bonjour foundationalism and the external world

8/12/2019 BonJour Foundationalism and the External World

1/22

Foundationalism and the External WorldAuthor(s): Laurence BonJourSource: Nos, Vol. 33, Supplement: Philosophical Perspectives, 13, Epistemology (1999), pp.229-249Published by: Wiley

Stable URL: http://www.jstor.org/stable/2676104.Accessed: 14/02/2014 20:22

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 [email protected].

.

Wileyis collaborating with JSTOR to digitize, preserve and extend access to Nos.

http://www.jstor.org

This content downloaded from 75.97.192.97 on Fri, 14 Feb 2014 20:22:45 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/action/showPublisher?publisherCode=blackhttp://www.jstor.org/stable/2676104?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/2676104?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=black


2/22

Philosophical Perspectives, 13, Epistemology, 1999

FOUNDATIONALISM AND THE EXTERNAL WORLD

LaurenceBonJourUniversity of Washington

My aims in this paperare,first, to outline what I take to be a tenableversionof a traditional oundationalistaccount of empirical ustification and, second, toconsiderin a necessarilypreliminaryand schematic way the implications of sucha view for the justification of beliefs about physical or material objects.1Myultimateconcern is thus withwhat reasons there are for thinkingthat ourfamiliarbeliefs aboutthe physical worldare true,where I have in mind reasons: (i) thatdonot in some way beg the question by presupposingthe acceptability of otherbeliefs aboutphysicalobjects;and(ii) thatareat least inprincipleavailable hroughreflection and analysis to believers more or less like ourselves. It is to this ques-tion thatI now think that a traditional oundationalismoffers the only hope of anon-skeptical answer.

Until roughly forty years ago, a foundationalistaccount of empirical ustifi-cation was generally acknowledged as obviously correct and indeed as more orless the only seriousepistemological alternative o apervasive skepticism. Intheinterveningperiod, however,empiricalfoundationalismof any sort has been sub-jected to incessant attack and has come to be widely regarded as an obviouslyuntenableand even hopeless view. As a result, a verysubstantialproportionof theepistemological work in thisperiod has been aimed at the delineation anddevel-opment of non-skepticalalternatives to empirical foundationalism.This efforthas spawned various widely discussed views, such as coherentism, contextual-ism, externalism, and a variety of others that are less easily labeled. (Thoughexternalismcan be regardedas a version of foundationalism, t rejects the inter-nalist requirement,common to all traditionalversions of foundationalism,thatthe justification for a belief must be cognitively accessible to the believer. Myconcernhere is to defend a foundationalistview of themore traditional, nternal-ist sort, and it is such a view that subsequentuses of the term "foundationalism"shouldbe taken to refer to.)I myself have played a role in these developments, offering some of theargumentsagainst foundationalism and attempting to develop and defend thecoherentistalternative n particular.But having labored ong in the intriguingbut

This content downloaded from 75.97.192.97 on Fri, 14 Feb 2014 20:22:45 PMAll use subject toJSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp


3/22

230 / LaurenceBonJourultimately barren abyrinths of coherentism, I have come to the conviction thatthe recent anti-foundationalist rend s a serious mistake, one that is taking epis-temological inquiry n largely the wrong direction and giving undeservedcredi-bility to those who would reject epistemology altogether.I have two main reasons for this conviction. One is a dissatisfaction withthese recently popularalternatives,none of which seem in the end to be compat-ible even in principlewith ourhaving internallyaccessible, non-question-beggingreasons for thinking that our beliefs about the world are true.2The second, elab-oratedbelow, is that I now think that I can see a way to develop a foundationalistposition that avoids the objections to such a view that are in my judgment mosttelling: first, objections that challenge the very possibility of the foundationalbeliefs themselves; and, second, the objection thatfoundationalism,at least of thetraditionalvariety to be defended here according to which the foundationalbe-liefs pertain solely to the contents of one's own conscious states of mind, yieldsno adequatebasis for an inference to the physical world.I. The Foundation1. The concept of a basic belief

I begin with a closer examination of the central concept of a basic or foun-dationalbelief. This will lead to one of the main objections to empirical founda-tionalism, in relation to which the specific account of the foundationalbeliefs tobe offered here will be developed.As reflected in the familiar epistemic regress argument or empirical foun-dationalism, which I will not take the time to rehearse here, a foundational or"basic"belief is supposedto be an empiricalbelief that (a) is adequately ustifiedin the epistemic sense, but (b) whose epistemic justification does not depend oninference from furtherempirical beliefs that would in turn have to be somehowjustified. The main problem is to understandhow these two elements can besuccessfully combined. To say that such a foundational belief is epistemicallyjustified is to say thatthere is some sort of reason or basis or warrant or thinkingthat t is trueor at least likely to be true one, I will assume here, that s availableor accessible to the person in question. But this reason or basis or warrant s notsupposed to take the form of a furtherempirical belief, for example a belief thatthe belief originally n questionhas some feature hatcan be independently hownto be indicative of truth.What form then does it take?

Here the obvious and, I now believe, correct thing to say is that basic orfoundationalbeliefs are ustified by appeal o experience.But the difficulty,whichturns out to be very formidable, is to give a clear and dialectically perspicuouspicture of how this is supposed to work.Foundationalistssuch as C. I. Lewis and RichardFumerton,3among manyothers, have spoken at this point of "immediateapprehensionof" or "directac-quaintancewith" the relevant experiential content. Contrary o my own earlier



4/22

Foundationalism 231arguments,4 now believe thatthere s a way to understand uch formulations hatleads to a defensible view. On the surface, however, this answer is seriouslyproblematic n the following way. The picture it suggests is that in a situation offoundational belief, there are two distinguishable elements, in addition to therelevant sensory experience itself. First, there is an allegedly basic or founda-tional belief whose content pertains to some aspect of that experience. Second,there s what appears o be a second, independentmental act, an act of immediateapprehensionof or directacquaintancewith the relevantexperiential eature.Andit is, of course, this second act that is supposed to supply the person's reason forthinking that the belief is true.

But the problem is to understand he natureand epistemic status of this sec-ond mental act itself. If it is construedas cognitive and conceptual, having as itscontent something like the proposition or claim that the experience in questionhas the specific character ndicated by the belief, then it is easy to see how thissecond mental act can, if it is itselfjustified,rovide a reason for thinkingthat thebelief is true, but hard to see why it does not itself require ustification of somefurthersort, some reason for thinkingthat ts propositionalor assertive content istrue or correct. And to say simply that acts of immediate apprehension,unlikeordinarybeliefs, somehow cannot by their very naturebe mistaken s to stipulatethatthis problemdoes not exist without offering any clear explanationof how andwhy this is so.

If, on the other hand, the mental act of immediate apprehensionor directacquaintance s construed as non-cognitive and non-conceptual n character,asnot involving any propositionalclaim about the characterof the experience, thenwhile no further ssue of justification is apparently aised, it becomes difficult tosee how such an act can provide any reason or other basis for thinking that theoriginal allegedly foundationalbelief is true. If one who is directly acquaintedwith an experience is not therebypropositionallyaware that t has such-and-suchfeatures, in what way is his belief that he has an experience with those featuresjustified by the act of direct acquaintance?

