North American Philosophical Publications

Can Empirical Have a Foundation? Author(s): Laurence Bonjour Source: American Philosophical Quarterly, Vol. 15, No. 1 (Jan., 1978), pp. 1-13 Published by: University of Illinois Press on behalf of the North American Philosophical Publications Stable URL: http://www.jstor.org/stable/20009690 . Accessed: 10/01/2015 18:24

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].

.

University of Illinois Press and North American Philosophical Publications are collaborating with JSTOR to digitize, preserve and extend access to American Philosophical Quarterly.

http://www.jstor.org

This content downloaded from 129.118.139.129 on Sat, 10 Jan 2015 18:24:08 PM All use subject to JSTOR Terms and Conditions American Philosophical Quarterly

Volume 15, Number 1, January 1978

I. CAN EMPIRICAL KNOWLEDGE HAVE A FOUNDATION? LAURENCE BONJOUR

HPHE idea that empirical knowledge has, and decisive argument which seems to rule out all non * must a common have, afoundation has been tenet skeptical alternatives to foundationism, thereby of most major epistemologists, both past and showing that some version of foundationism must There have as we shall see present. been, further be true (on the assumption that is false). below, many importantly different variants of this In a recent statement by Quinton, this argument idea. But the common denominator among them, runs as follows: the central thesis of epistemological foundationism If beliefs are to be at . . . must as I shall understand it here, is the claim that any justified all, there be some terminal beliefs that do not owe their . . . certain empirical beliefs possess a degree of credibility to others. For a to be justified it is not epistemic justification or warrant which does not enough for it to be accepted, let alone merely enter? depend, inferentially or otherwise, on the justifi? tained: there must also be good reason for accepting cation of other beliefs, but is instead an empirical it. Furthermore, for inferential belief to be justified somehow immediate or intrinsic. It is these non the beliefs that support itmust be justified themselves. inferentially justified beliefs, the unmoved (or self There must, therefore, be a kind of belief that does movers as not owe moved) of the epistemic realm Chisholm its justification to the support provided by has called them,1 that constitute the foundation others. Unless this were so no belief would be justified at for to belief would the ante? upon which the rest of empirical knowledge is all, justify any require cedent of an infinite series of beliefs. The alleged to rest. justification terminal . . . beliefs that are needed to the In recent years, the most familiar foundationist bring of to a need not be views have been to severe and continuous regress justification stop strictly subjected self-evident in the sense that somehow attack. But this attack has been aimed they justify rarely themselves. All that is is that should not at the central foundationist thesis required they directly itself, owe their justification to any other beliefs.2 and new versions of foundationism have been quick to emerge, often propounded by the erst? I shall call this argument the epistemic regress argu? while critics themselves. Thus foundationism has ment, and the problem which generates it, the a to come become philosophical hydra, difficult to epistemic regress problem. Since it is this argument grips with and seemingly impossible to kill. The which provides the primary rationale and argu? to a purposes of this paper are, first, distinguish and mentative support for foundationism, careful clarify the main dialectical variants of foundation? examination of itwill also constitute an exploration as ism, by viewing them responses to one funda? of the foundationist position itself. The main dia? mental problem which is both the main motivation lectical variants of foundationism can best be and the primary obstacle for foundationism; and understood as differing attempts to solve the regress second, as a result of this discussion to offer problem, and the most basic objection to the schematic reasons for doubting whether any version foundationist approach is that it is doubtful that of foundationism is finally acceptable. any of these attempts can succeed. (In this paper, reason The main for the impressive durability of I shall be concerned with the epistemic regress is not foundationism any overwhelming plaus? argument and the epistemic regress problem only ibility attaching to the main foundationist thesis in as they apply to empirical knowledge. It is obvious one itself, but rather the existence of apparently that an analogous problem arises also for a priori

1 Roderick M. Chisholm, Theory of Knowledge (Englewood Cliffs, N.J., p. 30. * 1966), The an Anthony Quinton, Nature of Things (London, 1973), p. 119. This is extremely venerable argument, which has played a central role in epistemological discussion at least since Aristotle's statement of it in the Posterior Analytics, Book I, ch. 2-3. have found an (Some anticipation of the argument in the at 209E-210B, but 's worry in that passage appears to be that the definition of to an proposed knowledge is circular, not that it leads infinite regress of justification.) I

This content downloaded from 129.118.139.129 on Sat, 10 Jan 2015 18:24:08 PM All use subject to JSTOR Terms and Conditions 2 AMERICAN PHILOSOPHICAL QUARTERLY

knowledge, but there it seems likely that the argu? solely because it is now seen to be inferentially ment would take a different course. In particular, a justifiable. Second, less obviously, a person for foundationist approach might be inescapable in an whom a belief is inferentially justified need not account of a priori knowledge.) have explicitly rehearsed the justificatory argu? ment in question to others or even to himself. It is enough that the inference be available to him if the I belief is called into question by others or by himself out The epistemic regress problem arises directly (where such availability may itself be less than as of the traditional conception of knowledge fully explicit) and that the availability of the infer? adequately justified true beliefz?whether this be taken ence be, in the final analysis, his reason for holding as a fully adequate definition of knowledge or, in the belief.6 It seems clear that many beliefs which light of the apparent counter-examples discovered are quite sufficiently justified to satisfy the justifi? as a by Gettier,4 merely necessary but not sufficient cation criterion for knowledge depend for their condition. (I shall assume throughout that the justification on inferences which have not been are elements of the traditional conception at least explicitly formulated and indeed which could not necessary for knowledge.) Now the most natural be explicitly formulated without considerable re? a a way to justify belief is by producing justificatory flective effort (e.g., my current belief that this is the some argument: beliefs is justified by citing other same piece of paper upon which I was typing (perhaps conjunctive) belief B> from which A is yesterday).7 inferable in some acceptable way and which is thus Suppose then that belief A is (putatively) justified offered as a reason for accepting A.5 Call this via inference, thus raising the question of how the inferential justification. It is clear, as Quinton points justifying premise-belief B is justified. Here again out in the passage quoted above, that for A to be the answer may be in inferential terms: B may be a genuinely justified by virtue of such justificatory (putatively) justified in virtue of being inferable argument, B must itself be justified in some from some further belief C. But then the same fashion; merely being inferable from an unsup? question arises about the justification of C, and so or would confer no an ported guess hunch, e.g., on, threatening infinite and apparently genuine justification upon A. vicious regress of epistemic justification. Each belief an Two further points about inferential justification, is justified only if epistemically prior be? as understood here, must be briefly noted. First, lief is justified, and that epistemically prior belief is the belief in question need not have been arrived at justified only if a still prior belief is justified, etc., as the result of an inference in order to be infer with the apparent result that justification can a never no entially justified. This is obvious, since belief get started?and hence that there is justi? some as a no arrived at in other way (e.g., result of fication and knowledge. The foundationist claim wishful thinking) may later come to be maintained is that only through the adoption of some version

