Laurence BonJour, Knowledge, Justification, and Truth

CHAPTER ONE

KNOWLEDGE AND THE GIVEN

1. The Problem of Knowledge.

The traditional philosophical explication of the concept of knowledge contain three main components. Knowledge is (i) belief, which is (ii) true; and (iii) justified. Although there is considerable body of recent work which suggests strongly that these three criteria are not, by themselves sufficient for knowledge,{2} at least not without considerable interpretation, it seems hard to deny that they are, if not too narrowly construed, at least necessary. Knowledge surely must involve a claim or assertion by the knower that something is the case. Further, this something must indeed be the case. And the would-be knower must have some ground or reason, meeting some appropriate standard, for thinking it to be the case; his claim must not be mere whim or guesswork. (Though there is, as we shall see, more than a little room for dispute over exactly what form this justification must take. It is also open to question whether or not the knower need be able to supply his justification explicitly if challenged.)

In any case, I shall assume for the purposes of this essay that the traditional explication of the concept of knowledge is essentially correct. The general question which I will discuss is whether or how our ordinary knowledge, and particularly our observational or perceptual knowledge, satisfies these three criteria, and especially the last.

One way of approaching these issues is by considering the following familiar problem. Consider the various ways which seem available for justifying one's belief or assertion of the proposition that-P.{3} One way would be to deduce that-P from some other proposition, that-Q, which is itself known to be true. Another way would be to justify the claim that-P "inductively." This last label, of course, covers a wide variety of sorts of justification, which are by no means obviously species of a common genus. If P has the form of a generalization, one may, in some cases at least, justify the claim that-P by citing other known propositions that-Q, that-R, . . . , which are particular instances of that-P ("instantial induction"). This may be only a specific form of a more general form of justification in which the propositions cited to justify that-P are simply deduced-consequences of that-P ("hypothetico-deductive method"). If P figures in some known law of either the form P => Q, or the form Q => P, where I use '=>' to represent "nomological" implication, then the claim that-P can, again in at least some cases be justified by citing that-Q and the law in question as both known.

This listing or modes of justification for propositional claims or beliefs is, of course, highly schematic and glosses over many major problems in epistemology. It will, however, suffice for my present purpose, which is to make the point that all of these ways of justifying a belief that-P, and thus of qualifying it as knowledge (if that-P is in fact true), seem to require that at least one further proposition, that-Q, and often more than one be antecedently known, which in turn requires that at least one further belief, the belief that-Q, be antecedently justified. But if the belief that-Q is also to be justified by citing a known proposition, the same need not for at least one antecedently known proposition will recur, and thus we seem threatened with an endless regress of justification, with the apparent consequence that something must be known before anything can be known, and thus that there is no genuine knowledge.

I shall refer to this problem as "the regress problem." Any answer or response to it must, it would seem take one of three possible forms:

  1. Although many, perhaps most, propositional claims are justified in terms of further propositional claims, the regress ends at a level or variety of propositions which are justified, not by other propositions, but by something non-propositional.

  2. Although all propositions are justified only by further propositions, this does not mean that knowledge is, considered as a whole, unjustified. The various propositions mutually support each other, any one or them being justifiable in terms of some of the others, and the framework of propositions as a whole is justified by this internal coherence.

  3. The regress is genuine, since no stopping point can be found, and mere coherence is insufficient. Thus human knowledge is really only knowledge so-called, and cannot be justified.

Variants of these basic positions are to be found throughout the epistemological literature.

Such a neat trichotomy tempts one to argue by elimination, and I shall succumb for the moment to this temptation. The basic objection to a position of sort (3) -- some variety of skepticism -- has been well-stated by C. I. Lewis:

I consider skepticism something worse than unsatisfactory; I consider it nonsense to hold or to imply that just any empirical judgment is as good as any other -- because none is warranted. A theory which implies or allows that consequence is not an explanation of anything but merely an intellectual disaster.{4}

This amounts to a sort of paradigm case argument. Clearly some of our beliefs are justified in a way in which others are not. One who wishes to deny this has either overlooked some possible mode of justification, or has set his standards of justification too high -- though, as we shall see, Lewis himself may be guilty of overlooking the latter alternative.

Consider then a position of sort (2), a coherence theory of justification (which need not, let it be noted, necessarily involve a coherence theory of truth in the familiar idealist style). Here too an objection advanced by Lewis seems obvious and devastating. There is no apparent reason to suppose that there is only one possible coherent and mutually supporting network of propositions.{5} But then why adopt this coherent framework rather than that? And won't any possible proposition fit into some such coherent framework? But surely the conclusion that all propositions are justified is just as absurd as the conclusion that none are.

Thus we seem left with a position of sort (1) as the only reasonable alternative. But can such a position in fact be developed? I.e., can a coherent account be given of how the framework of propositional knowledge can be finally justified in terms of something non-propositional? The next few paragraphs attempt to give a schematic outline of the usual way in which this task is attempted.

The question lurking behind all of these matters is that of how the structure of propositional knowledge and belief relates to the world which it is about. The thing that seems most obviously wrong with a coherence approach, after all, is that it seems to leave out altogether this most fundamental dimension of cognition, thus leaving one with an apparently arbitrary choice amongst competing coherent systems. One's intuitions cry out that the way to make this choice is to find out which propositional system depicts the way the world really is. And the way to do that, according to the empiricist tradition at least, is to observe the world via the senses, to look and see. Thus observation is identified as the key relation between the mind and the world.

There can, of course, be no question that in some sense this is so. What is crucial for epistemology is to say just how it is so, i.e. to give an adequate account of the notion of observation, and especially of how observation figures in the justification of knowledge. On the face of it this would seem to be a fairly simple problem with a quite obvious solution, but this impression is in fact misleading in the extreme. What one wants to say, of course, is something like the following. I observe, say, a tree directly in front of me, and thereby come to believe and/or assert, truly, that there is a tree before me. Clearly, it seems, this belief or assertion does not require justification in terms of some other (known) proposition. What justifies it is rather my "experience" of the tree. The tree is, one is tempted to say, "directly presented" to me visually, and hence I report it. What further justification could possibly be required? In semantic terms, one sees from one's experience that the world is (in fact) in such a state as to make the proposition that-P true, and hence proceeds to believe and/or assert that-P on the basis of that experience (thus following a "semantical rule").

It is this apparently simple and obvious view of observation, as involving a "direct experience" which is distinct from and justifies a propositional claim, which is a main root of the doctrine of the Given. Of course it is not trees and similar objects which finally turn out, on the usual view to be Given, but rather peculiar momentary entities called "sense-data." But the essence of the view is the same. The regress of propositional justification is held to end with a class of propositions which are justified, not by further propositions, but by one's "direct experience" of the entity which the proposition is about. (Or, alternatively, it may be said that it is just because they report what is "directly experienced," that such propositions require no justification or are self-justifying.)

The basic reason why ordinary physical objects are not normally taken as Given is of course the familiar problem of sensory error. If trees and similar objects were in fact directly Given in experience, so the argument goes, it would be impossible to account for the fact that we are sometimes mistaken in our observations of them. Givenness thus seems to be viewed as a direct confrontation of mind and object, so intimate as to allow no room for error to creep in. Now the implication of the fact of error with respect to even simple observations of ordinary physical objects might seem to be that this simple notion of direct presentation or confrontation, however obvious it may seem when we initially reflect on our ordinary perceptual experience of ordinary objects, is just too simple to adequately handle the full range of perceptual phenomena, perhaps too simple to do justice to the complex mechanisms involved in perceptual cognition. In fact, however, such a conclusion has seldom been drawn.{6} Instead the seemingly inevitable conclusion has been that the Givenness is mislocated by a view which takes ordinary objects to be Given. What is really Given is "sense-data," and our perceptual knowledge of ordinary objects is held to be indirect, being mediated by a direct knowledge of these other quasi-objects. The physical eye, having proved to be fallible, is replaced by the mental eye.

Thus, of our two more or less common-sensical intuitions, first that perception involves something like an element of presentation or Givenness, and second that it is ordinary physical objects which are thus presented, the latter is sacrificed to the former. This view was of course historically reinforced by the causal or scientific account of perception, which seems to show that ordinary physical objects are perceived very mediately indeed; that this conclusion involves a tacit confusion of causal with epistemic mediation was seldom noticed or appreciated.

Thus the regress of justification is held to end at a "level" of knowledge which, because it reports what is directly presented or Given in experience, requires no (further) justification. This level of knowledge represents the "foundation" of all the rest. It is in Givenness that the mind breaks out of the framework of propositional thought and makes direct contact with the world. The picture thus presented is, I think, an appealing and intuitively plausible one. Obviously the mind must at some point make "direct contact" with the world if knowledge is to be justified and the regress of propositional beliefs ended. Obviously too, the phenomenon of sensory experience is somehow basic to that mind-world contact. Clearly then, or so it has often seemed, propositions which formulate that of which we are immediately aware in experience, the sensory Given, will require no justification beyond the very awareness which they report, and will thus solve the regress and provide an adequate basis for knowledge.

