Foundations of Cognitive Psychology: Core Readings pot

in the brain, and the problem of other minds concerns how people come to believethat other people or animals are also conscious.1.1.1 The Mind-Body Problem Although there is a long histo

There are scientists from nearly every field engaged in this pursuit Physiciststry to understand how physical matter can give rise to that ineffable state wecall consciousness, and the decidedly nonphysical ‘‘mind stuff’’ that Descartesand other philosophers have argued about for centuries Chemists, biologists,and neuroscientists join them in trying to explicate the mechanisms by whichneurons communicate with each other and eventually form our thoughts, mem-ories, emotions, and desires At the other end of the spectrum, economists studyhow we balance choices about limited natural and financial resources, andanthropologists study the influence of culture on thought and the formation ofsocieties So at one end we find scientists studying atoms and cells, at the otherend there are scientists studying entire groups of people Cognitive psycholo-gists tend to study the individual, and mental systems within individual brains,although ideally we try to stay informed of what our colleagues are doing Socognition is a truly interdisciplinary endeavor, and this collection of readings isintended to reflect that.

Why Not a Textbook?

This book grew out of a course I took at the Massachusetts Institute of nology (MIT) in 1975, from Susan Carey and Merrill Garrett (with occasionalguest lectures by Mary Potter), and courses I taught at the University of Ore-

Tech-gon, Stanford University, and the University of California at Berkeley When Itook cognition at MIT, there were only two textbooks about cognition as a field(if it cou ld even be thou ght of as a field then): Ulric Neisser’s Cognitive Psy-chology and Michael Posner’s Cognition: An Introduction Professors Carey andGarrett supplemented these texts with a thick book of hand-picked readingsfrom Scientific American and mainstream psychology journals Reading journalarticles prepared the students for the debates that characterize science Susanand Merrill skillfully brought these debates out in the classroom, through inter-active lectures and the Socratic method Cognition is full of opposing theoriesand controversies It is an empirical science, but in many cases the same dataare used to support different arguments, and the reader must draw his or herown conclusions The field of cognition is alive, dynamic, and rediscoveringitself all the time We should expect nothing less of the science devoted tounderstanding the mind.

Today there are many excellent textbooks and readers devoted to cognition.Textbooks are valuable because they select and organize a daunting amount ofinformation and cover the essential points of a topic The disadvantage is thatthey do not reflect how psychologists learn about new research—this is mostoften done through journal articles or ‘‘high-level’’ book chapters directed tothe working researcher More technical in nature, these sources typically revealdetails of an experiment’s design, the measures used, and how the findings areinterpreted They also reveal some of the inherent ambiguity in research (oftenhidden in a textbook’s tidy summary) Frequently students, when confrontedwith the actual data of a study, find alternate interpretations of the findings,and come to discover firsthand that researchers are often forced to draw theirown conclusions By the time undergraduates take a course in cognition (usu-ally their second or third course in psychology) they find themselves wonder-ing if they ou ght to major in psychology, and a few even think about going tograduate school I believe they ought to know more about what it is like to readactual psychology articles, so they’ll know what they’re getting into

On the other hand, a book of readings composed exclusively of such primarysources would be difficult to read without a suitable grounding in the field andwould leave out many important concepts, lacking an overview That is, it mighttend to emphasize the trees at the expense of the forest

Therefore, the goal of this anthology is to combine the best of both kinds

of readings By compiling an anthology such as this, I was able to pick andchoose my favorite articles, by experts on each topic Of the thirty-nine selec-tions, ten are from undergraduate textbooks, six are from professional journals,sixteen are chapters from ‘‘high-level’’ books aimed at advanced students andresearch scientists, and seven are more or less hybrids, coming from sourceswritten for the educated layperson, such as Scientific American or popular books(e.g., Gardner, Norman) This book is not intended to be a collection of the mostimportant papers in the history of cognitive psychology; other authors havedone this extremely well, especially Lloyd Komatsuin his excellent Experiment-ing with the Mind (1994, Brooks/Cole) It is intended as a collection of readingsthat can serve as the principal text for a course in cognitive psychology or cog-nitive science

The particular readings included here owe their evolution to a course I taught

at the University of California at Berkeley in the fall of 1999, ‘‘FundamentalIssues in Cognitive Science.’’ The readings for that course had been carefullyhoned over ten years by Stephen Palmer and Alison Gopnik, outstandingteachers whose courses are motivated by an understanding of the philosophicalbasis for contemporary cognitive psychology I had never seen cognitive psy-chology taught this way, but once I did I couldn’t imagine teaching it any otherway A fundamental assumption I share with them is that cognitive psychology

is in many respects empirical philosophy By that I mean that the core questions

in cognitive psychology were for centuries considered the domain of phers Some of these questions include: What is the nature of thought? Doeslanguage influence thought? Are memories and perceptions accurate? How can

philoso-we ever know if other people are conscious?

Aristotle was the first information-processing theorist, and without tion one can argue that modern cognitive psychology owes him its heritage.Descartes launched modern approaches to these questions, and much currentdebate references his work But for Aristotle, Descartes, Hume, Locke, Husserl,and others, the questions remained in the realm of philosophy A century and

exaggera-a hexaggera-alf exaggera-ago this exaggera-all chexaggera-anged when Wundt, Fechner, Helmholtz, exaggera-and their cohortsestablished the first laboratories in which they employed empirical methods toprobe what had previously been impenetrable to true science: the mind Philos-ophers framed the questions, and mental scientists (as they were then some-times called) conducted experiments to answer them

Today, the empirical work that interests me most in the field of Cognition istheory-driven and builds on these philosophical foundations And a new group

of philosophers, philosophers of mind, closely monitor the progress made bycognitive psychologists in order to interpret and debate their findings and toplace them in a larger context

Who Is This For?

The book you have before you is intended to be used as a text for the graduate cognitive psychology class I teach at McGill University I hope thatothers will find some value in it as well It should also be suitable for studentswho wish to acquaint themselves through self-study with important ideas incognition The ambitious student or professor may want to use this to sup-plement a regular textbook as a way to add other perspectives on the topicscovered It may also be of use to researchers as a resource that gathers up keyarticles in one place It presupposes a solid background in introductory psy-chology and research methods Students should have encountered most of thesetopics previously, and this book gives them an opportunity to explore themmore deeply

under-How the Book Is Organized and under-How It Differs from Other Books

The articles in this reader are organized thematically around topics ally found in a course on cognitive psychology or cognitive science at the uni-

tradition-Preface xv

versity level The order of the readings could certainly be varied without loss ofcoherence, although I think that the first few readings fit better at the begin-ning After that any order should work.

The readings begin with philosophical foundations, and it is useful to keepthese in mind when reading the remainder of the articles This reflects the viewthat good science builds on earlier foundations, even if it ultimately rejectsthem

This anthology differs from most other cognition readers in its coverage ofseveral topics not typically taught in cognition courses One is human factorsand ergonomics, the study of how we interact with tools, machines, and arti-facts, and what cognitive psychology can tell us about how to improve the de-sign of such objects (including computers); this is represented in the excellentpapers by Don Norman Another traditionally underrepresented topic, evolu-tionary psychology, is represented here by two articles, one by David Buss andhis colleagues, and the other by John Tooby and Leda Cosmides Also unusualare the inclusion of sections on music cognition, experimental design, and asmentioned before, philosophical foundations You will find that there is some-what less coverage of neuroscience and computer science perspectives on cog-nition, simply because in our department at McGill, we teach separate courses

on those topics, and this reader reflects an attempt to reduce overlap

Acknowledgments

I would like to thank the many publishers and authors who agreed to let their works be included here, my students, and Amy Brand, Tom Stone, Carolyn Anderson, Margy Avery, and Kathleen Caruso at MIT Press I am indebted in particular to the following students from my cognition class for their tireless efforts at proofreading and indexing this book: Lindsay Ball, Ioana Dalca, Nora Hussein, Christine Kwong, Aliza Miller, Bianca Mugyenyi, Patrick Sabourin, and Hannah Wein- stangel I also would like to thank my wife, Caroline Traube, who is a constant source of surprise and inspiration and whose intuitions about cognitive psychology have led to many new studies Finally, I was extraordinarily lucky to have three outstanding scholars as teachers: Mike Posner, Doug Hintzman, and Roger Shepard, to whom this book is dedicated I would like to thank them for their patience, inspiration, support, and friendship.

Foundations—Philosophical Basis, The Mind/Body Problem

in the brain, and the problem of other minds concerns how people come to believethat other people (or animals) are also conscious.

1.1.1 The Mind-Body Problem

Although there is a long history to how philosophers have viewed the nature ofthe mind (sometimes equated with the soul), the single most important issueconcerns what has come to be called the mind-body problem: What is the relationbetween mental events (e.g., perceptions, pains, hopes, desires, beliefs) andphysical events (e.g., brain activity)? The idea that there is a mind-body prob-lem to begin with presupposes one of the most important philosophical posi-tions about the nature of mind It is known as dualism because it proposes thatmind and body are two different kinds of entities After all, if there were nofundamental differences between mental and physical events, there would be

no problem in saying how they relate to each other

Dualism The historical roots of dualism are closely associated with the ings of the great French philosopher, mathematician, and scientist Rene´Descartes Indeed, the classical version of dualism, substance dualism, in whichmind and body are conceived as two different substances, is often called Carte-sian dualism Because most philosophers find the notion of physical substancesunproblematic, the central issue in philosophical debates over substance dual-ism is whether mental substances exist and, if so, what their nature might be.Vivid sensory experiences, such as the appearance of redness or the feeling ofpain, are among the clearest examples, but substance dualists also include moreabstract mental states and events such as hopes, desires, and beliefs

writ-The hypothesized mental substances are proposed to differ from physicalones in their fundamental properties For example, all ordinary physical matter

From chapter 13 in Vision Science:Photons to Phenomenology (Cambridge, MA: MIT Press, 1999), 618–

630 Reprinted with permission.

has a well-defined position, occupies a particular volume, has a definite shape,and has a specific mass Conscious experiences, such as perceptions, remem-brances, beliefs, hopes, and desires, do not appear to have readily identifiablepositions, volumes, shapes, and masses In the case of vision, however, onemight object that visual experiences do have physical locations and extensions.There is an important sense in which my perception of a red ball on the table islocated on the table where the ball is and is extended over the spherical volumeoccupied by the ball What could be more obvious? But a substance dualistwould counter that these are properties of the physical object that I perceiverather than properties of my perceptual experience itself The experience is in

my mind rather than out there in the physical environment, and the location,extension, and mass of these mental entities are difficult to define—unless onemakes the problematic move of simply identifying them with the location, ex-tension, and mass of my brain Substance dualists reject this possibility, believ-ing instead that mental states, such as perceptions, beliefs, and desires, aresimply undefined with respect to position, extension, and mass In this case,

it makes sense to distinguish mental substances from physical ones on thegrounds that they have fundamentally different properties

We can also lookat the issue of fundamental properties the other wayaround: Do experiences have any properties that ordinary physical matter doesnot? Two possibilities merit consideration One is that experiences are subjectivephenomena in the sense that they cannot be observed by anyone but the personhaving them Ordinary matter and events, in contrast, are objective phenomenabecause they can be observed by anyone, at least in principle The other is thatexperiences have what philosophers call intentionality: They inherently refer tothings other than themselves.1Your experience of a bookin front of you rightnow is about the bookin the external world even though it arises from activity

in your brain This directedness of visual experiences is the source of the sion we mentioned in the previous paragraph about whether your perceptionshave location, extension, and so forth The physical objects to which such per-ceptual experiences refer have these physical properties, but the experiencesthemselves do not Intentionality does not seem to be a property that is shared

confu-by ordinary matter, and if this is true, it provides further evidence that scious experience is fundamentally different

con-It is possible to maintain a dualistic position and yet deny the existence ofany separate mental substances, however One can instead postulate that thebrain has certain unique properties that constitute its mental phenomena Theseproperties are just the sorts of experiences we have as we go about our every-day lives, including perceptions, pains, desires, and thoughts This philosophi-cal position on the mind-body problems is called property dualism It is a form

of dualism because these properties are taken to be nonphysical in the sense ofnot being reducible to any standard physical properties It is as though thephysical brain contains some strange nonphysical features or dimensions thatare qualitatively distinct from all physical features or dimensions

