SORITES ISSN 1135-1349
Issue #06. August 1996. Pp. 4-20.
Concepts play a vital role in the action-oriented foundationalism of C. I. Lewis. As well they should: without concepts to mediate the mind's interpretation of the sensuously given, experience itself would not be possible; neither would knowledge. Thus explains why he belabors the explication of conceptualization in both Mind and the World Order [MWO] and An Analysis of Knowledge and Valuation [AKV].<1>Foot note 1_1 For Lewis, all experience is indirect or «thick,» since concepts are always brought to bear upon the sensuously given. This suggests that conceptualization causally figures in the production of all [cognitive] behavior, a view today championed by proponents of folk psychology; indeed, it's championed by almost everyone currently engaged in explaining the mind. After all, conceptualization is explicit representation par excellence, and there can be «no intentional causation without explicit representation.»<2>Foot note 1_2 Or so we have been conditioned to believe. Still, the dearth of present-day foundationalists notwithstanding,<3>Foot note 1_3 Lewis, clearly, is not alone in privileging concepts when explaining why an agent does what she does.
Herein lies the problem: his action-oriented theory of knowledge is consistent with a minimally representational picture of mind; so too is his notion of concepts as social entities. As such, Lewis' picture of mind is far from being a species of the cognitivist orthodoxy that now dominates the philosophy of mind [and cognitive science].<4>Foot note 1_4 Indeed, his account is much less dependent on internal intensional entities than his claims about «thick experience» might lead one to believe. Since many cognitive scientists [including me] don't buy the cognitivist story, given the minimally representational picture of mind emerging from such fields as dynamic systems theory,<5>Foot note 1_5 situated action,<6>Foot note 1_6 and connectionism,<7>Foot note 1_7 all the more reason to give Lewis' theory a second look. Moreover, since Lewis contends that most of our behavior isn't mediated by conceptual interpretation -- which is another view [the above] opponents of cognitivism take seriously -- the time is right to reconsider Lewis' views on mind and knowledge. I aim to do just that.
My purpose for this paper is to resolve the apparent contradiction between Lewis' claim, on the one hand, that all experience is thick, and on the other, his claim that most of our behavior isn't mediated by conceptual interpretation. This is the major tension between Lewis' theories of knowledge and mind respectively. In short, I aim to resolve it. Toward that end, I begin by sketching-out his foundationalism. I then focus on what Lewis takes to be the nature of experience. Therein lies the major tension. Again, he claims that all experience is indirect or «thick» -- since concepts [or interpretation] are always brought to bear upon the sensuously given. But he also identifies a sort of experience that doesn't require conceptualization -- stimulus-driven experience -- the result of which are «unconsidered» responses (MWO, p. 403). At face value, Lewis' epistemology renders his theory of mind implausible: no agent could act if she were required to entertain the myriad beliefs that Lewis claims figures in the guidance of action. But rather than abandon Lewis' philosophy, I attempt to rehabilitate it. Rehabilitation is possible, I argue, for the following reasons. First, Lewis isn't claiming that his epistemology describes actual justificatory practices; rather, it describes what an agent could do. This renders his theory of mind more plausible. Second, the social character of concepts [and meaning] minimizes the need for internal concepts as causes for why an agent does what she does. And third, among his paradigm cases of cognitive behavior are paradigm cases of nonreflective action. Here's the rub: not only do such actions account for most of our behavior [as Lewis himself notes], nonreflective actions, though cognitive, don't require conceptualization.
Here come the qualifications. First, although I must summarize a great deal of material, I endeavor not to overlook any essential part of Lewis' views about empirical knowledge. So be warned: while my analysis is not exhaustive, an even-handed approach requires that I provide quite a bit of detail.
Second, most of my labor is directed toward explicating Lewis' theory [or rather, theories]. My analysis and conclusions, while controversial, pale in comparison to the controversial nature of antirepresentationalism. As such, this is not the place to evaluate Lewis' views from the standpoint of recent cognitive science. And though I am familiar with the controversies in cognitive science regarding representation[s],<8>Foot note 1_8 explicating them and doing justice to Lewis would entail making this paper far too technical and intolerably long.
