The Mystery of Consciousness By John R.

Searle
There is no general agreement in the interdisciplinary field known as consciousness research studies (or consciousness studies or consciousness research; take your pick).on exactly what the word consciousness means. This lack has not prevented a flourishing of such research, especially during the last two decades, any more than the a sence of a generally agreed upon definition of the word life has hindered the flourishing of the field of iology. The situation for consciousness research is actually more extreme than that, reminding one of the prover ial story of the four lind men and the elephant. !ersons claiming to e talking a out the mysteries of consciousness or to have solved them often seem to e talking right past each other a out some very different things. This ook contains reviews, originally written for the "ew #ork $eview of %ooks, of six significant ooks or sets of ooks y ma&or authors in the field. 'dditionally, it contains summaries of the views of the reviewer, (ohn )earle, a professor of !hilosophy at %erkeley and himself a ma&or figure in the field. Together they cover many, though y no means all, of the differing views on the nature of consciousness and why it is a mystery, if indeed it is. *t is my hope that this ook may serve as a sort of +liff,s "otes, providing summaries of the essential points in texts without having to read the original ook entire. -ne thing it does offer that a +liff,s "otes can not is, in two cases, sets of letters heatedly exchanged etween the reviewer and the person reviewed following a review,s original pu lication. * have read some ut y no means all of these ook reviewed ooks, and do hope that anyone who has read one or more of them will participate actively in our discussion and correct me if at any point my interpretation seems to e wrong. Some divisions within the conscious studies community and how they manifest here: . ' ma&or division in the consciousness studies community exists on this .uestion. *f we could learn enough a out rain functioning to completely descri e and predict the entire chain of events from sensation and prior rain state through ehavior and new rain state and do it every time would we then have created a complete description of consciousness. )ome argue that if we were a le to do this not only would we still not have a complete description of consciousness ut possi ly we would e no further towards one than we were efore. )ome claim this final knowledge will always e eyond our understanding. )urprisingly perhaps, none of those who hold this latter view today do so ecause they elieve in what is now known as su stance dualism, the idea that our physical rains are somehow connected to non/physical minds. *n a very road sense all persons * know of who are currently participating in consciousness studies de ates are what were traditionally called materialists. 0hy some of them would deny the possi ility of completely understanding consciousness through traditional, materialistic scientific

research and even refuse to e called materialists, sometimes throwing the word at their opponents as an accusation, is a matter much more su tle. !hilosopher 2aniel 2ennett has descri ed these two groups neutrally as the ' team and the % team. !sychologist 2aniel 0egner has descri ed them less neutrally as the ro o/geeks and the ad scientists, using the insult that each group would most likely throw at the other as an identifier. The ro o/geeks are the ones who elieve that such a complete sensation3 rain3 ehavior description, if could it e created, would e a complete description of consciousness. The ad scientists are the ones who think such a description would e insufficient and sometimes accuse the ro o/geeks of actually denying the existence of consciousness even as they claim to study it. *n this ook (ohn )earle himself serves as a nice example of the latter group while accusing 2ennett of eing a mem er of the former, one reason for the heartedness of their included exchange of letters. *n general, oth serve, to me, as exhi itions of a type of thinking a out consciousness that attempts to deal with seemingly fundamental issues which would exist as pro lems regardless of the detailed nature of the rain or the details of most ehavior. *ssues. The true relationship etween su &ectivity and o &ectivity and the possi ility of ever doing scientific research on the latter is a common point of contention for these people. )eemingly at the opposite end of ade.uacy for experimental research &ust now are the neuroscientists and clinical neurologists studying &ust what effect various regions of the rain have on consciousness and how they coordinate their efforts. *n this ook, )ir 4rancis +rick, *srael $osenfield and to some extent 5erald 6delman seem to fit. 7owever, rain scans and pathological dissections are not the only ways to study consciousness, even today. )ome researchers elieve that a more thorough understanding of the at a more fundamental iological level, elementary nerve nets within rain regions and the su /cellular functioning of neurons themselves (possi ly those of the other kinds of cells which together make up ninety percent of rain tissue as well) is needed. The con&ectures of mathematician3physicist $oger !enrose and some of the work of rain scientist 5erald 6delman serve as examples of this kind of research in this ook. 4inally, there is the most traditional of experimental study of consciousness, the study of the ehavior of intact organisms (fre.uently college students taking !sychology 181) which was already going on in the la oratories of 0illiam (ames and 0ilhelm 0undt well efore the end of the nineteenth century. There are, unfortunately, no examples of this type of consciousness study in )earle,s ook, ut most fortunately the current (2ecem er, 988:) issue of the )cientific 'merican contains and excellent example under the title of ;agic and the %rain which * shall e referring to later. "o papers from the artificial intelligence community are included, yet the influence of that topic is pervasive. 't the philosophical end of things the .uestion of whether or not a computer could ever e conscious seems to come up a out as often a discussions a out o &ectivity and su &ectivity. 'rtificial *ntelligence does not seem to have much influence on rain region studies, though computeri<ed analysis of rain scan data is often central