These mental features or dimensions are usually claimed to be emergent erties: attributes that simply do not arise in ordinary matter unless it reaches acertain level or type of complexity This complexity is certainly achieved in thehuman brain and may also be achieved in the brains of certain other animals

prop-4 Stephen E Palmer

The situation is perhaps best understood by analogy to the emergent property

of being alive Ordinary matter manifests this property only when it is nized in such a way that it is able to replicate itself and carry on the requiredbiological processes The difference, of course, is that being alive is a propertythat we can now explain in terms of purely physical processes Property dual-ists believe that this will never be the case for mental properties

orga-Even if one accepts a dualistic position that the mental and physical aresomehow qualitatively distinct, there are several different relations they mighthave to one another These differences form the basis for several varieties ofdualism One critical issue is the direction of causation: Does it run from mind

to brain, from brain to mind, or both? Descartes’s position was that both sorts

of causation are in effect: events in the brain can affect mental events, andmental events can also affect events in the brain This position is often calledinteractionism because it claims that the mental and physical worlds can interactcausally with each other in both directions It seems sensible enough at an in-tuitive level No self-respecting dualist doubts the overwhelming evidence thatphysical events in the brain cause the mental events of conscious experience.The pain that you feel in your toe, for example, is actually caused by the firing

of neurons in your brain Convincing evidence of this is provided by so-calledphantom limb pain, in which amputees feel pain—sometimes excruciating pain—

in their missing limbs (Chronholm, 1951; Ramachandran, 1996)

In the other direction, the evidence that mental events can cause physicalones is decidedly more impressionistic but intuitively satisfying to most inter-actionists They point to the fact that certain mental events, such as my havingthe intention of raising my arm, appear to cause corresponding physicalevents, such as the raising of my arm—provided I am not paralyzed and myarm is not restrained in any way The nature of this causation is scientificallyproblematic, however, because all currently known forms of causation concernphysical events causing other physical events Even so, other forms of causationthat have not yet been identified may nevertheless exist

Not all dualists are interactionists, however An important alternative sion of dualism, called epiphenomenalism, recognizes mental entities as being dif-ferent in kind from physical ones yet denies that mental states play any causalrole in the unfolding of physical events An epiphenomenalist would argue thatmental states, such as perceptions, intentions, beliefs, hopes, and desires, aremerely ineffectual side effects of the underlying causal neural events that takeplace in our brains To get a clearer idea of what this might mean, consider thefollowing analogy: Imagine that neurons glow slightly as they fire in a brainand that this glowing is somehow akin to conscious experiences The pattern

ver-of glowing in and around the brain (i.e., the conscious experience) is clearlycaused by the firing of neurons in the brain Nobody would question that Butthe neural glow would be causally ineffectual in the sense that it would notcause neurons to fire any differently than they would if they did not glow.Therefore, causation runs in only one direction, from physical to mental, in anepiphenomenalist account of the mind-body problem Although this positiondenies any causal efficacy to mental events, it is still a form of dualism because

it accepts the existence of the ‘‘glow’’ of consciousness and maintains that it isqualitatively distinct from the neural firings themselves

Idealism Not all philosophical positions on the mind-body problem are istic The opposing view is monism: the idea that there is really just one sort

dual-of stuff after all Not surprisingly, there are two sorts dual-of monist positions—idealism and materialism—one for each kind of stuff there might be A monistwho believes there to be no physical world, but only mental events, is called anidealist (from the ‘‘ideas’’ that populate the mental world) This has not been avery popular position in the history of philosophy, having been championedmainly by the British philosopher Bishop Berkeley

The most significant problem for idealism is how to explain the commonality

of different people’s perceptions of the same physical events If a fire engineraces down the street with siren blaring and red lights flashing, everyone lookstoward it, and they all see and hear pretty much the same physical events, al-beit from different vantage points How is this possible if there is no physicalworld that is responsible for their simultaneous perceptions of the sound andsight of the fire engine? One would have to propose some way in which theminds of the various witnesses happen to be hallucinating exactly correspond-ing events at exactly corresponding times Berkeley’s answer was that God wasresponsible for this grand coordination, but such claims have held little sway inmodern scientific circles Without a cogent scientific explanation of the com-monality of shared experiences of the physical world, idealism has largely be-come an historical curiosity with no significant modern following

Materialism The vast majority of monists believe that only physical entitiesexist They are called materialists In contrast to idealism, materialism is a verycommon view among modern philosophers and scientists There are actuallytwo distinct forms of materialism, which depend on what their adherentsbelieve the ultimate status of mental entities will be once their true physicalnature is discovered One form, called reductive materialism, posits that mentalevents will ultimately be reduced to material events in much the same way thatother successful reductions have occurred in science (e.g., Armstrong, 1968).This view is also called mind-brain identity theory because it assumes that mentalevents are actually equivalent to brain events and can be talked about more orless interchangeably, albeit with different levels of precision

A good scientific example of what reductive materialists believe will occurwhen the mental is reduced to the physical is the reduction in physics of ther-modynamic concepts concerning heat to statistical mechanics The temperature

of a gas in classical thermodynamics has been shown to be equivalent to theaverage kinetic energy of its molecules in statistical mechanics, thus replacingthe qualitatively distinct thermodynamic concept of heat with the more generaland basic concept of molecular motion The concept of heat did not then dis-appear from scientific vocabulary: it remains a valid concept within manycontexts Rather, it was merely given a more accurate definition in terms ofmolecular motion at a more microscopic level of analysis According to reduc-tive materialists, then, mental concepts will ultimately be redefined in terms

of brain states and events, but their equivalence will allow mental concepts

to remain valid and scientifically useful even after their brain correlates arediscovered For example, it will still be valid to say, ‘‘John is hungry,’’ ratherthan, ‘‘Such-and-such pattern of neural firing is occurring in John’s lateralhypothalamus.’’

6 Stephen E Palmer

The other materialist position, called eliminative materialism, posits that atleast some of our current concepts concerning mental states and events willeventually be eliminated from scientific vocabulary because they will be found

to be simply invalid (e.g., Churchland, 1990) The scenario eliminative alists envision is thus more radical than the simple translation scheme we justdescribed for reductive materialism Eliminative materialists believe that some

materi-of our present concepts about mental entities (perhaps including perceptualexperiences as well as beliefs, hopes, desires, and so forth) are so fundamen-tally flawed that they will someday be entirely replaced by a scientificallyaccurate account that is expressed in terms of the underlying neural events

An appropriate analogy here would be the elimination of the now-discreditedideas of ‘‘vitalism’’ in biology: the view that what distinguishes living fromnonliving things is the presence of a mysterious and qualitatively distinct force

or substance that is present in living objects and absent in nonliving ones Thediscovery of the biochemical reactions that cause the replication of DNA bycompletely normal physical means ultimately undercut any need for such mys-tical concepts, and so they were banished from scientific discussion, never to beseen again

In the same spirit, eliminative materialists believe that some mental concepts,such as perceiving, thinking, desiring, and believing, will eventually be sup-planted by discussion of the precise neurological events that underlie them.Scientists would then speakexclusively of the characteristic pattern of neuralfirings in the appropriate nuclei of the lateral hypothalamus and leave all talkabout ‘‘being hungry’’ or ‘‘the desire to eat’’ to historians of science who studyarchaic and discredited curiosities of yesteryear Even the general public wouldeventually come to thinkand talkin terms of these neuroscientific explanationsfor experiences, much as modern popular culture has begun to assimilate cer-tain notions about DNA replication, gene splicing, cloning, and related con-cepts into movies, advertising, and language

Behaviorism Another position on the mind-body problem is philosophical haviorism: the view that the proper way to talkabout mental events is in terms

be-of the overt, observable movements (behaviors) in which an organism engages.Because objective behaviors are measurable, quantifiable aspects of the physicalworld, behaviorism is, strictly speaking, a kind of materialism It provides such

a different perspective, however, that it is best thought of as a distinct view.Behaviorists differ markedly from standard materialists in that they seek toreduce mental events to behavioral events or dispositions rather than to neu-rophysiological events They shun neural explanations not because they dis-believe in the causal efficacy of neural events, but because they believe thatbehavior offers a higher and more appropriate level of analysis The radicalbehaviorist movement pressed for nothing less than redefining the scientificstudy of mind as the scientific study of behavior And for many years, theysucceeded in changing the agenda of psychology

The behaviorist movement began with the writings of psychologist JohnWatson (1913), who advocated a thoroughgoing purge of everything mental frompsychology He reasoned that what made intellectual inquiries scientific ratherthan humanistic or literary was that the empirical data and theoretical con-structs on which they rest are objective In the case of empirical observations,

objectivity means that, given a description of what was done in a ular experiment, any scientist could repeat it and obtain essentially the sameresults, at least within the limits of measurement error By this criterion, intro-spective studies of the qualities of perceptual experience were unscientific be-cause they were not objective Two different people could perform the sameexperiment (using themselves as subjects, of course) and report different expe-riences When this happened—and it did—there was no way to resolve dis-putes about who was right Both could defend their own positions simply byappealing to their private and privileged knowledge of their own inner states.This move protected their claims but blocked meaningful scientific debate.According to behaviorists, scientists should study the behavior of organisms

partic-in a well-defined tasksituation For example, rather than partic-introspect about thenature of the perception of length, behaviorists would perform an experiment.Observers could be asked to discriminate which of two lines was longer, andtheir performance could be measured in terms of percentages of correct andincorrect responses for each pair of lines Such an objective, behaviorally de-fined experiment could easily be repeated in any laboratory with different sub-jects to verify the accuracy and generality of its results Watson’s promotion ofobjective, behaviorally defined experimental methods—called methodologicalbehaviorism—was a great success and strongly shaped the future of psycho-logical research

Of more relevance to the philosophical issue of the relation between mindand body, however, were the implications of the behaviorist push for objectiv-ity in theoretical constructs concerning the mind It effectively ruled out refer-ences to mental states and processes, replacing them with statements about anorganism’s propensity to engage in certain behaviors under certain conditions.This position is often called theoretical behaviorism or philosophical behavior-ism Instead of saying, ‘‘John is hungry,’’ for example, which openly refers to

a conscious mental experience (hunger) with which everyone is presumablyfamiliar, a theoretical behaviorist would say something like ‘‘John has a pro-pensity to engage in eating behavior in the presence of food.’’ This propensitycan be measured in a variety of objective ways—such as the amount of a cer-tain food eaten when it was available after a certain number of hours since thelast previous meal—precisely because it is about observable behavior

But the behaviorist attempt to avoid talking about conscious experience runsinto trouble when one considers all the conditions in which John might fail toengage in eating behavior even though he was hungry and food was readilyavailable Perhaps he could not see the food, for example, or maybe he wasfasting He might even have believed that the food was poisoned It might seemthat such conditions could be blocked simply by inserting appropriate provi-sions into the behavioral statement, such as ‘‘John had a propensity to engage

in eating behavior in the presence of food, provided he perceived it, was notfasting, and did not believe it was poisoned.’’ This move ultimately fails, how-ever, for at least two reasons:

1 Inability to enumerate all conditionals Once one begins to thinkof ditions that would have to be added to statements about behavioral dis-positions, it quickly becomes apparent that there are indefinitely many

con-8 Stephen E Palmer

Perhaps John fails to eat because his hands are temporarily paralyzed,because he has been influenced by a hypnotic suggestion, or whatever.This problem undercuts the claim that behavioral analyses of mentalstates are elegant and insightful, suggesting instead that they are fatallyflawed or at least on the wrong track.