It is this dilemma5 hat has always seemed to me to be the most fundamentalobjectionto empirical oundationalism,andthe core of the first partof the presentpaper will be an account of how it can be answered. I will begin by consideringthe somewhat tangentialbut more easily accessible case of the justification of ameta-belief about a conscious, occurrentfirst-orderbelief or assertive thoughtand then turn to the justification of beliefs about sensory experience.2. Metabeliefs boutone'sownoccurrent eliefs.

As I work on this paper,I believe that I am having various occurrentbeliefsor assertive thoughts about foundationalism and its problems. For example, Ibelieve that I presently have the occurrentbelief or thoughtthat foundationalismis much more defensible than most philosophers think. What is my justificationfor this second-ordermeta-belief (assuming that there is any)?



5/22

232 / LaurenceBonJourAs already indicated, the naturalanswer to this question, which is also the

one that I want to elaborate and defend here, is to appeal to the conscious expe-rience nvolved in having the occurrentbelief or thought in question. But it iscrucial for present purposes that the nature and status of this experience be un-derstood n the right way. My suggestion is that an intrinsic and essential aspectof having an occurrentbelief is being consciously aware of the two correlativeaspects of its content: first, its propositionalcontent, in this case the propositionthatfoundationalism s much more defensible thanmost philosophers hink;and,second, the assertoryrather han, e.g., questioningcharacterof one's entertainingof that content. These two awarenesses (or rather wo aspects of one awareness)are, I am suggesting, not apperceptive or reflective in character: hey do notinvolve a second-ordermental act with the propositionalcontent that I have thebelief in question. Instead, they are constitutivef the first-level state of beliefitself in that they are what make it the very belief that t is, rather han some otherbelief or a different sort of conscious state altogether.The point here is simplythat occurrentbelief or thought is, after all, a conscious tate, and that what oneis primarilyconscious of in having such a belief is precisely its propositionalandassertive content; not to be consciously aware of that content would be not tohave that conscious, occurrentbelief at all.

It is this account of the experiential aspect of occurrentbelief that seems tome to allow an escape between the horns of the anti-foundationalistdilemmaposed in the previous section. The crucial point is that the most fundamentalexperience involved in having an occurrentbelief is neither second-orderap-perceptiveor reflective awareness hat t has occurrednor a purely non-cognitiveawarenessthat fails to reflect the specific characterof the belief and its content.Instead it is an intrinsic and constitutive awareness of the propositional and as-sertive content of the belief.6Because of its non-apperceptive, onstituentcharacter, his "built-in"aware-ness of content, as it might be described, neitherrequiresany justification itself,nor for that matter even admits of any. Indeed, as far as I can see, such a non-apperceptive, onstituentawarenessof content is strictlyinfalliblen prettymuchthe way that foundationalistviews have traditionallyclaimed, but which mosthave long since abandoned.Since it is in virtue of this constitutive or "built-in"awareness of content that the belief is the particularbelief that it is with thespecific content that it has, rather han some other belief or some other sort ofstate, there is apparentlyno way in which this awareness of content could bemistaken simply because there is no independentfact or situation for it to bemistaken about.It is by appeal to this intrinsic, constitutive awareness of propositional andassertive content that the meta-belief that I have the first-level belief can, I sug-gest, be justified. Such a constitutiveawarenessof contentseems obviously enoughto constitute n andby itself, at least if otherthings are equal, a reason for thinkingthat the meta-belief that I have an occurrentbelief with that very content is true(or, perhaps more realistically, for thinking that a meta-belief that gives a less



6/22

Foundationalism 233detailed, more abstract description of the first-order content, but one that theactual, more specific content falls under, s true). The point here, elaborated ur-ther below, is that the meta-belief is a descriptionf the very content involved inthe constitutive awareness of content, so that by consciously having that consti-tutive awareness,I am in an ideal position to judge whetheror not this descriptionis correct.

In this way, such a meta-belief can, I suggest, have precisely the epistemicstatus requiredby foundationalism: t can be justified in the sense of there beinga clear and internallyaccessible reason for thinking that it is true, but the reasonin question can be such as to avoid any appeal to a furtherbelief or belief-likestate that would itself be in need of justification though we now see that t is thefirst-level constitutive or "built-in"awareness of content, rather han the meta-belief that it justifies, that turns out to be the ultimate source of justification.The infallibility of the "built-in"awarenessdoes not, of course, extend to theapperceptivemeta-belief: it would still be possible to apperceptivelymisappre-hend one's own belief, i.e., to have a second-level belief that does not accuratelyreflect the content contained n the constitutive or "built-in"awarenessconstitu-tive of the first-level belief. Such a mistake might be a case of mere inattention,or it might result from the complexity or obscurity of the belief content itself orfrom some furthermistake or problem. But unless there is some special reason ina particularcase to think that the chances of such a misapprehensionare large,this possibility of error does not seem to prevent the second-level meta-belieffrom being justifiable by appeal to the first-level constituent awareness. This isjust to say thatwhile suchjustification is defeasible in variousways, it is adequateuntil and unless it is defeated, rather than requiring an independent and priorshowing of reliability.73. Beliefsabout ensory xperience.

If I am right, the foregoing provides at least a sketch of how a certain specificsort of belief, viz. an apperceptivemeta-belief about an occurrentbelief of one'sown, can be basic in the sense of there being an internally available reason whyit is likely to be true without that reason depending on any furtherbelief or othercognitive state that is itself in need of justification. I believe that an analogous,albeit somewhatmore complicated account can be given of how beliefs aboutthecontents of other kinds of experience, especially sensory or perceptual experi-ence, can be similarly basic.

Consider then a state of, e.g., visual experience, such as the one that I ampresentlyhaving as I sit at my desk. Like an occurrentbelief, such an experienceis a conscious state. What this means, I suggest, is that, in a way thatparallels theaccount of occurrentbelief or thought offered above, it automatically nvolves aconstitutiveor "built-in,"non-apperceptiveawarenessof its own distinctive sortof content, viz. sensory or experiential content. And, again in parallel fashion,such a constitutiveawarenessof sensory content is in no need of justification and



7/22

234 / LaurenceBonJouris indeed nfallible in the sense thatthere s no sort of mistakethat s even relevantto it: since it is this awareness of sensory content that gives my experientialstatethe specific content that it has and thus constitutes it as the specific experientialstatethat t is, there s no logical room for this awareness o be mistakenaboutthecontent in question. And thus such an awareness of sensory content is also ap-parentlyavailable to justify foundationalbeliefs.Before we embracethis idea too eagerly, however, there s a recently popularobjection that needs to be addressed.This objection, which is present with vari-ous degrees of explicitness in the thoughtof philosophers as differentas Popper,Sellars, Davidson, and Rorty,8begins with the idea that the distinctive content ofa sensory or perceptualexperience, that content the awareness of which makesthe experience the very experience that it is, is non-propositional and non-conceptual in character where what this means is at least that this most basicawarenessof sensory content is not couched in general or classificatory terms, isnot a propositionalawarenessthathe experience falls under general categoriesor universals. And from this the conclusion is drawn that such an awarenesscannot stand in any intelligible justificatory relation to a belief formulated inpropositionaland conceptual terms, and hence that the relation between the twomust be merely causal. As Davidson puts it:

The relationbetween a sensation and a belief cannot be logical, since sensations arenot beliefs or other propositionalattitudes.What then is the relation?The answer is,I think,obvious: the relation s causal. Sensations cause some beliefs and n this enseare the basis or groundof those beliefs. But a causal explanationof a belief does notshow how or why the belief is justified.9

And if this were correct, what I have been calling the constitutive or "built-in"awarenessof sensory content, even though it would still undeniablyexist, wouldbe incapableof playing anyjustificatoryole and thus would apparentlyhave noreal epistemological significance.