to it. 7owever, elow that level computer modeling is making a ig contri ution with artificial neural networks having moved eyond rain research and into a num er of applications, some of them .uite unexpected. 4urthermore, the power of massively multiprocessor computers (not artificially intelligent) is finally on the verge of permitting research on su /cellular processes y simulating the interactions of individual atoms. 4inally cognitive simulation of human psychology, originally named in the 1=>8,s, is often derided today as 5-4'*, standing for 5ood -ld 4ashion '*, ut like 5od it seems to keep hanging around however many times it is declared dead. Now on to the individual chapters in the boo ! chapter by chapter" #. John Searle and $the Chinese room% (The $ediscovery of ;ind and other works) )earle,s +hinese $oom thought experiment may e the most referred to and most critici<ed such in contemporary consciousness studies. 4ellow philosopher 2aniel 2ennett (.uoted elsewhere in this ook) may e correct that this is the only ma&or idea that )earle has ever had, ut even if this is true it still lifts )earle into the circle of ma&or philosophers currently working on !hilosophy of ;ind issues. The asic thought experiment is not difficult to understand. ;any variations upon it have een presented y oth )earle himself and (seemingly inumerical) critics of his over the years, ut the asic version is the one presented in this ook and is, * think, sufficient for our purposes. To paraphrase one of )earle,s own presentations of it, *magine that * (who do not understand +hinese) am locked in a room with oxes of +hinese sym ols and rule ooks descri ing what * am to do with these sym ols (my data ase). ?uestions in +hinese are passed to me ( unches of +hinese sym ols), and * look up in the rule ooks (my program) what * am supposed to do. * perform operations on the sym ols in accordance with the rules and from these generate unches of sym ols (answers to the .uestions) which * pass ack to those outside the room. 'gain imagine the room says )earle ut this time imagine that it contains a person fluent in +hinese who simply reads the passed in .uestions and, understanding them, simply writes out the answers in +hinese and passes those answers ack out of the room. )earle,s point is that something very different has happened in the room in each case, in one instance a clerk (or a computer) with no understanding of +hinese has created the output &ust y manipulating sym ols according to rules. *n the other case, the person fluent in +hinese and understanding to topics of the .uestions simply uses his or her understanding too translate and answer the .uestions, yet the outputs in each case may e identical. )earle sees this thought experiment as a refutation of the so/called Turing test, a thought experiment pu lished y that chemist, mathematician and computer scientist in the %ritish philosophical &ournal ;ind in 1=>>. *n Turing,s thought experiment &udges are allowed to communicate with parties to e tested only y Teletype or the e.uivalent. *n one version men and women all try to convince the &udges that they are really men. *n the

version, which interested Turing more, real humans and 'rtificial *ntelligence programs all try to convince the &udges that they are really human. 6xperiments of this kind have actually een carried out numerous times since the pu lication of Turing,s article. )ee, for example, an article in the on/line maga<ine )alon few years ago y &ournalist Tracy ?uan on 'rtificial *ntelligence and her participation in such an experiment, link toA archive.salon.com3may=B391st3artificial=B8>1>. *n fairness to Turing, he did not claim that his hypothesi<ed test would help in deciding the .uestion of machine consciousness. "either he nor any of his contemporaries that * am aware of ever discussed that issue. 0hat he and they were discussing in the mid/twentieth century was the possi ility of machine intelligence, which would seem to imply a strictly ehavioral trait, not a su &ective one. *n the early sixties when ;arvin ;insky and others at ;*T coined the term 'rtificial *ntelligence they defined it as referring to hardware3software systems which could perform acts which would e descri ed as intelligent if performed y a human. *mplicit and intended in that definition was the idea that the process y which such acts were performed might e nothing like the processes that would e used y a human. -nly the results were to count. * find it interesting that at the same time that the topic of machine consciousness has gained respecta ility on the intellectual scene the earlier .uestion of machine intelligence seems to have disappeared. -nce hotly de ated, it seems that no one much now wants to argue against the possi ility of any sort of strictly ehavioral intelligence eing shown y computer/like hardware and software. 4or examples of arguments emphatically made efore such skepticism disappeared dig up copies of philosopher 7u ert 2reyfus ooks 0hat +omputers +an,t 2o and 0hat +omputers )till +an,t 2o. )earle has now taken his argument against machine understanding much further in this ook and elsewhere than he did with his original version of the +hinese room. 7e argues that, for example, computers can not really do simple arithmetic. 0hat does happen, he says, is that computers are uilt or programmed to manipulate sym ols, and that it is humans not computers or programs that understand that the sym ols eing typed in or displayed descri e num ers and operations to e performed on num ers. *n the customary voca ulary of natural language research, among other areas, the term syntax is used to descri e grammar and other rules for forming and parsing sentences at the word level (i.e. rules a out language). *n contrast the term semantics is used to descri e rules for relating language statements to the thing eing descri ed, the meaning of sentences in other words. 7owever, )earle now insists repeatedly that since computers can never understand anything ( y his definition) programs can only do syntactic processing and never semantic processing of any kind, a highly idiosyncratic restriction on the use of these words. 0hat does )earle have to say a out consciousness then where it does show up, e.g. in rainsC 7e answers repeatedly that rains cause consciousness ecause it is a natural product of (some) iological systems, &ust as digestion is a natural, iological function of