This general sort of position, once so firmly entrenched as to be regarded as almost truistic, has of course fallen upon evil days, and is now explicitly advocated by almost no one. The reasons for this almost universal rejection are, however, by no means entirely clear, nor have the alternatives for epistemology been generally grasped. Often the arguments presented against the Given; or usually just against sense-data, are such as to make it quite obvious that something is wrong, without yielding any real insight into just what the problem is. I therefore want to begin this essay by presenting what I take to be the most basic and revealing argument against the doctrine of the Given, before going on to consider the alternative to Givenness as a solution to the regress problem, and to the problem of justification generally.

The argument to be presented is derived in its essentials from Wilfrid Sellars, and especially from his essay, "Empiricism and the Philosophy of Mind." I do not, however want to suggest that Sellars would endorse the argument in the form in which I shall present it. His own argument is considerably more complex, and indeed it seems quite likely that Sellars would regard my argument as a form of the "crude physicalism" which he in another place deplores [SM 22].

Although the present section has provided a rough sketch of the doctrine of Givenness, and of the way in which it is thought to solve the regress problem, it will obviously be necessary to consider the doctrine in much further detail before a counter-argument can be usefully formulated. In so doing, it will be useful to consider the work of a particular philosopher who has advocated the Given, both to provide concrete detail and to avoid any suggestion of strawmanishness. For this purpose I have selected C. I. Lewis, whose work represents perhaps the most carefully articulated version of an epistemology based on Givenness. It is to a presentation of Lewis' view of the Given that the next section will be devoted.

2. Lewis on the Given.

Lewis' primary account of the Given is to be found in his two books, Mind and the World Order (l929) [MWO] and An Analysis of Knowledge and Valuation (1946) [AKV].{7} Of the two books, it seems obvious that the later one should be regarded as the more accurate and definitive statement of Lewis' position, and it is to the account of the Given contained therein that my attention will be primarily devoted. But there seems to be no reason to think that Lewis intends AKV to replace, rather than to merely supplement, and perhaps in places correct, MWO, and hence it seems permissible to employ the earlier book, albeit with caution, to shed light on the later. Indeed, as we shall see, there are places where only such a procedure can make Lewis' views reasonable intelligible.

It is perhaps advisable to begin by quoting a passage which suggests the underlying rationale for the doctrine of the Given in terms which closely parallel our discussion of the regress problem in section one:

It is not, of course, intended to deny here that one objective statement can be confirmed by others; or to maintain that all corroborations of belief are by direct reference to immediate experience. Some objective beliefs are deductively derivable from others; and many -- or even most -- objective beliefs are inductively supported by other, and perhaps better substantiated objective beliefs. It is only contended that in such cases where one objective belief is corroborated or supported by another, (1) such confirmation is only provisional or hypothetical, and (2) it must have reference eventually to confirmations by direct experience, which alone is capable of being decisive and providing any sure foundation. . . . objective statements none of which could acquire probability by direct confirmations in experience, would gain no support by leaning up against one another in the fashion of the 'coherence theory of truth'. . . . [AKV 187]

Lewis' whole discussion of the Given is intended to give an account of immediate, direct, or Given experience (which terms are more or less synonymous for him), and of just how objective beliefs are eventually confirmed by that experience, which avoids the regress which would, in his view, leave them all without confirmation.

The key to Lewis' conception of the Given is the notion of immediacy or directness (which, as we have seen, are approximately synonymous with Givenness). The Given is "immediately presented" [AKV 3]. This notion of "immediacy" is to be understood by contrasting it with the notion of mediate or indirect perception. Something, x, is perceived mediately if it is perceived only via the perception of some other thing, y, which "represents" x [AKV 14]. Things which are not thus perceived immediately by perceiving an intermediary which "represents" them are perceived immediately. It would, I think, be a waste of time to inquire very closely at this point into what is intended by this notion of "representation"; to some extent it will be clarified in the sequel when we consider just how the Given is held by Lewis to function in the mediate perception of ordinary objects. It would seem, however, that the notion of representation is sufficiently broad that any case of clearly inferential knowledge would count as, for that reason, mediate. It is tempting to simply gloss 'immediate' and 'direct' as 'non-inferential', but this clearly will not do as an account of Lewis unless 'inferential' is in turn construed in such a way as to make our knowledge of ordinary objects, despite appearances to the contrary, inferential. For Lewis clearly wants to hold that such knowledge is not immediate.

One way of clarifying what is involved in this notion of immediacy is to ask why ordinary objects are not perceived immediately. The answer here is, I think, the one sketched in section one, viz. the problem of error. We do not perceive ordinary objects immediately because our perception of them is on occasion erroneous. Consider, in this regard, the following characterization of the Given:

Subtract, in what we say that we see, or hear, or otherwise learn from direct experience, all that conceivably could be mistaken; the remainder is the given content of the experience inducing this belief. [AKV l82f.]

There are two claims made in this passage, which is important to distinguish clearly. There is (i) the claim that the apprehension of the Given is immune to error or mistake. This could be taken as simply a matter of definition; i.e. immunity from error could be taken as one of the defining criteria of the Given. That such an interpretation of (i) would be wrong, or at least misleading, is shown by the second claim implicit in the quoted passage. This is (ii) roughly the claim that any "direct experience" (where 'direct' must be glossed as 'perceptual', rather than as 'given', since "direct experiences" are held to include elements which are subject to error) involves an element which is Given and thus immune from even the possibility of error. If (i) is merely a matter of definition, what reason could be given for thinking (ii) to be true?

I submit that there is a suppressed premiss at this point, the recognition of which will make the relation between (i) and (ii) quite clear. This premiss is roughly to the effect that the only way to account for error, or even the possibility of error, with respect to the perception of one entity, x, is by assuming that x is perceived only via the perception of some other entity, y, and thus that x is perceived only mediately. (Call this premiss alpha1.) The idea is that something like inference, though not necessarily ordinary conscious inference, is involved in any case in which error is possible. We perceive y and infer from there to x. If the putative perception of x turns out to be in error, the blame is to be laid at the feet of the quasi-inference from y to x, which depends on past regularities which may break down in the case in question; that such a regularity actually holds in a particular case is not, of course, itself perceived.

There are two reasons for thinking that Lewis holds some premiss like alpha1. The first is just that some such premiss or view is quite prevalent in the epistemological tradition to which Lewis generally belongs and that he does not repudiate it in the way in which he is generally quite careful to repudiate those aspects of views similar to his with which he does not agree. This reason, though weak in itself, is strongly buttressed by the second reason which is that such a premiss makes good sense of what would otherwise be a jumble of apparently unrelated and arbitrary claims. Thus if alpha1 were granted, it is clear why the perception of ordinary objects would be mediate. And alpha1 also has (i) as a more or less immediate consequence, and seems to require (ii) if a regress of mediate perceptions is to be avoided. If all error or possibility of error results from mediate perception, obviously what is perceived immediately cannot itself be susceptible to error; and something must be thus perceived without possibility of error so long as anything is perceived at all.

If this is right, it is alpha1 and its consequences (i) and (ii) which provide the real content of the notion of immediacy and thus of Givenness. (It is worth noting here that Lewis' repeated uses of the terms 'incorrigible' and 'indubitable' to characterize the Given are perhaps best viewed as stylistic variants, rather than genuine consequences, of (i); Lewis makes no clear distinction between certainty, incorrigibility, and indubitability.) This conception of the Given has the further consequence that our apprehension of a particular instance of the Given must be totally independent of any further apprehensions for its verification and truth:

. . . such apprehension neither has nor calls for any verification. The specific character of the presentation content, and the givenness of it is, so to say, its own verification. Any formulation of it will be independent, for its truth, of anything further and not contained in just this given experience itself. [AKV 26]

Such independence is a clear consequence of (i) and thus of alpha1. The point is just that if a particular instance of Givenness was subject to verification or falsification, or even to confirmation or disconfirmation, by further apprehensions, it could not be certain or immune from error. It would always remain possible that further experience would compel us to reject the original apprehension, and this very possibility would suffice to disqualify the apprehension in question as an instance of Givenness.

There is a problem lurking in all this which I want to at least suggest before proceeding further. Tho problem is that although the apprehension of the Given is held to be independent of other apprehensions in the way just outlined it seems clear that it cannot be totally independent of all inductive relations to other apprehensions, and these relations would seem to provide a basis for confirmation and disconfirmation. To see this, ask what relation exists between a particular apprehension of the Given and an objective apprehension of the physical world which is mediated by that apprehension. It is very hard to see how this relation could be construed in a way which would not make the discovery that the objective apprehension was in error at least a prima facie ground for doubting the truth of the apprehension of Givenness from which it was derived. I think that Lewis would reply that this sort of suggestion is at the very least misleading, since the objective judgment is never directly falsified, any more than it is directly verified. Rather its falsity, like its truth, is apprehended only via apprehensions of Given experience. But this only pushes the problem one step further off, by making it clear that one sort of Given apprehension must have at least lawlike relations to other sorts of Given apprehensions to which it is linked by the objective judgments which they all mediate. And then the question becomes why these relations cannot serve on occasion to disconfirm previous apprehensions of the Given. Why, that is, if Given experience alpha mediates my putative apprehension of objective state of affairs P, and P would also normally be accompanied by Given experiences B, C, D, E, etc., why cannot a failure to actually experience these latter instances of the Given tend to disconfirm, not only the apprehension of P, but also the apprehension of alpha? Clearly it must normally be the case that when alpha is experienced, B, C, D, E, etc. are also experienced, or else they could not all be linked to the same objective state of affairs. But why then is not the absence of the latter a good ground for doubting the presence of the former? I do not think that Lewis has any good answer to this question, beyond reiterating that the apprehension of the Given is absolutely certain, indubitable, and incorrigible. The problem for him, however, is to say just how this can be so if the Given is to have the sort of internal regularity which seems required if it is to support empirical knowledge.

