagine ourselves in a situation in which we
would steadfastly refuse to communicate, all the while
thinking our private thoughts, perhaps reflecting with
amusement on the difficulties that observers were having in
figuring out what , if anything , was going on in our minds .
Talking , no matter how conclusive its presence may b e , is
not necessary for having a mind. From this obvious fact we
are tempted to draw a problematic conclusion: there could
be entities who do have minds but who cannot tell us what
they 're thinking-not because they 're paralyzed or suffering
from aphasia (the inability to communicate verbally due to
localized brain damage) , but because they have no capacity
for language at all . Why do I say this is a problematic con
clusion?
First let's consider the case to be made in its favor. Surely,
tradition and common sense declare , there are minds with
out language. Surely our ability to discuss with others what
is going on in our minds is just a p eripheral talent , in the
sense in which one speaks of a computer's laser printer as a
peripheral device (the computer can go right on computing
without a printer attached ) . Surely nonhuman animals-at
least, some of them-have mental lives . Surely human
infants before they acquire language , and human deaf
mutes-even those rare deaf-mutes who have never acquired
even sign language-have minds. Surely. These minds may
doubtless differ in many hard-to-fathom ways from our
minds-the minds of those who can understand a conversa

tion such as this-but surely they are minds. Our royal road
to the knowledge of other minds-language-does not
extend to them, but this is just a limitation on our knowl
edge , not a limitation on their minds . The prospect arises,
then , that there are minds whose contents are systematically
inaccessible to our curiosity-unknowable, uncheckable,
impenetrable by any investigation.
The traditional response to this prospect is to embrace it.
Yes indeed, minds are the ultimate terra in cognita , beyond
the reach of all science and-in the case of languageless
minds-beyond all empathetic conversation as well. So
what? A little humility ought to temper our curiosity. Don't
confuse ontological questions (about what exists) with epis
temological questions (about how we know about it) . We
must grow comfortable with this wonderful fact about what
is off-limits to inquiry.
But before we get comfortable with this conclusion, we
need to consider the implications of some other facts about
our own case that are just as obvious . We find that we often
do clever things without thinking at all; we do them " auto
matically," or " unconsciously." What is it like , for instance ,
t o use information about the optic fl o w of shapes in periph
eral vision to adjust the length of your stride as you walk
across rough terrain? The answer is, It isn't like anything.
You can't pay attention to this process even if you try. What
is it like to notice, while sound asleep , that your left arm has
become twisted into a position in which it is putting undue
strain on your left shoulder? Like nothing; it is not part of
your experience. You swiftly and unconsciously shift to a
more " comfortable" position, without any interruption of
your sleep . If we are asked to discuss these p utative parts of
our mental lives , we draw a blank; whatever happened in us
to govern these clever behaviors wasn 't a part of our mental
lives at all. S o another prospect to consider is that among the
14 KINDS OF MINDS
creatures who lack language , there are some that do not have
minds at all , but do everything " automatically " or "uncon
sciously. "
The traditional response to this prospect , too , is to
embrace it. Yes indeed , some creatures entirely lack minds.
Surely bacteria are mindles s , and s o , probably, are amoebas
and starfish. Quite p ossibly even ants , for all their clever
activity, are mere mindless automata, trundling about in the
world without the s lightest experience or thought. What
about trout? What about chickens? What about rats? We may
never be able to tell where to draw the line between those
creatures that have minds and those that do not , but this is
j ust another aspect of the unavoidable limitations on our
knowledge. Such facts may be systematically unknowable,
not just hard to uncover.
Here , then , are two sorts of supposedly unknowable facts :
facts about what is going on in those who have minds but no
way of talking about their thoughts, and facts about which
creatures have minds at all. These two varieties of off-limits
ignorance are not equally easy to accept. The differences
between min ds might be differences whose maj or outlines
were readily discernible to objective observers but whose
minor details became harder and harder to determine-a
case of diminishing returns for labor invested. The unknown
leftovers would not be mysteries but j ust inevitable gaps in a
richly informative but finite catalog of similarities and dif
ferences. The differences between minds would then be like
the differences between languages , or styles of music or art
inexhaustible in the limit, but approachable to any degree of
approximation you like. But the difference between having a
mind and not having a mind at all-between being some
thing with its own subj ective p oint of view and being some
thing that is all outside and no inside , like a rock or a dis
carded sliver of fingernail-is apparently an all-or-nothing
difference. It is much harder to accept the idea that no
amount of further investigation will ever tell us wh eth er

there is anyone th ere to care inside a lobster's shell , for
instance , or behind the shiny fac;; a de of a robot.
The suggestion that such a morally important sort of fact
could be systematically unknowable by us is simply intoler
able. It means that no matter what investigations we con
ducted , we might , for all we could know, be sacrificing the
genuine moral interests of some for the entirely illusory ben
efit of mindless others . Unavoi dable ignorance of the conse
quences is often a legitimate excuse when we find we have
unwittingly produced some harm in the worl d , but if we
must declare ourselves at the outset to be unavoidably igno
rant of the very basis of all moral thinking , morality becomes
a sham . Fortunately, this conclusion is as incredible as it is
intolerable. The claim that , say, left-handed people are
unconscious zombies that may be dismantled as if they were
bicycles is preposterous . S o , at the other extreme , is the
claim that bacteria suffer, or that carrots mind being p lucked
unceremoniously from their earthy homes . Obviously, we
can know to a moral certainty (which is all that matters) that
some things have minds and other things don't.
But we don't yet know how we know these facts; the
strength of our intuitions about such cases is no guarantee of
their reliability. Consider a few cases, beginning with this
remark by the evolutionist Elaine Morgan :
The heart-stopp ing thing about the new-born is that , from

minute one , there is somebody there. Anyone who bends
over the cot and gazes at it is being gazed back at. ( 1 99 5 ,
p . 99)
As an observation about how we human observers

instinctively react to eye contact, this is right on target , but it
thereby shows how easily we can be misled. We can be
fooled by a robot , for instance. At the Artificial Intelligence
16 KINDS OF M I N DS
Lab at MIT, Rodney Brooks and Lynn Andrea Stein have

assembled a team of roboticists and others (myself included)
to build a humanoid robot, named Cog. Cog is made of metal
and silicon and glas s , like other robots , but the design is so
different , so much more like the design of a human being ,
that Cog may someday become the world 's first conscious
robot. Is a conscious robot possible? I have defended a the
ory of consciousness , the Multiple Drafts Model ( 1 9 9 1 ) , that
implies that a conscious robot is possible in principle , and
Cog is being designed with that distant goal in mind. But
Cog is nowhere near being conscious yet. Cog cannot yet see
or hear or feel at all , but its bodily parts can already move in
unnervingly humanoid ways . Its eyes are tiny video cam
eras , which saccade-- d art-to focus on any p erson who
enters the room and then track that p erson as he or she
moves . B eing tracked in this way is an oddly unsettling
experience, even for those in the know. Staring into Cog 's
eyes while Cog stares mindlessly back can be quite " heart
stopping" to the uninitiated , but there is nobody there-not
yet , in any case. Cog 's arm s , unlike those of standard robots
both real and cinematic , move swiftly and flexibly, like your
arms; when you press on Cog 's extended arm , it responds
with an uncannily humanoid resistance that makes you
want to exclaim , in stock horror-movie fashion, " It 's alive !
It 's alive ! " It isn't, but the intuition to the contrary is potent.
While we're imagining arms, let 's consider a variation
with a different moral : A man's arm has been cut off in a ter
rible accident, but the surgeons think they can reattach it.
While it i s lying there , still soft and warm , on the op erating
tab l e , does it feel pain? (If s o , we should inject some novo
caine into it-especially if we p lan to use a scalpel to cut
back any tissue on the amputated arm before attempting the
reunion. ) A silly suggestion, you reply; it takes a mind to
feel p ain , and as long as the arm is not attached to a body
with a mind , whatever you do to the arm can't cause suffer-
ing in any mind. But p erhaps the arm has a mind of its own.
Perhap s it has always had one but has j ust been unable to
talk to us about it! Wel l , why not? It does have a substantial
number of nerve cells in it, still firing away. If we found a
whole organism with that many active nerve cells in it , we
would be strongly inclined to suppose that it was capable of
experiencing pain, even if it couldn 't express itself in terms
we could understand. Here intuitions collide : arms don't
have minds , in spite of containing plenty of the processes
and materials that tend to persuade us that some nonhuman
animals do have minds .
Is it behavior that counts? Suppose you pinched the
thumb of the amputated arm and it pinched you back !
Would you then decide to give it novocaine? If not, why not?
Because its reaction would have to be an " automatic" reflex?
How can you be so sure? Is it something about the organiza
tion of those nerve cells that makes the difference?
These puzzle cases are fun to think about, and we learn
important facts about our naive concepts of mind when we
try to figure out why our intuitions line up the way they do,
but there must be a better way of investigating kinds of
minds-and nonminds that might fool us. The defeatist con
viction that we will never know should be postponed indefi
nitely, saved as a last-gasp conclusion to be reached only
after we have actually exhausted all other avenues and not
just imagined doing so. There may be surprises and illumi
nations awaiting us.
One prospect to consider, whether or not in the end we
rule it out, is that p erhaps language is not so p eripheral to
minds after all. Perhap s the kind of mind you get when you
add language to it is so different from the kind o f mind you
can have without language that calling them both minds is a
mistake. Perhaps , in other words , our sense that there are
riches in the minds of other creatures-riches inaccessible to
us but not, of course, to them-is an illusion. The philosopher
18 KINDS OF MINDS
Ludwig Wittgenstein famously sai d , "If a lion could talk, we

could not understand him . " ( 1 9 5 8 , p. 2 2 3 ) That 's one possi
bility, no doubt , but it diverts our attention from another
p ossibility: if a lion could talk, we could understand him
j ust fine-with the usual sorts of effort required for transla
tion between different languages-but our conversations
with him would tell us next to nothing about the minds of
ordinary lions , since his language-equipped mind would be
so different. It migh t be that adding language to a lion's
" mind" would be giving him a mind for the first time ! Or it
might not. In either case , we should investigate the prospect
and not just assum e , with tradition, that the minds of non
sp eaking animals are really rather like ours .
If we are to find some alternative path of investigation,
instead of just relying uncritically on our pretheoretical intu
itions , how might we begin? Let's consi der the historical ,
evolutionary path. There haven't always been minds . We
have minds, but we haven' t existed forever. We evolved from
beings with simpler minds (if minds they were) , who
evolved from beings with still simpler candidates for minds .
And there was a time, four or five billion years ago , when
there weren't any minds at all , simple or complex-at least ,
not on this planet. Which innovations occurred in what
order, and why? The major steps are clear, even if the details
about dates and places can be only speculative. Once we've
told that story, we will at least have a framework in which to
try to place our quandaries. Perhap s we will want to distin
guish classes of pseudominds , or protominds , or semiminds ,
or hemi-semi-demi-minds from the real thing. Whatever we
decide to call these ancestral arrangements , perhaps we can
agree upon a scale on which they mount, and the conditions
and principles that created the scale in the first place. The
next chapter develops some tools for this investigation.
CH A P T E R 2
I NT E N T I O N A L I TY:
T H E I N T E N T I O N A L S Y S T E M S A P P R O A CH
I notice something and seek a reason for it: this

means originally: I seek an intention in it, and
above all someone who has intentions , a sub
j ect , a doer: every event a deed-formerly one
saw intentions in all event s , this is our oldest
habit. Do animals also possess it?
Friedrich Nietzsche , The Will to Power
SIM PLE B EG I N N I NGS:

T H E B I RT H O F A G E N CY *
N o grain of sand has a mind; a grain of sand is t o o simp l e .

Even simpler, no carbon atom or water molecule h a s a mind.
I expect no serious disagreement about that. But what about
larger molecules? A virus is a single huge molecul e , a macro
molecule composed of hundreds of thousands or even mil
lions of parts , depending on how small the p arts are that we
*Portions of this section are drawn from my 1995 book, Darwin's

Dangerous Idea, with revisions.
19
count. These atomic-level parts interact , in their obviously

mindless ways , to produce some quite striking effects . Chief
among these effects , from the point of view of our investiga
tion , is self-replication . Some macromolecules have the
amazing ability, if left floating in a suitably well-furnished
medium , to mindlessly construct and then shed exact-or
nearly exact-copies of themselves. DNA and its ancestor,
RNA, are such macromolecules; they are the foundation of all
life on this planet and hence a historical precondition for all
minds-at least, all minds on this p lanet. For about a billion
years before simple single-celled organisms appeared on
earth , there were self-replicating macromolecules , ceaselessly
mutating, growing , even repairing themselves, and getting
better and better at it-and replicating over and over again.
This is a stupendous feat, still w e l l beyond the capacity
of any existing robot. Does that mean that such macromole
cules have minds like ours? Certainly not. They're not even
alive-they're just huge crystals , from the p oint of view of
chemistry. These gigantic molecules are tiny machines
macromolecular nanotechnology. They are , in effect, natural
robots. The possibility in principle of a self-replicating robot
was mathematically demonstrated by John van Neumann ,
one of the inventors of the computer, whose brilliant design
for a nonliving self-replicator anticipated many of the details
o f design and construction of RNA and DNA.
Through the microscope of molecular biology, we get to
witness the birth of agency, in the first macromolecules that
have enough complexity to perform actions, instead of just
lying there having effects. Their agency is not fully fledged
agency like ours . They know not what they do. We , in con
trast , often know full well what we do. At our best-and at
our worst-we human agents can perform in ten tional
actions , after having deliberated consciously about the rea
sons for and against. Macromolecular agency is different;
I NTENTIONALITY 21
there are reasons for what macromolecules do , but the

macromolecules are unaware of those reasons . Their sort of
agency is nevertheless the only p ossible ground from which
the seeds of our kind of agency could grow.
There is something alien and vaguely repellent about the
quasi agency we discover at this level-all that purposive
hustle and bustle, and yet " there's nobody home. " The molec
ular machines perform their amazing stunts , obviously
exquisitely designed and j ust as obviously none the wiser
about what they are doing. Consider this account of the activ
ity of an RNA phage-a replicating virus and a mo dern-day
descendant of the earliest self-replicating macromolecules:
First of all , the virus needs a material in which to pack

and protect its own genetic information. Secondly, it
needs a means of introducing its information into the
host cel l . Thirdly, it requires a mechanism for the specific
replication of its information in the presence of a vast
excess of host cell RNA. Finally, it must arrange for the
proliferation of its information , a process that usually
leads to the destruction of the host cell. . . . The virus
even gets the cell to carry out its replication; its only con
tribution is one protein factor, specially adapted for the
viral RNA. This enzyme does not become active until a
" password" on the viral RNA is shown . When it sees this ,
i t repro duces the viral RNA with great efficiency, while
ignoring the very much greater number of RNA molecules
of the host cell. Consequently the cell is soon flooded
with viral RNA. This is packed into the viru s ' coat pro
tein , which is also synthesized in large quantities, and
finally the cell bursts and releases a multitude of progeny
virus particles. All this is a programme that runs automat
ically and is rehearsed down to the smallest detail .
(Eigen , 1 9 9 2 , p . 4 0 )
22 KINDS OF MINDS
The author, the molecular biologist Manfred Eigen, has

helped himself to a rich vocabulary of agency words : in
order to repro duc e , the virus must " arrange for" the prolifer
ation of its information, and in furthering this goal it creates
an enzyme that " sees" its password and " ignores" other mol
ecules. This is poetic license, to be sure; these words have
had their meanings stretched for the occasion. But what an
irresistible stretch ! The agency words draw attention to the
most striking features of the phenomena: these macromole
cules are systematic. Their control systems are not just effi
cient at what they do; they are appropriately sensitive to
variation, opp ortunisti c , ingenious , devious. They can be
" fooled," but only by novelties not regularly encountered by
their ancestors.
These impersonal , unreflective , robotic, mindless little
scraps o f molecular machinery are the ultimate basis of all
the agency, and hence meaning , and hence consciousness , in
the world . It is rare for such a solid and uncontroversial sci
entific fact to have such potent implications for structuring
all subsequent debate about something as controversial and
mysterious as minds , so let's pause to remind ourselves of
these implications .
There is no l onger any serious informed doubt about this :
we are the direct descen dan ts of these .self-replicating
robots. We are mammals , and all mammals have descended
from reptilian ancestors whose ancestors were fish whose
ancestors were marine creatures rather like worms , who
descended in turn from simpler multicelled creatures several
hundred million years ago , who descended from single
celled creatures who descended from self-replicating macro
molecules, about three billion years ago. There is just one
family tree , on which all living things that have ever lived
on this planet can be found-not just animal s , but plants
and algae and bacteria as well. You share a common ancestor
with every chimpanzee, every worm , every blade of grass,
I N T E N T I O N A L I TY 23
every redwood tree . Among our progenitors , then , were

macromolecules .
To put it vividly, your great-great-. . . grandmother was a
robot ! Not only are you descended from such macromolecu
lar robots but you are composed of them: your hemoglobin
molecules , your antibodies , your neurons , your vestibular
ocular reflex machinery-at every level of analysis from the
molecular on up , your body (including your brain , of course)
is found to be composed of machinery that dumbly does a
wonderful, elegantly designed j ob .
We have ceased to shudder, perhaps , a t t h e scientific
vision of viruses and bacteria busily and mindlessly execut
ing their subversive projects-horrid little automata doing
their evil deeds. But we should not think that we can take
comfort in the thought that they are alien invaders , so unlike
the more congenial tissues that make up us. We are made of
the same sorts of automata that invade us-no special halos
of humanity distinguish your antibo dies from the antigens
they combat; your antibodies simply belong to the club that
is you, so they fight on your behalf. The billions o f neurons
that band together to make your brain are cells, the same sort
of biological entity as the germs that cause infections , or the
yeast cells that multiply in the vat when beer is fermenting
or in the dough when bread rises.
Each cell-a tiny agent that can p erform a limited number
of tasks-is about as mindless as a virus. Can it b e that if
enough of these dumb homunculi-little men-are put
together the result will be a real , conscious p erson , with a
genuine mind? According to modern science, there is no
other way of making a real p erson. Now, it certainly does not
follow from the fact that we are descended from robots that
we are robots ourselves . After all, we are also direct descen
dants of fish, and we are not fish; we are direct descendants
of bacteria, and we are not bacteria. But unless there is some
secret extra ingredient in us (which is what dualists and
vitalists used to think) , we are made of robots-or, what

comes to the same thing , we are each a collection of trillions
of macromolecular machines . And all of these are ultimately
descended from the original self-replicating macromole
cules. S o something made of robots can exhibit genuine con
sciousnes s , because you do if anything does .
To some people, all this seems shocking and unlikely, I
realize , but I suspect that they haven't noticed how desper
ate the alternatives are . Dualism (the view that minds are
composed of some nonphysical and utterly mysterious stuff)
and vitalism (the view that living things contain some spe
cial physical but equally mysterious stuff-elan vital) have
been relegated to the trash heap of history, along with
alchemy and astrology. Unless you are also prepared to
declare that the world is flat and the sun is a fiery chariot
pulled by winged horses-unless , in other words , your defi
ance of mo dern science is quite complete-you won 't find
any place to stand and fight for these obsolete ideas. So let 's
see what story can be told with the conservative resources of
science. Maybe the idea that our minds evolved from sim
pler minds is not so bad after all .
Our macromolecule ancestors (and that's exactly and
unmetaphorically what they were : our ancestors ) were
agent like in some ways, as the quotation from Eigen makes
clear, and yet in other ways they were undeniably passive ,
floating randomly around , pushed hither and yon-waiting
for action with their guns cocke d, you might say, but not
waiting h opefully or resolu tely or in ten tly. Their j aws might
have gap e d , but they were as mindless as a steel trap .
What changed? Nothing sudden. Before our ancestors got
minds , they got bodies. First, they became simple cells , or
prokaryote s , and eventually the prokaryotes took in some
invaders , or boarders , and thereby became complex cells
the eukaryotes . By this time , roughly a billion years after the
first appearance of simple cells , our ancestors were already
I NT E N T I O N A L I TY 25
extraordinarily complex machines (made of machines made

of machines) , but they still didn't have minds . They were as
passive and undirected in their traj ectories as ever, but now
they were equipped with many specialized subsystems , for
extracting energy and material from the environment and
protecting and repairing themselves when necessary.
The elaborate organization of all these coordinated p arts
was not very much like a mind. Aristotle had a name for it
or for its descendants : he called it a n u tritive soul. A nutri
tive soul is not a thing; it is not, for instance , one of the
microscopic subsystems floating around in the cytoplasm of
a cell. It is a principle of organization; it is form, not sub
stance , as Aristotle said. All living things-not only p lants
and animals but also unicellular organisms-have bodies
that require a self-regulative and self-protective organization
that can be differentially activated by different conditions .
These organizations are brilliantly designed , by natural
selection, and they are composed, at bottom, of lots of tiny
passive switches that can be turned ON or OFF by equally
passive conditions that the organisms encounter in their
wanderings.
You yourself, like all other animal s , have a nutritive
soul-a self-regulative , self-protective organization-quite
distinct from, and more ancient than , your nervous system:
it consists of your metabolic system, your immune system ,
and the other staggeringly complex systems o f self-repair
and health maintenance in your bo dy. The lines of commu
nication used by these early systems were not nerves but
blood vessels. Long before there were telephones and radi os,
there was the postal service, reliably if rather slowly trans
porting physical packages of valuable information around
the world. And long before there were nervous systems in
organisms , bodies relied on a low-tech postal system of
sorts-the circulation of fluids within the bo dy, reliably if
rather slowly transporting valuable packages of information
26 KINDS OF MINDS
to where they were needed for control and self-maintenance.

We see the descendants of this primordial postal system in
both animals and plants. In animal s , the bloo dstream carries
goods and waste , but it has also been, since the early days ,
an information highway. The motion of fluids within plants
also provides a relatively rudimentary medium for getting
signals from one p art of the plant to another. But in animals,
we c an s ee a maj or design innovation : t h e evolution of sim
ple nervous systems-ancestors of the autonomic nervous
system-capable of swifter and more efficient information
transmission but still devoted, in the main , to internal affairs.
An autonomic nervous system is not a mind at all but rather
a control system , more along the lines of the nutritive soul of
a plant, that preserves the basic integrity of the living system.
We sharply distinguish these ancient systems from our
minds , and yet , curiously, the closer we look at the details of
their op eration the more mindlike we find them to be ! The
little switches are like primitive sense organs , and the effects
that are produced when these switches are turned ON and
OFF are like intentional actions. How so? In being effects
produce d by information modulated goal seeking systems.
- , -
It is as if these cells and cell assemblies were tiny, simple

minded agen ts, specialized servants rationally furthering
their p articular obsessive causes by acting in the ways their
p erception of circumstances dictated. The world is teeming
with such entitie s , ranging from the molecular to the conti
nental in size and including not only " natural" obj ects , such
as plants , animal s , and their parts (and the parts of their
p arts) , but also many human artifacts . Thermostats , for
instance, are a familiar example of such simple pseudo
agents .
I call a l l these entities , fro m t h e simplest to the most com
plex, inten tional systems, and I call the perspective from
which their agenthood (pseudo or genuine) is made visible,
the in ten tional stance.
I NTENTIONALITY 27
A D O P T I N G T H E I N T E N T I O N A L S TA N C E
The intentional stance is the strategy of interpreting the

behavior of an entity (person, animal, artifact, whatever) by
treating it as if it were a rational agent who governed its
" choice" of " action" by a " consideration" of its "beliefs" and
" desires." These terms in scare-quotes have been stretched
out of their home use in what's often called " folk psychol
ogy," the everyday psychological discourse we use to dis
cuss the mental lives of our fellow human beings . The inten
tional stance is the attitude or perspective we routinely
adopt toward one another, so adopting the intentional stance
toward something else seems to be deliberately an thropo
morphizing it. How could this possibly be a good idea?
I will try to show that if done with care , adopting the
intentional stance is not just a good idea but the key to
unraveling the mysteries of the mind-all kinds of minds. It
is a method that exp loits similarities in order to discover dif
ferences-the huge collection of differences that have accu
mulated between the minds of our ancestors and ours , and
also between our minds and those of our fellow inhabitants
of the planet. It must be used with caution; we must walk a
tightrope between vacuous metaphor on the one hand and
literal falsehood on the other. Improper use of the inten
tional stance can seriously mislead the unwary researcher,
but properly understoo d , it can provide a sound and fruitful
perspective in several different fields , exhibiting underlying
unity in the phenomena and directing our attention to the
crucial exp eriments that need to be conducted.
The basic strategy of the intentional stance is to treat the
entity in question as an agent , in order to predict-and
thereby explain , in one sense-its actions or moves. The dis
tinctive features of the intentional stance can best be seen by
contrasting it with two more basic stances or strategies of
prediction: the physical stance and the design stance. The

physical stance is simply the standard laborious method of
the physical science s , in which we use whatever we know
about the laws of physics and the physical constitution of
the things in question to devise our prediction. When I pre
dict that a stone released from my hand will fall to the
ground , I am using the physical stance. I don 't attribute
beliefs and desires to the stone; I attribute mas s , or weight ,
to the stone , and rely on the law of gravity to yield my pre
diction. For things that are neither alive nor artifacts , the
physical stance is the only available strategy, though it can
b e conducted at various levels of detail , from the subatomic
to the astronomical . Explanations of why water bubbles
when it boils , how mountain ranges come into existence ,
and where the energy in the sun comes from are explana
tions from the physical stance. Every physical thing ,
whether designed or alive or not, is subject to the laws of
physics and hence behaves in ways that can be explained
and predicted from the physical stance . If the thing I release
from my hand is an alarm clock or a goldfish, I make the
same prediction about its downward traj ectory, on the same
basis. And even a model airplan e , or a bird , which may well
take a different traj ectory when released, behaves in ways
that obey the laws of physics at every scale and at every
moment.
Alarm clocks , being designed objects (unlike the rock) ,
are also amenable to a fancier style of prediction-predic
tion from the design stance. The design stance is a wonder
ful shortcut, which we all use all the time. Suppose someone
gives me a new digital alarm clock. It is a make and model
quite novel to me, but a brief examination of its exterior but
tons and displays convinces me that if I depress a few but
tons j ust s o, th en some hours later the alarm clock will make
a loud noise. I don't know what kind of noise it will b e , but
I NT E N T I O N ALITY 29
it will be sufficient to awaken me. I don't need to work out

the specific physical laws that explain this marvelous regu
larity; I don't need to take the thing apart , weighing its parts
and measuring the voltages . I simply assume that it has a
particular design-the design we call an alarm clock-and
that it will function prop erly, as designed. I ' m prepared to
risk quite a lot on this prediction-not my life , p erhap s , but
my waking up in time to get to my scheduled lecture or
catch a train. Design-stance predictions are riskier than
physical-stance predictions , because of the extra assump
tions I have to take on board : that an entity is designed as I
suppose it to be, and that it will operate according to that
design-that is, it will not malfunction. Designed things are
occasionally misdesigned , and sometimes they break. But
this moderate price I pay in riskiness is more than compen
sated by the tremendous ease of prediction. Design-stance
prediction, when applicable, is a low-cost , low-risk shortcut ,
enabling me to finesse the tedious application of my limited
knowledge of physics. In fact we all routinely risk our lives
on design-stance predictions : we unhesitatingly p lug in and
turn on electrical appliances that could kill us if miswired;
we voluntarily step into buses we know will soon accelerate
us to lethal speeds; we calmly press buttons in elevators we
have never been in before.
Design-stance prediction works wonderfully on well
designed artifacts , but it also works wonderfully on Mother
Nature 's artifacts-living things and their p arts. Long before
the physics and chemistry of p lant growth and reproduction
were understood, our ancestors quite literally bet their lives
on the re liability o f their design-stance knowl edge o f what
seeds were supposed to do when p l anted. If I press a few
seeds into the ground just s o , th en in a few month s , with a
modicum of further care from m e , there will be foo d here
to eat.
We have j ust seen that design-stance predictions are

risky, compared with physical-stance predictions (which are
safe but tedious to work out) , and an even riskier and swifter
stance is the intentional stance. It can be viewed, if you like ,
as a subspecies of the design stance , in which the designed
thing is an agent of sorts . Suppose we apply it to the alarm
clock. This alarm clock is my servant; if I comman d it to
wake me u p , by giving it to un derstan d a particular time of
awakening, I can rely on its internal abi lity to perceive when
that time has arrived and dutifully execute the action it has
promised. As soon as it comes to believe that the time for
noise is NOW, it will be " motivate d , " thanks to my earlier
instructions , to act accordingly. No doubt the alarm clock is
so simple that this fanciful anthropomorphism i s, strictly
speaking, unnecessary for our understanding of why it does
what it does-but notice that this is how we might explain
to a child how to use an alarm clock: " You tell it when you
want it to wake you up , and it remembers to do so, by mak
ing a loud noise."
Adoption o f the intentional stance is more useful
indeed, well-nigh obligatory-when the artifact in question
is much more complicated than an alarm clock. My favorite
example is a chess-playing computer. There are hundreds of
different computer programs that can turn a computer,
whether it's a laptop or a supercomputer, into a chess player.
For all their differences at the physical level and the design
level , these computers all succumb neatly to the same sim
ple strategy of interpretation : j ust think of them as rational
agents who wan t to win, and who know the rules and princi
ples of chess and the positions of the pieces on the board.
Instantly your problem of predicting and interpreting their
b ehavior is made vastly easier than it would be if you tried
to use the physical or the design stance . At any moment in
the chess game, simply look at the chessboard and draw up a
list of all the legal moves available to the computer when it
I N T E N T I O N ALITY 31
is its turn to play (there will usually be several dozen candi

dates) . Why restrict yourself to legal moves? Because, you
reason, it wants to play winning chess and knows that it
must make only legal moves to win, so, being rational, it
restricts itself to these. Now rank the legal moves from best
(wisest, most rational) to worst (stupidest, most self-defeat
ing) and make your predietion: the computer will make the
best move. You may well not be sure what the best move is
(the computer may " appreciate" the situation better than you
do ! ), but you can almost always eliminate all but four or five
candidate moves, which still gives you tremendous predic
tive leverage.
Sometimes, when the computer finds itself in a tough
predicament, with only one nonsuicidal move to make (a
" forced" move), you can predict its move with supreme con
fi dence. Nothing about the laws of physics forces this move,
and nothing about the specific design of the computer forces
this move. The move is forced by the overwhelmingly good
reasons for making it and not any other move . Any chess
player, constructed of whatever physical materials, would
make it. Even a ghost or an angel would make it! You come
up with your intentional-stance prediction on the basis of
your bold assumption that no matter how the computer pro
gram has been designed, it has been designed well enough to
be moved by such a good reason. You predict its behavior as
if it were a rational agent.
The intentional stance is undeniably a useful shortcut in
such a case, but how seriously should we take it? What does
a computer care, really, about whether it wins or loses? Why
say that the alarm clock desires to obey its master? We can
use this contrast between natural and artificial goals to
heighten our appreciation of the fact that all real goals ulti
mately spring from the predicament of a living, self-protec
tive thing. But we must also recognize that the intentional
stance works (when it does) whether or not the attributed
goals are genuine or natural or " really appreciated" by the

so-called agent, and this tolerance is crucial to understand
ing how genuine goal-seeking could be established in the
first place. Does the macromolecule really want to replicate
itself? The intentional stance explains what is going on,
regardless o f how we answer that question. Consi der a sim
ple organism-say, a planarian or an amoeba-moving non
randomly across the bottom of a laboratory dish , always
heading to the nutrient-rich end of the dish, or away from
the toxic end. This organism is seeking the good, or shun
ning the bad-its own good and bad, not those of some
human artifact-user. Seeking one 's own good is a fundamen
tal feature of any rational agent , but are these simple organ
isms seeking or j ust " seeking?" We don 't need to answer that
question. The organism is a predictable intentional system
in either cas e.
This is another way of making Socrates' point in the
Men o, when he asks whether anyone ever knowingly desires
evi l . We intentional systems do sometimes desire evi l ,
through misunderstanding or misinformation o r sheer
l unacy, but it is p art and parcel of rationality to desire what
is deemed good. It is this constitutive relationship between
the good and the seeking of the good that is endorsed-or
rather enforced-by the natural selection of our forebears:
those with the misfortune to be genetically designed so that
they seek what is bad for them l eave no descendants in the
long run. It is no accident that the pro ducts of natural selec
tion seek (or " seek" ) what they deem (or " deem") to be good.
Even the simplest organisms , if they are to favor what is
good for them, need some sense organs or discriminative
p owers-some simple switches that turn ON in the presence
of good and OFF in its absence-and these switches , or
transducers, must be united to the right bodily responses.
This requirement is the birth of function . A rock can't mal
function, for it has not been well- or ill-equipped to further
I N T E N T I O NA L I TY 33
any good. When w e decide t o interpret a n entity from the

intentional stance, it is as if we put ourselves in the role of
its guardian, asking ourselves, in effect, " If I were in this
organism 's predicament, what would I do?" And here we
exp ose the underlying anthropomorphism o f the intentional
stance: we treat all intentional systems as if they were just
like us-which of course they are not.
Is this then a misapplication of our own p erspective, the
perspective we min d-havers share? Not necessarily. From the
vantage point of evolutionary history, this is what has hap
pened: Over billions of years, organisms gradually evolved,
accumulating ever more versatile machinery designed to fur
ther their ever more complex and articulated goods. Eventu
ally, with the evolution in our species of language and the
varieties of reflectiveness that language p ermits (a topic for
later chapters ), we emerged with the ability to wonder the
wonders with which we began this book-wonders about
the minds of other entities. These wonders, naively con
ducted by our ancestors, led to animism , the idea that each
moving thing has a mind or soul (anima, in Latin) . We began
to ask ourselves not only whether the tiger wanted to eat
us-which it probably did-but why the rivers wanted to
reach the seas, and what the clouds wanted from us in return
for the rain we asked of them. As we became more sophisti
cated-and this is a very recent historical development, not
anything to be discerned in the vast reaches of evolutionary
time-we gradually withdrew the intentional stance from
what we now call inanimate nature, reserving it for things
more like us: animals, in the main, but also plants under
many conditions. We still " trick" flowers into blooming pre
maturely by " deceiving" them with artificial spring warmth
and light, and " encourage" vegetables to send down longer
roots by withhol ding from them the water they want so
badly. (A logger once exp lained to me how he knew we
would find no white pines among the trees in some high
ground in my forest-" Pines like to keep their feet wet.")