2 Inability to eliminate mental entities The other problem is that the ditionals that must be enumerated frequently make reference to just thesorts of mental events that are supposed to be avoided For example,whether John sees the food or not, whether he intends to fast, and what hebelieves about its being poisoned are all mentalistic concepts that have nowbeen introduced into the supposedly behavioral definition The amendedversion is therefore unacceptable to a strict theoretical behaviorist

con-For such reasons, theoretical behaviorism ultimately failed The problem, in anutshell, was that behaviorists mistookthe epistemic status of mental states(how we come to know about mental states in other people) for the ontologicalstatus of mental states (what their inherent nature is) (Searle, 1992) That is, wesurely come to know about other people’s mental states through their behavior,but this does not mean that the nature of these mental states is inherentlybehavioral

Functionalism Functionalism was a movement in the philosophy of mind thatbegan in the 1960s in close association with the earliest stirrings of cognitivescience (e.g., Putnam, 1960) Its main idea is that a given mental state can bedefined in terms of the causal relations that exist among that mental state,environmental conditions (inputs), organismic behaviors (outputs), and othermental states Note that this is very much like behaviorism, but with the im-portant addition of allowing other mental states into the picture This additionenables a functionalist definition of hunger, for example, to refer to a variety

of other mental states, such as perceptions, intentions, and beliefs, as gested above Functionalists are not trying to explain away mental phenomena

sug-as actually being propensities to behave in certain ways, sug-as behaviorists did.Rather, they are trying to define mental states in terms of their relations toother mental states as well as to input stimuli and output behaviors The picturethat emerges is very much like information processing analyses This is notsurprising because functionalism is the philosophical foundation of moderncomputational theories of mind

Functionalists aspired to more than just the overthrow of theoretical iorism, however They also attempted to blockreductive materialism by sug-gesting new criticisms of mind-brain identity theory The basis of this criticismlies in the notion of multiple realizability: the fact that many different physicaldevices can serve the same function, provided they causally connect inputs andoutputs in the same way via internal states (Putnam, 1967) For example, thereare many different ways of building a thermostat They all have the samefunction—to control the temperature in the thermostat’s environment—butthey realize it through very different physical implementations

behav-Multiple realizability poses the following challenge to identity theory pose there were creatures from some other galaxy whose biology was based

Sup-on silicSup-on molecules rather than Sup-on carbSup-on molecules, as ours is Let us also

suppose that they were alive (even though the basis of their life was not DNA,but some functionally similar self-replicating molecule) and that they even looklike people And suppose further not only that their brains were constructed ofelements that are functionally similar to neurons, but also that these elementswere interconnected in just the way that neurons in our brains are Indeed,their brains would be functionally isomorphic to ours, even though they weremade of physically different stuff.

Functionalists then claim that these alien creatures would have the samemental states as we do—that is, the same perceptions, pains, desires, beliefs,and so on that populate our own conscious mental lives—provided that theirinternal states were analogously related to each other, to the external world,and to their behavior This same approach can be generalized to argue for thepossibility that computers and robots of the appropriate sort would also beconscious Suppose, for example, that each neuron in a brain was replaced with

a microcomputer chip that exactly simulated its firing patterns in response toall the neuron chips that provide its input The computer that was thus con-structed would fulfill the functionalist requirements for having the same mentalstates as the person whose brain was ‘‘electronically cloned.’’ You should de-cide for yourself whether you believe that such a computer would actuallyhave mental states or would merely act as though it had mental states Onceyou have done so, try to figure out what criteria you used to decide (For twocontradictory philosophical views of this thought experiment, the reader is re-ferred to Dennett (1991) and Searle (1993).)

Multiple realizability is closely related to differences between the algorithmicand implementation levels The algorithmic level corresponds roughly to thefunctional description of the organism in terms of the relations among its in-ternal states, its input information, and its output behavior The implementa-tion level corresponds to its actual physical construction The functionalistnotion of multiple realizability thus implies that there could be many differentkinds of creatures that would have the same mental states as people do, at leastdefined in this way If true, this would undercut identity theory, since mentalevents could not then be simply equated with particular neurological events;they would have to be equated with some more general class of physical eventsthat would include, among others, silicon-based aliens and electronic brains.The argument from multiple realizability is crucial to the functionalist theory

of mind Before we get carried away with the implications of multiple ability, though, we must askourselves whether it is true or even remotely likely

realiz-to be true There is not much point in basing our understanding of ness on a functionalist foundation unless that foundation is well grounded Isit? More important, how would we know if it were? We will address this topicshortly when we consider the problem of other minds

conscious-Supervenience There is certainly some logical relation between brain activityand mental states such as consciousness, but precisely what it is has obviouslybeen difficult to determine Philosophers of mind have spent hundreds of yearstrying to figure out what it is and have spilled oceans of inkattacking anddefending different positions Recently, however, philosopher Jaegwon Kim(1978, 1993) has formulated a position with which most philosophers of mind

10 Stephen E Palmer

have been able to agree This relation, called supervenience, is that any difference

in conscious events requires some corresponding difference in underlying ral activity In other words, mental events supervene on neural events because

neu-no two possible situations can be identical with respect to their neural ties while differing in their mental properties It is a surprisingly weakrelation,but it is better than nothing

proper-Supervenience does not imply that all differences in underlying neural ity result in differences in consciousness Many neural events are entirely out-side awareness, including those that control basic bodily functions such asmaintaining gravitational balance and regulating heartbeat But supervenienceclaims that no changes in consciousness can take place without some change

activ-in neural activity The real trick, of course, is sayactiv-ing precisely what kactiv-inds ofchanges in neural events produce what kinds of changes in awareness

1.1.2 The Problem of Other Minds

The functionalist arguments about multiple realizability are merely thoughtexperiments because neither aliens nor electronic brains are currently at hand.Even so, the question of whether or not someone or something is conscious iscentral to the enterprise of cognitive science because the validity of such argu-ments rests on the answer Formulating adequate criteria for consciousness isone of the thorniest problems in all of science How could one possibly decide?Asking how to discriminate conscious from nonconscious beings brings usface to face with another classic topic in the philosophy of mind: the problem

of other minds The issue at stake is how I know whether another creature (ormachine) has conscious experiences Notice that I did not say ‘‘how we knowwhether another creature has conscious experiences,’’ because, strictly speak-ing, I do not know whether you do or not This is because one of the most pe-culiar and unique features of my consciousness is its internal, private nature:Only I have direct access to my conscious experiences, and I have direct accessonly to my own As a result, my beliefs that other people also have consciousexperiences—and your belief that I do—appear to be inferences Similarly, Imay believe that dogs and cats, or even frogs and worms, are conscious But inevery case, the epistemological basis of my belief about the consciousness ofother creatures is fundamentally different from knowledge of my own con-sciousness: I have direct access to my own experience and nobody else’s.Criteria for Consciousness If our beliefs that other people—and perhaps manyanimals as well—have experiences lik e ours are inferences, on what might suchinferences be based? There seem to be at least two criteria

1 Behavioral similarity Other people act in ways that are roughly similar

to my own actions when I am having conscious experiences When I perience pain on stubbing my toe, for example, I may wince, say ‘‘Ouch!’’and hold my toe while hopping on my other foot When other people dosimilar things under similar circumstances, I presume they are experienc-ing a feeling closely akin to my own pain Dogs also behave in seeminglyanalogous ways in what appear to be analogous situations in which theymight experience pain, and so I also attribute this mental state of being inpain to them The case is less compelling for creatures like frogs and

ex-worms because their behavior is less obviously analogous to our own, butmany people firmly believe that their behavior indicates that they alsohave conscious experiences such as pain.

2 Physical similarity Other people—and, to a lesser degree, various otherspecies of animals—are similar to me in their basic biological and physicalstructure Although no two people are exactly the same, humans are gen-erally quite similar to each other in terms of their essential biological con-stituents We are all made of the same kind of flesh, blood, bone, and soforth, and we have roughly the same kinds of sensory organs Many otheranimals also appear to be made of similar stuff, although they are mor-phologically different to varying degrees Such similarities and differencesmay enter into our judgments of the likelihood that other creatures alsohave conscious experiences

Neither condition alone is sufficient for a convincing belief in the reality ofmental states in another creature Behavioral similarity alone is insufficient be-cause of the logical possibility of automatons: robots that are able to simulateevery aspect of human behavior but have no experiences whatsoever We maythinkthat such a machine acts as if it had conscious experiences, but it couldconceivably do so without actually having them (Some theorists reject thispossibility, however [e.g., Dennett, 1991].) Physical similarity alone is insuffi-cient because we do not believe that even another living person is having con-scious experiences when they are comatose or in a dreamless sleep Only thetwo together are convincing Even when both are present to a high degree,

I still have no guarantee that such an inference is warranted I only know that

I myself have conscious experiences

But what then is the status of the functionalist argument that an aliencreature based on silicon rather than carbon molecules would have mentalstates like ours? This thought experiment is perhaps more convincing than theelectronic-brained automaton because we have presumed that the alien is atleast alive, albeit using some other physical mechanism to achieve this state ofbeing But logically, it would surely be unprovable that such silicon peoplewould have mental states like ours, even if they acted very much the same andappeared very similar to people In fact, the argument for functionalism frommultiple realizability is no stronger than our intuitions that such creatureswould be conscious The strength of such intuitions can (and does) vary widelyfrom one person to another

The Inverted Spectrum Argument We have gotten rather far afield from visualperception in all this talkof robots, aliens, dogs, and worms having pains, butthe same kinds of issues arise for perception One of the classic arguments re-lated to the problem of other minds—called the inverted spectrum argument—concerns the perceptual experience of color (Locke, 1690/1987) It goes like this:Suppose you grant that I have visual awareness in some form that includesdifferentiated experiences in response to different physical spectra of light (i.e.,differentiated color perceptions) How can we know whether my color experi-ences are the same as yours?