3 to as "Adequately justified" because a belief could be justified to some degree without being sufficiently justified qualify knowledge (if true). But it is far from clear just how much justification is needed for adequacy. Virtually all recent epistemo logists agree that certainty is not required. But the lottery paradox shows that adequacy cannot be understood merely in terms a see of some specified level of probability. (For useful account of the lottery paradox, Robert Ackermann, Knowledge and Belief that what is (Garden City, N.Y., 1972), pp. 39-50.) Armstrong, in Belief, and Knowledge (London, 1973), argues required is that one's reasons for the belief be "conclusive," but the precise meaning of this is less than clear. Ultimately, it may be that so to the concept of knowledge is simply too crude for refined epistemological discussion, that it may be necessary speak instead assume that of degrees of belief and corresponding degrees of justification. I shall (perhaps controversially) the proper solution reasons or to this problem will not affect the issues to be discussed here, and speak merely of the justification making the belief means. highly likely to be true, without trying to say exactly what this 4 See , "Is Justified True Belief Knowledge?" Analysis, vol. 23 (1963), pp. 121-123. Also Ackermann, op. cit., ch. V, and the corresponding references. 5 as rather more the For , I will speak of inference relations obtaining between beliefs than, accurately, between are to in a relation between two beliefs which believed. "Inference" is be understood here very broad sense; any count as which allows one, if accepted, to serve as a good reason for accepting the other will inferential. 6 reason a a It is difficult to give precise criteria for when a given reason is the for person's holding belief. G. Harman, in Thought a reason reason to he that (Princeton, 1973), argues that for a person to believe for given is for that explain why holds belief. a But this suggestion, though heuristically useful, hardly yields usable criterion. 7 Thus it is a mistake to conceive the regress as a temporal regress, as it would be if each justifying argument had to be expli? was citly given before the belief in question justified.

This content downloaded from 129.118.139.129 on Sat, 10 Jan 2015 18:24:08 PM All use subject to JSTOR Terms and Conditions CAN EMPIRICAL KNOWLEDGE HAVE A FOUNDATION? 3

can as of foundationism this skeptical consequence be opposed to how it might in principle be justified to avoided. ?would have involve the seemingly dubious Prima there seem to be four basic that an a facie, only thesis ordinary knower holds literally possibilities with regard to the eventual outcome infinite number of distinct beliefs. Thus it is not : no of this potential regress of epistemic justification surprising that important philosopher, with the with rather uncertain (i) the regress might terminate beliefs for exception of Peirce,9 seems to no of even a which justification any kind is available, have advocated such position. though they were earlier offered as justifying Alternative (iii), the view that justification ulti? premises; (ii) the regress might proceed infinitely mately moves in a closed curve, has been histori? ever more new backwards with premise beliefs cally more prominent, albeit often only as a dia? being introduced and then themselves requiring lectical foil for foundationism. At first glance, this justification; (iii) the regress might circle back alternative might seem even less attractive than upon itself, so that at some point beliefs which the second. Although the problem of the knower appeared earlier in the sequence of justifying argu? having to have an infinite number of beliefs is no ments are appealed to again as premises; (iv) the longer present, the regress itself, still infinite, now regress might terminate because beliefs are reached seems undeniably vicious. For the justification of are which justified?unlike those in alternative (i)? each of the beliefs which figure in the circle seems not now to own but whose justification does depend inferentially presuppose its epistemically prior on and not a other empirical beliefs thus does raise justification: such belief must, paradoxically, be any further issue of justification with respect to justified before it can be justified. Advocates of such beliefs.8 The foundationist opts for the last views resembling alternative (iii) have generally is that the to alternative. His argument other three tended respond to this sort of objection by a lead inexorably to the skeptical result, and that the adopting holistic conception of justification in second and third have additional fatal defects as which the justification of individual beliefs is sub? well, so that some version of the fourth, foundation? ordinated to that of the closed systems of beliefs a ist alternative must be correct (assuming that which such view implies; the property of such skepticism is false). systems usually appealed to as a basis for justification With to it seems is internal coherence. respect alternative (i), apparent Such coherence theories attempt that is correct. If this to the foundationist alternative evade the regress problem by abandoning the were correct, empirical knowledge would rest view of justification as essentially involving a linear on ultimately beliefs which were, from an epistemic order of dependence (though a non-linear view of at never standpoint least, entirely arbitrary and hence justification has been worked out in detail).10 of a incapable conferring any genuine justification. Moreover, such coherence theory of empirical What about the other two alternatives ? knowledge is subject to a number of other familiar The that alternative leads to a and argument (ii) seemingly decisive objections.11 Thus alter? outcome in skeptical has effect already been native (iii) seems unacceptable, leaving only sketched in the original formulation of the problem. alternative (iv), the foundationist alternative, as One who opted for this alternative could hope to apparently viable. avoid that the As thus skepticism only by claiming regress, formulated, the epistemic regress argu? is not vicious but seems to ment makes an though infinite, ; there be undeniably persuasive case for no a plausible way to defend such claim. Moreover, foundationism. Like any argument by elimination, a an as an defense of infinite regress view account however, it cannot be conclusive until the surviving of how is empirical knowledge actually justified? alternative has itself been carefully examined. The

8 these views could be Obviously combined, with different instances of the regress being handled in diflferent ways. I will not consider such combined views here. In would general, they simply inherit all of the objections pertaining to the simpler views. Peirce seems to a virtuous suggest regress view in "Questions concerning Certain Faculties Claimed for Man," Collected But view Papers V, pp. 135-155. the is presented metaphorically and it is hard to be sure exactly what it comes to or to what extent it bears on the present issue. 10 The statement of the non-linear view was original by Bernard Bosanquet in Implication and Linear Inference (London, 1920). For more recent see Gilbert discussions, Harman, Thought (Princeton, 1973); and Nicholas Rescher, ", Co herentism, and the Idea of Cognitive The Journal vol. 11 Systematization," of Philosophy, 71 (1974), pp. 695-708. I have to show how a coherence attempted view might be defended against the most standard of these objections in "The Coherence of Theory Empirical Knowledge," Philosophical Studies, vol. 30 (1976), pp. 281-312.