This way of thinking about plants is not only natural and
harmless but p ositively an aid to comprehension and an
imp ortant lever for discovery. When biologists discover that
a p l ant has some rudimentary discriminatory organ , they
immediately ask themselves what the organ is for-what
devious project does the plant have that requires it to obtain
information from its environment on this topic? Very often
the answer is an imp ortant scientific discovery.
In ten tional systems are , by definition, all and only those
entities whose behavior is predictable/explicable from the
intentional stance. Self-replicating macromolecules , ther
mostats , amoebas , plants , rat s, bats , people , and chess-play
ing computers are all intentional systems-some much more
interesting than others . Since the point of the intentional
stance is to treat an entity as an agent in order to predict its
action s , we have to suppose that it is a smart agent, since a
stupid agent might do any dumb thing at all. This bold leap
of supposing that the agent will make only the smart moves
(given its limited perspective) is what gives us the leverage
to make predictions. We describe that limited perspective by
attributing particular beliefs and desires to the agent on the
basis of its p erception of the situation and its goals or needs.
Since our predictive leverage in this exercise is critically
dependent on this particularity-since it is sensitive to the
p articular way the beliefs and desires are expressed by us,
the theorists , or represented by the intentional system in
question, I call such systems in ten tional systems. They
exhibit what philosophers call in ten tionality.
" Intentionality," in this special philosophical sense , is
such a controversial concept, and is so routinely misunder
stood and misused by nonphilosophers , that I must pause to
belabor its definition. Unfortunately for interdisciplinary
communicatio n , the philosophical term " intentionality" has
two false friends-perfectly good words that are readily con-
I NTENTI O NALITY 35
fused with it, and indeed are rather closely related t o it. One
is an ordinary term , the other is technical (and I will post
pone its introduction briefly) . In ordinary parlance, we often
discuss whether someone 's action was intentional or not.
When the driver crashed into the bridge abutment , was he
intentionally committing suicide , or had he fallen asleep?
When you called the policeman "Dad" just then , was that
intentional , or a slip of the tongue? Here we are asking , are
we not, about the intentionality of the two deeds? Yes , in the
ordinary sense; no, in the philosophical sense.
Intentionality in the philosophical sense is just
aboutness. Something exhibits intentionality if its compe
tence is in some way abou t something else. An alternative
would be to say that something that exhibits intentionality
contains a represen tation of something else-but I find that
less revealing and more problematic . Does a lock contain a
representation of the key that opens it? A lock and key
exhibit the crudest form of intentionality; so do the opioid
receptors in brain cells-receptors that are designed to
accept the endorphin molecules that nature has been provid
ing in brains for millions of years . B oth can be tricked-that
i s , opened by an impostor. Morphine molecules are artifac
tual skeleton keys that have recently been fashioned to open
the opioid-receptor doors too. (In fact it was the discovery of
these highly specific receptors which inspired the search
that led to the discovery of endorphins, the brain's own
painkillers . There must have been something already present
in the brain , reasearchers reasoned , for these specialized
receptors to have been abo u t in the first place . ) This lock
and-key variety of crude aboutness is the basic design ele
ment out of which nature has fashioned the fancier sorts of
subsystems that may more deservedly be called representa
tion systems , so we will have to analyze the aboutness of
these representations in terms of the (quasi?) aboutness of
locks-and-keys in any case. We can stretch a p oint and say
that the present shape of the bimetallic spring in a thermo

stat is a representation of the present room temperature, and
that the position of the thermostat's adjustable lever is a rep
resentation of the desired room temperature , but we can
equally well deny that these are , properly speaking, represen
tations. They do, however, embody information about room
temperature , and it is by virtue of that embodiment that they
contribute to the competence of a simple intentional system.
Why do philosophers call aboutness " intentionality"? It
all goes back to the medieval philosophers who coined the
term , noting the similarity between such phenomena and
the act of aiming an arrow at something (in ten dere arcum
in) . Intentional phenomena are equipped with metaphorical
arrow s , you might say, aimed at something or other-at
whatever it is the phenomena are about or refer to or allude
to. But of course many phenomena that exhibit this minimal
sort of intentionality do not do anything in ten tionally, in the
everyday sense of the term . Perceptual states , emotional
states , and states of memory, for example , all exhibit about
ness without necessarily being intentional in the ordinary
sense; they can be entirely involuntary or automatic
responses to one thing or another. There is nothing inten
tional about recognizing a horse when it looms into view,
but your state of recognition exhibits very particular about
nes s : you recognize it as a horse. If you had misperceived it
as a moose or a man on a motorcycl e , your perceptual state
would have had a different aboutness. It would have aimed
its arrow rather differently-at something nonexistent, in
fact, but nevertheless quite definite : either the moose that
never was , or the illusory motorcyclist. There is a large psy
chological difference between mistakenly thinking you 're in
the presence of a moose and mistakenly thinking you 're in
the presence of a man on a motorcycle , a difference with
predictable consequences . The medieval theorists noted that
I NTENTIONALITY 37
the arrow of intentionality could thus be aimed at nothing

while nevertheless being aimed in a rather p articular way.
They called the object of your thought , real or not, the in ten
tional object.
In order to think about something , you must have a
way-one way among many possible ways-of thinking
about it. Any intentional system is dependent on its particu
lar ways of thinking about-perceiving, searching for, identi
fying, fearing, recalling-whatever it is that its " thoughts"
are about. It is this dependency that creates all the opportu
nities for confusion, both practical and theoretical. Practi
cally, the best way to confuse a particular intentional system
is to exp loit a flaw in its way(s) of p erceiving or thinking
about whatever it needs to think about. Nature has explored
countless variations on this theme, since confusing other
intentional systems is a major goal in the life o f most inten
tional systems. After all , one of the primary desires of any
living intentional system is the desire for the food needed to
fuel growth , self-repair, and reproduction, so every living
thing needs to distinguish the food (the good material) from
the rest of the world. It follows that another primary desire is
to avoid becoming the food of another intentional system . So
camouflage, mimicry, stealth , an d a h o st of other stratagems
have put nature 's locksmiths to the test , provoking the evo
lution of ever more effective ways of distinguishing one
thing from another and keeping track of them. But no way is
ever foolproof. There is no taking without the possibility of
mistaking. That's why it's so important for us as theorists to
be able to identify and distinguish the different varieties of
taking (and mistaking) that can occur in intentional systems.
In order to make sense of a system 's actual " take" on its cir
cumstances , we have to have an accurate picture of its
dependence on its particular capacities for distinguishing
things-its ways of " thinking about" things.
Unfortunately, however, as theorists we have tended to

overdo it, treating our own well-nigh limitless capacity for
distinguishing one thing from another in our thoughts
(thanks to our abi lity to use language) as if it were the hall
mark o f all genuine intentionality, all aboutness worthy of
the name. For instance , when a frog's tongue darts out and
catches whatever is flying by, the frog may make a mistake
it may ingest a ball bearing thrown by a mischievous child,
or a fisherman's lure on a monofilament thread , or some
other inedible anomaly. The frog has made a mistake , but
exactly which mistake (s) has it made? What did the frog
" think" it was grabbing? A fly? Airborne food? A moving
dark convexity? We language users can draw indefinitely
fine distinctions of content for the candidate frog-thought,
and there has been an unexamined assumption that before
we can attribute any real intentionality to the frog we have
to narrow down the c ontent of the frog's states and acts with
the same precision we can use (in principle) when we con
sider human thoughts and their propositional content.
This has been a maj or source of theoretical confusion, and
to make matters wors e , there is a handy technical term from
logic that refers to just this capacity of language for making
indefinitely fine-grained discriminations : in tensionality.
With an s. Intensionality-with-an-s is a feature of languages;
it has no direct application to any other sort of representa
tional system (pictures , map s , graphs , " search images , " . . .
min ds) . According to standard usage among logicians , the
words or symbols in a language can be divided into the logi
cal , or functio n , words ( " i f, " " and , " " or, " "not , " " all , "
" some , " . . . ) and the terms o r predicates, which can b e as
various as the topic of discussion ( " red , " " tall , " "grandfa
ther, " " oxygen , " " second-rate composer of sonnets , " . . . ) .
Every meaningful term o r predicate o f a language has an
extension-the thing or set of things to which the term
refers-and an in tension-the p articular way in which this
I NTENTIO NALITY 39
thing o r set of things is picked out or determined. " Chelsea

Clinton's father" and " presi dent of the United States in
1 995 " name the very same thing-B ill Clinton-and hence
have the same extension, but they zero in on this common
entity in different ways, and hence have difference inten
sions . The term " equilateral triangle" p icks out exactly the
same set of things as the term " equiangular triangle," so
these two terms have the same extension, but clearly they
don 't mean the same thing : one term is abo u t a triangle 's
sides being equal and the other is abo u t the angles being
equal . So intension (with an s) is contrasted to extension,
and means, well, meaning. And isn't that what intentional
ity-with-a-t means, too?
For many purposes, logicians note, we can ignore differ
ences in the in tensions of terms and just keep track of exten
sions. After all, a rose by any other name would smell as
sweet, so if roses are the topic, the indefinitely many differ
ent ways of getting the class of roses into the discussion
should be equivalent, from a logical point of view. Since
water is H 2 0 , anything truly said of water, using the term
"water," will be just as truly said if we substitute the term
"H 2 0 " in its place-even if these two terms are subtly differ
ent in meaning, or intension. This freedom is particularly
obvious and useful in such topic areas as mathematics,
where you can always avail yourself of the practice of " sub
stituting equals for equals," replacing " 4 2 " by " 1 6 " or vice
versa, since these two different terms refer to one and the
same number. Such freedom of substitution within linguistic
contexts is aptly called referen tial transparency: you can see
right through the terms, in effect, to the things the terms
refer to. But when the topic is not roses but thinking-abo u t
roses, or talking-abou t-(thinking-abo u t}-roses, differences in
intension can matter. S o whenever the topic is intentional
systems and their beliefs and desires, the language used by
the theorist is intension-sensitive. A logician would say that
such discourse exhibits referen tial opacity; it is not transpar

ent; the terms themselves get in the way and interfere in sub
tle and confusing ways with the topic.
To see how referential opacity actually matters when we
adopt the intentional stance , let's consider a root case of the
intentional stance in action , applied to a human being. We
do this effortlessly every day, and seldom spell out what is
involved , but here is an exampl e , drawn from a recent philo
sophical article-an example that goes rather weirdly but
usefully into more detail than usual :
B rutus wanted to kill Caesar. He believed that Caesar was

an ordinary mortal, and that, given this , stabbing him (by
which we mean p lunging a knife into his heart) was a
way of killing him. He thought that he could stab Caesar,
for he remembered that he had a knife and saw that Cae
sar was standing next to him on his left in the Forum. So
Brutus was motivated to stab the man to his left . He did
so, thereby killing Caesar. (Israe l , Perry, and Tutiya, 1 99 3 .
p . 5 1 5)
Notice that the term " Caesar" is surreptitiously playing a

crucial double role in this explanation-not just in the nor
mal , transparent way of p icking out a man, Caesar, the chap
in the toga standing in the Forum , but in p icking out the
man in th e way Bru tus himself picks him out. It is not
enough for Brutus to see Caesar standing next to him; he has
to see that he is Caesar, the man he wants to kil l . If Brutus
mistook Caesar, the man to his left, for Cassius , then he
wouldn 't try to kill him : he wouldn't have been motivated,
as the authors say, to stab the man to his left, since he would
not have drawn the crucial connection in his mind-the link
identifying the man to his left with his goal.
I NTENTIONALITY 41
T H E M I S G U I D E D G OA L O F
PROPOSITIONAL PREC ISION
Whenever an agent acts , it acts on the basis o f a particular

understanding-or misunderstanding-of the circumstances ,
and intentional explanations and predictions rely on captur
ing that understanding. To predict the action of an inten
tional system , you have to know what things the beliefs and
desires of the agent are about , an d you have to know, at least
roughly, how those beliefs and desires are about those
things , so you can say whether the crucial connections have
been , or will be, drawn .
But notice that I said that when we adopt the intentional
stance we have to know at least roughly how the agent picks
out the obj ects of concern . Failing to notice this is a maj or
source of confusion. We typically don't need to know
exactly what way the agent conceives of his task. The inten
tional stance can usually tolerate a lot of slack, and that's a
blessing, since the task of expressing exactly how the agent
conceives of his task is misconceived , as p ointless an exer
cise as reading poems in a book through a microscope . If the
agent under examination doesn't conceive of its circum
stances with the aid of a language capable of making certain
distinctions , the sup erb resolving power of our language
can 't be harnessed directly to the task of expressing the par
ticular thoughts , or ways of thinking , or varieties of sensitiv
ity, of that agent . (Indirectly, however, language can be used
to describe those particularities in whatever detail the theo
retical context demands . )
This point often gets lost i n the mists of a spuriously per
suasive argument , along the following lines . D o dogs (for
example) think? If so, then of course they must think partic
ular thoughts. A thought couldn 't exist without being some
particular thought or other, could it? But a particular thought
42 KINDS OF MINDS
must be composed of particular concepts . You can 't think

the thought
th at my dish is full of beef
unless you have the concepts of dish and beef, and to have
these concepts you have to have a host of other concepts
(bucket, plate, cow, flesh , . . . ), since this particular thought
is readily distinguishable (by us) from the thought
that th e bucket is full of beef
as well as from the thought
that my plate is full of calves ' liver
to say nothing of the thought
th at the red, tasty stuff in the thing that I usually eat from
is not th e usual dry stuff th ey feed me
and so on and so forth, forever. Just which thought or

thoughts is the dog thinking? How can we express-in En
glish, say-exactly the thought the dog is thinking? If it can 't
be done (and it can 't) , then either dogs can't think thoughts
at all or dogs ' thoughts must be systematically inexpress
ible-and hence beyond our ken.
Neither alternative follows. The idea that a dog 's
" thought" might be inexpressible (in human language) for
the simple reason that expression in a human language cuts
too fine is often ignored , along with its corollary: the idea
that we may nevertheless exhaustively describe what we
can ' t express , leaving no mysterious residue at all. The dog
has to have its p articular ways of discriminating things , and
these ways get composed into quite particular and idiosyn

cratic " concepts ." If we can figure out how these ways work,
and describe how they work together, then we will know as
much about the content of the dog's thoughts as we ever l earn
about the content of another human being's thoughts through
conversation, even if we can't find a sentence (in English or
in any other human language) that expresses that content.
When we human mind-havers, from our uniquely ele
vated perspective, use our special trick of app lying the
intentional stance to other entities, we are imposing our
ways on them, and we risk importing too much clarity, too
much distinctness and articulation of content, and hence too
much organization, to the systems we are attempting to
understand. We also risk imp orting too much of the particu
lar kin d of organization of our own minds to our model of
these simpler systems. Not all of our needs, and hence
desires, and hence mental practices, and hence mental
resources, are shared by these simpler candidates for minds.
Many organisms " experience" the sun, and even guide
their lives by its passage . A sunflower may track the sun in a
minimal way, twisting to face it as it crosses the sky, maxi
mizing its daily exp osure to sunlight, but it can't cope with
an intervening umbrella. It can 't anticipate the sun 's reemer
gence at a calculable later time and adjust its slow, simple
"behavior" accordingly. An animal might well be capable of
such sophistication, modulating its locomotion to keep itself
hidden in shadows from its prey, or even anticipating where
to stretch out in the sun for a long nap, appreciating (dimly
and unthinkingly) that the tree 's shadow will soon lengthen.
Animals track and reidentify other things (mates, quarry, off
spring, favorite food sites), and they might similarly track
the sun. But we human beings don't just track the sun, we
make an ontological discovery about the sun: it's th e sun !
The very same sun each day.
The German logician Gottlob Frege introduced an exam

ple that logicians and philosophers have written about for
more than a century : the Morning Star, known to the
ancients as Phosphoru s , and the Evening Star, known to the
ancients as Hesperus , are one and the same heavenly body:
Venus . To day this is a familiar fact, but the discovery of this
identity was a substantial early advance in astronomy.
Which of us today could formulate the argument and amass
the crucial evidence without looking for help in a book?
Even as small children , we readily understand (and docilely
accept) the hypothesi s , however. It's hard to imagine that
any other creatures could ever be brought to formulate,
much less confirm , the hyp othesis that these small bright
spots are one and the same heavenly body.
Couldn 't those huge, hot disks that make a daily passage
across the skies be new every day? We 're the only species
that can even formulate the question . Compare sun and
moon to the seasons. Spring comes back each year, but we
don't ask (any more) if it 's the same spring , returned. Per
haps Spring , personified as a goddess in the old days , was
seen by our ancestors as a returning particular, not a recur
ring universal. But for other species this isn't even an issue .
S ome species have exquisite sensitivity to variations; they
can discriminate many more details, in some domains , than
we can with our naked senses (although as far as we know,
we can , with the aid of our prosthetic extensions-micro
scopes, spectroscopes , gas chromatographs , and so forth
make finer discriminations in every single modality than
any other creatures on the p lanet ) . But these other species
have a very limited ability to reflect, and their sensitivities
are channeled down rather narrow sets of possibilities , as we
shall see.
We , in contrast, are believe-alls. There is no limit , appar
ently, to what we can believe, and to what we can distin
guish in belief. We can distinguish between believing
I NTENTIO NALITY 45
that th e sun is an d always has been th e same star, each

day,
and believing
that th e sun has been the same star, each day, since Jan u
ary 1 , 1 900, when th e latest s u n took over its role from its
predecessor.
I take it that nobody believes the latter, but it is easy enough

to see what the belief is, and to distinguish it both from the
standard belief and from the equally daft but different belief,
that th e most recen t ch ange of suns happen ed on June 1 2,

1 986.
The fundamental form of all such attributions of mental

states to intentional systems are sentences that express what
are called propositional attitudes.
x believes that p.
y desires that q.
z wonders whether r.
Such sentences consist of three parts : a term referring to

the intentional system in question (x, y, z) , a term for the
attitude attributed to it (belief, desire, wonder, . . . ), and a
term for the particular content or meaning of that attitude
the proposition denoted in these dummy cases by the letters
p, q, and r. In actual attribution sentences, of course, these
propositions are expressed as sen ten ces (of English, or what
ever language the sp eaker is using), and these sentences con
tain terms that may not be substituted ad lib for coextensive
terms-that's the feature of referential opacity.
Propositions, then, are the theoretical entities with which
we identify, or measure , beliefs . For two believers to share a

belief i s , by definition, for them to believe one and the same
proposition. What then are prop ositions? They are , by mutu
ally agreed philosophical convention, the abstract meanings
shared by all sen tences that . . . mean the same thing. An
ominous circle emerges from the smoke of battle . Presum
ably, one and the same proposition is expressed by
1 . Snow is white.
2. La neige est blanche .
3 . Der Schnee ist weis s .
After all , when I attribute to Tom the belief that s n o w is

white , we want Pierre and Wilhelm to be able to attribute the
same belief to Tom in their own tongues. The fact that Tom
need not understand their attributions is beside the point.
For that matter, Tom need not understand my attribution, of
cours e , since p erhap s Tom is a cat , or a monolingual Turk.
But is one and the same proposition also shared by the
fol lowing?
4. Bill hit Sam.

5. Sam was hit by Bil l .
6 . It wa s Bill who w as t h e agent of t h e act of hitting of
which S am was the victim.
They all " say the same thing," and yet they all say " it" in dif
ferent ways. Should propositions line up with ways of say
ing or with things said? A simp l e , theoretically appealing
way of settling the issue would be to ask whether a believer
can believe one of these without believing another. If so,
then they are different propositions. After all , if propositions
are to be the theoretical entities that measure belief, we
wouldn't want this test to fail . But how can we test this if
Tom isn't an English speaker, or a speaker at all? We attribut-
I N T E N T I O N A L I TY 47
ors-at least when we express our attributions in language

must be bound by a system of expression, a languag e, and
languages differ in their structures as well as their term s . B y
being forced into one suc h language structure or another, we
willy-nilly take on more distinctions than the circumstances
may warrant. This is the p oint of the warning I issued earlier
about the rough attribution of content that suffices for the
success of the intentional stance .
The philosopher Paul Churchland ( 1 9 79 ) h a s likened
propositions to numbers-equally abstract obj ects used to
measure many physical prop erties.
x has weight-in-grams of 1 4 4 .
y has speed-in-meters-per-second of 1 2.
Obviously, numbers are well-behaved occupants of this

role . We can " substitute equals for equal s . " There is no diffi
culty in agreeing that x has weight-in-grams of 2x72 or that y
has speed-in-meters-per-second of 9+ 3. There is a difficulty,
as we have just seen, when we try to apply the same rules of
transformation and equivalence to different expressions of
what are putatively the same proposition. Propositions , alas ,
are not as well-behaved theoretical entities as numbers .
Propositions are more like dollars than numbers !
This goat is worth $50.
And how much is it worth in Greek drachmas , or Russian

rubles (on what day of the week ! )-and is it worth more or
less today than it was in ancient Athens or as p art of Marco
Polo 's exp editionary supplies? There is no doubt that a goat
always has a value to its owner, and there is no doubt that
we can fix a rough , operational measure of its value by exe
cuting-or imagining ourselves to execute-an exchange for
money, or gold dust, or bread , or whatever. But there is no
48 KIN DS 0F MI N DS
fixed , neutra l , eternal system of measuring economic value ,

and likewise there is no fixe d , neutral , eternal system for
measuring meaning by the propositionful. So what? It would
be nice , I gues s , if there were such systems ; it would make
for a neater worl d , and it might make the theoretician 's j ob
simpler. But such a single-standard , universal system of
measurement is unnecessary for theory in both economics
and intentional-system theory. S ound economic theory is
not threatened by the ineliminable imprecision in its mea
surement of economic value generalized to all circumstances
at all times . Sound intentional-system theory is not threat
ened by the ineliminable imprecision in its measurement of
meaning across the same universal spectrum. As l ong as we
are alert to the difficulty, we can deal with all local problems
quite satisfactorily, using whatever rough-and-ready system
we choose.
In subsequent chapters , we will find that when we take
our "believe-al l " competence and apply it to " l ower" crea
ture s , it handily organizes the data for u s : it tells us where to
look next, sets boundary conditions , and highlights patterns
of similarity and difference. But if we are not careful, as we
have already seen , it can also woefully distort our vision. It
is one thing to treat an organism , or any of its many subsys
tems , as a rudimentary intentional system that crudely and
un thinkingly pursues its undeniably sophi sticated ends , and
quite another to impute reflective appreciation to it of what
it is doing. Our kind of reflective thinking is a very recent
evolutionary innovation.
The original self-replicating macromolecules had reasons
for what they did , but had no inkling of them . We , in con
trast, not only know-or think we know-our reasons ; we
articulate them , discuss them, criticize them, share them.
They are not j ust the reasons we act; they are reasons for us.
In between the macromolecules and us there is quite a story
to be told. Consider, for instance , the fledgling cucko o ,

hatched in a n alien nest by unwitting adoptive p arents . Its
first action when it emerges from its egg is to roll the other
eggs out of the nest. This is not an easy task, and it is quite
astonishing to watch the ferocious single-mindedness and
resourcefulness with which the baby bird overcomes what
ever obstacles lie in its way to j ettison the other eggs . Why
does it do this? B ecause those eggs contain rivals for the
attentions of its surrogate providers . By disposing of these
rivals , it maximizes the food and protective care it will
receive. The newborn cuckoo is, of course , oblivious ; it has
no inkling of this rationale for its ruthless act, but the ratio
nale is th ere, and has undoubtedly shaped this innate b ehav
ior over the eons. We can see it , even if the cuckoo can ' t . I
call such a rationale " free floating , " because it is nowhere
represen ted in the fledgling , or anywhere else, even though
it is operative-over evolutionary time-in shaping and
refining the behavior in question (in providing for its infor
mational needs , for instance) . The strategic principles
involved are not explicitly encoded but j ust implicit in the
larger organization of designed features . How did those rea
sons get captured and articulated in some o f the minds that
have evolved? That 's a good question. It will occupy our
attention for several chapters , but before going on to con
sider it, I must address a residual suspicion some philoso
phers have aired , to wit: I have it exactly backward. I am
proposing to explain real intentionality in terms of pseudo
intentionality ! Moreover, it seems , I am failing to acknowledge
the important distinction between original or intrinsic inten
tionality and derived intentionality. What is the distinction?
O R I G I N A L A N D D E R I V E D I N T E N T I O N A L I TY
According to some philosophers , following John Searle

( 1 98 0 ) , intentionality comes in two varieties, intrinsic (or
original) and derived. Intrinsic intentionality is the about
ness of our thoughts , our beliefs , our desire s, our intentions
(intentions in the ordinary sense) . It is the obvious source of
the distinctly limited and derived sort of aboutness exhib
ited by some of our artifacts : our words , sentences , books ,
map s , p icture s , computer programs. They have intentional
ity only by courtesy of a kind of generous loan from our
minds. The derived intentionality of our artifactual repre
sentations is parasitic on the genuine , original, intrinsic
intentionality that lies behind their creation.
There is a lot to be said for this claim. If you close your
eyes and think about Pari s , or your mother, that thought of
yours is about its object in the most primary and direct way
that anything could be about anything. If you then write a
description of Pari s , or draw a sketch of your mother, the
representation on the paper is about Paris , or your mother,
only because that is your authorial intention (ordinary
sense) . You are in charge of your representations , and you
get to declare or decide what these creations of yours are
about. There are conventions of language that you rely on to
assist in this injection of meaning into brute marks on paper.
Unless you have just previously declared that henceforth
you shall mean to refer to Bo ston whenever you say or write
the word "Paris" or that you choose to call Michelle Pfeiffer
" Mother," the standard references agreed to by your linguis
tic community are assumed to be in force . These conven
tions , i n turn , depend on the communal intentions of that
community. So external representations get their mean
i ngs-their intensions and extensions-from the meanings
of the internal , mental states and acts of the people who
I NTENTIONALITY 51
make them and use them. Those mental states and acts have
original intentionality.
The p oint about the dependent status of artifactual repre
sentations is undeniable. Manifestly, the pencil marks in
themselves don't mean a thing . This is particularly clear in
cases of ambiguous sentences. The philosopher W. V. 0 .
Quine gives us the nice exampl e :
Our mothers bore us.
What is this thing abou t? Is this a present-tense complaint

about boredom or a past-tense truism about our origins? You
have to ask the p erson who created the sentence. Nothing
about the marks in themselves could p ossibly determine the
answer. They certainly don't have intrinsic intentionality,
whatever that might b e . If they mean anything at all , it is
because of the role they play in a system of representation
that is anchored to the minds of the representers .
But what of the states and acts of those minds? What
endows them with their intentionality? One popular answer
is to say that these mental states and acts have meaning
because they themselves , marvelously enough , are com
posed in a sort of language-the language of thought. Men
talese. This is a hopeless answer. It is hopeless not because
there couldn't be any such system to be found in the internal
goings-on in people 's brains. Indeed, there could be-though
any such system wouldn't be just like an ordinary natural
language, such as English or French. It is hopeless as an
answer to the question we posed, for it merely postpones the
question. Let there be a language of thought. Now whence
comes the meaning of its terms? How do you know what the
sentences in your language of thought mean? This problem
comes into sharp er focus if we contrast the language-of
thought hypothesis with its ancestor and chief rival , the p ic
ture theory of ideas . Our thoughts are like p icture s , runs this
52 KINDS OF MI NDS
view; they are about what they are about because , like pic
tures, they resemble their objects . How do I tell my idea of a
duck from my idea of a cow? By noting that my idea of a
duck looks like a duck, while my idea of a cow doesn 't!
Thi s , too , is hopeless , because it immediately raises the
question, And how do you know what a duck looks like?
Again , it's not hopeless because there couldn't be a system of
imagery in your brain that exp loits pictorial resemblances
between the brain's internal images and the things they rep
resent; indeed, there could be. In fact , there i s, and we are
beginning to understand how such a system works . It is
hopeless as an answer to our basic question, however,
because it depends on the very understanding that it's sup
posed to exp lai n, and hence goes round in circles.
The solution to this problem of our intentionality is
straightforward. We just agreed that representati onal arti
facts (such as written descriptions and sketches) possess
derived intentionality, by virtue of the role they play in the
activities of their creators . A shopping list written down on a
piece of paper has only the derived intentionality it gets
from the intentions of the agent who made it. Wel l , so does a
shopping list held by the same agent in memory ! Its inten
ti onality is exactly as derived as that of the external list, and
for the same reasons . Similarly, a merely mental image of
your mother-or Michelle Pfeiffer-is about its obj ect in just
as derived a way as the sketch you draw. It is internal, not
external , but it is still an artifact created by your brain and
means what it does because of its particular position in the
ongoing economy of your brain's internal activities and their
role in governing your body 's complex activities in the real ,
surrounding world.
And how did your brain come to have an organization of
such amazing states with such amazing powers? Play the
same card again : the brain is an artifact , and it gets whatever
intentionality its parts have from their role in the ongoing
I NT E N T I O NALITY 53
economy of the larger system of which it is a p art-or, i n

other words , from the intentions of its creator, Mother
Nature (otherwise known as the process of evolution by nat
ural selection) .
This idea that the intentionality of brain states is derived
from the intentionality of the system or process that
designed them is admittedly a strange and unsettling idea, at
first. We can see what it comes to by considering a context in
which it is surely correct: When we wonder about the
(derived) intentionality of the "brain" states of some manu
factured robot. Suppose we come across a robot trundling a
shopping cart through a supermarket and periodically con
sulting a slip of paper with symbols written on it. One line i s :
M I LK@.SxGAL i f P<2xQT\P e l s e 2xM I LK@QT
What , if anything , is this gibberish abo u t? We ask the robot.

It replies, " That's just to remind me to get a half gallon of milk,
but only if the price of a half gallon is less than twice the price
of a quart. Quarts are easier for me to carry. " This auditory
artifact emitted by the robot is mainly j ust a translation into
English of the written one, but it wears its derived meaning on
its sleeve , for our benefit. And where did either of these arti
facts get their derived intentionality? From the clever engi
neering work of the robot's designers , no doubt, but maybe
very indirectly. Maybe these engineers formulated and
directly installed the cost-conscious principle that has
spawned this particular reminder-a rather boring possibility,
but one in which the derived intentionality of these states
would definitely lead back to the human designers' own
intentionality as the creators of those states . It would be much
more interesting if the designers had done something deeper.
It is possible-just on the edge of technological capability
today-that they designed the robot to be cost-sensitive in
many ways and let it " figure out, " from its own " experience, "
that it should adopt some such principle. In this case, the

principle would not be hard-wire d but flexible, and in the
near future the robot might decide from its further " experi
ence " that this application was not cost-effective after all ,
and it would buy milk in convenient quarts no matter what
they cost. How much design work did the robot's designers
do , and how much did they delegate to the robot itself? The
more elab orate the system of controls , with its attendant
information-gathering and information-assessing subsys
tems , the greater the contribution of the robot itself, and
hence the greater its claim to be the " author" of its own
meanings-meanings that might , over time , become quite
inscrutable to the robot's designers .
The imagined robot does not yet exist , but someday it
might. I introduce it in order to show that within its world of
merely derived intentionality we can draw the very distinc
tion that inspired the contrast between original and derived
intentionality in the first place. (We had to " consult the
author" to discover the meaning of the artifact . ) This is
instructive, because it shows that derived intentionality can
be derive d from derived intentionality. It also shows how an
illusion of intrinsic intentionality (metaphysically original
intentionality) could arise. It might seem that the author of a
puzzling artifact would have to have intrinsic intentionality
in order to be the source of the artifact's derived intentional
ity, but this is not so. We can see that in this case, at least,
there is no work left over for in trinsic intentionality to do .
The imagined robot would be just as capable as we are of
delegating derived intentionality to further artifacts. It gets
around in the world , advancing its projects and avoiding
harm , on the strength of its " merely " derived intentionality,
the intentionality designed into it-first by its designers and
then, as it acquires more information about its world, by its
own processes of self-redesign. We may perhaps be in the
same predicament , living our lives by the lights of our
I NT E N T I O N ALITY 55
"merely" derived intentionality. What boon would intrinsic

intentionality (whatever that is) provide for us that could not
as well have been bequeathed to us, as evoluti on-designed
artifacts? Perhaps we are chasing a will-of-the-wis p .
It's a g o o d thing that this prospect h a s opened up for u s,
because t he intentionality that allows us to sp eak and write
and wonder all manner of wonders is undeniably a late and
complex product of an evolutionary process that has the
cruder sorts of intentionality-disparaged by S earle and oth
ers as " mere as if intentionality"-as both its ancestors and
its contemporary components. We are descended from
robots , and composed of robots , and all the intentionality we
enj oy is derived from the more fundamental intentionality of
these billions of crude intentional systems. I don't have it
backward; I have it forward. That 's the only promising direc
tion to travel . But the j ourney lies ahead.
CHAPTER 3
T H E B O DY A N D I T S M I N D S
In the distant future I see open fields for far

more important researches . Psychology will be
based on a new foundation, that of the neces
sary acquirement o f each mental p ower and
capacity by gradation. Light will be thrown on
the origin of man and his history.
Charles Darwin, The Origin of Species
F R O M S E N S I T I V I T Y TO S E N T I E N C E ?
At last , let's take the journey. Mother Nature-or, as we call

it today, the process of evolution by natural selection-has
no foresight at all , but has gradually built beings with fore
sight. The task of a mind is to produce future , as the poet
Paul Valery once put it. A mind is fundamentally an antici
pator, an expectation-generator. It mines the present for
clues , which it refines with the help of the materials it has
saved from the past, turning them into anticipations of the
57
58 KINDS OF MINDS
future . And then it acts , rationally, on the basis of those

hard-won anticipations .
Given the inescapable competition fur materials in the
world of living things , the task facing any organism can be
considered to be one version or another of the childhood
game of hide-and-seek. You seek what you need, and hide
from those who need what you have. The earliest replica
tors , the macromolecules , had their needs and developed
simple-rela tively simp l e ! -means of achi eving them. Their
seeking was just so much random walking , with a suitably
configured grabber at the business end. When they bumped
into the right things, they grabbed them . These seekers had
no plan , no "s earch image , " no representation of the sought
for items beyond the configuration of the grabbers . It was
lock-and-key, and nothing more. Hence the macromolecule
did not know it was seeking , and did not need to know.
The " need to know " principle is most famous in its appli
cation in the world of espionage , actual and fictional : No
agent should be given any more information than he
absolutely needs to know to perform his part of the project.
Much the same principle has been honored for billions of
years , and continues to be honored in a trillion ways , in the
design of every living thing. The agents (or microagents or
pseudoagents) of which a living thing is composed-like the
secret agents of the CIA or KGB-are vouchsafed only the
information they need in order to carry out their very lim
ited specialized tasks . In espionage , the rat ionale is security ;
in nature , the rationale is economy. The cheapest, least
intensively designed system will be " discovered" first by
Mother Nature , and myopically selected.
It is important to recognize, by the way, that the cheapest
design may well not be the most efficient , or the smal lest. It
may often be cheaper for Mother Nature to throw in-or
leave in-lots of extra , nonfunctioning stuff, simply because
such stuff gets created by the rep lication-and-development
THE BODY AND ITS M I N DS 59
process and cannot be removed without exorbitant cost. I t is

now known that many mutations insert a code that simply
" turns off" a gene without deleting it-a much cheaper move
to make in genetic space . A p arallel phenomenon in the
world of human engineering occurs routinely in computer
programming. When programmers improve a program ( creat
ing, say, WordWhizbang 7 . 0 to replace WordWhizbang 6 . 1 ) ,
the standard practice i s t o create the new source code adja
cent to the old code, simply by copying the old code and
then editing or mutating the copy. Then , before running or
compiling the new code, they " c omment out " the old code
they don't erase it from the source code file but isolate the
old version between special symbols that tell the computer
to skip over the bracketed stuff when compiling or executing
the program . The old instructions remain in the "genome , "
marked so that they are never " expressed" i n the phenotype .
I t costs almost nothing t o keep the o l d code along for the
ride , and it might come in handy some day. Circumstances
in the world might change , for instance-making the old ver
sion better after all. Or the extra copy of the old version
might someday get mutated into something of value. Such
hard-won design should not be lightly discarded , since it
would be hard to re-create from scratch. As is becoming ever
more clear, evolution often avails itself of this tacti c , reusing
again and again the leftovers of earlier design processes. (I
explore this principle of thrifty accumulation of design in
more depth in Darwin 's Dangerous Idea. )
The macromolecules had n o need to know, and their sin
gle-celled descendants were much more complex but also
had no need to know what they were doing , or why what
they were doing was the source of their livelihood. For bil
lions of years , then, there were reasons but no reason formu
lators , or reason representers , or even, in the strong sense ,
reason appreciators . (Mother Nature , the process of natural
selection, shows her appreciation of good reasons tacitly, by
wordlessly and mindlessly permitting the best designs to