The inverted spectrum argument refers to the possibility that my color riences are exactly like your own, except for being spectrally inverted In its

expe-12 Stephen E Palmer

literal form, the inversion refers to reversing the mapping between color riences and the physical spectrum of wavelengths of light, as though the rain-bow had simply been reversed, red for violet (and vice versa) with everything

expe-in between beexpe-ing reversed expe-in like manner The claim of the expe-inverted spectrumargument is that no one would ever be able to tell that you and I have differentcolor experiences

This particular form of color transformation would not actually workas tended because of the shape of the color solid (Palmer, 1999) The color solid isasymmetrical in that the most saturated blues and violets are darker than themost saturated reds and greens, which, in turn, are darker than the most satu-rated yellows and oranges (see figure 1.1A) The problem this causes for theliteral inverted spectrum argument is that if my hues were simply reversed,your experience of yellow would be the same as my experience of blue-green,and so you would judge yellow to be darker than blue-green, whereas I would

in-do the reverse This difference would allow the spectral inversion of my colorexperiences (relative to yours) to be detected

This problem may be overcome by using more sophisticated versions ofthe same color transformation argument (Palmer, 1999) The most plausible is

Figure 1.1

Sophisticated versions of the inverted spectrum argument Transformations of the normal color solid (A) that would not be detectable by behavioral methods include (B) red-green reversal, which reflects each color about the blue-yellow-black-white place; (C) the complementary transformation, which reflects each color through the central point; and (D) blue-yellow and black-white reversal, which is the combination of both the two other transformations (B and C) (After Palmer, 1999.)

red-green reversal, in which my color space is the same as yours except for flection about the blue-yellow plane, thus reversing reds and greens (see figure1.1B) It does not suffer from problems concerning the differential lightness

re-of blues and yellows because my blues correspond to your blues and myyellows to your yellows Our particular shades of blues and yellows would bedifferent—my greenish yellows and greenish blues would correspond to yourreddish yellows (oranges) and reddish blues (purples), respectively, and viceversa—but gross differences in lightness would not be a problem

There are other candidates for behaviorally undetectable color tions as well (see figures 1.1C and 1.1D) The crucial idea in all these versions ofthe inverted spectrum argument is that if the color solid were symmetric withrespect to some transformation—and this is at least roughly true for the threecases illustrated in figures 1.1B–1.1D—there would be no way to tell the dif-ference between my color experiences and yours simply from our behavior Ineach case, I would name colors in just the same way as you would, becausethese names are only mediated by our own private experiences of color It is thesameness of the physical spectra that ultimately causes them to be named con-sistently across people, not the sameness of the private experiences I wouldalso describe relations between colors in the same way as you would: that focalblue is darker than focal yellow, that lime green is yellower than emeraldgreen, and so forth In fact, if I were in a psychological experiment in which mytaskwas to rate pairs of color for similarity or dissimilarity, I would make thesame ratings you would I would even pickout the same unique hues as youwould—the ‘‘pure’’ shades of red, green, blue, and yellow—even though myinternal experiences of them would be different from yours It would be ex-tremely difficult, if not impossible, to tell from my behavior with respect tocolor that I experience it differently than you do.2

transforma-I suggested that red-green reversal is the most plausible form of color formation because a good biological argument can be made that there should

trans-be some very small numtrans-ber of seemingly normal trichromats who should trans-bered-green reversed The argument for such pseudo-normal color perception goes

as follows (Nida-Ru¨melin, 1996) Normal trichromats have three different ments in their three cone types (figure 1.2A) Some people are red-green colorblind because they have a gene that causes their long-wavelength (L) cones tohave the same pigment as their medium-wavelength (M) cones (figure 1.2B).Other people have a different form of red-green color blindness because theyhave a different gene that causes their M cones to have the same pigment astheir L cones (figure 1.2C) In both cases, people with these genetic defects losethe ability to experience both red and green because the visual system codesboth colors by taking the difference between the outputs of these two conetypes But suppose that someone had the genes for both of these forms of red-green color blindness Their L cones would have the M pigment, and their Mcones would have the L pigment (figure 1.2D) Such doubly color blind indi-viduals would therefore not be red-green color blind at all, but red-green-reversed trichromats.3 Statistically, they should be very rare (about 14 per10,000 males), but they should exist If they do, they are living proof that thiscolor transformation is either undetectable or very difficult to detect by purelybehavioral means, because nobody has ever detected one!

pig-14 Stephen E Palmer

These color transformation arguments are telling criticisms against the pleteness of any definition of conscious experience based purely on behavior.Their force lies in the fact that there could be identical behavior in response toidentical environmental stimulation without there being corresponding identi-cal experiences underlying them, even if we grant that the other person hasexperiences to begin with.

com-Phenomenological Criteria Let us return to the issue of criteria for ness: How are we to tell whether a given creature is conscious or not? Clearly,phenomenological experience is key In fact, it is the defining characteristic, thenecessary and sufficient condition, for attributing consciousness to something

conscious-I know that conscious-I am conscious precisely because conscious-I have such experiences This

is often called first-person knowledge or subjective knowledge because it is able only to the self (i.e., the first-person or subject) In his classic essay

avail-‘‘What Is It Like to Be a Bat?’’ philosopher Thomas Nagel (1974) identifies the

Figure 1.2

A biological basis for red-green-reversed trichromats Normal trichromats have three different ments in the retinal cones (A), whereas red-green color blind individuals have the same pigment in their L and M cones (B and C) People with the genes for both forms of red-green color blindness, however, would be red-green-reversed trichromats (D).

pig-phenomenological position with what it is like to be some person, creature, ormachine in a given situation In the case of color perception, for example, it

is what it is like for you to experience a particular shade of redness or paleblueness or whatever This much seems perfectly clear But if it is so clear, thenwhy not simply define consciousness with respect to such phenomenologicalcriteria?

As we said before, the difficulty is that first-person knowledge is availableonly to the self This raises a problem for scientific explanations of conscious-ness because the scientific method requires its facts to be objective in the sense

of being available to any scientist who undertakes the same experiment In allmatters except consciousness, this appears to workvery well But conscious-ness has the extremely peculiar and elusive property of being directly accessi-ble only to the self, thus blocking the usual methods of scientific observation.Rather than observing consciousness itself in others, the scientist is forced toobserve the correlates of consciousness, the ‘‘shadows of consciousness,’’ as itwere Two sorts of shadows are possible to study: behavior and physiology.Neither is consciousness itself, but both are (or seem likely to be) closelyrelated

Behavioral Criteria The most obvious way to get an objective, scientific handle

on consciousness is to study behavior, as dictated by methodological iorism Behavior is clearly objective and observable in the third-person sense.But how is it related to consciousness? The linkis the assumption that if some-one or something behaves enough like I do, it must be conscious like I am.After all, I believe I behave in the ways I do because of my own consciousexperiences, and so (presumably) do others I wince when I am in pain, eatwhen I am hungry, and duckwhen I perceive a baseball hurtling toward myhead If I were comatose, I would not behave in any of these ways, even in thesame physical situations

behav-Behavioral criteria for consciousness are closely associated with what iscalled Turing’s test This test was initially proposed by the brilliant mathemati-cian Alan Turing (1950), inventor of the digital computer, to solve the problem

of how to determine whether a computing machine could be called gent.’’ Wishing to avoid purely philosophical debates, Turing imagined an ob-jective behavioral procedure for deciding the issue by setting up an imitationgame A person is seated at a computer terminal that allows her to communicateeither with a real person or with a computer that has been programmed tobehave intelligently (i.e., like a person) This interrogator’s job is to decidewhether she is communicating with a person or the computer The terminal isused simply to keep the interrogator from using physical appearance as a factor

‘‘intelli-in the decision, s‘‘intelli-ince appearance presumably does not have any logical bear‘‘intelli-ing

on intelligence

The interrogator is allowed to askanything she wants For example, shecould askthe subject to play a game of chess, engage in a conversation on cur-rent events, or describe its favorite TV show Nothing is out of bounds Shecould even askwhether the subject is intelligent A person would presumablyreply affirmatively, but then so would a properly programmed computer If theinterrogator could not tell the difference between interacting with real people

16 Stephen E Palmer

and with the computer, Turing asserted that the computer should be judged

‘‘intelligent.’’ It would then be said to have ‘‘passed Turing’s test.’’

Note that Turing’s test is a strictly behavioral test because the interrogatorhas no information about the physical attributes of the subject, but only aboutits behavior In the original version, this behavior is strictly verbal, but there is

no reason in principle why it needs to be restricted in this way The tor could askthe subject to draw pictures or even to carry out tasks in the realworld, provided the visual feedbackthe interrogator received did not provideinformation about the physical appearance of the subject

interroga-The same imitation game can be used for deciding about the appropriateness

of any other cognitive description, including whether the subject is ‘‘conscious.’’Again, simply asking the subject whether it is conscious will not discriminatebetween the machine and a person because the machine can easily be pro-grammed to answer that question in the affirmative Similarly, appropriate re-sponses to questions asking it to describe the nature of its visual experiences orpain experiences could certainly be programmed But even if they could, wouldthat necessarily mean that the computer would be conscious or only that itwould act as if it were conscious?

If one grants that physical appearance should be irrelevant to whethersomething is conscious or not, Turing’s test seems to be a fair and objectiveprocedure But it also seems that there is a fact at issue here rather than just anopinion—namely, whether the target object is actually conscious or merely sim-ulating consciousness—and Turing’s test should stand or fall on whether itgives the correct answer The problem is that it is not clear that it will As criticsreadily point out, it cannot distinguish between a conscious entity and one thatonly acts as if it were conscious—an automaton or a zombie To assert thatTuring’s test actually gives the correct answer to the factual question of con-sciousness, one must assume that it is impossible for something to act as if it isconscious without actually being so This is a highly questionable assumption,although some have defended it (e.g., Dennett, 1991) If it is untrue, then pass-ing Turing’s test is not a sufficient condition for consciousness, because autom-atons can pass it without being conscious

Turing’s test also runs into trouble as a necessary condition for ness The relevant question here is whether something can be conscious andstill fail Turing’s test Although this might initially seem unlikely, consider aperson who has an unusual medical condition that disables the use of all themuscles required for overt behavior yet keeps all other bodily functions intact,including all brain functions This person would be unable to behave in anyway yet would still be fully conscious when awake Turing’s test thus runsafoul as a criterion for consciousness because behavior’s linkto consciousnesscan be broken under unlikely but easily imaginable circumstances

conscious-We appear to be on the horns of a dilemma with respect to the criteria forconsciousness Phenomenological criteria are valid by definition but do not ap-pear to be scientific by the usual yardsticks Behavioral criteria are scientific bydefinition but are not necessarily valid The fact that scientists prefer to rely onrespectable but possibly invalid behavioral methods brings to mind the street-light parable: A woman comes upon a man searching for something under astreetlight at night The man explains that he has lost his keys, and they both

search diligently for some time The woman finally asks the man where hethinks he lost them, to which he replies, ‘‘Down the street in the middle of theblock.’’ When she then asks why he is looking here at the corner, he replies,

‘‘Because this is where the light is.’’ The problem is that consciousness does notseem to be where behavioral science can shed much light on it

Physiological Criteria Modern science has another card to play, however, andthat is the biological substrate of consciousness Even if behavioral methodscannot penetrate the subjectivity barrier of consciousness, perhaps physiologi-cal methods can In truth, few important facts are yet known about the bio-logical substrates of consciousness There are not even very many hypotheses,although several speculations have recently been proposed (e.g., Baars, 1988;Crick, 1994; Crick & Koch, 1990, 1995, 1998; Edelman, 1989) Even so, it is pos-sible to speculate about the promise such an enterprise might hold as a way ofdefining and theorizing about consciousness It is important to remember that

in doing so, we are whistling in the dark, however

Let us suppose, just for the sake of argument, that neuroscientists discoversome crucial feature of the neural activity that underlies consciousness Perhapsall neural activity that gives rise to consciousness occurs in some particularlayer of cerebral cortex, or in neural circuits that are mediated by some partic-ular neurotransmitter, or in neurons that fire at a temporal spiking frequency ofabout 40 times per second If something like one of these assertions were true—and, remember, we are just making up stories here—could we then defineconsciousness objectively in terms of that form of neural activity? If we could,would this definition then replace the subjective definition in terms of ex-perience? And would such a biological definition then constitute a theory ofconsciousness?