The premise of this objection, viz. the claim that sensory experience is es-sentially non-conceptual n character,seems to me both true and important.Atleast part of the point is that the content of, e.g., the visual experience that I amhaving as I sit in my study is far too specific, detailed, and variegated to beadequately captured n any conceptual or propositional formulation that I ampresently able to formulateor even understand.Moreover,even if we imagine anideally complete andfine-grained onceptualdescription, t seems clear that hink-ing in conceptualtermsof, e.g., very specific shades of color in some complicatedpattern s not at all the same thing as actually experiencing the patternof colorsitself.But although I must confess to being one of those who has in the past beeninfluenced by this objection, it now seems to me that its conclusion simply doesnot follow from its premise. For even if we grant and indeed insist that the spe-cific content of the experience is itself non-propositionaland non-conceptual, t,



8/22

Foundationalism 235like various other kinds of non-conceptual phenomena, can of course still beconceptuallydescribedwith various degrees of detail and precision. The relationbetween the non-conceptualcontent and such a conceptual description thereofmay not be logical,as Davidson uses the term, but it is also obviously not merelycausal. Rather t is adescriptiveelation.And where such a relationof descriptionexists, the characterof the non-conceptualobject can obviously constitute a kindof reason or basis for thinking hatthe description s true or correct(or equally, ofcourse, untrueor incorrect).

Such a reason is, of course, only available to one who has some sort ofindependent access to the characterof the non-conceptual item, i.e., an accessthat does not depend on the conceptual description tself. In the most usual sortsof cases, e.g., where it is some physical object or situation hat s being described,one could have an access that is independentof the description n question onlyvia a second conceptual state embodying a second, perhaps more specific de-scription, and this second descriptionwould of course itself equally require us-tification, so thatno foundational ustificationwould result.But in the very specialcase we are concerned with, where the non-conceptual item being described isitselfa conscious state, my suggestion is that one can be awareof its characterviathe constitutive or "built-in"awareness of content without the need for a furtherconceptual description and thereby be in a position to recognize that a beliefabout that state is correct without raising any further ssue of justification.

Thus where I have a conscious sensory experience, I am, as already argued,awareof its specific sensory content simply by virtue of being in that experientialstate. And therefore if (i) an apperceptivebelief that I entertainpurportsto de-scribe or conceptually characterize hat perceptual content, albeit no doubt in-completely, and (ii) I understand the descriptive content of that belief, i.e.,understandwhat an experience would have to be like in order to satisfy thatconceptual description, then I seem to be in a good, indeed an ideal, position tojudge whetherthe conceptual description s accurateas far as it goes; and if it is,to be therebyjustified in accepting the belief. Here again there is no reason tothink that mistake is impossible and thus no reason to think that such an apper-ceptive belief is itself infallible. But as long as there is no special reason forsuspecting that a mistake has occurred, he fact that such a belief seems to accu-rately characterize he conscious experience that t purports o describe provides,I suggest, an entirely adequatebasis for thinking that the description is correct,and thus an adequatebasis for justification.

Here indeed we seem to have exactly the sort of direct comparison or "con-frontation"between a conceptual description and the non-conceptual chunk ofreality that it purports o describe which seems intuitively to be essential if ourconceptual descriptions are ever to make contact with independentreality in averifiable way, but which many philosophers, myself again alas included, haverejected as impossible. Such a confrontation s only possible, to be sure, wherethe reality in question is itself a conscious state and where the description inquestion pertains to the conscious content of that very state, but in that very



9/22

236 / LaurenceBonJourspecific case it seems to be entirelyunproblematic.Thus it turnsout thatthe givenis, after all, not a mythI am inclined to suspect that it is this sort of non-apperceptive, ntrinsicawareness of the content of a conscious state thatepistemologists such as thosementioned earlier had at least primarily in mind in their use of the notion of"immediateawareness"or"directacquaintance."But if this is right, thendiscus-sions of direct acquaintancewere often needlessly obscure, suggesting as theydid some sort of mysteriously authoritativeor infallible apprehensionof an in-dependentcognitive object, rather han an awareness that is simply constitutiveof aconscious state itself. Moreover,the suggestion of someproponentsof directacquaintancehatonemight possibly be directlyacquaintedwith physicalobjectsor their surfaces simply makes no sense on the present account of what directacquaintance eallyamountsto (thusvindicatingthe frequentclaim of other pro-ponents of this concept that one can be directlyacquaintedonly with one's ownmental states).10 also believe that it is this sort of constitutive or "built-in"awarenessof the content of aconscious state thatChisholmhad in mindin speak-ing of states that are "self-presenting," l a terminology that seems rathermoreappropriate o the phenomenon in question than"acquaintance."Theforegoingdiscussion seems to me to establish that apotentialfoundationforempirical ustificationgenuinelyexists, consisting, moreor less as traditionalfoundationalists hought,of beliefs about the content of sensoryexperience (andthecontent of otherconscious states).In the second part of the paper,I will try tosay somethingaboutwhether and how it is possible to exploit this foundation nsuch a way as to conferjustificationon beliefs aboutphysical objects.II. The Inference to the External World1. Theconceptualformulationof sensory experience.

The most immediatequestion atthis pointconcerns the specific characterofthe allegedly foundationalbeliefs about sensory experience. I am assuming thatfor the contentof such experience to play any epistemic role, it is necessary thatit be conceptuallyformulated n beliefs that areexplicitly about t. Things wouldbe far easier if it were plausible to hold, as some have,12 hat non-conceptualcontentcould somehowdirectly ustify other sortsof beliefs, e.g., beliefs that aredirectly aboutphysical objects, without thatcontent needing to be itself formu-lated in conceptual terms.I believe, however, that any such view is untenable,thatDavidson and the others are at least right in thinkingthat there is no intelli-gible relationof justification between non-conceptualsensory content and con-ceptual beliefs that are not directlyaboutthatcontent itself. Thus it is impossible,in my view, for a foundationalist o avoid the issue of whatform suchconceptualdescriptions of non-conceptualcontent might actually take.Here there are two main possibilities. One is that the content of sensoryexperience is apprehendedn phenomenologicalterms thatare as close as possi-



