Journal of Philosophy, Inc.

Lecture I: Sellars on Perceptual Experience Author(s): John McDowell Reviewed work(s): Source: The Journal of Philosophy, Vol. 95, No. 9 (Sep., 1998), pp. 431-450 Published by: Journal of Philosophy, Inc. Stable URL: http://www.jstor.org/stable/2564576 . Accessed: 13/05/2012 12:27
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Journal of Philosophy, Inc. is collaborating with JSTOR to digitize, preserve and extend access to The Journal of Philosophy.

http://www.jstor.org

THE JOURNAL OF PHILOSOPHY

VOLUMEXCV, NO. 9, SEPTEMBER 1998

HAVING THE WORLD IN VIEW:SELLARS, KANT, AND INTENTIONALITY* LECTURE I: SELL ARS ON PERCEPTUALEXPERIENCE tn his seminal set of lectures, "Empiricism and the Philosophy of | Mind,''1Wilfrid Sellars offers (among much else) the outlines of Sa deeply Kantian way of thinking about intentionality about howtllought and language are directed toward the world. Sellars describesScience Metap/zysics: and Variations KantianThemes,2 maon his jorwork of the next decade, as a sequel to "Empiricism and the Philosophyof Mind" (ibid.,p. viii). The later work makes explicit the Kantianorientation of the earlier; Sellars now shows a conviction that own thinking about his intentionality (and, indeed, about everything)can be well expounded through a reading of Kant. I do not think is far-fetched to attribute to it Sellars a belief on the following lines: one has come closer than Kant no to showing us how to find intentionalityunproblematic, and there is no better way for us to find intentionality unproblematic than by seeing what Kant was driving That means rethinking his at. thought for ourselves and, if necessary, correcting him at points where we think we see more clearly than did what he should have been he doing. Sellars does not hesitate claim, on some points, to have a to better understanding of the requirements Kantian thinking than Kant of himself achieved.
8 A

revisedversion of the WoodbridgeLectures, sponsored by Philosophy, ColumbiaUniversity,on April 15, 16, and 17, 1997. the Departmentof tIn Herbert Feigl and Michael Scriven,eds., Minnesota in of Science, Volume 1 (Minneapolis:MinnesotaUP, 1956); Studies thePhilosophy reprinted (withsome added footnotes)Sellars'sScience, in Perception Reality and (London: Routledge, 1963; reissued, Atascadero,CA:Ridgeview,1991). I shall give page references to the original printing, followed in parenthesesby references to the 2 London: Routledge, 1967;reissued,Atascadero, reprintedversion. CA:Ridgeview,1992. 0029-362X/98/9509/431-90 431 (C) 1998 The Journal of Philosophy,Inc.

432

THEJOURNALOF PHILOSOPHY

Now, I share this belief I have read into Sellars, that there is no better way for us to approach an understanding of intentionality tllan by working toward understanding Kant. I also believe that coming to terms with Sellars's sustained attempt to be a Kantian is a fine way to begin appreciating Kant, and thereby given the first belief to become philosophically comfortable with intentionality. I mean this as a partly backhanded compliment to Sellars. He makes perfectly clear how he thinks he has to correct Kant, and I want to suggest that the divergence is revealing. I think a fully Kantian vision of intentionality is inaccessible to Sellars, because of a deep structural feature of his philosophical outlook. I believe we can bring into clearer focus the way Kant actually thought about intentionality, and thereby given that first belief how we ourselves ought to think about intentionality, by reflecting on the difference between what Sellars knows Kant wrote and what Sellars thinks Kant should have written.3 The reading of Kant that I aim to sketch in these lectures is under construction in a collaborative enterprise I am privileged to be engaged in with my colleagues James Conant and John Haugeland. Here I want to make a standard prefatory remark, which I mean in a less ritualistic manner than is perhaps usual. Conant and Haugeland should receive full credit for anything in what follows that is helpful toward the understanding of Kant, and thereby toward the understanding of intentionality. The blame for anything unhelpful, or simply wrong, is mine alone. In particular, Conant and Haugeland should not be held responsible for the perhaps perverse idea that we can approach an understanding of Kant through seeing how close Sellars comes to Kant's picture; nor should they be held responsible for the details of my reading of Sellars.4
3 It is a measure of how difficult it is to come to terms with Kantthat this sort of indirect approachcan be helpful. Dieter Henrich describesmy references to Kant, in my earlier engagement with the issues I shall be considering in these lectures (Mind and World(Cambridge: Harvard, 1994; reissued with an Introduction, 1996)), as "platitudinnahen" "ZweiNaturalismen auf englisch," Merkur, DLXV (1996): 334-43.No doubt it is nearlyplatitudinousthat sensibilitymust have a central role in any even approximatelyKantianattempt at making intelligible the very idea of intentionality,the directedness of subjectivestates or episodes towardobjects. But that is nearly platitudinousjust because it is neutral between Sellars's reading of Kantand the quite different picture I was tryingto give. Sellarsthinks a properlyKantianposition requiresthat conceptual episodes occur in perception in a way that is guided by "sheerreceptivity." do not believe that is a correct picture I of the transcendentalrole of sensibilityin a properlyKantianposition. If this belief were platitudinous,Sellarscould not have understood Kantianthinking as he does. I hope this will become clearerin these lectures. 4 I have also benefited from years of fruitful exchange with Robert Brandom, and from his veryhelpful comments on a draftof these lectures.