The first important observation about such an enterprise is that biology not really give us an objective definition of consciousness independent of itssubjective definition The reason is that we need the subjective definition todetermine what physiological events correspond to consciousness in the firstplace Suppose we knew all of the relevant biological events that occur in hu-man brains We still could not provide a biological account of consciousnessbecause we would have no way to tell which brain events were conscious andwhich ones were not Without that crucial information, a biological definition

can-of consciousness simply could not get can-off the ground To determine the logical correlates of consciousness, one must be able to designate the events

bio-to which they are being correlated (i.e., conscious ones), and this requires asubjective definition

For this reason, any biological definition of consciousness would always bederived from the subjective definition To see this in a slightly different way,consider what would constitute evidence that a given biological definition wasincorrect If brain activity of type C were thought to define consciousness, itcould be rejected for either of two reasons: if type C brain activity were found

to result in nonconscious processing of some sort or if consciousness werefound to occur in the absence of type C brain activity The crucial observationfor present purposes is that neither of these possibilities could be evaluatedwithout an independent subjective definition of consciousness

18 Stephen E Palmer

Correlational versus Causal Theories In considering the status of physiologicalstatements about consciousness, it is important to distinguish two differentsorts, which we will call correlational and causal Correlational statements con-cern what type of physiological activity takes place when conscious experiencesare occurring that fail to take place when they are not Our hypothetical ex-amples in terms of a specific cortical location, a particular neurotransmitter, or

a particular rate of firing are good examples The common feature of thesehypotheses is that they are merely correlational: They only claim that the des-ignated feature of brain activity is associated with consciousness; they don’texplain why that association exists In other words, they provide no causalanalysis of how this particular kind of brain activity produces consciousness.For this reason they fail to fill the explanatory gap that we mentioned earlier.Correlational analyses merely designate a subset of neural activity in the brainaccording to some particular property with which consciousness is thought to

be associated No explanation is given for this association; it simply is the sort

of activity that accompanies consciousness

At this point we should contrast such correlational analyses with a goodexample of a causal one: an analysis that provides a scientifically plausibleexplanation of how a particular form of brain activity actually causes consciousexperience Unfortunately, no examples of such a theory are available In fact,

to this writer’s knowledge, nobody has ever suggested a theory that the tific community regards as giving even a remotely plausible causal account ofhow consciousness arises or why it has the particular qualities it does Thisdoes not mean that such a theory is impossible in principle, but only that noserious candidate has been generated in the past several thousand years

scien-A related distinction between correlational and causal biological definitions

of consciousness is that they would differ in generalizability Correlational yses would very likely be specific to the type of biological system within whichthey had been discovered In the best-case scenario, a good correlational defi-nition of human consciousness might generalize to chimpanzees, possibly even

anal-to dogs or rats, but probably not anal-to frogs or snails because their brains aresimply too different If a correlational analysis showed that activity mediated

by a particular neurotransmitter was the seat of human consciousness, for ample, would that necessarily mean that creatures without that neurotrans-mitter were nonconscious? Or might some other evolutionarily related neuraltransmitter serve the same function in brains lacking that one? Even moredrastically, what about extraterrestrial beings whose whole physical make-upmight be radically different from our own? In such cases, a correlational analy-sis is almost bound to break down

ex-An adequate causal theory of consciousness might have a fighting chance,however, because the structure of the theory itself could provide the lines alongwhich generalization would flow Consider the analogy to a causal theory oflife based on the structure of DNA The analysis of how the double helicalstructure of DNA allows it to reproduce itself in an entirely mechanistic waysuggests that biologists could determine whether alien beings were alive in thesame sense as living organisms on earth by considering the nature of their mo-lecular basis and its functional ability to replicate itself and to support theorganism’s lifelike functions An alien object containing the very same set of

four component bases as DNA (adenine, guanine, thymine, and cytosine) insome very different global structure that did not allow self-replication wouldnot be judged to be alive by such biological criteria, yet another object contain-ing very different components in some analogous arrangement that allowed forself-replication might be Needless to say, such an analysis is a long way off inthe case of consciousness.

Notes

1 The reader is warned not to confuse intentionality with the concept of ‘‘intention’’ in ordinary language Your intentions have intentionality in the sense that they may refer to things other than themselves—for example, your intention to feed your cat refers to your cat, its food, and yourself—but no more so than other mental states you might have, such as beliefs, desires, per- ceptions, and pains The philosophical literature on the nature of intentionality is complex and extensive The interested reader is referred to Bechtel (1988) for an overview of this topic.

2 One might thinkthat if white and blackwere reversed, certain reflexive behaviors to light would somehow betray the difference This is not necessarily the case, however Whereas you would squint your eyes when you experienced intense brightness in response to bright sunlight, I would also squint my eyes in response to large amounts of sunlight The only difference is that

my experience of brightness under these conditions would be the same as your experience of darkness It sounds strange, but I believe it would all work out properly.

3 One could object that the only thing that differentiates M and L cones is the pigment that they contain, so people with both forms of red-green color blindness would actually be normal tri- chromats rather than red-green-reversed ones There are two other ways in which M and L cones might be differentiated, however First, if the connections of M and L cones to other cells of the visual system are not completely symmetrical, they can be differentiated by these connections independently of their pigments Second, they may be differentiable by their relation to the genetic codes that produced them.

References

Armstrong, D M (1968) A materialist theory of the mind London: Routledge & Kegan Paul Baars, B (1988) A cognitive theory of consciousness Cambridge, England: Cambridge University Press.

Churchland, P M (1990) Current eliminativism In W G Lycan (Ed.), Mind and cognition:A reader (pp 206–223) Oxford, England: Basil Blackwell.

Crick, F H C (1994) The astonishing hypothesis:The scientific search for the soul New York : Scribner Crick, F H C., & Koch, C (1990) Toward a neurobiological theory of consciousness Seminars in the Neurosciences, 2, 263–275.

Crick, F H C., & Koch, C (1995) Are we aware of neural activity in primary visual cortex? Nature,

375, 121–123.

Crick, F H C., & Koch, C (1998) Consciousness and neuroscience Cerebral cortex, 8, 97–107 Cronholm, B (1951) Phantom limbs in amputees Acta Psychiatrica Scandinavica, 72 (Suppl.) Dennett, D (1991) Consciousness Explained Boston: Little, Brown.

Edelman, G M (1989) The remembered present:A biological theory of consciousness New York : Basic Books.

Kim, J (1978) Supervenience and nomological incommensurables American Philosophical Quarterly,

15, 149–156.

Kim, J (1993) Supervenience and mind Cambridge, England: Cambridge University Press.

Locke, J (1690/1987) An essay concerning human understanding Oxford, England: Basil Blackwell Nagel, T (1974) What is it like to be a bat? The Philosophical Review, 83, 435–450.

Palmer, S E (1999) Color, consciousness, and the isomorphism constraint Behavioural and Brain Sciences, 22(6), 923–989.

Putnam, H (1960) Minds and machines In S Hook(Ed.), Dimensions of mind New York : Collier Books.

20 Stephen E Palmer

Putnam, H (1967) Psychological predicates In W Captain & D Merrill (Eds.), Art, mind, and gion (pp 35–48) Pittsburgh: University of Pittsburgh Press.

reli-Ramachandran, V S., Levi, L., Stone, L., Rogers-reli-Ramachandran, D., McKinney, R., Stalcup, M., Arcilla, G., Sweifler, R., Schatz, A., Flippin, A (1996) Illusions of body image: What they reveal about human nature In R R Llinas and P S Churchland (Eds.), The mind-brain con- tinuum:Sensory processes (pp 29–60) Cambridge, MA: MIT Press.

Searle, J R (1992) The rediscovery of mind Cambridge, MA: MIT Press.

Turing, S (1959) Alan M Turing Cambridge, England: W Heffer.

lib-Several years ago I was approached by Pentagon officials who asked me tovolunteer for a highly dangerous and secret mission In collaboration withNASA and Howard Hughes, the Department of Defense was spending billions

to develop a Supersonic Tunneling Underground Device, or STUD It was posed to tunnel through the earth’s core at great speed and deliver a speciallydesigned atomic warhead ‘‘right up the Red’s missile silos,’’ as one of the Pen-tagon brass put it

sup-The problem was that in an early test they had succeeded in lodging awarhead about a mile deep under Tulsa, Oklahoma, and they wanted me toretrieve it for them ‘‘Why me?’’ I asked Well, the mission involved some pio-neering applications of current brain research, and they had heard of my inter-est in brains and of course my Faustian curiosity and great courage and soforth Well, how could I refuse? The diffi culty that brought the Pentagon to

my door was that the device I’d been asked to recover was fiercely radioactive,

in a new way According to monitoring instruments, something about the ture of the device and its complex interactions with pockets of material deep inthe earth had produced radiation that could cause severe abnormalities in cer-tain tissues of the brain No way had been found to shield the brain from thesedeadly rays, which were apparently harmless to other tissues and organs of thebody So it had been decided that the person sent to recover the device shouldleave his brain behind It would be kept in a safe place where it could execute itsnormal control functions by elaborate radio links Would I submit to a surgicalprocedure that would completely remove my brain, which would then beplaced in a life-support system at the Manned Spacecraft Center in Houston?Each input and output pathway, as it was severed, would be restored by a pair

na-of microminiaturized radio transceivers, one attached precisely to the brain, theother to the nerve stumps in the empty cranium No information would be lost,all the connectivity would be preserved At first I was a bit reluctant Would itreally work? The Houston brain surgeons encouraged me ‘‘Think of it,’’ theysaid, ‘‘as a mere stretching of the nerves If your brain were just moved over an

From chapter 17 in Brainstorms (Cambridge, MA: MIT Press, 1978), 310–323 Reprinted with mission.

per-inch in your skull, that would not alter or impair your mind We’re simply ing to make the nerves indefinitely elastic by splicing radio links into them.’’

go-I was shown around the life-support lab in Houston and saw the sparklingnew vat in which my brain would be placed, were I to agree I met the largeand brilliant support team of neurologists, hematologists, biophysicists, andelectrical engineers, and after several days of discussions and demonstrations, Iagreed to give it a try I was subjected to an enormous array of blood tests,brain scans, experiments, interviews, and the like They took down my auto-biography at great length, recorded tedious lists of my beliefs, hopes, fears,and tastes They even listed my favorite stereo recordings and gave me a crashsession of psychoanalysis

The day for surgery arrived at last and of course I was anesthetized and member nothing of the operation itself When I came out of anesthesia, Iopened my eyes, looked around, and asked the inevitable, the traditional,the lamentably hackneyed post-operative question: ‘‘Where am I?’’ The nursesmiled down at me ‘‘You’re in Houston,’’ she said, and I reflected that this stillhad a good chance of being the truth one way or another She handed me amirror Sure enough, there were the tiny antennae poking up through theirtitanium ports cemented into my skull

re-‘‘I gather the operation was a success,’’ I said, re-‘‘I want to go see my brain.’’They led me (I was a bit dizzy and unsteady) down a long corridor and into thelife-support lab A cheer went up from the assembled support team, and Iresponded with what I hoped was a jaunty salute Still feeling lightheaded, Iwas helped over to the life-support vat I peered through the glass There,floating in what looked like ginger-ale, was undeniably a human brain, though

it was almost covered with printed circuit chips, plastic tubules, electrodes, andother paraphernalia ‘‘Is that mine?’’ I asked ‘‘Hit the output transmitter switchthere on the side of the vat and see for yourself,’’ the project director replied Imoved the switch to off, and immediately slumped, groggy and nauseated, intothe arms of the technicians, one of whom kindly restored the switch to its onposition While I recovered my equilibrium and composure, I thought to my-self: ‘‘Well, here I am, sitting on a folding chair, staring through a piece of plateglass at my own brain But wait,’’ I said to myself, ‘‘shouldn’t I havethought, ‘Here I am, suspended in a bubbling fluid, being stared at by my owneyes’?’’ I tried to think this latter thought I tried to project it into the tank, of-fering it hopefully to my brain, but I failed to carry off the exercise with anyconviction I tried again ‘‘Here am I, Daniel Dennett, suspended in a bubblingfluid, being stared at by my own eyes.’’ No, it just didn’t work Most puzzlingand confusing Being a philosopher of firm physicalist conviction, I believedunswervingly that the tokening of my thoughts was occurring somewhere in

my brain: yet, when I thought ‘‘Here I am,’’ where the thought occurred to mewas here, outside the vat, where I, Dennett, was standing staring at my brain