a stomach. 0hile the +hinese room argument would seem to counter top/down arguments for consciousness eing deriva le y writing programs to simulate externally o serva le ehavior, it does not seem to me that it counters the opposite, top/down though experiment. -ne of creating simulated rains y simulating the interactions of the atoms that make up molecules and so on up. Ep in this case would include a simulation not only an entire rain ut also as much of the rest of the nervous system, the ody and it,s environment as necessary to reach a point where attachment to real world interfaces are possi le. )ome who have presented this argument have suggested that an appropriate real/world interface might e a humanoid ro ot with its sensors feeding into the simulated sensory nerves and the simulated motor nerves feeding into the ro ot ody,s effectors. )earle has countered this argument y saying that such a simulation would e only a simulation. #et in other places he proclaims himself a materialist which presuma ly means that he is not a vitalist who elieves that the physical laws governing iological systems are somehow different from those governing non/ iological ones. 't the same time he has said repeatedly that he elieves consciousness might e caused y systems made of materials other than the normal iological ones. ;ight not one not say then that such a simulation might e simply a consciousness causing system whose materials are simulated atoms rather than real ones. 'spects of this argument will come up again when discussing )earle,s review of 2avid +halmers ideas. &. Sir 'rancis Cric and the $bindin(% problem (The )cientific )earch for the )oul) 2espite its arresting title, +rick,s ook is a very mainstream example of contemporary, experimental rain science studies. *n this case tracing the flow of information through various speciali<ed rain centers (specifically the visual pathways) and attempting to prove or at least con&ecture how the ensem le manages to ehave, at least su &ectively, as a single entity. This is one of those ooks that may have een more important at time of pu lication for who wrote it than for what it contained. )ir 4rancis +rick, co/discoverer with 0atson of the structure of 2"', was pro a ly, especially to the general pu lic, one of the most recogni<ed and respected scientists of his time. The fact that he had now een devoting himself for some time to the scientific study of the relationship of consciousness to the rain legitimi<ed such studies to a significant degree as something perused y reputa le scientists and not y &ust y somewhat weird people out on the fringe. The pro lem +rick was dealing with was one that has long een recogni<ed in consciousness research. *n the 1F88,s the polymath philosopher 2escartes tried to formali<e his ideas a out the relationship etween mind and rain, the famous +artesian dualism. 7e con&ectured (and presented it only as a con&ecture) that the interface where rain and mind connected to each other was the penal gland near the center of the rain. 7is reason for this con&ecture was the knowledge that of all rain features then known only the penal gland was not lo ed into left and right halves. To 2escartes this was

>

suggestive ecause he was sure through introspection that consciousness seems entirely unary despite the ilateral structure of most of the ody with two arms, legs and eyes etc.. %y the time +rick egan his work on rains and consciousness much was known a out the visual pathways from retinas through the occipital lo e at the ack of the head and various mid/ rain structures including the thalamus on multiple ranching and intersection routes up to and within the cere ral cortex. *n fact, more was known then and pro a ly still is now a out information flow within the visual system than a out any other sense, making it an excellent focus for study. The inding pro lem as +rick reduced it to a rain function pro lem was the pro lem of how neurons temporarily ecome active as a unit. -ther researchers had already suggested that the solution might involve the synchroni<ed firing of neurons in areas responsive to different features of an o &ect such as shape, color and movement. +rick and his colleague +hris Goch took this idea further and suggested that particular firing rates in the range of thirty/five to seventy/five cycles per second ut most often around forty might e the rain correlate of visual consciousness. 'n o &ection can e raised (and )earle dutifully raises it).that what +rick and Goch seem to have discovered is an important mechanism underlying consciousness rather than consciousness itself. 7owever, as more and more rain research indicates that consciousness resides in a dynamic network of interacting ut highly distri uted areas and processes it would seem to ecome increasingly hard to tell one from the other if indeed they can e told apart. -n this point, * wish that these reviews of )earles, had included on of ;arvin ;insky,s ook )ociety of ;ind. This is ecause it would serve oth as an example of a ma&or ook on consciousness y an eminent 'rtificial *ntelligence researcher and also ecause its title. That title, it seems to me, so well captures the current view of Hmind from rain science, a view so different from the unary one which we, like 2escartes, seem to know so intimately through introspection. )ince the pu lication of +rick and Goch,s 'stonishing 7ypothesis research findings have een very good to ideas of temporal synchroni<ation as dynamic organi<ing principles in more and more types of rain activity, where in many cases emerging potential sources for attention competitively recruit other areas for temporary colla oration. 4or a very ela orate metaphor of how this works from a more recent ook see chapter four, ;aking +onsciousation, in $ita +arter,s 6xploring +onsciousness. ). *erald +delman! brain maps! robots , much else (The $emem ered !resent etc.) 5erald 6delman is a rain scientist in a road sense, writing, researching and speculating on everything from molecular em ryology through neural networks and on into rain area coordination and theories of conscious functioning including the linguistic and sym olic processes. 4urthermore, he has where appropriate resorted to computer modeling including modeling of simulated ro ots, a whole area of consciousness research not previously in this ook and unfortunately not to e discussed again.