10/22

Foundationalism 237ble to the apparentcharacterof the given experience itself in terms of some-thing ikethe puresense-datum onceptsenvisaged byvariousphilosophersearlierin this century.13The advocates of such views have usually assumed that theresultingdescription of, e.g., visual experience would be in terms of patches ofcolor arranged n visual space,14and I am inclined to tentatively accept such apicture.The second main possibility is that we conceptually grasp sensory contentprimarily n termsof thephysicalobjects andsituations hatwe would be inclinedon thebasis of thatexperience, otherthingsbeing equal,to thinkwe areperceiv-ing. Thuson this view, for example, my primaryconceptual graspof my presentvisual experience would characterize t roughlyas the sort of experience that inthe absence of countervailingconsiderationswould in fact lead me to thinkthat Iam sitting about a foot away from a largewooden table,uponwhich a computer,severalbooks,many pieces of paper,etc., areresting (all of whichcould be spelledout at great length). The usual way of puttingthis is to say thatwhat I am con-ceptually aware of is certainphysical-objectappearances or apparentphysicalobjects or, in adverbialterms, ways of being "appeared o" thatarecharacter-ized in physical object terms. Where the appearance n question is a visual ap-pearance,we may say alternatively hat t looksas thoughthere areobjectsof thesorts indicated,andanalogously for othersensorymodalities.15These two views arenot entirelyincompatible: t would be possible to holdthat we apprehendsensory contentpartly or even entirely in each of these twoways. But the main questionis which is epistemologically primary,and heretheissues arequite complicated.One obvious concern is the availability of each sort of characterization oordinarybelievers. Contrary o whathas sometimes been suggested, I can see noreason why it would not be possible for each of us to have the conceptual re-sources to give a phenomenological description of experience to any level ofprecision and accuracy desired, even though it seems obvious that we wouldalways fall short of an ideally complete description.But it seems abundantlyclear thatmostpeople do not in fact possess suchconceptualresources.Ordinarypeople are capableof giving reasonablyprecise andaccuratephenomenologicalor at least quasi-phenomenologicaldescriptionsof some aspects of theirexperi-ence, anda person,such as an artistora wine taster,who cultivatesthis abilitycanoften do a good deal betterin a particulararea.But it is doubtfulwhethereventhose whose abilities of this sort arethe best developed are in a position to con-ceptually formulate a strictly phenomenological characterizationof their totalsensoryexperiencethat is sufficientlydetailedandprecise to captureall or evenmost of itsjustificatorysignificancefor claims aboutthephysical world (assum-ing for the momentthat it has such significance).This difficulty may seem to show that the alternativecharacterizationofexperience in terms of physical-objectappearancesmust be the one that is pri-mary forjustification. But do such characterizationsprovide an adequatebasisfor thejustificationof physical object claims?



11/22

238 / LaurenceBonJourIn thinkingabout this question, it is importantnot to misunderstand he sig-nificance of such a description of experience in physical-object terms. In partic-

ular, t is crucially mportant o distinguisha descriptionof experiencethatmerelyindicateswhat sort of physical objects and situationsseem or appear,on the basisof that experience, to be present from one that embodies some furthercausal orrelationalclaim aboutthe connection between experience and the physical realm,one whose justification would clearly have to appeal to something beyond theexperienced content itself. Thus the claims about physical appearancesor waysof being appeared o that constitute our conceptual formulationsof sensory ex-perience must be understood n what Chisholm has called the "descriptive,non-comparative" ense of the terms or concepts in question,16 or only in that sensecan the claim to be "appeared o" in a certain way be adequately ustified simplyby appeal to our constitutive or "built-in"awareness of the non-conceptualsen-sory content in the way indicated above.On reflection, however, it is far from obvious exactly what such "descriptive,non-comparative" haracterizations f experience in physical object termsreallyamount o or what their epistemic significance might be. In giving them, we seemto be relying on a tacitly grasped and, we think, mutually understoodcorrelationor association, perhaps learned or perhaps at least partially innate, between thespecific characterof our sensory experiences and the physical situationsof whichthey are taken to be appearances,a correlation hat we are confidently guided byin the vast majority of cases, even though we are unable to formulate it at allexplicitly. To speak here of a "correlation"might suggest that it is a mere orre-lation, that the experiential content and the correspondingpropositional claimaboutphysical objects are only externally coordinated,without being connectedwith each other in any more intimate way, but this seems quite obviously wrong.On the contrary, t seems intuitively prettyclear that the experientialcontent is initself somehowtrongly suggestive of and in some interestingway isomorphic tothe correlatedphysical situation. Ultimately I will want to suggest that this iso-morphismneeds to be spelled out in something like sense-datumor phenomeno-logical terms in order for the justificatory force of sensory experience to beadequately captured. But the point for the moment is that it cannot simply beassumedhat this correlationbetween sensory content and physical objects re-flected in such descriptions of experience is reliable, that beliefs adopted on thebasis of it are likely to reflect in an accurateway what is really going on in thephysical world.

Given this understanding f a physical-objectappearance haracterization fthe content of sensory experience, the question is then how an inference fromsuch a characterization o claims about the physical world might be justified?Here there seem to be only two alternativesworthy of mention.First. One answer is advocated by H. H. Price17and, in what seems to be aratherseriously qualified form, by Chisholm.18The core idea of this view is thatthe mere occurrenceof a physical-object appearanceor state of being appeared oconfers prima acie justification on the correspondingphysical claim. Chisholm's



12/22

Foundationalism 239somewhat more explicit version of this solution appeals to a supposed logicalrelation of "tending o make evident" that is alleged to exist between claims orbeliefs about sensory appearancesand the correspondingclaims or beliefs aboutthe actual perception of physical objects. Thus, it is claimed, my belief that mypresent visual experience can be correctly described as an appearanceof a com-puter on a table (or my belief that I am being appeared o in the correspondingway) tends o makeevidentmy belief that I am actually perceiving such a com-puter and table and thus that they really exist in the physical world. Such anevidential tendency is capable of being defeated by countervailingevidence, butwhere no such defeater is present, Chisholm suggests, the claim of genuine per-ception and so of correspondingphysical reality is justified.The difficulty with this sort of view is that it seems very implausible tosuppose that such a logical relation of "tendingto make evident" or "tendingtojustify" genuinely exists between an individual belief about particularphysical-object appearancesand the corresponding belief about physical reality. To besure, Chisholm'sclaim is not thatany such relation s discerniblea priori n itself,but only that it is an a prioriconsequence of the "general presupposition"or"faith,"roughly, that epistemological success is possible19 a view that alreadyseems to have serious skeptical implications. But over and above that, the prob-lem is that if a belief about a physical-object appearance s construed, as I havearguedthat it must be construed, as merely a useful though oblique way of de-scribing the non-conceptualcontent of sensory experience, then there is no ap-parentreason to think that a single such belief could by itself have any direct orimmediatebearingon the truthor likely truthof the correspondingphysical claim.Why should the occurrence of a single conscious state with a particularsort ofnon-conceptualcontenthave by itself any bearingat all on whethera specific sortof physical object exists in the mind-independentworld? It thus seems hard tomake sense of any such relation of "tending to make evident" or "tending tojustify," whethera priori ustifiable or not.Second. It might be suggested that the basis for the needed inference fromsensory appearanceto physical reality is to be found in two facts about suchphysical-object appearances hat were noticed by Locke and Berkeley, amongothers:first, the involuntary, pontaneouscharacterof their occurrence;and sec-ond, the fact that the physical descriptions they embody fit together with eachother n a coherentashion, presentinga relatively seamless and immensely com-plicated picture of an ongoing physical world.20These two facts are the centralones appealedto by Locke, in justifying his inference from sensory ideas to theexternalworld; and also by Berkeley, in justifying his inference to the God whois supposed to produce our ideas. In both cases, the underlying dea, rathermoreexplicit in Berkeley, is that some explanations needed for the combination ofinvoluntarinessand coherence, with the conclusion advocatedby the philosopherin question being claimed to be the best explanation of the facts in question.21I once believed that an inference on something like this basis, to Locke'sconclusion rather han Berkeley's, was ultimately cogent, but this now seems to



13/22

240 / LaurenceBonJourme mistaken.No doubtthe combination of spontaneity andintricatecoherencerequires some explanation,which is just to say that it is unlikely to result frommere chance.But why anexplanation n physicalobjectterms, rather han any oftheotherpossibilities (includingBerkeley's) thatso obviouslyexist?Whatmakesthe physical explanationso obviously salient is our ingrained nclination to de-scribe the experientialcontent in physical-object terms (or, indeed, to leap di-rectly to a physical claim with no explicit acknowledgmentof the experientialpremise). Since it is, however, this very correlationbetween sensory experienceandphysical object claims whose justification is at issue, no appeal to that cor-relation can as such have anyjustificatoryweight. Thus some further, ndepen-dentreasonneeds to be offered for thinking thatthe physical-objectexplanationis the best one.