I shall leave this issue hanging for the moment, and continue my exposition of Lewis. I do not think it is possible to give any further non-metaphorical characterization of the notions of immediacy and Givenness. Any further content the notion may have seems formulable only in terms of the analogy with a naively realistic view of ordinary perception, and the attendant metaphor of the mental eye. Some attempt to suggest this dimension of the notion was made in section one. I suspect that something like this notion is also what is involved in Lewis' frequent use of the metaphor of "presentation" to characterize Givenness.

What is it that we are thus aware of as immediately presented or Given, according to Lewis? One answer, literally tautologous of course for sense perception, is "sense data" (or "data of sense"). Lewis frequent use of this term [e.g. AKV 28, 30, 186, 190, etc.] is nevertheless quite significant, because it indicates his basic acceptance of the accompanying philosophical tradition, and thus explains in part his failure to give any very extensive characterization of particular instances of Givenness. He does, however, provide some few hints in this direction which are worth considering. What is Given is a sensuous "content" [AKV 30], "a certain complex of sensa or qualia" [AKV 188], which are absolutely specific in character [AKV 25]. According to MWO, an instance of the Given will be "either a specific quale (such as the immediacy of redness or loudness) or something analyzable into a complex of such" [MWO 60]. The Given is describable in what Lewis calls "expressive language" (of which more later) "by the use of adjectives of color, shape, size, and so on" [AKV l88]. This sort of characterization of what is Given is not very much help, employing as it does the equally unclear notions of "sense," "qualia," "immediacy," and "expressive language," but it is all that we are likely to get. The general intent, though by no means clear, should at least be familiar enough.

Somewhat more light, though by no means as much as one would like, is shed by the account of the Given in MWO. There we are provided with two criteria for the Given. The first of these is its "specific sensuous or feeling-character" [MWO 66]; this has already been discussed in the preceding paragraph. The other criterion is what Lewis calls the "brute fact" character or the Given [MWO 57]: "the mode of thought can neither create nor alter it -- . . . it remains unaffected by any change of mental attitude or interest" [MWO 66]. Without such an element of Givenness our knowledge would be "contentless and arbitrary" [MWO 39]. The same sort of point is made in AKV, where we are told that the Given is that element in experience "which we do not invent and cannot have as we will but merely find" [ARV 182]. This is more than a little puzzling, for it would seem that all elements in experience are like this, certainly more of them than Lewis wants to recognize as Given. If I look out of my window and see a tree, it seems clear that I do not "invent" the tree, nor is it something which is "arbitrary" or which I can "have as I will." In an obvious sense, I "merely find" it. But trees are not Given for Lewis.

Once again error rears its familiar head. What Lewis is concerned about is the possibility that my tree might turn out to be illusory in some way (his own example is an illusory deer) [AKV lB3]. But surely just the fact that the tree turned out not to exist would not mean that I "invented" it or that my apprehension of it was "arbitrary?" Clearly we are in the neighborhood of another suppressed premiss, this time to the effect that the only way in which an apprehension which is subject to actual or possible error can be other than invented or arbitrary is by containing or being mediated by a further apprehension which is immune to all error; it is this further apprehension which makes the difference between a false perceptual belief and an arbitrary guess or hunch. (Call this premiss alpha2.)

A further revealing statement about the nature of the Given occurs in MWO where Lewis says that the Given would be "qualitatively no different" if the subject were "an infant or an ignorant savage" [MWO 50]. The implication seems to be that the Given is independent of the training, background of experience, and degree of mental development of the subject. And this would further indicate that Lewis takes the ability to apprehend the Given to be unacquired.

The view that the capacity to apprehend the Given is unacquired fits well with Lewis' discussion of the linguistic expression of the Given. He sharply distinguishes between "the specific character of the presentation-content" and its "linguistic formulation" [AKV 26]. The latter makes use of what Lewis calls "expressive language":

. . . The distinctive character of expressive language . . . is that such language signifies appearances. And in thus referring to appearances or affirming what appears, such expressive language neither asserts any objective reality of what appears nor denies any. It is confined to description of the content of presentation itself. [AKV 179]

Such expressive language employs the phrase "looks like," "feels like," etc., to qualify an otherwise objective report so as to restrict it to the Given; an example would be: "I see what looks like granite steps before me" [AKV 173f.].

The key point to be made, however, is that Lewis regards the question of how the Given is to be linguistically expressed, even of whether it can be thus expressed, as only an interesting side issue, "a relatively inessential consideration for the analysis of knowledge" [AKV 182]. He concedes quite readily that actual reports of the pure Given are seldom, if indeed ever, made, and that they are difficult or impossible to make in ordinary language [AKV 182]. ("Expressive language" is not really ordinary, since considerable philosophical commentary is required, at the very least, to restrict the content of "looks like" statements to the pure Given; I shall have more to say on this point further on.) But such difficulties merely constitute "a comment upon the inessential character of language in its relation to the cognitive process" [AKV 204]:

Knowledge itself might well get along without the formulation of the immediately given: what is thus directly presented does not require verbalization. [AKV 183]

. . . That which we should thus attempt to formulate plays the same role whether it is expressed, or could be expressed, or not. [AKV 182]

In short, the Given and our apprehension thereof are fundamentally independent of language and linguistic expression. The influence of the mental eye approach seems clearly visible here, especially in the first of the two quoted passages.

I have now said everything I want to say about the Given itself, as opposed to the "interpretation" thereof, and it may be well to summarize briefly before proceeding. The Given is immediately apprehended, which means at least that such apprehensions are (i) non-inferential, (ii) certain, indubitable, and incorrigible, and (iii) totally independent of other apprehensions for their verification and truth. Further, (iv) what is thus apprehended is one or more absolutely specific sensuous qualia. (v) The capacity to apprehend the Given seems to be regarded by Lewis as unacquired. Finally, (vi) the apprehension of the Given does not depend in any way on being able to express linguistically what is thus apprehended, though such expression may in fact be possible through expressive language.

Lewis holds the Given to be present in every perceptual experience, whether veridical or not. There seem to be at least three arguments for such a view, though it is not clear that Lewis distinguishes them. The first of these is the regress argument, as sketched in section one (perhaps buttressed by some sort or symmetry consideration, since it is not at alI clear that the task of justification by itself requires that all perceptual experiences, rather than merely a strategic few, contain the Given). This argument is rather weak in itself, since it depends on rejecting coherence as an adequate means of justification, and is thus open to the counter-argument that coherence is all there is, and that Lewis' demands for an ultimate justification simply ask too much. The second and third arguments, which Lewis may believe to be mere elaborations of the first, depend on the suppressed premisses alpha1 and alpha2, respectively. The second argument thus claims that Givenness is required to account for erroneous perception, and the third that Givenness is required to account for the fact that erroneous perception is nevertheless not arbitrary. (Both arguments would presumably use some sort of symmetry argument to extend the presence of the Given from erroneous to veridical perception.) It is important to realize that alpha1 and alpha2 are the core of the notion of immediacy and thus of the notion of the Given; they alone link the various features of the notion in a coherent fashion and justify the epistemological claims which Lewis makes for it.

The Given thus characterized is, for Lewis, only one of the two key aspects or phases of perceptual knowledge. The other is the "interpretation" put upon the Given "in the light of past experience" [AKV l88]. It is this interpretation "which constitutes belief in or assertion of some objective fact" [AKV 109], and which thus derives from the narrow basis of the Given the whole elaborate structure of human knowledge. To see how the Given finally functions in the justification of human knowledge, we shall have to consider carefully just how it is "interpreted."

Perhaps the best way to approach the subject of the interpretation of the Given is via a consideration of Lewis' discussion of meaning. Lewis distinguishes with great care several "modes of meaning," the most important of which is "intensional meaning":

. . . The intensional is that aspect of apprehension in which it is significant of a classification made and of a criterion in mind which is the cue to this classification and the determinant of cognitively guided reaction . . . [AKV 72]

Very roughly, intensional meaning has to do with the criteria according to which a classification, as represented by a word which may be either applied or refused application, is made; while extensional meaning, in contrast, has to do with the entities which are thus classified or to which the word is applied. (It is worth noting that Lewis holds meaning to be prior to and independent of language, which is needed only for communication [AKV 72].)

Lewis distinguishes between two alternative ways of specifying the intension of a word or expression. "Linguistic meaning" is "constituted by the pattern of definitive and analytic relationships of the word or expression in question to other words and other expressions" [AKV 131]. It is what one could find out through the use of a dictionary, even if one didn't understand the language [AKV 132]. "Sense meaning," in contrast, is "the criterion in terms of sense by which the application of expressions is determined" [AKV 131]. 0f these two aspects of intension, it is sense meaning which for Lewis is the more basic and which ultimately determines linguistic meaning. And sense meanings are formulated purely in terms of the Given. Thus the idea is that we apply language to the world on the basis of a (logically) prior recognition that what is Given satisfies the appropriate criteria for the bit of language in question.