5iven the readth of 6delman,s interests it was necessary for )earle in reviewing several of his ooks together to skip around somewhat and * shall do the same. 7owever, a recurrent theme in his work has een processes of self/organi<ation from the neuronal through rain region co/evolution and on into the development of conscious skills such a language and ruminative self/awareness. )earle, with what * see as un&ustified reification of consciousness and its cause, critici<es 6delman for failing to precisely delineate the point at which unconscious processes ecome conscious ones or explain exactly how or why that transition occurs. 7owever, * elieve that the lack of such explanations is precisely the point, that proto/conscious processes evolve from, with and into conscious ones in a coordinated and constantly shifting manner. To egin at the ottom and e more specific, 6delman elieves that rain structures are not genetically programmed to grow gradually ut deterministically into some final form. $ather, he thinks, the rain comes e.uipped with an oversupply of neuronal groups some of which die out while others flourish due to self/organi<ing interaction with oth stimuli from the outside world and each other. This "eural 2arwinism (the title of his ook on the su &ect) would e .uite in accord with developmental processes of many organs in many different creatures. 4or example, a utterfly,s wings while in the cocoon are initially solid, and the ela orate structure they exhi it on the emergent creature is the result of the dying off of the unneeded parts, a literal cutting or whittling away. 't the next level of organi<ation 6delman, like +rick, takes on the inding pro lem, discussed in the last section. . 7owever, where +rick,s interest was in the inding of different feature recognition areas within the single sensory modality of vision, 6delman is most interested in the inding across sensory modalities leading progressively to the recognition and classification of o &ects repeatedly encountered. 7e does agree, however, that some primitive levels of self3non/self recognition are genetically uilt in. *t would not do to have a a y chew off it,s fingers in the process of learning that they are part of its unary ody. *n attacking the multi/sensory inding pro lem 6delman emphasi<es what he calls reentry mapping ut most would refer to simply as feed ack from sheets of receptor cells providing more a stract representations of the individual,s world ack to those closer to the sensory sources which provide the more primitive and underlying representations. These maps which 6delman refers to are literal, physiological structures in the rain, more than thirty of them in the visual cortex alone. Through Hreentry entire ensem les evolve together. 6xtensive work has een done on the feasi ility of such process through computer simulations of such networks, and the importance of feed ack structures in such networks has een amply demonstrated. 6delman,s group has gone eyond neural net simulations using simple sensory inputs such as lack white imaging of pictures into nets which self/organi<e so as to recogni<e letters in a variety of typefaces. They have programmed complete simulated ro ots can learn coordination of simulated hands and eyes to explore their environments, and since

this ook was written such work has een extended into constructing real ro ots capa le of exploring the real (la oratory) world and at times interacting with humans there. 4rom these approaches and results 6delman has moved up to con&ecturing similar self/organi<ing hierarchies to explain the gradual ac.uisition of increasingly self/aware consciousness. Iital within this approach is a conception of memory as something constantly and dynamically organi<ed as part of the evolving complex rather than any sort of separate and passive storehouse, hence his title The $emem ered !resent. 0e experience time and se.uence directly at higher levels &ust as we sense motion directly at lower levels (all the way ack to the retinal nerves within the eyeJ) and not as a deduction from a succession of still images. This dynamic view of memory will come up again when discussing the work of neurologist *srael $osenfield. . -. Ro(er .enrose! *odel/s Theorem and 0uantum Computin( ()hadows of the ;ind) *n a nutshell, !enrose argues that the a ility of humans to comprehend 5odel,s theorem in mathematical logic proves that all human thinking is not done using the "ewtonian physics that underlies conventional chemistry ut must utili<e some su /atomic properties of .uantum mechanics. 7e further argues that this limit would also apply to conventional computers ut perhaps not to future .uantum information processing systems. )ince oth 5odel,s Theorem and .uantum computing are, to most people, very ig and difficult su &ects * feel like saying here, more than at any other point in this precis go read the ooks, meaning oth )earle,s ook and !enrose ook. )till, here goesK Gurt 5odel developed his theorem in the mid/thirties. *t proves that for any system of formal mathematics capa le of representing oth the addition and multiplication of positive whole num ers there will always e some true statements that can never e proven either true or false within the system itself. This only holds for true statements since, theoretically, any false statement can always e proven false y showing one particular instance where it is false. 4or true statements, y contrast, it must e possi le to prove that every single instance is true, and for theorems a out sets of natural num ers the num er of specific instances can e literally, mathematically infinite. *magine some theorem which states that for ever num er it is true that (something or other) ut re.uires one proof in the case of the num er one, a different proof if the num er is two and so on /L perhaps you can now get the general idea. The way in which 5odel proved this was to show a way in any such num er system of constructing a statement that could never e proven either true or false ecause it was self/reverently contradictory. 's a very simple example of such a statement consider this one, it says, this statement is false. *f it is true then it must e correct when it says that it is false, ut if it is false then it must e e.ually certain that it is true, and so on ad infinitum.

* have a ook on 5odel,s theorem which * very much wish * could lay my hands on &ust now to give here as a citation. The reason is that it contains as an appendix a translation from the original 5erman of the complete theorem done y Gurt 5odel himself when he was a 4ellow at the !rinceton *nstitute of 'dvanced )tudies, in the early sixties. The complete paper is a out thirty/five pages long and like much in pure mathematics is actually more tedious than difficult. 0hat 5odel demonstrated in that paper was a way (&ust one of many) to encode any finite string of characters in a finite alpha et as a num er consisting of the product o tained y multiplying together a collection of prime num ers (num ers greater than one with no divisors except themselves and one) with the position of a prime in the se.uence of primes starting with the num er two representing the position of a character and the num er of times that prime occurs representing the value of the character in that position. 7ere,s an exampleA assume we have a two or more letter alpha et with the value of the first letter, call it ', eing one and the value of the second letter, call it %, eing two. 5iven those definitions the 5odel num er for the two character string '% can e calculated a followsA the first position is represented y a two, the first of all prime num ers, and the value of the character, ', in that position is one, so we have one two. The second position is represented y three, the second prime num er the se.uence of all prime num ers, and the value of the character, %, in that position is two, so we are going to have two threes. The 5odel num er for our string '% is therefore going to e 9 x (@ x @) or 9 x = or 1:, and given the num er 1: we could work ackward and unam iguously discover that the original string consisted of '%. 7aving devised his coding scheme and assigned values to a minimal set of logical operators plus some varia les what 5odel did next was devise se.uences of arithmetic operations which would do the same as the normal character manipulations in sym olic logic or, more precisely, second order prepositional calculus. That was the hard part. 0ith all of this paraphernalia in hand he could then show how to represent a self/contradictory, self/referential statement of the sort exemplified a few paragraphs ack and he was done. ' way of always eing a le to create an unprova le and also undisprova le statement in any such num er system or one containing it had now een demonstrated. !eople have een using 5odel,s theorem to try to prove that there are limits on what computers can do that humans are not su &ect to since at least the 1=D8,s. They have usually failed ecause ehavior L human or machine L is usually imitated y simulation, not y proving a stract properties. 'lan Turing himself disproved some of these early attempts that were already around in his time. $oger !enrose has given the est attempt to use 5odel,s theorem in this way that * have ever come across. *t has not convinced me, ut it is long and su tle enough so that * could not reproduce or criti.ue it without having !enrose, ook at hand to go y, which * do not currently have. %acking away for a moment and ack to the sort of issues that often preoccupy )earle it is never clear to me &ust what !enrose means y seeing the truth of 5odel,s theorem. *f he means a ehavior such a reading a statement or set of statements and then printing out

