WHAT

MARY DIDN'T

KNOW

291

throughblack-and-white books and throughlectures reon black-and-white television. In thiswayshe learns everything thereis to knowabout the physicalnature of the world. She knowsall the physicalfactsabout us and our environment, in a wide sense of 'physical'whichincludeseverything in completed physics, chemistry, and neurophysiology, and all thereis to know about the causal and relationalfactsconsequent upon all this,includingof course functional roles. If physicalism is true,she knowsall thereis to know.For to suppose otherwise is to suppose thatthereis more to know than every physical fact, and that is just what physicalism denies. Physicalism is not thenoncontroversial thesisthattheactual world is largely physical, but the challenging thesisthatit is entirely physical. This is whyphysicalists musthold thatcompletephysical knowledge is complete knowledgesimpliciter. For suppose it is not complete: thenour worldmustdiffer froma world,W(P), forwhichit is complete,and the difference must be in nonphysical facts;for our world and W(P) agree in all mattersphysical.Hence, physicalism would be falseat our world [thoughcontingently so, forit would be true at W(P)].I It seems, however,thatMarydoes not knowall thereis to know. For whenshe is let out of the black-and-white room or givena color television, she willlearn whatit is liketo see something red, say.This is rightly describedas learning-she willnot say "ho, hum." Hence, physicalism is false.This is theknowledgeargument againstphysicalism in one of its manifestations.2 This note is a replyto threeobjections to it mounted by Paul M. Churchland.t v slayed a
lThe claim here is not that, if physicalismis true, only what is expressed in explicitly physicallanguage is an itemof knowledge.It is that,ifphysicalism is true, thenifyou knoweverything expressed or expressiblein explicitly physicallanguage, you knoweverything. Pace Terence Horgan, "Jacksonon PhysicalInformation and Qualia," Philosophical Quarterly, XXXIV, 135 (April 1984): 147-152. 2 Namely, that in my "Epiphenomenal Qualia," ibid., xxxii, 127 (April 1982): 127-136. See also Thomas Nagel, "What Is It Like to Be a Bat?", Philosophical Review, LXXXIII, 4 (October 1974): 435-450, and Howard Robinson, Matter and Sense (New York: Cambridge, 1982). t "Reduction, Qualia, and the Direct Introspectionof Brain States," thisJOURNAL, LXXXII, 1 (anuary 1985): 8-28. Unless otherwisestated, futurepage references are to this paper. 0022-362X/86/8305/0291$00.50 (? 1986 The Journal of Philosophy,Inc.
* I am much indebted to discussionswithDavid Lewis and withRobert Pargetter.

COMMENTS AND CRITICISM WHAT MARY DIDN'T KNOW*

r AARY is confined to a black-and-white room,is educated

292

THE JOURNAL I. THREE

OF PHILOSOPHY

CLARIFICATIONS

The knowledgeargumentdoes not rest on the dubious claim that logically you cannot imaginewhatsensingred is likeunless you have sensed red. Powersof imagination are not to the point.The contentionabout Maryis not that,despiteher fantastic graspof neurophyselse physical, she could not imagine whatit is iologyand everything like to sense red; it is that,as a matterof fact,she would not know. But if physicalism is true,she would know; and no great powers of imagination would be called for. Imaginationis a faculty thatthose who lack knowledgeneed to fall back on. Secondly,the intensionality of knowledgeis not to the point. The argument does not reston assumingfalsely that,ifS knowsthata is F and ifa = b, thenS knowsthatb is F. It is concernedwiththenature of Mary'stotalbody of knowledgebefore she is released: is it complete, or do some factsescape it?What is to the point is thatS may knowthata is F and knowthata = b, yetarguablynot knowthatb is F, by virtue of not being sufficiently logicallyalert to follow the consequences through.If Mary's lack of knowledgewere at all like in it. But it is veryhard this,therewould be no threatto physicalism to believe thather lack of knowledgecould be remedied merelyby her explicitly following throughenough logical consequences of her vast physicalknowledge. Endowing her with great logical acumen and persistenceis not in itselfenough to fill in the gaps in her knowledge.On being let out, she willnot say "I could have worked all thisout beforeby makingsome more purelylogical inferences." the knowledgeMary lacked whichis of particularpoint Thirdly, forthe knowledgeargumentagainstphysicalism is knowledgeabout theexperiences not about her own. When she is let out, she ofothers, has new experiences,color experiencesshe has neverhad before. It is not, therefore, an objection to physicalism that she learns somethingon being let out. Before she was let out, she could not have knownfactsabout herexperienceof red,fortherewereno such facts alikecan agree on. After to know.That physicalist and nonphysicalist she is let out, thingschange; and physicalism can happilyadmitthat she learns this; afterall, some physicalthingswill change, for instance, her brain states and theirfunctionalroles. The trouble for is that,after Maiy sees her firstripe tomato, she will physicalism realize how impoverished her conceptionof the mentallifeof others has been all along. She will realize thattherewas, all the timeshe into the neurophysiolwas carrying out her laborious investigations roles of theirinternalstates, ogies of othersand into the functional somethingabout these people she was quite unaware of. All along theirexperiences (or manyof them,those got fromtomatoes,the