What is involved in this distinction is the problem, briefly discussed in section one, of how language and propositional thought relate to the world. Linguistic meaning is a purely intra-language matter, consisting solely of relations among words. It involves criteria of application for some words only in terms of the prior application of other words. But sense meaning involves a relation between the linguistic order and something fundamentally nonlinguistic, viz. the Given; it is via this relation that language ultimately gains empirical application. All this is made quite clear by the way in which Lewis argues for the distinction:

. . . Whoever approaches an empirical situation with intent to apply or refuse to apply an expression, or assert something as evidenced or its falsity as evidenced, must . . . be somehow prepared to accept or reject what he finds as falling under or confirming what he thus intends. Otherwise applicability could never be determined at all, and there would be no such thing as apprehensible empirical fact or empirical truth or falsity. [AKV 135]

And mere linguistic meaning obviously cannot serve this indispensable function. "Meaning as language-pattern abstracts altogether from that function of language by which it empirically applies" [AKV 140].

It is worthwhile to pause a moment here to reflect on the foregoing argument for sense meaning and, by implication, for the Given. In particular, I want to ask how Lewis moves from the premiss that linguistic meaning alone will not suffice to determine the empirical application of language to the world (which I will grant) to the conclusion that such application must be determined via a recognition by the language user that certain criteria are satisfied by the Given. No doubt the application of language is in some way mediated by sensory stimulation, but why must the situation be construed in terms of something like deliberate action or decision on the basis or a cognitive recognition? It would seem that we have uncovered another suppressed premiss (alpha3), this time to the effect that the only way to account for the intelligent application of language is to assume that such application is mediated by a pre-linguistic recognition that criteria are satisfied.

The three suppressed premisses, alpha1, alpha2 and alpha3, which I have argued are assumed by Lewis, are related in an interesting way. In each case, we have a situation which seems to require some sort of what may be broadly called mediation. The fact that we on occasion perceive erroneously, and also the fact that such errors are not arbitrary or invented, seems to suggest something which intervenes between us and the external object, something whose characteristics would explain both the fact and the nature of our error. Similarly, the application of language to the world seems to require some link between the two to explain just how the application is achieved. All this is very fuzzy, but I think it contains some element of truth. But the premisses in question all go beyond this to claim that such mediation must involve an awareness of the Given. And the question to be asked is why any such awareness that something is the case, anything which could usefully be described as a perception or recognition, anything in short which could be cited as a justification for knowledge, must be present. And this question suggests a generalization of alpha1, alpha2 and alpha3, a general implicit premiss or assumption which underlies Lewis' whole position, to the effect that a conscious cognitive state must be mediated, if at all, only by another state of roughly the same sort. Call this premiss alpha. It is easy to see how something like alpha could be derived from a roughly dualist view of the mind-body relationship.

It is an apparent implication of alpha that our apprehension of the Given is unmediated in any way, even causally. And it seems clear that Lewis is in fact committed to something like this. The point is just that if Lewis were once to admit that apprehension of the Given was causally, though not consciously or cognitively, mediated, perhaps along the lines of the scientific account of perception, he would be open to the question of why ordinary perception both erroneous and veridical, could not similarly result directly from causal mediation with no apprehension of the Given intervening. Here Lewis could only fall back upon the claim that such perception would not be capable of being justified.

I shall have much more to say about this sort of issue further on. For the moment I want to provisionally grant Lewis the distinction between linguistic meaning and sense meaning, and see just what he does with it.

The central thesis of Lewis' account of sense meaning is that the sense meaning of an objective judgment can never be exhausted by particular passages of Given experience, but will always involve further possible experiences which could be realized if appropriate action were taken. This feature of objective judgments explains both their necessary lack of certainty (because it is always possible that the future experiences might not in fact occur when the appropriate actions were taken, thus falsifying, or tending to falsify, the objective judgment), and their usefulness as a guide to intelligent action. The experiential implications of an objective judgment may be formulated in what Lewis calls "terminating judgments" [AKV 181]. The general form of a terminating judgment is "S being given, if alpha then E," where 'S' represents the initial Given sensory cue, 'alpha' represents the action taken, and 'E' represents the experience which results, and where all three are formulated in expressive language and thus refer only to Given experience. (Thus what 'alpha' actually represents is not an objective action, but the subjective Given "feel" of the action; what 'alpha' represents could be realized even though no action were in fact objectively performed [AKV 184].) Any objective judgment will have an unlimited number of such terminating judgments as consequences [AKV 177], and for this reason Lewis calls such objective judgments "non-terminating judgments" [AKV 181]. The sense meaning of a particular objective or non-terminating judgment is just all of the associated or entailed terminating judgments [AKV 211].

It may help to clarify what is involved in this distinction of terminating from non-terminating judgments if we briefly consider one of Lewis' examples. Consider then the objective belief that one is now being confronted by a sheet of white paper:

. . . The reason I believe this is that I see it: a certain visual presentation is given. But my belief includes the expectation that so long as I continue to look in this same direction, this presentation, with its qualitative character essentially unchanged, will persist; that if I move my eyes right, it will be displaced to the left in the visual field; that if I close them it will disappear; and so on. . . . [AKV 174]

To the extent that these actions are in fact taken, and the anticipated results indeed ensue, the belief is corroborated. But the belief has many further consequences which still leave open the possibility of disconfirmation and falsification; e.g.:

. . . that what I see could be folded without cracking, as a piece of celluloid could not; that it would tear easily, as architect's drawing-cloth would not; that this experience will not be followed by waking in quite different surroundings; and others too numerous to mention. . . . [AKV 175]

And it should once again be understood that all of these actions and anticipated results are to be formulated in expressive language, so that they are susceptible of complete verification. Thus, e.g., the experienced result of what subjectively seems like tearing the paper will be what looks like torn paper, i.e. the Given appearance of torn paper [AKV 175].

If we let 'P' stand for an objective or non-terminating judgment, 'S1', 'S2', etc., for initially Given sensory cues, 'alpha1', 'alpha2', etc., for Given apparent actions, and 'E1', 'E2', etc., for Given sensory results, Lewis' view of the relation between objective judgments and their experiential consequences can be represented as follows:
P >( S1&alpha1=>E1)
P >( S1&alpha2=>E2)
P >( S2&alpha1=>E3)
P >( S2&alpha2=>E4)
. . . .

[AKV 248f.]

Here '>' represents strict or analytic implication; each of the above statements is for Lewis an analytic truth, knowable a priori solely by understanding the meanings involved. In contrast, '=>' represents what Lewis calls "real connection," and what we would today call "nomological" implication. It is the connection involved in a law of nature and of the sort which supports counterfactual conditionals [AKV 223-27]. If such a list of statements is to completely specify the sense meaning of 'P', it must include all terminating judgments whose truth or falsity could possibly be relevant to the truth of 'P'; i.e. it must be the case that if all the statements on the right-hand side of the '>' symbol were true, then the truth of 'P' would be absolutely and totally verified, and could no longer be rationally challenged [AKV 193]. Thus the list must be infinite in length if one grants only that there could never come a time when an objective belief was theoretically, as opposed to practically, immune from falsification [AKV 176f.].

The picture just sketched of the interpretation of the Given is modified in various ways by Lewis, especially by introducing probability qualifications, but these refinements do not matter for our present purposes.

At long last we are in a position to see how the Given operates in Lewis' epistemology to provide an ultimate justification of human knowledge and solve the regress problem. Lewis' view distinguishes, as we have seen, three classes of empirical statements: (i) objective or non-terminating judgments, (ii) terminating judgments, and (iii) statements of the pure Given. (Lewis, as we have already seen, does not consider it an important issue whether or not statements of type (iii) are or even can actually be made; he takes the same view with regard to statements of type (ii) [AKV 204].) Objective judgments are related in various ways, inductively and deductively, and thus can often be justified in terms or further objective judgments. Not all objective judgments, however, can be so justified, by the regress argument and the argument against the coherence theory of truth. Some therefore must be justified in terms of terminating judgments. This is possible because the meaning of an objective judgment, in the most basic sense of meaning, can be represented by an infinite collection of terminating judgments, and hence if some of these terminating judgments are known to be true, the objective judgment is in some degree corroborated. (Just to what degree is a complicated question which occupies Lewis at great length; it need not be gone into here, however, since it does not affect the basic question of how objective beliefs come to have any degree of justification at all.) And the terminating judgments are known to be true simply because the passages of Given experience which they predict actually occur. Thus judgments of type (i) are justified in terms of judgments or type (ii), which are in turn justified in terms of apprehensions of type (iii). And these last, as we have seen, require no justification. (This picture is again rather seriously oversimplified, and requires elaboration especially with regard to the roles played by probability and memory; but again these refinements make no real difference for our central problem.)

This completes my initial presentation of Lewis' account of the Given and its role in the justification of knowledge. Before concluding the present section, however, it may be worthwhile to suggest briefly and in a preliminary way certain problems which seem to be involved in this account .

The best way to begin, I think, is by considering just what sort of apprehension the apprehension of the Given is. One way to approach this question is to ask whether or not that apprehension is to be considered knowledge. The answer would seem to be obvious and trivial. Apprehension of the Given surely seems to involve a belief in the broad sense of a cognitive claim; what is believed is just what a statement of the Given in expressive language would "formulate," viz. that the Given content is thus-and-so. Further, this belief is true and justified. So it is obviously a case of knowledge, by the criteria set forth in the first section.