some kind of statement of the result, something that either a human or a machine, conscious or unconscious in the latter case, can do. %ut &ust what would those input statements and the output statement look like in detail. * do not know, and if !enrose does * wish he would pu lish them as examples in some new ook. *f !enrose means understand in some su &ective sense L which * think he does not, ut &ust in case L then we are ack to )earle,s argument from his first chapter that no computer can ever understand anything. "ow finally on to the second half of !enrose assertion, that the a ility of humans to understand 5odel,s theorem proves that they must e a le to do .uantum information processing of some kind. ?uantum information is a very real topic currently eing frantically researched y the E) "ational )ecurity 'gency among others. This is ecause it seems potentially capa le of decomposing very large num ers into their prime factors, and the ina ility to do so in any reasona le amount of time is the key (pun unavoida leC) to so/called trap/door algorithms. These in turn are the asis of cryptographic schemes that allow someone to encrypt a message without a clue as to how to decrypt it and vice versa, and such codes are vital today to many oth military and commercial applications, ?uantum information processing is not a candidate to replace your !+ or even most scientific super/computers. Their physical set/ups look like arrangements for very expensive physics experiments and asically are. The ways in which they will e used when practical are likewise much more like physics experiments than like computer programming. ' ma&or pro lem in making .uantum computing practical is entangling (* won,t go into that here. #ou should e a le to sort of guess at the meaning * hope, and for exact definitions there are entire ooks out thereJ) enough atoms to represent enough ./ its (again, either guess or read) to do re.uisite calculations while keeping them separate from all other atoms for the entire (miniscule) time that the computations are going on. !enrose thinks that water molecules isolated in the cytoskelatical tu ules that help neurons keep their shape could provide suita le environments in which this could happen. 0hen * first egan to read an article y him on this * suspected that was where he was heading and was pleased with myself when proven right. +urrently, many experts are saying that those microtu ules would not e suita le for that purpose. ;ore fundamentally, the .uantum laws !enrose wants to depend on are not those currently known ut new ones which he suspects may come to light from attempts to formulate )uper 5ravity unified field theories, another topic that * am not going to even try to get into hereJ 1. 2aniel 2ennett and consciousness re3ected or not (+onsciousness 6xplained et al.) *n this chapter of this ook )earle states repeatedly and often condescendingly that 2ennett does not elieve in consciousness, that he re&ects the idea of its existence completely. 2ennett sees it differently, 7e states that he does elieve in consciousness ut that it,s not what )earle and many others intuitively think that it is. 7e states this not only

18

in the ooks of his that )earle reviewed ut also in an exchange of letters etween them following )earle,s review which, wonderfully and for a wonder, )earle includes here immediately after the copy of his original review. 0hat 2ennett elieves and )earle disagrees on seems actually to consist of two different things. -ne is epistemic, i.e. a .uestion of what can e known accurately and how. 2ennett holds to the ehaviorist/methodology, which as )earle takes pains to point out was also held y 2ennett,s philosophical mentor 5il ert $yle, which claims that relia le science can only e ased on things o serva le in the third person, prefera ly y several persons who can then compare their o servations. This puts him very s.uarely and proudly on the ' or ro o/geek team explained in my introduction, the group that elieves that if rain and ehavior could e completely explained that would also amount to a complete explanation of consciousness. *n detail and in practice, what this methodological restriction says is that when we study consciousness we never actually do so directly. *nstead, it says, we infer the consciousness of experimental su &ects from their ver al ehavior, rain scans etc. 0e should always make this clear in research studies wherever there might e any confusion, which in practice seems to happen .uite seldom. 0hat this methodological !uritanism, to use a term coined y one of its critics, excludes from good science is any first person report y the experimenter of anything supposedly learned directly from his or her introspective experience. To include such experiences scientifically they must e reported and evaluated in exactly the same way that a third person report from any other experimental su &ect would e. ;ost of the time of course, almost all of us including committed ehaviorists intuitively (as 2ennett would pro a ly say) use a mixture of approaches. 0e infer the su &ective states of others y comparing them to real or imagined su &ective states of our own. Then we assume, unless the evidence forces us to think otherwise, that others have functioned internally and su &ectively pretty much as we know that we would have from our direct and introspective examination of ourselves. $ecent neural research on oth humans and animals has shown that this process appears to occur at a very asic and often unconscious level in so called mirror neurons that fire as if we were performing an act when we see someone else, even of a different species, doing so. 2ennett does not deny that we use such methods or that they are fre.uently effective and have stayed with us through evolution precisely ecause they so often are. 7e simply denies that they constitute a valid scientific method of drawing scientifically accurate conclusions. )earle, in contrast, seems to think that su &ective experience provides the most certain evidence that we have a out consciousness. This is not ecause it always provides us with an accurate representation of the outside world ut ecause we can not deny the reality of our own thoughts and sensations as things that we have su &ectively experienced. This is emphatically not a new issue in the study of consciousness. 2escartes is still remem ered a third of a millennium later for having said, * think therefore * am.