This content downloaded from 129.118.139.129 on Sat, 10 Jan 2015 18:24:08 PM All use subject to JSTOR Terms and Conditions 4 AMERICAN PHILOSOPHICAL QUARTERLY

foundationist turn out to be status or for position may subject epistemic thinking that in fact they to thus a re equally serious objections, forcing do. There might even be some sort of indirect a examination of the other alternatives, search for argument to show that such a status is a con? a or further non-skeptical alternative, conceivably sequence of the sorts of epistemic properties which con? are the reluctant acceptance of the skeptical directly required to solve the regress problem. is not on an clusion.12 In particular, it clear the basis But until such argument is given (and it is of the argument thus far whether and how founda? doubtful that it can be), the question of whether can tionism itself solve the regress problem; and basic beliefs are or can be certain, infallible, etc., will remain a thus the possibility exists that the epistemic regress relatively unimportant side-issue. argument will prove to be a two-edged sword, as Indeed, many recent foundationists have felt to as it is to even lethal the foundationist his opponents. that the relatively modest version of strong foundationism outlined above is still too strong. II Their alternative, still within the general aegis of The most straightforward interpretation of alter? the foundationist position, is a view which may be native (iv) leads directly to a view which I will here called weak foundationism. Weak foundationism call strong foundationism. According to strong accepts the central idea of foundationism?viz. foundationism, the foundational beliefs which that certain empirical beliefs possess a degree of terminate the regress of justification possess sufficient independent epistemic justification or warrant epistemic warrant, independently of any appeal which does not derive from inference or coherence to inference from (or coherence with) other em? relations. But the weak foundationist holds that pirical beliefs, to satisfy the justification condition these foundational beliefs have only a quite low of knowledge and qualify as acceptable justifying degree of warrant, much lower than that attributed premises for further beliefs. Since the justification to them by even modest strong foundationism and of these basic beliefs, as they have come to be called, insufficient by itself to satisfy the justification con? is thus allegedly not dependent on that of any dition for knowledge or to qualify them as accept? other empirical belief, they are uniquely able to able justifying premises for other beliefs. Thus this secure warrant must provide starting-points for the justification independent somehow be augmented of empirical knowledge and stopping-points for the if knowledge is to be achieved, and the usual appeal regress of justification. here is to coherence with other such minimally The position just outlined is in fact a fairly warranted beliefs. By combining such beliefs into modest version of strong foundationism. Strong larger and larger coherent systems, it is held, their foundationists have typically made considerably initial, minimal degree of warrant can gradually be stronger claims on behalf of basic beliefs. Basic enhanced until knowledge is finally achieved. Thus beliefs have been claimed not only to have suffi? weak foundationism, like the pure coherence cient non-inferential justification to qualify as theories mentioned above, abandons the linear knowledge, but also to be certain, infallible, indubit? conception of justification.14 able, or incorrigible (terms which are usually not Weak foundationism thus represents a kind of very carefully distinguished).13 And most of the hybrid between strong foundationism and the major attacks on foundationism have focused on coherence views discussed earlier, and it is often these stronger claims. Thus it is important to point thought to embody the virtues of both and the out that nothing about the basic strong founda? vices of neither. Whether or not this is so in other to tionist response to the regress problem demands respects, however, relative the regress problem that basic beliefs be more than adequately justified. weak foundationism is finally open to the very There might of course be other reasons for re? same basic objection as strong foundationism, with quiring that basic beliefs have some more exalted essentially the same options available for meeting

12 a The presumption against a skeptical outcome is strong, but I think it is mistake to treat it as absolute. If no non-skeptical theory can be found which is at least reasonably plausible in its own right, skepticism might become the only rational alternative. 13 For some useful distinctions among these terms, see , "Varieties of Privileged Access," American Philosophical Quarterly, vol. 8 (1971), pp. 223-241. 14 For discussions of weak foundationism, see , Human Knowledge (New York, 1949), part II, ch. 11, and part V, chs. 6 and 7 ;, "Sense and Certainty," Philosophical Review, vol. 61 (1952), pp. 160-167 ; Israel Schemer, Science and Subjectivity (New York, 1967), chapter V; and Roderick Firth, "Coherence, Certainty, and Epistemic Priority," The Journal of Philosophy, vol. 61 (1964), pp. 545-557.

This content downloaded from 129.118.139.129 on Sat, 10 Jan 2015 18:24:08 PM All use subject to JSTOR Terms and Conditions CAN EMPIRICAL KNOWLEDGE HAVE A FOUNDATION? 5

etc. none it. As we shall see, the key problem for any version dogmas; But of these other varieties of of foundationism is whether it can itself solve the justification can satisfy the justification condition regress problem which motivates its very existence, for knowledge. Knowledge requires epistemic justifi? without resorting to essentially ad hoc stipulation. cation, and the distinguishing characteristic of this The distinction between the two main ways of particular species of justification is, I submit, its meeting this challenge both cuts across and ismore essential or internal relationship to the cognitive basic than that between strong and weak founda? goal of truth. Cognitive doings are epistemically tionism. This being so, it will suffice to concen? justified, on this conception, only if and to the trate here on strong foundationism, leaving the extent that they are aimed at this goal?which means application of the discussion to weak foundation? roughly that one accepts all and only beliefs ism largely implicit. which one has good reason to think are true.15 To The fundamental concept of strong foundation? accept a belief in the absence of such a reason, a or even ism is obviously the concept of basic belief. It is by however appealing mandatory such accept? an appeal to this concept that the threat of infinite ance might be from other standpoints, is to neglect regress is to be avoided and empirical knowledge the pursuit of truth ; such acceptance is, one might can given a secure foundation. But how there be say, epistemically irresponsible. My contention is that any empirical beliefs which are thus basic? In the idea of being epistemically responsible is the fact, though this has not always been noticed, the core of the concept of epistemic justification.16 very idea of an epistemically basic empirical belief A corollary of this conception of epistemic is extremely paradoxical. For on what basis is justification is that a satisfactory defense of a parti? such a belief to be justified, once appeal to further cular standard of epistemic justification must con? empirical beliefs is ruled out? Chisholm's theo? sist in showing it to be truth-conducive, i.e. in logical analogy, cited earlier, is most appropriate: showing that accepting beliefs in accordance with a basic belief is in effect an epistemological un? its dictates is likely to lead to truth (and more moved (or self-moved) mover. It is able to confer likely than any proposed alternative). Without justification on other beliefs, but apparently has such a meta-justification, a proposed standard of no need to have justification conferred on it. But is epistemic justification lacks any underlying ration? a status to an such any easier understand in episte ale. Why after all should epistemically respon? mology than it is in theology? How can a belief sible inquirer prefer justified beliefs to unjustified impart epistemic "motion" to other beliefs unless ones, if not that the former are more likely to be it is itself in "motion"? And, even more para? true ?To insist that a certain belief is epistemically doxically, how can a belief epistemically "move" justified, while confessing in the same breath that itself? this fact about it provides no good reason to think This intuitive difficulty with the concept of a that it is true, would be to render nugatory the basic empirical belief may be elaborated and whole concept of epistemic justification. clarified by reflecting a bit on the concept of epis? These general remarks about epistemic justifi? a temic justification. The idea of justification is cation apply in full measure to any strong founda? to generic one, admitting in principle of many specific tionist position and its constituent account of Thus the of an a varieties. acceptance empirical basic beliefs. If basic beliefs are to provide secure belief might be morally justified, i.e. justified as foundation for empirical knowledge, if inference morally obligatory by reference to moral principles from them is to be the sole basis for the justification and standards; or pragmatically justified, i.e. of other empirical beliefs, then that feature, what? justified by reference to the desirable practical ever it may be, in virtue of which a belief qualifies as must a reason consequences which will result from such accept? basic also constitute good for or ance; religiously justified, i.e. justified by refer? thinking that the belief is true. If we let '