The obvious response at this point would appeal to the idea, briefly men-tioned earlier,that the correlationbetween experientialcontent andphysical ob-jects is not amerecorrelation, hatthere aredetailed featuresof the experientialcontent itself that are strongly suggestive of and systematically isomorphic orstructurally imilar to the correlatedphysical situations. It is apparentlyonly ifsomething like this is so that therecould be any very clear reason to prefer thephysical explanationof experience to the variousothers thatmight be given. Butthis seems to indicate n turn hattheconceptualcharacterization f experience inphysical objecttermscannot be takenas epistemologicallyprimaryafterall, thatsomethingmuch closer to a sense-datumorphenomenologicalcharacterizationsneeded if anadequatebasis for the inference to thephysicalworld is to be found.In the next section, relying on the pioneering work of Price,22 will attempttosketch, in a necessarilybrief andschematic way, some of the features of experi-ence that mightplay such a role. Then, in the subsequentsection, I will give anequally schematic accountof how this account of experiencecan be used to jus-tify an inference to the externalphysical world.2. Apreliminaryphenomenological characterizationof sensory experience.

We arenow in aposition to see thatthecharacterization f thenon-conceptualcontent of sensory experience in termsof physical-objectappearances s episte-mologically unsatisfactory n two closely relatedways: First, it depends on thevery correlation between experience and physical objects whose reliability isultimately at issue, thus making it difficult or impossible to avoid begging im-portantquestions. Second, and even more importantly, t tends to obscure thevarious fine details and nuances of experience upon which, I am suggesting,thejustificationthat s sought mustultimatelydepend.Think,forexample, of theextremely large number of differentvisual presentationsthat could be lumpedtogether as appearancesof a table or even as appearancesof a relatively specificsort of table at close range.A crucialpart of the overall issue is precisely what itis that warrantsviewing these qualitatively distinguishableexperiences as allappearancesof one and the same specific sort of physical object, something thatcanbe adequatelydealtwith, I believe, only by consideringthatqualitativechar-



14/22

Foundationalism 241acter n its own right,rather hanburying t underthephysical-objectappearancedescription.

Such a qualitativecharacterization f sensory experience in its own rightissomething that very few philosophers have even attemptedto give: partly be-cause of the extremedifficulty of doing so in a clear way while operatingwith apublic language whose conceptualresourceshave been shapedby morenarrowlypracticalconcerns; andeven morebecause of the greatlength thatwould seem-ingly be required to do an even approximatelyadequatejob. Given the spacelimitationsof the presentessay, I propose to simplify thetask in threeways. First,I will largely confine myself to a sketch of some of the featuresof experiencepointed out by Price in his much more extensive account in Perception,23 anaccount that while surely debatableon many points of detail, still seems to mefundamentallycorrect in its main outlines. Second, I will make use of a tool,mainly for the sake of brevity, thatalthoughacceptablein a preliminarysketch,would be obviously objectionable n a full account:namely, the use of physical-object descriptions to direct the reader'sattention to the featuresof experiencethat are at issue. While the justificatory force of such experiential features inrelation to physical-object claims depends on their being ultimatelydescribablein ways that are logically independent of such claims, the plausibility of thegeneral sort of justificatory argumentthat is in question can, I think, be ade-quately appreciated or presentpurposeswithout actually giving such an inde-pendentdescription.Third, will employsense-datum erminologyasaconvenientway of referring to various relatively specific aspects and features of sensoryexperience, even thoughI thinkthat the idea that sensory experience is literallyanacquaintancewithentities of the sort thatsuchterminologysuggests is almostcertainlymistaken.24

In these terms, the conclusion at which I am aiming is that the specificcharacteristicsof our sense-data,andespecially theirspatialcharacteristics,aresuch as to be easily and naturallyexplainable by supposing that they are sys-tematically caused by a relatively definite world of (mostly) solid objects ar-ranged in 3-dimensionalspace, andby no otherhypothesis that is not, in a wayto be furtherexplained below, essentially parasitic on that one. What then arethese characteristics?First.Think of the visual sense-dataexperiencedwhile doing what we com-monsensically think of as moving around a medium-sized physical object, per-haps the matchboxinvoked by Price, at close range,observing its various sidesfrom variousdistancesand angles underrelativelyuniform and"normal" ondi-tions of lighting. In relation to each side of the box, therewill be a collection ofspatialsense-data,varying more or less continuously in shape, intuitively as theangle fromwhich the object is being perceivedis altered.Withineach such col-lection, therewill be a much smaller set of what might be called central sense-data, having geometrically similar2-dimensional shapes and also satisfying thefollowing two conditions: (i) The 2-dimensional shapes of the other data in thecollection canbe regardedas perspectivaldistortions in variousdirections of theshapesof thesecentraldata.(ii) At aconstantapparentdistance,the2-dimensional



15/22

242 / LaurenceBonJourshapes of the central data in the collection corresponding to each side of theobject are such as could be fitted together in three dimensions to form a closed3-dimensional shape, one which is in fact the same as the shape commonsensi-cally ascribed to the box; and perspectivally distorted versions of two or threesuch 2-dimensional shapes can be experienced as adjoinedto each other in waysthatcan be viewed as perspectivallydistorted mages of that3-dimensional hape.25

Second. Think now of the sense-data corresponding o intuitively less ade-quatevisual perceptionsof the matchbox:perceptionsat greaterdistances, hroughdistortingmedia of various sorts (e.g., wavy glass), and undervaried conditionsof lighting. Here too the sense-data experienced can be regardedas related to theintuitively more adequatedata already discussed via various sorts of distortionthat are still perspectival in a somewhat broadersense. And it is in general pos-sible, at least in principle, to experience what Price calls a "gradual ransitionseries" eading from one of these intuitively less adequate ense-datato one of theintuitively more adequate ones via a series of intermediatesense-data in whichthe changes between any two members in the series are as small as one likes.Third.Considernow the tactual sense-data that would be experienced whiletouching the matchbox, running one's hands over its various sides and edges.Here too, there will be central sense-data having shapes that correspond intu-itively to the various faces of the box, and that can be experienced much moredirectly than the visual data as fitting together to form a 3-dimensional shape.Therewill also be other sense-datathat depart rom these centraldata via varioussorts of (milder) distortion and incomplete perception. Price's view, which I aminclined to accept, is roughly that the collection of tactual sense-data can becoordinatedwith the largerand more complicated collection of visual sense-databy virtue of the geometrical propertiesthat sense-data in each group, especiallythe central ones, jointly instantiate.26Taking the visual and tactual sense-datatogether,we have corresponding o our commonsensical matchboxroughly whatPrice calls a "family"of sense-data,where the centralstructure f such a "family"is constitutedby the relations n which its membersstandto the centralsense-dataandthereby o the 3-dimensionalshape to which those data are related.His claim,which I believe to be correct (though this surely has not been adequately estab-lished by the foregoing sketch), is that all of this structureamong the sense-data(as well as the furtherpoints to follow) is in principle discernible without refer-ence to the physical objects to which we intuitively think the sense-data n ques-tion are related.

