1 ing Qualia: Know Ja ckson A Reply to chland (in put ational Perspective , MIT Press, 1989) Paul M. Chur A Neurocom erning the direct introspection of brain states (1985b) In a recent paper conc I leveled ledge ent." At stake was his bol d "know argum Jackson's three criticisms against Frank for all mental phe nom ena. of mind can pos claim that no materialist account sibly account e criticisms in his 1986 . It is to thos e replies, and to the issues Jackson has replied to thos pted them, that the present chapter is directed. that prom vocation The Persistent Equi 1 Jackson conc edes the criticism I leveled at my ow n statement of his argum ent -- specifically, that it invol ation on 'kn ow s about ' -- but he insists that my ves an equi voc truction doe recons represent the argum ent he wishes to defend. I accept his s not instruction, turn my attention to the sum mary of the argument he provi des at the and tom of page 293. will recall, has been raised in innoc ence of any color bot Mary, you has an exha uroscience. mand of ne rience, but expe ustive com Mary (before her release) know s everything phys ical there is to know about other (1) e. peopl (2) s not know everything there is to know about other Mary (before her release) doe e (because she about som ething peopl them on her release). learns There are truths about other peopl e (and herself) which escape the phys icalist (3) story. 1 Regimenting sake, yields the follow ing. further, for clarity's (1) ( x)[(Hx & Px)  Kmx] ∀ ∃ x)[Hx & ~Kmx] (viz., "what it is like to see red") (2) ( ∴ ∃ x)[Hx & ~Px] (3) ( m = Mary; Kyx = y know s about x; Hx = x is about persons ; Here = x is about Px som phys ical in character; and x range s ove r "know ables," generous ly cons ething trued so as not to beg any que stions about whether they are propos itiona l or otherwise in nature. Thus expr essed, the argum ent is formally valid: the salient move is a modus tollens that appl ies the second unc t of premise (2), '~Kmx', to the waiting cons eque nt of conj 'Kmx'. are whether the premises are jointly true, and stions now premise (1), The que arances. Here lam r the crucial notion is uni whethe al in bot h of its appe 'Kmx' voc surprised that Jackson sees any progr ess at all with the above formulation, since I cont inue to see the same equi voc ation found in my earlier casting of his argument. Specifically story about Mary's color - , premise (1) is plausibly true, within Jackson's 1 [Translation: (1) Fo r all knowables x, if x is about human s and physical , then Mary knows x. (2) There is somet hing x that is about human s but Mary does not know it. (3) Therefo re, there is hing x about humans that is not physical .] somet 1

2 free upbr ingi of 'kno ws about ' that casts the obj ect of ng, only on the interpretation ledge as som propos itiona l, as som ething adequa tely expr essible in an know ething e. Mary, to put it briefly, gets 100 every written and oral ish sentenc Engl percent on the truth of any given sentence about e on ical exam; she can pronounc the phys characteristics of persons , especially the states of their brains. Her "know ledge by of phys i cal facts about persons is without lacuna e . description" Premise (2), how ever, is plausibly true onl the interpretation of 'know s about ' that y on itiona ect of know ething nonp ropos as som l, as som ething inarticulable, casts the obj ledge ething that is non – as som - va luable. What Mary is missing is som e form of truth "know ledge by acqua intance," acqua intance with a sensory character, prototype , or uni versal, perhaps. Given this prima facie difference in the sense of 'know s about ', or the kind of ledge appe in each p remise, we are still looki ng at a prima facie case of an know aring lid by reason of equi voc ation on a critical term. Replace either of the 'K's argum ent inva with a distinct letter, as acknow ledgm above ty demands , and the ent of the ambigui inference to (3) evapor . The burden of articulating som e specific and uni tary sense of ates s about ', and of argui ng that bot h premises are true unde r that interpretation of the 'know epistemic ope rator, is an undi scharged bur den that still belongs to Jackson. It is also a heavy n, since the resour ces of modern cogni tive neurobi ology already burde de us with a plausible account of what the difference in the two kinds of know ledge provi amount s to, and of how it is pos sible to have the one kind without the other. Let me illustrate with a that at issue, so as not to beg any que stions . case distinct from Any com petent gol fer has a detailed representation (perhaps in his cerebellum , perhaps motor in his motor cortex) of a gol f swing. It is a it cons representation, and titutes his "know ing " to execute a prope r swing. The same gol fer will also have a discursive how of a golf swing (perhaps in his language cortex, or in the neighbor ing representation parietal regions al and f swing or perhaps ), which allow tempor s him to describe a gol are qui te distinct. aper. The motor and draw it on p the discursive representations Localized brain