15 How good a reason must one have ? Presumably some justification accrues from any reason which makes the belief even more to minimally likely be true than not, but considerably more than this would be required to make the justification ade? quate for knowledge. (See note 3, above.) (The James-Clifford controversy concerning the "will to believe" is also relevant here. I am agreeing with Clifford to the extent of saying that epistemic justification requires some positive reason in favor of the belief and not just the absence of any reason against.) 16 a use For similar of the notion of epistemic irresponsibility, see , "How Do You Know?" American Philosophical Quarterly, vol. 11 (1974), p. 117.

This content downloaded from 129.118.139.129 on Sat, 10 Jan 2015 18:24:08 PM All use subject to JSTOR Terms and Conditions 6 AMERICAN PHILOSOPHICAL QUARTERLY

basic in an acceptable foundationist account, the basic beliefs at least, it is sufficient that the premises premises of the following justificatory argument for an argument of that general sort (or for some must themselves be at least justified:17 favored particular variety of such argument) or merely be true, whether not that person (or (i) Belief B has feature . anyone else) justifiably that they are true. (ii) Beliefs having feature are highly likely to Second, he might grant that it is necessary both be true. that such justification exist and that the person for whom the belief is basic be in Therefore, B is highly likely to be true. cognitive possession of it, but insist that his cognitive grasp of the Notice further that while either taken premise premises required for that justification does not turn out to be on an separately might justifiable involve further empirical beliefs which would then a basis on the choice of priori (depending particular require justification, but instead involves cognitive it seems clear that could not both be thus a more ), they states of rudimentary sort which do not For B is ex an justifiable. hypothesi empirical belief, themselves require justification: intuitions or im? and it is hard to see how a particular empirical mediate apprehensions. I will consider each of these on a a belief could be justified purely priori basis.18 alternatives in turn. And ifwe now assume, reasonably enough, that for a a B to be justified for particular person (at parti? Ill cular it is necessary, not merely that a justi? time) The who has come the closest to an fication for B exist in the abstract, but that the philosopher explicit advocacy of the view that basic beliefs may person in question be in cognitive possession of that be justified even though the person for whom they justification, we get the result that B is not basic are basic is not in any way in cognitive possession after all since its justification depends on that of at of the is D. M. least one other belief. If this is correct, appropriate justifying argument empirical In his recent Truth and foundationism is untenable as a solution to Armstrong. book, Belief, strong a an Knowledge,19 Armstrong presents version of the the regress problem (and analogous argument regress problem (though one couched in will show weak foundationism to be similarly un? epistemic . terms of knowledge rather than justification) and tenable) defends what he calls an "Externalist" solution: The foregoing argument is, no doubt, exceedingly to accounts obvious. But how is the strong foundationist According to 'Externalist' of non-inferential seem to two what makes a true non-inferential belief a answer it? Prima facie, there be only knowledge, sorts of answer which are even case of knowledge is some natural relation which holds general remotely between the belief-state . . . and the situation which so long as the strong foundationist plausible, true. It a matter a remains within the confines of the traditional makes the belief is of certain relation holding between the believer and the world. of knowledge, avoids tacitly embracing conception [157]. skepticism, and does not attempt the heroic task of that an belief could be own candidate for this "natural re? arguing empirical justified Armstrong's' must be a connection on a purely a priori basis. First, he might argue that lation' is "that there law-like a between the state of affairs a's although it is indeed necessary for belief to be Bap [i.e. believing a that and the state of affairs that makes true justified and afortiori for it to be basic that justi? p] '/>' it must be the case that fying argument of the sort schematized above be in such that, given Bap, A similar view seems to be in principle available in the situation, it is not always p." [166] implicit in See? necessary that the person for whom the belief is Dretske's account of perceptual knowledge or even and with the variation that Dretske basic (or anyone else) know justifiably ing Knowing, believe that it is available; instead, in the case of requires for knowledge not only that the relation in

17 true as in order to avoid But I shall In fact, the premises would probably have to be well, Gettier-type counterexamples. ignore this refinement here. 18 of be the case that On a Carnap-style a priori theory of probability it could, course, very general empirical propositions in which are true outnumber those in which were more likely to be true than not, i.e. that the possible state-descriptions they not to be true in a sense which would allow the detached assertion of the they are false. But clearly this would make them likely this fact seems to such from for in question (on pain of contradiction), and preclude justification being adequate knowledge. 19 in this section are to this book. Armstrong, op. cit., chapters 11?13. Bracketed page references