WHAT

MARY DIDN'T

KNOW

293

sky,. . .) had a featureconspicuous to thembut untilnow hidden fromher (in fact,not in logic). But she knew all the physicalfacts about themall along; hence, whatshe did not knowuntilher release is not a physicalfactabout theirexperiences.But it is a factabout them.That is the trouble for physicalism.
II. CHURCHLAND'S THREE

(i) Churchland's firstobjection is that the knowledge argument containsa defectthat"is simplicity itself"(23). The argument equivocates on the sense of 'knowsabout'. How so? Churchlandsuggests thatthefollowing is "a conveniently version"of theknowltightened edge argument:
(1) Maryknowseverything thereis to knowabout brainstatesand their properties. thereis to knowabout (2) It is not the case thatMaryknowseverything sensationsand theirproperties. Therefore,by Leibniz's law, (3) Sensationsand theirproperties# brain statesand theirproperties (23).

OBJECTIONS

Churchland observes, plausiblyenough, that the type or kind of knowledgeinvolvedin premise 1 is distinct fromthe kindof knowledge involvedin premise2. We might followhis lead and tag thefirst 'knowledgeby description',and the second 'knowledgeby acquainthatthedisplayedargument tance'; but,whatever the tags,he is right involvesa highly dubious use of Leibniz's law. My replyis thatthe displayedargument maybe convenient, but it is not accurate.It is not theknowledgeargument. Take, forinstance, premise1. The whole thrust of theknowledgeargument is thatMary (beforeher release) does notknoweverything thereis to knowabout brain statesand theirproperties,because she does not know about certainqualia associated withthem.What is complete,accordingto the argument,is her knowledgeof mattersphysical.A convenient and accurate way of displaying the argumentis:
(1)' Mary(beforeher release) knowseverything thereis to know physical about other people. thereis to know (2)' Mary(beforeher release) does not knoweverything about other people (because she learns something about themon her release). Therefore, (3)' There are truths about otherpeople (and herself)whichescape the physicalist story.

What is immediately to the point is not the kind,manner,or type of knowledgeMary has, but what she knows. What she knows be-