Last, I assume that the story Lewis tells in AKV is an extension and refinement of the story he tells in MWO. To be sure, inconsistencies between the two accounts can be found. But bear this in mind: my aim is to rehabilitate Lewis, not to bury him.
1. THE BIG PICTURE
«[W]ithout concepts, there is no knowledge.» (MWO, p. 121)
«The primary and pervasive significance of knowledge lies in its guidance of action: knowing is for the sake of doing.» (AKV, p. 3)
The action-dependent nature of knowledge is one of the hallmarks of Lewis' epistemology. This is both a boon and a bane. The boon is that it makes Lewis' philosophy relevant to current cognitive science, much of which is concerned with modeling minds [and cognitive processing] in an action-dependent way. The bane is that in cashing-out all that is involved when a knowing agent to acts, minds would be too occupied with conceptualizing ever to do anything: concepts simply get in the way. First things first.
1.1 Lewis' Methodology
Before sketching-out Lewis' theory of knowledge, it will be worthwhile to say just a bit about his methodology.
First, Lewis considers his theory to be reflective rather than speculative. Reflective philosophizing -- an activity not limited to philosophers alone (see MWO, p. 2) -- involves reflecting on and clarifying what is already given in commonsense. In short, it is the business of philosophy «to investigate what we already know» (MWO, p. 2). Notwithstanding his references to Socrates (see MWO, p. 19), he doesn't have Platonic anamneis in mind. Rather, he means that «it is the business of philosophy to analyze and interpret our common experience, and by reflection, to bring to clear and cogent expression those principles which are implicit because they are brought to experience by the mind itself» (MWO, p. 36). Above all, the reflective method is empirical, analytic [i.e., critical], and pragmatic.
Second, although Lewis is willing to defer to the special sciences regarding «certain questions of phenomenal fact» (MWO, p. 4), he maintains that epistemology «is a subject too fundamental to rest upon distinctions drawn from the particular sciences» (MWO, p. 56):
It is not the business of philosophy, as it is in the natural sciences, to add to the sum total of phenomena with which men are acquainted. Philosophy is concerned with what is already familiar. (MWO, pp. 2-3)
The delineation of the fundamental concepts `mind' and `mental' is a truly philosophic enterprise. (MWO, p. 6)
Such passages strongly suggest that Lewis accepts a strict demarcation between the realms of philosophy and science. As such, he is clearly not a naturalist in the Quineian sense. Neither does he think that philosophers can contribute much to empirical model building [even when the special sciences haven't yet determined what the facts are]. Of course, there is no reason why Lewis should have been less dichotomous. Still, given that much philosophy today is offered in close partnership with the special sciences, the sharp division between philosophy and science that Lewis defends is worth noting.
The task now is to sketch-out his theory. I'll save most of my analysis of it for later [Section 3].
1.2 Lewis' theory of knowledge: Part I [`Action']
As noted above, Lewis claims that «knowing is for the sake of doing.» Only «active beings» can have knowledge, for knowledge is necessary if a creature is to evaluate which course of action will best achieve its ends. After all, «[f]or a being which did not assign comparative values, deliberate action would be pointless» (AKV, p. 3). Thus, knowledge involves two general species of activity: action [which is the output of comparative evaluation] and evaluation [which is the process by which a purposive action is selected and justified].
The sort of `acts' [or behavior] that depend on knowledge are all and only those acts that involve the anticipation of consequences -- actions that may be called `deliberate', `intentional', or `purposive'. Such acts arise only in creatures capable of exercising «explicit foresight» -- going beyond what is immediately present to the senses and anticipating possible experiences (AKV, p. 5). «To know,» Lewis says, «is to apprehend the future as qualified by values which action may realize» (AKV, p. 4). But while all knowledge-dependent action depends on evaluation, not all actions are knowledge-dependent. On this, Lewis is quite explicit:
[M]uch of our own behavior for which we are taken to be responsible is hardly such as is instigated by explicit foresight and assignment of values. Deliberate action shades off, in one direction, into that which represents instinctive tendencies and automatic responses, and in another, into that which has become habitual and is no longer attended by any definite prevision or assignment of consequences. Somewhere here a line must be drawn -- or more than one. Our own deliberately judged conduct belongs on one side, and those processes called actions or doings of inanimate objects and unconscious organisms plainly belong to the other. But between these two there remains an indeterminate middle ground -- e.g., what we do habitually and without consideration -- often covered by the broad term `behavior'. (AKV, p. 5; my emphasis)
Thus, Lewis' theory isn't intended to capture the process by which every action is selected, but rather capture only those acts that are deliberate or purposive -- the sort of acts requiring «explicit prevision of consequences and evaluation of these» (AKV, pp. 6-7). For now, I shall leave it an open question as to which of the following labels best describes most of our behavior: `knowledge-mediated' or `unconsidered responses'?