This content downloaded from 129.118.139.129 on Sat, 10 Jan 2015 18:24:08 PM All use subject to JSTOR Terms and Conditions CAN EMPIRICAL KNOWLEDGE HAVE A FOUNDATION? 7

question obtain, but also that the putative knower further empirical beliefs in need of justification and believe that it obtains?though not that this belief no regress. seems be justified.20 In addition, it likely that Now it is clear that such an externalist position of an various views ordinary-language stripe which succeeds in avoiding the regress problem and the appeal to facts about how language is learned anti-foundationist argument. What may well be either to justify basic belief or to support the claim doubted, however, is whether this avoidance that no justification is required would, if pushed, deserves to be considered a solution, rather than an turn out to of this sort. I be positions general Here essentially ad hoc evasion, of the problem. Plainly to is shall mainly confine myself Armstrong, who the sort of "external" relation which Armstrong one the only of these philosophers who is explicitly has in mind would, if known, provide a basis for a concerned with the regress problem. justifying argument along the lines sketched earlier, some as to There is, however, uncertainty how roughly as follows: views of this sort in general and Armstrong's view (i) Belief Z? is an instance of kind K. in particular are properly to be interpreted. On the (ii) Beliefs of kind K are connected in a law-like one hand, Armstrong might be taken as offering an way with the sorts of states of affairs which account of how basic beliefs (and perhaps others would make them true, and therefore are as well) satisfy the adequate-justification condition highly likely to be true. for knowledge; while on the other hand, he might be taken as simply repudiating the traditional con? Therefore, B is highly likely to be true. ception of knowledge and the associated concept of epistemic justification, and offering a surrogate But precisely what generates the regress problem in conception in its place?one which better accords the first place is the requirement that for a belief with the "naturalistic" world-view which Arm? B to be epistemically justified for a given person P, strong prefers.21 But it is only when understood in it is necessary, not just that there be justifiable or the former way that externalism (to adopt Arm? even true premises available in the situation which strong's useful term) is of any immediate interest could in principle provide a basis for a justification here, since it is only on that interpretation that it of B, but that P himself know or at least justifiably constitutes a version of foundationism and offers a believe some such set of premises and thus be in a to the anti-foundationist to the direct response argument position employ corresponding argument. on The externalist seems to amount set out above. Thus I shall mainly focus this position merely to this interpretation of externalism, remarking only waiving general requirement in cases where the a briefly at the end of the present section on the justification takes certain form, and the question alternative one. iswhy this should be acceptable in these cases when Understood in this way, the externalist solution it is not acceptable generally. (If it were acceptable seem to the regress problem is quite simple: the person generally, then it would that any true belief who has a basic belief need not be in possession of would be justified for any person, and the dis? any justified reason for his belief and indeed, tinction between knowledge and true belief would in not even a move seems except Dretske's version, need think collapse.) Such rather analogous to that there is such a reason ; the status of his belief solving a regress of causes by simply stipulating that as most events must a constituting knowledge (if true) depends solely although have cause, events of on the external relation and not at all on his a certain kind need not. subjective view of the situation. Thus there are no Whatever plausibility attaches to externalism

20 Fred I. Dretske, Seeing and Knowing (London, 1969), chapter III, especially pp. 126-139. It is difficult to be quite sure of Dretske's view, however, since he is not concerned in this book to offer a general account of knowledge. Views which are in some ways similar to those of Armstrong and Dretske have been offered by Goldman and by Unger. See , "A Causal Theory of Knowing," The Journal of Philosophy, vol. 64 (1967), pp. 357-372; and Peter Unger, "An Analysis of Factual Knowledge," The Journal of Philosophy, vol. 65 (1968), pp. 157-170. But both Goldman and Unger are explicitly concerned with the and not at all with the regress problem, so it is hard to be sure how their views relate to the sort of externalist view which is at issue here. 21 one a On the hand, Armstrong seems to argue that it is not requirement for knowledge that the believer have "sufficient a evidence" for his belief, which sounds like rejection of the adequate-justification condition. On the other hand, he seems to want to a a say that the presence of the external relation makes it rational for person to accept belief, and he seems (though is not to have in no this clear) epistemic rationality mind; and there appears to be substantial difference between saying that a belief is epistemically rational and saying that it is epistemically justified.

This content downloaded from 129.118.139.129 on Sat, 10 Jan 2015 18:24:08 PM All use subject to JSTOR Terms and Conditions 8 AMERICAN PHILOSOPHICAL QUARTERLY

seems to derive from the fact that if the external surely preclude its being justified, and it is hard to relation in question genuinely obtains, then P will see why the result should be any different for an not go wrong in accepting the belief, and it is, in a allegedly basic belief. sense, not an accident that this is so. But it remains Thus it finally seems possible to make sense of unclear how these facts are supposed to justify P's externalism only by construing the externalist as an acceptance of B. It is clear, of course, that simply abandoning the traditional notion of epis? that P external observer who knew both accepted temic justification and along with it anything B and that there was a law-like connection between resembling the traditional conception of know? a such acceptance and the truth of B would be in ledge. (As already remarked, this may be precisely an position to construct argument to justify his own what the proponents of externalism intend to be P serve as a are acceptance of B. could thus useful doing, though most of them not very clear on epistemic instrument, a kind of cognitive ther? this point.) Thus consider Armstrong's final sum? mometer, for such an external observer (and in fact mation of his conception of knowledge : the of a thermometer is exactly the example Knowledge of the truth of particular matters offact is a which to illustrate the analogy Armstrong employs belief which must be true, where the 'must' is a matter which is to obtain between relationship supposed of law-like necessity. Such knowledge is a reliable who has the belief and the external or the person representation 'mapping' of reality. [220]. state of affairs [i66ff.]). But P himself has no Nothing is said here of reasons or justification or reason at all for thinking that B is likely to be true. evidence or having the right to be sure. Indeed From his perspective, it is an accident that the the whole idea, central to the western belief is true.22 And thus his acceptance of B is no epistemo of as the or an logical tradition, knowledge essentially more rational responsible from epistemic product of reflective, critical, and rational inquiry than would be the acceptance of a standpoint has vanished without a trace. It is similar belief for which the external seemingly subjectively of course that such an altered of relation in failed to obtain.23 possible conception question be or even in some Nor does it seem to matters to move from knowledge may inescapable way help but it constitutes a solution to the which desirable, regress Armstrong's version of externalism, requires problem or any problem arising out of the tradi? only that the requisite relationship between the tional conception of knowledge only in the radical believer and the world obtain, to the superficially and relatively uninteresting sense that to reject less radical version held Dretske, apparently by that is also to the which that P also believe that the ex? conception reject problems requires out of it. In this I shall confine not arising paper, myself ternal relation obtains, but does require that to less radical solutions. this latter belief be justified. This view may seem slightly less implausible, since it at least requires that the person have some idea, albeit unjustified, IV of why B is likely to be true. But this change is not to save to see this is The externalist solution discussed enough externalism. One way just represents to suppose that the person believes the requisite a very recent approach to the justification of basic relation to obtain on some totally irrational and beliefs. The second view to be considered is, in a tea so that it to irrelevant basis, e.g. as result of reading contrast, venerable deserves be called to leaves or studying astrological charts. If B were an the standard foundationist solution the problem course to ordinary, non-basic belief, such a situation would in question. I refer of the traditional

22 a true or in contrast it is an accident that it One way to put this point is to say that whether belief is likely to be whether is be true of one and the same belief that it is "a belief true depends significantly on how the belief is described. Thus it might on connected in a law-like way with the state of affairs which it describes,, and also that it is "a belief adopted the basis of no on and to be true on the second. The claim apparent evidence"; and it might be likely to be true the first description unlikely here is that it is the believer's own conception which should be considered in deciding whether the belief is justified. (Some? a to be in terms of that thing analogous seems to be true in ethics : the moral worth of person's action is correctly judged only not to result from person's subjective conception of what he is doing and in light of what happens, willy-nilly, it.) 23 to some dis? Notice, however, that if beliefs standing in the proper external relation should happen possess subjectively believer were to notice tinctive feature (such as being spontaneous and highly compelling to the believer), and if the empirically, a to construct a that beliefs having this feature were true a high proportion of the time, he would then be in position justifica? course a tion for a new belief of that sort along the lines sketched at the end of section II. But of belief justified in that way would no longer be basic.