But here we are confronted at once by the fact that Lewis explicitly denies that the apprehension of the Given constitutes knowledge. He gives the following reason:

Cognition generally, or the content of it, must have meaning in the sense that something is signified, believed in, or asserted which lies beyond or outside of the cognitive experience itself. When such cognition is veridical or is knowledge, it must correspond to, accord with, or be true of what is thus meant or affirmed. [AKV 27]

The key to this passage is obviously the phrase "beyond or outside of the cognitive experience itself," which is unfortunately far from clear. One problem is to discover just what is meant by the "cognitive experience." If this simply refers to the apprehension of the Given, then it is most unclear why the quoted criterion is not satisfied. Surely the Given content (consisting of sensuous qualia) is distinct from our apprehension thereof; the latter, but not the former, is true, certain, justified, etc.

Perhaps what is meant is that the Given depends for its existence on being apprehended, so that it is not "outside of" or "beyond" such apprehension in the sense of being capable of independent existence. But Lewis, unlike other sense-datum theorists, nowhere makes this sort of claim. And in any case, why should this sort of dependence prevent the apprehension of the Given from being knowledge? The picture which comes to mind is the idealist one of the mind "creating" the world which it knows. But we need not go into the puzzling question of whether the apprehension by the mind of a world which is thus its own "creation" could count as knowledge. Lewis, as we have seen already, emphatically rejects the idealist picture by insisting that the Given is a brute fact of experience, "which we do not invent and cannot have as we will but merely find" [AKV ] 182]. It would surely seem that this feature of the Given is sufficient to place it "beyond" the apprehension of it in the only way required for that apprehension to count as knowledge. Of course, if Lewis wants for whatever reason to reserve the term 'knowledge' for apprehensions where error is, on his view, possible, he is free to do so; but this makes the whole issue seem, as Lewis himself suggests, purely terminological [AKV 29f.].

I think, however, that there is a great deal more involved than mere terminology. On the contrary, Lewis' doubts about whether apprehension of the Given is to count as knowledge seem to me to be symptomatic of a serious difficulty in his position, about which he seems to have been uneasy, but which he never seems to have explicitly grasped. This difficulty lies closer to the surface in MWO, and has been buried to a considerable extent by the time we get to AKV. In the earlier work, Lewis discusses and rejects the familiar Russellian doctrine of "Knowledge by Acquaintance," on the grounds that "knowledge always transcends the immediately given" [MWO 118]. The chief reason for this is that:

. . . there is no concept the denotation of which does not extend beyond the immediately given, and beyond what could be immediately given. And without concepts there is no knowledge. [MWO 121]

According to this paragraph, apprehension of the Given is not knowledge because (i) knowledge requires concepts, and (ii) concepts do not apply to the Given. I think that (i), however opaque, represents an important insight. The immediate problem, however, is to understand (ii).

It is hard to reconcile the paragraph just quoted with the one which immediately follows it in MWO. There Lewis claims that:

There are recognizable qualitative characters of the given, which may be repeated in different experiences, and are thus a sort of universals; I call these "qualia." . . . [MWO 121]

And further on:

Qualia are universals, and they are universals such that without the recognition of them by the individual nothing presented in experience could be named or understood or known at all. . . . [MWO 123]

Now the obvious question to be asked is why these "recognizable qualitative characters," these "universals," are not concepts. (Which may be the same question as why, if they can be recognized to apply to the Given, the recognition that they thus apply does not constitute knowledge.)

The answer which Lewis gives to this question in MWO is that concepts are fundamentally public and relational in character, while qualia are essentially private and non-relational -- indeed ineffable. The conceptual meaning of a term is constituted by its relations to other terms; Lewis employs here the analogy of a relational view of space and the points contained therein [MWO 81ff.]. The relations in question here are clearly the same ones that are involved in the linguistic meaning of AKV. This conceptual network is common to the speakers of a particular language. Qualia, in contrast, are involved only in the application of the network in the experience of a particular individual. Such application has the effect of conveying the relations of various qualia within the experience of that one individual, but the qualia themselves are necessarily private in the very strong sense of being incommunicable. It would make no difference if the particular qualia associated with a particular word or concept varied widely from individual to individual, so long as the variation was systematic and the pattern of relations of those qualia to the qualia associated with other concepts remained constant or approximately so [MWO 124f.]. We are clearly in the dialectical vicinity of familiar controversies about the possibility or impossibility of "private languages"; indeed MWO contains a very clear anticipation of Wittgenstein's famous "private diary" argument, the point of which is not, however, to show that the notion of such private qualia makes no sense, but only that the apprehension of them is incorrigible [MWO 125].

I should like, however, to avoid becoming embroiled in these issues, for the moment at least. The point of the last few pages has simply been to suggest that there is something at least a bit peculiar about the apprehension of the Given, in Lewis' account. That apprehension, though certain and justified, is not knowledge. The qualia thus apprehended, though universals, are not concepts. Further, though they play a very fundamental role in the acquisition and justification of knowledge, such qualia are absolutely incommunicable. And finally, such incommunicability in no way impairs their epistemological function. All this, if not obviously wrong, is at least puzzling. It will, I think, help to shed some light on the situation if we pause in our exposition of Lewis to draw a fundamental distinction between two very different sorts of apprehension. This distinction will be the subject of the next section.

3. Two Kinds of Apprehension.

The distinction which needs to be drawn at this point is between what I shall call conceptual apprehension, on the one hand, and discriminative apprehension, on the other.{8} The best way to begin is perhaps by considering some examples. A paradigm of the first sort of apprehension would be the case of someone who is familiar with English, whose sensory apparatus is functioning normally, who is confronted with a tree, and who says, "There is a tree." What he conceptually apprehends is (the fact) that there is a tree at the indicated place. Since we are supposing that there really is a tree, this apprehension may be said to be true or veridical. If we grant that it is justified in some way (just how is, of course, the whole problem), the apprehension seems to qualify as knowledge, indeed as a paradigm of empirical knowledge. The man in question knows that there is a tree at the indicated place. The conceptual apprehension is clearly a cognitive or epistemic matter.

In contrast, what I want to call discriminative apprehension occurs in any case in which one's behavior or dispositions to behave are influenced causally by some state of affairs. Any psychology book is full of examples. One sort of case is illustrated by any moderately complex activity, such as walking. Performance of such activities involves a complicated discriminative apprehension of what the various muscles are doing. This is shown clearly by the fact that when the appropriate nerve impulses are cut off or not received in the brain, one's ability to perform the activity is impaired or destroyed.{9} But only someone with a very advanced knowledge of human physiology will have even a remote conceptual apprehension of what is going on, i.e. will have propositional knowledge about the nerves, etc., of the same sort as the man in the case first considered has of the tree.

An interesting and rather different sort of case is provided by investigations into the ability of blind people to "perceive" (this is the psychologist's term) the presence of silent obstacles and avoid contact with them. It was known for some time that blind people had this ability, but no one, including the very blind people themselves, had at first any idea of how they did it. A series of experiments established that such obstacle-avoidance was achieved through the sense of hearing, and depended on the reflection of high-pitched sounds by the obstacles.{10} Clearly the blind people possessed a discriminative apprehension of the reflected sounds; i.e. the presence or absence of the sounds made a clear difference in their behavior and dispositions to behave. But it is equally clear that they had nothing like a conceptual or cognitive apprehension of the sounds. And the interesting thing about this case is that the discriminative apprehension of one thing, the sounds, was apparently causally responsible for the conceptual apprehension of quite a different thing, the obstacles in question. (Note that the obstacle was also presumably a causal antecedent of the sounds.)

The examples could be easily multiplied. Very many details of our own personal anatomy obviously influence our behavior, even though we are quite incapable of apprehending them conceptually. It also seems to be a psychological fact that people can be trained or conditioned to make various sorts of responses without having any idea of what they are responding to. Psychologists tell us that we respond to a wide variety of perceptual cues, which we do not in any ordinary sense know to be present, and do not even in general possess the concepts to formulate. All these sorts of phenomena are cases of discriminative, but not conceptual, apprehension. Of course, all cases of conceptual apprehension will also be cases of discriminative apprehension. The point is just that the category of discriminative apprehension seems to be significantly wider than the category of conceptual apprehension, in that people can be influenced in their behavior by many things of which they are not conceptually aware.

I have attempted to suggest the distinction in question by citing examples. Such a procedure is fine up to a point, but now it is time to ask what exactly is the difference between these two sorts of apprehension. What I want to suggest is that discriminitive apprehension is to be regarded as in itself a purely causal phenomenon, depending on external and internal stimuli of various sorts, but not in general or necessarily involving any cognitive or conscious component whatsoever. The support for such a view derives from considerations of simplicity and theoretical economy. There seems to be no good reason to deny that a purely causal process could produce behavioral responses of a systematic sort. Nor is there any real temptation to postulate a genuinely cognitive element in the discriminative behavior of an amoeba or an insect. But then why suppose that anything of a radically different sort need be present in human behavior in cases where no conceptual or cognitive apprehension seems to be evidenced?