SELLARS ON PERCEPTUAL EXPERIENCE

433

Sellars's master thought in "Empiricism and the Philosophy of Mind"is this. There is a special categoryof characterizations states of or episodes that occur in people's lives for instance, characterizations of states or episodes as knowings; and, we might add, corresponding characterizations the people in whose lives the states or of episodes occur for instance, characterizations people as knowers. of In giving tllese characterizations, place whateverthey characterwe ize in "the logical space of reasons"(op. cit.,36, pp. 298-99 (169)). Sellars'sthesis is that the conceptual apparatuswe employ when we place things in the logical space of reasonsis irreducibleto any conceptual apparatusthat does not serve to place things in the logical space of reasons.So tlle masterthought as it were drawsa line: above the line are placingsin the logical space of reasons,and below it are characterizations do not do that. that That is a merelynegativespecificationof whatwe must distinguish from placings in the logical space of reasons. But Sellars is concerned to warnagainsta particularphilosophicalpitfall,namely,the temptation to suppose, of certain specific below-the-linecharacterizations,that they can fulfill tasksthat can, in fact, be fulfilled only by above-the-line characterizations. This temptationis urgent in respect of some, in particular,of the characterizations that function below Sellars'sline, and we need a positivespecificationof the characterizations that activatethe temptation. Sellarssometimes suggests this helpful wayof putting his thought: characterizations affirmepisthat temic facts need to be distinguished from characterizationsthat affirm naturalfacts.5In these terms, his central thesis is that we must not suppose we can understandepistemicstatesor episodes in terms of the actualization of merely natural capacities capacities that their subjectshave at birth, or acquire in the course of merely animal maturation.I think 'epistemic'here amounts to something like 'conceptinvolving'; shalljustifythis interpretationshortly. I Assuming this interpretationfor the moment, we can bring Sellars'sthought into direct contact with Kant.The logical space of reasons,on this reading,is the logical space in which we place episodes orstateswhen we describethem in termsof the actualizationof conceptual capacities.What corresponds in Kant to this image of the

5 See 17, p. 274 (146), and note the echo of S, p. 257 (131), where Sellars warns a mistakein epistemologythat is "ofa piece with the so-called 'naturalistic of fallacy' ethics."At 36, pp. 298-99 (169), he contrastsplacing things in the space in ofreasonswith "elupirical description"; think this formulationis less helpful. I

434

THEJOURNALOF PHILOSOPHY

logical space of reasons is the image of the realm of freedom. The way to understand the correspondence is to focus on the Kantian idea that conceptualcapacitiesare essentiallyexercisablein judging. It is true, and important,thatjudging is not the only mode of actualization of conceptual capacities;I shall be exploiting the point in these lectures. But even so, judging can be singled out as the paradigmatic mode of actualizationof conceptual capacities,the one in termsof which we should understandthe veryidea of conceptualcapacities in the relevant sense. And judging, making up our minds what to think, is something for which we are, in principle, responsible something we freely do, as opposed to something that merely happens in our lives. Of course, a belief is not always,or even typically,a result of our exercisingthis freedom to decide what to think. But even when a belief is not freely adopted, it is an actualizationof capacitiesof a kind, the conceptual,whose paradigmatic mode of actualizationis in the exercise of freedom that judging is. This freedom, exemplified in responsible acts of judging, is essentially a matterof being answerableto criticismin the light of rationallyrelevant considerations.So the realm of freedom, at least the realm of the freedom ofjudging, can be identifiedwith the space of reasons. Sellarsdescribesthe logical space of reasonsas the space "ofjustifying and being able to justify what one says."6 can see this as a We distinctivelytwentieth-century elaboration of a Kantianconception: the conception of a capacityto exercise, paradigmatically judgin ment, a freedom that is essentiallya matterof responsivenessto reasons. The twentieth-centuryelement is the idea that this capacity comes with being initiatedinto language.
II

At a pivotalpoint in "Empiricism the Philosophyof Mind,"Selland ars addressesthe questionwhether empiricalknowledgehas foundations (op. cit., VIII,pp. 293-300 (16670)). His answeris nuanced. In an empiricisticfoundationalismof the usual kind, it is not just that the credentials of all knowledge are ultimately grounded in knowledgeacquiredin perception. Beyond that, the grounding perceptual knowledgeis atomistically conceived. Traditionalempiricists take it that each element of the grounding knowledgecan in principle be acquired on its own, independently not only of other elements of the grounding perceptualknowledge but also of anything

6 36, p. 299 (169). This connects with the perhaps infelicitouslylabeled thesis of "psychological nominalism": 29, p. 289 (160); 31, p. 291 (162); and, for an see anticipationearlyin the lectures, 6, p. 258 (131).

SELLARS ON PERCEPTUAL EXPERIENCE

435

in the world view that is grounded on this basic stratum of knowledge. What Sellars objects to in traditional empiricism is just this supposed independence. He writes: There is clearlysome point to the pictureof human knowledgeas resting on a level of propositions-observatioll reports which do not rest on other propositionsin the same wayas other propositionsrest on them. On the other hand, I do wish to irlsistthat the metaphor of "foundation"is misleadingin that it keeps us from seeing that if there is a logical dimension in which other empirical propositions rest on observation reports,there is anotherlogical dimensionin which the latter rest on the former (op. cit., 38, p. 300 (170); cf. 19, p. 275 (148)). Sellars does not deny that there is a logical dimension in which observation reports are basic. His point is just to insist on the otller logical dimension, in which observation reports depend on the world view that is grounded on them, in the logical dimension to which a traditional empiricism restricts itself. The result is a picture that is still in a way empiricist, by virtue of its acknowledgment of the one logical dimension, though it is separated from traditional empiricism by virtue of its insistence on the other. Of course Sellars's point here is at least partly epistemological, in an intelligibly narrow sense; he is telling us how we should conceive the credentials in virtue of which a world view counts as knowledgeably held. But the divergence from traditional empiricism means that we cannot take Sellars to be doing epistemology in some sense that contrasts with reflecting about intentionality. It is indeed perceptual knowledge (knowledge expressed in observation reports) about which he is here urging that it depends, in respect of the concepts that figure in it, on a world view. But that is just a case of something more general; his thought is that the conceptual equipment that is operative in perceptual experience generally, whether the experience is such as to yield knowledge or not, is dependent on a world view, in the logical dimension that the metaphor of "foundation" risks leading us to forget. We can capture this part of the picture by saying that the intentionality, the objective purport, of perceptual experience in general whether potentially knowledge yielding or not depends, in that logical dimension, on having the world in view, in a sense that goes beyond glilupses of the here and now. It would not be intelligible that the relevant episodes present themselves as glimpses of the here and now apart from their being related to a wider world view in the logical dimension Sellars adds. But the wider world view depends