1.3 Lewis' theory of knowledge: Part II [`Evaluation']
Appropriately enough, Lewis directs most of his energy toward explicating the `evaluation' component of empirical knowledge, which breaks down into `transcen-dence' and `justification'. Justification, in turn, breaks down into `verification' and `conceptualization'. Conceptualization breaks down into `interpretation' and `the given'. There endeth the reduction. This description suggests that evaluation is rather convoluted. It is. Still, the basics are straightforward; identifying them will serve to summarize his theory.
All knowledge, empirical or otherwise, «has an eventual empirical significance.» What this means is that everything that is knowable or thinkable must ultimately refer to meanings that are «sense-representable» (AKV, p. 171). Why this is so, and how it works, will become clearer presently. What I wish to emphasize here is that Lewis is wholly wedded to there being two general types of statements: analytic statements and nonanalytic statements [or synthetic statements]. The former «assert some relation of meanings amongst themselves.» The latter «require relation of a meaning to what is found on particular occasions of experience.» «It is the latter class alone,» Lewis says, «which may express empirical knowledge» (AKV, p. 171).
For a synthetic statement to count as an expression of knowledge, it must go beyond -- «transcend» -- the mere reporting of what is immediately given in experience (MWO, p. 132). Not all synthetic statements have this feature, but all judgments do. Here's why: Judgments make a prediction about future possible experience, so they are subject to error. Such isn't the case with expressive statements (AKV, p. 184) -- direct reports of the «momentarily given» (MWO, p. 275). These statements express only the content of one's subjective experience. They make no prediction. They are not subject to any possible error. For example, `The thing in front of me seems furry' and `I see what seems to be a white object' are reports of the content of my immediate experience. Because mere reports of one's «apprehensions of the given» make no prediction, don't need to be verified, and are immune from error, they are not judgments (AKV, p. 183). As such, Lewis denies the possibility of direct knowledge by perception alone: «there is no knowledge by acquaintance; ... knowledge always transcends the immediately given» (MWO, p. 118).
What Lewis is trying to capture by `expressive statements' is the awareness one has about the content of one's immediate experience. Reports of such content don't count as expressions of knowledge because empirical knowledge is never certain, only probable. Since one is always certain about the content of one's experience, these reports cannot be expressions of knowledge. The difficulty here is in grasping the distinction between apprehensions of the given versus reports of apprehensions. Without language, however, it is impossible to talk about the former, much less express the latter (see AKV, p. 183). As such, there is no way of talking about the content of one's experience without employing concepts. Therein lies the problem, for concepts are not supposed to figure in direct experience. What is, rather, is only one's awareness of what is immediately given:
[T]here is such a thing as experience, the content of which we do not invent and cannot have as we will but merely find. ... [The] given is an element in perception but not the whole of perceptual cognition. 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. If there were no such hard kernel in experience -- e.g., what we see when we think we see a deer but there is no deer -- then the word `experience' would have nothing to refer to. (AKV, pp. 182-183; my emphasis)
In the next section, I shall say quite a bit about the contributions made to experience by the given and conceptualization. For now, again, the main point is: knowledge requires judgments, judgments require conceptualization, and conceptualization involves prediction -- going beyond what is immediately given to the senses. Ultimately, all empirical knowledge will reduce to the given and its interpretation. At this stage, the given is all there is. Thus, direct reports -- expressive statements -- constitute elements «in» knowledge rather than expressions «of» knowledge.