traum either one while sparing the other. a, or surgery, coul d remove t of that, an inarticulate gol f champion might have a supe rb representation of the Shor but a feeble representation of the latter kind. And a phys icist or spor former kind, ts phys st might have a detailed and penetrating representation of the mechanics of a iologi swing, yet be una ble to duf f the ball more than ten feet because he lacks an good and te motor representation, of the desired behavioral seque nce, in the brain areas that adequa cont rol his limbs. Inde phys icist is chroni cally disabled in his motor capacities, ed, if our motor representation he may have no f swing whatsoever. In one medium of of a gol representation, his representationa l achievements on the topi c may be com plete; while in anot her medium he has not hing. of representation, already rast between "know " and "know ing that" is one how acknow ledge d A cont ing mon sense, and thus in com surprising that som e of the earliest replies to it is not Jackson' s argum ent (Nemirow 1980; Lewis 1983) tried to portray its equi voc ation in these familiar terms, and tried to expl icate Mary's missing know ledge solely in terms of ze red, to imagine red, etc.). While the her missing or more abi liti es (to recogni e one som appr oach is well motivated, this binary distinction in type s of know ledge barely begins to sugge st the range and variety of different sites and type s of internal representation to be found in a nor mal brain. There is no reason why we must be bound by the crude divisions ication of itive expl of our pre scientific idiom s when we attempt to give a precise and pos 2

3 the equi voc displayed in Jackson's argum ent. And there are subs tantial grounds for ation g a som ewhat different story conc the sort of nondi scursive know ledge at tellin erning and qua entarily aside, I will tell such a story . caution lification mom issue. Putting (i.e., with three type s of retinal cone In creatures with trichrom atic vision ), color rmation is code d as a pattern of spiking freque ncies across the axona l fibers of the info ellular subs ystem of the opt ic nerve. That massive cable of axons leads to a second parvoc ation of cells in a central body called the lateral geniculate nuc leus (LGN) , whos e popul l projections axona lead in turn to the several areas of the visual cortex at the rear of the brain's cerebral hemisphe res, to VI, V2, and ultimately to V 4, which area appe ars to be ed to the processing and representation of color i nformation (Zeki 1980; especially devot Van Essen and ell 1983; Hube l and Livings Mauns 1987) . Hum an cogni tion divides a tone smoot h cont inuum of color input s into a finite num ber of prototypi cal categor ies. The laminar structure at V4 is perhaps the earliest place in the pr ocessing to which hierarchy ascribe that familiar taxonom petent to make reliable color we might y. A creature com discriminations has there develope d a representation of the range of familiar colors, a that appe ars to cons ist in a specific co nfigur representation of weight ed syna ptic ation conne ctions meeting the millions of neurons that make up area V4. That conf iguration of syna ptic weight s partitions the "activation space" of the neurons in area V4: it partitions that abstract space into a structured set o paces, one for each f subs cal color. Input s from the eye will each occasion prototypi a specific pattern of activity across these cortical neurons , a pattern or vector that falls within one of thos e subs paces. In such a pigeon holing, it now appe ars, doe s v isual recogni tion of a color cons ist (see chapters 5 and 9 for the general theory of information processing here appe aled to). This sessing recogni tion depends upon the creature pos - a prior representation a learned conf igur of syna pses meeting the rele vant pop ulation of cells - that antecedently ation the creature's visual taxonom y so it can respond appr opr iately partitions selectively and the retina and arriving from LGN . to the flux of visual stimulation is not remotely propos itiona l or discursive, but it is This distribut ed representation , even thos e without any lingui stic atic animals have one entirely real. All trichrom rently makes pos sible the many abilities we expe ct from color - capacity. It appa petent com creatures: discrimination, tion, imaginatio n, and so on. Such a representation is recogni ably what a person y in stunt upbr ingi ng woul d lack, or pos sess onl presum ed with Mary's plete form. Her representationa l space within the relevant area of neurons woul d or incom ain onl cont pace for black, whit e, and the intervening shades of gray, for the y the subs visual examples that have shaped her syna ptic conf igur ation were limited to these. There is thus more than just a clutch of abilities missing plex in Mary: there is a com - framework t hat deserves to be called "cogni tive" - that she representation a processing ed form. There is inde ed som ething she "doe s not either lacks or has in reduc ." know Jackson' e, is thus true on these whol s premise (2), we may assum ly materialist assum ptions . These same assum ptions are entire ly cons istent with the further assum ption that elsewhere in Mary's brain in the langua ge areas, for example - she has stored a detailed - and ustive set of discursive, propos itiona l, truth - valuable representations of even exha what goe s on in peopl e's brains du ring the expe rience of color, a set she has brought into pleted cogni being the exha ustive reading of author itative texts in a com by tive neuroscience. She may even be able to expl ain her ow n representationa l deficit, as s premise (1), we may thus sical detail. Jackson' sketched above , in com plete neurophy 3