436

THEJOURNALOF PHILOSOPHY

in turn, in the logical dimension that figures in traditional empiricism, on perceptual experience that is capable of yielding knowledge, in the form of glimpses of the here and now. With this mutual dependence, the nontraditional empiricism that Sellars espouses constitutes a picture both of the credentials of empirical knowledge and of the intentionality of empirical thought in general. This makes it unsurprising that we find Sellars speaking of "the epistemic character, the 'intentionality'," of expressions such as 'thinking of a celestial city'.7 When he introduces the image of the logical space of reasons, he singles out the episodes or states whose characterizations place them in the space of reasons as episodes or states of knowing(op. cit., 36, pp. 298-99 (169)). And of course episodes or states of knowing would have an epistemic character in an etymologically obvious sense. But it would be wrong to conclude that Sellars's concern is narrowly epistemological. In the remark about 'thinking of a celestial city', he makes this clear by showing that he is willing to equate epistemic character with intentionality, and to talk of epistemic character in a case in which there need be no question of knowing. In the remark about 'thinking of a celestial city', 'epistemic' can amount to no more than 'concept involving'.8 This is the interpretation I announced and promised to vindicate. I have been urging that Sellars's nontraditional empiricism is not only a picture of the credentials of empirical knowledge, a topic for epistemology in a narrow sense, but also a picture of what is involved in having one's thought directed at the world at all, the topic of reflection about intentionality. This enables me to forestall a possible objection to the proposal that we read Kant, Sellars's model, as a philosopher of intentionality. I do not mean the feeble objection that 'intentionality' is not a Kantian term. 'Intentionality' is a scholastic term, which did not come back into mainstream philosophical currency until (I think) Brentano, but obviously that does not prevent us from supposing that the topic is a Kantian topic. What I have in mind is rather the potentially more challenging objection that Kant's concern is epistemological. As a putative reason for supposing that Kant is not concerned with intentionality, I can neutralize this by saying: certainly, Kant's concern is epistemological in just the way in which Sellars'sis.

7 7, p. 260 (133). Compare 24, 2S, pp. 283-85 (155-56). Consider also the implication, at 17, p. 274 (146), that looking red is an epistemic as opposed to natural fact about objects. Looking red is not an epistemic fact in the etymologically obvioussense that I mention in the text below. 8 See Science and Metaphysics, 23: for purposes of the philosophy of mind, "the p. intentional is that which belongs to the conceptual order."

SELLARS ON PERCEPTUAL EXPERIENCE

437

Against a "neo-Kantian"reading of Kant, Heidegger says: "The Crz-

tique PureRBason nothing to do with a 'theory of knowledge'."9 I of has

think we can make the point Heidegger is trying to make more effectively certainly we can put it in a form in which it is easier to swallow by saying, not that epistemology is noconcern of the first Critique, but that it is no more theconcern of the first Critique it is of "Emthan piricism and the Philosophy of Mind"or of Science Metap/zysics. and
III

Early in "Empiricism and the Philosophy of Mind," Sellars diagnoses "the classical concept of a sense datum" as "a mongrel resulting from a crossbreeding of two ideas":first, an idea of non-concept-involving sensory episodes, such as sensations of red, and, second, an idea of noninferential knowings that such-and-such is the case (op. cit., 7, p. 259 (132)). This is a mongrel, a conflation, because attributions of nonconcept-involving episodes belong below the line drawn by Sellars's master thought, whereas attributions of knowings belong above it. When Sellars repeats the diagnosis a few pages later, he extracts from it a program for the rest of the lectures: "to examine these two ideas and determine how that which survives criticism in each is properly to be combined with the other" (op. cit.,10, p. 267 (140)). The program, then, is to arrive at an acceptable picture of how the sensory and the conceptual sensibility and understanding combine so as to provide for the intentionality of perceptual experience, and (the same function viewed from a different angle) to provide for how perceptual experience figures in the acquisition of a knowledgeable view of the world. We have looked ahead in Sellars'swork, so we know that this case of intentionality, the intentionality of perceptual experience, is going to be in one way basic to intentionality in general, though not in a way that involves its being intelligible in advance of the idea of having a world view that goes beyond the immediate deliverances of perception. The above-the-line element in the mongrel conflation is the idea of noninferential knowings. Sellars mostly focuses on one sensory modality, and considers seeings.l But in pursuing his program in connection with this particularization to one sensory modality of the
9 Kant and theProblem Metaphysics, of RichardTaft, trans. (Bloomington:Indiana UP, 1990, fourth edition), p. 11. Heidegger's word is 'Erkenntnistheorie',which might have been translated'epistemology';see Taft'snote, p. 188. t For a self-conscious comment on this, see Science Metaphysics, 9. There is a and p. minor complication (nothing turns on it): seeings are not, as such, noninferential knowingsor acquirings knowledge(thatwashow the above-the-line of elementsin the mongrel conflationwere first introduced), but ratheropportunitiesto know,which maynot be taken.Considerhow it mightbe intelligibleto saythis:"Ithoughtit merely looked to me as if the tie wasgreen,but I now realizethatI wasseeing it to be green."

438

THEJOURNALOF PHILOSOPHY

above-the-lineelement in the mongrel, he expands the topic from seeings to a wider class of experiences, which he initiallyintroduces as ostensible seeings. Seeings are a singled-outsubclassof ostensible seeings.ll EvidentlySellars takes it that for purposes of separating and correctlycombining what survivescriticismin the ideas that are conflated into the mongrel, what mattersis to understandthe wider class. The goal is to understand the intentionalityof visual experience in general,whetherpotentiallyknowledgeyieldingor not. Ostensible seeings are experiences in which it looks to their subject as if things are a certainway,and Sellarsdevotes some effort to elucidatingthat idea. Centrally importanthere is the image of an experience as, "so to speak, making an assertionor claim"or as "containing"a claim (op. cit., 16, pp. 271-72 (144)). Sellarsintroduces this image in an explicitlypromissoryway, pointing forwardto the culminationof "Empiricism the Philosophyof Mind."There, he and vindicatesa notion of nonovert conceptual episodes, on the ground that they can be understoodby analogicalextension from overt conceptual episodes,linguisticacts.l2 Visualexperiences "make" "conor tain" claims in that they are conceptual episodes, actualizationsof conceptual capacities, and as such are to be understood on the model of linguisticperformancesin which claimsare literallymade. This bears some elaboration.I have mentioned the Kantianview that conceptualcapacitieshave their paradigmatic mode of actualization in judgings.We can approachthe idea thatvisualexperiencesare conceptualepisodes,and as such "make" "contain" or claims,through this identificationof judging as the paradigmatic kind of conceptual episode. Consider,say, judging that there is a red cube in front of one. There is a conceptualcapacity thatwouldbe exercisedboth in making thatjudgment and in judging that there is a red pyramidin front of one, and anotherconceptualcapacitythatwouldbe exercisedboth in judging that there is a red cube in front of one and in judging that there is a blue cube in front of one. Injudging that there is a red cube in front of one, one wouldbe exercising(at least) these two capacities together.Whatdoes 'together'mean here?Notjust that one wouldbe