To be an expression «of» knowledge, the empirical statement must be a judgement. There are two sorts of judgments. The more general [and common] of the two are nonterminating judgments -- «statements of objective fact.» Judgments of the other sort are called terminating judgments -- «predicative and verifiable statements» (AKV, p. 185). Because terminating judgments are composed of expressive statements, since I have already explained what expressive statements are, let me turn now to terminating judgments.
Terminating judgments, obviously enough, are judgments, so they can express knowledge. Such isn't the case, recall, with expressive statements. But like expressive statements, terminating judgments are formulated in expressive language. Expressive language is used to convey the content of one's immediate presentation or subjective experience. Examples include `...seems like...', `...looks like...', etc. Expressive language is not used to make any assertion about objective reality. To do that, one uses objective language. The principal difference between terminating judgments and nonterminating ones lies in the fact that the latter are formulated in objective language. I shall have more to say about this presently.
Here's the rub: terminating judgments make a prediction; expressive statements do not. For example, the expressive statement `I see what seems to be a white cat' «neither asserts any objective reality of what appears nor denies any. It is confined to description of the content of presentation itself» (AKV, p. 179). Though direct awareness or perception of the given need not involve conceptual interpretation, the expression of such awareness clearly does. Such statements involve not merely the «conceptual interpretation» of the given, they imply «much which is not given» (MWO, p. 275). Therein lies the cognitive significance of direct perception and expressive statements: they function as «cues» for predictions; and predictions are expressed in judgments.
There are three «elements» to any cognitive situation that engenders a judgement [and judgments precede knowledge]. First, there is the presentation of the given, followed by its interpretation. This is how experience is made. While perception -- experience -- is sufficient to serve as a cue for prediction, let's suppose Stage 1 in the production of empirical knowledge ends with an expressive statement. The second and third elements -- respectively, an envisaged action and an expected consequence -- are the two aspects of the prediction [which is implied by the interpretation]. Thus, Stage 2 in the production of empirical knowledge ends with a judgement. Cognitive judgments all have the following form: `Given S, if I act in manner A, then E', where `S' is the sensory cue, `A' is some possible mode of action, and `E' represents an eventuality of experience. For terminating judgments, both the action and the empirical eventuality should be formulated in expressive language; e.g., `Given that there appears to be a cat on my desk, if I pull its tail, then I should aurally experience what would seem to be a meow sensation'.<9>Foot note 1_9 For nonterminating judgments, both the action and the empirical eventuality must be formulated in objective language; e.g., `Given that there appears to be a cat on my desk, if I pull its tail, then it should produce a meow'. Stage 3 in the production of empirical knowledge lies in testing the prediction -- i.e., performing the requisite action needed to verify [or falsify] the prediction. terminating judgments «admit of decisive and complete verification or falsification.» nonterminating judgments admit of only «partial» verification or falsification (AKV, p. 181). Thus, only if after pulling the cat's tail, I hear what sounds like a meow, does `There appears to be a cat on my desk' count as an instance of knowledge: this statement, having been verified [let us suppose], is what the above terminating judgment expresses. And only if after pulling the cat's tail, I hear what sounds like a meow, does `There is a cat on my desk' count as an instance of empirical knowledge: this statement, having been partially verified [let us suppose], is what the above nonterminating judgment expresses [all things being equal].
Believe it or not, the above sketch of the evaluation of nonterminating judgments was grossly oversimplified. Let me add a few wrinkles.
First, «an objective and nonterminating judgement must be translatable into judgments of the terminating kind. Only so could confirmation of it in experience comeabout» (AKV, p. 181). Thus, the verification of nonterminating judgments must ultimately hinge upon the verification of some terminating judgments, which in turn ultimately reduces to experience, which in turn reduces to the given and its interpretation. So, let `P' stand for what is expressed [or inferred] by a verified nonterminating judgment. An expression of empirical knowledge of this sort is probable only; its probability depends on its «grounds,» which are some given data of sense -- `D'. The data are certain. Thus, D probilifies P [together with «principles of probability or rules of induction» (AKV, p. 321)]. «Such,» says Lewis, «is the general character of my cognition at each successive instant» (AKV, p. 321). I shall have more to say about this below [in Section 3]. I've summarized the bottom-up version of his story in Figure 1.