Conceptual apprehension, in contrast, seems to involve more than simply causal processes of the sort involved in purely discriminative behavior. (Which is not to suggest that it might not turn out to involve only causal processes of a more complicated and sophisticated sort.) But exactly what "more" is involved? The obvious answer is that conceptual apprehension involves the use of "concepts," of a "conceptual framework." But this more or less tautologous answer is of little help unless we can achieve some clearer notion of just what is involved in it. The account of concepts and conceptual frameworks which I want to suggest is very similar to that which Lewis develops in MWO and which has already been discussed to some extent in the preceding section: concepts are essentially relational in character, involving a complicated network of connections with other concepts. But to make this answer plausible will require some preliminary development.

It should be clear at least that the "more" involved in a conceptual apprehension is not merely the utterance of a series of sounds which would normally represent an appropriate sentence in some appropriate language. (I shall simplify for the moment by considering only those cases of conceptual apprehension which involve the actual utterance of a sentence in some language which expresses what we may call the "propositional content" of the apprehension.) If one can be trained to behave discriminatively with respect to a certain sort of stimulus, there seems to be no reason why the learned discriminative behavior could not, given the appropriate sort of training, be the utterance of a certain sequence of sounds. But it seems quite clear that one so trained need not have any notion of what the sounds he is uttering mean, or indeed that they have any sort of meaning at all, as opposed to being merely a string of nonsense syllables. Thus more is required for a conceptual apprehension than merely that a person respond to the state of affairs that-P with the utterance of a string of sounds which in fact normally represents (in some language) an utterance of the statement that-P. This alone would not suffice to make the sounds in question meaningful speech and the expression of a conceptual apprehension, as opposed to merely a discriminative response which just happened to take an unusual form. But what "more" is required?

I think that some light can be shed on this question by considering a particular case. Consider, then, a man with a broken collarbone. Clearly, in the absence of painkillers, nerve defects, etc., such a man will be discriminatively aware of the injury, as will be quite evident from his behavior. But clearly too, he need not and usually will not conceptually apprehend that his collarbone is broken. He may have no notion at all of the relevant anatomy or even of bones in general. Of course he will in general at least conceptually apprehend that he is injured, or that he is in pain, but these apprehensions are of course quite different from the apprehension that his collarbone is broken. And in any case, I can see no a priori reason why there could not be a person thus afflicted who, perhaps because of an extremely sheltered life, had never previously been himself injured or seen someone else injured, and thus possessed no concept of injury or even of pain, and hence could not conceptually apprehend even that he was injured. It seems conceivable that such an admittedly unlikely person might have virtually no conceptual apprehension at all with respect to the injury; he would quite literally "not know what to say" (or think). But he would still discriminatively apprehend that he was injured and that his collarbone was broken (assuming, as seems reasonable, that his behavior was different from the broken collarbone than it would have been for other sorts of injury). Further, as I have suggested already, it would seem to make no difference whether or not such a person came to be trained, as he might very well be through repeated breakings of his collarbone and associated conditioning, to utter a string of sounds which in some language, perhaps different from his own, would ordinarily be taken to mean, "My collarbone is broken."

Contrast the person just described with a trained doctor, possessing a thorough knowledge of physiology, who has learned through repeated experience to recognize the fact that his collarbone is broken, and who thus on a particular occasion reports, "My collarbone is broken." What is the difference between him and the other person who has been trained or conditioned to utter the same sounds? Why do we want to say that the doctor, but not the other man, has the concept of a broken collarbone and conceptually apprehends that his collarbone is broken? I submit that the difference consists in the fact that the doctor's utterance does not occur in a vacuum, but rather fits into and connects up with a complicated pattern of further utterances (and other responses) which the doctor makes or is able to make in specifiable circumstances. E.g., if asked, "What's a collarbone?" he will be able to give a more or less elaborate account. And similarly for further questions like: "What does it mean for a collarbone to be 'broken'?"; "Is it serious?"; etc. And whether questioned or not, the doctor will be able to draw further conclusions, which might or might not be overtly expressed, from the apprehended fact that his collarbone is broken.

Exactly what further verbal responses would be required to establish that a particular utterance in question was meaningful and thus expressed a conceptual apprehension is hard to say; it seems quite likely that there is in any case no sharp line to be drawn. But that some such pattern of associated responses and abilities to respond is what enables us to differentiate an utterance expressing a conceptual apprehension from one representing a purely discriminative response seems hard to deny, especially in the absence of any clear alternative suggestion. At least it is true that the demonstrated ability to make such responses is a sufficient condition for someone's utterance to be regarded as meaningful speech, and thus as expressing a conceptual apprehension. Clearly, if a person makes enough responses of an appropriate sort, the question, "But does he really understand what he is saying?" will soon lose all point. It also seems to be true that the ability to make responses of the indicated sort is a necessary condition for an utterance of a pattern of sounds to qualify as meaningful speech. If no further responses are forthcoming spontaneously, and no amount of questioning is able to elicit any such, it will soon become quite reasonable to conclude that the utterance was not meaningful speech, and hence was not the expression of a conceptual apprehension.

I have been arguing that once the distinction between discriminative and conceptual apprehension has been made, it becomes clear that a particular utterance can reasonably be regarded as the expression of a conceptual apprehension if and only if it occurs in the context of further utterances and dispositions to such utterances. Nor does it seem reasonable to regard this point as having purely epistemological significance, i.e. as merely a limitation on when we can know that an utterance is meaningful. This would amount to saying that one of two apparently similar utterances might express a conceptual apprehension and the other not, even though we could never tell from the outside which was which. Rather the contrast between our two original cases suggests that apparent speech about broken collarbones qualifies as genuine speech about broken collarbones only in the context of other apparent speech about bones, nerves, pain, etc., Just because the concept of a broken collarbone depends essentially on the concepts of these other things. (And of course vice versa, to some extent.) Once it is made clear that one could still apprehend in a discriminative fashion, the idea that one could have conceptual apprehension of or (possible) knowledge about broken collarbones without, e.g., knowing what a bone is becomes very dubious indeed.

Thus I conclude that one could not have the concept of a broken collarbone, and thus could not conceptually apprehend that he had a broken collarbone, unless he had these further concepts as well. It should also be clear that many more basic concepts, such as the concepts of a physical object, of a person, etc., are also involved. And there seems to be no reason to suppose that what is true of the concept of a broken collarbone is not true of concepts and conceptual apprehension generally. Thus we arrive at something at least highly similar to Lewis' view of concepts in MWO. Concepts are essentially relational entities; their nature and very being is constituted by their relations to each other in "logical space." It is this essentially relational character of our concepts which accounts for the essentially relational character of meaningful speech.

One way of putting all this is to say that our doctor has the concept of a broken collarbone and is thus in a position to conceptually apprehend that he has a broken collarbone just because he is also prepared to make various further inferences from such an initial apprehension, specifically those inferences which are involved in the use of the English words 'broken' and 'collarbone' [cf. RLG]. Traditionally it would have been thought that only deductive or "analytic" inferences were involved here, but once we have, on roughly Quinean grounds, abandoned the analytic/synthetic distinction, there seems to be no reason to exclude "causal," "nomological," or "inductive" inferences, or even to admit them on a different footing from those traditionally regarded as "analytic." It is, I am suggesting, just this associated pattern of inference which distinguishes a conceptual apprehension from one which is purely discriminative.{11} To suppose otherwise would be to suppose that one could have an apprehension from which he could draw no conclusions whatsoever, no matter how rudimentary, because it had no associated pattern of inference, and which was yet more than merely discriminative in a way which made it conceptual. I simply have no idea what this "more" could consist in.

So far in this section I have simplified things by fudging over the question of just how language relates to conceptual apprehension and concentrating on just those conceptual apprehensions which are in fact linguistically expressed. But of course this picture is too simple. It would be absurd to deny that our doctor might very well conceptually apprehend that his collarbone is broken without in fact making any overt statement to that effect. Clearly too, he need not explicitly make all of the inferences which I have claimed to constitute the difference between conceptual and discriminative apprehension, even in foro interno. A little reflection will suggest that it is quite impossible that he even could make all these inferences in explicit fashion, even to himself. So to this extent conceptual apprehension and overt verbal expression are independent of one another. There are nevertheless at least two reasons for regarding the capacity for conceptual apprehension and the capacity for verbal expression as intimately linked, so intimately indeed that the former could not reasonably be supposed to exist apart from the latter.

First. It seems quite clear that we could never know that some person or other creature possessed a capacity for conceptual apprehension unless he sometimes expressed such apprehensions and the inferences therefrom verbally. Here the point is very similar to that made in our treatment of the isolated utterance. The reason is that the complexity and subtlety of the pattern of inference bound up with even the simplest conceptual apprehension is such that only verbal behavior or behavior equally complex could hope to convey it. And of course any behavior as complex as verbal behavior, and which succeeded in conveying a conceptual claim, would arguably be just some alternative sort of verbal behavior, perhaps some sort of sign language. I am not here claiming merely that although in the absence of verbal behavior one might think that some creature possessed a capacity for conceptual apprehension, one could never be sure in such a case. The claim is rather the much stronger one that only verbal behavior could provide adequate grounds for reasonably supposing that conceptual apprehension was present. Any less complicated form of behavior could be accounted for merely in terms of discriminative apprehension; to even hypothesize the presence of conceptual apprehension in such a case would be as otiose as hypothesizing that plants have to think to turn toward the light. (The operative principle here is, of course, just some version of Ockham's Razor.) Thus the capacity for conceptual apprehension and tho capacity for verbal behavior are at least inseparable insofar as our knowledge of others is concerned.