" For seeings as veridicalmembers of a class of ostensible seeings, see "Empiricism and the Philosophyof Mind,"7, pp. 260-61 (133-34). This points towardthe discussionof 'looks' statementsin III,pp. 267-77 (140-49). I shall comment in my third lecture on the idea, which is implicit in at least the first version of "Empiricism and the Philosophy of Mind,"that veridicalityis all it takes for an ostensible . . seelng to Dea seelng. 12 The first phase of the myth of Jones; for the application to perceptual experience, see 60, p. 321 (190).

SELLARS ON PERCEPTUAL EXPERIENCE

439

exercisingthe two capacitiesin a single act of judgment; that would not distinguish judging that there is a red cube in front of one from judging, say, that there is a red pyramidand a blue cube in front of one. In a judgment that there is a red cube in front of one, the two conceptual capacitiesI have singled out would have to be exercised with a specificmode of togetherness: togethernessthat is a countera partto the "logical" semanticaltogethernessof tlle words'red' and or 'cube' in the verbalexpressionof the judgment, 'Thereis a red cube in front of me'. Here we see the point of the idea that nonovertCOl1ceptual episodes are to be understoodon analogywith linguisticacts; it affordsa wayto makea distinctionthatwe need to make.13 The conceptual episodes Sellars is concerned with, when he speaks of visual experiences as "containing" claims, are not as such cases of judging. Even if one does judge that things are as they look, having them look that way to one is not the same as judging that they are that way.In soznecases,perhaps,one doesjudge that things are a certainwaywhen they look that way acquiringthe belief that they are that wayby freely making up one's mind that they are that way.But more typically, perceptualbelief acquisitionis not a matter of judging, of activelyexercising control over one's cognitivelife, at all. Unless there are groundsfor suspicion,such as odd lightlng conditions, having it look to one as if things are a certainway ostensibly seeing things to be that way becomes accepting that things are that way by a sort of default, involving no exercise of the freedom that figuresin a Kantianconception ofjudgment. So thereis a disconnectonbetweenperceptual experienceandjudging. But evenso, we can exploitthe apparatus the conceptionofjudgof ing I havesketchedin orderto vindicateSellars's image of experiences as "containing" claims.A free, responsibleexerciseof certainconceptual capacities, includingat least the two I mentioned,with a suitable mode of togetherness wouldbejudging thatthereis a red cube in front of one. Now we can saythat in an ostensibleseeing that there is a red cube in frontof one an experiencein whichit looksto one as if there is a red cube in front of one the same conceptualcapaciteswouldbe
13 I mean this quick sketch of a conception of judging, as the joint exercise of different conceptual capacities,to recall GarethEvans'sdiscussionof "the Generality Constraint": Varieties Reference The of (New York:Oxford, 1982), pp. 100-05.Evans's discussionhas its roots in P. T. Geach'saccount of judging on analogywith saying, in Mental Acts(London:Routledge, 1957). Geach'sanalogicalaccount of judging is roughly contemporarywith Sellars'sin "Empiricism and the Philosophyof Mind"; independent of it; and (I would argue) more satisfactory, being free of the scienin tistic baggagewith which Sellarsencumbershis version.But I shall not be considering the scientistic details of Sellars's version in these lectures (though Sellars's scientismwill matterin other contexts).

440

THEJOURNALOF PHILOSOPHY

actualized withthe samemode of togetherness. This cashesout the idea thatan experienceso described "contains" claim,whosecontentisjust a whatone wouldbejudgingin the corresponding judglnent.But thisactualizationof the relevantconceptualcapacities,unlike the one that would be involvedin the corresponding judgment,would be involuntary; is whyI say'actualization' that ratherthan 'exercise'. This idea of conceptualcapacitiesbeing involuntarily actualizedin perceptual experiencepartly captures point of a striking the remark Sellarsmakesaboutthe wayan experience"contains" claim;he saysthat a the claim"is,so to speak,evokedor wrungfromthe perceiver the obby ject perceived" cit.,16 bis, p. 272 (144)). When Sellars this,he (op. says is talkingaboutexperiencesof seeing,but the point he is makingsurely appliesalso to membersof the widerclass,ostensibleseeings,even the ones that are not seeings. Ostensibleseeings are experiencesthat, as conceptualepisodes,"contain" claims,but in a specialwaythatdifferentiates them from conceptualepisodes of other kinds. They "contain" their claimsas ostensibly necessitated an objectostensibly by seen. In Science and Metaphysics, Sellarsputs the same point by sayingthat, if one saysit looks to so-and-so though there were a red and rectangular as physical objectin frontof him, one is attributing so-and-so to (whoInay, of course,be oneselp a conceptualrepresentation, a particular of kind, that there is a red and rectangular physical objectin front of him; and the kindis "that kindof conceptual representation whichis being under the visualimpression that...there (or of there being) a red and rectanis gularphysicalobjectin front of one" (op. cit.jp. 14; my emphasis).In the languageof "Empiricism the Philosophy Mind," is to say and of this that ostensibleseeings "contain" their claimsin a distinctive way,one that distinguishes them from other conceptualepisodes;they "contain" theirclaimsas ostensibly visually imposed impressed theirsubject.l4 or on

14 I have, I think charitably,discounted "evoked or wrung from theperceivef' in the formulation Sellarsuses in "Empiricism and the Philosophyof Mind."A claim evoked from a perceiverwould surely be a claim that the perceiver makes. But it seems wrong to imply that a perceivermakes the claim his experience "contains"wrong even before we widen the focus from seeings to ostensible seeings. Whether an ostensible seeing is a seeing turns on whether its "contained" claim is true, and that is a separate question from whether its subject makes (endorses) the claim. (See footnote 10 above.) So even a seeing, let alone a merely ostensible seeing, does not necessarily"contain" claim made by its subject.Where I have markedan a omission in my citation of the parallel remarkfrom Science Metaphysics, and Sellars glosses "being under the visual impression that"with "(visually taking it to be the case that),"and this seems wrong in the same way.We can correct Sellarson this without posing a threat to something he wants to insist on: that one gets to have conceptual episodes (representations)of the relevantkind occur in one's life at all only by acquiringthe capacityto make the claims-they"contain."