4 assum e, is also true on ptions . these whol ly materialist assum date for the correct story of sensory codi The view sketched above ng is alive candi and it is true, it is sibility. But whether or not at least a logical pos sensory recogni tion. entirely phy sical mode Accordingl y, what we have sketched here is a cons istent but l (i.e., conc lusion is false) in which bot h of Jackson's premises are l in which Jackson's a mode opr iate interpretation. They can hardly entail a conc true unde then, that r the appr lusion, istent with phys pos sibility, on pur ely phys icalist assum ptions , is incons icalism. Their com of representation erically different medium resides in the different character and the num argu ment, to re file the charge, equi voc ates premises. Jackson's at issue in each of the two 'know '. on s about 2 Other Invalid Instanc es ent form with one inva lid instance can be expe cted to have others. This was the An argum poi nt of a subs ection in my 1985b paper: if vali d, Jackson's argum ent, or one idiary obj d also serve to refute the possibility of subs e dual ism. I did not formally parallel, woul tanc nt with not I accept respons ever, and ess my poi ibility for there expr able clarity, how te missing my intention. Let me try again. Jackson' s qui nt is that the canoni cal presentation of the know ledge argum ent, as The basic poi lined on p. 67 above , woul d be just as valid if the predicate term 'P' were everyw here out 'E'. the resulting premises woul d be just as plausibly tru e if replaced by And 'E' stood (1) som ething ectopl asmic in character' (where 'ectopl asm' for 'is about is an arbitrary name for the dua list's nonphys ical subs tance), and (2) the story is altered so that Mary becom es an exha ustive expe rt on a com pleted ectopl asmic science of hum a n nature. The plausibility woul d be com it, because a long discursive lecture on parable, I subm ective, statable, law - rned prope rties of ectopl asm, whatever they might be, the obj gove acquai useless, Mary to d be exactly as useful, or - by - in helping n tanc e “what it woul know as woul d along discursive lecture on the obj ective, statable, law - is like to see red,” rned prope gove ical matter of the brain. Even if subs tance dua lism were rties of the phys ectopl ly parallel 'know ledge true, therefore, and asm were its heroic principal, an exact ent" woul d "show " that there are som e aspects of cons cious argum ness that must forever escape the asmic story . ectopl Given Jackson's antiphys icalist intentions , it is at least an irony that the same form of argum ent shoul d incidentall y serve to blow subs tance dua lism out of the water . lam hardly a subs list (and neither is Jackson) , I do regard subs tance Though tance dua lism as a theoretical pos sibility, one that might conc eivably succeed in expl icating dua the psychol cal ont ology of com ogi rlying prope rties and law - mon sense in terms of the unde gove rned behavior of the nonm aterial subs tance it pos tulates. And I must protest that the parallel know argum ent against subs tance dualism woul d be wildly unf air, and for ledge the very same reason that its analogue against phys icalism is unf air: it woul d equi voc ate on s about '. It woul d be no more effective against dua lism than it is against 'know materialism. The parallel unde r examination cont ains a further lesson. If it works at all, Jackson's e defect that is uni to que ar gum ent works against phys icalism not because of som 4

5 phys icalism; e no am ount of discursive know ledge , on any topi c, will it works becaus titute the nondi scursive form of know that Mary lacks. Jackson's argum ent is cons ledge ance of an indi tioni ant ireduc inst st form of argum ent. If it works at all, scriminately one propos ed reduc tive, discursive, obj ective account of the will work against any an analog ective expe rience, no matter what the reduc ing theory might happe nature of our subj to n be. I see this as a further sym of the logi cal pathol ogy described earlier . Since the ptom ent "works" for reasons hing essential to do with phys icalism, it argum that have not d "work" against the expl anatory aspirations of other ont ologi shoul . And so it es as well "doe s." The price of embracing Jackson's argum ent is thus dramatically highe r than first appe ars. For it makes any scientific account of our sensory expe rience entirely impos sible, no employe