This content downloaded from 129.118.139.129 on Sat, 10 Jan 2015 18:24:08 PM All use subject to JSTOR Terms and Conditions CAN EMPIRICAL KNOWLEDGE HAVE A FOUNDATION? 9

doctrine of cognitive givenness, which has played a they might be thought to be entailed by the only central role in epistemological discussions at least very obvious intuitive picture of how the view is since Descartes. In recent years, however, the supposed to work. In the second place, givenism as concept of the given, like foundationism itself, has understood here does not involve the usual stipu? come under serious attack. One upshot of the lation that only one's private mental and sensory a states can or resulting discussion has been realization that be given. There may may not be other re? reasons there are many different notions of givenness, for thinking that this is in fact the case, but lated to each other in complicated ways, which such a restriction is not part of the position itself. almost certainly do not stand or fall together. Thus Thus both positions like that of C. I. Lewis, for it will be well to begin by formulating the precise whom the given is restricted to private states appre? notion of givenness which is relevant in the present hended with certainty, and positions like that of context and distinguishing it from some related Quinton, for whom ordinary physical states of affairs are with no or conceptions. given claim of certainty In the context of the epistemic regress problem, incorrigibility being involved, will count as versions givenness amounts to the idea that basic beliefs are of givenism. justified by reference, not to further beliefs, but As already noted, the idea of givenness has been rather to states of affairs in the world which are roundly criticized in recent philosophical discussion as "immediately apprehended" or "directly pre? and widely dismissed a piece of philosophical sented" or "intuited." This justification by refer? mythology. But much at least of this criticism has ence to non-cognitive states of affairs thus allegedly to do with the claim of certainty on behalf of the or avoids the need for any further justification and given with the restriction to private, subjective thereby stops the regress. In a way, the basic gambit states. And some of it at least has been mainly of givenism (as I shall call positions of this sort) concerned with issues in the philosophy of mind thus resembles that of the externalist positions which are only distantly related to our present considered above. In both cases the justificatory epistemological concerns. Thus even if the objec? are appeal to further beliefs which generates the re? tions offered cogent against other and stronger an versions of it gress problem is avoided for basic beliefs by givenness, remains unclear whether appeal directly to the non-cognitive world; the and how they apply to the more modest version at crucial difference is that for the givenist, unlike the issue here. The possibility suggests itself that modest state not be a even more am? externalist, the justifying of affairs in the givenness may myth, if some a world is allegedly apprehended in way by the bitious varieties are, result which would give believer. the epistemological foundationist all he really even The givenist position to be considered here is needs, though he has usually, in a spirit of significantly weaker than more familiar versions of philosophical greed, sought considerably more. In the doctrine of givenness in at least two different what follows, however, I shall sketch a line of respects. In the first place, the present version does argument which, if correct, will show that even an not claim that the given (or, better, the appre? modest givenism is untenable position.24 hension thereof) is certain or even incorrigible. As The argument to be developed depends on a discussed above, these stronger claims are in? problem within the givenist position which is sur? essential to the strong foundationist solution to the prisingly easy to overlook. I shall therefore proceed regress problem. If they have any importance at all in the following way. I shall first state the problem in this context it is only because, as we shall see, in an initial way, then illustrate it by showing how

24 I suspect that something like the argument to be given here is lurking somewhere in Sellars' " and the Philo? sophy of Mind" (reprinted in Sellars, Science, , and Reality [London, 1963], pp. 127-196), but it is difficult to be sure. A more recent argument by Sellars which is considerably closer on the surface to the argument offered here is contained in "The Structure of Knowledge," his Machette Foundation Lectures given at the University of Texas in 1971, in Hector-Nerl Casteneda (ed.), Action, Knowledge, and Reality: Critical Studies inHonor of (Indianapolis, 1975), Lecture III, sections III-IV. A similar line of argument was also offered by Neurath and Hempel. See Otto Neurath, "Protocol Sentences," tr. in A. J. Ayer (ed.), Logical (New York, 1959), pp. 199-208; and Carl G. Hempel, "On the Logical Positivists' Theory of Truth," Analysis, vol. 2 (1934-5), pp. 49-59- The Hempel paper is in part a reply to a foundationist critique of Neurath by Schlick in "The Foundation of Knowledge," also translated in Logical Positivism, op. cit., pp. 209-227. Schlick replied to Hempel in "Facts and Propositions," and Hempel responded in "Some Remarks on Tacts' and Propositions," both in Analysis, vol. 2 (1934-5), PP* ^5~7? and 93-96, respectively. Though the Neurath-Hempel argument conflates issues having to do with truth a and issues having to do with justification in confused and confusing way, it does bring out the basic objection to givenism.