prosper. ) We late-blooming theorists are the first to see the
patterns and divine these reasons-the free-floating ratio
nales of the designs that have been created over the eons .
We describe the patterns using the intentional stance.
Even some of the simplest design features in organisms
p ermanent features even simpler than ON/OFF switches
can be installed and refined by a process that has an inten
tional-stance interpretation. For instance , p lants don't have
minds by any stretch of the theorist's imagination, but over
evolutionary time their features are shaped by comp etitions
that can be modeled by mathematical game theory-it is as if
the plants and their competitors were agents like us ! Plants
that have an evolutionary history of being heavily preyed
upon by herbivores often evolve toxicity to those herbivores
as a retaliatory measure . The herbivores , in turn , often
evolve a specific tolerance in their digestive systems for
those specific toxins , and return to the feast, until the day
when the pl ants , foiled in their first attempt , develop further
toxicity or prickly barb s , as their next move in an escalating
arms race of measure and countermeasure . At some point,
the herbivores may " choose " not to retaliate but rather to
discriminate, turning to other food sources , and then other
nontoxic plants may evolve to " mimic " the toxic plants ,
blindly exploiting a weakness in the discriminatory sys
tem-visual or olfactory-of the herbivores and thereby
hitching a free ride on the toxicity defense of the other p lant
species . The free-floating rationale is clear and predictive ,
even though neither the p lants nor the digestive systems of
the herbivores have minds in anything like the ordinary
sense.
All this hap p ens at an achingly slow pace , by our stan
dards. It can take thousands of generations , thousands of
years , for a single move in this game of hide-and-seek to be
made and responded to (though in some circumstances the
TH E BODY A N D ITS M I N DS 61
pace is shockingly fast) . The patterns of evolutionary change

emerge so slowly that they are invisible at our normal rate of
information uptake , so it's easy to overlook their intentional
interpretation , or to dismiss it as mere whimsy or metaphor.
This bias in favor of our normal pace might be called
timescale cha u vinism . Take the smartest, quickest-witted
person you know, and imagine filming her in action in ultra
slow motion-say, thirty thousand frames per second, to be
projected at the normal rate of thirty frames per second. A
single lightning riposte , a witticism offered " without skip
ping a beat , " would now emerge like a glacier from her
mouth , boring even the most patient moviegoer. Who could
divine the intelligence of her performanc e , an intelligence
that would be unmistakable at normal speed? We are also
charmed by mismatched timescales going in the other direc
tion , as time-lapse photography has vividly demonstrated.
To watch flowers growing , budding, and blooming in a few
seconds, is to be drawn almost irresi stibly into the inten
tional stance. See how that plant is striving upward , racing
its neighbor for a favored place in the sun, defiantly thrust
ing its own leaves into the light, parrying the counterblows ,
ducking and weaving like a boxer ! The very same pattern s ,
projected a t different speeds , c a n reveal or conceal the pres
ence of a mind, or the absence of a mind-or so it seems .
(Spatial scale also shows a powerful built-in bias ; if gnats
were the size of seagull s , more people would be sure they
had minds , and if we had to look through microscopes to see
the antics of otters , we would be less confident that they
were fun-loving. )
I n order for u s t o see things a s mindfu l , they have t o hap
pen at the right pace , and when we do see something as
mindful , we don't have much choice; the p erception is
almost irresistible. But is this just a fact about our bias as
observers , or is it a fact about minds? What is the actual role
of speed in the phenomenon of mind? Could there be minds ,
as real as any minds anywhere , that conducted their activi

ties orders of magnitude slower than our minds do? Here is a
reason for thinking that there could be: if our planet were
visited by Martians who thought the same sorts of thoughts
we do but thousands or millions of times faster than we do ,
we would seem to them to be about as stupid as trees , and
they would be inclined to scoff at the hypothesis that we
had minds. If they did, they would be wrong , wouldn't
they-victims of their own timescale chauvinism. So if we
want to deny that there could be a radically slow-thinking
mind, we will have to find some grounds other than our
preference for the human thought rate. What grounds might
there be? Perhaps , you may think, there is a minimum speed
for a mind, rather like the minimum escape velocity
required to overcome gravity and leave the planet. For this
idea to have any claim on our attention, let alone allegiance ,
we would need a theory that says why this should be. What
could it be about running a system faster and faster that
eventually would " break the mind barrier" and create a
mind where before there was none? Does the friction of the
moving p arts create heat , which above a certain temperature
leads to the transformation of something at the chemical
level? And why would that make a mind? Is it like particles
in an accelerator approaching the speed o f light and becom
ing hugely massive? Why would that make a mind? Does the
rapi d spinning of the brain parts somehow weave a contain
ment vessel to prevent the escape of the accumulating mind
p articles until a critical mass of them coheres into a mind?
Unless something along these lines can be proposed and
defen ded, the idea that sheer speed is essential for minds is
unappealing, since there is such a good reason for holding
that it's the relative speed that matters : perception , delibera
tion , and action all swift enough-relative to the unfolding
environment-to accomplish the purposes of a mind. Pro
ducing future is no use to any intentional system if its "pre-
T H E BODY AND ITS M I N DS 63
dictions " arrive t o o late to be acted on. Evolution will

always favor the quick-witted over the slow-witted, other
things being equal, and extinguish those who can 't meet
their deadlines well on a regular basis.
B u t what if there were a planet on which the speed of
light was 1 00 kilometers per hour, and all other physical
events and processes were slowed down to keep pace? Since
in fact the pace of events in the physical worl d can't be sped
up or slowed down by orders of magnitude (except in
philosophers ' fantastic thought experiments) , a relative
speed requirement works as well as an absolute speed
requirement. Given the speed at which thrown stones
approach their targets , and given the speed at which light
bounces off those incoming stone s, and given the speed at
which audible warning calls can be propagated through the
atmosphere , and given the force that must be marshaled to
get 1 00 kilograms of body running at 20 kilometers per hour
to veer sharp ly to the left or right-given these and a host of
other firmly fixed performance specifications , useful brains
have to operate at quite definite minimum speeds , indepen
dently of any fanciful " emergent properties " that might also
be produced only at certain speeds. These speed-of-operation
requirements , in turn , force brains to use media of informa
tion transmission that can sustain those speeds. That's one
good reason why it can matter what a mind is made of.
There may be others .
When the events in question unfold at a more stately
pace , something mindlike can occur in other media. These
patterns are discernible in these phenomena only when we
adopt the intentional stance . Over very long periods of time ,
species or lineages of p lants and animals can be sensitive to
changing conditions , and respon d to the changes they sense
in rational ways . That 's all it takes for the intentional stance
to find predictive and exp lanatory leverage. Over much
shorter periods of tim e , individual plants can resp ond
64 KINDS OF MINDS
appropriately to changes they sense in their environment,

growing new leaves and branches to exploit the available
sunlight , extending their roots toward water, and even (in
some species) temporarily adjusting the chemical composi
tion of their edible p arts to ward off the sensed onsla ugh t of
transient herbivores.
These sorts of slow-paced sensitivity, like the artificial
sensitivity of thermostats and computers , may strike us as
mere second-rate imitations of the phenomenon that really
makes the difference: sen tience. Perhaps we can distinguish
" mere intentional systems " from " genuine minds " by asking
whether the candidates in question enj oy sentience. Wel l ,
what is i t ? " Sentienc e" h a s never been given a proper defini
tion , but it is the more or less standard term for what is
imagined to be the lowest grade of consciousness. We may
wish to entertain the strategy, at about thi s point, of contrast
ing sentience with mere sensitivity, a phenomenon exhibited
by single-celled organisms , plants , the fuel gauge in your car,
and the film in your camera. Sensitivity need not involve
consciousness at all . Photographic film comes in different
grades of sensitivity to l ight; thermometers are made of
materials that are sensitive to changes in temp erature ; litmus
p ap er is sensitive to the presence of acid. Popular opinion
proclaims that p lants and perhaps " lower" animals-jelly
fish, sponge s , and the l ike-are sensitive without being sen
tient , but that " higher" animals are sentient. Like us, they
are not merely endowed with sensitive equipment of one
sort or another-equipment that responds differentially and
appropriately to one thing or another. They enj oy some fur
ther property, called sentience-so says popular opinion.
But what is this commonly proclaimed property?
What does sentience amount to, above and beyond sensi
tivity? This is a question that is seldom asked and has never
been properly answered. We shouldn 't assume that there 's a
good answer. We shouldn't assume , in other words , that it's
a good question. I f w e want t o u s e the concept of sentience ,

we will have t o construct i t from parts w e understand .
Everybody agrees that sentience requires sensitivity plus
some further as yet uni dentified factor x, so if we direct our
attention to the different varieties of sensitivity and the roles
in which they are exploited, keeping a sharp lookout for
something that strikes us as a crucial addition, we may dis
cover sentience along the way. Then we can add the phe
nomenon of sentience to our unfol ding story-or, alterna
tively, the whole idea of sentience as a special category may
evaporate. One way or another, we will cover the ground
that separates conscious us from the merely sensitive , insen
tient macromolecules we are descended from . One tempting
place to look for the key difference between sensitivity and
sentience is in the materials involved-the media in which
information travels and is transformed.
T H E M E D I A A N D T H E M E S S AG E S
We must look more closely at the development I sketched at

the beginning of chapter 2 . The earliest control systems were
really just body protectors . Plants are alive , but they don 't
have brains. They don't need them, given their lifestyle.
They do, however, need to keep their bodies intact and prop
erly situated to benefit from the immediate surroundings ,
and for this they evolved systems of self-governance or con
trol that took account of the crucial variables and reacted
accordingly. Their concerns-and hence their rudimentary
intentionality-was either directed inward , to internal con
ditions, or directed to conditions at the all-important bound
aries between the body and the cruel world. The responsibil
ity for monitoring and making adj ustments was distributed ,
not centralized. Local sensing of changing conditions could
be met by local reactions , which were largely independent of

each other. This could sometimes lead to coordination prob
lems , with one team of microagents acting at cross-purp oses
to another. There are times when independent decision mak
ing is a bad idea; if everybody decides to lean to the right
when the boat tips to the left , the boat may well tip over to
the right. But in the main , the minimalist strategies of plants
can be well met by highly distributed " decision making , "
mo destly coordinated b y the slow, rudimentary exchange of
information by diffusion in the fluids coursing through the
plant bo dy.
Might pl ants then just be " very slow animals , " enjoying
sentience that has been overlooked by us because of our
timescale chauvinism? Since there is no established mean
ing to the word " s entience , " we are free to adopt one of our
own choosing, if we can motivate it. We could refer to the
slow but reliable responsiveness of plants to their environ
ment as " s entienc e " if we wanted, but we would need some
reason to distinguish this quality from the mere sensitivity
exhibited by bacteria and other single-celled life-forms (to
say nothing of light meters in cameras) . There 's no ready
candidate for such a reason, and there 's a fairly compelling
reason for reserving the term " sentience " for something
more special : animals have slow body-maintenance systems
rather like those of p lants, and common opinion differenti
ates between the op eration of these systems and an animal 's
sentience .
Animals have h a d slow systems of body maintenance for
as long as there have been animals . Some of the molecules
floating along in such media as the bloo dstream are them
selves operatives that directly " do things " for the body (for
instance, some of them destroy toxic invaders in one-on-one
combat) , and some are more like messengers, whose arrival
at and " recognition " by some larger agent tells the larger
agent to " do thing s " (for instance , to speed up the heart rate
o r initiate vomiting) . Sometimes the larger agent is the entire

body. For instance , when the pineal gland in some species
detects a general decrease in daily sunlight , it broadcasts to
the whole body a hormonal message to begin preparing for
winter-a task with many subtasks , all set into motion by
one message . Although activity in these ancient hormonal
systems may be accompanied by powerful instances of what
we may presume to be sentience (such as waves of nausea,
or dizzy feelings , or chills , or pangs of lust) , these systems
operate independently of those sentient accompaniments
for instance , in sleeping or comatose animal s . Doctors speak
of brain-dead human beings kept alive on respirators as
being in a " vegetative state , " when these body-maintenance
systems alone are keeping life and limb together. Sentience
is gone , but sensitivity of many sorts p ersists , maintaining
various bodily balances. Or at least that 's how many people
would want to apply these two words .
In animals , this complex system of biochemical packets
of control information was eventually supplemented by a
swifter system , running in a different medium : traveling
pulses of electrical activity in nerve fibers . This opened up a
space of opportunities for swifter reactions , but also permit
ted the control to be differently distributed , because of the
different geometries of connection possible in this new sys
tem , the autonomic nervous system. The concerns of the
new system were still internal-or, at any rate, immediate in
both space and time : Should the body shiver now, or should
it sweat? Should the digestive processes in the stomach be
postponed because of more press ing needs for the blood sup
ply? Should the countdown to ej aculation begin? And so
forth . The interfaces between the new medium and the old
had to be worked out by evolution, and the history of that
development has left its marks on our current arrangements ,
making them much more complicated than one might have
expected. Ignoring these complexities has often led theorists
68 KINDS OF MINDS
of mind astray-myself included-so we should note them,

briefly.
One of the fundamental assumptions shared by many
modern theories of mind is known as fun ctionalism . The
basic idea is well known in everyday life and has many
proverbial expressions , such as handsome is as han dsome
does. What makes something a mind (or a belief, or a pain,
or a fear) is not what it is made of, but what it can do. We
appreciate this princip le as uncontroversial in other areas ,
especially in our assessment of artifacts. What makes some
thing a spark p lug is that it can be p lugged into a particular
situation and deliver a spark when called upon . That's all
that matters ; its color or material or internal complexity can
vary ad l ib , and so can its shap e , as long as its shape permits
it to meet the specific dimensions of its functional role. In
the world of living things , functionalism is widely appreci
ate d : a heart i s something for pumping blood, and an artifi
cial heart or a p ig 's heart may do just about as wel l , and
hence can be substituted for a diseased heart in a human
bo dy. There are more than a hundred chemically different
varieties of the valuable protein lysozyme. What makes them
all instances of lysozyme is what makes them valuable: what
they can do. They are interchangeable, for almost all intents
and purposes.
In the standard j argon of functionalism, these function
ally defined entities admit m ultiple realizations. Why
couldn't artificial minds , like artificial hearts , be made
real-realized-out of almost anything? Once we figure out
what minds do (what pains do, what beliefs do, and so on) ,
we ought to be able to make minds (or mind parts) out of
alternative materials that have those competences. And it
has seemed obvious to many theorists-myself included
that what minds do is process information ; minds are the
control systems of bodies , and in order to execute their
app ointed duties they need to gather, discriminate , store ,
transform , and otherwise process information about the con

trol tasks they perform . So far, so good. Functionalism, here
as elsewhere , promises to make life easier for the theorist by
abstracting away from some of the messy particularities of
performance and focusing on the work that is actually get
ting done . But it's almost standard for functionalists to over
simplify their conception of this task, making life too easy
for the theorist.
It's tempting to think of a nervous system (either an auto
nomic nervous system or its later companion, a central ner
vous system) as an information network tied at various spe
cific places-transducer (or inp ut) nodes and effector (or
output) nodes-to the realities of the body. A transducer is
any device that takes information in one medium (a change
in the concentration of oxygen in the blood, a dimming of
the ambient light , a rise in temperature) and translates it into
another medium. A photoelectric cell transduces light, in
the form of impinging photons, into an electronic signal , in
the form of electrons streaming through a wire . A micro
phone transduces sound waves into signals in the same elec
tronic medium. A bimetallic spring in a thermostat trans
duces changes in ambient temperature into a bending of the
spring (and that , in turn , is typically translated into the
transmission of an electronic signal down a wire to turn a
heater on or off) . The rods and cones in the retina of the eye
are the transducers of light into the medium of nerve signals ;
the eardrum transduces sound waves into vibrations , which
eventually get transduced (by the hair cells on the basilar
membrane) into the same medium of nerve signals . There
are temperature transducers distributed throughout the
bo dy, and motion transducers (in the inner ear) , and a host
of other transducers of other information. An effector is any
device that can be directed, by some signal in some medium ,
to make something happen in another " medium " (to bend an
arm , close a pore , secrete a flui d , make a noise) .
In a computer, there is a nice neat boundary between the

" outside" world and the information channels. The input
devices , such as the keys on the keyboard , the mouse , the
microphone , the television camera , all transduce informa
tion into a common medium-the electronic medium in
which "bits " are transmitted, store d , transformed. A com
puter can have internal transducers too , such as a tempera
ture trans ducer that " informs " the computer that it is over
heating, or a transducer that warns it of irregularities in its
p ower supp ly, but these count as inpu t devices , since they
extract information from the (internal) environment and put
it in the common medium of information processing.
It would be theoretically clean if we could insulate infor
mation channels from " outs ide " events in a body's nervous
system, so that all the important interactions happened at
identifiable transducers and effectors . The division of labor
this would p ermit is often very illuminating. Consider a ship
with a steering wheel located at some great distance from the
rudder it control s . You can connect the wheel to the rudder
with rop e s , or with gears and bicycle chains , wires and pul
leys , or with a hydraulic system of high-pressure hoses filled
with oil (or water or whiskey ! ) . In one way or another, these
systems transmit to the rudder the energy that the helmsman
supplies when turning the wheel. Or you can connect the
wheel to the rudder with nothing but a few thin wires ,
through which electronic signals pass . You don't have to
transduce the energy, just the information abo u t how the
helmsman wants the rudder to turn. You can transduce this
information from the steering wheel into a signal at one end
and put the energy in locally, at the other end , with an effec
tor-a motor of some kind . (You can also add " feedback"
message s , which are transduced at the motor-rudder end and
sent up to control the resistance-to-turning of the wheel, so
that the helmsman can sense the pressure of the water on the
rudder as it turns. This feedback is standard, these days, in
power steering in automobiles, but was dangerously missing

in the early days of power steering . )
If y o u o p t for this sort of system-a p ure signaling system
that transmits information and almost no energy-then it
really makes no difference at all whether the signals are elec
trons passing through a wire or photons passing through a
glass fiber or radio waves passing through empty space. In
all these cases , what matters is that the information not be
lost or distorted because of the time lags between the turning
of the wheel and the turning of the rudder. This is also a key
requirement in the energy-transmitting systems-the sys
tems using mechanical linkages , such as chains or wires or
hoses . That 's why elastic bands are not as good as unstretch
able cables , even though the information eventually gets
there , and why incompressible oil is better than air in a
hydraulic system . *
In modern machines , it is often possible in this way to iso
late the control system from the system that is controlled, so
that control systems can be readily interchanged with no loss
of function. The familiar remote controllers of electronic
appliances are obvious examples of this , and so are electronic
ignition systems (replacing the old mechanical linkages) and
other computer-chip-based devices in automobiles. And up to
a point , the same freedom from particular media is a feature of
animal nervous systems, whose parts can be quite clearly seg
regated into the peripheral transducers and effectors and the
intermediary transmission pathways. One way of going deaf,
for instance , is to lose your auditory nerve to cancer. The
* The example of the steering gear has an important historical p e di

gree. The term "cybernetics" was coined by Norbert Wiener from
the Greek word for " h e lmsman " or " steerer. " The word " governor "
comes from the sam e source. These ideas about how control is
accomp lished by the transmission and processing of information
were first clearly formulated by Wiener in Cybernetics; or, Control
and Communication in the Animal and the Machine ( 194 8 ) .
sound-sensitive parts of the ear are still intact, but the trans
mission of the results of their work to the rest of the brain has
been disrupted . This destroyed avenue can now be replaced
by a prosthetic link, a tiny cable made of a different material
(wire , just as in a standard computer) , and since the inter
faces at both ends of the cable can be matched to the require
ments of the existing healthy material s , the signals can get
through. Hearing is restored. It doesn 't matter at all what the
medium of transmission is, just as long as the information
gets through without loss or distortion.
This important theoretical idea sometimes leads to seri
ous confusions , however. The most seductive confusion
could be called the Myth of Double Transduction : first , the
nervous system transduces light , sound, temperature , and so
forth into neural signals (trains of impulses in nerve fibers)
and second , in some special central place , it transduces
these trains of impulses into some oth er medium, the
medium of consciousness ! That 's what Descartes thought ,
and he suggested that the pineal gland, right in the center of
the brain, was the place where this second transduction took
place-into the mysterious , nonphysical medium of the
mind. To day almost no one working on the mind thinks
there is any such nonphysical medium. Strangely enough ,
though , the idea of a second transduction into some special
physical or material medium , in some yet-to-be-identified
place in the brai n , continues to beguile unwary theorists . It
is as if they saw-or thought they saw-that since peripheral
activity in the nervous system was mere sensitivity, there
had to be some more central place where the sentience was
created . After all, a live eyeball , disconnected from the rest
of the brai n, cannot see, has no conscious visual experience,
so that must happen later, when the mysterious x is added to
mere sensitivity to yield sentience .
The reasons for the persistent attractiveness of this idea
are not hard to find. One is tempted to think that mere nerve
impulses couldn 't be the stuff of consciousness-that they

need translation, somehow, into something else. Otherwise ,
the nervous system woul d be like a telephone system with
out anybody home to answer the phone, or a television net
work without any viewers-or a ship without a helmsman . It
seems as if there has to be some central Agent or Boss or
Audience, to take in (to transduce) all the information and
appreciate it, and then " steer the ship . "
The idea that the network itself-by virtue o f its intricate
structure , and hence powers of transformation, and hence
capacity for controlling the body-could assume the role of
the inner Boss and thus harbor consciousness, seems prepos
terous. Initially. But some version of this claim is the materi
alist 's best hope. Here is where the very complications that
ruin the story of the nervous system as a pure information
processing system can be brought in to help our imagina
tions , by distributing a portion of the huge task of " ap precia
tion " back into the body.
" M Y B O DY H A S A M I N D O F I T S OW N ! "
Nature appears to have built the apparatus of

rationality not just on top of the apparatus of bio
logical regulation , but also from it and with it.
Antonio Damasio , Descartes ' Error: Emotion,
Reason, and th e Human Brain
The medium of information transfer in the nervous system is

electrochemical pulses traveling through the long branches
of nerve cells-not like electrons traveling through a wire at
the speed of light, but in a much-slower-traveling chain reac
tion . A nerve fiber is a sort of elongated battery, in which
chemical differences on the inside and outside of the nerve

cell's wall induce electric activities that then propagate
along the wall at varying speeds-much faster than molecule
packets could be shipped through fluid, but much, much
slower than the speed of light. Where nerve cells come in
contact with each other, at junctures called synapses , a
microeffector/microtrans ducer interaction takes place: the
electrical pulse triggers the release of neurotransmitter mole
cules , which cross the gap by old-fashioned diffusion (the
gap is very narrow) and are then transduced into further
electrical pulses. A step backward, one might think, into the
ancient world of molecular lock-and-key. Especially when it
turns out that in addition to the neurotransmitter molecules
(such as glutamate) , which seem to be more or less neutral
all-purp ose synap se crossers , there are a variety of neuro
modulator molecules , which, when th ey find the " locks " in
the neighboring nerve cells , produce all sorts of changes of
their own. Would it be right to say that the nerve cells trans
duce the presence of these neuromodulator molecules , in
the same way that other transducers " notice " the presence of
antigens , or oxygen, or heat? If so, then there are transducers
at virtually every j o int in the nervous system , adding input
to the stream of information already being carried along by
the electrical pulses. And there are also effectors every
where , secreting neuromodulators and neurotransmitters
into the " outside" world of the rest of the bo dy, where they
diffuse to produce many different effects . The crisp bound
ary between the information-processing system and the rest
of the world-the rest of the body-breaks down.
It has always been clear that wherever you have transduc
ers and effectors , an information system's " media-neutrality, "
or multiple realizability, disapp ears . In order to detect light ,
for instance, you need something photosensitive-some
thing that will respond swiftly and reliably to photons ,
amplifying their subatomic arrival into larger-scale events
that can trigger still further events . (Rho dopsin is one such
photosensitive substance, and this protein has been the
ma terial of ch oke f o a]] n a t u ral ey es, from ants to fish to
eagles to people. Artificial eyes might use some other photo
sensitive element , but not just anything will d o . ) In order to
identify and disable an antigen, you need an antibody that
has the right shape, since the identification is by the lock
and-key method. This limits the choice of antibody building
materials to molecules that can fold up into these shapes ,
and this severely restricts the molecules' chemical composi
tion-though not completely (as the example of lysozyme
varieties shows ) . In theory, every information-processing
system is tied at both ends , you might say, to transducers
and effectors whose physical composition is dictated by the
jobs they have to do; in between, everything can be accom
plished by media-neutral processes.
The control systems for ships , automobiles , oil refineries,
and other complex human artifacts are media-neutral , as
long as the media used can do the j ob in the avai lable time .
The neural control systems for animal s, however, are not
really media-neutral-not because the control systems have
to be made of particular materials in order to generate that
special aura or buzz or whatever, but because they evolved
as the control systems of organisms that already were lav
ishly equipped with highly distributed control system s , and
the new systems had to be built on top of, and in deep col
laboration with , these earlier system s , creating an astronomi
cally high number of points of transduction. We can occa
sionally ignore these ubiquitous interpenetrations of
different media-as , for instance , when we rep lace a single
nerve highway, like the auditory nerve, with a prosthetic
substitute-but only in a fantastic thought experiment could
we ignore these interpenetrations in general.
For example: The molecular keys needed to unlock the
locks that control every transaction between nerve cells are
glutamate molecules , dopamine molecules , and norepineph

rine molecules (among others) ; but " in principle" all the
locks could be changed-that i s , replaced with a chemically
different system . After all , the function of the chemical
depends on its fit with the lock, and hence on the subse
quent effects triggered by the arrival of this turn-on message ,
and not on anything else. But the distribution of responsibil
ity throughout the body makes this changing of the locks
practically impossible. Too much of the information process
ing-and hence information storage-is already embedded
in these particular materials. And that 's another good reason
why, when you make a mind, the materials matter. So there
are two good reasons for thi s : speed, and the ubiquity of
trans ducers and effectors throughout the nervous system . I
don't think there are any other good reasons.
These considerations lend support to the intuitively
appealing claim often advanced by critics of functionalism:
that it really does matter what you make a mind out of. You
couldn 't make a sen tien t mind out of silicon chips , or wire
and glas s , or beer cans tied together with string. Are these
reasons for abandoning functionalism? Not at all . In fact, they
depend on the basic insight of functionalism for their force.
The only reason minds depend on the chemical composi
tion of their mechanisms or media is that in order to do the
things these mechanisms must do, they have to be made , as
a matter of biohistorical fact , from substances compatible
with the preexisting bodies they control. Functionalism is
opposed to vitalism and other forms of mysticism about the
" intrinsic properties " of various substances . There is no
more anger or fear in adrenaline than there is silliness in a
bottle of whiskey. These substances , per se, are as irrelevant
to the mental as gasoline or carbon dioxide. It is only when
their abilities to function as comp onents of larger functional
systems depend on their internal composition that their so
called " intrinsic nature " matters .
The fact that your nervous system, unlike the control sys
tem of a modern shi p , is not an insulate d , media-neutral
control system-the fact that it " effects " and " transduces " at
almost every juncture-forces us to think about the func
tions of their parts in a more complicated (and realistic) way.
This recognition makes life slightly more difficult for func
tionalist philosophers of mind. A thousand philosophical
thought experiments (including my own story, " Where am
I?" [ 1 9 78 ] ) have exploited the intuition that I am not my
body but my body's . . . owner. In a heart transp lant opera
tion , you want to be the recipient, not the donor, but in a
brain transplant operation, you want to be the donor-you go
with the brain, not the body. In principle (as many philoso
phers have argued) , I might even trade in my current brain
for another, by rep lacing the medium while preserving only
the message. I could travel by teleportation, for instance, as
long as the information was perfectly preserved. In princi
ple, yes-but only because one would be transmitting infor
mation about the whole body, not just the nervous system.
One cannot tear me apart from my body leaving a nice clean
edge , as philosophers have often supposed. My body contains
as much of me, the values and talents and memories and dis
positions that make me who I am , as my nervous system does.
The legacy of Descartes's notorious dualism of mind and
body extends far beyond academia into everyday thinking :
"These athletes are prepared both mentally and physically, ''
and " There's nothing wrong with your bo dy-it's all in your
mind. " Even among those of us who have battled Descartes 's
vision , there has been a powerful tendency to treat the mind
(that is to say, the brain) as the body's boss , the pilot of the
shi p . Falling in with this standard way of thinking , we
ignore an important alternative : viewing the brain (and
hence the mind) as one organ among many, a relatively
recent usurper of contro l , whose functions cannot properly
be understood until we see it not as the boss but as just one
78 KI N DS OF M I N DS
more somewhat fractious servant , working to further the

interests of the body that shelters and fuels it and gives its
activities meaning.
This historical or evolutionary p erspective reminds me of
the change that has come over Oxford in the thirty years
since I was a student there . It used to be that the dons were
in charge, and the bursars and other bureaucrats, right up to
the vice chancellor, acted under their guidance and at their
b ehest . Nowadays the dons , like their counterparts on Amer
ican university faculties, are more clearly in the role of
employees hired by a central administration . But from
where , finally, does the University get its meaning? In evolu
tionary history, a similar change has crept over the adminis
tration of our bodies . But our bodies , like the Oxford dons ,
still have some power of decision-or, at any rate , some
p ower to rebel when the central administration acts in ways
that run counter to the sentiments of " the body politic . "
It is harder to think functionalistically about the mind
once we abandon the crisp identification of the mind with
the brain and let it spread to other p arts of the body, but the
compensations are enormous. The fact that our control sys
tems , unlike those o f ships and other artifacts , are so nonin
sulated permits our bodies themselves (as distinct from the
nervous systems they contain) to harbor much of the wis
dom that " w e " exploit in the course of daily decision mak
ing . Friedrich Nietzsche saw all this long ago , and put the
case with characteristic brio , in Th us Spake Zarath ustra (in
the section aptly entitled " On the Despisers of the Body " ) :
" B ody a m I , and soul "-thus speaks the child. And why
should one not speak like children? But the awakened
and knowing say : body am I entirely, and nothing else;
and soul is only a word for something about the bo dy.
The body is a great reason, a p lurality with one sense ,
a war and a p eace, a herd and a shepherd. An instrument
of your body is also your little reas on, my brother, which

you call " spirit"-a little instrument and toy of your great
reason. . . . B ehind your thoughts and feeling s , my
brother, there stands a mighty ruler, an unknown sage
whose name is self. In your body he dwells; he is your
body. There is more reason in your body than in your best
wisdom. (Kaufmann translation, 1 9 5 4 , p . 1 4 6 )
Evolution embodies information in every part of every

organism. A whale's baleen embodies information about the
food it eats, and the liquid medium in which it finds its
food. A bird's wing embodies information about the medium
in which it does il3 work. A chame leon's skin , more dramati
cally, carries information about its current environment. An
animal 's viscera and hormonal systems embody a great deal
of information about the world in which its ancestors have
lived. This information doesn 't have to be copied into the
brain at all. It doesn't have to be " represented" in " data
structures" in the nervous system. It can be exp loited by the
nervous system, however, which is designed to rely on, or
exploit, the information in the hormonal systems just as it is
designed to rely on, or exp loit, the information embodied in
the limbs and eyes . S o there is wisdom, particularly about
preferences, embodied in the rest of the bo dy. By using the
old bodily systems as a sort of sounding board , or reactive
audience, or critic, the central nervous system can be
guided-sometimes nudged, sometimes slammed-into wise
policies. Put it to the vote of the body, in effect. To be fair to
poor old Descartes , we should note that even he saw-at
least dimly-the imp ortance of this union of body and mind:
By means of these feelings of pain, hunger, thirst, and so

on, nature also teaches that I am present to my body not
merely in the way a seaman is present to his ship , but
that I am tightly joined and, so to speak, mingled together
80 KIN DS 0F MIN DS
with it, so much so that I make up one single thing with

it. (Meditation S ix)
When all goes wel l , harmony reigns and the various

sources of wisdom in the body coop erate for the benefit of
the whole, but we are all too familiar with the conflicts that
can provoke the curious outburst " My body has a mind of its
own ! " S ometime s , apparently, it is tempting to lump
together some of this embodied information into a separate
mind. Why? B ecause it is organized in such a way that it can
sometimes make somewhat independent discriminations ,
consult preferences, make decisions , enact policies that are
in competition with your mind. At such times , the Cartesian
perspective of a puppeteer self trying desperately to control
an unruly body-puppet is very p owerfu l . Your body can vig
orously betray the secrets yo u are desperately trying to
keep-by blushing and trembling or sweating, to mention
only the most obvious cases . It can " decide " that in spite of
your well-laid p lans , right now would be a good time for sex,
not intellectual discussion , and then take embarrassing steps
in preparation for a coup d ' etat. On another occasion , to
your even greater chagrin and frustration , it can turn a deaf
ear on your own efforts to enlist it for a sexual campaign,
forcing you to raise the volume , twirl the dial s , try all man
ner of preposterous cajo lings to persuade it.
But why, if our bodies already had minds of their own,
did they ever go about acquiring additional minds-our
minds? Isn't one mind per body enough? Not always . As we
have seen , the old body-based minds have done a robust j ob
of keeping life and limb together over billions of years , but
they are relatively slow and relatively crude in their discrim
inatory powers . Their intentionality is short-range and easily
tricked. For more sophisticated engagements with the world,
a swifter, farther-seeing mind is called for, one that can pro
duce more and better future .
CH A P T E R 4
· · · · · · · · · · · · · · · · · • · · · · · · · · · · · · · · · · · · · · · · · · • · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · ·
H OW I N T E N T I O N A L I T Y
CA M E I N T O F O C U S
T H E TOW E R O F G E N E R AT E - A N D - T E S T *
In order t o see farther ahead i n time , i t helps t o see farther

into space. What began as internal and p eripheral monitor
ing systems slowly evolved into systems that were capable
of not just proximal (neighboring) but distal (distant) dis
crimination. This is where perception comes into its own.
The sense of smel l , or olfaction, relies on the wafting from
afar of harbinger keys to local locks . The traj ectories of these
harbingers are relatively slow, variable, and uncertain ,
because of random dispersal and evaporation; thus informa
tion about the source they emanate from is limited. Hearing
depends on sound waves striking the system's transducers ,
and because the paths of sound waves are swifter and more
regular, perception can come closer to approximating
" action at a distance. " B ut sound waves can deflect and
bounce in ways that obscure their sourc e. Vision depends on
*This section is drawn, with revisions , from Darwin's Dangerous