11

Enfortunately (and to me frustratingly) this is misremem ered as eing the heart of his philosophy. *n fact it was simply the start of in example in chapter four of his 2iscourse on ;ethod of how his method for reaching relia le conclusions (which was the heart of his philosophy) could e applied to the pro lem of metaphysics. *n discovering this, as something that could not e dou ted (the first step in his method he was specifically contrasting its certainty with all other seemingly true ut potentially fallacious eliefs. 'n example was the elief that he had a ody and that there was an outside world rather than, for instance, eing a spirit, alone in the universe who was having a dream. (!roving that there did exist in the universe something esides himself was the next step in his example.) *n other words, he was concluding only that something existed, not at all the nature of that something. This leads us nicely to the second of the two points a out consciousness on which 2ennett and )earle disagree so violently. The first point of their disagreement was3is epistemic, how to gain accurate ( e a modern and say scientific) knowledge a out consciousness and how to make sure that it is accurate. The second point of contention etween them is ontological, i.e. is consciousness real and if so what does it mean to say that it is real. They do at least agree that these are the two points in contention and use the same words for themJ The reality of consciousness and what that means is a slippery issue now &ust as it has een for centuries for people trying to understand 2escartes. 0hen it comes to 2ennett,s views on the matter *,m inclined to say don,t trust me too much; go and read his ooks, prefera ly several of them. 4ortunately, in this little ook we also have the exchange of letters etween the two men to go y. These are two men, oth considered to e currently eminent philosophers in the area of philosophy of mind, really seem to genuinely despise each other, perhaps somewhat ecause they oth are considered so eminent. 4orgetting the relevance to the topic of consciousness, these letter might e read for fun simply as examples of how right intellectuals can sometimes go at each other in pu lic like small oys in a schoolyard and apparently oth en&oy it. 0hat )earle elieves a out consciousness seems to e easier to understand, he elieves it to e something unary and fundamental, not to e dou ted y anyone except as an example of intellectual pathology, which is something he explicitly accuses 2ennett of at one point in this pu lished review. * always get a it edgy when * come across someone emphatically asserting that something can not possi ly e dou ted y any sane person. * lean toward the modern methodological doctrine of falsa ility, that something which can not even potentially e proven to e false can pro a ly not e shown to e true in the usual sense either ut is perhaps an artifact of our language or some such. * do however .uite agree with )earle that those facts seem intuitively, un.uestiona ly true. !erhaps surprisingly, 2ennett agrees with that, he simply holds that such intuitions (he uses that word) however hard to shake are in fact wrong. 0hat does 2ennett elieve then a out consciousnessC 4or one thing he elieves what he calls his multiple drafts theory a out it. *f * understand this (and * am not at all sure that * do) it takes the known fact of neurologically distri uted consciousness processing in the

19

and con&ectures that it applies to su &ective states as well. This is not the sort of conscious3unconscious distinction of 4reud as * understand it; it seems to have more similarity to the known separate minds within split rain patients, the ver al on the dominant side and the silent ut visually and manipulatively acute one on the other. 2ennett himself descri es the separate conscious states he is talking a out as eing like successive drafts of an article, hence the name. 7ow it is that we have the illusion of a single stream of consciousness with no perception of these concurrent drafts either as they are going on or after one is selected and others discarded (if indeed that is what he thinks happens) * do not understand. 7owever, it does seem to rule out their eing like different personalities in a person with multiple personality disorder since in those cases some of the personalities are persistent and very much aware of each other even though only one may e in control at a time. The final point a out 2ennett,s ideas that arouses )earle,s wrath is his re&ection of the possi ility of philosophers, <om ies and alleged acceptance of the possi ility of strong '*. The philosophers, <om ie is a common creature in consciousness researcher thought experiments. )upposedly he or she would e like us in all ways including showing emotions and talking a out consciousness, yet they would in fact e totally unconscious and reactive or ro otic. )ome who have considered them accept the idea that they could in theory exist, though * know of no one who has ever suggested that they exist in practice; others have argued that without consciousness such ehavior would simply e impossi le. 2ennett argues that any such eing, exhi iting such ehavior would inevita ly be conscious, which seems at the least to fly in the face of what )earle,s +hinese room thought experiment, discussed earlier, is supposed to prove. )earle also claims that 2ennett is a eliever in strong '*. )trong '* claims that it is possi le to uild computer hardware3software systems that are truly intelligent or conscious. 0eak '*, claims that such systems, however they might appear would only e simulating intelligence or consciousness. *t is logically possi le to split the difference of this, to claim for example that computer systems could e genuinely intelligent (strong '*) ut never do more than simulate consciousness (weak '*). )earle however, elieves that neither consciousness nor genuine intelligence can ever e manifested, at least y the sort of computer systems we now know. 7is reasons for those eliefs * have also tried to explicate in my earlier discussion of )earle and his +hinese room. * have never come across a passage in any of 2ennett,s works that * have read or scanned which seemed to explicitly either accept or re&ect strong '* as the hypothesis relates to either intelligence or consciousness. 7owever, * suspect that )earle is right and that 2ennett would or does accept it in oth cases. 0hat is certain is that )earle claims to e sure that 2ennett accepts the strong '* possi ility as it relates to consciousness. 7e seems to take this as a final proof in his argument that 2ennett does not really elieve in consciousness at all, however paradoxical that may seem, ecause he claims to elieve that 2ennett has a severe case of intellectual pathology as mentioned a ove. 4. 2avid Chalmers! $the hard problem% and panpsychism (The +onscious ;ind)