Second. There is also the problem of concept formation, of how the capacity for conceptual apprehension is acquired. Once it is realized that conceptual capacity essentially involves the capacity for a pattern of responses which at least mirror very closely the pattern of linguistic responses involved in the use of a word for that concept in an appropriate language, the view that conceptual capacity could be acquired apart from verbal capacity becomes far less appealing. Certainly one way in which such conceptual capacity might be acquired is via the same process of conditioning through which one learns a language. On this view, the unexpressed conceptual apprehension is just a state of the perceiver which is linked to various other (conceptual) states of the perceiver by conditioned connections which parallel and mirror the inference relations which obtain between the sentence which would verbally express that apprehension and the other sentences which would analogously express the other conceptual states. Such a state is obviously also linked to the overt utterance which would express it, so that it becomes plausible to regard it as simply a disposition to such utterance, which of course may be inhibited by other factors. The conditioning which establishes the links between the conceptual states is, on this view, the very same conditioning via which the capacity for the associated verbal responses is acquired. One learns to apprehend and think conceptually only by learning a language in which one can express such thoughts and apprehensions [cf. RLG 333].

I shall have at least somewhat more to say about this picture of concepts and conceptual states further on. The present point is that one who would deny this intimate link between conceptual capacity and linguistic capacity must provide some alternative account of concept acquisition. (The alternative would be to regard conceptual capacity as unacquired; it does not seem to me that Lewis at least would want to do this.) And once the distinction between discriminative apprehension and conceptual apprehension has been drawn, such an alternative is by no means easy to find. Certainly the traditional empiricist or abstractive theory of concept formation loses much of its appeal, once it is realized that possession of a concept, as opposed to merely the capacity for a discriminative response, intrinsically involves possession of further concepts related inferentially to the first. The idea that the capacity for making this set of inferential moves could be acquired simply by exposure to a few instantiations of the concept in question seems highly dubious at best. ln any case, moreover, the empiricist theory seems on reflection to be either clearly circular, presupposing the very capacity for which it is intended to account, or else quite hopelessly vague and unclear. To see this, one need only ask whether our apprehension of the instances from which we are said to "abstract" the concept is discriminative or conceptual apprehension. If the latter, then the account is plainly circular. But if our apprehension of the instances is of a merely discriminative sort, then it becomes most unclear just what "abstraction" is supposed to mean. The picture suggested by the term is of someone thinking that this and that are similar in such-and-such a respect, and thereby abstracting a concept of the respect in which they are similar. But if our awareness of this similarity is purely discriminative, if we have no conceptual knowledge of it, what reason can there be to suppose that anything which can usefully be called "abstraction" can take place [cf. EPM 176]? Further, as I have already tried to suggest, discriminative apprehension quite commonly takes place without any concept acquisition resulting. So what then is "abstraction" supposed to mean if it is to be more than a mere label for we-know-not-what?

The conclusion which I draw from the above considerations is that the capacity for conceptual apprehension depends essentially on the capacity for linguistic expression of such apprehension, and cannot be reasonably supposed to exist without it. This is essentially Sellars' thesis of psychological nominalism [EPM 160]. Something like it is also very likely present somewhere in the maze of the Philosophical Investigations.

To sum up very briefly the distinction which I have been attempting to draw. Discriminative apprehension occurs in any case in which a person's (or animal's) behavior is influenced, presumably causally, by any state of affairs. In a pure case, there is no reason to regard it as more than a merely causal matter, without any element of cognitive awareness. Conceptual apprehension, on the other hand, occurs only when a person responds to a stimulus in a certain specific way, viz. roughly by occupying a certain position in a pattern of inference relationships. On this view, the whole difference between conceptually apprehending that-P and merely discriminatively apprehending that-P consist in an awareness of the (deductive and inductive) consequences of the proposition that-P. And this awareness consists solely in being able to actually make the relevant inferences when the occasion warrants. Finally, there is no reason to think that such a conceptual apprehension and the accompanying awnreness of consequences could exist apart from the capacity to verbally express both the apprehension and the consequences.

The upshot of all this, as I have suggested at various points, is that it seems reasonable to regard conceptual apprehension, but not discriminative apprehension, as constituting knowledge in a strict sense. In a way this is a purely terminological recommendation which need not be obeyed. But it is most important to understand the rationale behind it. Conceptual apprehension clearly involves a judgment or propositional claim that something is the case. Discriminative apprehension, on the other hand, involves judgment or assertion only in whatever weak sense in which one can judge or assert that-P without being able either to draw any further conclusions, no matter how simple, from that judgment or to provide any justification for it. For it is just the inferential relations which are lacking in the case of a purely discriminative apprehension which provide the means both for drawing conclusions and for offering justification.

There are many contexts in which this difference may reasonably be regarded as having little importance. But in the context of an epistemological discussion, particularly one concerned with the problem of justification, the difference looms truly enormous. In the next section I shall try to show why by applying the distinction to Lewis' discussion of the Given.

4. Why the Given is a Myth.

The dialectical position at which I have arrived is as follows. In section two, I attempted to give an account of Lewis' conception of the Given and of its epistemological function, and ended up by raising the question of what sort of apprehension the apprehension of the Given would have to be if it is to play the role which Lewis assigns to it. In section three, I attempted to show the point of this question by distinguishing two rather radically different sorts of apprehension which are arguably present in human cognitive behavior. Obviously the next question to ask is whether or not either of these two sorts of apprehension will do as an account of the apprehension of the Given. In what follows, I want to argue that neither a conceptual nor a discriminative apprehension of the Given will do for Lewis' purposes, and also that no clear third alternative seems to be available, so that the apprehension of the Given is best regarded as muddled conflation of our two sorts of apprehension and thus as not a genuine feature of human cognition.

There are several excellent reasons for thinking that if the apprehension of the Given is one of our two varieties of apprehension, it must be discriminative rather than conceptual. In the first place, only such a view seems to make sense of the contrast between the Given and its interpretation, of which Lewis makes so much. This is clearest in MWO, where Lewis develops a relational view of concepts which, as I have noted already, is quite similar to the view developed in section three, and explicitly characterizes the element of interpretation as conceptual interpretation [MWO, ch. 3 passim). In AKV, the term "concept" is no longer used at this point, but the basic view seems unaltered, though less explicitly spelled out. The interpretation of a particular instance of the Given still consists basically in relating it to other such instances [AKV 188-901. It is just because these relations yield predictions which may not in fact be satisfied that the interpretation of the Given, unlike the pure apprehension of the Given, is subject to error, and is not independent of other apprehensions insofar as questions of truth and confirmation are concerned.

Secondly, the view that the capacity for apprehension of the Given is independent of linguistic capacity and perhaps even unacquired seems to strongly favor discriminative apprehension over conceptual apprehension. At least the most basic requirement for discriminative apprehension, viz. that differential stimuli result causally in differential states of the sensory apparatus, seems quite plausibly to represent an unacquired capacity, even if most or all behavioral responses to these differential states have to be acquired. In contrast, it seems quite dubious to suppose that the complex capacity for conceptual apprehension could be either unacquired or independent of language.

Thirdly, discriminative apprehension seems sufficient to satisfy at least one of Lewis' arguments for the Given, viz. that knowledge must not be "contentless and arbitrary" [MWO 39], that knowledge must involve an element "which we do not invent and cannot have as we will" [AKV 182]. Our discriminative responses are neither "arbitrary" nor "invented"; we clearly cannot have them "as we will." (This suggests the possibility that our conceptual apprehensions might be non-arbitrary just because they depend causally on discriminative apprehensions, even though containing no element of Givenness.)

Fourthly, it seems clear that a conceptual apprehension would not solve the regress problem, but would simply add a further step to the regress. If the regress is to end, the last propositional claim must be justified by something non-propositional. To turn the apprehension of the Given into just a further propositional claim is to abandon this sort of approach, and thus to be left with either coherence or skepticism. Why should a propositional claim about the Given have any more authority than any other propositional claim, unless it can appeal to something beyond itself for justification? This is the most basic reason why the apprehension of the Given cannot be conceptual.

Can the apprehension of the Given then be construed as discriminative in character? Unfortunately, there seem to be equally powerful reasons why discriminative apprehension will not work any better than conceptual apprehension. In the first place, the apprehension of the Given is said by Lewis to be certain, indubitable, and incorrigible. But these terms simply seem to have no clear application to a case of discriminative apprehension, where no genuinely cognitive claim is made. It is hard to see how such an apprehension can even reasonably be called true or false, let alone certain. Of course a discriminative response can be appropriate or adequate to the situation, but it can also be inappropriate or inadequate, so that it would not come out as certain, even is one were willing to gloss 'true' as 'appropriate.' And in any case such a crude pragmatism simply will not do. Appropriateness or adequacy is just not the same as truth, in the short run at least. The latter, but not the former, is a distinctively cognitive or epistemic notion.