Idea.
81
82 KINDS OF MINDS
the much swifter arrival of photons bounced off the things in

the world , on definitively straight-line traj ectories, so that
with a suitably shap ed pinhole (and optional lens) arrange
ment, an organism can obtain instantaneous high-fidelity
information about events and surfaces far away. How did
this transition from internal to proximal to distal intention
ality take p lace? Evolution created armies of specialized
internal agents to receive the information available at the
p eripheries of the body. There is just as much information
encoded in the light that falls on a pine tree as there is in the
l ight that falls on a squirrel , but the squirrel is equipped
with millions of information-seeking microagents , specifi
cally designed to take in, and even to seek out and interpret
this information.
Animals are not just herbivores or carnivores. They are ,
in the nice coinage of the p sychologist George Miller, infor
mavores. And they get their epistemic hunger from the com
bination, in exquisite organization, of the specific epistemic
hungers of millions of microagents, organized into dozens or
hundreds or thousands o f subsystems . Each of these tiny
agents can be conceived of as an utterly minimal intentional
system , whose life project is to ask a single question, over
and over and over-" Is my message coming in NOW? " "Is
my message coming in NOW? "-and springing into limited
but appropriate action whenever the answer is YES . Without
the epistemic hunger, there is no perception , no uptake .
Philosophers have often attempted to analyze perception
into the Given and what is then done with the Given by the
mind. The Given i s , of cours e , Taken , but the taking of the
Given is not something done by one Master Taker located in
some central headquarters of the animal 's brain. The task of
taking is distributed among all the individually organized
takers . The takers are not just the p eripheral transducers
the rods and cones on the retina of the eye , the specialized
cells in the epithelium of the nose-but also all the internal
H O W I N T E N T I O NALITY CAM E I N TO F O C U S 83
functionaries fed by them, c el l s an d groups of ce l l s con

nected in networks throughout the brain . They are fed not
patterns of light or pressure (the pressure of sound waves
and of touch) but patterns of neuronal impulses ; but aside
from the change of diet, they are playing similar roles. How
do all these agents get organized into larger systems capable
of sustaining ever more sophisticated sorts of intentionality?
By a process of evolution by natural selection, of course, but
not j ust one process .
I want t o propose a framework in which w e can place the
various design options for brains , to see where their p ower
comes from. It is an outrageously oversimplified structure ,
but idealization is the price one should often be willing to
pay for synoptic insight. I call it the Tower of Generate-and
Test. As each new floor of the Tower gets constructed, it
empowers the organisms at that level to find better and bet
ter moves , and find them more efficiently.
The increasing power of organisms to produce future can
be represented , then , in a series of steps . These steps almost
certainly don 't represent clearly defined transitional periods
in evolutionary history-no doubt such steps were taken in
overlapping and nonuniform ways by different lineages-but
the various floors of the Tower of Generate-and-Test mark
important advances in cognitive power, and once we see in
outline a few of the highlights of each stage , the rest of the
evolutionary steps will make more sense.
In the beginning , there was Darwinian evolution of
species by natural selection. A variety of candi date organ
isms were blindly generated , by more or less arbitrary
processes of recombination and mutation of genes . These
organisms were fiel d-tested , and only the best designs sur
vived. This is the ground floor of the tower. Let us call its
inhabitants Darwinian creatures.
This process went through many millions of cycles , pro
ducing many wonderful designs , both plant and animal.
84 KINDS 0F MINDS
�
Darwinian creatures, different
"hardwired" phenotypes
�
$
...
.
0
::i
a
g
...,
selection of one favored

phenotype
multiplication of the favored

genotype
F I G U RE 4. 1
Eventually, among its novel creations were some designs

with the prop erty of ph enotypic plasticity: that is, the indi
vi dual candidate organisms were not wholly designed at
birth; there were elements of their design that could be
adjusted by even ts that occurred during th e field tests. Some
of these candidate s , we may suppose, were no better off than
their cousins , the hardwired Darwinian creatures , since they
had no way of favoring (selecting for an encore) the behav-
H OW I N T E N T I O NALITY CAM E I NTO F O C U S 85
ioral options they were equipped to " try out. " But others , w e
may suppose, were fortunate enough to have wired-in " rein
forcers " that happened to favor Smart Moves-that i s ,
actions that were better for the candidates than the available
alternative actions. These individuals thus confronted the
environment by generating a variety of actions , which they
tried out, one by one, until they found one that worked .
They detected that it worked only by getting a p ositive o r
negative signal fr o m t h e environment, which adjusted the
probability of that action 's being repro duced on another
occasion. Any creatures wired up wrong-with positive and
negative reinforcement reversed-would be doomed, of
course. Only those fortunate enough to be born with appro
priate reinforcers would have an advantage. We may call this
subset of Darwinian creatures Skinnerian creatures, since , as
the behaviorist psychologist B. F. Skinner was fon d of point
ing out, such " operant conditioning " is not just analogous to
Darwinian natural selection; it is an extension o f it: " Where
inherited behavior leaves off, the inherited modifiability of
the process of conditioning takes over. " ( 1 9 5 3 , p. 8 3 )
The cognitive revolution that emerged in the 1 9 70s
ousted behaviorism from its dominant p osition in p sychol
ogy, and ever since there has been a tendency to underesti
mate the power of Skinnerian conditioning (or its variations)
to shape the behavioral comp etence of organisms into highly
adaptive and discerning structures. The flourishing work on
neural networks and " connectionis m " in the 1 99 0 s , how
ever, has demonstrated anew the often surprising virtuosity
of simple networks that begin life more or less randomly
wired and then have their connections adjusted by a simple
sort of " experience "-the history o f reinforcement they
encounter.
The fundamental idea of letting the environment p lay a
blind but selective role in shaping the mind (or brain or con
trol system) has a pedigree even older than Darwin. The
intellectual ancestors of today's connectionists and yester-
86 KINDS OF MINDS
Skinnerian creature
. . . until one is selected by

"reinforcement. "
Next time , the creature's

first choice will be the
reinforced response.
F I G U RE 4 . 2
day's behaviorists were the associationists : such philoso

phers as David Hum e , who tried in the eighteenth century to
imagine how mind parts (he called them impressions and
i d eas) could become self-organizing without benefit of some
all-too-knowing director of the organization. As a student
once memorably said to me , " Hume wanted to get the ideas
to think for themselves . " Hume had wonderful hunches
about how impressions and ideas might link themselves
together by a process rather like chemical bonding, and then
create beaten p aths of habit in the min d , but these hunches
H O W I NT E N T I O N A L I TY CA M E I N TO F O C U S 87
were t o o vague to be tested. Hum e 's ass ociati onism was ,

however, a direct inspiration for Pavlov's famous experi
ments in the conditioning of animal behavior, which led in
turn to the somewhat different conditioning theories of E . L .
Thorndike , Skinner, a n d the other behaviorists in psychol
ogy. S ome of these researchers-Donald Hebb , in particu
lar-attempted to link their behaviorism more closely to
what was then known about the brain. In 1 94 9 , Hebb pro
posed models of simple conditioning mechanisms that could
adjust the connections between nerve cells. These mecha
nisms-now called Hebbian learning rules-and their
descendants are the engines of change in connectionism , the
latest manifestation of this tradition.
Associationism, behaviorism, connectionism-in histori
cal and alphabetical order we can trace the evolution of
models of one simple kind of learning , which might well be
called ABC learning. There is no doubt that most animals are
capable of ABC learning; that i s , they can come to modify (or
redesign) their behavior in appropriate directions as a result
of a long, steady process of training or shaping by the envi
ronment. There are now good model s , in varying degrees of
realism and detai l , of how such a process of conditioning or
training can be nonmiraculously accomplished in a network
of nerve cells .
For many life-saving purposes (pattern recognition, dis
crimination, and generalization, and the dynamical control of
locomotion, for instance) , ABC networks are quite wonder
ful-efficient, compact, robust in performanc e , fault-tolerant,
and relatively easy to redesign on the fly. Such networks ,
moreover, vividly emphasize Skinner's point that it makes
little difference where we draw the line between the pruning
and shaping by natural selection which is genetically trans
mitted to offspring (the wiring you are born with) , and the
pruning and shaping that later takes place in the individual
(the rewiring you end up with, as a result of experience or
training) . Nature and nurture blend seamlessly together.
There are , however, some cognitive tricks that such ABC net
works have not yet been trained to perform , and-a more
telling criticism-there are some cognitive tricks that are
quite c learly not the result of training at all. Some animals
seem to be capable of " one-shot learning " ; they can figure
some things out without having to endure the arduous
process o f trial-and-error in the harsh world that is the hall
mark o f all ABC learning.
S kinnerian conditioning is a good thing as long as you
are not killed by one of your early errors . A better system
involves preselection among all the possible behaviors or
actions , so that the truly stupid moves are weeded out before
they're hazarded in " real life . " We human beings are crea
tures capable of this p articular refinement , but we are not
alone. We may call the beneficiaries of this third floor in the
Tower Popperian creatures, since , as the philosopher Sir
Karl Popper once elegantly put it, this design enhancement
" p ermits our hypotheses to die in our stead. " Unlike the
merely Skinnerian creatures , many of whom survive only
because they make lucky first move s, Popperian creatures
survive because they 're smart enough to make better-than
chance first moves. Of course they 're just lucky to be smart ,
but that's better than being j ust lucky.
How is this preselection in Popp erian agents to be done?
There must be a filter, and any such filter must amount to a
sort of inn er en vironmen t, in which tryouts can be safely
executed-an inner something-or-other structured in such a
way that the surrogate actions it favors are more often than
not the very actions the real world would also bless , if they
were actually p erformed. In short , the inner environment,
whatever it i s , must contain lots of information about the
outer environment and its regularities . Nothing else (except
magic ) could provide preselection worth having. (One could
always flip a coin or consult an oracle , but this is no
improvement over blind trial and error-unless the coin or
HOW I NT E N T I O NALITY CAM E I NTO FOCUS 89
oracle is systematically biased by someone or something that

has true information about the worl d . )
The beauty of Popper's idea is exemplified in the recent
development of realistic flight simulators used for training
airplane pilots . In a simulated worl d , pilots can learn which
moves to execute in which crises without ever risking their
lives (or expensive airplanes ) . As examples o f the Popperian
trick, however, flight simulators are in one regard mislead
ing : they reproduce the real worl d too literally. We must be
very careful not to think of the inner environment of a Pop
perian creature as simply a replica o f the outer worl d , with
all the physical contingencies of that world repro duced. In
such a miraculous toy world , the little hot stove in your
head would be hot enough to actually burn the little finger
in your head that you placed on it! Nothing of the sort needs
to be supposed. The information about the effect of putting a
Popperian creature has

an inner selective
environment that
previews candidate acts.
First time, the creature

acts in an insightful
way (better than
chance).
FIGURE 4.3
90 KINDS OF MINDS
finger on the stove has to be in there , and it has to be in there

in a form that can produce its premonitory effect when
called upon in an internal trial , but this effect can be
achieved without constructing a replica worl d. After all , it
would be equally Popperian to educate pilots just by having
them read a book that explained to them all the contingen
cies they might encounter when they eventually climbed
into the cockpit. It might not be as powerful a method of
learning , but it would be hugely better than trial-and-error in
the sky ! The common element in Popperian creatures is that
one way or another (either by inheritance or by acquisition)
information is installed in them-accurate information
about the world that they (probably) will encounter-and
this information is in such a form that it can achieve the pre
selective effects that are its raison d ' etre .
One of the ways Popperian creatures achieve useful filter
ing is by putting candidate behavioral options before the
b odily tribunal and exploiting the wisdom, however out-of
date or shortsighted, accumulated in those tissues. If the
b o dy rebels-for example, in such typical reactions as nau
sea , vertig o , or fear and trembling-this is a semireliable sign
(better than a coin flip) that the contemplated act might not
be a good idea. Here we see that rather than rewiring the
brain to eliminate these choice s, making them strictly
unthinkable, evolution may simply arrange to respond to
any thinking of them with a negative rush so strong as to
make them highly unlikely to win the competition for execu
tion. The information in the body that grounds the reaction
may have been placed there either by genetic recipe or by
recent individual experience . When a human infant first
learns to crawl , it has an innate aversion to venturing out
onto a pane of supportive glas s , through which it can see a
" visual cliff. " Even though its mother beckons it from a few
feet away, cajoling and encouraging, the infant hangs back
fearfully, desp ite never having suffered a fall in its life . The
H OW I NTENTIONALITY CAM E I NTO FOCUS 91
experience of its ancestors is making it err on the side of

safety. When a rat has eaten a new kind of foo d and has then
been injected with a drug that causes it to vomit, it will sub
sequently show a strong aversion to food that looks and
smells like the food it ate just before vomiting. Here the
information leading it to err on the side o f safety was
obtained from its own experienc e . Neither filter i s p erfect
after all , the pane o f glass is actually safe , and the rat's new
food is actually nontoxic-but better safe than sorry.
Clever experiments by p sychologists and ethologists sug
gest other ways in which animals can try actions out " in
their heads " and thereby reap a Popperian benefit. In the
1 9 30s and 1 94 0 s , behaviorists demonstrated to themselves
time and again that their experimental animals were capable
of " latent learning" about the world-learning that was not
specifically rewarded by any detectable reinforcement.
(Their exercise in self-refutation is itself a prime example of
another Popperian theme : that science makes progress only
when it poses refutable hypotheses . ) If left to explore a maze
in which no food or other reward was present , rats would
simply learn their way around in the normal course of
things ; then, if something they valued was introduce d into
the maze, the rats that had learned their way around on ear
lier forays were much better at finding it (not surprisingly)
than the rats in the control group , which were seeing the
maze for the first time. This may seem a paltry discovery.
Wasn't it always obvious that rats were smart enough to
learn their way around? Yes and n o . It may have seemed
obvious , but this is just the sort of testing-testing against
the background of the null hypothesis-that must b e con
ducted if we are going to be sure just how intelligent , how
mindfu l , various species are . As we shall see, other experi
ments with animals demonstrate surprisingly stupid
streaks-almost unbelievable gap s in the animal s ' knowl
edge of their own environments .
The behaviorists trie d valiantly to accommo date latent

learning into their ABC models. One of their most telling
stopgaps was to postulate a " curiosity drive , " which was sat
isfied (or " reduce d , " as they said) by exploration. There was
reinforcement going on after all in those nonreinforcing
environments . Every environment, marvelous to say, is full
of reinforcing stimuli simply by being an environment in
which there is something to learn. As an attempt to save
orthodox behaviorism, this move was manifestly vacuous ,
but that does not make it a hopeless idea in other contexts ; it
acknowledges the fact that curiosity-epistemic hunger
must drive any p owerful learning system.
We human beings are conditionable by ABC training, so
we are Skinnerian creatures , but we are not just Skinnerian
creatures . We also enjoy the benefits of much genetically
inherited hardwiring , so we are Darwinian creatures as well.
But we are more than that. We are Popperian creatures.
Which other animals are Popperian creatures , an d which are
merely Skinnerian? Pigeons were Skinner's favorite experi
mental animal s , and he and his fol lowers developed the
technology of operant conditioning to a very sophisticated
leve l , getting p igeons to exhibit remarkably bizarre and
sophisticated learned behaviors . Notoriously, the Skinneri
ans never succeeded in proving that pigeons were not Pop
p erian creatures ; and research on a host of different species ,
from octopuses to fish to mammals , strongly suggests that if
there are any purely Skinnerian creatures , capable only of
blind trial-and-error learning, they are to be found among
the simple invertebrates. The huge sea slug (or sea hare)
Aplysia californica has more or less rep laced the pigeon as
the focus of attention among those who study the mecha
nisms of simple conditioning.
We do not differ from all other species in being Popper
ian creatures then . Far from it; mammals and birds , reptiles,
amphibian s , fish, and even many invertebrates exhibit the
H O W I NT E N T I O NA LITY CA M E I N TO F O C U S 93
capacity to use general information they obtain from their

environments to presort their behavioral options before
striking out. How does the new information about the outer
environment get incorp orated into their brains? By percep
tion, obviously. The environment contains an embarrass
ment of riches , much more information than even a cogni
tive angel could use. Perceptual mechanisms designed to
ignore most of the flux of stimuli concentrate on the most
useful , most reliable information. And how does the infor
mation gathered manage to exert its selective effect when the
options are " considere d , " help ing the animal design ever
more effective interactions with its world? There are no
doubt a variety of different mechanisms and methods , but
among them are those that use the body as a sounding board.
T H E S EARC H F O R S E N T I E N C E :
A P R O G R E S S R E P O RT
We have been gradually adding elements to our recipe for a

mind. Do we have the ingredients for sentience yet? Cer
tainly the normal behavior of many of the animals we have
been describing passes our intuitive tests for sentience with
flying col ors. Wa tching a p uppy or a baby tremble wi th fear
at the edge of an apparent precipice , or a rat grimacing in
apparent disgust at the odor of supposedly toxic foo d , we
have difficulty even entertaining the hypothesis that we are
not witnessing a sentient being. But we have also uncovered
substantial grounds for caution: we have seen some ways in
which surprisingly mindlike behavior can be produced by
relatively simple , mechanical , apparently unmindlike con
trol systems. The p otency of our instinctual responses to
sheer speed and lifelikeness of motion, for instance , should
alert us to the genuine-not merely philosophical-p ossibil-
ity that we can be fooled into attributing more subtlety, more

understanding , to an entity than the circumstances warrant.
Recognizing that observable behavior can enchant us, we
can appreciate the need to ask further questions-about
what lies behind that behavior.
Consider pain. In 1 98 6 , the British government amended
its laws protecting animals in experiments , adding the octo
pus to the privileged circle of animals that may not be oper
ated upon without anesthesia. An octopus is a mollusk,
physiologically more like an oyster than a trout (let alone a
mammal ) , but the behavior of the octopus and the other
cephalopods (squid , cuttlefish) is so strikingly intelligent
and-apparently-sentient that the scientific authorities
decided to let behavioral similarity override internal differ
ence : cephalopods (but not other mollusks) are officially
presumed to be capable of feeling pain-j ust in case they
are . Rhesus monkeys , in contrast, are physiologically and
evolutionarily very close to us , so we tend to assume that
they are capable of suffering the way we d o , but they exhibit
astonishingly different behavior on occasion. The primatolo
gist Marc Hauser has told me in conversation that during
mating season the male monkeys fight ferociously, and it is
not uncommon to see one male pin another down and then
bite and rip out one of its testicles . The injured male does
not shriek or make a facial expression but simply licks the
wound and walks away. A day or two later, the wounded
animal may be observed mating ! It is hard to believe that
this animal was experiencing anything like the agonies of a
human being similarly afflicted-the mind reels to think of
it-in spite of our biological kinship . S o we can no longer
hope that the physiological and behaviorial evidence will
happily converge to give us unequivocal answers , since we
already know cases in which these two sorts of compelling if
inconclusive evidence pull in opposite directions. How then
can we think about this issue?
HOW I NTENTIO NALITY CAM E I NTO FOCUS 95
A key function of pain is negative reinforcement-the

" punishment " that diminishes the likelihood of a repeat p er
formance-and any Skinnerian creature can be trained by
negative reinforcement of one sort or another. Is all such neg
ative reinforcement pain? Experienced pain? Could there be
unconscious or unexperienced pain? There are simple mech
anisms of negative reinforcement that provide the behavior
shaping or pruning power of pain with apparently no further
mindlike effects , so it would be a mistake to invoke sentience
wherever we find Skinnerian conditioning. Another func
tion of pain is to disrupt normal patterns o f bodily activity
that might exacerbate an inj ury-pain causes an animal to
favor an injured limb until it can men d , for instance-and
this is normally accomplished by a flood of neurochemicals
in a self-sustaining loop of interaction with the nervous sys
tem. Does the presence of those substances then guarantee
the occurrence of pain? No , for in themselves they are just
keys floating around in search of their locks ; if the cycle of
interaction is interrupted, there is no reason at all to suppose
that pain persists. Are these particular substances even neces
sary for pain? Might there be creatures with a different system
of locks and keys ? The answer may depend more on histori
cal processes of evolution on this p lanet than on any intrin
sic properties of the substances . The example o f the octopus
shows that we should look to see what variations in chemi
cal implementation are to be found, with what differences in
function , but without expecting these facts in themselves to
settle our question about sentience.
What then about the other features of this cycle of inter
action? How rudimentary might a pain system be and still
count as sentience? What would be relevant and why? Con
sider, for instance , a toad with a broken leg. Is this a sentient
being experiencing pain? It is a living being whose normal life
has been disrupted by damage to one of its parts , preventing
it from engaging in the behaviors that are its way of earning a
96 K I N DS OF M I N DS
living. It is moreover in a state with p owerful negative-rein

forcement p otential-it can readily be conditioned to avoid
such states of its nervous system . This state is maintained by
a cycle of interaction that somewhat disrupts its normal dis
p ositions to leap-though in an emergency it will leap any
way. It is tempting to see all this as amounting to pain. But it
is also tempting to endow the toad with a soliloquy, in
which it dreads the prospect of such an emergency, yearns
for relief, deplores its relative vulnerability, bitterly regrets
the foolish actions that led it to this crisis, and so forth , and
these further accompaniments are not in any way licensed
by anything we know about toads. On the contrary, the more
we learn about toads , the more confident we are becoming
that their nervous systems are designed to carry them through
life without any such expensive reflective capacities.
S o what? What does sen tience have to do with such fancy
intellectual talents? A good question, but that means we
must try to answer it, and not just use it as a rhetorical ques
tion to deflect inquiry. Here is a circumstance in which how
we ask the questions can make a huge difference , for it is
possible to bamboozle ourselves into creating a phantom
problem at this point. How? By losing track of where we
stand in a process of addition and subtraction. At the outset,
we are searching for x, the special ingredient that distin
guishes mere sensitivity from true sentience , and we work
on the project from two directions . Working up from simple
cases , adding rudimentary versions of each separate feature ,
we tend to be unimpressed: though each of these p owers is
arguably an essential component of sentience, there is surely
more to sentience than that-a mere robot could well exhibit
th at without any sentience at all ! Working down , from our
own richly detailed (and richly appreciated) experience , we
recognize that other creatures manifestly lack some of the
p articularly human features of our experience , so we sub
tract them as inessential . We don't want to be unfair to our
H OW I N T E N T I O NA L I TY CAM E I NTO F O C U S 97
animal cousins . S o while we recognize that much of what

we think of when we think of the awfulness of p ain (and
why it matters morally whether someone is in p ain) involves
imagining just these anthrop omorphic accompaniments , we
generously decide that they are j ust accompaniments , not
" essential " to the brute phenomenon of sentience (and its
morally most significant instance, pain) . What we may tend
to overlook, as these two ships pass in the night, is the possi
bility that we are subtracting, on one p ath , the very thing we
are seeking on the other. If that's what we're doing , our con
viction that we have yet to come across x-the "missing
link" of sentience-would be a self-induced illusion.
I don 't say that we are making an error of this sort , but
j ust that we migh t well be doing so . That's enough for the
moment, since it shifts the burden of proof. Here , then, is a
conservative hypothesis about the problem of sentience: There
is no such extra phenomenon. " S entienc e " comes in every
imaginable grade or intensity, from the simplest and most
" robotic , " to the most exquisitely sensitive, hyp er-reactive
" human. " As we saw in chapter 1 , we do indeed have to
draw lines across this multistranded continuum of case s ,
because having moral policies requires i t , b u t t h e prospect
that we will discover a threshold-a morally s ignificant
" step , " in what is otherwise a ramp-is not only extremely
unlikely but morally unappealing as well .
Consider the toad once again in this regard. On which
side of the line does the toad fall? (If toads are too obvious a
case for you one way or the other, choose whatever creature
seems to occupy your penumbra of uncertainty. Choose an
ant or a j ellyfish or a pigeon or a rat . ) Now suppose that " sci
ence confirms " that there is minimal genuine sentience i n
th e toad-that a toad's " p ai n " is real , experienced p ain , for
instance. The toad now qualifies for t h e special treatment
reserved for the sentient. Now suppose instead that the toad
turns out not to have x, once we have determined what x i s .
In this case , the toad's status falls to " mere automaton , "
something that w e may interfere with i n any imaginable way
with no moral compunction whatever. Given what we
already know about toads , does it seem plausible that there
could be some h eretofore unimagined feature the discovery
o f which could j usti fy this enormous difference in our atti
tude? Of cours e , if we discovered that toads were really tiny
human beings trapped in toad bodies , like the prince in the
fairy tal e , we would immediately have grounds for the
utmost solicitude, for we would know that in spite of all
behavioral appearances , toads were capable of enduring all
the tortures and anxieties we consider so important in our
own cases. But we already know that a toad is no such thing.
We are being asked to imagine that there is some x that is
nothing at all like being a human prince trapped in a toad
skin, but is nevertheless morally compelling. We also
already know, however, that a toad is not a simple wind-up
toy but rather an exquisitely complex living thing capable of
a staggering variety of self-protective activities in the further
ance of its preordained task of making more generations of
toads . Isn't that already enough to warrant some special
regard on our p art? We are being asked to imagine that there
is some x that is nothing at all like this mere sophistication
o f-control-structure , but that nevertheless would command
our moral appreciation when we discovered it. We are being
aske d , I suspect, to indulge in something beyond fantasy.
But let us continue with our search, to see what comes next ,
for we are still a long way from human minds .
F R O M P H OTOTA X I S TO M E TA P H Y S I C S
Once we get t o Popperian creatures-creatures whose brains

have the p otential to be endowed , in inner environments ,
H OW I NT E N T I O NALITY CAM E I NTO FOCUS 99
with preselective prowess-what happens next? Many dif

ferent things , no doubt, but we will concentrate on one par
ticular innovation whose powers we can clearly see. Among
the successors to mere Popperian creatures are those whose
inner environments are informed by the designed p ortions of
the outer environment. One of Darwin 's fundamental
insights is that design is expensive but copying designs is
cheap ; that i s , making an all new design is very difficult, but
redesigning old designs is relatively easy. Few of us could
reinvent the whee l , but we don't have to , since we acquired
the wheel design (and a huge variety of others) from the cul
tures we grew up in. We may call this sub-sub-subset of Dar
winian creatures Gregorian creatures, since the British psy
chologist Richard Gregory is to my mind the preeminent
theorist of the role of information (or more exactly, what
Gregory calls Potential Intelligence) in the creation of Smart
Moves (or what Gregory calls Kinetic Intelligence ) . Gregory
observes that a pair of scissors , as a well-designed artifact, is
not just a result of intelligence but an endower of intelli
gence (external potential intelligence) , in a very straightfor
ward and intuitive sense: when you give someone a p air of
scissors , you enhance their potential to arrive more safely
and swiftly at Smart Moves. ( 1 9 8 1 , pp. 3 1 1 ff. )
Anthropologists have long recognized that the advent of
tool use accompanied a major increase in intelligence . Chim
panzees in the wild go after termites by thrusting crudely
prepared fishing sticks deep into the termites' underground
homes and swiftly drawing up a stickful of termites , which
they then strip off the stick into their mouths . This fact takes
on further significance when we learn that not all chim
panzees have hit upon this trick; in some chimpanzee " cul
tures , " termites are an unexp loited food source. This
reminds us that tool use is a two-way sign of intelligenc e ;
not only d o e s i t require intelligence to recognize a n d main
tain a tool (let alone fabricate one ) , but a tool confers intelli-
I 00 KI N DS 0F MI N DS
gence on those lucky enough to be given one. The better

designed the tool (the more information there is embedded
in its fabrication) , the more p otential intelligence it confers
on its user. And among the preeminent tools , Gregory
reminds u s , are what he calls mind tool s : words .
Words and other mind tools give a Gregorian creature an
inner environment that p ermits it to construct ever more
subtle move generators and move testers . Skinnerian crea
tures ask themselve s , " What do I do next?" and haven 't a
clue how to answer until they have taken some hard knocks .
Popperian creatures make a big advance by asking them
selve s , " What should I think about next?" before they ask
themselves , " What should I do next?" (It should be empha
sized that neither Skinnerian nor Popp erian creatures actu
ally need to talk to themselves or think these thoughts. They
are simply designed to operate as if they had asked them
selves these questions . Here we see both the p ower and the
risk of the intentional stance : The reason that Popperian
creatures are smarter-more successfully devious , say-than
Skinnerian creatures is that they are adaptively responsive
Gregorian creature imports mind tools from the (cultural)

environment; these improve both the generators and the testers.
F I G U RE 4.4
HOW I N T E N T I O N A L I TY CAM E I NTO F O C U S 101
to more and better information , in a way that we can vividly

if loosely describe from the intentional stance, in terms of
these imaginary soliloquies. But it would be a mistake to
impute to these creatures all the subtleties that go along with
the ability to actually formulate such questions and answers
on the human model of exp licit self-questioning . ) Gregorian
creatures take a big step toward a human level of mental
adroitness , benefiting from the exp erience of others by
exploiting the wisdom embodied in the mind tools that
those others have invented , improve d , and transmitted;
thereby they learn how to think better about what they
should think about next-and so forth , creating a tower o f
further internal reflections with no fixed or discernible limit.
How this step to the Gregorian level might be accomplished
can best be seen by once more backing up and l ooking at the
ancestral talents from which these most human mental tal
ents must be constructed.
One of the simplest life-enhancing practices found in
many species is phototaxis-distinguishing light from dark
and heading for the light. Light is easily transduced , and
given the way light emanates from a sourc e , its intensity
diminishing gradually as you get farther away, quite a simple
connection between trans ducers and effectors can produce
reliable phototaxis . In the neuroscientist Valentino Braiten
berg's elegant little book Vehicles, we get the simplest
model-the vehicle in figure 4 . 5 . It has two light transduc
ers , and their variable output signals are fed , crossed, to two
effectors (think of the effectors as outboard motors) . The
more light trans duced, the faster the motor runs . The trans
ducer nearer the light source will drive its motor a bit faster
than the transducer farther from the light , and this will
always turn the vehicle in the direction of the l ight, till even
tually it hits the light source itself or orbits tightly around it.
The world of such a simple being is graded from light to
not-so-light to dark, and it traverses the gradient. It knows ,
1 02 K I N DS OF M I N D S
'\. I I /
:::Q::I I
"
\
\
F I G U RE 4.5
and needs to know, nothing else. Light recognition is almost

for free-whatever turns on the transducer is light, and the
system doesn't care whether it's the very same light that has
returned or a new light. In a world with two moons , it might
make a difference , ecologically, which moon you were track
ing ; moon recognition or identification could be an addi
tional problem that needed a solution. Mere phototaxis
would not be enough in such a worl d . In our worl d , a moon
is not the sort of obj ect that typically needs reidentifying by
a creature ; mothers , in contrast, often are .
Mama ta xis-homing in on Mother-is a considerably
more sophisticated talent. If Mama emitted a bright light ,
phototaxis might do the j ob , but not if there were other
mothers in the vicinity, all using the same system. If Mama
HOW I NTENTIO NALITY CAM E I NTO FOCUS 1 03
then emitted a particular blue light, different from the light

emitted by every other mother, then putting a p articular
everything-but-blue filter on each of your phototransducers
would do the j ob quite well . Nature often relies on a similar
princip l e, but using a more energy-efficient medium. Marna
emits a signature odor, distinguishably different from all
other odors (in the immediate vicinity) . Marnataxis (rnother
reidentification and horning) is then accomplished by odor
trans duction, or olfaction. The intensity of o dors is a func
tion of the concentration of the molecular keys as they
diffuse through the surrounding medium-air or water. A
transducer can therefore be an appropriately shaped lock,
and can follow the gradient of concentration by using an
arrangement just like that in Braitenberg's vehicle. Such
olfactory signatures are ancient, and potent. They have been
overlaid, in our species , by thousands of other mechanisms ,
but their position in the foundation is still discernible. In
spite of all our sophistication, odors move us without our
knowing why or how, as Marcel Proust famously noted . *
Technology honors the same design principle in yet
another medium : the EPIRB (Emergency Position Indicating
Radio B eacon) , a self-containe d , battery-powered radio
transmitter that repeats over and over again a p articular sig
nature at a particular frequency. You can buy one in a
marine hardware store and take it with you on your sailboat.
Then if you ever get in distress , you turn it on. Immediately
* O dors are not used only for identification signals. They often p lay
powerful roles in attracting a mate or even suppressing the sexual
activity or maturation of one 's rivals. Signals from the olfactory
bulb bypass the thalamus on their way to the rest of the brain, so in
contrast to the signals arising in vision, hearing , and even touch ,
olfactory commands go directly to the old control centers, elimi
nating many middlemen. It is likely that this more direct route
helps to exp lain the peremptory , nearly hypnotic power some
odors have over us.
1 04 KINDS OF M I N DS
a worldwide tracking system senses your EPIRB 's signal and

indicates its position with a blip on an electronic map . It
also looks up the signature in its giant table of signatures
and thereby identifies your boat. Identification greatly sim
p lifies search and rescue, since it adds redundancy: the bea
con can be homed in on blindly by radio receivers (transduc
ers) , but as the rescuers get close it helps if they know
whether they are looking (with their eyes) for a black fishing
trawler, a small dark-green sailboat , or a bright-orange rub
ber raft. Other sensory systems can be brought in to make the
final approach swifter and less vulnerable to interruption
(should the EPIRB 's battery run down , for instance ) . In ani
mals , odor tracking is not the only medium of Mamataxis.
Visual and auditory signatures are also relied on, as the
ethologist Konrad Lorenz has notably demonstrated in his
pioneering studies of " imprinting " in young geese and
ducks. Chicks that are not imprinted shortly after birth with
a proper Mama signature will fix on the first large moving
thing they see and treat it as Mama thereafter.
Beacons (and their complement of beacon sensors) are
good design solutions whenever one agent needs to track
(recognize, reidentify) a particular entity-typ ically another
agent, such as Mama-for a long time . You just install the
beacon in the target in advance , and then let it roam . (Anti
car-theft radio beacons that you hide in your car and then
remotely turn on if your car is stolen are a recent manifesta
tion. ) But there are costs , as usual . One of the most obvious
is that friend and foe alike can use the tracking machinery to
home in on the target. Predators are typ ically tuned to the
same olfactory and auditory channels as offspring trying to
stay in touch with Mama, for instance .
Odors and sounds are broadcast over a range that is not
easily in the control of the emitter. A low-energy way of
achieving a more selective beacon effect would be to put a
particular blue spot (pigment of one sort or another) on
HOW I NTENTIO NALITY CAM E I NTO FOCUS 1 05
Mama , and let the reflected light of the sun create a beacon
visible only in particular sectors of the world and readily
extinguished by Mama's simply moving into the shadows .
The offspring can then follow the blue spot whenever it is
visible. But this setup requires an investment in more
sophisticated photosensitive machinery : a simple eye , for
instance-not just a pair of photocells.
The ability to stay in reliably close contact with one par
ticular ecologically very important thing (such as Mama)
does not require the ability to conceive of this thing as an
enduring particular entity, coming and going. As we have
just seen , reliable Mamataxis can be achieved with a bag of
simple tricks . The talent is normally robust in simple envi
ronments, but a creature armed with such a simple system is
easily " foole d , " and when it is fooled, it trundles to its mis
fortune without any appreciation of its folly. There need be
no capability for the system to monitor its own success or
reflect on the conditions under which it succeeds or fails ;
that's a later (and expensive) add-on.
Cooperative tracking-tracking in which the target pro
vides a handy beacon and thus simplifies the task for the
tracker-is a step on the way toward competitive tracking , in
which the target not only provides no unique signature bea
con but actively tries to hide, to make itself untrackable.
This move by prey is countered by the development in preda
tors of general-purpose, track-anything system s , designed to
turn whatever aspects a trackworthy thing reveals into a sort
of private and temp orary beacon-a " search image , " created
for the nonce by a gaggle of feature-detectors in the predator
and used to correlate , moment by moment , the signature of
the target , revising and updating the search image as the tar
get changes , always with the goal of keeping the picked-out
obj ect in the cross-hairs .
It is important to recognize that this variety of tracking
does not require categorization of the target. Think of a prim-
1 06 K I N DS O F M I N DS
itive eye , consisting of an array of a few hundred photocells ,

transducing a changing pattern of pixels, which are turned
on by whatever is reflecting light on them . Such a system
could readily deliver a message of the following sort : "X, the
whatever-it-is responsible for the pixel-clump currently
under investigation, has just dodged to the right. " (It would
not have to deliver this message in so many words-there
need be no words , no symbols , in the system at all . ) So the
only identification such a system engages in is a degenerate
or minimal sort of moment-to-moment reidentification of the
something-or-other being tracked. Even here , there is toler
ance for change and substitution. A gradually changing
clump of p ixels moving against a more or less static back
ground can change its shape and internal character radically
and still be trackable, so long as it doesn't change too fast.
(The phi phen omen on , in which sequences of flashing lights
are involuntarily interpreted by the vision system to be the
trajectory of a moving obj ect, is a vivid manifestation of this
built-in circuitry in our own vision systems . )
What happens when X temporarily goes behind a tree?
The simpleminded solution is to keep the most recent ver
sion of the search image intact and then just scan around at
random, hoping to lock back onto this temp orary beacon
once again when it emerges , if it ever does . You can improve
the odds by aiming your search image at the likeliest spot for
the reappearance of the temporary beacon. And you can get
a better-than-a-coin-flip idea of the likeliest spot just by sam
pling the old traj ectory of the beacon and plotting its future
continuation in a straight line. This yields instances of pro
ducing future in one of its simplest and most ubiquitous
form s , and also gives us a clear case of the arrow of inten
tionality poised on a nonexistent but reasonably hoped-for
target.
This ability to " keep in touch with " another object (liter
ally touching and manipulating it, if possible) is the prereq-
H OW I NT E N T I O NALITY CAM E I NTO F O C U S 1 07
uisite for high-quality p erception. Visual recognition of a