1@

5iven that )earle so dislikes 2aniel 2ennett, a leader of the ' team ro o/geeks, one might expect him to like 2avid +halmers, a leader if the % team ad scientists, ut not so. This chapter on +halmers is, * think, the worst in this entire ook, saved somewhat y the inclusion of a response y +halmers to the review at its end. 4or an excellent summary of +halmers views see the transcript of his dialog with )usan %lackmore in a ook we discussed in this group a few months ago, +onversations on +onsciousness. *n this chapter )earle manages to drastically misstate not only +halmers views ut also some rather asic facts a out oth ehaviorism and functionalism as twentieth century intellectual movements. )pecifically and to start with, * have never come across anything y or a out any ehaviorist denying the existence of consciousness, though * have come across that charge many times. 0hat ehaviorists did consistently deny, rightly or wrongly, was the legitimacy of su &ective experience as such as valid raw data for scientific investigations, a point &ust ela orated on with regard to 2aniel 2ennett. 4unctionalism, which is still around, a stracts analogous properties from different situations and tries to explain those similarities as alternate ways of achieving analogous functional results. )earle mentions electric and mechanical clocks in passing and showing how similar su /functions in such physically different devices are achieved and why, for example, explaining why oth need of some actuator with a very steady eat, whether it e a pendulum or an alternating current from a wall socket. 4unctionalism demonstrates correlations not causation etween examined systems. ;echanical clocks with their pendulums did not cause electric clocks that plug into '+ outlets, rather oth needed parts of some kind to accomplish an analogous function. +halmers is most noted for having coined the phrase the hard pro lem to contrast the difficulty of explaining why we have consciousness at all, rather than eing consciousless philosophers <om ies of the sort &ust previously discussed, with the easy pro lems. $elating conscious perceptions and intentions to external stimuli and ehaviors are the (relatively) easy pro lems y his and most others, standards. +uriously, )earle never even mentions the hard pro lem when discussing +halmers ook, and * wonder why. *n his response to )earle, printed in this ook, +halmers descri es himself as eing an agnostic a out most aspects of the mind/ ody pro lem, and a very thoroughgoing one he is. To return to the philosophers, <om ies who ehave like us ut without any speck of consciousness +halmers concedes that we can imagine them ut asks whether they are possi le in this physical universe and if not would they e possi le in some other universe with different physical laws. The elief that some assem lages of matter can exhi it mental properties as well as physical ones is what is now called property dualism, and as a way of trying to simultaneously hold on to oth materialism and consciousness has een around for some time. %ertrand $ussell proposed such a dual aspect theory in explicit analogy to the wave3particle dualism thus then eing elucidated in .uantum physics as far ack as the 1=98,s.

1D

*n investigating &ust how far down such dualism might e a le to extend, eyond the level of lower animals +halmers asked whether inanimate o &ects might also have a conscious aspect of some sort. *n the ook reviewed y )earle as well as in other places he has tried to imagine &ust what the phenomenological life of a thermostat might e like, very simple he concluded. )peculating still further, he asked whether consciousness might e a fundamental property of this universe like space and time and coextensive with them. This is where the panpsychism comes inA if everything including empty space might have a conscious aspect of some sort wouldn,t that imply a universal consciousness and why would we not e aware of itC )uch a consciousness for most of the universe would e very uninteresting he felt, ut when entangled with informationaly complex entities such as ourselves or even lower animals it might manifest as the sort of consciousness that we would recogni<e as such. (ust what defines informational complexity and why consciousness would ecome entangled with it does not seem clear, and )earle is .uite right a out that. 7owever, +halmers does not present this as a con&ecture, merely as a logical possi ility that seems hard to exclude simply on the asis of logic and challenges us to discover &ust how it might e re&ected. *n sum, +halmers does not affirm so/called su stance dualism (minds and rains connected ut composed of different su stances) or mentalism (the idea that everything is really mind with matter eing merely an illusion). 7owever, in his demand for good reasons to exclude such possi ilities he examines them more closely than anyone else * know of doing consciousness studies today from a secular perspective. 5. 6srael Rosenfield! the Self , Body 6ma(e (The )trange, the 4amiliar and 4orgotten) $osenfield, a clinical neurologist, uses different source materials from others reviewed y )earle, clinical case histories of the results of various forms of neural damage, and uses them to address a somewhat different topic, not consciousness as such ut the normal, seemingly continuous sense of self. This is a topic with a long history in consciousness studies, !sychologist 0illiam (ames was writing a out it more than a century ago and with the return of interest in consciousness studies within the last couple of decades a num er of researchers have grappled with it, usually with difficulty. )ome have considered it an achievement to prove to their own satisfaction that there is no such thing as the self, often relying on common experiences reported in meditative states as evidence. 2aniel 2ennett has suggested that while the self seems real it is in fact an illusion, a sort of virtual, se.uential machine that runs atop the multiprocessing rain. $osenfield sees a coherent self as eing a product of dynamic memory uilt around changes and sta ilities in ones ongoing, physical ody image. This claim of ody image as central to sense of self and even more is not new.. 7u ert 2reyfus in his 0hat +omputers ()till) +an,t 2o ooks, previously mentioned, centered much of his argument on the assertion of "a<i philosopher ;artin 7eidigger that not only consciousness ut intelligence re.uires the existence of a ody. Mikewise, a num er of '* researchers have felt that providing real or simulated ro ot odies to their computer programs was essential if those programs were to e a le to learn3evolve in any deep sense.