d. matter what the ontology A Genui Argum qui voca l Know ledge 3 ent nely None eciate the equi voc ation more deeply if we expl ore a version of Jackson's We can appr argum ent that doe s not equi voc ate on 'know s abou t'. The equi voc ation can qui ckly be closed, if we are determined to do so, and the results are revealin g. Given that the em is a variety in the pos ing, let us simply rewrite the argum ent probl sible forms of know with suitable qua ove r the relevant forms of know ing. The first premise must ntification assert that, for any know able x, and for any form f of know l edge , if x is about hum ans and t x is phys ical in character, then Mary know s(f) abou The second x. premise is modified in 2 the same mode the conc lusion is identical. Canoni cally, st fashion, and (1') x)( ∀ f)[(Hx & Px)  K(f)mx] ( ∀ ∃ f)[Hx & ~K(f)mx] ∃ ( (2') x)( ∴ (3') ( ∃ x)[Hx & ~Px] ent is also formally valid, and its premises expl icitly encom pass whatever This argum ing. What can we say about ss? variety there may be in forms of know its soundne e that Mary has had the upbr ng described in Jackson's story , and thus lacks Assum ingi know ledge - by - acqua intance with "what it is like to see red." Premise (2') will then be any true, as and for the reasons story requires. What will be the truth value of that Jackson's on these assum ? premise (1') ptions Premise (1') is now a very strong claim inde ed, much stronge r than the old premise (1), and a materialist will be sure to insist that it is false. The reason offered will be that, because of her deprived upb ringi te clearly lacks one form of know ledge of a ng, Mary qui phys e. Specifically, she lacks a prope r conf iguration of certain ical aspect of peopl ptic conne ctions meeting the neurons in the appropr iate area of her visual cortex. She syna thus lacks an appr iately partitione d activation vector space across thos e neurons , an d opr representation, at that site, of the full range of sensory codi ng vectors therefore has no that might som eday com e from the retina and the LGN. In other words, there is n som ical about persons (their color sensations , or identically, their codi phys g ething 2 [T ran slation: (1) Fo r all knowables x and forms of knowledge f, if x is about human s and physical knows x in the f way. (2) There is a knowable x and a form of knowledge f , then Mary such that x is about human is does not know it in the f way. (3) Therefo re, there s and Mary .] s that is not physical somet hing x that is about human 5

6 vectors in their visual pathw ays), and there is som ledge (an antecedently e form of know d prelingui stic taxonom ledge of that partitione y), such that Mary lacks that form of know y, premise (1') is false and lusion (3') is not . Accordingl aspect of persons the conc sustained. a materialist's poi nt of view, it is obvi ous that (1') will be false on the From ptions of Jackson' s story. For that story denies her the upbr inging that nor mally assum s and shapes the developm ent of the relevant representation ac ross the provoke opr ation of cortical neurons . And so, of course, there is a form of appr iate popul ledge , of a phys ical aspect of persons , that Mary doe s not have. As just illustrated, know ledge the materialist can even specify that form of know its obj ects, i n neural terms. , and But this means that premise (l'), as prope ntified at last, is false. Mary doe s not rly qua have know ledge of everything phys ical about persons , in every way that is pos sible for her. (That is why premise (2') is true.) o gua rantee that the materialist's account and There is, of course, n of sensations tion the expe rimental and theoretical evidence for a sensory recogni is correct (although cont inue s to accum ulate). But neither is Jackson view of this general kind ition to in a pos insist that it d beg the very que stion at issue: whether must be mistaken. That woul sensory qua ically distinct class of phe nom ena beyond the scope of lia form a metaphys phys ical science. To sum marize, if we write a deliberately non equi voc al form of Jackson's argum ent, - that qua ntifies appr opriately ove r all of the relevant forms of know ledge , then the one first premise must almost certainly be false unde r the condi tions of his ow n story .So, at . Jackson's any pos ition to argue rate, is the materialist in a strong expr es sed hope for highl realized in (1'). The original premise (1) was of course y plausible premises" is not much more plausible. But it failed to sustain a valid argum ent, and y it was plausible onl ess all the relevant forms of know led because it failed to addr ge. erting - Person Account into a First Conv Person Account a Third 4 - ection to Jackson was aimed more at breaking the grip of the ideology My final obj his argum ent than at the argum ent itself. That ideology include s a dom ain of behind rties prope the qua lia of subj ective expe rience - that are held to be metaphys ically distinct - the obj rties addr ical prope from essed by orthodox science. It is not a surprise, ective phys this view, that one might know