294

THE JOURNAL

OF PHILOSOPHY

forehandis ex hypothesi everything physical thereis to know,but is it everything thereis to know?That is the crucial question. There is, though,a relevantchallenge involving questions about kindsof knowledge.It concernsthesupportforpremise2'. The case for premise 2' is that Mary learns somethingon her release, she acquires knowledge, and thatentailsthather knowledgebeforehand (what she knew,never mind whetherby description, acquaintance, or whatever) was incomplete. The challenge, mounted by David Lewis and Laurence Nemirow,is thaton her release Marydoes not learn somethingor acquire knowledgein the relevantsense. What Maryacquires when she is released is a certainrepresentational or imaginative ability; it is knowledgehow ratherthanknowledgethat. Hence, a physicalist can admit that Mary acquires somethingvery of significant a knowledge kind-which can hardly be deniedwithout admitting thatthisshowsthather earlierfactualknowledge is defective.She knewall that therewas to know about the experiences of others beforehand,but lacked an abilityuntil after her release.3 true that Marywill acquire abilitiesof various Now it is certainly kindsafterher release. She will,forinstance,be able to imaginewhat whatit is like,and be able to seeing red is like,be able to remember understand whyher friends regardedher as so deprived(something which,untilher release, had alwaysmystified her). But is it plausible that thatis all she will acquire? Suppose she received a lecture on about other minds while she was incarcerated.On her skepticism release she sees a ripe tomato in normal conditions,and so has a reactionis to saythatshe now knowsmore sensationof red. Her first about the kind of experiences others have when looking at ripe the lectureand startsto worry. Does tomatoes.She thenremembers she reallyknowmore about whattheirexperiencesare like,or is she indulgingin a wild generalizationfrom one case? In the end she is mistaken(even if,like decides she does know,and thatskepticism its errors).What so manyof us, she is not sure how to demonstrate about-her abilities?Surelynot; her repwas she to-ingand fro-ing What else resentational abilitieswere a knownconstantthroughout. then was she agonizing about than whetheror not she had gained factual knowledge of others? There would be nothingto agonize was all she acquired on her release. about if ability
3See Laurence Nemirow, review of Thomas Nagel, Mortal Questions, Philosophical Review, LXXXIX, 3 (July1980): 473-477, and David Lewis, "Postscriptto 'Mad Pain and Martian Pain'," Philosophical Papers, vol. I (New York: Oxford, 1983). Churchland mentions both Nemirow and Lewis, and it may be that he intended his objection to be essentially the one I have just given. However, he says quite explicitly(bottom of p. 23) that his objection does not need an "ability" analysisof the relevantknowledge.

WHAT

MARY DIDN'T

KNOW

295

I grantthatI have no proofthatMaryacquires on her release, as well as abilities,factualknowledgeabout the experiences of others -and notjust because I have no disproofof skepticism. My claimis thatthe knowledgeargument is a validargument fromhighly plausible, thoughadmittedly not demonstrable, premisesto theconclusion that physicalism is false. And that, after all, is about as good an objection as one could expect in thisarea of philosophy. (ii) Churchland's second objection (24/5) is that there must be something wrongwiththe argument,for it proves too much. Suppose Mary received a special series of lectures over her black-andwhitetelevision froma full-blown dualist,explainingthe "laws" governing the behavior of "ectoplasm" and tellingher about qualia. This would not affect the plausibility of the claim thaton her release she learns something. So if the argument worksagainstphysicalism, it worksagainstdualism too. My replyis thatlecturesabout qualia over black-and-white television do not tellMaryall thereis to knowabout qualia. Theymaytell her some things about qualia, forinstance,thatthey do not appear in thephysicalist's story, and thatthe quale we use 'yellow'foris nearly as different fromthe one we use 'blue' foras is whitefromblack.But whyshould it be supposed thattheytellher everything about qualia? On the otherhand, it is plausible thatlecturesover black-and-white television mightin principletell Maryeverything in the physicalist's You do not need color television story. to learn physics or functionalist psychology. To obtain a good argumentagainstdualism (attribute dualism;ectoplasmis a bitof fun),thepremisein theknowledge argumentthat Maryhas the fullstoryaccording to physicalism beforeher release,has to be replaced bya premisethatshe has the full story accordingto dualism.The former is plausible; the latteris not. Hence, thereis no "parityof reasons" troublefordualistswho use the knowledgeargument. (iii) Churchland'sthirdobjection is thatthe knowledgeargument claims"that Marycould not even imagine whatthe relevantexperience would be like, despite her exhaustiveneuroscientific knowledge, and hence must stillbe missingcertaincrucial information" (25), a claim he goes on to argue against. But,as we emphasizedearlier,theknowledgeargument claimsthat Marywould not knowwhatthe relevantexperienceis like.What she could imagineis another matter.If her knowledgeis defective, despitebeing all thereis to knowaccordingto physicalism, thenphysicalism is false,whateverher powers of imagination. Monash University
FRANK JACKSON