Second, with every interpretation of any external object, the mind «implicitly predicts» a host of further experiences. Hence, no one action will ever verify any judgment about objective reality (MWO, p. 277). Indeed, «[f]or the object presented to be real, there must be more to it than could be given in any single experience» (MWO, p. 135). This is so because external objects, for Lewis, are conjunctions of a vast set of possible experiences: That is an X = [(If I should do A1, then E1 will result) & (if I should do A2, then E2 will result) & (if I should do A3, then E3 will result) & ... (if I should do An, then En will result)]. Lewis hangs his existential hat upon such conjunctions of possible verifications:
The whole content of our knowledge of reality is the truth of such `If-then' propositions, in which the hypothesis is something we conceive could be made true by our mode of acting and the consequent presents a content of experience which, though not actual now and perhaps not to become actual, is a possible experience connected with the present. (MWO, p. 142)
And knowing the «empirical eventualities» of a given presentation constitutes the a priori element to empirical knowledge (MWO, p. 294).<10>Foot note 1_10
In all this talk of terminating judgments, nonterminating judgments, evaluation, verification, and the like, it is easy to loose site of what motivates this paper: resolving some of the tensions between Lewis' theory of knowledge and his theory of mind. Detailing his epistemology was therefore crucial. It's take-home message is: All knowledge ultimately reduces to the content of direct experience.<11>Foot note 1_11 Cashing-out the nature of experience [and his too active conception of the mind] is the task to which I now turn.
2. EXPERIENCE = THE GIVEN + CONCEPTUALIZATION
«Experience does not categorize itself. The criteria of interpretation are of the mind; they are imposed upon the given by our active attitude.» (MWO, p. 14)
Experience does not spring fully formed from sense presentations alone. Rather, experience is constructed by the mind via interpretation of the data of sense. Thus, experience comprises «two elements»: the given and the conceptual interpretation put upon the given (MWO, p. 48). Although each element can be disassociated conceptually, in experience they are inseparable. All experience is a continuous train of the sensuously given, which the mind then actively interprets.
Both the given and the interpretation are each necessary, and neither by itself is sufficient to engender cognitive experience [and hence knowledge]. If there were no given, then knowledge would be «contentless and arbitrary; there would be nothing which it must be true to» (MWO, p. 39). Should one's theory deny this, Lewis says, one has placed it «beyond the pale of plausibility» (MWO, p. 48). And if there were no interpretation, then thought would be «rendered superfluous, the possibility of error becomes inexplicable, and the distinction of true and false [would be] in danger of becoming meaningless» (MWO, p. 39). Knowledge, recall, arises only where error is possible. So, if knowledge were based on the given alone, then all content must be veridical. But since we are creatures prone to illusion, hallucination, etc., content isn't always veridical. Thus, knowledge can't be based on the given alone (MWO, p. 43). Moreover, where error occurs, it arises «directly» from the conceptual interpretation put upon the given (MWO, p. 158). Since the possibility of error is one of the defining features of judgments, and judgments are necessary for knowledge, all knowledge involves conceptualization -- the taking of some attitude that «serves practical action and relates it to what is not given.» Interpretation serves as a «conceptual go-cart» to get one over the interval between the presentations of the given, and the end projected by one's purpose (MWO, p. 119).
Since it is `concepts' that are doing the work when the mind interprets or categorizes the given, at last we come to the nature of concepts.
2.1 Concepts
Lewis appeals to two senses of `concept', one is public, objective, and external; the other is private, subjective, and internal. The former he calls the `pure concept'. It is «that meaning which must be common to two minds when they understand each other by the use of a substantive or its equivalent» (MWO, p. 70). Let's call the latter the `private concept'. Its meaning is idiosyncratic and subjective. Both types of concepts can undergo evolution or a «succession of different meanings» (MWO, p. 68).
Lewis, however, is not primarily interested in private concepts. Instead, and for the following reasons, he focuses on pure concepts.