SELLA1AS ON PERCEPTUAL EXPERIENCE

441

So it is not simply that conceptual episodes of the relevant kind consistin actualizations conceptualcapacitiesthat are involuntary. of (We have that also with other kinds of conceptual episode; for instance, when one is, as we say,struckby a thought.) In visualexperiences, conceptual capacities are actualized with suitable modes of togetherness;this is how we cash out the idea that the episodes "contain"claims.But they are actualizedwith an involuntariness a speof cific kind; in a visualexperience an ostensiblyseen object ostensibly itupressesitselfvisuallyon the subject.Presumably parallelthings are to be said about other sensorymodalities.l5
IV

I have been consideringwhat survivescriticism,from the above-theline element of the mongrel conflation, in a conception of visualexperience that would be acceptable by Sellars'slights. His program for "Empiricism and the Philosophy of Mind"presupposesin addition that something survivescriticism from the below-the-lineelement in the mongrel conflation, and needs to be acceptably combined with conceptual episodes of the distinctive kind I have been discussing, in a total picture of visual experience. In the course of executing his program,Sellarssaysit is "clear" that there is more to visual experience than conceptual episodes of that distinctive kind; and, specif1cally, that a full picture lnust also include non-concept-involving episodes of the kind exemplif1ed,in the original description of the mongrel conflation, by sensations of red.l6 But whyis this supposed to be clear? The question is especially pressing when we realize how much goes into whatwe alreadyhave above Sellars'sline how much goes into the idea of a conceptualepisode of the relevantkind. Even after we have said not just that a visualexperience "contains" claim but a also that the "contained" claim is, so to speak, "evoked" an ostenby siblyseen object, Sellarsstill saysit is clear that we need to add something about visual episodes of a nonconceptual kind. In his view,

15 Compare the conception of experience I recommended in Mind and World, where I wrote of states or episodes in which conceptual capacitiesare operativein sensibility. I think such a formulation simply captures, in explicitly Kantian language, the way Sellars shows us how to conceive perceptual experience at any rate what he sees as the above-the-lineelement in the total truth about perceptual experience. In Mind and World (for example, pp. 140-41) I focused on the belowthe-line role that Sellars credits to sensibility,and missed the fact that he has an above-the-lineconception of perceptualimpressionsthat matches the conception I was recommending. 16 "Empiricism and the Philosophyof Mind,"16 bis, p. 272 (14645). Compare the use of 'of course' at 22, p. 279 (151); and again at 45, p. 305 (175).

442

THEJOURNALOF PHILOSOPHY

conceptual episodes of the relevantkind are already,as the conceptual episodes they are, cases of being under the visual impressionthat such-and-such the case. It is not that as conceptual episodes they is are phenomenologicallycolorless, so that they would need to be associated with visual sensations in order that some complex composed of these conceptual episodes and the associated visual sensationscan be recognizably visual.These conceptualepisodes are already,as the conceptual episodes they are, shapingsof visual consciousness.l7 we need a below-the-lineelement in our picture, it is If not in order to ensure that the picture depicts states or episodes of
. .

vlsua

consclousness.

So why does Sellars think our total account of visual experience needs to include visualsensationsas well?About the presence of the correspondingelement in the mongrel conflation,he says:
[This] idea clearlyarisesin the attemptto explain the factsof sense perception in scientificstyle.How does it happen that people can have the experiencewhich they describeby saying"Itis as though I were seeing a red and triangular physicalobject"when either there is no physicalobject there at all, or, if there is, it is neither red nor triangular? The explanation, roughly,posits that in every case in which a person has an experience of this kind, whetherveridicalor not, he has whatis called a 'sensation'or 'impression''of a red triangle' (op. cit., 7, p. 259 (13233)).

Arld in the view that emerges as his own, in the course of "Empiricism and the Philosophyof Mind,"it is not in this explanatorymotivation for its below-the-line element that the mongrel conflation goes wrong, but just in the way it conflates the below-the-lineelement so motivated with the above-the-lineelement, episodes that would have to be actualizationsof conceptual capacities.Sensations figure in the picture, at least initially,as posited on the ground that they are needed for an explanatorypurpose.l8 What explanatorypurpose?In "Empiricism the Philosophyof and Mind,"the envisaged explanation is, as we have seen, "in scientif1c
17 Contrast, for instance, Robert B. Brandom, Making Explicit: It Reasoning, Representin$ and Discursive Commitment (Cambridge:Harvard,1994). In his chapter 4, Brandom undertakesto give an account of observationalclaims and observational knowledge while sedulouslyavoiding any mention of sensory consciousness. Brandom here divergesfrom something that is quite central to Sellars'sthinking. 18 See 21-22,pp. 277-80 (149-52); a programmatic passage, and he returns to its programin 4S, pp. 305-07 (175-77), and esecutes it in the rest of the lectures. Much of the workneeded is in qualifyingthe idea of sensationsas posited, in order to make room for immediate self-attribution sensations.This is why I say 'at least of initially'.

