A major American pragmatist educated at Harvard, Lewis taught at the University of California from 1911 to 1919 and at Harvard from 1920 until his retirement in 1953. Known as the father of modern modal logic and as a proponent of the given in epistemology, he also was an influential figure in value theory and ethics.
In "The Pragmatic Element in Knowledge", Lewis extended his pragmatism about the a priori to the theory of knowledge. Here, following Peirce and Royce, he identifies three elements in knowledge which are separable only by analysis: the element of experience which is given to an agent, the structure of concepts with which the agent interprets what is given, and the agent's act of interpreting what is given by means of those concepts. The distinctively pragmatic character of this theory lies both in the fact that knowledge is activity or interpretation and that the concepts with which the mind interrogates experience reflect fallible and revisable commitments to future experiential consequences. Knowledge is an interpretation of the experiential significance for an agent with certain interests of what is given in experience; a significance testable by its consequences for action.
A priori truth is independent of experience because it is purely analytic of our concepts and can dictate nothing to the given. The formal sciences depend on nothing which is empirically given, depending purely on logical analysis for their content. So a priori truth is not assertive of fact but is instead definitive. There is logical order arising from our definitions in all knowledge. Ordinarily we do not separate out this logical order, but it is always possible to do so, and it is this element which minds must have in common if they are to understand each other. As Lewis puts it, "At the end of an hour which feels very long to you and short to me, we can meet by agreement, because our common understanding of that hour is not a feeling of tedium or vivacity, but means sixty minutes, one round of the clock…". In short, shared concepts do not depend upon the identity of sense feeling, but in their objective significance for action.
The concept, the purely logical pattern of meaning, is an abstraction from the richness of actual experience. It represents what the mind brings to experience in the act of interpretation. The other element, that which the mind finds , or what is independent of thought, is the given. The given is also an abstraction, but it cannot be expressed in language because language implies concepts and because the given is that aspect of experience which concepts do not convey. Knowledge is the significance which experience has for possible action and the further experience to which such action would lead.
Mind and the World Order
Lewis first major book, Mind and the World Order (MWO) develops these results in three principal theses: first, a priori truth is definitive and offers criteria by means of which experience can be discriminated; second, the application of concepts to any particular experience is hypothetical and the choice of conceptual system meets pragmatic needs; and third, the susceptibility of experience to conceptual interpretation requires no particular metaphysical assumption about the conformity of experience to the mind or its categories. These principles allow Lewis to present the traditional problem of knowledge as resting on a mistake. There is no contradiction between the relativity of knowledge to the knowing mind and the independence of its object. The assumption that there is, is the product of Cartesian representationalism, the 'copy theory' of thought, in which knowledge of an object is taken to be qualitative coincidence between the idea in the mind and the external real object. For Lewis knowledge does not copy anything but concerns the relation between this experience and other possible experiences of which this experience is a sign. Knowledge is expressible not because we share the same data of sense but because we share concepts and categorial commitments.
All knowledge is conceptual; the given, having no conceptual structure of its own, is not even a possible object of knowledge. Foundationalism of the classical empiricist sort is thus directly precluded. Lewis's task for MWO is in effect a pragmatic solution to Hume's problem of induction: an account of the order we bring to experience which renders knowledge possible but makes no appeal to anything lying outside of experience. Prefiguring contemporary externalist accounts of representation, Lewis argues that both representative realism and phenomenalism are incoherent. Knowledge as correct interpretation is independent of whether the phenomenal character of experience is a "likeness" of the real object known, because the phenomenal character of experience only receives its function as a sign from its conceptual interpretation, that is, from its significance for future experience and action. The question of the validity of knowledge claims is thus for Lewis fundamentally the question of the normative significance of our empirical assessments for action.
Lewis argued that our spontaneous interpretation of experience by way of concepts that have objective significance for future experience constitutes a kind of diagnosis of appearance . If we could not recognize a sensuous content in our classification of it with qualitatively similar ones which have acquired predictive significance in the past, interpretation would be impossible. Despite the fact that such recognition is spontaneous and unconsidered it has the logical character of a generalization. To recognize an object — "this is a round penny" — is to make a fallible empirical claim, but to recognize the appearance is to classify it with other qualitatively similar appearances. The basis of the empirical judgment lies in the fact that past instances of such classification have been successful. Our empirical knowledge claims are dependent for their justification upon this body of conceptual interpretations in two ways. First, the world, in the form of future events implicitly predicted (or not) by our empirical judgments, will confirm or disconfirm those judgments: all empirical knowledge is thus merely probable. But secondly, the classification of immediate apprehensions by way of concepts justifying particular empirical judgments is itself generalization even when those concepts have come to function as a criterion of sense meaning. Concepts become criteria of classification because they allow us to make empirically valid judgments, and because they fit usefully in the larger structure of our concepts.