Fourth. Sense-data pertaining intuitively to other senses, mainly aural, ol-factory, and thermal sense-data, can also be viewed as members of such "fami-lies," with the main connection being the way in which their strengthor intensityis coordinatd with the visual or tactualdatathat are simultaneouslyexperienced,increasing in general as one "moves"througha "gradual ransitionseries" in thedirection of the central data.Fifth. Consider now the sequences of sense-data from different "families"that intuitively correspond to moving around in space, experiencing first oneobject, then an object adjacent o the first (where it is usually possible to simul-



16/22

Foundationalism 243taneouslyexperiencemembersof the two families), then a furtherobject adjacentto the second one, etc., etc., in various directions, perhapsreturningultimately tothe original object from a different perspective.

Sixth. Considerthe various patternsof sense-data that intuitively reflect theways in which one object can block, either wholly or partially, our view of an-other and the way in which these patternsrelate to those connected with move-ment, as just discussed.

Seventh. Consider he extremely varied sense-datathat ntuitively reflect thecausal actions of common-sense objects on each other, e.g., such things as theeffects of a source of heat on objects that are subject to melting or burning, to-gether with the ways in which the sense-data corresponding o such effects arecorrelatedwith those intuitively reflecting movement from one such object toanother.Here the central idea is that the sense-data corresponding o causal ef-fects in general vary in intensity in a way that correlates with the intuitive dis-tance between the two objects as reflected in the sense-data ntuitively correlatedwith movement. There are also patternsof sense-data that reflect what we intu-itively think of as various shielding effects.Eighth. Think, finally, of the ways in which the families of sense-data andtheir relations to each other as thus far specified change over time in ways thatintuitivelyreflect both changes in the objects in question andthe movement of theobserver.

Though the foregoing is only the merest sketch of matters hat would in factrequire a book much longer than Price's to be described in anything like fulldetail, I hope that it is adequateto indicate in a very approximateway the factsaboutour experience upon which the inference to the physical world must, in myjudgment, inally rest.27One importantpoint, which I have so far ignored, should,however, be added:The features of experience that I have described are not re-alized in any very complete way by the sense-data actually experienced by agiven person or even by all persons taken together.Many of them pertain n largepart to sense-datathat are merely possible or (better) obtainable: sense-data thatwould be experienced if, from an intuitive standpoint, the observer's point ofview or the conditions of observation were to change in ways that it does not infact change. This means that there are really two distinguishable nferences (orattempted nferences) involved in the attempted ustification of belief in the phys-ical or material world on the basis of sense experience: one from actual experi-ence to obtainable xperience,which seems to be essentially nductive n character;and a second abductive or explanatory nference from actual and obtainable ex-perience to physical objects. While the first of these two inferences is by nomeans entirelyunproblematic, will focus my attentionhere on the second, wherethe problems and difficulties are obviously much greater.283. The inference to the physical world.

On the view that I want to defend, one main premise for the inference fromexperienceto the physical world is the detailed accountof our sensory experience



17/22

244 / LaurenceBonJourof which a very preliminary sketch has just been given. A second is that someexplanation s needed for this complicated patternof experience, that t cannot beplausibly viewed as either ust a matterof chance or as somehow an ultimate andnot furtherexplicable brute fact (the latterbeing essentially the view of the phe-nomenalist). The intuitive credentials of this second premise seem pretty obvi-ous, and I will not offer any more explicit defense of it here.It seems quite clear that at least one possible explanation for the experien-tial pattern n question is the approximatelycommonsensical idea that my sen-sory experiences are systematically caused by a realm of 3-dimensional objects:(i) having at least approximately he shapes defined by what I have called thecentralsense-data,(ii) throughwhich I move in such a way as to change my pointof view, (iii) which are spatially related to each other in the ways reflected in theexperientialsequences producedby my apparentmovement, and (iv) which havecausal propertiesand change over time in the ways corresponding o the relevantfurtheraspects of experience. Obviously all of this could and ultimately shouldbe spelled out in vastly greaterdetail, but the general idea should be clear enoughfor presentpurposes.Assuming then that this explanation, which I will henceforth refer to as thequasi-commonsensicalypothesis (with the reason for the qualification to bediscussed later), is at least one relatively adequateexplanation of the details ofour sensory experience, the question is what other explanations are available.Here it will be helpful to draw a distinction between two fundamentallydifferentkinds of explanation. On the one hand there are explanations, such as the oneinvolving the quasi-commonsensicalhypothesis just indicated, that explain thefeaturesof experience by appealdirectly to the basic featuresof the objects in thehypothesized world; on the other hand, there are explanations, of which Berke-ley's is one obvious example, that explain experience by appeal to the combina-tion of (i) something like a representationf a world, together with (ii) someagent or mechanism that produces experience in perceivers like us in a way thatmimics the experience that we would have if the representedworld were actualand we were located in it, even though neither of these things is in fact the case.Adopting a useful technological metaphor,I will refer to the former sort of ex-planationas an analogexplanation and the latter as a digitalexplanation.In these terms, it is, I submit, far from obvious that there is any alternativeanalogexplanationavailable that s even approximatelyas good as thatprovidedby the quasi-commonsensicalhypothesis. In brief, the main points that make theprospectsfor such an alternative eem so dim are the following: (i) It is very hardto see how the spatialfeaturesof experience could be explained in analog fashionby anythingotherthana spatialworld:certainly a multi-dimensionalworld seemsneeded to account for the various sorts of experiential paths that returnto thesame experientialstartingpoint. (ii) For essentially the same reason, a world thatis spatially 2-dimensional does not seem to have enough internal structuretoaccountfor all of the possible experientialsequences and variations.(iii) Thus analternativeanalog hypothesis would apparentlyhave to involve a 3-dimensionalspatial world containing objects whose shapes and spatial relations differ sys-