SELLARS ON PERCEPTUAL EXPERIENCE

443

style."And the question that the explanation is to answerseems to be this, to put it in termsthat become availableduring the execution of Sellars'sprogram:How is it that the same claim would be "contained" in, say, each member of a trio of possible experiences of which one is a case of seeing that there is a red and triangularphysical object in front of one, one is a case in which something in front of one looks red and triangularalthough it is not, and one is a case in which it looks to one as if there is something red and triangular in front of one although tllere is nothing there at all?l9 If whatwe still need to ask for is an explanationof the sameness of claims "contained" such a trio of experiences, the request for an in explanationapparentlyassumesthat we are alreadyentitled to think of experiences as "containing" some claims or other, independently of the explanationwe are askingfor. After all, we might say,the idea that experiences "contain"claims is already accounted for in the part of the storythat belongs above the line; we have alreadyentitled ourselvesto it by talking about actualizationsof conceptual capacities, before questionsarise about a below-the-lineelement in the total story.But in that case, it is not clear why we should suppose that our explanatoryneed can be met only by finding a samenessat the level of visualsensations items in consciousness between the members of such a trio, as opposed to a samenessat the level of, say,patterns of light impingingon retinas.When he elaborates"theattempt to explain the facts of sense perception in scientificstyle"by positing sensations, Sellarshimself says:"The core idea is that the proxienate cause of such a sensation is only for the most part brought about by the presence in the neighborhood of the perceiverof a red and triangular physicalobject" (op. cit., 7, pp. 259-60 (133); einphasis altered). In this retnark,Sellarsis suggestingthat we should expect to fUlnd sameness between seeings and ostensible seeings that "cona tain"the same claims,at a level that he here specifUles that of proxias mate causes of sensations for instance, at the level of retinal images. But then why not suppose a sameness at this level will do the explanatorywork for which Sellarsthinks we need to appeal to sensations? Conceptual episodes of the relevant kind are triggered by impactsfrom the environmenton a perceiver'ssensoryequipment.
t9 See 45, p. 305 (175) for a formulation on these lines. Sellars notes (22, p. 279 (151)) that it is not strictlyaccurateto say that the same claim is "contained" in each member of such a trio: the claim "contained" each of the first two is referin ential, whereasthe claim "contained" the third is not. This will startto be signifiin cant in my second and third lectures, but it does not matter here, any more than it does at the point where Sellarsacknowledgesit.

444

THEJOURNALOF PHILOSOPHY

If the impactsare suitablysimilar,there is nothing puzzling about a similarity between the conceptual episodes they trigger.And it is not clear why it should seem necessaryto describe these suitablysimilar impacts in terms of nonconceptual impingements on consciousness (sensations), as opposed to saying that consciousness comes into play only with conceptual episodes, triggered by nonmentalistically described impacts on sensoryequipment. It seems that what Sellars here introduces as proximate causes of sensations can themselves meet the explanatoryneed, conceived as he seems to conceive it in "Empiricism the Philosophyof Mind."The sensationslook like and idle wheels. In Scienceand MetapSlysics, Sellars explicitly confronts an objection on these lines (op. cit., p. 18); and he responds in a way that changes the picture rather radicallyfrom the one he seemed to be giving in "Empiricismand the Philosophy of Mind." First, he no longer formulatesthe explanation-seekingquestion in terms of the sameness of the claims "contained"in different possible experiences as if we could anyway help ourselvesto the idea that experiences "contain" claims at all. The explanation-seeking question now is: How is it that sensory relatedness to the environment takes the form of conceptual episodes, episodes that, in the terminology of "Empiricism and the Philosophy of Mind,""contain"claims, at all?20 Second, the explanatoryneed that sensations are supposed to satisfyis not a need for scientifUlc understanding,as it seemed to be in "Empiricism and the Philosophy of Mind";rather, it is transcendental.2l think these are two waysof putting the same thought: the I reason Sellars thinks our complete account of visual experience must include visual sensations nonconceptual visual episodes is that he thinks this is the only way we can fUlnd intelligible that it there should so much as be the conceptual shaping of sensory consciousness that constitutes the above-the-line element in his account of visual experience.

20 See op. cit., p. 18, where, in urging that the explanatoryquestion is not specially about nonveridical experiences, Sellars writes: "even in normal [veridical] cases there is the genuine question, 'Whydoes the perceiver conceptually represent a red (blue, etc.) rectangular(circular,etc.) object in the presence of an object having these qualities?'" 21 See op. cit., p. 9. Sellars says that manifolds of sensation are "postulatedon general epistemological or, as Kant would say, transcendental grounds."I think this equation, in the context of the understandingI am offering of the sense-impression inference, reinforces the impressiongiven by his willingnessto equate 'intentionality' with 'epistemic character' in "Empiricismand the Philosophy of Mind";he does not conceive epistemologynarrowly.

SELLARS ON PERCEPTUAL EXPERIENCE

445

'Transcendental' figures here in a recognizably Kantian sense. The explanation Sellars envisages is transcendental because it is needed, he thinks,in order to vindicatethe legitimacyof the apparatus the talk of experiences as actualizationsof conceptual capacities, which as such "contain"claims, but in a distinctivelysensory way in terms of which we enable ourselvesto conceive experiences as ostensibly of objectsat all.22Sellars thinks his picture, with sensations playing such a transcendental role, just is the picture Kant would have given if he had been fully clear about the drift of his own thinking. On this reading of the "sense-impression inference"as it H1gures in Scienceand Metaphysics (for the phrase,see op. cit., p. 17), visualsensations or sense impressionsare not simplyan extra part of the truth about visualexperiences, over and above the part that deals with the distinctiveway in which visual experiences "contain" claims. That is how it might have seemed from "Empiricism the Philosophyof and Mind."But in the view Sellarsurges in Scienceand Metaphysics, is it not that visual experiences "contain" claims in their distinctiveway, and then there is a simplyadditionalfact about them, namely, that they involve visual sensations. The reason we have to acknowledge the "additional" fact, in Sellars'sview, is that only so can we be entitled to have spoken as we did when we gave our above-the-line characterization of visual experiences when we spoke of visual experiences as "containing" claims,and so having objectivepurport, in the distinctivewaythey do. Sellars's sense-impressioninference is a piece of transcendental philosophy,in the followingsense: it is directed towardshowing our entitlement to conceive subjectiveoccurrences as possessing objective purport.Notice that that descriptionof transcendentalphilosophy implies nothing in particularabout the nature of the activity. There is a temptation to suppose transcendentalphilosophy would have to be done at a standpoint external to that of the conceptual goings-onwhose objectivepurportis to be vindicated a standpoint at which one could contemplate the relation between those conceptual goings-on and their subject matter from sidewayson. Sellars's move f1tsthis conception; he undertakesto vindicate the objective purport of conceptual occurrencesfrom outside the conceptual order. I shall be taking issue with this conception of transcendental
22 See, for example, Critique of Pure Reason, Norman Kemp Smith, trans. (London: Macmillan,1929), A11-12/B25: "Ientitle transcendentalall knowledgewhich is occupied not so much with objects as with the mode of our knowledge of objects in so far as this mode of knowledgeis to be possible a priori."