This structure, looked at apart from experience is an a priori system of concepts. The application of one of its constituent concepts to any particular is a matter of probability but the question of applying the system in general is a matter of the choice of an abstract system and can only be determined by pragmatic considerations. The implications of a concept within a system become criteria of its applicability in that system. If later experience does not accord with the logical implications of our application of a concept to a particular, we will withdraw the application of the concept. Persistent failure of individual concepts to apply fruitfully to experience will lead us to readjust the system as a whole. Our conceptual interpretations form a hierarchy in which some are more fundamental than others; abandoning them will have more radical consequences than abandoning others. Lewis's account of inquiry offers both a non-metaphysical account of induction and an early version of the so called 'theory-ladenness of observation terms'. There is no need for synthetic a priori or metaphysical truths to bridge the gap between abstract concepts in the mind and the reality presented in experience. Lewis offers a kind of 'Kantian deduction of the categories' providing a pragmatic vindication of induction but without Kant's assumption that experience is limited by the modes of intuition and fixed forms of thought. Without a system of conceptual interpretation, no experience is possible, but which system of interpretation we use is a matter of choice and what we experience is given to us by reality. The importance of the given in this story is its independence . Our conceptual system can at best specify a system of possible worlds; within it the actual is not to be deduced but acknowledged. In short, Lewis's theory of knowledge in MWO is a pragmatic theory of inquiry which combines rationalist and naturalistic elements to make knowledge of the real both fallible and progressive without recourse to transcendental guarantees.
The Analysis of Knowledge and Valuation
In 1946 The Analysis of Knowledge and Valuation (AKV) was published, and Lewis was awarded the Edgar Pierce Professorship at Harvard, the chair which had been held by Perry and would be held by Quine after Lewis. AKV was the most widely discussed book of its day.
The pragmatic psycho-biological model of inquiry which Lewis adopted from Peirce and James is even more visibly a part of AKV than it was in MWO. Knowledge, action and evaluation are essentially connected animal adaptive responses. Cognition, as a vital function, is a response to the significance which items in an organism's experiential environment have for that organism. Any psychological attitude which carries cognitive significance as a response will exhibit some value character of utility or disutility which can ground the correctness or incorrectness of that response as knowledge. Cognitively guided behavior is a kind of adaptive response, and the correctness of behavior guiding experience, to the extent that it carries cognitive significance, depends simply on whether the expectations lodged in it come about as the result of action. Meaning, in this sense is anticipation of further experience associated with present content and the truth of it concerns the verifiability of expected consequences of action. It is because of this that sense-apprehension is basic and underlies other forms of empirical cognition. Perceptual cognition involves a sign-function connecting present experience and possible future eventualities grounded in some mode of action which, pervading the experience in its immediacy, gives it its cognitive content.
The signifying character of the expectancies lodged in immediate experience is enormously expanded by the web of concepts we inherit as language users. Lewis did not, however, identify meaning with linguistic signs. Linguistic signs are secondary to something more basic in our experience which we share with animals generally and which occurs when something within our experience stands for something else as a sign. When the cat comes running because she hears you opening a can and takes it as a sign of dinner, she is responding to the meaning of her experience. While this meaning is independent of whether or not you are opening a can of cat food her expectation will be confirmed if the can contains cat food and disconfirmed if it doesn't.
Meaning in this sense of empirical significance could only be available to a creature who can act in anticipation of events to be realized or avoided. Accordingly, the possible is epistemologically prior to the actual. Only an agent, for whom experience could have anticipatory significance, could have a concept of objective reality as that which is possible to verify or change. In addition to meaning as empirical significance Lewis distinguished the kind of meaning involved in the apprehension of our concepts. A definition represents a mode of classification, and although alternative modes of classification can be more or less useful, classification cannot be determined by that which is to be classified. Knowledge of meanings in this sense is analytic.