This content downloaded from 129.118.139.129 on Sat, 10 Jan 2015 18:24:08 PM All use subject to JSTOR Terms and Conditions 10 AMERICAN PHILOSOPHICAL QUARTERLY

one recent it arises in version of givenism, and reminded here of Chisholm's claim that certain finally consider whether any plausible solution is cognitive states justify themselves,25 but that possible. (It will be useful for the purposes of this extremely paradoxical remark hardly constitutes an discussion to make two simplifying assumptions, explanation of how this is possible. would more com? on an without which the argument be If, the other hand, intuition is not a cog? not state plicated, but essentially altered. First, I shall nitive and thus involves no cognitive grasp of assume that the basic belief which is to be justified the state of affairs in question, then the need for a by reference to the given or immediately appre? justification for the intuition is obviated, but at the hended state of affairs is just the belief that this serious cost of making it difficult to see how the same state of affairs obtains. Second, I shall assume intuition is supposed to justify the belief. If the that the given or immediately apprehended state person in question has no cognitive grasp of that of affairs is not itself a belief or other state of affairs of cognitive (or any other) by virtue of having such an then how state.) intuition, does the intuition give Consider then an allegedly basic belief that-/> him a reason for thinking that his belief is true or which is supposed to be justified by reference to a likely to be true ?We seem again to be back to an or given immediately apprehended state of affairs externalist position, which it was the whole point that-?. Clearly what justifies the belief is not the of the category of intuition or givenness to avoid. an state of affairs simpliciter, for to say that would be As illustration of this problem, consider Quin? to to a as return form of externalism. For the givenist, tan's version of givenism, outlined in his book what justifies the belief is the immediate apprehension The Nature of Things.2* As noted above, basic or intuition of the state of affairs. Thus we seem to beliefs may, according to Quinton, concern ordin? have three items present in the situation : the belief, ary perceptible states of affairs and need not be or the state of affairs which is the object of the belief, certain incorrigible. (Quinton uses the phrase as and the intuition or immediate apprehension of "intuitive belief" I have been using "basic that state of affairs. The problem to be raised belief" and calls the linguistic expression of an revolves around the nature of the last of these intuitive belief a "basic statement"; he also seems items, the intuition or immediate apprehension to pay very little attention to the difference between (hereafter I will use mainly the former term). It beliefs and statements, shifting freely back and seems to a a forth between I be cognitive state, perhaps somehow of them, and will generally follow more rudimentary sort than a belief, which in? him in this.) Thus "this book is red" might, in an a volves the thesis or assertion that-/>. Now if this is appropriate context, be basic statement express? to a a or correct, it is easy enough understand in rough ing basic intuitive belief. But how are such sort of way how an intuition can serve to justify a basic statements (or the correlative beliefs) sup? belief with this same assertive content. The problem posed to be justified? Here Quinton's account, is to understand why the intuition, involving as it beyond the insistence that they are not justified by does the cognitive thesis that-?, does not itself reference to further beliefs, is seriously unclear. He answer rather that the is require justification. And if the is offered says vaguely person "aware" that the intuition is justified by reference to the [129] or "directly aware" [139] of the appropriate or state of affairs that-^, then the question will be why state of affairs, that he has "direct knowledge" this would not require a second intuition or other [126] of it, but he gives no real account of the or apprehension of the state of affairs to justify the nature epistemological status of this state of one. For one and same or "direct original otherwise the "direct awareness" knowledge," though cognitive state must somehow constitute both an it seems clear that it is supposed to be a cognitive apprehension of the state of affairs and a justifi? state of some kind. (In particular, it is not clear cation of that very apprehension, thus pulling what "direct" means, over and above "non own itself up by its cognitive bootstraps. One is inferential.")27

25 Chisholm, "Theory of Knowledge," in Ghisholm et al., Philosophy (Englewood Cliffs, N.J., 1964), pp. 270fr. 26 Op. cit. Bracketed page references in this section will be to this book. 27 Quinton does offer one small bit of clarification here, by appealing to the notion of ostensive definition and claiming in effect that the sort of awareness involved in the intuitive justification of a basic belief is the same as that involved in a situation as of ostensive definition. But such a comparison is of little help, for at least two reasons. First, Wittgenstein, Sellars, and others have argued, the notion of ostensive definition is itself seriously problematic. Indeed, an objection quite analogous to the present one against the notion of a basic belief could be raised against the notion of an ostensive definition; and this objection, if

This content downloaded from 129.118.139.129 on Sat, 10 Jan 2015 18:24:08 PM All use subject to JSTOR Terms and Conditions CAN EMPIRICAL KNOWLEDGE HAVE A FOUNDATION? II

The difficulty with Quintan's account comes out by claiming that an intuition is a semi-cognitive or most clearly in his discussion of its relation to the quasi-cognitive state,29 which resembles a belief in : correspondence theory of truth its capacity to confer justification, while differing from a belief in not itself. In The theory of basic statements is closely connected requiring justification some such seems to be in with the correspondence theory of truth. In its classical fact, conception implicit most if not all But when stated form that theory holds that to each true statement, givenist positions. whatever its form may be, a fact of the same form thus baldly, this "solution" to the problem seems corresponds. The theory of basic statements indicates hopelessly contrived and ad hoc. If such a move is the point at which correspondence is established, at acceptable, one is inclined to expostulate, then which the of beliefs makes its contact once sort system justifying again any of regress could be solved in with the world. [139]. similar fashion. Simply postulate a final term in the And further on he remarks that the truth of basic regress which is sufficiently similar to the previous statements "is directly determined by their cor? terms to satisfy, with respect to the penultimate sort or respondence with fact" [143]. (It is clear that term, the of need impetus which originally means "determined" here "epistemically deter? generated the regress; but which is different a from terms so as not to mined.") Now it is familiar but still forceful enough previous itself a idealist objection to the correspondence theory of require satisfaction by further term. Thus we truth that if the theory were correct we could never would have semi-events, which could cause but our were not know whether any of beliefs true, since need be caused; semi-explanatia, which could we no our need not have perspective outside system of be? explain but be explained; and semi liefs from which to see that they do or do not beliefs, which could justify but need not be justified. seems to is not such a move in? correspond. Quinton, however, suppose The point that is always rather blithely that intuition or direct awareness correct (though I suspect that it is), but simply that a nature a provides just such perspective, from which we the and possibility of such convenient at to can in some cases apprehend both beliefs and regress-stopper needs the very least be clearly world and judge whether or not they correspond. and convincingly established and explained before can a to And he further supposes that the issue of justification it constitute satisfactory solution any not somehow does arise for apprehensions made regress problem. from this perspective, though without giving any The main account which has usually been offered or so. account of how why this is by givenists of such semi-cognitive states is well no account can terms or My suggestion here is that such suggested by the in which immediate are be given. As indicated above, the givenist is caught intuitive apprehensions described: "immedi? in a fundamental dilemma: if his intuitions or ate," "direct," "presentation," etc. The under? are as : immediate apprehensions construed cog? lying idea here is that of confrontation in intuition, or nitive, then they will be both capable of giving mind consciousness is directly confronted with sort justification and in need of it themselves; if they its object, without the intervention of any of are not non-cognitive, then they do need justifi? intermediary. It is in this sense that the object is are to root cation but also apparently incapable of pro? given the mind. The metaphor underlying at this whole is mind or consciousness viding it. This, bottom, is why epistemological picture vision: a an givenness is myth.28 is likened to immaterial eye, and the object of Once the problem is clearly realized, the only intuitive awareness is that which is directly before seems to to meta possible solution be to split the difference the mental eye and open its gaze. If this