446

THEJOURNALOF PHILOSOPHY

philosophy. It is importantto see that this is not to take issue with the veryidea of transcendental philosophy.23
v

When Sellarsvindicatesthe idea of inner episodes, at the culmination of "Empiricism the Philosophyof Mind,"he does so in two and phases: fUlrst conceptual episodes and then for nonconceptual for episodes. Each phase has two stages, and the structureis parallelin each phase. First,there is an account of how concepts of episodes of the relevant kind could have been introduced in the context of a theory; at this stage, the episodes are envisaged as attributable,to others or oneself, only inferentially,in a waymediated by the theory. But then, second, there is an account of how a noninferential self-attributing ("reporting") employmentof the relevantconceptual apparatuscould have been introduced, by trainingpeople in such a wayas to leave them immediatelydisposedto make self-attributions'immediately'in the sense that they do not need to advertto the evidence that the theoryprovidesfor on occasionswhen, accordingto the theory,those attributionsare correct.By the end of"Empiricism and the Philosophyof Mind,"conceptual episodes, including those which "contain" claims in the distinctiveway in which visual experiences do, and nonconceptual sensory (and in particular visual) episodes impressions or sensations are on a level, in respect of being available noninferentialself-attribution.24 for In Science Metaphysics, and SellarsmodiEles picture in a waythat this belongs, I think, with the fact that he now explicitlysees the senseimpression inference as transcendentallydriven. He suggests that the visualimpressionsor sensationsthat the sense-impression inference requiresus to posit are states of consciousness,but not objects of consciousness;or the same thought differentlyexpressed that

23 The idea that transcendentalphilosophy would have to be done from a special standpoint is implicit in Richard Rorty, Philosophy and the Mirrorof Nature (Princeton:UniversityPress, 1979), p. 293, where Rortywritesof the "demand...for some transcendentalstandpoint outside our present set of representationsfrom which we can inspect the relations between those representations and their object." Kant distinguishes "transcendental" and "transcendent"(see, for example, AS96/B352-53). In Rorty'sphrase, 'transcendental'could be replaced by 'transcendent'. Not that that shows Rortyto be misusing 'transcendental';he is suggesting that transcendentalphilosophy requires a transcendentstandpoint. That is what I think we should dispute. When I wrote disparagingly about Kant's"transcendental story"in Mind and World (pp. 41-43, 95-98), I was acquiescing, in a way I now regret, in a reading of parts of Kant's transcendentalactivitythat fits Rorty'sphrasing. (I would still disparagethe philosophysuch a reading finds in Kant.) 24 See 59, p. 320 (189) for conceptual episodes, and 6S, p. 327 (194-95) for nonconceptual episodes.

SELLARS PERCEPTUAL ON EXPERIENCE

447

these impressionsor sensations are states of consciousness that are not apperceived,where "apperception" be explained can as "noninferentialself-knowledge."25 is suggesting,then, that the He transcendentally posited visual impressions or sensations figure in visual consciousnessin a way that does not amount to their achieving the immediate or noninferentialattributability oneself that he to works to secure at the culmination of "Empiricism and the Philosophyof Mind." Now it is hard to see how, on Sellarsianor indeed any principles, there could be a classof items in consciousnesswhose memberswere permanentlyand constitutionally incapableof being apperceived,incapableof being directlyavailablefor self-attribution. Sellarscannot mean to be suggestingthat the visualimpressionsor sensationsthat, accordingto his transcendental sense-impression inference, must figure in episodes of "outersense"are, simplyas the visual impressions or sensationsthey are, incapableof being objectsof consciousness.I thinkhis thought must rather be on the following lines. The visual impressions sensationsin question are not apperceivedwhen or they are playingtheirtranscendental That is not to say that they are role. not apperceivable. just to say that if they do get to be apperceived It is if theydo become objects for consciousness they can no longer be playing their transcendental role, that of enabling episodes of "outer sense," episodes that "contain" claims about the environment.One canfocus one's attention on the manifold of "sheer receptivity" that was, moment before, enabling one's attention to be a directed toward ostensiblyseen environment.But in doing so in the bringing it within the scope of one's apperception one ensuresthat it ceases to performthat function. I think this thought strengthensSellars'sposition by immunizing itagainst a certain objection. If it were right to endorse Sellars's sense-impression inference, it would be a good idea to construe its conclusion this way.Consideringonly "Empiricism the in and Philosophy Mind,"I used to think one could complain that of the belowthe-line items that figure in Sellars's picture of visual experience would opaque; not something through which the be environment could intelligiblybe revealedto us, but at best something on the basis which, if we knew enough about how features of of the environment cause these affectionsof our sensorycapacities,we could infer conclusions about the environment.26 "Empiricism the PhilosIn and
25 26

Science Metaphysics,pp. 10, 11. For the gloss on and "apperception," see p. 72. SeeMind and World,p. 145Z.