18/22

Foundationalism 245tematically fromthose thatareactuallyreflected in ourexperience, with the ex-periences in question thusreflecting what amounts to a systematicdistortion ofwhat is really to be found in the world. But if the differences in question aresupposedto be large enoughto be interesting, hen thevery featuresthatmake thequasi-commonsensicalhypothesiswork so well also make it hard o see how suchan alternativehypothesiscould workat all. Here the crucial point is thatwhile itis easy to imagine a systematicdistortion of secondaryqualities like color andtaste,whose relationsto otherqualitiesof thesame sortand of othersorts areverylimited and unstructured,and thus which can be varied more or less indepen-dently, it is difficult or impossible to imagine a similardistortionof systemati-cally interconnectedprimaryqualities like shape, size, and spatial orientation,where distortions in one respect demandcompensatingdistortions in other re-spects, and those in still otherrespects, etc., with no prospect of ever achieving astable and coherent view.29At least it seems fair to say that no one has everactuallydescribedan alternativehypothesis involving such systematicprimaryqualitydistortion n even an approximateway, making it reasonableto place theburdenof proof on those, if there areany, who seriously believe that such alter-natives genuinely exist.But even if I am rightthat thereare no serious contendersfor an alternativeanalogexplanationof sensoryexperience, thereare still of coursevariouspossi-ble digitalexplanations.Indeed, it is obvious that given any analogexplanation,therearealwaysguaranteed o be indefinitelymanypossible digitalexplanations,since anything hatcan be explainedby appealto thefeaturesof agiven worldcanalso be explainedby appealto arepresentationof thatworld (perhaps n concep-tual terms or perhaps n a form analogous to a map) together with a translatingmechanism of some appropriate ort. In the presentcase, as alreadysuggested,suchexplanationswill involve some kind of representation f theworlddepictedby the quasi-commonsensicalhypothesis, together with some agent or mecha-nism (Berkeley's God, Descartes's demon, or the computerthatfeeds electricalimpulsesto abrain-in-a-vat) hatgeneratesexperiencesof the sortthat we wouldhave if the representedquasi-commonsensicalworld were actual.Is there thenanyrationalbasis for preferring he analogexplanationofferedby thequasi-commonsensicalhypothesisto thevariousdigitalalternatives?Whilean extended discussion of this issue is not possible within the confines of thepresentpaper, t seems to me thatthere is. Consideragain the basic modusope-randi f adigitalexplanation:rather hansupposing thatthefeaturesof theworldthat explain experience are as close as possible to those actually reflected inexperience, such anexplanationclaims insteadthatthemainfeaturesof theworldareutterly ifferent romthose that arereflectedin experience. It seems clearthatthere is somethingratherarbitrary bout such a view, as partiallyreflected in thefact thatthere is and apparentlycould be no basis at all for preferringone suchdigitalmechanismto another.My tentativesuggestion would be that it is unrea-sonable to opt for such a further,arbitrarymechanismuntil andunless there issome specific featureof experiencethatrequires t. Theunderlyingprinciplehereis something like Ockham'sRazor: in relationto the correspondinganalog ex-



19/22

246 / LaurenceBonJourplanation,digital explanations involve an additional evel of complexity that, inaddition to being arbitrary n its specific details, is quite unnecessary.

My tentative thesis is thus that the quasi-commonsensicalhypothesis pro-vides the best explanation or the detailed contoursof our sensory experience andthat we are accordingly ustified in accepting it. I will conclude this necessarilysketchy discussion with two furthercomments about this result.First. At best this explanatoryargument ustifies attributing o physical ob-jects only those features that are clearly required o explain the characterof ourexperience. As already noted, this will arguably nclude spatial, temporal, andcausal properties.But it seems clear, for reasons alreadypartiallytouchedon, thatit will not include secondary qualities like color, but only the causal powers re-quired o produce experiences of such qualities in us, thus leading to the familiarLockean view of the status of such qualities. (This is the main reason that thehypothesis in question is only "quasi-commonsensical.")The results of modernscience seem to show that t is also possible to infer in the same basic way also toother,more "theoretical" ropertiesof such objects, i.e., such propertiesas mass,electric charge, etc. (But it is at least questionablewhether any of the things thatcan be arrivedat in this way really amountto intrinsic (as opposed to relational)propertiesof the objects in question.)

Second. Even if everything else about this argument urns out to work, it isclearly not plausible to claim that anything like it is explicitly in the minds ofordinarypeople when they make claims about the physical world (even thoughthey are arguably aware at some level of the relevant features of experience).Thus if this is the best justification available for such claims, it will follow thatmost ordinary eople areunjustified n making hem (andso do not possess "knowl-edge" of such matters).The most that can be said is thatthe essential elements forsuch an argumentare present n their conscious experience and in this way avail-able to them. This is of course to some degree a skeptical result. But it seems tome nonetheless a substantially ess troublingversion of skepticism than is to befound in views like externalism,which make any such internallyaccessible rea-

. ... * * * .son 1mposslole1npr1nc1ple.Notes

1. I say "justification" ather han "knowledge,"because I want to sidestep issues aboutwhether ustification s a requirement or knowledge and aboutthe, to my mind, rathervexed concept of knowledge itself.

2. For a discussion of coherentism and its problems in particular,along with an earliersketch of the foundationalistview presented here, see my paper "The Dialectic ofFoundationalismand Coherentism," n The Blackwell Guide to Epistemology, ed.John Greco and Ernest Sosa pp. 117-42.3. See Lewis, AnAnalysis of Knowledgeand Valuation La Salle, Ill.: Open Court, 1946);and Fumerton,Metaepistemologyand Skepticism(Lanham,Md.: Rowman & Little-field, 1995).

4. In The Structureof Empirical Knowledge (Cambridge, Mass.: HarvardUniversityPress, 1985), chapter4. (This book will hereafterbe referred o as SEK.)



20/22

Foundationalism 2475. The original source of the dilemma is WilfridSellars's "Empiricismand the Philos-ophy of Mind,"reprinted n Sellars,SciencePerception ndReality London:Rout-ledge & Kegan Paul, 1963), pp. 127-96, esp. pp. 131-2; and "The Structure of

Knowledge," in Hector-NeriCastaneda ed.), Action,Knowledge,ndReality:Crit-icalStudiesnHonor fWilfrid ellars Indianapolis:Bobbs-Merrill,1975),pp.295-347. For my own previousdevelopment andelaborationof it, see SEK, hapter4.6. I interpret his as "goingbetween the horns of the dilemma,"because I amconstruingthe horns as embodying thedevelopment andelaboration ndicated in the earlierdis-cussion and just summarized n the text, according to which the conceptualhorn in-volves aconceptualorpropositionalawareness hata stateof thespecified sortoccurs.Though the constitutiveawareness of the content of the belief of course involves theconcepts that figure in thatcontent, it is not a conceptualor propositionalawarenessthat have abelief with the content nquestion,and so, as explainedfurther n thetext,does not raiseany issue of justification. But one could instead nterpret hepresent ar-gumentas showing thattheconceptualside of theconceptual non-conceptualdichot-omy is notnecessarily ncompatiblewithfoundationalismafterall, because itneed notinvolve suchapropositional wareness;his wouldamount o"grasping ne of thehornsof thedilemma" ather hangoingbetween them.(On thislatter nterpretation,heanal-ogous possibility for sensoryexperience,discussed below, would show that the non-conceptualhornalso includes apossibility that s compatiblewithfoundationalism.)amgrateful o MatthiasSteup forhelping me to see thisalternativeway of viewing therelation of the constitutiveawarenessof contentto the Sellarsiandilemma.7. The view thatconsciousness is a "built n" or intrinsicfeatureof certain mentalstatesis in oppositionto the "higher-orderhought"account of consciousness proposed byDavid Rosenthal,according to which consciousness always derives from a higher-order,apperceptivestate. See David Rosenthal,"Two Concepts of Consciousness,"Philosophical tudies, ol. 94 (1986), pp. 329-59; reprinted n Rosenthal (ed.), TheNature f Mind New York:Oxford,University Press, 1991), pp. 462-77. I have noroomhere for a discussionof Rosenthal'sinteresting,but, I believe, ultimatelypara-doxical view; butsee mypaper"TowardaDefense of EmpiricalFoundationalism," nearlierversionof the present paper,forthcomingin a volume edited by Michael De-Paul.There are also possible objections from the direction of externalisttheories ofmental content, views which seem to me extremely implausible but which there isagainno space to go intohere;for some relevantdiscussion,see my paper"IsThoughta Symbolic Process?"Synthese,ol. 89 (1991), pp. 331-52.8. See KarlPopper,TheLogicofScientificDiscoveryNew York:Harper,1959), 25-30; Wilfrid Sellars, "Empiricismand the Philosophy of Mind," reprinted n his Sci-ence, Perception ndReality London: Routledge & Kegan Paul, 1963); DonaldDavidson, "ACoherenceTheory of TruthandKnowledge," in Dieter Henrich (ed.),KantoderHegel(Stuttgart:Klett-Cotta, 1983), pp. 423-38; andRichardRorty,Phi-losophy nd heMirror fNaturePrinceton:PrincetonUniversityPress,1979), chap-ters 3 and 4. (As will be obvious, this objection is not unrelated to the Sellarsiandilemmadiscussed above, though still differentenough in its explicit formulation owarrant eparate reatment.)9. Davidson, op.cit., p.428.