particular person or obj ect , for instance , is almost impossible
if the image of the obj ect is not kept centered on the high
resolution fovea of the eye for an appreciable length of time.
It takes time for all the epistemically hungry microagents to
do their feeding and get organized. S o the ability to maintain
such a focus of information abo u t a particular thing (the
whatever-it-is I'm visually tracking right now) i s a precondi
tion for developing an identifying description o f the thing. *
The way to maximize the likelihood of maintaining or
restoring contact with an entity being tracked is to rely on
multiple independent systems , each fallible but with over
lapping domains of competence. Where one system lets
down the side, the others take over, and the result tends to
*This point about the primacy of tracking over descri ption is, I
think, the glimmer of truth in the otherwise forlorn philosophical
doctrine that there are two varieties of belief-de re beliefs, which
are somehow " directly" about their objects , and de dicta beliefs ,
which are about their objects only through the mediation of a dic
tum , a definite descri ption (in a natural language , or in some " lan
guage of thought " ) . The contrast is i l lustrated (supposedly) by the
difference between
believing that Tom ( that guy, right over there) is a man ,
and
believing that whoever it was that mailed this anonymous letter

to me is a man.
The intentionality in the first case is supposed to be somehow more

direct , to latch onto its object in a more primitive way. But, as we
have seen, we can recast even in the most direct and primitive
cases of perceptual tracking into the de dicta mode (the x such that
x is whatever is responsible for the p ixel-clump currently under
investigation has j ust j umped to the right) in order to bring out a
feature of the mechanism that mediates this most " immediate " sort
of reference. The difference between de re and de dicta is a differ
ence in the speaker's perspective or emphasis , not in the phenome
non. For more on this, see Dennett , " B eyond Belief" ( 198 2 ) .
be smooth and continuous tracking composed of intermit

tently functi oning elements .
How are these multiple systems linked together? There
are many p ossibilities . If you have two sensory systems , you
can link them by means of an AND-gate : they both have to
be turned ON by their input for the agent to respond posi
tively. (An AND-gate can be implemented in any medium ; it
isn't a thing , but a principle of organization. The two keys
that have to be turned to open a safe deposit box, or fire a
nuclear missi l e, are linked by an AND-gate. When you fasten
a garden hose to a spigot and put a controllable nozzle on
the other end, these ON-OFF valves are linked by an AND
gate; both have to be open for water to come out . ) Alterna
tively, you can link two sensory systems with an OR-gate :
either one by itself, A or B (or both together) , will evoke a
positive response from the agent. OR-gates are used to
include backup or spare subsystems in larger systems: i f one
unit fail s , the extra unit's activity is enough to keep the sys
tem going. Twin-engined p lanes link their engines by an OR
gate: two in working order may be best, but in a pinch , one is
enough.
As you add more systems , the possibility of linking them
in intermediate ways looms . For instance , you can link them
so that IF system A is ON, then if either B or C is ON, the
system is to respond p ositively; otherwise , both systems B
and C must be on to produce a p ositive response. (This is
equivalent to a maj ority rule linking the three systems ; if the
maj ority-any maj ority-is ON, the system will respond
p ositively. ) All the possible ways of linking systems with
AND-gates and OR-gates (and NOT-gates , which simply
reverse or invert the output of a system , turning ON to OFF
and vice versa) are called Boolean functions of those sys
tems , since they can be precisely described in terms of the
logical operators AND , OR, and NOT, which the nineteenth
century English mathematician George Boole first formal-
H OW I NTENTIO NALITY CAM E I NTO FOCUS 1 09
ized. But there are also non-B oolean ways that systems can
intermingle their effects . Instead of bringing all the contribu
tors to a central voting plac e , giving them each a single vote
(YES or NO, ON or OFF) , and thereby channeling their con
tribution to behavior into a single vulnerable decision point
(the summed effect of all the B oolean connections) , we
could let them maintain their own indep endent and contin
uously variable links to behavior and have the worl d extract
an outcome behavior as the result of all the activity.
Valentino Braitenberg 's vehicle, with its two cross-wired
phototransducers , is an utterly simple case in point. The
" decision " to turn left or right emerges from the relative
strength of the contributions of the two transducer-motor
systems, but the effect is not efficiently or usefully repre
sented as a Boolean function of the respective " arguments "
of the trans ducers . (In principle, the input-output behavior
of any such system can be approximated by a B oolean func
tion of its components , suitably analyzed, but such an ana
lytic stunt may fail to reveal what is imp ortant about the
relationships . Considering the weather as a B oolean system
is possible in principle, for instance , but unworkable and
uninformative . )
B y installing dozens o r hundreds o r thousands of such
circuits in a single organism , elaborate life-protecting activi
ties can be reliably controlled, all without anything happen
ing inside the organism that looks like thinking specific
though ts. There is plenty of as if decision making , as if rec
ognizing , as if hiding and seeking. There are also lots of
ways an organism, so equipped, can " make mistakes , " but its
mistakes never amount to formulating a representation of
some false proposition and then deeming it true.
How versatile can such an architecture be? It is hard to
say. Researchers have recently designed and test-driven arti
ficial control systems that produce many of the striking
behavioral patterns we observe in relatively simple life-
1 10 KINDS OF M I NDS
forms , such as insects and other invertebrates ; so it is tempt

ing to believe that all the astonishingly complex routines of
these creatures can be orchestrated by an architecture like
thi s , even if we don't yet know how to design a system of the
required complexity. After all , the brain of an insect may
have only a few hundred neurons in it, and think of the elab
orate engagements with the world such an arrangement can
oversee. The evolutionary biologist Robert Trivers notes , for
example:
Fungus-growing ants engage in agriculture . Workers cut

leave s , carry these into the nest, prepare them as a
medium for growing fungus , plant fungus on them, fertil
ize the fungus with their own droppings , weed out com
petitive species by hauling them away, and , finally, har
vest a special part of the fungus on which they feed.
( 1 98 5 , p . 1 7 2 )
Then there are the prolonged and intricately articulated mat

ing and child-rearing rituals of fish and birds . Each step has
sensory requirements that must be met before it is under
taken , and then is guided adaptively through a field of obsta
cles. How are these intricate maneuvers controlled? Biolo
gists have determined many of the conditions in the
environment that are used as cue s , by painstakingly varying
the available sources of information in experiments , but it is
not enough to know what information an organism can pick
up . The next difficult task is figuring out how their tiny
brains can be designed to put all this useful sensitivity to
information to good use.
If you are a fish or a crab or something along those lines ,
and one of your projects is, say, building a nest of pebbles on
the ocean floor, you will need a pebble-finder device, and a
way of finding your way back to your nest to deposit the
found pebble in an appropriate place before heading out
H OW I NTENTIO NALITY CAM E I NTO FOCUS 1 1 1
again. This system need not be foolproof, however. Since

impostor pebble-nests are unlikely to be surreptitiously
erected in place of your own during your foray (until clever
human experimenters take an interest in you ) , you can keep
your standards for reidentification quite low and inexpen
sive. If a mistake in " identification" occurs , you probably go
right on building, not just taken in by the ruse but com
pletely incapable of recognizing or appreciating the error,
not in the slightest bit troubled. On the other han d , if you
happen to be equipped with a backup system of nest i denti
fication , and the impostor nest fails the backup test, you will
be thrown into disarray, pulled in one direction by one sys
tem and in another by the other system. These conflicts hap
pen , but it makes uo sense to ask, as the organism rushes
back and forth in a tizzy, "Just what is it thinking now? What
is the propositional con ten t of its confused state? "
I n organisms such a s us-organisms equipped with many
layers of self-monitoring systems , which can check on and
attempt to mediate such conflicts when they arise-it can
sometimes be all too clear just what mistake has been made.
A disturbing example is the Capgras delusion , a bizarre
affliction that occasionally strikes human beings who have
suffered brain damage. The defining mark of the Capgras
delusion is the sufferer's conviction that a close acquain
tance (usually a loved one) has been replaced by an impostor
who looks like (and sounds like , and acts like) the genuine
companion, who has mysteriously disappeare d ! This amaz
ing phenomenon should send shock waves through philoso
phy. Philosophers have made up many far-fetched cases of
mistaken identity to illustrate their various philosophical
theories , and the literature of philosophy is crowded with
fantastic thought experiments about spies and murderers
traveling incognito , best friends dressed up in gorilla suits ,
and long-lost identical twins , but the real-life cases of Cap
gras delusion have so far escaped philosophers ' attention.
1 12 KINDS OF MINDS
What is p articularly surprising about these cases is that they

don't depend on subtle disguises and fleeting glimpses. On
the contrary, the delusion p ersists even when the target indi
vidual is closely scrutinized by the agent , and is even plead
ing for recognition. Capgras sufferers have been known to
murder their spouses , so sure are they that these look-alike
interlopers are trying to step into shoes-into whole lives
that are not rightfully theirs ! There can be no doubt that in
such a sad cas e , the agent in question has deemed true some
very specific propositions of nonidentity: This man is not
my h u sban d; this man is as qualitatively similar to my hus
band as ever can be, and yet he is not my husband. Of p artic
ular interest to us is the fact that people suffering from such
a delusion can b e quite unable to say why they are so sure .
The neuropsychologist Andrew Young ( 1 994) offers an
ingenious and p lausible hypothesis to explain what has
gone wrong. Young c ontrasts Capgras delusion with another
curious affliction caused by brain damage: prosopagnosia.
People with prosopagnosia can't recognize familiar human
faces. Their eyesight may be fine, but they can ' t i dentify
even their closest friends until they hear them sp eak. In a
typical experiment, they are shown collections of pho
tographs: some photos are of anonymous individuals and
others are of family members and celebrities-Hitler, Mari
lyn Monro e , John F. Kennedy. When asked to pick out the
familiar faces , their p erformance is no better than chance.
But for more than a decade researchers have suspected that
in spite of this shockingly poor p erformance, something in
some prosopagnosics was correctly identifying the family
members and the famous people , since their bodies react dif
ferently to the familiar faces. If, while looking at a photo of a
familiar face, they are told various candidate names of the
p erson p icture d , they show a heightened galvanic skin
response when they hear the right name. (The galvanic skin
response is the measure of the skin's electrical conductance
H OW I NT E N T I O N A LITY CAM E I N TO F O C U S 1 13
and is the primary test relied on in polygraphs, or " lie detec

tors . " ) The conclusion that Young and other researchers
draw from these results is that there must be two (or more)
systems that can identify a face, and one of these is spared in
the prosopagnosics who show this response. This system
continues to do its work wel l , covertly and largely unnoticed.
Now suppose , Young says , that Capgras sufferers have j ust
the opposite disability: the overt , conscious face-recognition
system ( or systems) works just fine-which is why Capgras
sufferers agree that the " impostors " do indeed look j ust like
their loved ones-but the covert system (or systems) , which
normally provides a reassuring vote of agreement on such
occasions , is impaired and ominously silent. The absence of
that subtle contribution to i dentification is so upsetting
( " Something's missing ! " ) that it amounts to a pocket veto on
the positive vote of the surviving system: the emergent result
is the sufferer's heartfelt conviction that he or she is looking
at an impostor. Instead of blaming the mismatch on a faulty
perceptual system, the agent blames the world , in a way that
is so metaphysically extravagant , so improbabl e, that there
can be little doubt of the power (the p olitical p ower, in
effect) that the impaired system normally has in us all. When
this particular system 's epistemic hunger goes unsatisfied, it
throws such a fit that it overthrows the c ontributions of the
other systems.
In between the oblivious crab and the bizarrely mistaken
Capgras sufferer there are intermediate cases. Can't a dog
recognize , or fail to recognize , its master? According to
Homer, when Ulysses returns to Ithaca after his twenty-year
odyssey, disguised in rags as a beggar, his old dog, Argos ,
recognizes him , wags his tai l , drops his ears , and then dies.
(And Ulysses , it should be remembered , secretly wipes a tear
from his own eye . ) Just as there are reasons for a crab to (try
to) keep track of the i dentity of its own nest, there are rea
sons for a dog to (try to) keep track of its master, among
many other important things in its worl d. The more pressing

the reasons for reidentifying things , the more it pays not to
make mistake s , and hence the more investments in percep
tual and cognitive machinery will pay for themselves.
Advanced kinds of learning depen d , in fact , on prior capaci
ties for (re-) identification. To take a simple case, suppose a
dog sees Ulysses sober on Monday, We dnes day, and Friday,
but sees Ulysses drunk on Saturday. There are several con
clusions that are logically available to be drawn from this set
of experiences: that there are drunk men and sober men , that
one man can be drunk on one day and sober on another, that
Ulysses is such a man. The dog could not-logical ly, could
not-learn the second or third fact from this sequence of
separate exp eriences unless it had some (fallible , but relied
upon) way of reidentifying the man as the same man from
experience to experience . (Millikan , forthcoming) (We can
see the same principle in a more dramatic application in the
curious fact that you can 't-as a matter of logic-learn what
you look like by looking in a mirror unless you have some
oth er way of i dentifying the face you see as yours. Without
such an independent identification, you could no more dis
cover your app earance by looking in a mirror than you could
by looking at a photograph that happened to be of you . )
Dogs live in a behavioral world much richer and more
complex than the world of the crab , with more opportunities
for subterfuge , bluff, and disguise, and hence with more ben
efits to derive from the rej ection of misleading clues. But
again, a dog 's systems need not be foolproof. If the dog
makes a mistake of identification (of either sort ) , we can
characterize it as a case of mistaken identity without yet
having to conclude that the dog is capable of thinking the
proposition which it behaves as if it believes. Argos 's behav
ior in the story is touching, but we mustn 't let sentimentality
cloud our theories . Argos might also love the smells of
autumn , and respond with j oy each year when the first whiff
H OW I N T E N T I O NALITY CAM E I NTO F O C U S 1 15
of ripe fruit met his nostrils , but this would not show that he
had any way of distinguishing between recurring season
types, such as autumn, and returning individual s , such as
Ulysses. Is Ulysses , to Argos , just an organized c ollection of
pleasant smells and sounds , sights and feelings-a sort of
irregularly recurring season (we haven 't had one for twenty
years ! ) , during which particular behaviors are favored? It is a
season that is usually sober, but some instances of it have
been known to be drunk. We can see, from our peculiar
human perspective , that Argos 's success in this world will
often depend on how closely his behavior approximates the
behavior of an agent who , like us adult human beings ,
clearly distinguishes between individual s . S o we find that
when we interpret his behavior from the intentional stan ce,
w e do w el l to attribute beliefs to Argos that distinguish
Ulysses from other peop l e , strong rival dogs from weaker
rival dogs , lambs from other animals , Ithaca from other
places , and so forth . But we must be prepared to discover
that this apparent understanding of his has shocking gaps in
it-gaps inconceivable in a human being with our concep
tual scheme , and hence utterly inexpressible in the terms of
a human language.
Tales of intelligence in p ets have been commonplace for
millennia. The ancient Stoic philosopher Chrysippus
reported a dog that could perform the following feat of rea
son: coming to a three-way fork, he sniffed down p aths A
and B , and witho u t sniffing C , ran down C , having reasoned
that if there is no scent down A and B, the quarry must have
gone down C. People are less fond o f telling tales of j aw
dropping stupidity in their p ets , and often resist the implica
tions of the gaps they discover in their p ets ' competences.
Such a smart doggie , but can he figure out how to unwind
his leash when he runs around a tree or a lamppost? This is
not, it would seem, an unfair intelligence test for a dog
compared , say, with a test for sensitivity to irony in poetry,
or appreciation of the transitivity of warmer-than (if A is

warmer than B, and B is warmer than C , then A is [warmer
than? colder than?] C) . But few if any dogs can pass it. And
dolphin s , for all their intelligence, are strangely unable to
figure out that they could easily leap over the surrounding
tuna net to safety. Leaping out of the water is hardly an
unnatural act for them , which makes their obtuseness all the
more arresting. As researchers regularly discover, the more
i ngeniously you investigate the competence of nonhuman
animal s , the more likely you are to discover abrupt gaps in
competence . The ability of animals to generalize from their
p articular exploitations of wisdom is severely limited. (For
an eye-opening account of this pattern in the investigation of
the minds of vervet monkeys , see Cheney and Seyfarth , How
Monkeys See the World, 1 99 0 . )
We human beings , thanks t o the p erspective w e gain from
our ability to reflect in our special ways , can discern failures
of tracking that would be quite beyond the ken of other
beings . Suppose Tom has been carrying a lucky penny
around for years . Tom has no name for his penny, but we
shall call it Amy. Tom took Amy to Spain with him, keeps
Amy on his bedside table when he sleep s , and so forth. Then
one day, on a trip to New York City, Tom impulsively throws
Amy into a fountai n, where she blends in with the crowd of
other pennies , utterly indistinguishable, by Tom and by u s,
from all the others-at least , all t h e others that have the
same date of issue as Amy stampe d on them. Sti l l , Tom can
reflect on this development. He can recognize the truth of
the proposition that one, and only one , of those pennies is
the lucky p enny that he had always carried with him . He
can be bothered (or j ust amused) by the fact that he has irre
mediably lost track of something he has been tracking , by
one method or another, for years . Suppose he picks up an
Amy-candidate from the fountain. He can appreciate the fact
H OW I NTENTIO NALITY CAM E I NTO FOCUS 1 17
that one , and exactly one , of the fol lowing two propositions
is true:
1. The penny now in my hand is the p enny I brought with

me to New York.
2. The penny now in my hand is not the p enny I brought
with me to New York.
It doesn't take a rocket scientist to appreciate that one or

the other of these has to be tru e , even if neither Tom nor any
body else in the history of the world , past and future , can
determine which. This cap acity we have to frame , and even
under most circumstances test , hypotheses about i dentity is
quite foreign to all other creatures . The practices and proj
ects of many creatures require them to track and reidentify
individuals-their mothers , their mates, their prey, their
superiors and subordinates in their band-but no evidence
suggests they must appreciate that this is what they are
doing when they do it. Their intentionality never rises to the
p itch of metaphysical particularity that ours can rise to.
How do we do it? It doesn't take a rocket scientist to
think such thoughts, but it does take a Gregorian creature
who has language among its mind tools. But in order to use
language , we have to be specially equipped with the talents
that permit us to extract these mind tools from the (social)
environment in which they reside.
C HAPTER S
T H E C R E AT I O N O F T H I N K I N G
U N T H I N K I N G N AT U RA L P S Y C H O L O G I S T S
Language was invented so that people could

conceal their thoughts from each other.
Charles-Maurice de Tal leyrand
Many animals hide but don't think they are hiding. Many
animals flock but don't think they are flocking. Many animals
pursue , but don't think they are pursuing. They are all the
beneficiaries of nervous systems that take care of the controls
of these clever and appropriate behaviors without burdening
the host's head with thoughts , or anything arguably like
thoughts-the thoughts we thinkers think. Catching and eat
ing, hiding and fleeing , flocking and scattering all seem to be
within the competence of unthinking mechanisms . But are
there clever behaviors that must be accompanied by, pre
ceded and controlled by, clever thoughts?
If the strategy of adopting the intentional stance is as
great a boon as I have claimed , then an obvious p lace to look
for a breakthrough in animal minds is in those intentional
1 19
systems who themselves are capable of adopting the inten

tional stance toward others (and toward themselves) . We
should look for behaviors that are sensitive to differences in
the (hypothesized) thoughts of other animals . An old j oke
about behaviorists is that they don't bel ieve in beliefs , they
think that nothing can think, and in their opinion nobody
has opinions. Which animals are stuck as behaviorists,
unable even to entertain hypotheses about the minds of oth
ers? Which animals are force d , or enabled, to graduate to a
higher level? There seems to be something paradoxical about
a thoughtless agent concerning itself with the discovery and
manipulation of the thoughts of other agents, so perhaps
here we can fin d a level of sophistication that forces think
ing to evolve.
Might thinking pull itself into existence by its own boot
straps? (If you 're going to think about my thinking , I'm going
to have to start thinking about your thinking to stay even
an arms race of reflection. ) Many theorists have thought that
some version of this arms race explains the evolution of
higher intelligence. In an influential paper (" Nature 's Psy
chologists , " 1 9 78 ) , the psychologist Nicholas Humphrey
argued that the development of self-consciousness was a
stratagem for develop ing and testing hypotheses about what
was going through the minds of oth ers. It might seem that an
ability to make one's behavior sensitive to , and manipulative
of, the thinking of another agent would automatically carry
with it an ability to make one's behavior sensitive to one's
own thinking. This might be either because , as Humphrey
suggeste d , one uses one's self-consciousness as a source of
hypotheses about other-consciousness , or because when one
gets into the habit of adopting the intentional stance toward
others , one notices that one can usefully subject oneself to
the same treatment. Or for some combination of these reasons,
the habit of adopting the intentional stance could spread to
cover both other-interpretation and self-interpretation.
T H E C R EAT I O N O F T H I N K I N G 121
In an essay entitled " Conditions of Personhood" ( 1 9 7 6 ) , I

argued that an important step toward becoming a p erson was
the step up from a first-order intentional system to a secon d
order intentional system . A first-order intentional system
has beliefs and desires about many things, but not about
beliefs and desires. A second-order intentional system has
beliefs and desires about beliefs and desire s , its own or those
of others. A third-order intentional system would be capable
of such feats as wan ting you to believe that it wan ted some
thing , while a fourth-order intentional system might believe
you wan ted it to believe that you believed something , and so
forth . The big step , I argued, was the step from first-order to
second-order; the higher orders were j ust a matter o f how
much an agent can keep in its head at one time , and this
varies with the circumstances , even within a single agent.
Sometimes higher orders are so easy as to be involuntary.
Why is the fellow in the movie trying so hard to avoid smil
ing? In the context it's deliciously obvious : his effort shows
us he knows she doesn't realize he already kn ows she wan ts
him to ask her to the dance , and he wan ts to keep it that
way ! Other times , simpler iterations can stump us. Are you
sure that I want you to believe that I want you to believe
what I'm saying here?
But if higher-order intentionality is, as I and others have
argued, an important advance in kinds of minds , it is not as
clearly the watershed we are looking for between thinking and
unthinking cleverness. Some of the best studied examples of
(apparent) higher-order intentionality among nonhuman crea
tures still seem to fall on the side of unreflective adroitness.
Consider " distraction display, " the well-known behavior of
low-nesting birds , who , when a predator approaches the nest,
move surreptitiously away from their vulnerable eggs or
nestlings and begin in the most ostentatious way to feign a
broken wing, fluttering and collapsing and calling out most
piteously. This typically leads the predator far away from the
I ll KINDS OF M I N DS
nest on a wild goose chase, in which it never quite catches the

" easy" dinner it is offered. The free-floating rationale of this
behavior is clear, and, following Richard Dawkins 's useful
practice in his 1 9 76 book, The Selfish Gene, we can put it in
the form of an imaginary soliloquy:
I'm a low-nesting bird , whose chicks are not protectable

against a predator who discovers them. This approaching
predator can be expected soon to discover them unless I
distract it; it could be distracted by its desire to catch and
eat me , but only if it th ough t there was a reasonable
chance of its actually catching me (it's no dummy) ; it
would contract just that belief if I gave it evidence that I
couldn't fly anymore ; I could do that by feigning a broken
wing, etc . (From Dennett , 1 98 3 )
I n the case of Brutus stabbing Caesar, discussed i n chap

ter 2 , it was within the bounds of plausibility to suppose
that Brutus actually went through something like the solilo
quy process outlined for him-though normally, in even the
most loquacious self-addresser, much of it would go without
saying. It defies credence, however, to suppose that any bird
goes through anything like the soliloquy here . Yet that solil
oquy undoubtedly expresses the rationale that has shaped
the behavior, whether or not the bird can appreciate the
rational e . Research by the ethologist Carolyn Ristau ( 1 9 9 1 )
has shown that in a t least one such species-the pip ing
plover-individuals govern their distraction displays with
quite sophisticated controls. For instanc e , they monitor the
direction of the predator's gaze , turning up the volume of
their display if the predator seems to be losing interest, and
in other ways adapt their behavior to features detected in the
predator's . Plovers also discriminate on the basis of an inter
loper's shape and size : since cows aren't carnivorous , a cow
T H E C R EAT I O N O F T H I N K I N G 1 23
is not apt to be attracted by t h e prospect of a n easy bird

meal , so some plovers treat cows differently, squawking and
pecking and trying to drive the beast away instead of luring
it away.
Hares apparently can size up an approaching predator,
such as a fox, and make an estimate of its dangerousness
(Hasson, 1 99 1 , Holley, 1 994 ) . If the hare determines that a
particular fox has somehow managed to get within striking
distance , it will either crouch and freeze-counting on escap
ing the notice of the fox altogether-or crouch and scurry as
swiftly and quietly as it can , ducking behind whatever cover
is available. But if the hare determines that this fox is unlikely
to succeed in its chase, it does a strange and wonderful
thing. It stands up on its hind legs , most conspicuously, and
stares the fox down ! Why? Because it is announcing to the
fox that the fox ought to give up. " I 've already seen you, and
I'm not afraid. Don't waste your precious time and even
more precious energy chasing me. Give it up ! " And the fox
typically draws just this conclusion, turning elsewhere for
its supper and leaving the hare , which has thus conserved its
own energy, to continue its own feeding.
Once again, the rationale of this behavior is almost cer
tainly free-floating . It is probably not a tactic the hare has fig
ured out for itself, or been capable of reflecting on. Gazelles
being chased by lions or hyenas often do something similar,
called stotting. They make ridiculously high leap s , obviously
of no benefit to their flight but designed to advertise their
superior speed to the pre dators . " Don't bother chasing me.
Chase my cousin. I'm so fast I can waste time and energy
doing these silly leaps and still outrun you . " And it appar
ently works ; predators typically turn their attention to other
animals .
Other varieties of predator and prey behavior could be
cited , all with elaborate rationales but little or no evidence
that the animals actually represent these rationales to them

selves in any fashion. If th ese creatures are to be considered
"natural psychologists " (to use Humphrey's term) , they are
apparently unthinking natural psychologists. These creatures
don't represent the minds of those they interact with-that is,
they don't need to consult any i nternal " model " of the mind
of another in order to anticipate the other's behavior and
hence govern their own behavior. They are well-supplied
with a largish " list" of alternative behaviors , nicely linked to
a largish list of perceptual cue s, and they don't need to know
any more. Does this count as mind reading? Are piping
p lovers , or hares, or gazelles higher-order intentional sys
tems or not? That question begins to appear less important
than the question of how such an apparent mind-reading
competence might be organized. When, then, does the need
arise to go beyond these large lists? The ethologist Andrew
Whiten has suggested that the need arises simply when the
lists get too long and unwieldy to be supplemented. Such a
list of pairs amounts , in logicians ' terms, to a conjunction of
conditional s , or if-then pairs :
[If you see x, do A] , and [if you see y, do Bl , and [ i f you

see z, do C] , . . .
Depending on just how many independent conditionals

there are , it may become economical to consolidate them
into more organized representations of the worl d. Perhaps in
some species-which species remains an open question
the brilliant innovation of exp licit generalization enters the
picture , permitting the lists to be broken down and rebuilt
on demand from first principles , as new cases arise. Con
sider Whiten 's diagram of the complexity that would get
organized around an internal representation by one animal
of a specific desire in another animal .
Y PERCEIVES Y PREDICTS
(and --. TAKES ACTION)
X constantly watches X wi ll take the meat

the meat if someone leaves it
----- GUARD
X grooms those who

have meat X will be unwilling
to give up meat once
she is holding it
X tries to grab the ---1 .- APPROACH
..
meat WITH CARE
X WANTS
X threatens lower THE MEAT X 's grooming may
ranked animals who j ust be means to end ,
come near the meat terminated if chance
to grab meat
__,.. BE SKEPTICAL
X picks up any meat
scra s she can find
Giving X meat may
increase friendliness
X follows those with of X
meat ----1..� GIVE MEAT
FIGURE 5. 1
As before , we can see the rationale behind such consoli da

tion , but this rationale need not be entertained in any fash
ion by the minds of the consolidators . If they are lucky
enough to hit upon this design improvement, they could
simply be the beneficiaries of it without appreciating why or
how it worked . But is this design really the improvement it
appears to be? What are its costs and benefits? And its value
aside , how could it have come into existence? Did it just
arise one day, in random and desperate reaction to a growing
problem of " overhead "-too many conditional rules to keep
in service simultaneously? Perhaps , but nobody yet knows
any plausible upper bound on the number of concurrent
semi-independent control structures that can coexist in a
1 26 KINDS OF MINDS
nervous system. (In a real agent with a real nervous system ,

there may not be any. Maybe a few hundred thousand such
p erceptuo-behavioral control circuits can mingle together
efficiently in a brain-how many might be called for?)
Might there not be some other sort of selective pressure
that could have led to the reorganization of control struc
ture s , yielding a capacity for generalization as a bonus? The
ethologist David McFarland ( 1 989) has argued that the
opportunity for communication provi des just such a design
pressure , and moreover, Talleyrand 's cynical suggestion at
the opening of this chapter is close to an important truth.
When communication arises in a species , he claims , pure
honesty is clearly not the best policy, since it will be all too
exploitable by one's competitors (Dawkins and Krebs, 1 9 7 8 ) .
T h e competitive context is clear in a l l cases of communica
tion between predator and prey, such as the minimal com
munication practices exhibited by the stotting gazelle and
the hare staring down the fox; and here it is obvious how the
opp ortunity for bluffing arises. In the arms race of producing
future , you have a tremendous advantage if you can produce
more and better future about the other than the other can
produce about you , so it always behooves an agent to keep
its own control system inscrutable. Unpredictability is in
general a fine protective feature , which should never be
squandered but always spent wisely. There is much to be
gained from communication if it is craftily doled out
enough truth to keep one's credibility high but enough false
hood to keep one's options open. (This is the first point of
wisdom in the game of poker: he who never bluffs never
wins ; he who always bluffs always loses . ) It takes some
stretching of the imagination to see the hare and fox as coop
erating on their j oint problems of resource management , but
in fact they are both better off for their occasional truces.
The prospects for expanding cooperation and hence mul
tiplying its benefits is much more clearly visible in the con-
text of communication with members o f one 's o w n species.

Here food sharing, and sharing the costs and risks of child
care and defense of the group , and so forth, provi de plenty of
opportunities for cooperati on, but o nl y if t h e rather stringent
conditions for exp loiting these opportunities can be met.
Cooperation between parents, or between parents and off
spring, cannot be taken as a given in nature ; the omnipresent
possibility of competition still lies behind any mutually use
ful conventions that emerge , and this context of competition
must be taken into account.
According to McFarland, the need for an exp licit , manip
ulable representation of one's behavior arises only when the
option of potentially cooperative but still self-protective
communication emerges , for then a new form of behavior
must come under the agent 's contro l : the behavior of explic
itly communicating something about one's other behavior.
("I'm trying to catch fish , " or " I ' m looking for my mother, " or
'Tm just resting. " ) Confronted with the task of shaping and
executing such a communicative act , the agent 's problem is a
version of the very problem confronting us as observing the
orists : How should the agent 's own tangle o f competing ,
enhancing , merging , intertwining behavioral control circuits
be carved up into competing " alternatives " ? Communication
favors clear-cut answers . As the saying goes , " Are you going
to fish or cut bait? " So the demands of communication, by
forcing an agent into declaring a category, may often create a
distortion-rather like the distortion you recognize when
required to check off just one item in a p oorly designed mul
tiple-choice test : if " none of the above " is not an available
option, you are forced to settle for whatever you take to be
the least objectionable near miss. McFarland suggests that
this task of carving where nature has provided no salient
j oints is a problem the agent solves by what we might call
approximating confabulation . The agent comes to label its
tendencies as if they were governed by explicitly represented
I 28 KINDS 0F MINDS
goals-blueprints for actions-instead of trends of action

that emerge from the interplay of the various candidates.
Once such represen tations of in ten tions (in the everyday
sense of intentions) come into existence in this backhanded
way, they may succeed in convincing the agent itself that it
has these clear-cut prior intentions governing its actions . In
order to solve its communication problem, the agent has
made a special user-interface for itself, a menu of explicit
options from which to choose, and then has been to some
degree taken in by its own creation.
Opportunities to put such communications to good use
are strictly limited , however. Many environments are inhos
p itable to secret keeping , quite independently of any pro
clivities or talents of the agents in that environment; and if
you can't keep a secret there is little role for communication
to p lay. According to ancient folk wisdom, people who live
in glass houses shoul dn't throw stones , but animals who live
in the natural equivalent of glass houses have no stones to
throw. Animals who live close together in groups in open
territory are seldom if ever out of sight and hearing (and
smell and touch) o f their conspecifics for very long, and thus
have no opportunities to satisfy the conditions under which
secrets can flourish. Suppose that p is an ecologically valu
able fact, and suppose that you know that p and nobody else
does-yet. If you and the other potentially competitive
agents in the vicinity all have access to pretty much the
same information about the environment , then it is next to
impossible for circumstances to arise in which you can turn
such a temporary information-gradient to your advantage.
You may b e the first wildebeest to see or smell the lion to the
northwest, but you can't really hoard (or sell) this informa
tion, because those standing shoulder to shoulder with you
will soon have it themselves . Since there is scant possibility
that such a temporary information advantage can be con
trolled, a devious wildebeest (for example) would have pre-
T H E C R E AT I O N O F T H I N K I N G 1 29
cious little opp ortunity to benefit from its talent. Just what
could it do to gain sneaky advantage over the others?
The intentional stance shows us readily that the appar
ently simple behavior of keeping a secret-a null behavior,
from most vantage points-is in fact a behavior whose suc
cess depends on satisfying a rather demanding set of condi
tions. Suppose that Bill is keeping some secret, p, from Jim.
The following conditions must be met:
1. Bill knows (believes) that p.