I tried and tried to think myself into the vat, but to no avail I tried to build

up to the task by doing mental exercises I thought to myself, ‘‘The sun is ing over there,’’ five times in rapid succession, each time mentally ostending adifferent place: in order, the sun-lit corner of the lab, the visible front lawn ofthe hospital, Houston, Mars, and Jupiter I found I had little difficulty in getting

shin-my ‘‘there’s’’ to hop all over the celestial map with their proper references I

could loft a ‘‘there’’ in an instant through the farthest reaches of space, and thenaim the next ‘‘there’’ with pinpoint accuracy at the upper left quadrant of afreckle on my arm Why was I having such trouble with ‘‘here’’? ‘‘Here inHouston’’ worked well enough, and so did ‘‘here in the lab,’’ and even ‘‘here inthis part of the lab,’’ but ‘‘here in the vat’’ always seemed merely an unmeantmental mouthing I tried closing my eyes while thinking it This seemed tohelp, but still I couldn’t manage to pull it off, except perhaps for a fleeting in-stant I couldn’t be sure The discovery that I couldn’t be sure was also unset-tling How did I know where I meant by ‘‘here’’ when I thought ‘‘here’’? Could Ithink I meant one place when in fact I meant another? I didn’t see how thatcould be admitted without untying the few bonds of intimacy between a per-son and his own mental life that had survived the onslaught of the brainscientists and philosophers, the physicalists and behaviorists Perhaps I wasincorrigible about where I meant when I said ‘‘here.’’ But in my present cir-cumstances it seemed that either I was doomed by sheer force of mental habit

to thinking systematically false indexical thoughts, or where a person is (andhence where his thoughts are tokened for purposes of semantic analysis) is notnecessarily where his brain, the physical seat of his soul, resides Nagged byconfusion, I attempted to orient myself by falling back on a favorite philoso-pher’s ploy I began naming things

‘‘Yorick,’’ I said aloud to my brain, ‘‘you are my brain The rest of my body,seated in this chair, I dub ‘Hamlet.’ ’’ So here we all are: Yorick’s my brain,Hamlet’s my body, and I am Dennett Now, where am I? And when I think

‘‘where am I?’’ where’s that thought tokened? Is it tokened in my brain,lounging about in the vat, or right here between my ears where it seems to

be tokened? Or nowhere? Its temporal coordinates give me no trouble; must itnot have spatial coordinates as well? I began making a list of the alternatives

1 Where Hamlet goes, there goes Dennett This principle was easily refuted byappeal to the familiar brain transplant thought-experiments so enjoyed by phi-losophers If Tom and Dick switch brains, Tom is the fellow with Dick’s formerbody—just ask him; he’ll claim to be Tom, and tell you the most intimatedetails of Tom’s autobiography It was clear enough, then, that my currentbody and I could part company, but not likely that I could be separated from

my brain The rule of thumb that emerged so plainly from the thought ments was that in a brain-transplant operation, one wanted to be the donor, notthe recipient Better to call such an operation a body-transplant, in fact So per-haps the truth was,

experi-2 Where Yorick goes, there goes Dennett This was not at all appealing, ever How could I be in the vat and not about to go anywhere, when I was soobviously outside the vat looking in and beginning to make guilty plans to re-turn to my room for a substantial lunch? This begged the question I realized,but it still seemed to be getting at something important Casting about for somesupport for my intuition, I hit upon a legalistic sort of argument that mighthave appealed to Locke

how-Suppose, I argued to myself, I were now to fly to California, rob a bank, and

be apprehended In which state would I be tried: In California, where the bery took place, or in Texas, where the brains of the outfit were located? Would

rob-I be a California felon with an out-of-state brain, or a Texas felon remotely

Where Am I? 25

controlling an accomplice of sorts in California? It seemed possible that I mightbeat such a rap just on the undecidability of that jurisdictional question, thoughperhaps it would be deemed an inter-state, and hence Federal, offense In anyevent, suppose I were convicted Was it likely that California would be satisfied

to throw Hamlet into the brig, knowing that Yorick was living the good life andluxuriously taking the waters in Texas? Would Texas incarcerate Yorick, leav-ing Hamlet free to take the next boat to Rio? This alternative appealed to me.Barring capital punishment or other cruel and unusual punishment, the statewould be obliged to maintain the life-support system for Yorick though theymight move him from Houston to Leavenworth, and aside from the unpleas-antness of the opprobrium, I, for one, would not mind at all and would con-sider myself a free man under those circumstances If the state has an interest

in forcibly relocating persons in institutions, it would fail to relocate me inany institution by locating Yorick there If this were true, it suggested a thirdalternative

3 Dennett is wherever he thinks he is Generalized, the claim was as follows:

At any given time a person has a point of view, and the location of the point ofview (which is determined internally by the content of the point of view) is alsothe location of the person

Such a proposition is not without its perplexities, but to me it seemed a step

in the right direction The only trouble was that it seemed to place one in aheads-I-win/tails-you-lose situation of unlikely infallibility as regards location.Hadn’t I myself often been wrong about where I was, and at least as often un-certain? Couldn’t one get lost? Of course, but getting lost geographically is notthe only way one might get lost If one were lost in the woods one could at-tempt to reassure oneself with the consolation that at least one knew whereone was: one was right here in the familiar surroundings of one’s own body.Perhaps in this case one would not have drawn one’s attention to much to bethankful for Still, there were worse plights imaginable, and I wasn’t sure Iwasn’t in such a plight right now

Point of view clearly had something to do with personal location, but it wasitself an unclear notion It was obvious that the content of one’s point of viewwas not the same as or determined by the content of one’s beliefs or thoughts.For example, what should we say about the point of view of the Cineramaviewer who shrieks and twists in his seat as the roller-coaster footage over-comes his psychic distancing? Has he forgotten that he is safely seated in thetheater? Here I was inclined to say that the person is experiencing an illusoryshift in point of view In other cases, my inclination to call such shifts illusorywas less strong The workers in laboratories and plants who handle dangerousmaterials by operating feedback-controlled mechanical arms and hands under-

go a shift in point of view that is crisper and more pronounced than thing Cinerama can provoke They can feel the heft and slipperiness of thecontainers they manipulate with their metal fingers They know perfectly wellwhere they are and are not fooled into false beliefs by the experience, yet it is as

any-if they were inside the isolation chamber they are peering into With mentaleffort, they can manage to shift their point of view back and forth, rather likemaking a transparent Neckar cube or an Escher drawing change orientation

before one’s eyes It does seem extravagant to suppose that in performing thisbit of mental gymnastics, they are transporting themselves back and forth.Still their example gave me hope If I was in fact in the vat in spite of myintuitions, I might be able to train myself to adopt that point of view even as amatter of habit I should dwell on images of myself comfortably floating in myvat, beaming volitions to that familiar body out there I reflected that the ease ordifficulty of this task was presumably independent of the truth about the loca-tion of one’s brain Had I been practicing before the operation, I might now befinding it second nature You might now yourself try such a tromp l’oeil Imag-ine you have written an inflammatory letter which has been published in theTimes, the result of which is that the Government has chosen to impound yourbrain for a probationary period of three years in its Dangerous Brain Clinic inBethesda, Maryland Your body of course is allowed freedom to earn a salaryand thus to continue its function of laying up income to be taxed At this mo-ment, however, your body is seated in an auditorium listening to a peculiaraccount by Daniel Dennett of his own similar experience Try it Think yourself

to Bethesda, and then hark back longingly to your body, far away, and yetseeming so near It is only with long-distance restraint (yours? the Govern-ment’s?) that you can control your impulse to get those hands clapping inpolite applause before navigating the old body to the rest room and a well-deserved glass of evening sherry in the lounge The task of imagination is cer-tainly difficult, but if you achieve your goal the results might be consoling.Anyway, there I was in Houston, lost in thought as one might say, but not forlong My speculations were soon interrupted by the Houston doctors, whowished to test out my new prosthetic nervous system before sending me off on

my hazardous mission As I mentioned before, I was a bit dizzy at first, andnot surprisingly, although I soon habituated myself to my new circumstances(which were, after all, well nigh indistinguishable from my old circumstances)

My accommodation was not perfect, however, and to this day I continue to beplagued by minor coordination difficulties The speed of light is fast, but finite,and as my brain and body move farther and farther apart, the delicate interac-tion of my feedback systems is thrown into disarray by the time lags Just asone is rendered close to speechless by a delayed or echoic hearing of one’sspeaking voice so, for instance, I am virtually unable to track a moving objectwith my eyes whenever my brain and my body are more than a few milesapart In most matters my impairment is scarcely detectable, though I can nolonger hit a slow curve ball with the authority of yore There are some com-pensations of course Though liquor tastes as good as ever, and warms mygullet while corroding my liver, I can drink it in any quantity I please, withoutbecoming the slightest bit inebriated, a curiosity some of my close friends mayhave noticed (though I occasionally have feigned inebriation, so as not to drawattention to my unusual circumstances) For similar reasons, I take aspirinorally for a sprained wrist, but if the pain persists I ask Houston to administercodeine to me in vitro In times of illness the phone bill can be staggering.But to return to my adventure At length, both the doctors and I were sat-isfied that I was ready to undertake my subterranean mission And so I left mybrain in Houston and headed by helicopter for Tulsa Well, in any case, that’s

Where Am I? 27

the way it seemed to me That’s how I would put it, just off the top of my head

as it were On the trip I reflected further about my earlier anxieties and decidedthat my first post-operative speculations had been tinged with panic The mat-ter was not nearly as strange or metaphysical as I had been supposing Wherewas I? In two places, clearly: both inside the vat and outside it Just as one canstand with one foot in Connecticut and the other in Rhode Island, I was in twoplaces at once I had become one of those scattered individuals we used to hear

so much about The more I considered this answer, the more obviously true itappeared But, strange to say, the more true it appeared, the less important thequestion to which it could be the true answer seemed A sad, but not unprece-dented, fate for a philosophical question to suffer This answer did not com-pletely satisfy me, of course There lingered some question to which I shouldhave liked an answer, which was neither ‘‘Where are all my various and sundryparts?’’ nor ‘‘What is my current point of view?’’ Or at least there seemed to besuch a question For it did seem undeniable that in some sense I and not merelymost of me was descending into the earth under Tulsa in search of an atomicwarhead

When I found the warhead, I was certainly glad I had left my brain behind,for the pointer on the specially built Geiger counter I had brought with me wasoff the dial I called Houston on my ordinary radio and told the operation con-trol center of my position and my progress In return, they gave me instructionsfor dismantling the vehicle, based upon my on-site observations I had set towork with my cutting torch when all of a sudden a terrible thing happened Iwent stone deaf At first I thought it was only my radio earphones that hadbroken, but when I tapped on my helmet, I heard nothing Apparently the au-ditory transceivers had gone on the fritz I could no longer hear Houston or myown voice, but I could speak, so I started telling them what had happened Inmid-sentence, I knew something else had gone wrong My vocal apparatus hadbecome paralyzed Then my right hand went limp—another transceiver hadgone I was truly in deep trouble But worse was to follow After a few moreminutes, I went blind I cursed my luck, and then I cursed the scientists whohad led me into this grave peril There I was, deaf, dumb, and blind, in a ra-dioactive hole more than a mile under Tulsa Then the last of my cerebral radiolinks broke, and suddenly I was faced with a new and even more shockingproblem: whereas an instant before I had been buried alive in Oklahoma, now Iwas disembodied in Houston My recognition of my new status was not im-mediate It took me several very anxious minutes before it dawned on me that

my poor body lay several hundred miles away, with heart pulsing and lungsrespirating, but otherwise as dead as the body of any heart transplant donor, itsskull packed with useless, broken electronic gear The shift in perspective I hadearlier found well nigh impossible now seemed quite natural Though I couldthink myself back into my body in the tunnel under Tulsa, it took some effort tosustain the illusion For surely it was an illusion to suppose I was still in Okla-homa: I had lost all contact with that body