First, pure concepts figure in communication as well as in the transmission of knowledge; private concepts do not. The reason, of course, is that the individuation of pure concepts depend on «common,» «sharable,» and «expressible» meanings -- the sort of meanings that are necessary for communication (MWO, p. 80):
[I]t is obvious that common meanings do transcend such individual differences of perception as imagery. We use language to convey thought. If language really conveys anything, then there must be something which is identical in your mind and in mine when we understand each other. (MWO, p. 73; cf. AKV, p. 143)
The individuation of private concepts, however, depend on idiosyncratic meanings. Such meanings are «direct,» «nonsharable,» and purely individualistic (MWO, p. 81); i.e., they are relations between one's given and one's unique learning history, feelings, sensations, or imagery.<12>Foot note 1_12
Second, and relatedly, there are objective standards for the ascription of pure concepts; there aren't such standards for the ascription of private ones. As it would happen, these standards are also the «only practical and applicable criteria of common knowledge»: (1) we should share common definitions of the terms we use; and (2) «we should apply these terms identically to what is presented» (MWO, p. 76; also see p. 84). «Congruity of behavior,» Lewis says, «is the ultimate practical test of common understanding.» «Speech,» he continues, «is only that part of behavior which is most significant of meanings and the most useful for securing human cooperation» (MWO, p. 90). Therefore, it shouldn't be surprising that we share pure concepts. Aside from the fact that we are «confronted by a common reality,» we are «creatures fundamentally alike, having in the large the same needs and interests and powers of discrimination and relation» (MWO, p. 91; also see p. 110).
Third, although knowledge is relative to minds, pure concepts extend knowledge outside of minds:
Relativity is not incompatible with, but requires, an independent character in what is thus relative. And second, though what is thus relative cannot be known apart from such relation, still the other term or terms of the relation being given, all such relative knowledge is true knowledge of that independent character, together with the other term or terms of this relationship, determines this content of our relative knowledge. The concept, or conceptual interpretation, transcends this relativity precisely because what the concept comprises is this relational pattern in which the independent nature of what is apprehended is exhibited in experience. (MWO, pp. 172-173; final emphasis is mine)
Such is the story Lewis tells in MWO. The story he tells in AKV is slightly different. So, before I begin my analysis, let me conclude this section with a brief comment or two about `meaning'.
2.1.1 Meanings. In AKV, Lewis says «[m]eanings are not ... creatures of language.» Rather, they are «antecedent,» and sometimes even «independent» of language (AKV, p. 131). He makes a distinction between two general types of meaning [or intension]: linguistic meaning and sense meaning. The former, which roughly corresponds to the `pure concept', includes «the pattern of definitive and analytic relationships of the word or expression in question to other words and expressions.» The latter, which corresponds [more or less] to the `private concept', is the «criterion in mind» by which the application of a word or expression is determined (AKV, p. 131):
What we indicate by this phrase sense meaning is intension as a criterion of mind, by reference to which one is able to apply or refuse to apply the expression in question in the case of presented, or imagined, things or situations. (AKV, p. 133)
It is worth noting that linguistic meaning and sense meaning are «supplementary, not alternative.» They are separable only by «abstraction» (AKV, p. 133).
Still, of the two, sense meaning is more important. Here's why. «[C]onditions for determining applicability ... do not always exist ready-made.» Rather, such conditions need «to be sought out or created» (AKV, p. 136). The making -- and application -- of the requisite connections is the mind's contribution to experience. Hence, although we use language to convey thought, and pure concepts and linguistic meanings often constrain private ones, it is the private ones that ultimately must do the work when an agent cognizes. For example, though you and I may correctly use the term `cat', and my cat-labeling behavior is sufficient for you to ascribe to me the concept `CAT', in the end, there must be something about my cats-presentations that permits me to pick-out cats when the need arises. But while our expressions may be the same -- «Lo, a cat» -- the subjective, private, and idiosyncratic conditions for determining the applicability of `cat' need not. While the given remains [more or less] constant across each presentation, its interpretation, its «character as sign, its classification, and its relation to other things and to action are differently taken» by different people (MWO, p. 50). Thus, sense meanings arise from an interpretation of the the given via some activity of the mind. Such meanings aren't only alterable, they are relative to one's interest, action, or will (MWO, p. 51).
Contrary to appearances, I don't think Lewis is being inconsistent when he privileges pure concepts in MWO and sense meanings in AKV. My reasons for thinking this are offered below [Section 3.3]. And having belabored the explication of Lewis' theory -- which, at times, required quite a bit of interpretation -- my analysis can therefore be much more concise.