2 . Bill believes that Jim does not believe that p.
3 . Bill wants Jim not to come to believe that p.
4 . Bill believes that Bill can make it the case that Jim not
come to believe that p.
It is this last condition that restricts advanced secret-keeping

(for instance , about features of the external environment) to
quite specific behavioral environments . This was clearly
brought out by experiments in the 1 9 70s by the primatolo
gist Emil Menzel ( 1 9 7 1 , 1 9 74) , in which individual chim
panzees were shown the location of hidden foo d , and
thereby given the opp ortunity to deceive the other chim
panzees about its location . They often rose to the opp ortu
nity, with fascinating results , but this behavior always
depended on the experimenters ' producing a state of affairs
in the laboratory (a cage adj acent to a larger fenced enclo
sure , in this case) that would only rarely occur in the wild:
the chimpanzee who sees the hidden foo d must be in a posi
tion to know that the oth er chimpanzees do not see him see
ing the food. This was achieved by keeping all the other
chimpanzees locked in a common cage while the chosen
chimpanzee was taken alone into the larger enclosure and
shown the hidden food. The chosen chimpanzee could come
to learn that it alone was learning that p--t hat its informative
adventures in the enclosure were not visible to the others in
the cage . And , of course, there had to be something the chimp

with the secret could do to protect its secret-at least , for a
while-once the others were released.
Chimpanzees in the wild do frequently wander far
enough away from their groups for long enough to acquire
secrets within their contro l , so they are a good species to
examine with such tests . In animals whose evolutionary his
tory has not unfolded in environments in which such oppor
tunities naturally and frequently arise, there is little likeli
hood that the capacity to exploit such opp ortunities has
evolved . Discovering (in the lab) a heretofore unused talent
is not impossibl e , of course , since unused talent must sur
face , rarely, in the real worl d , whenever innovation occurs .
Such a talent will typ ically be a by-product of other talents
developed under other selection pressures. In general , how
ever, since we expect cognitive complexity to coevolve with
environmental complexity, we should look for cognitive
complexity first in those species that have a long history of
dealing with the relevant sort of environmental complexity.
Taken together, these p oints suggest that thinking-our
kind of thinking-had to wait for talking to emerge, which in
turn had to wait for secret keeping to emerge , which in turn
had to wait for the right complexification of the behavioral
environment. We should be surprised to find thinking in any
species that hasn't made it to the bottom of this cascade of
sieves . As long as the behavioral options are relatively sim
ple-witness the piping plover's predicament-no fancy
central representation needs to occur, so in all likelihood it
doesn't. The sort of higher-order sensitivity required to meet
the needs of a piping plover or a hare or a gazelle can proba
bly be provi ded by networks designed almost entirely by
Darwinian mechanisms , abetted here and there by Skinner
ian mechanisms . ABC learning, then, could probably suffice
to produce such a sensitivity-though this is an empirical
issue that is nowhere near settled. It will be interesting to
TH E C R EAT I O N O F T H I N K I N G 131
discover if there are cases in which we have clear evidence

of differential treatment of specific in dividuals (a piping
p lover that doesn' t waste its ruses on a p articular reidenti
fied dog, say, or a hare that , after a p articular close c al l , dras
tically increases its stare- down distance for a specific fox. )
Even in these cases , w e may be abl e t o account for the learn
ing via relatively simple models : these animals are Popper
ian creatures-creatures who can be guided by past experi
ence to rej ect tempting but untested candidates for
action-but still not explicit thinkers.
As long as the natural p sychologists don't have an oppor
tunity or an obligation to communicate with each other
about their attributions of intentionality to themselves or
others , as long as they never have an opp ortunity to compare
notes , to dispute with others , to ask for the reasons that
ground the conclusions they are curious about, it seems that
there is no selective pressure on them to represent those rea
sons , and hence no selective pressure on them to forsake the
Need to Know principle in favor of its familiar opposite, the
Commando Team Principle: give each agent as much knowl
e dge about the total project as possibl e , so that the team has
a chance of ad-libbing appropriately when unanticipated
obstacles arise. (Many films , such as Th e Guns of Navarone,
or The Dirty Dozen , make this principle visible by present
ing the exploits of such versati l e an d knowing teams ; h ence
my name for it.)
The free-floating rationales that explain the rudimentary
higher-order intentionality of birds and hares-and even
chimpanzees-are honored in the designs of their nervous
systems , but we are looking for something more ; we are look
ing for rationales that are represen ted in those nervous sys
tems.
Although ABC learning can yield remarkably subtle and
p owerful discriminatory competences , capable o f teasing
out the patterns lurking in voluminous arrays of data, these
competences tend to be anchored in the specific tissues that

are modified by training. They are " embedded" competences,
in the sense that they are incapable of being " transported "
readily t o be brought t o bear o n other problems faced by the
individual , or shared with other individuals. The philosopher
Andy Clark and the psychologist Annette Karmiloff-Smith
( 1 99 3 ) have recently been exploring the transition from a
brain that has only such embedded knowledge to a brain that ,
as they say, " enriches itself from within by re-representing
the knowledge that it has already represented. " Clark and
Karmiloff-Smith note that while there are clear benefits to a
design policy that " intricately interweave [s] the various
aspects of our knowledge about a domain in a single knowl
e dge structure , " there are costs as well : " The interweaving
makes it practically impossible to operate on or otherwise
exp loit the various dimensions of our knowledge indepen
dently of one another. " S o opaquely is such knowledge hid
den in the mesh of the connections that " it is knowledge in
the system , but it i s not yet knowledge to the system "-like
the wisdom revealed in the precocious single-mindedness
with which the newly hatched cuckoo shoulders the com
p eting eggs out of the nest. What would have to be added to
the cuckoo 's computational architecture for it to be able to
appreciate , understand , and expl oit the wisdom interwoven
in its neural nets?
A popular answer to this question, in its many guises, is
" symbols ! " The answer is well-nigh tautological , and hence
is bound to be right in som e interpretation. How could it not
be the case that implicit or tacit knowledge becomes explicit
by being expressed or rendered in some medium of
" explicit" representation? Symbol s , unlike the nodes woven
into connectionist networks , are movable; they can be
manipulated ; they can be composed into larger structures , in
which their contribution to the meaning of the whole can be
a definite and generatable function of the structure-the syn-
tactic structure-of the parts. There is surely something right

about this , but we must proceed cautiously, since many pio
neers have posed these questions in ways that have turned
out to be misleading.
We human beings have the capacity for swift , insightful
learning-learning that does not depend on laborious train
ing but is ours as soon as we contemplate a suitable sym
bolic representation of the knowledge . When psychologists
devise a new experimental setup or paradigm in which to
test such nonhuman subjects as rats or cats or monkeys or
dolphins , they often have to devote dozens or even hun
dreds of hours to training each subj ect on the new tasks .
Human subj ects , however, can usually j ust be told what is
desired of them. After a brief question-and-answer session
and a few minutes of practice, we human subj ects will typi
cally be as competent in the new environment as any agent
ever could be. Of cours e , we do have to un derstan d the rep
resentations that are presented to us in these tests , and that 's
where the transition from ABC learning to our kind of learn
ing is still shrouded in fog. An insight that may help clear it
is a familiar maxim of artifact making : if you " do it your
self, " you understand it. To anchor a free-floating rationale
to an agent in the strong way, so that it is th e agen t 's own
reason, the agent must " make " something. A representation
of the reason must be composed, designed , edite d , revise d ,
manipulated, endorsed. H o w d o e s any agent come to b e
able t o d o such a wonderful thing? D o e s it have to grow a
new organ in its brain? Or can it build this competence out
o f the sorts of external-world manipulations it has already
mastered?
M A K I N G T H I N G S TO T H I N K W I T H
Just as you cannot d o very much carpentry with

your b are hands , there is not much thinking you
can do with your bare brain.
B o Dahlbom and Lars-Erik Janlert ,
Comp u ter Fu ture (forthcoming)
Every agent faces the task of making the best use of its envi
ronment. The environment contains a variety of goods and
toxin s , mixed in with a confusing host of more indirect
clues : harbingers and distractors , stepping-stones and pit
falls . These resources often amount to an embarrassment of
riches in competition for the agent's attention; the agent's
task of resource management (and refinement) is thus one in
which time is a crucial dimension. Time spent in a futile
pursuit of prey, or bracing oneself to withstand illusory
threats , is time wasted , and time is precious.
As suggested in figure 4.4, Gregorian creatures take in
from the environment various designed entities and use
them to improve the efficiency and accuracy of their hypoth
esis testing and decision making , but the diagram is mislead
ing as it stands. How much room is there in the brain for
these artifacts , and how do they get installed? Is the brain of
a Gregorian creature so much more capacious than the
brains of other creatures? Our brains are modestly larger
than the brains of our nearest relatives (although not larger
than the brains of some dolphins and whales) , but this is
almost certainly not the source of our greater intelligence.
The primary sourc e , I want to suggest, is our habit of off
loading as much as possible of our cognitive tasks into the
environment itself-extruding our minds (that i s, our mental
projects and activities) into the surrounding world, where a
host of peripheral devices we construct can store , process,
and re-represent our meanings , streamlining, enhancing, and

protecting the processes of transformation that are our think
ing. This widespread practice of off-loading releases us from
the limitations of our animal brains.
An agent faces its environment with its current repertoire
of skills, perceptual and behavioral. If the environment is
too complicated for these skills to cope , the agent is in trou
ble unless it can develop new skills, or simplify its environ
ment. Or both. Most species rely on natural landmarks to
find their way around , and some species have developed the
trick of adding landmarks to the world for their subsequent
use. Ants, for instance , lay down pheromone trails-odor
trails-leading from nest to food and back, and the individu
als in many territorial species mark the boundaries of their
territories with idiosyncratic aromatic compounds in their
urine . Posting your land in this way warns o ff trespassers ,
but it also provides a handy device you can use yourself. It
saves you from needing some other way to remember the
boundary of that part of the environment in which you have
invested significant efforts of resource refinement-or even
cultivation. As you approach the boundary, you can smell it.
You let the outside world store some easily transduced infor
mation about where the important jo ints in nature are , so
that you can save your limited brain for other things . This is
good stewardship . Putting deliberate marks on the environ
ment to use in distinguishing what are for you its most
important features is an excellent way of reducing the cogni
tive load on your perception and memory. It's a variation o n ,
a n d enhancement of, evolution's g o o d tactic of installing
beacons where most needed.
For us human beings , the benefits of labeling things in
our environments are so obvious that we tend to overlook
the rationale of labeling, and the conditions under which it
works . Why does anyone ever label anything , and what does
it take to label something? Suppose you were searching
through thousands of boxes of shoes , looking for a house key

that you thought was hidden in one of them. Unless you 're
an idiot, or so frantic in your quest that you cannot pause to
consider the wisest course , you will devise some handy
scheme for getting the environment to assist you with your
problem. You want in particular to avoid wasting time by
looking more than once in each box. One way would be to
move the boxes one at a time from one stack (the unexam
ined stack) to another stack (the examined stack) . Another
way, p otentially more energy efficient , is to put a check
mark on each box as you examine it, and then adopt the rule
o f never looking into a box with a check mark on it. A check
mark makes the world simpler, by giving you a simple per
ceptual task in place of a more difficult-perhaps impossi
ble-memory and recognition task. Notice that if the boxes
are all lined up in a row, and you don't have to worry about
unnoticed reorderings of the queue , you don't need to put
check marks on them ; you can just work your way from left
to right , using the simple distinguisher that nature has
already provided you with-the left/right distinction.
Now let's concentrate on the check mark itself. Will any
thing do as a check mark? Clearly not. " I 'll put a faint
smudge somewhere on each box as I examine it. " "I'll bump
the corner of each box as I examine it. " Not good choices ,
since the likelihood is too high that something else may
already have inadvertently put such a mark on a box. You
need something distinctive, something that you can be con
fi dent is the result of your labeling act and not some extrane
ously produced blemish. It should also be memorable, of
cours e , so that you won't be beset by confusions about
whether or not some salient label you encounter is a label
yo u put there , and if so, what policy you meant to follow
when you adopted it. There 's no use tying a string around
your finger as a reminder if, when it later catch es yo ur eye
(thereby fulfilling its function as a self-control beacon off-
T H E C R EATI O N O F T H I N K I N G 1 37
loaded into the environment) , you can 't remember why you
tied it. Such simple deliberate marks on the world are the
most primitive precursors of writing, a step toward the cre
ation in the external world of dedicated p eripheral informa
tion-storage systems. Notice that this innovation does not
depend on there being a systematic language in which such
labels are composed. Any nonce system will d o , as long as it
can be remembered during use.
Which species have discovered these strategies? Some
recent experiments give us a tantalizing , if inconclusive,
glimpse into the possibilities. B irds that hide caches of seeds
at many specific locations are astonishingly successful at
retrieving their secret stores after long interval s . Clark's nut
crackers , for instance, have been experimentally studied by
the biologist Russell Balda and his colleagues in an enclosed
laboratory setting-a large room with either a dirt floor or a
floor provided with many holes filled with san d , and further
furnished with various landmarks . The birds may make more
than a dozen caches with seeds provided to them, and then
return , days later, to recover them. They are remarkably good
at relying on multiple cues , finding most of their caches even
when the experimenters move or remove some of the land
marks . But they do make mistakes in the laboratory, and most
of these mistakes seem to be errors of self-contro l : they waste
time and energy by revisiting sites they have already cleaned
out on earlier expeditions . Since these birds may make sev
eral thousand caches in the wil d , and visit them over a
period of more than six months , the frequency of such wasted
revisits in the wild is almost impossible to record , but it
stands to reason that revisiting would be a costly habit to fall
into , and other species of caching birds , such as chickadees ,
are known to be able to avoid such revisits.
In the wil d, Clark's nutcrackers are observed to eat the
seeds where they dig them up , leaving behind a mess of pic
nic litter that could remind them , on another fly-by, that
they had already opened that particular shoebox. Balda and

his colleagues designed experiments to test the hypothesis
that the birds relied on such marks to avoid revisits . In one
condition, the birds ' disturbances of the visited sites were
carefully erased between sessions , and in another the tell
tale disturbances were left . In this laboratory setting, how
ever, the birds did not do significantly better when the dis
turbances were left , so it has not been shown that the birds
do rely on these cues. Perhaps they couldn't in the wild,
since such cues are often soon obliterated by weather in any
case , as Balda note s. He also points out that the experiments
to date are inconclusive ; the cost of error in the laboratory
setting is slight-a few seconds wasted in the life of a well
fed bird .
It is also possible that putting the birds in a laboratory
setting inadvertently renders them relatively incompetent,
since their everyday habits of distributing part of the task of
self-control to the environment may depend on further cues
that are inadvertently absent in the laboratory. It is com
monly observed-but not commonly enough !-that old folks
removed from their homes to hospital settings are put at a
tremendous disadvantage, even though their basic bodily
needs are well provided for. They often appear to be quite
demented-to be utterly incapable of feeding, clothing , and
washing themselves , let alone engaging in any activities of
greater interest. Often, however, if they are returned to their
homes , they can manage quite well for themselves. How do
they do this? Over the years , they have loaded their home
environments with ultrafamiliar landmarks , triggers for
habits , reminders of what to d o , where to find the foo d , how
to get dressed, where the telephone is, and so forth. An old
p erson can be a veritable virtuoso of self-help in such a
hugely overlearned world , in spite of his or her brain's
increasing imperviousness to new bouts of learning-of the
ABC variety or any other. Taking them out of their homes is
literally separating them from large p arts of their minds

potentially just as devastating a development as undergoing
brain surgery.
Perhaps some birds unthinkingly make check marks as a
by-product of their other activities . We human beings cer
tainly rely on many check marks unwittingly placed in our
surroundings . We pick up helpful habits that we vaguely
appreciate without ever stopp ing to understand why they 're
such treasures . Think of trying to do multidigit multiplica
tion problems in your head. How much is 2 1 7 times 4 3 6 ? No
one would try to answer this without the help of pencil and
pap er, except as a stunt. The tally on paper serves more than
one useful function; it provides a reliable store for the inter
mediate results , but the individual symbols also serve as
landmarks that can be followed, reminding you , as your eyes
and fingers reach each point , of what the next step in the
overlearned recipe should be. (If you doubt the second con
tribution, just try doing multi digit multiplication in which
you write down the intermediate results on separate slips of
paper placed in a nonstandard arrangement in front of you ,
instead of lining them up in the canonical way. ) We Grego
rian creatures are the beneficiaries of literally thousands of
such useful technologie s, invented by others in the dim
recesses of history or prehistory but transmitted via cultural
highways , not via the genetic pathways of inheritance. We
learn , thanks to this cultural heritage, how to spread our
minds out in the worl d , where we can put our beautifully
designed innate tracking and pattern-recognizing talents to
optimal use.
Making such a change in the world doesn't just take a
load off memory. It may also permit the agent to bring to bear
some cognitive talent that otherwise would be underutilized,
by preparing special materials for it-in the minimal case ,
unwittingly. The roboticist Philippe Gaussier ( 1 994) has
recently provided a vivid illustration of this possibility,
1 40 KINDS OF MINDS
using tiny robots that first alter their environment and then
have their own behavioral repertoire altered in turn by the
new environment they have created. These robots are real
world Braitenberg vehicles-called Kheperas (the Italian
word for scarab beetles) by their creator, the roboticist
Francesco Mondada. They are somewhat smaller than
hockey pucks , and they roll around on two tiny wheels and
a castor. The robots have extremely rudimentary visual sys
tems-just two or three photocells-connected to their
wheels in such a way that signals from them turn the robots
away from collisions with the walls that surround their
tabletop world . S o these robots are innately equipped, you
might say, with a visually guided wall-avoi dance system.
Small , movable " pegs "-little cylinders of wood-are scat
tered about on the tabletop , and the robots ' innate vision
systems cause them to duck around these lightweight obsta
cles too, but wire hooks on their backs typically snag the
p egs as the robots go by. They scurry around in random
walks on the tabletop , unwittingly picking up pegs and then
depositing them whenever they swerve sharp ly in the direc
tion of a carrie d peg. (See figure 5 . 2 ) Over time , these
encounters redistribute the pegs in the environment, and
whenever two or more p egs happen to be deposited next to
each other, they form a group that the robots subsequently
" misperceive " as a bit of wall-to be avoided. In short order,
and without further instruction from any Central Headquar
ters , the robots will line up all the pegs that have been scat
tered in their environment, organizing their environment
into a series of connected walls. The Kheperas ' random
walks in an initially random environment first structure that
environment into something like a maze , and then use that
structure to shape their own behavior; they become wall fol
lowers .
This is as simple a case as can be imagined of a tactic that
include s , at the sophisticated end of the sp ectrum, all dia-
T H E C R EAT I O N O F T H I N K I N G 141
Philippe Gaussier's robots
wall � / pegs '\i. "wall "

+
o r
·· · o 0 0
<' · ·
··
9 :.
· .
F I G U RE 5 . 2
,
9
gram drawing and model building. Why do we ever draw a
diagram-for instance, on a blackboard or (in earli er days)
on the floor of the cave with a sharp stick? We do so because
by re-representing the information in another format, we
make it presentable to one special-purpose p erceptual com
petence or another.
Popperian creatures-and their subvariety, the Gregorian
creatures-live in an environment that can be roughly
divided into two part s : the " external " and the " internal. "
The denizens o f the " internal " environment are distin
guished not so much by which side of the skin they are
found on (as B . F. Skinner has remarked ( 1 96 4 , p. 84] , " The
skin is not that important as a boundary " ) as by whether
they 're portable, and hence largely omnipresent , and hence
relatively more controllable and better known, and hence
more likely to be designed for an agent's benefit. (As we
noted in chapter 2 , the shopping list on the slip of p aper gets
its meaning in exactly the same way as a shopping list mem
orized in the brain.) The " external " environment changes in
many hard-to-track ways , and is, in the mai n , geographically
outside the creature. (The limits of geography in drawing
this distinction are nowhere more vividly illustrated than in
the case of antigens , evil invaders from the outside, and anti
bodies , loyal defenders from the inside, both of which min
gle with friendly forces-like the bacteria in your gut , with
out whose labors you would die-and irrelevant bystanders ,
in the crowds of microbe-sized agents populating your body
space . ) A Popperian creature 's portable knowledge about the
world has to include some modicum of knowledge-know
how-about the omnipresent p art of its world that is itself. It
has to know which limbs are its own , of course, and which
mouth to feed , but it also has to know its way around in its
own brai n , to some extent. And how does it do that? By
using the same old metho ds: by p lacing landmarks and
labels wherever they would come in handy ! Among the
resources to b e managed under time pressure by an agent are
the resources of its own nervous system. This self-knowl
edge need not itself be represented explicitly, any more than
the wisdom of an unthinking creature needs to be represented
explicitly. It can be mere embedded know-how, but it is cru
cial know-how about how to manipulate that curiously docile
and relatively unfleeting part of the world that is oneself.
You want these refinements of your internal resources to
simplify your life , so that you can do more things better and
do them faster-time is always precious-with your avail
able repertoire of talents . Once again, there is no use creating
an internal symbol as a tool to use in self-control if when it
" catches your mind's eye " you can't remember why you cre
ated it. The manipulability of any system of pointers , land
marks , l abels , symbols, and other reminders depends on the
underlying robustness of your native talents at tracking and
reidentification, providing you with redundant, multimodal
p aths of accessibility to your tools. The resource manage
ment techniques you are born with make no distinction
between interior and exterior things . In Gregorian creatures,
such as us , the representations of features and things in the
( external or internal) world become obj ects in their own
right-things to be manipulated , tracked , move d , hoarde d ,

lined up , studied, turned upside down , a n d otherwise
adjusted and exploited.
In her book On Ph otography ( 1 9 7 7 ) , the literary critic
Susan Sontag points out that the advent of high-speed still
photography was a revolutionary technological advance for
science because it permitted human beings , for the first time
ever, to examine complicated temporal phenomena not in
real time but in their own good tim e-in leisurely, methodi
cal , backtracking analysis of the traces they had created of
those complicated events. As noted in chapter 3, our natural
minds are equipped to deal with changes that occur only at
particular paces . Events that happen faster or slower are sim
ply invisible to us. Photography was a technological advance
that carried in its wake a huge enhancement in cognitive
power, by permitting us to re-represent the events of interest
in the world in a format , and at a rate, that was tailor-made
for our particular senses.
Before there were cameras and high-speed film, there
were plenty of observational and recording devices that per
mitte d the scientist to extract data precisely from the world
for subsequent analysis at his leisure . The exquisite dia
grams and illustrations of several centuries of science are
testimony to the power of these metho ds , but there is some
thing special about a camera : it is " stupi d . " In order to " cap
ture " the data represented in its products , it does not have to
understand its subj ect in the way a human artist or illustra
tor must. It thus passes along an unedited , uncontaminated ,
unbiased but still re-represented version of reality t o the fac
ulties that are equipped to analyze, and ultimately under
stand, the phenomena. This mindless mapping of complex
data into simpler, more natural or user-friendly formats is, as
we have seen, a hallmark of increasing intelligence.
But along with the camera , and the huge pile of still pho
tographs that poured out of it, came a resource problem: the
photos themselves needed to be labeled. It does scant good

to capture an event of interest in a still picture , if you can 't
remember which of thousands of prints lying arou nd the
office is the one that represents the event of interest. This
" matching problem " doesn 't arise for simpler, more direct
varieties of tracking , as we have seen , but the cost of solving
it should often be borne ; the trick can pay for itself (time is
money) in cases in which it permits indirect tracking of
important things that cannot be tracked directly. Think of
the brilliant practice of sticking colored pins in a map to
mark the location of each of a large number of events we are
trying to understand. An epidemic may be diagnosed by see
ing-seeing, thanks to color coding-that all the cases of one
sort line up on the map alongside one or another inconspic
uous or even heretofore undepicted feature-the water main ,
or the sewage system , or perhaps the route of the postman. A
s erial killer 's secret base of operations may sometimes be
homed in on-a variety of villaintaxis-by plotting the geo
graphic center of the cluster of his attacks . The dramatic
improvements in all our kinds of investigations , from the
foraging strategies of our hunter-gatherer days to the contem
porary investigations by our police , poetry critics , and
physicists , are due in the main to the exp losive growth in
our technologies of re-representation.
We keep " pointers " and " indices " in our brains and leave
as much of the actual data as we can in the external worl d ,
i n o u r address books , libraries , notebooks , computers-and,
indeed, in our circle of friends and associates. A human
mind is not only not limited to the brain but would be rather
severely disabled if these external tools were removed-at
least as disabled as the near-sighted are when their eye
glasses are taken away. The more data and devices you off
load , the more dep endent you become on these peripherals ;
nevertheles s , t h e more intimately familiar you become with
the peripheral objects thanks to your practice in manipulat

ing them, the more confidently you can then do without them,
sucking the problems back into your head and solving them in
an imagination dis � iplined by its external practice. (Can you
alphabetize the words in this sentence in your head?)
A particularly rich source of new techniques of re-repre
sentation is the habit that we-and only we-have devel
oped of deliberately mapping our new problems onto our
old problem-solving machinery. Consider, for instance , the
many different methods we have developed for thinking
about time by actually thinking about space (Jayne s , 1 9 7 6 ) .
We have all sorts of conventional ways of mapping past,
present , and future , before and after, sooner and later-dif
ferences that are virtually invisible in unrefined nature
onto left and right , up and down , clockwise and counter
clockwise . Monday is to the left of Tues day for most of u s ,
while ( i n a valuable convention that is fading fr o m our cul
ture , sad to say) four o ' clock is tucked under three o ' clock on
the right hand side of every day or night. Our spatialization
of time doesn't stop there . In scienc e , particularly, it extends
to graphs , which have by now become a familiar system of
diagrams for almost all literate people. (Think of the profits ,
or the temperature , or the loudness of your stere o , rising up
up up from left to right with the passage of time . ) We use our
sense of space to see the passage of time (usually from left to
right , in standard convention, except in evolutionary dia
grams, in which earlier eras are often shown at the bottom ,
with today at the top ) . As these examples show-the absence
of any figures in the text at this point is deliberate-our abil
ity to imagine these diagrams when verbally invited to do so
is itself a valuable Gregorian competenc e , with many uses.
Our ability to imagine these diagrams is p arasitic on our
ability to draw and see them, off-loading them at least tem
porarily into the external worl d.
1 46 KI N DS 0F MI N DS
Thanks to our prosthetically enhanced imaginations , we

can formulate otherwise imponderable, unnoticeable meta
physical p ossibilities , such as the case of Amy the lucky
p enny, discussed at the end of chapter 4 . We need to be able
to imagine the otherwise invisible traj ectory line linking the
genuine Amy of yesterday with just one of the look-alike
pennies in the pile-we need to draw it " in our mind 's eye. "
Without such visual aids , internal or external, we would
have great difficulty following , let alone contributing to ,
these metaphysical observations . (Does that mean that some
one born blind couldn't participate in such metaphysical
discussions? No, because the blind develop their own meth
ods of spatial imagining, concerned, just as a sighted per
son's imagining i s , with keep ing track of moving things in
space , one way or another. But an interesting question is
what differences , if any, can be found in the styles of abstract
thinking adopted by those born blind or deaf. ) Armed with
these mind tools , we tend to forget that our ways of thinking
about the worl d are not the only ways , and in particular are
not prerequisites for engaging the world successfully. It
probably seems obvious , at first , that since they are so mani
festly intelligent, dogs and dolphins and bats must have con
cepts more or less like ours , but on reflection it shouldn't
seem obvious at all. Most of the questions we 've raised from
our evolutionary p erspective about the ontology and ep iste
mology o f other creatures have not yet been answered , and
the answers will no doubt be surprising. We have taken only
the first step : we've seen some possibilities to be investi
gated that we overlooked before.
Of all the mind tools we acquire in the course of furnish
ing our brains from the stockpiles of culture , none are more
important , of cours e , than words-first sp oken , then written.
Words make us more intelligent by making cognition easier,
in the same way (many times multiplied) that beacons and
landmarks make navigation in the world easier for simple
creatures . Navigation in the abstract multidimensional world

of ideas is simply impossible without a huge stock of mov
able, memorable landmarks that can be shared , criticized ,
recorded , and looked at from different p erspectives. It's
important to remember that speaking and writing are two
entirely distinct innovations , separated by many hundreds
of thousands (and maybe millions) of years , and that each
has its own distinct set of powers . We tend to run the two
phenomena together, especially when theorizing about the
brain or mind. Most of what has been written about the pos
sibilities of a " language of thought " as a medium of cognitive
operations presupposes that we're thinking of a written lan
guage of thought-"brain writing and mind reading, " as I put
it some years ago . We can get a better perspective on how the
advent of language might magnify our cognitive p owers if
we concentrate instead on why and how a spoken language
of thought-an offspring of our natural, public language
might do some good work.
TA L K I N G TO O U R S E LV E S
If the untrained infant's mind i s to become an

intelligent one, it must acquire both discipline
and initiative .
Alan Turing
There is no step more uplifting, more explosive , more

momentous in the history of mind design than the invention
of language. When Homo sapiens became the beneficiary of
this invention, the species stepped into a slingshot that has
launched it far beyond all other earthly species in the p ower
to look ahead and reflect. What is true of the species is just as
true of the individual. No transition is more astronomically
I 48 KI N DS 0F M I N DS
enabling in the life of an individual person than " learning "

to speak. I must put the word in scare-quotes , since we have
come to realize (thanks to the work of linguists and psycho
linguists) that human infants are genetically predesigned for
language in many ways. As the father of modern linguistics ,
Noam Chomsky, o ften says (with excusable exaggeration) ,
birds don't have to l earn their feathers and babies don't have
to learn their language. Much of the hard work of designing
a language user (or a feather user) was accomplished eons
ago and is provided to the infant in the form of innate talents
and dispositions , readily adapted to local conditions of
vocabulary and grammar. Children acquire language at
breathtaking speed , p icking up new words at an average rate
of a dozen a day, for years on end , until they become adoles
cents , when the rate slows to a trickle. They master all but
the finest points of their grammar before they enter school.
In addition to all their linguistic interactions with their fam
ily members (and pets) , babies and toddlers spend many
hours vocalizing to themselves , first babbling, then indulging
in marvelous mixtures of words and nonsense syllables richly
endowed with different tones of voice-hortatory, soothing,
explanatory, caj o ling-and eventually evolving into elaborate
self-commentary.
Children enjoy talking to themselves. What might this be
doing to their minds? I cannot answer that question yet , but I
have some speculative suggestions for further research. Con
sider what happens early in the linguistic life of any child.
"Hot ! " says Mother. "Don't touch the stove ! " At this point ,
the child doesn 't have to know what "hot" or "touch " or
" stove " means-these words are primarily just sounds , audi
tory event-types that have a certain redolence, a certain
familiarity, a certain echoing memorability to the child.
They come to conj ure up a situation-type-stove-approach
and-avoidance-which is not just a situation in which a spe
cific prohibition is typically heard but also a situation in
which a mimicking auditory rehearsal is encountere d .

Crudely simplifying, let's suppose that the c h i l d acquires the
habit of saying to itself (aloud) " Hot ! " "Don't touch ! " with
out much of an idea what these words mean, voicing them
merely as an associated p art of the dril l that goes with
approaching and then avoiding the stove-and also as a sort
of mantra , which might be uttered at any other time. After
all , children are taken with the habit of rehearsing words
they have just heard-rehearsing them in and out o f context
and building up recognition links and association p aths
between the auditory prop erties and concurrent sensory
properties, internal state s , and so forth.
That's a rough sketch of the sort of process that must go
on. This process could have the effect of initiating a habit of
what we might call semi- un derstood self-commen tary. The
child, prompted initially by some insistent auditory associa
tions provoked by its parents ' admonitions , acquires the
habit of adding a sound track to its activities-" comment
ing" on them. The actual utterances would consist at the out
set of large measures of " scribble "-nonsense talk composed
of wordlike sounds-mixed with real words mouthed with
much feeling but little or no appreciation of their meaning,
and a few understood words . There would be mock exhorta
tion, mock prohibition, mock praise, mock description, and
all these would eventually mature into real exhortation , pro
hibition, praise, and description. But the habit of adding
" labels " would thus be driven into place before the labels
themselves were understoo d , or even p artially understood.
I'm suggesting that it's such initially " stupi d " practices
the mere mouthing of labels , in circumstances appropriate
and inappropriate-that could soon turn into the habit of
representing one's own states and activities to oneself i n a
new way. As the child lays down more associations between
the auditory and articulatory processes on the one han d ,
and patterns of concurrent activity on t h e other, this would
create nodes of saliency in memory. A word can become

familiar even without being understood. And it is these
anchors of familiarity that could give a label an independent
identity within the system. Without such independence ,
labels are invisible. For a word to serve as a useful, manipu
lable label in the refinement of the resources of a brain, it
must be a ready enh ancer of sought-for associations that are
already to some extent laid down in the system. Beyond
that , words can be arbitrary, and their arbitrariness is actu
ally p art of what makes them distinctive: there is little risk
of failing to notice the presence of the label ; it doesn 't just
blend into its surroundings , like a dent in the corner of a
shoebox. It wears the deliberateness of its creation on its
sleeve.
The habit of semi-understood self-commentary could, I
am suggesting, be the origin of the practice of deliberate
labeling , in words (or scribble words or other private neolo
gisms) , which in turn could lead to a still more efficient
practice : dropping all or most of the auditory and articula
tory associations and j ust relying on the rest of the associa
tions (and association-possibilities) to do the anchoring. The
chi ld, I suggest, can abandon out-loud mouthings and create
private, unvoiced neologisms as labels for features of its own
activities.
We can take a linguistic object as a fo und object (even if
we have somehow blundered into making it ourselves rather
than hearing it from someone else) and store it away for fur
ther consideration, off-line. Our ability to do this depends
on our ability to re-identify or recognize such a label on dif
ferent occasions , and this in turn depends on the label hav
ing some feature or features by which to remember it-some
guise independent of its meaning. Once we have created
labels and acquired the habit of attaching them to experi
enced circumstance s , we have created a new class of objects
that can themselves become the objects of all the pattern-
T H E C R EATI O N O F T H I N K I N G 151
recognition machinery, association-building machinery, and

so forth. Like the scientists lingering retrospectively over an
unhurried examination of the photographs they took in the
heat of experimental battle , we can reflect on whatever pat
terns there are to be discerned in the various labeled exhibits
we dredge out of memory.
As we improve, our labels become ever more refin e d ,
more p erspicuous , ever better articulated , a n d t h e p o i n t is
finally reached when we approximate the near-magical
prowess we began with : the mere con templation of a repre
sentation is sufficient to call to mind all the appropriate
lessons. We have become understan ders of the objects we
have created. We might call these artifactual nodes i n our
memories , these pale shadows of articulated and heard
words , concepts. A concept, then, is an internal label which
may or may not include among its many associations the
auditory and articulatory features of a word (public or pri
vate ) . But words , I am suggesting , are the prototypes or fore
bears of concepts . The first concepts one can manipulate , I
am suggesting , are " voiced" concepts , and only concepts that
can be manipulated can become objects of scrutiny for us.
Plat o , in the The<FJtetus, compares human memory to a
huge cage o f birds :
S o cRATE s : Now consider whether knowledge is a thing you

can possess in that way without having it about you ,
like a man who has caught some wild birds-p igeons
or what not-and keeps them in an aviary for them at
home. In a sense , of cours e , we might say that he
" has " them all the time inasmuch as he possesses
them, mightn't we?
THEIETETu s : Yes .
S o cRAT E s : B u t in another sense he " has " none of them,
though he ha s g o t control of them , now that he has
made them captive in an enclosure of his own ; he can
take and have hold of them whenever he likes by

catching any bird he chooses , and let them go again;
and it is open to him to do that as often as he pleases.
( 1 9 7c-d, Cornford translation)
The trick is: getting the right bird to come when you need
it. How do we do it? By means of technology. We build elab
orate systems of mnemonic association-p ointers , labels ,
chutes and ladders , hooks and chains . We refine our
resources by incessant rehearsal and tinkering, turning our
brains (and all the associated peripheral gear we acquire)
into a huge structured network of competences. No evidence
yet unearthed shows that any other animal does anything
like that.
C HAPTER 6
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
O U R M I N D S A N D OT H E R M I N D S
Once the child has learned the meaning of