It occurred to me then, with one of those rushes of revelation of which weshould be suspicious, that I had stumbled upon an impressive demonstration ofthe immateriality of the soul based upon physicalist principles and premises.For as the last radio signal between Tulsa and Houston died away, had I not

changed location from Tulsa to Houston at the speed of light? And had Inot accomplished this without any increase in mass? What moved from A to B

at such speed was surely myself, or at any rate my soul or mind—the masslesscenter of my being and home of my consciousness My point of view had laggedsomewhat behind, but I had already noted the indirect bearing of point ofview on personal location I could not see how a physicalist philosopher couldquarrel with this except by taking the dire and counter-intuitive route of ban-ishing all talk of persons Yet the notion of personhood was so well entrenched

in everyone’s world view, or so it seemed to me, that any denial would be ascuriously unconvincing, as systematically disingenuous, as the Cartesian nega-tion, ‘‘non sum.’’1

The joy of philosophic discovery thus tided me over some very bad minutes

or perhaps hours as the helplessness and hopelessness of my situation becamemore apparent to me Waves of panic and even nausea swept over me, made allthe more horrible by the absence of their normal body-dependent phenomen-ology No adrenalin rush of tingles in the arms, no pounding heart, no pre-monitory salivation I did feel a dread sinking feeling in my bowels at onepoint, and this tricked me momentarily into the false hope that I was under-going a reversal of the process that landed me in this fix—a gradual undis-embodiment But the isolation and uniqueness of that twinge soon convinced

me that it was simply the first of a plague of phantom body hallucinations that

I, like any other amputee, would be all too likely to suffer

My mood then was chaotic On the one hand, I was fired up with elation at

my philosophic discovery and was wracking my brain (one of the few familiarthings I could still do), trying to figure out how to communicate my discovery

to the journals; while on the other, I was bitter, lonely, and filled with dreadand uncertainty Fortunately, this did not last long, for my technical supportteam sedated me into a dreamless sleep from which I awoke, hearing withmagnificent fidelity the familiar opening strains of my favorite Brahms pianotrio So that was why they had wanted a list of my favorite recordings! It didnot take me long to realize that I was hearing the music without ears Theoutput from the stereo stylus was being fed through some fancy rectificationcircuitry directly into my auditory nerve I was mainlining Brahms, an unfor-gettable experience for any stereo buff At the end of the record it did not sur-prise me to hear the reassuring voice of the project director speaking into amicrophone that was now my prosthetic ear He confirmed my analysis of whathad gone wrong and assured me that steps were being taken to re-embody me

He did not elaborate, and after a few more recordings, I found myself driftingoff to sleep My sleep lasted, I later learned, for the better part of a year, andwhen I awoke, it was to find myself fully restored to my senses When I lookedinto the mirror, though, I was a bit startled to see an unfamiliar face Beardedand a bit heavier, bearing no doubt a family resemblance to my former face,and with the same look of spritely intelligence and resolute character, but defi-nitely a new face Further self-explorations of an intimate nature left me nodoubt that this was a new body and the project director confirmed my con-clusions He did not volunteer any information on the past history of my newbody and I decided (wisely, I think in retrospect) not to pry As many philoso-phers unfamiliar with my ordeal have more recently speculated, the acquisition

Where Am I? 29

of a new body leaves one’s person intact And after a period of adjustment to anew voice, new muscular strengths and weaknesses, and so forth, one’s per-sonality is by and large also preserved More dramatic changes in personalityhave been routinely observed in people who have undergone extensive plasticsurgery, to say nothing of sex change operations, and I think no one conteststhe survival of the person in such cases In any event I soon accommodated to

my new body, to the point of being unable to recover any of its novelties to myconsciousness or even memory The view in the mirror soon became utterlyfamiliar That view, by the way, still revealed antennae, and so I was not sur-prised to learn that my brain had not been moved from its haven in the life-support lab

I decided that good old Yorick deserved a visit I and my new body, whom

we might as well call Fortinbras, strode into the familiar lab to another round

of applause from the technicians, who were of course congratulating selves, not me Once more I stood before the vat and contemplated poor Yorick,and on a whim I once again cavalierly flicked off the output transmitter switch.Imagine my surprise when nothing unusual happened No fainting spell, nonausea, no noticeable change A technician hurried to restore the switch to on,but still I felt nothing I demanded an explanation, which the project directorhastened to provide It seems that before they had even operated on the firstoccasion, they had constructed a computer duplicate of my brain, reproducingboth the complete information processing structure and the computationalspeed of my brain in a giant computer program After the operation, but beforethey had dared to send me off on my mission to Oklahoma, they had run thiscomputer system and Yorick side by side The incoming signals from Hamletwere sent simultaneously to Yorick’s transceivers and to the computer’s array

them-of inputs And the outputs from Yorick were not only beamed back to Hamlet,

my body; they were recorded and checked against the simultaneous output ofthe computer program, which was called ‘‘Hubert’’ for reasons obscure to me.Over days and even weeks, the outputs were identical and synchronous, which

of course did not prove that they had succeeded in copying the brain’s tional structure, but the empirical support was greatly encouraging

func-Hubert’s input, and hence activity, had been kept parallel with Yorick’s ing my disembodied days And now, to demonstrate this, they had actuallythrown the master switch that put Hubert for the first time in on-line control of

dur-my body—not Hamlet, of course, but Fortinbras (Hamlet, I learned, had neverbeen recovered from its underground tomb and could be assumed by this time

to have largely returned to the dust At the head of my grave still lay the nificent bulk of the abandoned device, with the word STUD emblazoned on itsside in large letters—a circumstance which may provide archeologists of thenext century with a curious insight into the burial rites of their ancestors.)The laboratory technicians now showed me the master switch, which hadtwo positions, labeled B, for Brain (they didn’t know my brain’s name wasYorick) and H, for Hubert The switch did indeed point to H, and they ex-plained to me that if I wished, I could switch it back to B With my heart in mymouth (and my brain in its vat), I did this Nothing happened A click, that wasall To test their claim, and with the master switch now set at B, I hit Yorick’soutput transmitter switch on the vat and sure enough, I began to faint Once

mag-the output switch was turned back on and I had recovered my wits, so tospeak, I continued to play with the master switch, flipping it back and forth Ifound that with the exception of the transitional click, I could detect no trace of

a difference I could switch in mid-utterance, and the sentence I had begunspeaking under the control of Yorick was finished without a pause or hitch ofany kind under the control of Hubert I had a spare brain, a prosthetic devicewhich might some day stand me in very good stead, were some mishap to be-fall Yorick Or alternatively, I could keep Yorick as a spare and use Hubert

It didn’t seem to make any difference which I chose, for the wear and tearand fatigue on my body did not have any debilitating effect on either brain,whether or not it was actually causing the motions of my body, or merelyspilling its output into thin air

The one truly unsettling aspect of this new development was the prospect,which was not long in dawning on me, of someone detaching the spare—Hubert or Yorick, as the case might be—from Fortinbras and hitching it to yetanother body—some Johnny-come-lately Rosencrantz or Guildenstern Then (ifnot before) there would be two people, that much was clear One would be me,and the other would be a sort of super-twin brother If there were two bodies,one under the control of Hubert and the other being controlled by Yorick, thenwhich would the world recognize as the true Dennett? And whatever the rest

of the world decided, which one would be me? Would I be the Yorick-brainedone, in virtue of Yorick’s causal priority and former intimate relationship withthe original Dennett body, Hamlet? That seemed a bit legalistic, a bit too redo-lent of the arbitrariness of consanguinity and legal possession, to be convincing

at the metaphysical level For, suppose that before the arrival of the secondbody on the scene, I had been keeping Yorick as the spare for years, and lettingHubert’s output drive my body—that is, Fortinbras—all that time The Hubert-Fortinbras couple would seem then by squatter’s rights (to combat one legalintuition with another) to be the true Dennett and the lawful inheritor ofeverything that was Dennett’s This was an interesting question, certainly, butnot nearly so pressing as another question that bothered me My strongest intu-ition was that in such an eventuality I would survive so long as either brain-body couple remained intact, but I had mixed emotions about whether I shouldwant both to survive

I discussed my worries with the technicians and the project director Theprospect of two Dennetts was abhorrent to me, I explained, largely for socialreasons I didn’t want to be my own rival for the affections of my wife, nor did Ilike the prospect of the two Dennetts sharing my modest professor’s salary.Still more vertiginous and distasteful, though, was the idea of knowing thatmuch about another person, while he had the very same goods on me Howcould we ever face each other? My colleagues in the lab argued that I wasignoring the bright side of the matter Weren’t there many things I wanted to

do but, being only one person, had been unable to do? Now one Dennett couldstay at home and be the professor and family man, while the other could strikeout on a life of travel and adventure—missing the family of course, but happy

in the knowledge that the other Dennett was keeping the home fires burning

I could be faithful and adulterous at the same time I could even cuckoldmyself—to say nothing of other more lurid possibilities my colleagues were all

Where Am I? 31

too ready to force upon my overtaxed imagination But my ordeal in Oklahoma(or was it Houston?) had made me less adventurous, and I shrank from thisopportunity that was being offered (though of course I was never quite sure itwas being offered to me in the first place).

There was another prospect even more disagreeable—that the spare, Hubert

or Yorick as the case might be, would be detached from any input from tinbras and just left detached Then, as in the other case, there would be twoDennetts, or at least two claimants to my name and possessions, one embodied

For-in FortFor-inbras, and the other sadly, miserably disembodied Both selfishness andaltruism bade me take steps to prevent this from happening So I asked thatmeasures be taken to ensure that no one could ever tamper with the transceiverconnections or the master switch without my (our? no, my) knowledge andconsent Since I had no desire to spend my life guarding the equipment inHouston, it was mutually decided that all the electronic connections in the labwould be carefully locked: both those that controlled the life-support systemfor Yorick and those that controlled the power supply for Hubert would beguarded with fail-safe devices, and I would take the only master switch, out-fitted for radio remote control, with me wherever I went I carry it strappedaround my waist and—wait a moment—here it is Every few months I recon-noiter the situation by switching channels I do this only in the presence offriends of course, for if the other channel were, heaven forbid, either dead orotherwise occupied, there would have to be somebody who had my interests atheart to switch it back, to bring me back from the void For while I could feel,see, hear and otherwise sense whatever befell my body, subsequent to such aswitch, I’d be unable to control it By the way, the two positions on the switchare intentionally unmarked, so I never have the faintest idea whether I amswitching from Hubert to Yorick or vice versa (Some of you may think that inthis case I really don’t know who I am, let alone where I am But such reflections

no longer make much of a dent on my essential Dennett-ness, on my own sense

of who I am If it is true that in one sense I don’t know who I am then that’sanother one of your philosophical truths of underwhelming significance.)

In any case, every time I’ve flipped the switch so far, nothing has happened

So let’s give it a try

‘‘thank god! i thought you’d never flip that switch! You can’t imaginehow horrible it’s been these last two weeks—but now you know, it’s your turn

in purgatory How I’ve longed for this moment! You see, about two weeksago—excuse me, ladies and gentlemen, but I’ve got to explain this to my

um, brother, I guess you could say, but he’s just told you the facts, so you’llunderstand—about two weeks ago our two brains drifted just a bit out ofsynch I don’t know whether my brain is now Hubert or Yorick, any more thanyou do, but in any case, the two brains drifted apart, and of course once theprocess started, it snowballed, for I was in a slightly different receptive state forthe input we both received, a difference that was soon magnified In no time atall the illusion that I was in control of my body—our body—was completelydissipated There was nothing I could do—no way to call you you didn’t evenknow i existed!It’s been like being carried around in a cage, or better, like be-ing possessed—hearing my own voice say things I didn’t mean to say, watch-ing in frustration as my own hands performed deeds I hadn’t intended You’d

scratch our itches, but not the way I would have, and you kept me awake, withyour tossing and turning I’ve been totally exhausted, on the verge of a nervousbreakdown, carried around helplessly by your frantic round of activities, sus-tained only by the knowledge that some day you’d throw the switch.