1>

2evelopmental studies of infants oth efore and after irth have usually reached similar conclusions, and psycho/linguist 5eorge Makoff has tried in a rather large ook to demonstrate that even seemingly very a stract ranches of mathematics rely on, often unconscious, ody image metaphors for more than is customarily reali<ed. The negative evidence, if any, regarding the importance of ody image and coherent sense of self to intelligence and consciousness seems to come from recent studies of memory. The episodic memory, which is what one thinks of when thinking of cases of amnesia appears to e only one of several types which seem to e associated in differing ways with differing rain regions. -thers commonly recogni<ed today are the semantic, the procedural and the emotional as well as the working memory, which does play such an important part in $osenfield,s examples. 4ew seem to consider in the case of the prover ial amnesiac of 7ollywood films how it is that the person who has forgotten even his or her name is nonetheless a le to use their native language and tie their shoes without difficulty. Conclusion7addendum! findin(s from modern psycholo(ists/ consciousness studies 0hile )earle,s reviews unfortunately do not contain any examples of conscious research carried out y psychologists, we are fortunate that there is an excellent example of such research in an article entitled ;agic and the %rain in the current (2ecem er 988:) issue of )cientific 'merican. 4urthermore, it seems to dovetail nicely with the matter that )earle rings up in his conclusion as an important topic for future conscious research, the phenomenon of lindsight. This is a neurological condition in which an individual can not consciously see out of all or part of his or her visual field yet can relia ly answer .uestions a out things which are going on in that lind area such as how many fingers am * holding upC The lindsighted person will insist that he or she can not see what is going on in that area, ut when asked to guess will usually guess correctly and e very much surprises at how often those guesses turn out to e correct. The neural areas and pathways whose damage produces the lindsight phenomenon are now pretty well understood, ut a relevant .uestion * want to ring up that )earle seems to have missed. 0ho if anyone in this person,s rain is conscious of what is going on in that reportedly lind areaC "ot the self doing the ver al reporting o viously, ut does that mean that there was no consciousness in the lindsighted person,s rain aware of what was going on in that areaC $emem er the results of the tests that can e performed on persons whose interhemispheric connections have een cut. -nly one side is a le to speak ut oth sides seem to e a le to think and to feel emotions. $emem er too the .uestions raised y oth 2aniel 2ennett and 2avid +halmers a out &ust how many consciousnesses may e simultaneously active in even a normal rain. $emem er finally that in the case of the split rain patients the speaking side usually has and elieves reasona le sounding ut factually .uite incorrect interpretations of ehaviors initiated y the other side, other half. *n the )cientific 'merican article an experiment on perfectly normal people, that seems nonetheless to remind one of those split rain experiments. *n the experiment descri ed in

1F

the current article each su &ect was shown a pair of photographs and asked which one he or she found more attractive. The photographs were then covered momentarily and switched y the experimenter, using slight of hand. 6ach su &ect was then asked to explain why he or she found the person in the selected photograph (really the person in the photograph not selected) to e more attractive. -nly a out one forth of the su &ects reali<ed that the photographs had een switched, even though they had een out of sight only momentarily. -f those who did not catch on to the switching almost all were a le to give detailed, introspective reports of why they had &ust made the selection that they did, even though they had, in fact, &ust made exactly the opposite selection. 4our other types of cognitive illusions are also descri ed. +hange lindnessA a viewer misses changes made to a scene during a rief interruption. * have seen some extraordinary examples of this in ooks on the su &ect where two photographs were displayed side y side. $epeatedly, * was una le to spot what, after the fact, were indeed o vious changes from one to the other even though the only interruption was the movement of my eyes from one side of the page to the other. *nattentional lindnessA a person does not perceive items that are plainly in view. 6xampleA a person in a gorilla suit wanders across a scene of a group playing asket all, stopping riefly in the middle to pound its chest, ut goes completely unnoticed. #es, this is a real experiment that has een done repeatedly with variations. +hoice lindnessA the experiment with the photographs, &ust descri ed, is an example of this one, and finally illusory correlationA a stage magician waves a wand and a ra it appears. -ne * find even more dramatic was an experiment in which an experimenter with a clip oard pretended to e taking a survey. !art way through, two men, also part of the experiment, carrying a large oard passed etween the supposed interviewer and the person eing interviewed. -ut of sight, the initial interviewer swung up to hang ehind the oard and a different interviewer, differently dressed swung down and, once the oard had passed, continued the interview as if nothing had happened. *ncredi ly, when asked a out the incident &ust afterward more than half the su &ects had not noticed or could not remem er that the switch of clip oard wielding interviewers had occurredJ 0hat * take from all these examples as well as the lindsight and split rain results is that 2ennett is right, though perhaps not in the sense he meant it, that consciousness isn,t what you think it is. 'lso, that the difference etween consciousness and unconsciousness is not nearly so clear cut as )earle assumes.

1B