" why " and "becaus e , " he has become a fully
paid-up member of the human race.
Elaine Morgan, The Descen t of the Child:
Human Evolu tion from a New Perspective
OUR CONSCIOUSN ESS, T H E I R M I N DS
A mind looks less miraculous when one sees how it might

have been put together out of part s, and how it still relies on
those parts . A naked human mind-without paper and pen
cil, without speaking, comparing note s, making sketches-is
first of all something we have never seen. Every human
mind you've ever looked at-including most especially your
own , which you look at " from the inside "- is a product not
just of natural selection but of cultural redesign o f enormous
proportions. It's easy enough to see why a mind seems
miraculous , when one has no sense of all the components
1 53
and of how they got made. Each component has a long

design history, sometimes billions of years long.
Before any creature could think, there were creatures
with crude, unthinking intentionality-mere tracking and
discriminating devices that had no inkling of what they were
doing or why. But they worked well. These devices tracked
things , reliably responding to their twists and turns , keeping
on target for the most part , and seldom straying for long
before returning to their task. Over much longer time spans ,
the designs of these devices could also be said to track some
thing : not evasive mate s , or prey, but something abstract
the free-floating rationales of their own functions. As cir
cumstances changed , the designs of the devices changed in
appropriate reponse to the new conditions , keeping their
owners well equipped without burdening them with the rea
sons. These creatures hunted, but didn't think they were
hunting, fled but didn't think they were fleeing. They had
the kn ow-h ow they needed. Know-how is a kind of wisdom ,
a kind of useful information, but it is not represented knowl
e dge .
Then some creatures began to refine that part of the envi
ronment that was easiest to contro l , putting marks both
inside and outside-off-loading problems into the world ,
and j u s t i n to other parts of their brains. They began making
and using representations , but they didn't know they were
doing s o . They didn't need to know. Should we call this sort
of unwitting use of representations " thinking "? If so, then
we would have to say that these creatures were thinking , but
didn't know they were thinking ! Unconscious thinking
those with a taste for " p aradoxical " formulations might favor
this way of speaking , but we could less misleadingly say that
this was in telligen t b u t un thinking behavior, because it was
not just not reflective but also not reflectable-upon.
We human beings do many intelligent things unthink
ingly. We brush our teeth, tie our shoes , drive our cars , and
OUR M I N DS AND OTHER M I N DS 1 55
even answer questions without thinking. But most of these

activities of ours are different , for we can think about them
in ways that other creatures can 't think about their unthink
ing but intelligent activiti es. Indeed, many of our unthinking
activities, such as driving a car, could become unthinking
only after passing through a long period of design develop
ment that was explicitly self-conscious. How is this accom
plished? The improvements we install in our brains when
we learn our languages permit us to review, recal l , rehearse ,
redesign our own activities , turning our brains into echo
chambers of sorts, in which otherwise evanescent processes
can hang around and become objects in their own right.
Those that persist the longest , acquiring influence as they
persist , we call our conscious thoughts .
Mental contents become conscious not by entering some
special chamber in the brain, not by being transduced into
some privileged and mysterious medium , but by winning
the competitions against other mental contents for domina
tion in the control of behavior, and hence for achieving long
lasting effects-or as we misleadingly say, " entering into
memory. " And since we are talkers , and since talking to our
selves is one of our most influential activitie s , one o f the
most effective ways for a mental content to become influen
tial is for it to get into position to drive the language-using
parts of the controls.
A common reaction to this suggestion about human con
sciousness is frank bewilderment, expressed more or less as
follows : " Suppose all these strange competitive processes
are going on in my brain, and suppose that , as you say, the
conscious processes are simply those that win the competi
tions . How does that make them conscious? What happens
next to them that makes it true that I know about them? For
after all , it is my consciousness, as I know it from the first
person point of view, that needs explaining ! " Such questions
betray a deep confusion, for they presuppose that what you
1 56 KINDS OF MINDS
are is something else, some Cartesian res cogitans in addi

tion to all this brain-and-body activity. What you are , how
ever, just is this organization of all the competitive activity
between a host of competences that your body has devel
oped. You " automatically" know about these things going on
in your b ody, because if you didn 't, it wouldn 't be your
body ! (You could walk off with somebody else 's gloves , mis
takenly thinking they were your gloves , but you couldn 't
sign a contract with somebody else's han d , mistakenly
thinking it was your han d , and you couldn't be overcome by
somebody else's sadness or fear, mistakenly thinking it was
your own . )
The acts and events you can tell u s about, and the rea
sons for them , are yours because you made them-and
because they made you. What you are is that agent whose
life you can tell about. You can tell us, and you can tell your
self. The process of self-description begins in earliest child
hood and includes a good deal of fantasy from the outset.
(Think o f Snoopy in the Pean u ts cartoon, sitting on his dog
house and thinking , " Here's the World War I ace, flying into
battle . ") It continues through life . (Think of the cafe waiter
in Jean-Paul Sartre 's discussion of "bad faith" in Being and
Nothingness, who is all wrap ped up in learning how to live
up to his self-description as a waiter. ) It is what we do. It is
what we are .
Are other minds really so different from human minds?
As a simple experiment, I would like you to imagine some
thing that I daresay you've never imagined before . Please
imagine , in some detai l , a man in a white lab coat climbing
hand over hand up a rope while holding a red plastic bucket
in his teeth. An easy mental task for you. Could a chim
panzee p erform the same mental task? I wonder. I chose the
elements-man , rop e , climbing , bucket , teeth-as familiar
objects in the perceptual and behavioral world of a labora
tory chimp . I'm sure that such a chimp can not only perceive
OUR M I N DS AND OTH ER M I N DS 1 57
such things but see them as a man , a rope , a bucket , and so

forth . In some minimal sense , then , I grant that the chimp
has a con cept of a man , a rope , a bucket (but does not have
concepts , presumably, of a lobster, or a limerick, or a
lawyer) . My question i s , What can a chimp do with its con
cepts? Back during World War I , the German psychologist
Wolfgang Kohler did some famous experiments with chimps
to see what sort of problems they could solve by thinking :
Could a chimp figure out how to stack some boxes in its cage
to get at some bananas hanging from the ceiling which were
too high to for it to reach? Alternatively, could it figure out
how to fasten two sticks together into one long enough to
knock the food down? The p opular lore holds that Kohler's
chimps could indeed figure out these solution s , but in fact
the animals were quite unimpressive in action; some solved
the problems only after many, many trials , and others never
saw the light. Later studies , including some current ones
that are much more subtle , still fail to settle these apparently
simple questions about what a chimp can think when pro
vided with all the clues. But let 's suppose for the moment
that Kohler's experiments did answer the question they are
commonly reputed to have answered : that a chimp can
indeed discover the solution to a simple problem of this sort ,
provided that the elements of the solution are visible and
ready to hand-available for trial-and-error manipulation.
My question is different : Can a chimpanzee call to min d
the elements of a solution when these elements are not pres
ent to provide the chimp with visible reminders of them
selves? The exercise you engaged in was provoked by a ver
bal suggestion from me. I am sure that you can just as readily
make suggestions to yourself, and then take your own sug
gestions , thereby framing mental imagery of consi derable
novelty. (That's one of the things we know about ourselves
that we all enjoy engaging in elaborate exercises of imagina
tion carefully tailored to meet our interests of the moment . )
I 58 KIN DS 0F MI N DS
The account I've sketched in previous chapters of how non

human minds work implies that chimps should be incapable
o f such activities . They might somehow happen to put
together the relevant concepts (their sort of concepts) by
accident, and then have their attention drawn to any
serendipitously interesting results , but even that , I suspect ,
is beyond the limits of their resources' movability or manip
ulability.
These questions about the minds of chimps are rather
simp l e , but nobody knows the answers-yet. The answers
are not impossible to acquire , but devising suitable experi
ments is not easy. Notice that these questions are not the sort
that can be addressed by looking at the relative size of the
animal's brain, or even gauging its brute cognitive capacity
(of memory, of discriminatory prowess) . Surely there is
plenty of machinery in a chimp 's brain to store all the infor
mation needed as raw material for such a task; the question
is whether the machinery is organized in such a way as to
permit this s ort of expl oitation. (You have a big aviary, with
plenty of birds; can you get them to fly in formation?) What
makes a mind powerful-indeed, what makes a mind con
scious-is not what it is made of, or how big it is, but what it
can do. Can it concentrate? Can it be distracted? Can it recall
earlier events? Can it keep track of several different things at
once? Which features of its own current activities can it
notice or monitor?
When such questions as these are answered, we will
know everything we need to know about those minds in
order to answer the morally important questions . These
answers will capture everything we want to know about the
concept of consciousnes s , except the idea of whether, as one
author has recently sai d , " the mental lights would be out " in
such a creature . But that 's just a bad idea-in spite of its
popularity. Not only has it never been defined or even clari
fied by any of its champions ; there is no work for such a
OUR M I N DS AND OTH ER M I N DS 1 59
clarification or definition to do. For suppose that we have

indeed answered all the other questions about the mind of
some creature , and now some philosophers claim that we
still don't know the answer to that all-imp ortant question, Is
the mental light on-yes or no? Why would either answer be
important? We are owed an answer to this question, before
we need to take their question seriously.
Does a dog have a concept of cat? Yes an d n o . No matter
how close a dog's " concept" of cat is to yours extensionally
(you and the dog discriminate the same sets of entities as
cats and noncats) , it differs radically in one way: the dog
cannot consider its concept. It cannot ask itself whether it
knows what cats are ; it cannot wonder whether cats are ani
mals; it cannot attempt to distinguish the essence o f cat (by
its lights) from the mere accidents . Concepts are not things
in the dog's world in the way that cats are . Concepts are
things in our worl d , because we have language. A polar bear
is competent vis-a-vis snow in many ways that a lion is not,
so in one sense a polar bear has a concept that the lion
lacks-a concept of snow. But no languageless mammal can
have the concept of snow in the way we can , because a lan
guageless mammal has no way of considering snow " in gen
eral " or " in itself. " This is not for the trivial reason that it
doesn't have a (natural-language) word for snow but because
without a natural language it has no talent for wresting con
cepts from their interwoven connectionist nests and manip
ulating them . We c an speak of t h e polar bear's implicit or
procedural knowledge of sno w (the p olar bear's snow-how) ,
and we can even investigate, empirically, the extension of
the polar bear's embedded snow-concept, but then b ear in
mind that this is not a wieldable concept for the p olar bear.
" It may not be able to talk, but surely it thinks ! "-one of
the main aims of this book has been to shake your confi
dence in this familiar reaction. Perhap s the biggest obstacle
in our attempts to get clear about the mental competences of
nonhuman animals is our almost irresistible habit of imagin

ing that they accompany their clever activities with a stream
of reflective consciousness something like our own. It is not
that we now kn ow that they don't do any such thing; it is
rather that in these early days of our investigations we must
not assume that they do. B oth the philosophical and scien
tific thinking about this issue has been heavily influenced by
the philosopher Thomas Nagel 's classic 1 9 74 paper, "What
Is It Like to Be a Bat?" The title itself sets us off on the wrong
foot, inviting us to ignore all the different ways in which
bats (and other animals) might accomplish their cunning
feats without its "being like " anything for them . We create a
putatively impenetrable mystery for ourselves if we presume
without further ado that Nagel 's question makes sense, and
that we know what we are asking.
What is it like for a bird to build a nest? The question
invites you to imagine how you would build a nest and then
to try to imagine the details of the comparison. But since
nest buil ding is not something you habitually do, you
should first remind yourself of what it's like for you to do
something familiar. Wel l , what is it like for you to tie your
shoelaces? S ometimes you pay attention; sometimes it gets
done by your fingers without any notice at all , while you
think of other things . S o mayb e , you may think, the bird is
daydreaming or p lotting tomorrow 's activities while it exe
cutes its constructive moves. Maybe, but in fact the evidence
to date strongly suggests that the bird is not equipped to do
any such thing. Indeed , the contrast you note between pay
ing attention and doing the task while your mind was other
wise occupied probably has no counterpart at all in the case
of the bird. The fact that you couldn't build a nest without
thinking carefully and reflectively about what you were
doing and why is not at all a good reason for assuming that
when the bird builds its nest , it must think its birdish
thoughts about what it is doing (at least for its first nest,
OUR M I N DS AND OTHER M I N DS 161
before it has mastered the task) . The more we learn about

how brains can engage in processes that accomplish clever
deeds for their nonhuman owners , the less these processes
look like the thoughts we had dimly imagined to b e doing
the work. That doesn't mean that our thoughts are not
processes occurring in our brains , or that our thoughts are
not playing the critical roles in governing our behavior that
we normally assume they are . Presumably some of the
processes in our own human brains will eventually be dis
cernible as the thoughts we know so intimately, but it
remains to be seen whether the mental competences of any
other species depend on their having mental lives the way
we do.
PA I N A N D S U F F E R I N G : W H AT M AT T E R S
There is always a well-known solution to every

human problem-neat, p lausible, and wrong.
H . L. Mencken, Preju dices (second series)
It would be reassuring if we had come to the end o f our story

and could say something along the lines of " An d so we see
that it follows from our discoveries that insects and fish and
reptiles aren't sentient after all-they are mere automata
but amphibians , birds , and mammals are sentient or con
scious just like us! And, for the record , a human fetus
becomes sentient at between fifteen and s ixteen weeks . "
Such a neat, plausible solution t o some o f our human prob
lems of moral decision making would be a great relief, but
no such story can be told yet, and there is no reason to
believe that such a story will unfold later. It is unlikely that
we have entirely overlooked a feature o f mentality that
would make all the difference to morality, and the features
we have examined seem to make their appearance not just

gradually but in an unsynchronized, inconsistent , and
p atchy fashion, both in evolutionary history and in the
development of individual organisms. It is possible, of
cours e , that further research will reveal a heretofore unde
tectable system of c;imilarities and differences which will
properly impress us , and we will then be able to see , for the
first time, where nature has drawn the line , and why. This is
not a possibility on which to lean , however, if we can 't even
imagine what such a discovery might be, or why it would
strike us as morally relevant. (We might just as well imagine
that one fine day the clouds will part and God will tell us,
directly, which creatures to include and which to exclude
from the charmed circ l e. )
In our survey of kinds o f minds (and protominds) there
does not seem to be any c lear threshold or critical mass
until we arrive at the sort of consciousness that we language
using human beings enj oy. That variety of mind is unique ,
and orders of magnitude more p owerful than any other vari
ety of mind, but we probably don 't want to rest too much
moral weight on it. We might well think that the capacity for
suffering counts for more , in any moral calculations , than
the capacity for abstruse and sophisticated reasoning about
the future (and everything else under the sun ) . What , then , is
the relationship between pain, suffering , and consciousness?
While the distinction between pain and suffering is, l ike
most everyday, nonscientific distinctions , somewhat blurred
at the edges , it is nevertheless a valuable and intuitively sat
isfying mark or measure of moral importance. The phenome
non of p ain is neither homogeneous across species , nor sim
ple. We can see this in ourselves , by noting how unobvious
the answers are to some simple questions. Are the stimuli
from our pain receptors-stimuli that prevent us from allow
ing our limbs to assume awkward, j oint-damaging positions
while we sleep-experienced as pains? Or might they be
OUR M I N DS AND OTHER M I N DS 1 63
properly called unconscious pains? Do they have moral sig

nificance , in any case? We might call such body-protecting
states of the nervous system " sentient " states , without
thereby implying that they were the experiences of any self,
any ego , any subject. For such states to matter-whether or
not we call them pains , or conscious state s , or experiences
there must be an enduring subj ect to wh om they matter
because they are a source of suffering.
Consider the widely reported phenomenon of dissocia
tion in the presence of great pain or fear. When young chil
dren are abused , they typically hit upon a desperate but
effective stratagem : they " leave . " They somehow declare to
themselves that it is not they who are suffering the pain.
There seem to be two main varieties of dissociators : those
who simply reject the pain as theirs and then witness it from
afar, as it were ; and those who split at least momentarily into
something like multiple personalities ( " I " am not undergo
ing this pain , "she " i s ) . My not entirely facetious hypothesis
about this is that these two varieties of children differ in
their tacit endorsement of a philosophical doctrine: Every
experience must be the experience of some subj ect. Those
chil dren who reject the principle see nothing wrong with
simply disowning the pain, leaving it subj ectless to wander
around hurting nobody in p articular. Those who embrace
the principle have to invent an alter to be the subj ect-" any
body but me! "
Whether or not any such interpretation of the phenome
non of dissociation can be sustaine d , most psychiatrists
agree that it does work , to some degree. That i s , whatever
this psychological stunt of dissociation consists in, it is gen
uinely analgesic-or, more precisely, whether or not it
diminishes the pain , it definitely obtunds s uffering. S o we
have a modest result of sort s : the differenc e, whatever it i s ,
between a nondissociated child and a dissociated child is a
difference that markedly affects the existence or amount of
suffering. (I hasten to add that nothing I have said implies

that when children dissociate they in any way mitigate the
atrocity of the vile behavior of their abusers ; they do, how
ever, dramatically diminish the awfulness of the effects them
selves-though such children may pay a severe price later in
life in dealing with the aftereffects of their dissociation. )
A dissociated child does n o t suffer a s much a s a nondis
sociated chi l d. But now what should we say about creatures
that are n a turally dissociated-that never achieve, or even
attempt to achieve, the sort of complex internal organization
that is standard in a normal child and disrupted in a dissoci
ated child? An invited conclusion would be: such a creature
is constitutionally incapable of undergoing the sort or
amount of suffering that a normal human can undergo. But if
all nonhuman species are in such a relatively disorganized
state , we have grounds for the hypothesis that nonhuman
animals may indeed feel p ain but cannot suffer the way we
can.
How convenient ! Animal lovers can be expected to
respond to this suggestion with righteous indignation and
deep suspicion. Since it does indeed promise to allay many
of our misgivings about common human practices , absolving
our hunters and farmers and experimenters of at least some
of the burden of guilt that others would place on their shoul
ders , we should be particularly cautious and even-handed in
considering the grounds for it. We should be on the lookout
for sources of illusion-on both sides of this stormy issue.
The suggestion that nonhuman animals are not susceptible
to human levels of suffering typically provokes a flood of
heart-wrenching stories-mostly about dogs. Why do dogs
predominate? Could it be that dogs make the best counterex
amples because dogs actually do have a greater capacity for
suffering than other mammals? It could b e , and the evolu
tionary p erspective we have been pursuing can explain why.
Dogs , and only dogs among domesticated species ,
OUR M I N DS AND OTH ER MI NDS 1 65
respond strongly to the enormous volume of what we might

call " humanizing " behavior aimed at them by their owners .
We talk to our dogs , commiserate with our dogs , and in gen
eral treat them as much like a human companion as we
can-and we delight in their familiar and positive res ponse
to this friendliness. We may try it with cats , but it seldom
seems to take . This is not surprising, in retrospect; domestic
dogs are the descendants of social mammals , accustomed
over millions of years to living in cooperative , highly inter
active groups, while domestic cats spring from asocial lin
eages. Moreover, domestic dogs are importantly unlike their
cousins , the wolves and foxes and coyote s , in their respon
siveness to human affection. There is no mystery about why
this should be s o. Domestic dogs have been selected for j ust
these differences for hundreds of thousands of generations.
In The Origin of Species, Charles Darwin pointed out that
whereas deliberate human intervention in the reproduction
of domesticated species has worked for several thousand
years to breed faster horses , woollier sheep , beefier cattle,
and so forth , a more subtle but still powerful force has been
at work for a much longer time shaping our domesticated
species. He called it unconscious selection. Our ancestors
engaged in selective breeding , but they didn't think they
were doing so. This unwitting favoritism, over the eons , has
made our dogs more and more like us in ways that appeal to
us . Among other traits we have unconsciously selected for, I
suggest, is susceptibility to human socializing , which has , in
dogs , many of the organizing effects that human socializing
also has on human infants . By treating them as if they were
human , we actually succeed in making them more human
than they otherwise would be. They begin to develop the
very organizational features that are otherwise the sole
province of socialized human beings . In short , if human con
sciousness-the sort of consciousness that is a necessary
condition for serious suffering-is , as I have maintaine d , a
1 66 KINDS OF MI NDS
radical restructuring of the virtual architecture of the human

brai n , then it should follow that the only animals that would
be capable of anything remotely like that form of conscious
ness would be animals that could also have imposed on
them , by culture , that virtual machine. Dogs are clearly clos
est to meeting this condition.
What about p ain? When I step on your toe, causing a brief
but definite (and definitely conscious) p ain , I do you scant
harm-typically none at all. The pain, though intense , is too
brief to matter, and I have done no long-term damage to your
foot. The idea that you " suffer" for a second or two is a risi
ble misap p lication of that important notion , and even when
we grant that my causing you a few seconds of pain may irri
tate you for a few seconds or even minutes more-especially
if you think I did it deliberately-the pain itself, as a brief,
negatively-signed experience, is of vanishing moral signifi
cance. (If in stepping on your toe I have interrupted your
singing of the aria , thereby ruining your operatic career, that
is quite another matter. )
Many discussions seem to assume tacitly ( 1 ) that suffer
ing and p ain are the same thing , on a different scale; (2) that
all pain is " experienced p ain " ; and ( 3 ) that the " amount of
suffering" is to be calculated ( " in principl e " ) just by adding
up all the p ains (the awfulness of each of which is deter
mined by duration-times-intensity) . These assumptions ,
looked at dispassionately in the cold light of day (a difficult
feat for some p artisans) , are ludicrous . A little exercise may
help : Suppose, thanks to some " miracle of modern medi
cine , " you could detach all your pain and suffering from the
c ontexts in which it occurred, postponing it all, say, to the
end of the year, when it could be endured in one horrible
week of unremitting agony, a sort of negative vacation , or-if
the formula of assumption ( 3 ) is to be taken seriously-trad
ing off duration for intensity, so that a year's misery could be
OUR M I N DS AN D OTH ER M I N DS 1 67
packed into one excruciating lump-sum j olt lasting , say, five

minutes . A whole year without so much as a mild annoy
ance or headache in exchange for a brief and entirely
reversible descent into hell-without-anesthesia-would you
accept such a bargain? I certainly would, if I thought it made
sense. (We are assuming , of cours e , that this horrible episode
would not kill me or render me insane in the aftermath
though I ' d be quite happy to be insane during the jolt itself! )
In fact , I ' d gladly take the bargain even if it meant " dou
bling" or " quadrupling " the total amount of suffering, just as
long as it would be all over in five minutes and leave no last
ing debilities . I expect anybody would be happy to make
such a deal , but it doesn 't really make sense. (It would imply,
for instance , that the benefactor who provided such a service
gratis to all would, ex hypothesi, double or quadruple the
world's suffering-and the world would love him for it.)
What 's wrong with this scenario is, of cours e , that you
can't detach pain and suffering from their contexts in the
imagined way. The anticipation and aftermath , and the
recognition of the implications for one's life p lans and
prospects , cannot be set aside as the " merely cognitive "
accompaniments of the suffering . What is awful about losing
your j ob , or your leg, or your reputation , or your loved one is
not the suffering this event ca uses in you , but the suffering
this event is. If we are concerned to discover and ameliorate
unacknowledged instances of suffering in the world , we
need to study creatures ' lives, not their brains . What hap
pens in their brains is of course highly relevant as a rich
source of evidence about what they are doing and how they
do it, but what they are doing is in the end just as visible-to
a trained observer-as the activities of plants , mountain
streams, or internal combustion engines. If we fail to find
suffering in the lives we can see (studying them diligently,
using all the metho ds of science) , we can rest assured that
1 68 KINDS OF MINDS
there is no invisible suffering somewhere in their brains. If

we find suffering, we will recognize it without difficulty. It
is all too familiar.
This book began with a host of questions , and-since this
is a book by a philosopher-it ends not with the answers ,
but , I hope , with better versions of the questions themselves .
At least we can see some paths to p ursue , and some traps to
avoid, in our ongoing exploration of the different kinds of
minds .
F U RT H E R R E A D I N G
It might seem that there would be little point in your reading

the books that have influenced me the most in writing this
book, since if I have done my work wel l , I have already
extracted the best bits , saving you the time and trouble. That ' s
true of some of them, perhaps , but n o t t h e books I l i s t here.
These are books that I particularly want my readers to read , if
they haven 't already read them , and read again if they have . I
have learned a lot from them-but not enough ! I am acutely
aware , in fact, that there is much more for me (and everybody
else) to find in these books , and in some ways this book is
meant as an inducement and guide.
First , I submit two fa mous and influential but often misun
derstood books by philosophers : Th e Con cept of Mind ( 1 949) ,
by Gilbert Ryle, and Philosophical In vestigations ( 1 9 5 8 ) , by
Ludwig Wittgenstein. B oth Ryle and Wittgenstein were quite
hostile to the idea of a scientific investigation of the min d , and
the standard wisdom in the " c ognitive revolution" is that we
have seen through and beyond their ruthlessly unscientific
analyses of the mental. Not true. One has to tolerate their often
frustrating misperception of good scientific questi ons , and their
almost total ignorance of biology and brain science, but they
still managed to make deep and important observations that
most of us are only now getting into position to appreciate.
Ryle's account of " knowing how " (as distinct from " knowing
that ") has long attracted the attention and approval of cognitive
scientists , but his notorious claims that thinking could happen
out in the public world and didn't have to go on in some private
1 69
1 70 F U RTH ER READ I N G
thinking place have seemed perverse and ill motivated to most

readers . Some of them no doubt were, but it is surprising to see
how much of Ryl e ' s thought shines when new light is directed
upon it. Wittgenstein, meanwhile, has suffered the admiration
of a horde of misunderstanders who share his antipathy to sci
ence but not his vision. They can be safely ignored; go to the
original , and read it through the lens I have tried to provide. A
similarly placed figure is the psychologist James J. Gibson ,
whose amazingly original book Th e Senses Considered as Per
cep tu al Systems ( 1 968) has been a lightning ro d for mis directed
attacks from cognitive scientists and a holy text for an all-too
devoted cabal of radical Gibsonians. Read it; save them for
later.
Valentino Braitenberg ' s Vehicles: Experimen ts in Syn th etic
Psychology ( 1 984) has inspired a generation of roboticists and
other cognitive scientists and i s , simply, a classic. It will
change the way you think about the mind, if my book has not
already accomplished that transformation. Another philosopher
who has drunk deeply at Braitenberg' s well is Dan Ll oyd , and
his 1 98 9 book, Simple Min ds, covers much of the ground that
this book does , with somewhat different emphases but , I think,
no major disagreements . Dan Lloyd was my informal student
and j unior colleague at Tufts when he was working on his book.
I simply cannot tell what he has taught me and I him; there is a
lot to learn from his book in any case. I could say the same
about some other colleagues of mine at the Center at Tufts,
Kathleen Akins , Nicholas Humphrey , and Evan Thompson. It
was Akins who first showed me, back in the mid-1 980 s , why
and how we must escape old-fashioned epistemology and
ontology when thinking about animal minds. See, for instance,
her essays " Science and our Inner Lives : B irds of Prey , Beasts ,
and the Common (Featherless) Biped" and " What Is It Like to
B e B oring and Myopic?" Nicholas Humphrey came to work
with me for several years in 1 9 8 7 , but I still haven't come to
terms with all the ideas in his A History of th e Min d ( 1 9 9 2 ) , in
spite of many hours of discussion. While Evan Thompson was
at the Center, he was finishing his coauthored book, with Fran
cisco Varela and Eleanor Rosch, Th e Embodied Min d ( 1 990 ) ,
a n d the influences of that b o o k in this b o o k can be readily seen,
F U RT H E R R E A D I N G 171
I am sure. More recently, Antonio Damas i o ' s Descartes ' Error:

Emotion , Reason , and th e Human Brain ( 1 994 ) consolidates
and advances some of the themes in these works , in addition to
opening up new ground of its own .
For a deeper understanding of the role of evolution in
designing the minds of all creatu res , you should read all of
Richard Dawkins' books , beginning with Th e Selfish Gene.
Robert Trivers ' Social Evolu tion is an excellent introduction to
the fine points of sociobiology. The new field of evolutionary
psychology is well represented in an anthology edited by
Jerome Barkow, Leda Cosmides , and John Tooby , Th e A dapted
Min d: Evolu tionary Psychology and th e Gen era tion of Culture
( 1992 ) , and for an eye-opening rethinking of child psychology
and child biology , read Elaine Morgan , Th e Descen t of th e
Child: Human Evolu tion from a New Perspective ( 1995 ) .
O n another front , the cognitive ethologists have filled out
philosophers ' (and psychologists ' ) fantasies about the mental
lives and powers of nonhuman animals with a flood of fascinat
ing experimental and observational work. Donald Gri ffin is the
fa ther of the field. His books Th e Question of Animal A ware
ness ( 19 76) , Animal Thinking ( 1984 ) , and Anim al Min ds ( 1992 )
but even more important, his pioneering investigations of bats '
echolocati on , opened the minds of many to the possibilities in
this field. An exemplary study is Dorothy Cheney and Robert
Seyfarth's work with vervet monkeys , How Monkeys See th e
World ( 1990 ) . Andrew Whiten and Richard Byrne 's anthology ,
Machiavellian Intelligen ce ( 198 8 ) , and Carolyn Ristau ' s anthol
ogy , Cognitive Ethology ( 1991) . provide both classic texts and
astringent analyses of the problems ; and a beautifully illus
trated book by James and Carol Gould , Th e Animal Min d
( 1994 ) , should flavor the theoretical imaginations of everybody
who thinks about animal minds. For the very latest on animal
thinking and communicati on, see Marc Hauser's new book, Th e
Evolu tion of Comm unication , and Derek Bickerton ' s Langu age
and Human Behavior. Patrick Bateson 's 199 1 essay , " Assess
ment of Pain in Animals , " is a valuable overview of what is
known and still unknown about animal pain and suffering.
In chapter 4, I passed swiftly (but reluctantly so) over a large
and fascinating literature on higher-order intentionality-children
1 72 FU RTH ER READ I N G
and animals as " natural p sychologists . " I could get away with
this swiftnes s , I decided, because the topic has received so
much good attention elsewhere recently. Two excellent
books-among many-that explain both the details and why it
is important are Janet Astington's Th e Child 's Discovery of the
Mind ( 1 9 9 3 ) and Simon B aron-Cohen ' s Mindblindn ess ( 1 9 9 5 ) .
I also skimped o n the important topic of A B C learning and
its most promising current models . For the details (and some
nontrivial differences of philosophical opinion well worth con
sidering) see Andy Clark, A ssocia tive Engines: Conn ectionism,
Concepts and Represen tational Ch ange ( 1 9 9 3 ) , and Paul
Churchland , The Engine of Reason , th e Seat of th e Soul ( 1 99 5 ) .
Those who want to get even more serious about the details
(which I recommend) can start with Patricia Churchland and
Terence Sejnowski , Th e Comp u ta tional Brain ( 1 9 9 2 ) . Consider
these books an important reality check on some of my more
impressionistic and enthusiastic speculations . Two more
philosophers whose work should be consulted by anyone who
wants to evaluate the claims I have advanced here by triangu
lating them with some related but quite orthogonal visions are
Gareth Evan s , The Varieties of Reference ( 1 9 8 2 ) , and Ruth Gar
rett Millikan, Language Though t and Other Biological Cate
gories ( 1 984) and White Queen Psychology and Oth er Essays for
Alice ( 1 9 9 3 ) .
The discussion of making things t o think with i n chapters 5
and 6 was inspired not just by Richard Gregory 's Mind in Sci
ence ( 1 9 8 1 ) and Andy Clark and Annette Karmiloff-Smith's
1993 paper, but also by Karmiloff-Smith ' s book Beyond Modu
larity ( 1 9 9 2 ) , a n d by several earlier books that have been fruit
fully rattling around in my brain for years : Julian Jaynes ' Th e
Origins of Consciousn ess in th e Breakdown of th e Bicam eral
Mind ( 1 9 76 ) , George Lakoff and Mark Johnson's Metaphors We
Live By ( 1 980) , Philip Johnson-Laird 's Men tal Models ( 1 9 8 3 ) ,
and Marvin Minsky' s Th e Society of Mind ( 1 9 8 5 ) . A new book
that presents the first actual models of some of these quintes
sentially human activities is Douglas Hofstadter' s Fluid Con
cepts and Crea tive Analogies: Compu ter Models of th e Funda
m en tal Mech anisms of Though t ( 1 9 9 5 ) .
F U RT H E R R E A D I N G 1 73
My 1 9 9 1 book Consciousn ess Explained was primarily

about human consciousness , saying little about the minds of
other animals except by implication. Since some readers who
tried to work out those implications arrived at positions they
found dubious or even alarming, I realized that I had to clarify
my theory of consciousness , extending it explicitly to other
species. Kinds of Minds is one result; another is " Animal Con
sciousness : What Matters and Why , " my contribution to the
conference " In the Company of Animals , " held at the New
School for Social Research, in New York City , April 1 9 9 5 . The
evolutionary underpinnings of my theory of consciousness
have also met with skepticism, which I have addressed in my
1 9 9 5 book, Darwin 's Dangerous Idea. Many of the claims I
advance in Kinds of Minds are drawn from , or are elaborated
upon in, other articles of mine listed in the bibliography.
B I B L I O G RA P H Y
Akins , Kathleen , " Science and Our Inner Lives : B irds of Prey,
Beasts , and the Common (Featherless) Bipe d , " in Marc
Bekoff and Dale Jamieson , eds . , In terpretation and Explana
tion in th e Study of Animal Beha vior, Vol. 1 (Boulder, Col o . :
Westview, 1 990 ) , 4 1 4-4 2 7 .
--- , " What I s I t Like t o Be B oring and Myopic?" in Dahlbom,
ed. , Dennett and His Critics.
Astington, Janet , Th e Child 's Discovery of th e Min d (Cambridge :
Harvard University Press , 1 993 ) .
Balda, Russell P . , and R . J . Turek, " Memory i n B irds , " i n Her
bert L. Roitblat , Thomas G. Bever, and Herbert S . Terrace ,
eds . , Animal Cognition (Hillsdale, N.J. : Erlbaum, 1984 ) ,
5 1 3-5 3 2 .
--- , Alan C. Kamil , and Kristie Grim, " Revisits t o Emptied
Cache Sites by Clark's Nutcrackers (Nucifraga columbiana) , "
Animal Beh a vior 3 4 ( 1 98 6 ) , 1 2 89-1 298.
Barkow , Jerome , Leda Cosmides , and John Tooby, Th e A dapted
Min d: Evolu tionary Psych ology and th e Generation of Cul
ture (Oxford: Oxford University Press, 1 992 ) .
Baron-Cohen, Simon , Min dblin dness : A n Essay on A u tism and
Theory of Min d (Cambridge : MIT Press/A Bradford B ook,
1 995 ) .
Bateson , Patrick, " Assessment o f Pain i n Animals , " Animal
Beh avior 4 2 ( 1 991 ) . 8 2 7-8 39.
Bickerton , Derek, Langu age and Human Beh a vior (Seattle: Uni
versity of Washington Pres s , 1 995 ) .
1 75
1 76 B I B L I O G RA P H Y
Braitenberg , Valentino , Vehicles : Experimen ts in Syn th etic Psy

ch ology (Cambridge , MIT Press/ A Bradford Book, 1 984) .
Cheney , Dorothy, and Robert Seyfarth, How Monkeys See th e
World (Chicago : University of Chicago Press , 1 99 0 ) .
Churchlan d , Patricia, a nd Terence Sejnowski , Th e Compu ta
tional Brain (Cambridge : MIT Press/A Bradford Book, 1 99 2 ) .
Churchland, Pau l , Scien tific Realism a n d th e Plasticity of Min d
(Cambridge , U . K . : Cambridge University Press , 1 9 79 ) .
--- , Th e Engine of Reason, the Seat of th e Soul (Cambridge :
MIT Press/A Bradford Book, 1 9 9 5 ) .
Clark, Andy , A ssocia tive Engines: Conn ectionism, Concepts
and Represen tational Change (Cambridge: MIT Press/ A
Bradford Book, 1 9 9 3 ) .
--- , and Annette Karmiloff-Smith, " The Cognizer's Innards:
A Psychological and Philosophical Perspective on the
Development of Thought , " Min d and Language 8 ( 1 9 9 3 ) ,
4 8 7-5 1 9 .
Dahlbom, B o , ed. , Dennett and His Critics : Demystifying Min d
(Oxford: B l ackwell , 1 9 9 3 ) .
Damas i o , Antonio , Descartes ' Error: Em otion, Reason, an d the
Human Brain (New York: Grosset/Putnam , 1 9 9 4 ) .
Darwin, Charles, Th e Origin of Species (London : Murray , 1 8 5 9 ) .
Dawkins , Richard, The Selfish Gene (Oxford : Oxford University
Press, 1 9 76 ; revised e dition , 1 9 8 9 ) .
--- , a n d John R. Kreb s , " Animal Signal s : Information or
Manipulation?" in John R. Krebs and Nicholas B. Davies,
e ds . , Beh a vioural Ecology, 2 d ed. (Sunderland, Mass . : Sin
auer Associate s , 1 9 78 ) , 2 8 2-309.
Dennett, Daniel , " B rain Writing and Mind Reading , " in K. Gun
derson, ed. , Language, Min d an d Kn o wledge, Minnesota
Stu dies in th e Philosophy of Science, Vol . 7 (Minneapolis:
University ofMinnesota Press , 1 9 7 5 ) . Reprinted in Dennett ,
Brainstorms and later with a postscript in D . Rosenthal , ed. ,
Th e Na ture of Min d (Oxford : Oxford University Press , 1 9 9 1 ) .
--- , " Conditions of Personhood , " i n Amelie Rorty , ed. , Th e
Iden tities of Persons (Berkeley: University of California
Press , 1 9 76 ) . Reprinted in Dennett, Brain storms.
--- , Brainstorms (Cambridge : MIT Press/ A Bradford Book,
1978).
B I B L I O G RA P H Y 1 77
--- , " Where A m I?" i n Dennett , Brainstorms.