Secondly, the Given is said by Lewis to consist of specific sensuous "qualia." But it is most unclear why cases of discriminative apprehension must involve such qualia. Why isn't simply a stimulus plus a behavioral response mediated by some sort of neurological circuitry sufficient? Consider in this regard the case of the discriminative apprehension by blind people of the high-pitched sounds, which apprehension allows them to know that an obstacle is present. Presumably the qualia in question would have to be aural qualia. But then we are struck by the fact that the blind people had no idea that hearing had anything whatsoever to do with their awareness of obstacles. Is it then possible to apprehend qualia without even knowing which sense they are associated with, let alone what particular sort of qualia they are? And would such a nebulous sort of apprehension provide any sort of a basis for justifying anything? At this point the already mysterious notion of "qualia" threatens to dissolve into the mist.

Thirdly, and most basically, a purely discriminative apprehension cannot solve the regress problem, for reasons already implicit in the preceding two paragraphs. Discriminative apprehension, I have argued, is a purely causal and non-cognitive phenomenon. But how can something of this sort provide justification for anything? It has no inferential connections, so there can be no element of inference in the process of justification. But how can there be justification without inference? It is not enough that the assertion of the last proposition in the chain result causally from a discriminative apprehension of the Given. The perceiver must be aware of the Given in a sense which would allow him to do something like citing it as evidence. But such an awareness seems clearly to go beyond a merely discriminative apprehension, and in fact to be basically conceptual in character. There is just no reason to think that discriminative apprehension in general involves any sort of awareness that could be used in a process of justification. Here again the case of the blind people is helpful. What reason is there to think that they were aware of anything which could have served to justify their conceptual claim that an obstacle was present? [Cf. RLG 356].

We seem to have reached something of an impasse. The apprehension of the Given, if it is to play the epistemological role which Lewis wants it to play, can be neither a conceptual nor a discriminative apprehension, but must be an apprehension of some third sort which somehow shares some of the features of each of the other two. Such an apprehension must not be conceptual, i.e. must not essentially involve any relations to other apprehensions which could not be indubitably and independently apprehended, but yet must be more than purely discriminative. But what does this "more" consist in? What could it consist in?

It is important to note that the argument here does not depend on the existence of a rigid line between conceptual and discriminative apprehension. It is quite possible, indeed likely, that as a person gradually learns the set of responses which make up a language, there is no precise point at which his apprehensions cease to be purely discriminative and become conceptual. But this fact will not help Lewis. If neither a purely discriminative nor a purely conceptual apprehension of the Given will serve his purposes, it is most difficult to see how a more or less random blend of the two can do any better. To the extent that such an apprehension is conceptual it will be open to one set of objections. To the extent to which it is discriminative it will be open to the other set of objections. I can see no reason why these objections from opposite quarters should somehow cancel each other out.

At this point it is useful to recall that, even for Lewis, the Given is admittedly an abstraction which never appears alone in experience, unaccompanied by any conceptual interpretation [MWO 66; AKV 26]. But now, in the light of our failure to find any acceptable account of just what sort of apprehension the apprehension of the Given is, it is worth asking how Lewis can be so sure that anything beyond a purely discriminative apprehension remains once the conceptual content has been excised from an ordinary perceptual apprehension. Why must there be some third sort of apprehension, an apprehension of peculiar sensuous qualia? Lewis, I think, has only two sorts of reply at this point. One is an appeal to the regress argument. This, I have suggested, is very weak by itself, and amounts simply to saying that something, we know not what, must be present if knowledge is to be justified. But it is open to a critic either to reject Lewis' standards of justification, opting for coherence, or even to embrace skepticism. The second sort of reply would appeal to premiss (alpha) and the variants thereof. But why should we believe (alpha)?

At this point one may begin to suspect that Lewis' view of the apprehension of the Given represents a mistaken conflation of conceptual and discriminative apprehension, an essentially incoherent combination of the two. Certainly it is easy to see how such a confusion could come about, if one realized that apprehension of some sort must be present prior to language acquisition and concept formation, without realizing that such an apprehension can be equally accounted for in purely causal and discriminative terms. Something like this might also account for the acceptance of a premiss like (alpha).

This sort of suspicion will be reinforced is we reflect briefly on Lewis' account of meaning and especially on the linguistic meaning/sense meaning distinction. The point of that distinction was the claim that the application of language to the world must finally be determined by a recognition that certain criteria are satisfied by the Given. Now if the discussion in section three is basically correct, the application of language to the world is identical with the application of concepts to the world. Thus the view seems to be that concepts are applied to the world only via the recognition that criteria are satisfied. But then both the criteria and our recognition that they are satisfied must be non-conceptual in character, on pain of circularity. And this seems to deprive the notions of "criteria" and of "recognition" of all real content.

Similar points could be made with respect to the account of error in terms of mediate perception, which we have found to be implicitly involved in the notion of immediacy, and also with respect to the claim that knowledge without the Given must be "contentless and arbitrary." In all of these cases, Lewis' use of the notion of Givenness combines a realization that some further sort of apprehension beyond the conceptual is involved with a failure to realize that this further apprehension need be only discriminative and non-cognitive in character. The result is a tendency, along the lines of premiss (alpha), to view this sort of apprehension on the model of conceptual apprehension, despite the obvious differences. Thus it is viewed as a kind of quasi-cognitive awareness, though not a conceptual one; as involving "qualia," which are universals, though not concepts; as true and justified, though not genuine knowledge; as providing justification for further claims, though not requiring it from them.

I suggest that at least most of the epistemological chores which Lewis regards the apprehension of the Given as performing can be handled by discriminative apprehension without significant loss. Thus the problem of the application of language to the world, as a solution to which Lewis introduces the notion of sense meaning, can be handled in terms of a discriminative apprehension where the behavioral response is just the application of a particular concept or concepts, i.e. the asserting of or becoming disposed to assert a particular sentence, with its accompanying inferential connections. Similarly, perceptual error can be accounted for as the same sort of conditioned conceptual response, where either the stimulus or the state of the perceiver is in some way abnormal, resulting in a conceptual response which does not accurately reflect the environmental situation; in other words, the familiar causal or scientific account of error can be applied. And conceptual apprehensions are not "contentless and arbitrary," not because they contain an element of "sheer datum," but because they are conditioned to external stimuli [cf. RLG]. All of this amounts, of course, to a denial of premiss (alpha) and of its variants, (alpha1), (alpha2), and (alpha3). (These points will be further elaborated in the next chapter.)

I conclude that Lewis' Given is indeed a myth. No coherent account can be given of how it is apprehended which squares with other plausible considerations about human cognitive apprehension; the account of the nature of the Given and of its omnipresence in perceptual knowledge is puzzling indeed and depends upon certain premisses which we have found cause to reject; and the only remaining argument for the existence of the Given is a general one which depends on the elimination of all other alternatives on grounds which, though persuasive, are by no means conclusive.

I think that the argument can be fairly easily generalized so as to rule out the whole general approach to the problem of justification which the doctrine of the Given presents. The core of this approach is the idea that the elaborate structure of conceptual and propositional claims represented by human knowledge is finally to be justified by something outside that structure, something which is thus itself non-conceptual. But any such view, no matter what sort of thing it takes this ultimate foundation to consist in, must give an account of how we apprehend it in a way which will allow it to serve as justification for our conceptual knowledge. But until some third sort of apprehension is suggested and clearly characterized, the only alternatives for such apprehension are conceptual and discriminative apprehension. And neither of these can do the job here for reasons which we have already examined at length. So the framework of conceptual claims cannot be justified from without, but rather must be justified from within if it is to be justified at all. The alternatives, that is, are coherence and skepticism. Thus my next task will be to examine coherence and the objections to coherence,


[Table of Contents] -- [Go to Chapter 2]

Notes

{2} The article which initiated this discussion is Edmund Gettier's "Is Justified True Belief Knowledge?" Analysis 23 (1963), pp. 121-23, reprinted in A. Phillips Griffiths (ed.), Knowledge and Belief (Oxford: Oxford University Press, 1967). See the bibliography in Griffiths for the numerous discussions of and replies to Gettier's paper. [Back]

{3} It will simplify matters to speak here and throughout of propositions. Anyone whose nominalistic inclinations are offended by this should find it relatively simple to translate what is said into talk of sentences. [Back]

{4} Lewis, "The Given Element in Empirical Knowledge," Philosophical Review 61 (1952), 175. [Back]

{5} Ibid., pp. 168-69. [Back]

{6} The only exception I can think of is Critical Realism in general, and Roy Wood Sellars in particular. See his Philosophy of Physical Realism (New York: Macmillan, 1932; New York: Russell, l966). [Back]

{7} References to those two books will use the indicated abbreviations and will be placed in the text. [Back]

{8} This distinction is, I think, implicit in EPM and RLG. It is also formulated in P. T. Geach, Mental Acts (London: Routledge & Kegan Paul, 1957). That Sellars' argument can be usefully formulated in terms of such a distinction was first pointed out to me by Richard Rorty. [Back]

{9} Cf., e.g., Norman L. Munn, Psychology, 5th ed. (Boston: Houghton Mifflin, 1966), pp. 52f. [Back]

{10} Cf. Sellars, "Is There a Synthetic A Priori?" [Back]

{11} Cf. Sellars, "Is There a Synthetic A Priori?"sn, reprinted in Science, Perception, and Reality. [Back]


[Table of Contents] -- [Go to Chapter 2]