3. RESOLVING THE TENSIONS
My aim for this section is to identify a few of the tensions that have emerged from the preceding analysis. The task then will be to resolve them. I'll do so, for the most part, by turning Lewis upon himself. Still, given the nature of some of his commitments, <13>Foot note 1_13there are limits as to how far his theory can be rehabilitated.
3.1 Is Lewis' theory of mind plausible?
Lewis is an epistemologist, not a philosopher of mind. Epistemologists are concerned primarily with issues of justification. Philosophers of mind are concerned primarily with how minds work. Since `the mind' and `cognition' are among the notions shared by both disciplines, given that epistemology is a normative discipline, it is hardly surprising that Lewis' theory makes claims about how minds ought to work. After all,
[k]nowledge is not descriptive but a normative category: it claims correctness; mental states are classified as genuine knowing only on assumption of such correctness. Epistemology is not psychological description of such mental states, but is critique of their cognitive claim; the assessment of their veracity and validity, and the eliciting of those criteria by which such claim may be attested. (AKV, pp. 10-11; my emphasis)
The problem isn't merely that Lewis does describe psychological states, some of his descriptions of cognitive [or mental] activity are wildly implausible. For example,
I see something in the distance moving toward me, and believe it is my dog. This object moves closer and closer; I have more and more corroborating evidence; my belief becomes stronger and stronger. ... I know this is my dog. ... There has been here a series of cognitive apprehensions, differing from one another in degree, all the way from doubt to practical certainty. But even from the start there has been something entirely certain; namely, some visually apprehended content of sense. I could not well express these visually given data with any accuracy, but such relatively inexpressible content of experience was indubitable fact. From moment to moment, these visual data were increasingly clear and detailed; and increasingly adequate grounds for the judgment, «This is my dog.» Correspondingly, there was a growing conviction, from initial doubt to finally complete -- or nearly complete -- assurance. I have made successive inferences (so it would appear if I should analyze my successive apprehensions from the point of view of their cognitive validity)<14>Foot note 1_14 based on these successively given and successively more adequate data. The validity of the inference, in each case, is attested by certain rules, called the principles of probability or rules of induction. The data are empirical and certain, and the principles assure that the conclusion validly follows. If D, then probably P; so application of the principles tells me. And `D' is given. Therefore, probably P. Such is the general character of my cognition at each successive instant. (AKV, pp. 320-321; my emphasis in bold)
At each perceiving instant, do agents really make the sort of inferences that Lewis describes? Of course not. Do they make such inferences even for rather ordinary empirical beliefs such as `That's my dog'? Again, the answer is no. As such, the above passage<15>Foot note 1_15 is at odds with his reflective methodology [which is supposed to be an extension of commonsense]; it also stands in tension to the business of epistemology [which isn't to provide «psychological descriptions» of mental states]. Thus, if Lewis is claiming that cognizers actually go through the various stages of interpretation, prediction, test, and the like, then it is his theory that, as it were, goes beyond the pale of plausibility.
How might this tension be resolved? The answer, I think, is implicit in his qualification above. Namely, if epistemology isn't in the business of explaining actual mental processes [which would be a causal story], but it is rather in the business of prescribing justificatory practice, then his psychological descriptions are not meant to be accounts of actual cognitive processing. Lewis seems to agree. Following the above passage Lewis writes: «Both verbal and mental economy, and the necessity of decision, require us to think and act in terms of what approximates to complete assurance, omitting the strictly called-for qualification» (AKV, p. 321). Moreover, this very issue arose early in AKV: «The question is not so much ... whether the behavior was deliberately initiated through explicit appraisal and decision as whether it could have been and would have been if question of consequences and their desirability had been raised» (AKV, p. 8; my emphasis). As such, what really matters for Lewis is whether an agent could produce and justify empirical beliefs in the probabilistic manner he defends. The issue isn't whether they do so actually and continuously. His views on the mind, therefore, have been rendered all the more plausible.
3.2 Is all experience thick?
If Lewis' descriptions of psychological processing do not imply that minds are continuously cognizing, and if his theory is meant to prescribe how empirical statements could be justified [when queried or when the need otherwise arises], then there is no reason for him to maintain that conscious minds are constantly interpreting an ever-changing given.