‘‘Now it’s your turn, but at least you’ll have the comfort of knowing I knowyou’re in there Like an expectant mother, I’m eating—or at any rate tasting,smelling, seeing—for two now, and I’ll try to make it easy for you Don’t worry.Just as soon as this colloquium is over, you and I will fly to Houston, and we’llsee what can be done to get one of us another body You can have a femalebody—your body could be any color you like But let’s think it over I tell youwhat—to be fair, if we both want this body, I promise I’ll let the project direc-tor flip a coin to settle which of us gets to keep it and which then gets to choose

a new body That should guarantee justice, shouldn’t it? In any case, I’ll takecare of you, I promise These people are my witnesses

‘‘Ladies and gentlemen, this talk we have just heard is not exactly the talk Iwould have given, but I assure you that everything he said was perfectly true.And now if you’ll excuse me, I think I’d—we’d—better sit down.’’2

Notes

1 Cf Jaakko Hintikka, ‘‘Cogito ergo sum: Inference or Performance?’’ The Philosophical Review, LXXI, 1962, pp 3–32.

2 Anyone familiar with the literature on this topic will recognize that my remarks owe a great deal

to the explorations of Sydney Shoemaker, John Perry, David Lewis and Derek Parfit, and in particular to their papers in Amelie Rorty, ed., The Identities of Persons, 1976.

Where Am I? 33

Can Machines Think?

Daniel C Dennett

Much has been written about the Turing test in the last few years, some of it terously off the mark People typically mis-imagine the test by orders of magnitude.This essay is an antidote, a prosthesis for the imagination, showing how huge the taskposed by the Turing test is, and hence how unlikely it is that any computer will everpass it It does not go far enough in the imagination-enhancement department, how-ever, and I have updated the essay with two postscripts

prepos-Can machines think? This has been a conundrum for philosophers for years,but in their fascination with the pure conceptual issues they have for the mostpart overlooked the real social importance of the answer It is of more thanacademic importance that we learn to think clearly about the actual cognitivepowers of computers, for they are now being introduced into a variety of sen-sitive social roles, where their powers will be put to the ultimate test:In a widevariety of areas, we are on the verge of making ourselves dependent upon theircognitive powers The cost of overestimating them could be enormous

One of the principal inventors of the computer was the great British matician Alan Turing It was he who first figured out, in highly abstract terms,how to design a programmable computing device—what we now call a uni-versal Turing machine All programmable computers in use today are in es-sence Turing machines Over thirty years ago, at the dawn of the computer age,Turing began a classic article, ‘‘Computing Machinery and Intelligence,’’ withthe words:‘‘I propose to consider the question, ‘Can machines think?’ ’’—butthen went on to say this was a bad question, a question that leads only to steriledebate and haggling over definitions, a question, as he put it, ‘‘too meaningless

mathe-to deserve discussion’’ (Turing, 1950) In its place he substituted what he mathe-took

to be a much better question, a question that would be crisply answerableand intuitively satisfying—in every way an acceptable substitute for the philo-sophic puzzler with which he began

First he described a parlor game of sorts, the ‘‘imitation game,’’ to be played

by a man, a woman, and a judge (of either gender) The man and woman arehidden from the judge’s view but able to communicate with the judge by tele-type; the judge’s task is to guess, after a period of questioning each contestant,which interlocutor is the man and which the woman The man tries to convincethe judge he is the woman (and the woman tries to convince the judge of the

From chapter 1 in Brainchildren (Cambridge, MA:MIT Press, 1995/1998), 3–29 Reprinted with permission.

truth), and the man wins if the judge makes the wrong identification A littlereflection will convince you, I am sure, that, aside from lucky breaks, it wouldtake a clever man to convince the judge that he was a woman—assuming thejudge is clever too, of course.

Now suppose, Turing said, we replace the man or woman with a computer,and give the judge the task of determining which is the human being andwhich is the computer Turing proposed that any computer that can regularly

or often fool a discerning judge in this game would be intelligent—would

be a computer that thinks—beyond any reasonable doubt Now, it is important

to realize that failing this test is not supposed to be a sign of lack of gence Many intelligent people, after all, might not be willing or able to playthe imitation game, and we should allow computers the same opportunity todecline to prove themselves This is, then, a one-way test; failing it provesnothing

intelli-Furthermore, Turing was not committing himself to the view (although it iseasy to see how one might think he was) that to think is to think just like a hu-man being—any more than he was committing himself to the view that for aman to think, he must think exactly like a woman Men and women, and com-puters, may all have different ways of thinking But surely, he thought, if onecan think in one’s own peculiar style well enough to imitate a thinking man orwoman, one can think well, indeed This imagined exercise has come to beknown as the Turing test

It is a sad irony that Turing’s proposal has had exactly the opposite effect

on the discussion of that which he intended Turing didn’t design the test as auseful tool in scientific psychology, a method of confirming or disconfirmingscientific theories or evaluating particular models of mental function; he de-signed it to be nothing more than a philosophical conversation-stopper Heproposed—in the spirit of ‘‘Put up or shut up!’’—a simple test for thinking thatwas surely strong enough to satisfy the sternest skeptic (or so he thought)

He was saying, in effect, ‘‘Instead of arguing interminably about the ultimatenature and essence of thinking, why don’t we all agree that whatever thatnature is, anything that could pass this test would surely have it; then we couldturn to asking how or whether some machine could be designed and builtthat might pass the test fair and square.’’ Alas, philosophers—amateur andprofessional—have instead taken Turing’s proposal as the pretext for just thesort of definitional haggling and interminable arguing about imaginary coun-terexamples he was hoping to squelch

This thirty-year preoccupation with the Turing test has been all the more grettable because it has focused attention on the wrong issues There are realworld problems that are revealed by considering the strengths and weaknesses

re-of the Turing test, but these have been concealed behind a smokescreen re-ofmisguided criticisms A failure to think imaginatively about the test actuallyproposed by Turing has led many to underestimate its severity and to confuse

it with much less interesting proposals

So first I want to show that the Turing test, conceived as he conceived it, is(as he thought) plenty strong enough as a test of thinking I defy anyone toimprove upon it But here is the point almost universally overlooked by theliterature:There is a common misapplication of the sort of testing exhibited by

36 Daniel C Dennett

the Turing test that often leads to drastic overestimation of the powers of ally existing computer systems The follies of this familiar sort of thinkingabout computers can best be brought out by a reconsideration of the Turing testitself.

actu-The insight underlying the Turing test is the same insight that inspires thenew practice among symphony orchestras of conducting auditions with anopaque screen between the jury and the musician What matters in a musician,obviously, is musical ability and only musical ability; such features as sex, hairlength, skin color, and weight are strictly irrelevant Since juries might bebiased—even innocently and unawares—by these irrelevant features, they arecarefully screened off so only the essential feature, musicianship, can be exam-ined Turing recognized that people similarly might be biased in their judg-ments of intelligence by whether the contestant had soft skin, warm blood,facial features, hands and eyes—which are obviously not themselves essentialcomponents of intelligence—so he devised a screen that would let through only

a sample of what really mattered:the capacity to understand, and think erly about, challenging problems Perhaps he was inspired by Descartes, who

clev-in his Discourse on Method (1637) plausibly argued that there was no moredemanding test of human mentality than the capacity to hold an intelligentconversation:

It is indeed conceivable that a machine could be so made that it wouldutter words, and even words appropriate to the presence of physical acts

or objects which cause some change in its organs; as, for example, if it wastouched in some spot that it would ask what you wanted to say to it; if inanother, that it would cry that it was hurt, and so on for similar things.But it could never modify its phrases to reply to the sense of whateverwas said in its presence, as even the most stupid men can do

This seemed obvious to Descartes in the seventeenth century, but of course thefanciest machines he knew were elaborate clockwork figures, not electroniccomputers Today it is far from obvious that such machines are impossible, butDescartes’s hunch that ordinary conversation would put as severe a strain onartificial intelligence as any other test was shared by Turing Of course there isnothing sacred about the particular conversational game chosen by Turing forhis test; it is just a cannily chosen test of more general intelligence The as-sumption Turing was prepared to make was this:Nothing could possibly passthe Turing test by winning the imitation game without being able to performindefinitely many other clearly intelligent actions Let us call that assumptionthe quick-probe assumption Turing realized, as anyone would, that there arehundreds and thousands of telling signs of intelligent thinking to be observed

in our fellow creatures, and one could, if one wanted, compile a vast battery ofdifferent tests to assay the capacity for intelligent thought But success on hischosen test, he thought, would be highly predictive of success on many otherintuitively acceptable tests of intelligence Remember, failure on the Turing testdoes not predict failure on those others, but success would surely predict suc-cess His test was so severe, he thought, that nothing that could pass it fair andsquare would disappoint us in other quarters Maybe it wouldn’t do everything

we hoped—maybe it wouldn’t appreciate ballet, or understand quantum

physics, or have a good plan for world peace, but we’d all see that it was surelyone of the intelligent, thinking entities in the neighborhood.

Is this high opinion of the Turing test’s severity misguided? Certainly manyhave thought so—but usually because they have not imagined the test insufficient detail, and hence have underestimated it Trying to forestall thisskepticism, Turing imagined several lines of questioning that a judge mightemploy in this game—about writing poetry, or playing chess—that would betaxing indeed, but with thirty years’ experience with the actual talents andfoibles of computers behind us, perhaps we can add a few more tough lines ofquestioning

Terry Winograd, a leader in artificial intelligence efforts to produce sational ability in a computer, draws our attention to a pair of sentences (Wino-grad, 1972) They differ in only one word The first sentence is this:

conver-The committee denied the group a parade permit because they advocatedviolence

Here’s the second sentence:

The committee denied the group a parade permit because they fearedviolence

The difference is just in the verb—advocated or feared As Winograd points out,the pronoun they in each sentence is officially ambiguous Both readings of thepronoun are always legal Thus we can imagine a world in which governmen-tal committees in charge of parade permits advocate violence in the streets and,for some strange reason, use this as their pretext for denying a parade permit.But the natural, reasonable, intelligent reading of the first sentence is that it’sthe group that advocated violence, and of the second, that it’s the committeethat feared violence

Now if sentences like this are embedded in a conversation, the computermust figure out which reading of the pronoun is meant, if it is to respondintelligently But mere rules of grammar or vocabulary will not fix the rightreading What fixes the right reading for us is knowledge about the world,about politics, social circumstances, committees and their attitudes, groups thatwant to parade, how they tend to behave, and the like One must know aboutthe world, in short, to make sense of such a sentence

In the jargon of Artificial Intelligence (AI), a conversational computer needs alot of world knowledge to do its job But, it seems, if somehow it is endowed withthat world knowledge on many topics, it should be able to do much more withthat world knowledge than merely make sense of a conversation containingjust that sentence The only way, it appears, for a computer to disambiguatethat sentence and keep up its end of a conversation that uses that sentencewould be for it to have a much more general ability to respond intelligently toinformation about social and political circumstances, and many other topics.Thus, such sentences, by putting a demand on such abilities, are good quick-probes That is, they test for a wider competence

People typically ignore the prospect of having the judge ask off-the-wallquestions in the Turing test, and hence they underestimate the competence acomputer would have to have to pass the test But remember, the rules of the

38 Daniel C Dennett