all phys ical facts, and yet be igno rant of som e then, on dom ain of these nonphys ical qua lia. The cont rast between what is know n and what is not know n simply reflects an antecedent metaphys in the furniture of the world. ical division her way to look one that finds no such division. Our But there is anot at the situation, zing a range of (currently) inarticulable features in our subj ective capacity for recogni rience is easily expl expe I materialist principles; the relevant sketch appe ars ained on earlier in this essay and um e (chapter 5, section 7). Our discursive elsewhere in this vol inarticulation of thos e features is no surprise either, and signi fies not hing about their metaphys ical status (chapter 10, 5). Inde ed, that veil of inarticulation may itself section suitable learning. be swept aside by able spo ntaneous ly to repor t about What we are now our internal states and cogni tive activities need not define the limit on what we might be opriate able to repor aneous ly and accurately, if we were taught a more appr t, spont conc eptual scheme in which to expr ess our discriminatio ns. In closing, let me again urge pos sibility. on Jackson this exciting 6

7 The intricacies of brain func tion ectively opa que to us now , but they need may be subj remain that way forever. Neuroscience may appe ar to be defective in provi a not ding - " of mind, but onl y familiarity of idiom and spont aneity of ely "third person account pur conc red to make it a "first - person account " as well. What makes eptual respons e are requi account a 'first - person account " is not an ent of that account , but the fact that on e the cont has ane ous conc eptual ization in introspection and learned to use it as the vehicle of spont - We all of us, as children learned to use the framework of current folk self description. ogy in this role. But it is entirely pos sible for a person psychol ure to learn and use or cult som e other framework in that role, the framework of cogni tive neuroscience, perhaps. Given a deep and practiced familiarity with the developi ng idiom s of cogni tive neurobi ology, learn to discriminate by introspection the codi ng vectors in our we might l pathw patterns across salient neural popul ations , and internal axona ays, the activation myriad other things besides. d that ever happe n, it woul d then be obv Shoul to everyone who had made the ious conc eptual shift that a com pleted cogni ti ve neuroscience woul d cons titute not a pinched and exclusiona ry picture of hum an cons cious ness, one blind to the subj ective dimension of self as Jackson's ent sugge sts. Rather, it woul d be the vehicle of a grand argum recons truction and expa nsion of our sub jective conscious ness, since it woul d provi de us with a conc eptual framework that; unl ike folk psychol ogy, is at last equa l to the kinematical and dyna mical intricacies of the world within. (See also chapter 1 of this um vol e l and Chur chland 1979, section 16. ) Real precedents for such a reformation can be drawn from our n history .We did not ow act with a metaphys ically distinct dimension d seeing lose cont of reality when we stoppe strewn qui re each time we looke d to the - ntessential crystal sphe an immutable sparkle t and giant stars heavens, and began to see instead an infinite space of gas and dus l attractions and violent nuclear processes. On the cont rary, we gravitationa structured by now see far more than we used to, even with the unaided eye. The diverse " colors of the us to see directly their absolute temperatures. Stellar temperature is a func tion stars allow stellar masses. The intrinsic lum inos ity or of stellar mass, so we are just as reliably seeing ness of a star is tight atures, and thus is also visually bright ly tied to these same fe matter how bright or faint the star may appe ar from Earth. Appa rent available, no bright ness is visually obvi also, of cour se, and the cont rast between the appa rent and ous nesses gives you the star's ro ugh distance from Earth. In this way is the the intrinsic bright character and three - dimensiona l distribut ion of complex stellar obj ects in a vol um e of interstellar space hundr eds of light r una ided years on aside made visually available to you your n back yard, giv en onl y the right conc eptual framework for grasping it, eyes from ow obs ervationa l practice in using that framework. From within the new framework, one and finds ficance in expe riential details that hitherto went largely or entirely asystematic signi iced (com unnot Feyerabend 1963b ). pare The case of inne r space is pot entially the same. We will not lose cont act with a metaphys of self when we stop introspecting inarticulable qua lia, ically distinct dimension and start introspecting "instead" sensory codi ng vectors and sund ry activation patterns within the vector spaces of our ution in accessible cortical areas. As with the revol astronom rather we shoul d welcom e as metaphys ically liberating, y, the prospect is one ossible. ically imp than deride as metaphys ically irrelevant or metaphys 7

Related documents