10. I am limiting my attentionhere to claims of directacquaintancewith mattersof con-crete and contingent fact. The application of the idea of direct acquaintance o nec-essary truthsandabstractentities generally raises issues that lie beyond the scope ofthe presentpaper.



21/22

248 / LaurenceBonJour11. See, e.g., Chisholm,Theoryof Knowledge, 3rd ed. (Englewood Cliffs, N.J.:PrenticeHall, 1989), pp. 18-19.12. See, e.g., PaulMoser,KnowledgeandEvidence (London:CambridgeUniversityPress,

1989). Moser's view is thatnon-conceptualcontents ustify physical objectclaims invirtue of the fact that the latterexplain the former,but he says almost nothing abouthow the explanatoryrelation in question is supposed to work.13. In sayingthis, I do not mean to be committingmyself to an ontology of sense-data.Infact, I am inclined to think that the so-called "adverbial"account of the contents ofexperience is almostcertainly correct,the main reason being that there is no accept-able accountto be given of the relationthat would have to exist betweenontologicallyindependent ense-dataand the mindthat apprehends hem that s independentof theadverbial heory: f such sense-datasomehow affect the mind, then the resulting ad-verbiallycharacterizable tates of mind are really all thatmatter,making the sense-data hemselvessuperfluous;and f they do notaffect the mind,then theirapprehensionby thatmindis difficult or impossible to makesense of. But the most importantpointfor present purposes is that any characterizationof sensory experience that can begiven in sense-datum terms can equally well be adopted by an adverbialtheorist,simply by construing a comprehensivesense-datumdescriptionof one's sensory ex-perience as characterizing he specific manner n which one is adverbially"appearedto." This is why the issue betweenthese two views, while important n otherrespects,makes no difference at all, in myjudgment, to the epistemological issues that we aremainlyconcerned with here.14. Such views are divided on the issue of whetherthe visual space in question is 2- or3-dimensional. This, however, is an issue of detail that is beyond the scope of thepresentdiscussion.15. Some philosophers have objectedstrenuouslyto the idea that,e.g., ordinary"looks"statementscan be construed as descriptionsof non-conceptualsensorycontent. See,e.g., WilfridSellars, "Empiricismandthe Philosophy of Mind," n his Science, Per-ception, and Reality (London:Routledge & Kegan Paul, 1963), 10-23. I do nothave space to enter ntothiscontroversyhere andmust contentmyself with saying thatthe objections in question seem to me to show at most that there are other senses of"looks"besides the one thatI want here (e.g., one that ndicatesa tentativeor guardedopinion about what is actually there),but have no serious tendency to show that thesense of "looks"presently at issue does not exist.16. See, e.g., Theoryof Knowledge, 3rded., p. 23.17. See H. H. Price, Perception, 2nd ed. (London:Methuen, 1950), chapter7.18. Theoryof Knowledge,3rd ed., pp. 46-54, 64-68, 71-74.19. Theoryof Knowledge,3rd ed., pp. 4-6, 72-73.20. See John Locke, An Essay Concerning Human Understanding,ed. P. H. Nidditch(Oxford:Oxford University Press, 1975), Book IV,ch. xi; andGeorgeBerkeley, ThePrinciples of HumanKnowledge, in C. M. Turbayne ed.), Berkeley,Principles, Di-alogues,andPhilosophicalCorrespondenceIndianapolis:Bobbs-Merrill,1965),28-30. While Locke and Berkeley do not claim explicitly that it is the physical-objectcharacterizations f experience thatfit together in this way, that is the most obviousand naturalway of construingtheirdiscussions.21. For a useful discussion andelaborationof Locke's argument, ee J. L. Mackie, Prob-lemsfrom Locke (Oxford:OxfordUniversity Press, 1976), chapter2. See also SEK,chapter8, for a somewhatdifferentversion of thesameunderlying dea,couched there(with some strain) n terms of a coherence theoryof justification.



22/22

Foundationalism 24922. In Perception,cited above. Anothervery useful attempt n the same direction s to befound in C. D. Broad,Scientific Thought London:Routledge, 1923), but in the spaceavailable here I will limit myself to ideas drawnfromPrice.23. Mainly in chapters4 and 8 of Perception. Since I amcondensing Price's accountveryradically, I will not attempt or the most part to give more specific references.24. See the discussion in note 14. Restating the following discussion in adverbialtermswould greatlycomplicate the formulation,while not altering the essential content.25. Thesecentralsense-dataareroughlywhat Price calls "nuclearsense-data," houghhisaccountdiffersby claimingthatatrelativelyclose range, heshapeandsize of thesense-dataareconstant.This goes along withhis view thatsensedataare ocated and iterallyface differentdirections n a 3-dimensionalspace. See Perception, pp. 218-19.26. Whatis mainly at issue here is, of course, Molyneux'sproblem. Price takes it asjust

obvious that thevery samegeometricalpropertiesareexemplified by both visual andtactual data, so that there is no need for an empiricalcorrelationbetween the proper-ties of one andthe propertiesof the other. (He also seems to thinkthat this coordina-tion can be done for the sense-data corresponding intuitively to a single object,considered by themselves;whereas I am inclined to believe, because of the problemposedby multiple objectswith approximately he samegeometricalproperties, hat twill requireconsidering the sense-datacorresponding ntuitively to a relativelylargegroupof spatially relatedobjects.)27. For reasons that are too complicated and idiosyncratic to be discussed here, Pricehimself rejectsany such inferenceand attemptsto deal with the problem in the wayalready discussed.

28. I am also ignoring here the problem of other minds, which would also have to beconsidered in a fuller account.29. See JonathanBennett'sdiscussion of "sizeblindness" n his "Substance,Reality,andPrimaryQualities,"AmericanPhilosophical Quarterly,vol. 2 (1965).

bonjour foundationalism and the external world

Documents