While all knowledge-dependent action depends on evaluation, not all actions are knowledge-dependent. As noted earlier, Lewis' theory isn't meant to capture the process by which every action is selected. Instead, it is meant capture only those acts that are deliberate or purposive -- the sort of acts requiring «explicit prevision of consequences and evaluation of these» (AKV, pp. 6-7; my emphasis). Note the emphasis upon `explicit prevision', for «much of our own behavior,» as Lewis recognizes, are unconsidered responses -- actions for which we are responsible, yet which are «hardly ... instigated by explicit foresight and assignment of values» (AKV, p. 5). Thus, there is no need to claim that minds are continuously interpreting [and hence forming judgments about] the given.
Moreover, «esthetic apprehension» in particular, and `direct perception' in general, do not involve conceptualization.<16>Foot note 1_16 Rather, they occur when one apprehends the given, but makes no judgments about it. Though such `experience' functions as cues [and ultimately the «grounds»] for empirical judgments, `direct apprehension' per se, though sufficient to trigger habitual, unconsidered behavior, doesn't always involve interpretation. As such, not all experience is thick. And given our reliance upon habits and unconsidered responses, it doesn't need to be.
3.3 Concepts and meaning revisited
I left `meaning' [Section 2.3] having identified another tension in Lewis' philosophy;
viz., he privileges pure concepts in MWO, though he privileges sense meanings in AKV. Thus, which type of concept is doing the work when one conceptualizes, `pure concepts' or `private ones'? They both are: each fulfills a different, though essential, function. Still, it's pure concepts that are doing most of the work.
Pure concepts, recall, figure in communication as well as in the transmission of knowledge. They do so because the individuation of pure concepts depends on «common,» «sharable,» and «expressible» meanings -- the sort of meanings that are necessary for communication (MWO, p. 80). The individuation of private concepts depends on idiosyncratic meanings, meanings that «direct,» «nonsharable,» and purely individualistic (MWO, p. 81). And while there are objective standards for the ascription of pure concepts [above all, «congruity of behavior»], there aren't such standards for the ascription of private ones. Thus, since «speech is that part of behavior which is most significant of meanings and the most useful for securing human cooperation» (MWO, p. 90), and linguistic meaning and sense meaning are «supplementary» rather than «alternative» (AKV, p. 133), pure concepts [and hence `linguistic meaning'] are doing most of the work in Lewis' system. They even do some of the work when an agent cognizes. The remainder, as noted above, depends on `private concepts'.
Here's the rub: If what matters most in questions of knowledge is the justification of empirical statements -- but not their actual causal history -- then given the role of pure concepts in language-mediated thought, communication, and knowledge [common or otherwise], then there are reasons for de-emphasizing internal, subjective intensions in favor of external, objective ones. In fact, simplicity and commonsense dictate that we do so.
4. CONCLUSION
My purpose has been to explore and resolve some of the tensions between Lewis' theory of knowledge and his theory of mind. Doing so required that I sketch-out his foundationalist epistemology. Particular attention was directed toward the role of conceptualization [or interpretation] in the production of experience, action, and knowledge. I have shown that if Lewis isn't claiming that his epistemology describes actual psycho-epistemic processes, but rather what an agent could do to justify her empirical beliefs, then his theory of mind is rendered all the more plausible. Moreover, the social character of concepts [and meaning] considerably reduces the need for appealing to internal concepts when explaining why an agent does what she does. And finally, among his paradigm cases of cognitive behavior include paradigm cases of nonreflective action; not only do such actions account for most of our behavior [as Lewis himself notes], they don't require conceptualization.
Insofar as knowledge is wedded to language, concepts, albeit public ones, will remain a feature of any plausible epistemic story. Nevertheless [and possibly quite despite himself] Lewis has shown something about cognition that isn't often raised in polite philosophical circles. Namely, though concepts are important, we can do quite a bit without out them.<17>Foot note 1_17
Robert S. Stufflebeam
Washington University
Department of Philosophy, Campus Box 1073
One Brookings Dr., St. Louis, MO 63130-4899
Dept: (314) 935-6670
Fax: (314) 935-7349
e-mail: rob@twinearth.wustl.edu