448

THEJOURNALOF PHILOSOPHY

ophy of Mind,"the only waySellarsconsidersfor something to be in consciousnessis for it to be an object of consciousness.This leaves it seeming that the sensationsthat are part of Sellars'spicture of perceptual experience would have to be objects of consciousness, on pain of not figuringin consciousnessat all. If that were not whatSellarswants,why does he need to workat securing that they can be objects of consciousness,as he does at the culminationof "Empiricism and the Philosophyof Mind"? And if we tryto make out that they are objectsof consciousness,it does seem that they would engross the attention, and preventit from fixing itself except indirectly,through inference on environmentalobjects,although it was perception of environmentalobjects that we were supposed to be trying to make intelligible. But the picture Sellarsgivesin Science Metaphysicsimmune to and is any such objection.Now he can concede that, if a manifold of visual sensationsis figuring as an object for its subject'sconsciousness,the subject'sattention can no longer pass through it to features of the environment directly, but at best inferentially. But this leaves unthreatened the idea that such a manifold is transcendentally required for perceptual awarenessof the environment.Sellarsis now equipped to saythat the transcendentally requiredmanifold,when it is doing its transcendental job, figuresin consciousnessnot as an object, in which case it would indeed prevent the free passage to the environmentof the subject'sattention, but preciselyas that through which the subject's attention is directed without hindrance to features of the ostensiblyseen environment.The idea is that attention, which involvesapperception,can be directed either at the ostensibly seen environmentor at the visualsensationsthat were enabling the environment to be ostensiblyseen, but not both; if the attention is directed at the sensations,they can no longer be enabling the ostensible seeing of environmentalobjects. This complication, however, does not undermine the fact that a Sellarsianaccount of the noninferentialself-attribution visualsenof sations when it does occur, as it surelycan will have the structure establishedin "Empiricism and the Philosophyof Mind."The concepts under which visualsensationsare apperceivedwhen they arewhich is not when they are enabling episodes of "outersense" will be concepts whose originalhome is a transcendentally requiredtheory concerning how manifolds of sensations enable episodes of "outersense."When visualsensationsbecome objectsfor consciousness, it will be under concepts whose originalfunction is to connect these episodes, in a theory-mediated way,with the claim-"containing"

SELLARS PERCEPTUAL ON EXPERIENCE

449

characterof visualexperiences. Some people think we can vindicate a role for sensationsin the total picture of visualexperience on the basisof a theoreticallyinnocent introspection.The idea is that sensational properties are introspectivelyavailablein any case, whatever we say about what figures, in this conception, as the subsequent question how, if at all, the sensationalpropertiesof experiences relate to their claim-"containing" character.I have ignored this conception, because it makesno contactwith Sellars'sthinking.27 It is important not to be misled by the fact that Sellars uses the word 'impression'both in the phrase 'being under the visualimpression that... ', which characterizes kind of conceptual episode, and a for the below-the-lineelement in his picture of perceptual experience.28Conceptual episodes that belong to the kind, being under the visual impression that.., are, simply as conceptual episodes, available for apperception when they occur.29Impressions in the other sense, in contrast,can be apperceivedonly when they are not

27 See, for example, Christopher Peacocke, Sense and Content (NewYork:Oxford, 1983), chapter 1. Peacocke saysthe sensationalproperties of visual experience are arrayedin a two-dimensional visualfield, and he does not suggest that 'two-dimensional' here means anything different from what it might mean in describing, say, a surfacein the environment.By Sellars'slights this is naive;for Sellars,the spatiality of the arrangementof visualsensationsis not the spatialityof "outer" configurations, but something that needs to be understood by analogical extension fi-omit, in a sophisticatedexercise of concept formation. Even on their own terms, I think Peacocke's phenomenological arguments are unconvincing,but I shall not argue this here. Perhaps the thinness of the supposedly independent phenomenological considerations reveals that Peacocke's conception of what a supposedly innocent introspection would yield is really controlled by an implicit acceptance of something like Sellars's transcendental thought. There is another putative ground for supposing that visual experiences must have a sensational aspect, equally non-Sellarsian,which I shall also not discuss in these lectures. This is the thought that there must be a two-dimensionalsensationalarrayto serve as a vehicle for the representationalcontent of a visualexperience, somewhat as an arrangement of pigment on a surface is a vehicle for the representational content of a picture. 28 See Scienceand Meta%71lysics, p. 19, where Sellarsdistinguishes"animpressionof a red rectangle"from "an impression of a man lurking in the corner."The latter would,as he says,be "a conceptual state" (or episode); one of the kind identified atp. 14, the kind "beingunder the visualimpressionthat...." This is not to say that they are actuallyapperceived.The 'I think' of apperceptionmust be ableto accompanyall my representations( Critiqueof Pure Reason, B131), which not to saythat it actuallyaccompaniesthem. But conceptualrepresentations is areavailable apperceptionin a waythat differsfrom that in which, in a plausibly for Sellarsian picture, perceptualsensationsare; apperceivingthe latter would require equippingoneself with something new, a conceptual representationinvolvingconcepts whose primary home is the transcendental theoryof how conceptualrepresentations outer realityare guided by manifoldsof "sheerreceptivity." of
99

450

THEJOU1tNAL PHILOSOPHY OF

serving as the below-the-lineelement in the total truth about some perceptualexperiences. The fact that the same word naturallyacquires these two uses is perhapssuggestiveabout the shape of Sellars'spicture.When a conceptual episode is apperceivedas belonging to the kind, being under the visualimpressionthat..., what is apperceptively available,according to Sellars'spicture, is thatthe flow of one's conceptual representations, of the sort involved in normal perceptual activity,is being guided into "containing" relevantclaim by the flow of one's imthe pressionsin the below-the-linesense. Perhapsthis is why 'being under the visual impression that...' is an appropriatespecification of the kind to which the conceptual representationapperceivably belongs. But apperception does not embrace the specifics of howthis guidance is effected; if the formerlyguiding items get to be apperceived, they can no longer be performingtheir guiding function.30
VI

Sellarsthinks this picture is essentiallythe one Kantis aiming at, although he has to acknowledgethat it is not to be found adequately set out on Kant'spages. For one thing, Kant "tends to restrict the term 'consciousness'to apperceivingand the apperceivedas such," which makes it difficult to find in him the idea that impressionsor sensations can figure in consciousness without being apperceived (ibid.,p. 11). A more substantialproblem is that Sellarshas to find Kantseriouslyconfused in his thesis that space is the form of outer sense.3l One is bound to wonder whether Sellars has Kant wrong; and, since Sellars'sreading of Kantis, perfectlyproperly,shaped by Sellars'sown conviction about how we should conceive perceptual experience, that is inextricablybound up with wondering whether Sellars is mistaken in thinking that sound philosophy requires impressions or sensationsto be credited with the role he attributesto them. These are my questionsfor my next lecture.

30 For the image of guiding, see Scienceand Metaphysics,p. 16. 3' See ibid., p. 8: "the idea that Space is the form of outer sense is incoherent."