--- , " B eyond Belief, " in Andrew Woodfiel d , ed. , Th o ugh t
and Object (Oxford: Oxford University Pres s , 1 98 2 ) .
Reprinted i n Dennett , Th e In ten tional Stan ce.
--- , " Intentional Systems in Cognitive Ethology : The ' Pan
glossian Paradigm' Defended , " Beh a vioral and Brain Sci
ences 6 ( 1 9 8 3 ) , 3 4 3-3 9 0 .
-- , The In ten tional Stance (Cambridge : M I T Press/ A Brad
ford Book, 1 9 8 7) .
--- , Conscio usn ess Explained (Boston: Little , Brown , 1 9 9 1 ) .
--- , " Learning and Labeling" (commentary on Clark and
Karmiloff-Smith) , Min d an d Langu age 8 ( 1 9 9 3 ) , 540-5 4 8 .
-- , " The Message Is: There Is No Medium " (reply to Jack
son, Rosenthal , Shoemaker, and Tye) , Philosophy & Phe
n omen ological Research , December 1 9 9 3 , 889-9 3 1 .
--- , " B ack from the Drawing B oard " (reply t o critics) , in
Dahlbom, ed. , Dennett and His Critics.
--- , Darwin 's Dangero us Idea (New York: Simon & Schuster,
1995).
--- , " Get Real " (reply t o critics) , i n Philosophical Topics, 2 2
( 1 9 9 5 ) , 5 0 5-5 6 8 .
--- , " Animal Consciousness : What Matters a n d Why , " i n
Social Research , 6 2 ( 1 9 9 5 ) , 691-7 1 0 .
--- , forthcoming: " Consciousness : More like Fame than
Television , " for volume from the conference " Interfaces
Brain-Computer, " Christa Maar, Ernst Poppel, and Thomas
Christaller, eds. , to be published by Rowohlt.
--- , forthcoming: "Do Animals Have Beliefs ? " in Herbert L .
Roitblat, ed. , Comparative Approaches t o Cognitive Sciences,
MIT Press.
Eigen , Manfred , Steps Towards Life (Oxford: Oxford University
Press , 1 99 2 ) .
Evans, Gareth, Th e Varieties of Reference (Oxford: Clarendon
Press , 1 9 8 2 ) .
Gaussier, Philippe, and Zrehen , S . , A Constructivist Approach
for Autonomous Agents , " in Adia Magnenat Thalmann and
Daniel Thalmann , eds . , Artificial Life and Virtual Reality
(London: Wiley, 1 994) .
-- , " Avoiding the World Model Trap : An Acting Robot
1 78 B I B L I O G RAPHY
Does Not Need to Be So Smart , " Robotics an d Comp u ter

Integra ted Man ufacturing 1 1 ( 1 994 ) , 2 7 9-2 86.
Gibson , James J. , Th e Senses Considered as Perceptual Systems
(London: Allen & Unwin, 1 96 8 ) .
Goul d , James , a n d Carol Goul d , The Animal Min d (New York:
HPHLP , Scientific American Library , 1 99 4 ) .
Gregory , Richard L . , Min d i n Scien ce : A History of Explan ations
in Psych ology (Cambri dge , U.K. : Cambridge University
Pres s , 1 9 8 1 ) .
Griffin, Donald , Th e Question of Animal A wareness (New York:
Rockefeller University Pres s , 1 9 76 ) .
--- , Anim al Thinking (Cambri dge : Harvard University Press,
1 984) .
--- , Animal Min ds (Chicago : University of Chicago Press ,
1992).
Hasson , 0 . , " Pursuit-Deterrent Signals: Communication
between Predator and Prey , " Tren ds in Ecology and Evolu
tion 6 ( 1 9 9 1 ) , 3 2 5-3 2 9 .
Hebb , Donald , Th e Organiza tion of Beh avior: A Neuropsych o
logical Th eory (New York: Wiley , 1 94 9 ) .
Hofstadter , Douglas R. , Fluid Con cepts a n d Creative Analogies :
Compu ter Models of th e Fun damen tal Mech anisms of
Th o ugh t (New York: Basic Books , 1 9 9 5 ) .
Holley , Tony , "No Hide , N o Seek , " Natural History 7 ( 1 994) ,
42-4 5 .
Humphrey, Nicholas , " Nature 's Psychologists , " New Scien tist
2 9 ( June 1 9 7 8 ) , 900-904. Reprinted in Conscio usn ess
Regained (Oxford : Oxford University Press , 1 9 8 3 ) .
-- , A History of the Min d (London: Chatto & Windus,
1992).
Israel , David , John Perry , and Syun Tutiya, " Executions , Moti
vations and Accomplishments , " Philosophical Review 1 0 2
( 1 9 9 3 ) , 5 1 5-54 0 .
Jaynes , Julian , The Origins of Conscio usn ess in the Breakdown
of th e Bicam eral Min d (B oston: Houghton Mifflin, 1 9 7 6 ) .
Johnson-Laird , Philip N . , Men tal Models (Cambri dge , U.K. :
Cambridge University Press , 1 9 8 3 ) .
Kamil , Alan C. , Russell P. Balda, Deborah J . Olson , and Sally
Good, " Returns to Emptied Cache Sites by Clark's Nutcrackers ,
B I BLI O G RAPHY 1 79
Nucifraga columbiana : A Puzzle Revisite d , " Animal Behav

ior 4 5 ( 1 9 9 3 ) , 241-2 5 2 .
Karmiloff-Smith , Annette , Beyond Modularity: A Developmen
tal Perspective on Cognitive Science (Cambridge : MIT
Press/A Bradford Book, 1 9 9 2 ) .
Lakoff, George , and Mark Johnson, Metaph ors We Live By
(Chicago : University of Chicago Press, 1 98 0 ) .
Lloyd , Dan , Simple Min ds (Cambridge : MIT Press/ A Bradford
Book, 1 98 9 ) .
McFarland , David , 1 98 9 , " Goals, No-Goals a n d O w n Goals , " in
Alan Montefiore and Denis Noble , e ds . , Goals, No-Goals an d
Own Goals : A Debate on Goal-Directed and In ten tional
Beh aviour (London: Unwin Hyman, 1 989) , 3 9-5 7 .
Menzel , Emil W. , Jr. , 1 9 7 1 , " Communication about the Environ
ment in a Group of Young Chimpanzees , " Folia Primatolo
gia 1 5 ( 1 9 7 1 ) , 2 2 0-2 3 2 .
--- , " A Group o f Young Chimpanzees i n a One-Acre Fiel d , "
in A . M . Schreier and F. Stolnitz , e ds . , Beh a vior of Nonh u
m a n Prima tes, V o l . 5 (New York: Academic Press, 1 9 74 ) ,
8 3-1 5 3 . Reprinted i n Ristau , Cognitive Eth ology.
Millikan , Ruth Garrett, Language, Th o ugh t, and Oth er Biologi
cal Categories (Cambridge : MIT Press/ A Bradford Book,
1 98 4 ) .
--- , White Queen Psych ology an d Other Essays for Alice
(Cambridge : MIT Press/A Bradford Book, 1 9 9 3 ) .
-- , " A Common Structure for Concepts of Individual s ,
Stuffs , and Basic Kin : More Mama , More Milk, a n d More
Mouse , " Beha vioral and Brain Sciences, forthcoming.
Minsky , Marvin , The Society of Min d (New York: Simon &
Schuster, 1 9 8 5 ) .
Morgan , Elaine , Th e Descen t of th e Child: Human Evolu tion
from a New Perspective (Oxford : Oxford University Press,
1 99 5 ) .
Nagel , Thomas , " What Is I t Like t o B e a Bat?" Philosophical
Review 83 ( 1 9 74 ) , 4 3 5-4 5 0 .
Nietzsche , Friedrich, Th us Spake Zara th ustra , Walter Kauf
mann , trans. (New York: Viking, 1 9 5 4 ) .
Ristau , Carolyn , ed. , Cognitive Eth ology (Hillsdal e , N.J. : Erl
baum , 1 9 9 1 ) .
1 80 B I B L I O G RAPHY
--- , " Aspects o f the Cognitive Ethology of an Injury-Feign

ing Bird, the Piping Plover , " in Ristau , ed. , Cognitive Eth ol
ogy, 9 1 -1 2 6 .
Ryl e , Gilbert , Th e Concept of Min d (London: Hutchinson,
1 94 9 ) .
Sartre , Jean Pau l , Being a n d Nothingness (L 'Etre et l e Nean t},
1 94 3 , Hazel Barnes , trans. (New York: Philosophical
Library , 1 9 5 6 ; paperback ed. , 1 96 6 ) .
Searl e , John , " Minds , Brains a n d Programs , " Beh avioral and
Brain Scien ces 3 ( 1 980) , 4 1 7-4 5 8 .
Skinner , B . F . , Science a n d Human Beha vior (New York:
Macmillan, 1 9 5 3 ) .
--- , " Behaviorism a t Fifty , " i n T. W. Wann , ed . , Behavior
ism an d Ph en omen ology (Chicago : University of Chicago
Pres s , 1 964 ) , 79-1 0 8 .
Sontag , Susan , O n Ph otography (New York: Farrar , Straus &
Giroux , 1 9 7 7 ) .
Thomas , Elizabeth Marshall , Th e Hidden Life of Dogs (Boston:
Houghton Mifflin, 1 9 9 3 ) .
Trivers , Robert , Social Evolu tion (Menlo Park, Calif. : Benj amin
Cummings , 1 9 8 5 ) .
Varela, Francisco J. , Evan Thompson , and Eleanor Rosch , The
Embodied Min d: Cognitive Scien ce and Human Experience
(Cambridge : MIT Press/A Bradford B ook, 1 9 9 1 ) .
Whiten , Andrew , " Grades of Mind Reading , " i n Charlie Lewis
and Peter Mitche l l , eds. , Children 's Early Un derstan ding of
Min d: Origins an d Developmen t (Hillsdale, N.J. : Erlbaum,
1 9 94 ) , 4 7-70.
--- , and Richard W. Byrne, eds. , Machiavellian In telligen ce
(Oxford : Oxford University Press, 1 98 8 ) .
Wiener, Norbert , Cybernetics; or, Control a n d Comm unica tion
in th e Animal and the Machin e (New York: Wiley, 1 94 8 ) .
Wittgenstein , Ludwig , Philosophical Investiga tions (Oxford :
Blackwe l l , 1 9 5 8 ) .
Young, Andrew, " The Neuropsychology of Awareness , " in
Antii Revonsuo and Matti Kampp inen , Conscio usness in
Philosophy an d Cognitive Neuroscien ce (Hillsdale, N.J. : Erl
baum, 1 994) , 1 7 3-2 0 3 .
I NDEX
ABC learn i n g , 8 7-8 8 , 9 2 , Baron-Cohen , S i m o n , C o g (robot ) , 1 6

1 3 0-3 1 1 72 c o m m u n i c at i o n :
abortion debate , a n d Bates o n , Patric k , 1 7 1 acknowledgment o f
attribution o f m i n d s , 6 behaviori s m , 85-92 ( see m i n d-havers in, 4 ;
aboutness ( see intention- also Skinnerian crea- approximating con-
ality) tures) fabulation in, 1 2 7-2 8 ;
agen c y : birth of, 1 9- 2 0 ; Being and Nothingness comp uter-driven read-
intentional stance ( Sartre ) . 1 5 6 i n g devices a n d , 8 ;
a n d , 2 7-2 8 ; macro- b e l iefs , 44-4 5 ; de re and c o o perat i o n and ,
molecular, 2 0-2 3 ; d e dicta, 1 0 7 n ; inten- 1 2 6-2 7 ; eye-contact
understan d i ng cir- tional stance of other in, 1 5- 1 6 ; incommu-
cumstances of, 4 1-4 3 entities a n d , 4 1 -4 3 ; nicative m i n d s a n d ,
animals : " automat i c " or propositions a n d , 1 2- 1 8 ; sharing experi-
" unconscious" activ- 4 5-46 ences w i t h , 9-1 0 ; s i n -
ity of, 1 3- 1 4 ; c i rcula- B i ckert o n , Derek, 1 7 1 cerity o r deceptions
tion system in, 2 6 , birds: d i stracti o n d i s - i n , 1 0- 1 1
66-6 7 ; insti nctive play b y , 1 2 1-2 3 ; seed comp uters , 59; chess-
understanding among, caching b y , 1 3 7- 3 9 p l a y i n g , 3 0-3 1 ; input
10; neural control sys- B o o l e , George, 1 0 8-9 devices i n , 7 0 ; as
terns i n , 7 5 - 7 6 ; pain B o olean functions , reading devices , 8
i n , 94 ; proof of mind- 1 08-9 concept s , words as pro-
having for, 6-7 ; ques- Braitenberg, Valent i n o , totypes of, 1 5 1 - 5 2
l i o n s about having 1 0 1 , 1 0 9 , 1 70 " C o n d i t i o n s of Person-
minds or being mind- B rooks , Rodney, 1 6 h o o d " (Dennett ) , 1 2 1
l e s s , 1 -2 , 3 -4 Byrne, Richard , 1 7 1 connect i o n i s m , 8 5 , 8 7
animism, 3 3 consciousness: eye-con-
antelopes , stotting Capgras delus i o n , t a c t a n d , 1 5- 1 6 ; M u i -
behavior o f, 1 2 3 , 1 2 6 1 1 1- 1 2 t i p l e Drafts M o d e l of,
anthropomorp h i s m , 2 7 , Cheney , Dorothy, 1 1 6 , 16
30, 3 3 , 97 171 control systems : l i fe-pro-
approximating confabu- chess-playing comput- tecting activities a n d ,
l a t i o n , 1 2 7-28 ers , 3 0-3 1 1 0 9-1 1 ; m e d i a-neu-
Aristotl e , 25 c h i l dren : dissociation tra l , 7 5 - 7 6 ; response
Artifi c i a l Intell igence in, 1 6 3-64 ; language t o environment and ,
Lab , MIT, 1 5- 1 6 acqu i s i t i o n b y , 6 5 - 6 6 ; transducers
associat i o n i s m , 86-8 7 1 4 8-5 0 ; seeing a n d effectors i n ,
Astingt o n , Janet , 1 7 2 humans a s m i n d l e s s 69-73
Atkin s , Kathleen , 1 70 machines , 2 C o s m i d e s , Leda, 1 7 1
auditory signat ures , and Chomsky , Noam , 1 4 8 cybern eti c s , 7 1 n
trac king, 1 04 Chry s i p p u s , 1 1 5
"automatic" activity, 1 3-14 Churc h l a n d , Patric i a , Damas i o , Anton i o , 1 7 1
"automat i c " reflex, 1 7 1 72 Darwi n , Charle s , 9 9 , 1 6 5
autono m i c nervous sys- Churc h l a n d , Pau l , 4 7 , Darwin i a n creature s ,
t e rn , 6 7 1 72 83-85
circulation systems , Darwin 's Dangero u s
B a l d a , Russel l , 1 3 7- 3 8 2 5 - 2 6 , 66-6 7 Idea (Dennett ) , 1 9n ,
Barkow, Jerome, 1 7 1 Clark, A n d y , 1 3 2 , 1 7 2 59, 81n, 1 73
181
1 82 I N D EX
Dawk i n s , Richard , 1 2 2 , game theory , 60 intention ( i n the ordi-

126, 171 Gaussier, P h i l i p p e , nary sense ) . d i stin-
decept i o n i n communi- 1 3 9-40 guished from the
cat i o n , 1 0-1 1 generat e-and-test , 8 3 - 9 3 p h i l osophical term ,
de dicta beli efs , 1 0 7 n Gibs o n , J a m e s J . , 1 7 0 34-3 5
Dennett, Dan i e l , 1 9n , glutamate m o l e c u l e s , i nten t i o n a l i t v , 34-3 5 ,
107n, 1 7 3 6-7 , 74 1 5 4 ; abou iness is,
de re b e l i e fs , 1 0 7 n Gould , James a n d Caro l , 3 5-3 7 ; approximating
derived intentiona l i t y , 171 confabu lation a n d ,
5 0-5 5 Gregorian creature s , 9 9 , 1 2 8 ; derived v s . origi-
design stance , 28-30 1 00 , 1 0 1 , 1 1 7 , 1 3 4 , nal , 50-5 5 ; i n early
Descart e s , Rene, 1 -2 , 3 , 141 . 142 control system s , 6 5 ;
7 2 , 7 7 , 79-80 Gregory, Ric hard . kee p i n g a secret a n d ,
d i s s o c i a t i o n , 1 6 3-64 99-1 0 0 , 1 7 2 1 2 9-3 0 ; mistaking
d i straction d i s p l a y , Griffi n , Dona l d , 1 7 1 and , 3 7-38
1 2 1-23 intentional object, 3 7
DNA, 2 0 Hasso n , 0 . . 1 2 3 in tentional stance,
d o g s : identification of Hauser, Marc, 9 4 , 1 7 1 2 6-4 0 , 1 0 0-1 0 1 ;
masters b y , 1 1 3- 1 5 ; hearing, 8 1 adapted toward
insti nctive under- Hebb , Dona l d , 8 7 plants , :l 3-3 4 ; adopt-
standing among, 1 0 ; Hebbian learning ru l e s , ing, 2 7-40 , 1 1 9-2 0 ;
i n t e l l igence o f, 87 anthropomorphism
1 1 5- 1 6 ; response to Hidden Life of Dogs, Th e of, 2 7 , 3 0 , 3 3 ; chess-
human behav ior, (Thomas) . 1 0 playing computers
1 64-6 5 ; u n c o n s c i o u s H ofstadter, Douglas, 1 7 2 an d , 3 0-3 1 ; im preci -
selection of human Holley, Tony, 1 2 3 sion of, 4 1 -4 3 ; natural
traits in, 1 6 5 ; use of Homer, 1 1 3 selection and , 60-6 1 ,
concepts a n d , 1 5 9 human beings : a s a class 63; physical stance
d o l p h i n s , i n t e l ligence o f m i nd-havers , 4-5 ; and design stance
o f, 1 1 6 descent from s e l f- con trasted with,
d u a l i s m , 2 3-2 4 , 7 7 re pl icating robot s , 2 8-3 0 ; referential
2 2- 2 4 ; as m i n d less opacity o f, 40; und er-
effectors, in nervous sys- machines , 2 standing agency a n d ,
terns , 6 9- 7 0 , 74-75 H u m e , Davi d , 86-8 7 41
Eigen , Manfred, 2 1 -2 2 , Humphrey, N i c h o l a s . i n t e n t i o n a l systems , 2 6 ,
24 1 20, 1 2 4 , 1 70 3 4 ; birth of function
EPIRB (Emergency P o s i - hypothesis-testing, 1 1 7 , a n d , 3 2 -3 3 ; desiring
tion I n d i cating Radio 1 2 0 , 1 34 evil a n d , 3 2 ; higher
Beacon ) , 1 0 3 -4 order, 1 1 9- 2 0 . 1 2 1 ;
epistemology, 2 i d e a s , p i cture theory of, mi staking a n d , 3 7-3 8 ;
Evan s , Garet h , 1 7 2 5 1 -52 p r o p o s i t i o n a l atti-
evil , and i ntentional sys- i d e n t i fi c a t i o n : of a mas- tudes and, 4 1 -4 5 ; sen-
tern s , 3 2 ter b y a dog, 1 1 3-1 5 ; tience of, 64
evolut i o n : brain and , 7 9 ; o l faction a n d , 1 0 3 , intri nsic intentionality,
natural selection and, 1 04 ; v i s u a l and audi- 5 0- 5 5
8 3 ; perception a n d , t o r y signatures i n , Israe l , Davi d , 40
82; unconscious 1 04
selection i n , 1 6 5 imprinting, 1 04 Jaynes , J u l i a n , 1 4 5 , 1 7 2
exten s i o n , 3 8- 3 9 i n fants , a n d eye-contact , J o h n s o n , Mark, 1 7 2
eye contact , 1 5 15 Johnson-Laird , Phi l i p .
i n formation process i n g , 172
face identification, 1 1 2-1 3 68-6 9 , 73-74
fantas y , 1 1 intel l igence; Gregorian Karm i l o ff-Smith ,
fl owers , moral standing creatures and . 9 9 ; Annett e . 1 3 2 , 1 7 2
of, 5 kinetic and essent i a l , Khepera robot s , 1 4 0
fo l k psychology , 2 7 9 9 ; p e t s and . 1 1 5- 1 6 ; Kinetic Intell igence , 9 9
Frege, Gott lob, 4 4 tool u s e and , 9 9 - 1 00 Kohler, W o l fgang,
functi o n , birth of, 3 2 - 3 3 inten s i o n , 38-39 1 5 7-58
funct i o n a l i s m , 68-6 9 , 76 inten s i o n a l i t y , 3 8-4 0 Kreb s , john R. , 1 2 6
I N D EX 1 83
l abeling, 1 3 5-3 9 , 1 4 9-50 off-l oading cognitive original intent i o n a l i t y ,

Lakoff, George, 1 7 2 t a s ks b y , 1 34-3 5 ; 5 0-5 5
language , 1 1 7 ; acq u i s i - overattribu t i n g , 5 , Origin of Species, The
tion o f, 1 4 8-5 0 ; 6-7 ; protom i n d s , 1 8 ; (Darwin ) , 1 65
incommuni cative p s e u d o m i n d s , 1 8 ; sci-
m i n d s and lack o f, e n t i fi c proof o f, 6-7 : p a i n : dissociation a n d ,
1 2-1 8 ; intensiona l i t y u n d erattributing, 5-6; 1 6 3-64 ; distinction
i n , 38--40: intrinsic using words a n d , 8-9 between s u ffering a n d ,
intentionality a n d , M i n s k y , Marv i n , 1 7 2 1 6 2 , 1 66-6 7 ; feeling
5 0-5 1 : m i nd-havers mist akes , and tracking, of, 1 6- 1 7 ; i n animals,
and u s e o f, 8-9 , 1 1 1- 1 2 94 ; negative rei n force-
1 7-1 8 ; o f thought, 5 1 ; mistaking, and inten- ment a n d , 95-96; sen-
sharing experi ences t i o n a l i t y , 3 7-38 t i ence a n d , 95-98
and, 9-1 0 : trans lation MIT, 1 6 P a v l o v , Ivan , 8 7
and, 9 , 1 8 Mondada, Francesco, percepti o n : as action at a
l atent learning, 9 1 140 d i stance , 8 1-8 2 ; evo-
Lloyd. D a n , 1 7 0 Morgan, E l a i n e , 1 5 , 1 5 3 , l u t i o n o f, 8 2 ; face
Lorenz, Kar l , 1 04 171 i d e n t i fication a n d ,
M u l t i p l e D r a ft s M o d e l o f 1 1 2- 1 3 ; i d e n t i fication
McFarland , Dav i d , consciousness , 1 6 of a master by a dog,
1 2 6-2 7 mutations, 5 9 1 1 3- 1 5
mach i n e s . human bei ngs Perry , J oh n , 4 0
as m i n d l e s s , 2 ( see Nage l , Thomas , 1 60 phenotypic p l a s t i c i t y , 84
also robot s ) natural select i o n , 5 7 , 8 3 , p h i phenomenon , 1 0 6
macrom olecules: agency 1 5 3 ; design a n d , p h otograph y , 1 4 3--44
a n d , 20-2 3 ; self-re p l i - 5 8-60 ; intentional- phototax i s , 1 0 1 -2 , 1 09
cating, 2 1 -2 2 , 4 8 , 5 8 stance i nterpretati o n p h y s i c a l stance, 2 8 - 3 0
Mamata x i s , 1 02--4 o f, 60-6 1 ; pace o f, p i c t u re theory of i deas ,
Massachusetts Institute 6 1 -6 3 ; sensitivity t o 5 1 -5 2
of Tech n o l ogy ( M IT) , changing conditions p i n e a l g l a n d , 67, 7 2
16 i n , 63-65 p l ants : adopting inten-
measurement, and " Nature ' s Psychologi s t s " t i o n a l s t a n c e toward ,
propos i t i o n s , 4 7--4 8 ( H u m phrey ) . 1 2 0 3 3-3 4 ; circulation
Meno ( P l ato ) . 32 negative rei n forcement, system in, 2 6 ; early
Menze l , Emi l , 1 2 9 and p a i n , 95-96 control systems a n d ,
M i l ler, George , 82 nervous system: i n for- 65-66; response to
M i l l i k a n , Ruth Garrett , mation transfer i n , evolutionary change
1 1 4 , 1 72 7 3 - 7 4 ; transdu cers i n , 6 3-64
mind-havers: difference and effectors i n , Plato , 3 2 , 1 5 1 - 5 2
between n o t having a 69-7 3 , 74-75 Popper, Karl , 8 8
m i n d and, 1 4-1 5 ; feel- nest-b u i l di n g , 1 1 0- 1 1 , Popperian creatures ,
ing pain and , 1 6-1 7 ; 1 60-6 1 88-9 3 , 98-9 9 ,
knowledge of member- neurotransmitt ers , 6-7 . 1 00-1 0 1 , 1 3 1 , 1 4 1--4 2
ship in class of, 4-5 ; 74 Potential Inte l l igence, 99
language use and, 8-9 ; Nietzsche, Friedrich , propositional attitudes ,
moral-standing and, 78-79 4 1 -4 5
4-5 : proof of presence n u l l hypothes i s , 7 prop o s i t i o n s : express i o n
of mind and, 6-7 nutritive sou l , 2 5 as sentences , 4 5 --4 7 ;
m i n d s : anticipations o f l ike d o l lars , 4 7 ; mea-
the fut ure in, 5 7- 5 8 ; o ffl o a d i n g , 1 3 4-3 5 surement a n d , 4 7-48
body and, 78-8 0 ; o l fact i o n , 8 1 , 1 0 3 , 1 04 prosopagn o s i a , 1 1 2- 1 3
hemi-semi-demi- ON/OFF switches: birth Pro u s t , Marce l , 1 0 3
m i n d s , 1 8 ; i n forma- of fu nction and , p s e u do-agents , 2 6
lion processing b y , 3 2-3 3 ; nutritive s o u l
68-6 9 ; knowing and , 2 5 Qui n e , W . V . 0 . , 5 1
abo u t , 1 - 3 ; language On Ph o tography (Son-
use a n d , 8-9 , 1 7-1 8 : tag ) . 1 4 3 rea ding devices , 8
moral standing and ontology . 2 referential opacity, 4 0 ,
possession o f , 4-5 ; opacity, referential, 4 0 , 4 5 45
1 84 I N D EX
referential trans p aren c y , Seyfart h , Robert , 1 1 6 , eating Ra dio Beacon)

39 1 71 for, 1 0 3-4 ; l i n king of
reflexes , 1 7 si ncerity i n communica- multiple systems i n ,
reinforcement ( s e e also t i o n , 1 0-1 1 1 0 7-9 ; Mamataxis as,
behaviori s m ; Skinner- Skinner, B. F . , 8 5 , 1 4 1 1 02-3 ; mi stakes i n ,
i a n creature s ) : i n con- Skinnerian condition- 1 1 1 - 1 2 ; phototaxis as,
nectionism, 85; latent ing, 8 5 , 8 8 , 95 1 0 1 -2 , 1 0 9
learning a n d , 9 1 Skinnerian creatures , 8 5 , transducers , i n nervous
re-representati o n , 8 8 , 9 2 , 9 5 , 1 00-1 0 1 syste m s , 69-7 3 , 74-75
1 44-4 5 S mart Moves , 9 9 translation, 9 , 1 8
Ristau, Caro l y n , 1 2 2 , smel l , sense o f, 8 1 transparency, referen-
171 Socrates, 3 2 tial, 3 9
RNA , 2 0 solipsism , 2 , 3 , 4 Trivers , Robert , 1 1 0 , 1 7 1
RNA phage, 2 1 - 2 2 S o n t a g , Susan , 1 4 3 Tutiya, Syun, 4 0
rob o t s , 1 3 ; Cog, 1 6 ; eye- Stei n , Lynn Andrea, 1 6
contact with, 1 5- 1 6 ; stotting, 1 2 3 , 1 2 6 " unconscious" activity ,
intrinsic intentional- s u ffering, distinction 1 3- 1 4
i t y a n d , 5 3-5 5 ; Khep- between pain and, unconscious selectio n ,
eras , 1 4 0 ; organizing 1 6 2 , 1 66-6 7 1 65
barriers in the envi- symbo l s , 1 3 2- 3 3 unconscious thinking,
ronment b y , 1 3 9-4 1 ; synapses , 74 1 5 4-5 5
s e l f-rep licating, 2 0 ,
2 2-24 talking: acknowledg- Valery , Pau l , 5 7
Rosch, E leanor, 1 70 ment of m i nd-havers Vare l a , Franc i s c o , 1 70
Ryle, Gilbert, 1 6 9 i n , 4 ; comp uter- vegetative state , 67
driven reading Vehicles ( B raitenberg ) ,
S artre , Jean-Pa u l , 1 5 6 devices a n d , 8 ; shar- 1 0 1 , 1 70
Searle , Joh n , 5 0 , 5 5 ing experiences with, viruse s , self-replicating,
secret-keeping, 1 2 9- 3 0 9-1 0 ; sincerity or 2 1 -22
S e j nowski , Terence, 1 7 2 deceptions i n , 1 0- 1 1 ; v i s i o n , 8 1 -8 2
s e l f-consciousnes s , a n d to ourselves , 1 4 7- 5 2 v i s u a l signatures , a n d
hypothesis-testing, T a l l eyran d , Charles- tracking, 104, 1 0 7
120 Maurice d e , 1 2 6 vitalism , 2 4 , 7 6
Selfish Gene, The Th eretetus (Plato ) , von Neumann , John, 2 0
(Dawkin s ) . 1 2 2 , 1 7 1 1 5 1-52
s e l f-replication, 2 0 ; Thomas , E l i zabeth M ar- Whiten , An drew ,
macromolecular, shall , 10 1 2 4-2 5 , 1 7 1
2 1 - 2 2 , 4 8 , 5 8 ; rob o t s , Thompson , Evan, 1 70 Wiener, Norbert , 7 1 n
2 0 , 2 2- 2 3 Thorndike, E . L . , 8 7 Wittgenstei n , Ludwig,
sensitivity: evolutionary Th u s Spake Zarath ustra 1 8 , 169
change a n d , 6 3 -6 5 ; (Nietzsche) . 78-79 word s , 8-1 1 , 1 46-4 7 ;
sentience a n d , 64-6 5 ; timescale chauvinism , computer-driven
vegetative state a n d , 6 1 -64 reading devices a n d ,
67 Tooby , J o h n , 1 7 1 8 ; Gregorian creatures
sentences , propositions t o o l u s e , and intelli- a n d , 100; intensional-
expressed as , 4 5 -4 7 gence, 9 9- 1 0 0 ity a n d , 3 8-40 ; mind-
sentience: a n i m a l body- Tower o f Generate-and- havers and user of,
maintenance systems Tes t , 83-93 8-9 ; orototypes o f
a n d , 66-6 7 ; nervous tracking: complex, l i fe- concepts , 1 5 1 -5 2 ;
system a n d creation protecting activities sharing experiences
of, 72-7 3 ; pain per- and, 1 09-l l ; c o opera- with, 9-1 0 ; sincerity
ception and, 9 5 -9 8 ; t i v e versus competi- or deceptions i n
saarch for , 93-9 8 ; sen- tive, 1 0 5-6 ; discern- using, 1 0-1 1 ; trans la-
sitivity and, 64-6 5 ; i n g fai l ures i n , tion a n d , 9
vegetative state a n d , 1 1 6-1 7 ; EPIRB ( Erner-
67 gency Position Indi- Young, Andrew , 1 1 2- 1 3

Kinds of Minds Toward An Understanding of Consciousness - Daniel C Dennett

Uploaded by

Copyright:

Available Formats

Kinds of Minds Toward An Understanding of Consciousness - Daniel C Dennett

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

Kinds of Minds Toward An Understanding of Consciousness - Daniel C Dennett

Uploaded by

Copyright:

Available Formats

KINDS OF MINDS

Toward an Understanding of Consciousness

The Science Masters name and marks are

Copyright © 1996 by Daniel Dennett.

All rights reserved. Printed in the United States

Designed by Joan Greenfield

What Kinds of Minds Are There? 1

Knowing Your Own Mind 1

Words and Minds 8

The Problem of In comm unica tive Minds 12

2 Intentionality: The Intentional Systems Approach 19

Simple Beginnings : Th e Birth of Agency 19

Adopting th e In ten tional Stance 27

Th e Misguided Goal of Propositional Precision 41

Original and Derived In ten tionality 50

3 The Body and Its Minds 57

From Sensitivity to Sen tience? 57

The Media and th e Messages 65

"My Body Has a Mind of Its Own!" 73

4 How Intentionality Came into Focus 81

Th e Tower of Generate-and- Test 81

The Search for Sen tien ce: A Progress Report 93

From Phototaxis to Metaphysics 98

s The Creation of Thinking 119

Un thinking Natural Psychologists 119

Making Things to Think With 134

Talking to Ourselves 147

6 Our Minds and Other Minds 153

Our Consciousn ess, Their Minds 153

Pain and S uffering: Wha t Matters 161

Further Reading 169

I am a philosopher, not a scientist, and we philosophers are

my earlier questions. Other philosophers have offered rival

Can we ever really know what is going on in someone else's

maintained this in the seventeenth century. Might he have

these are different kinds of knowledge . You know you have a

If we want to consider the question of whether nonhuman

" We left Houston at dawn, headin ' down the road-just

Strange . If this fellow thinks his truck is such a worthy com­

reasons , or (if I'm punishing you j ustly for a misdeed of

feel if you were treated as an inanimate object? (Notice how

signaling between nerve cells) amounts to the presence o f

It is beyond serious dispute , however, that you and I each

eagerly you would both seek to confirm your hard-won

pilot. We can know much more about these topics than we

difficulty of determining whether a speaker is sincere .

minds-the minds of those who can understand a conversa­

amount of further investigation will ever tell us wh eth er

The heart-stopp ing thing about the new-born is that , from

As an observation about how we human observers

Lab at MIT, Rodney Brooks and Lynn Andrea Stein have

Ludwig Wittgenstein famously sai d , "If a lion could talk, we

I notice something and seek a reason for it: this

SIM PLE B EG I N N I NGS:

N o grain of sand has a mind; a grain of sand is t o o simp l e .

*Portions of this section are drawn from my 1995 book, Darwin's

count. These atomic-level parts interact , in their obviously

there are reasons for what macromolecules do , but the

First of all , the virus needs a material in which to pack

Strange . If this fellow thinks his truck is such a worthy com

minds-the minds of those who can understand a conversa

It is as if these cells and cell assemblies were tiny, simple

is its turn to play (there will usually be several dozen candi

that the present shape of the bimetallic spring in a thermo

such discourse exhibits referen tial opacity; it is not transpar

these ways get composed into quite particular and idiosyn

The German logician Gottlob Frege introduced an exam

ors-at least when we express our attributions in language

as real as any minds anywhere , that conducted their activi

transform , and otherwise process information about the con