answerable at all, could only be answered by construing the awareness involved in ostensi?n in such a way as to be of no help to the foundationist in the present discussion. Second, more straightforwardly, even if the notion of ostensive definition were entirely unobjectionable, there is no need for the sort of awareness involved to be justified. If all that is at issue is learning the meaning of a word (or acquiring a concept), then justification is irrelevant. Thus the existence of ostensive definitions would not show how there could be basic beliefs. 28 Notice, however, that to reject an epistemological given does not necessarily rule out other varieties of givenness which may have importance for other philosophical issues. In particular, there may still be viable versions of givenness which pose an obstacle to materialist views in the philosophy of mind. For useful distinctions among various versions of givenness and a discussion of their relevance to the philosophy of mind, see James W. Cornman, "Materialism and Some Myths about Some Givens," The Monist, vol. 56 (1972), pp. 215-233. 29 Compare the Husserlian notion of a "pre-predicative awareness."

This content downloaded from 129.118.139.129 on Sat, 10 Jan 2015 18:24:08 PM All use subject to JSTOR Terms and Conditions 12 AMERICAN PHILOSOPHICAL QUARTERLY

to on phor were to be taken seriously, it would become which both enables it confer justification relatively simple to explain how there can be a other states and also requires that it be justified no cognitive state which can justify but does not itself. If this is right, then it does good to intro? an require justification. (If the metaphor is to be taken duce semi-cognitive states in attempt to justify seriously enough to do the foundationist any real basic beliefs, since to whatever extent such a state good, it becomes plausible to hold that the in? is capable of conferring justification, it will to that same even tuitive cognitive states which result would after all very extent require justification. Thus have to be infallible. For if all need for justification if such states do exist, they are of no help to the answer is to be precluded, the envisaged relation of con? givenist in attempting to the objection at must as too issue here.30 frontation seemingly be conceived to intimate to allow any possibility of error. To the Hence the givenist response the anti-founda? extent that this is so, the various arguments which tionist argument seems to fail. There seems to be no to a have been offered against the notion of infallible way explain how basic cognitive state, states count also this version of whether called a belief or an can be congitive against intuition, givenism.) directly justified by the world without lapsing back Unfortunately, however, it seems clear that the into externalism?and from there into skepticism. comments mental eye metaphor will not stand serious I shall conclude with three further at sorts scrutiny. The mind, whatever else itmay be, is not aimed warding off certain likely of mis? so as we an First. It natural in this connection an eye or, far know, anything like eye. understanding. is to to to Ultimately the metaphor is just far too simple to be attempt justify basic beliefs by appealing a even minimally adequate to the complexity of experience. But there is familiar ambiguity in the term in over mental phenomena and to the variety of conditions "experience," which fact glosses the upon which such phenomena depend. This is not crucial distinction upon which the foregoing argu? ment rests. mean an to deny that there is considerable intuitive appeal Thus "experience" may either a or to the confrontational model, especially as applied experiencing (i.e., cognitive state) something an of And once to perceptual consciousness, but only to insist that experienced (i.e., object cognition). is the of this appeal is far too vague in its import to ade? this ambiguity resolved, concept experi? sorts ence seems to be of no to the quately support the very specific of episte? particular help Second. I have for the mological results which the strong foundationist givenist. concentrated, of on version of needs. In particular, even if empirical knowledge at sake simplicity, Quintan's givenism some sort or in which states of affairs are some point involves of confrontation ordinary physical which are But the of seeming confrontation, this by itself provides no among the things given. logic the same if it clear reason for attributing epistemic justification the argument would be essentially were to a more traditional version like or reliability, let alone certainty, to the cognitive applied which Lewis's in which it is which are states, whatever they may be called, private experiences I cannot see that the end result would result. given, and vicissitudes of be it be harder to Moreover, quite apart from the different?though might discern, are in? in cases where the basic belief the mental eye metaphor, there powerful especially allegedly to is a belief about another state. Third. dependent reasons for thinking that the attempt cognitive a that the raised here with defend givenism by appeal to the idea of semi Notice carefully problem to is a a cognitive or quasi-cognitive state is fundamentally respect givenism logical problem (in broad sense of Thus it would be a misguided. The basic idea, after all, is to dis? "logical"). to think that it can be solved tinguish two aspects of a cognitive state, its capacity mistake simply by some sort of state which seems intui? to justify other states and its own need for justifi? indicating to sorts of character? cation, and then try to find a state which possesses tively have the appropriate is to understand how it is only the former aspect and not the latter. But it istics; the problem state to have those characteristics. seems clear on reflection that these two aspects possible for any is one and the same mistake would be to one occasion? cannot be separated, that it (The analogous a its made in connection with the free-will : feature of cognitive state, viz. assertive content, ally problem

30 an in an earlier "The Problem It is interesting to note that Quinton seems to offer analogous critique of givenness paper, and New of Perception," reprinted in Robert J. Swartz (ed.), Perceiving, Sensing, Knowing (Garden City, York, 1965), pp. 497 526; cf. especially p. 503.

This content downloaded from 129.118.139.129 on Sat, 10 Jan 2015 18:24:08 PM All use subject to JSTOR Terms and Conditions CAN KNOWLEDGE HAVE A EMPIRICAL FOUNDATION? 13

the mistake of attempting to solve the logical prob? cation. How then is the epistemic regress problem lem of how an action can be not determined but to be solved ? The natural direction to look for an also not merely random by indicating a subjective answer is to the coherence theory of empirical or seems act of effort similar state, which intuitively knowledge and the associated non-linear concep? to satisfy such a description.) tion of justification which were briefly mentioned are Thus foundationism appears to be doomed by its above.31 But arguments by elimination danger? own internal momentum. No account seems to be ous at best : there may be further alternatives which available of how an empirical belief can be have not yet been formulated; and the possibility genuinely justified in an epistemic sense, while still threatens that the epistemic regress problem avoiding all reference to further empirical beliefs or may in the end be of aid and comfort only to the cognitions which themselves would require justifi skeptic.32

The Received October 7, igy?

31 For a discussion of such a coherence theory, see my paper cited in note 11, above. 32 am I grateful to my friends Jean Blumenfeld, David Blumenfeld, Hardy Jones, Jeff Pelletier, and Martin Perlmutter for extremely helpful comments on an earlier version of this paper.

This content downloaded from 129.118.139.129 on Sat, 10 Jan 2015 18:24:08 PM All use subject to JSTOR Terms and Conditions