In AKV, Lewis distinguishes between four modes of meaning: (1) the denotation or extension of a term is the class of actual things to which the terms applies; (2) the comprehension of a term is the class of all possible things to which the term would correctly apply; (3) the signification of a term is the character of things the presence or absence of which determines the comprehension of the term; (4) the intension of a term is the conjunction of all the other terms which must be correctly applicable to anything to which the term correctly applies. A proposition is a term capable of signifying a state of affairs; it comprehends any possible world which would contain the state of affairs it signifies. The intension of a proposition includes whatever the proposition entails and thus comprises whatever must be true of any possible world for that proposition to be true of it.
Intentional and denotational modes of meaning are two aspects of cognitive apprehension in general, the denotational being that aspect of apprehension which, given our classifications, is dependent upon how experience turns out, and the intentional being that aspect of apprehension which reflects the classifications or definitions we have made and is thus independent of experience. Our choice of classification is essentially pragmatic, however, so what may count as an empirical matter in one context may count legislatively in another, generalizations may be corrected by future experience and our definitions replaced on the grounds of inadequacy. The analytic element in knowledge is indispensable because unless our intensions are fixed our terms have no denotation, but nothing determines how we shall fix our intensions save the superior utility of one set of terms over others.
While intensional meaning is primary for him, Lewis distinguishes between two different ways in which we can think of it. First, linguistic meaning is intension as constituted by the pattern of definitions of our terms. Secondly, sense meaning is intension as the criterion in terms of sense by which the application of terms to experience is determined. Sense meaning is more fundamental. Learning involves the extension of generalizations to unobserved cases and correlatively recognizing in new experiences the correct applicability of our terms. The sense meaning of a term is our criterion for applying the term correctly. In a thought experiment anticipating Searle's "Chinese Room," Lewis imagines a person who somehow learns Arabic using only an Arabic dictionary thus learning all the linguistic patterns in the language. This person would grasp the linguistic meanings of all the terms in Arabic but might nonetheless not know the meaning of any of the terms in the sense of knowing their application to the world. The language would remain a meaningless and arbitrary system of syntactic relationships. Linguistic meaning is nonetheless central in communication because what can be shared is conceptual structure. Understanding between two minds depends not on postulated identity of imagery or sensation but on shared definitions and concepts.
The validation of empirical knowledge has two dimensions, its verification and its justification. Verification is predictive and formulates our expectations for verification or falsification. Justification looks to the rational credibility of those expectations prior to their verification. In the acquisition of knowledge these dimensions support each other. The warrant which our present beliefs have is shaped by the history of past verifications of similar beliefs. Reflection on the warranted expectancies in our present beliefs leads us to formulate new generalizations and normative principles we can subject to tests. The common stock of concepts in our language embeds such principles and empirical generalizations in the intensions of terms. As a result our use of terms decisively shapes what is warranted and verifiable for us.
Lewis distinguishes between three classes of empirical statements. First, there are what he calls expressive statements which attempt to express what is presently given in experience. An ordinary perceptual judgment, say seeing my cat by the fridge, outstrips what is presently evident. This added content is carried by the intensions of the concepts in the judgment insofar as they convey the expectancies found in the experience. These expectancies, although partly a function of past learning and knowledge of the intension of terms, are simply given in the experience, they are the part we do not invent and cannot change but merely find. Lewis suggests that we can use language expressively to capture this presentational content by stripping our meaning of its ordinary implication of objective content. Secondly, there are statements which formulate predictions. The judgment that if I do action A the outcome will include E, where E indicates an aspect of experience expressively characterized, is one which can be completely verified by putting it to the test. Upon acting the content E will either be given or it will not. Lewis calls empirical judgments of this sort terminating judgments. Finally, there are judgments which assert the actuality of some state of affairs. Although they can be rendered increasingly probable by tests, no set of eventualities envisioned can exhaust their significance. Lewis calls these judgments non-terminating because there are indefinitely many further tests which could, theoretically speaking, falsify the prediction and any actual verification can be no more than partial.
The ground of empirical judgments is past experience of like cases. At bottom those experiences have a warrant-producing character for a particular response because of the directly apprehended qualitative character of the signal combined with the expectations due to similar experiences in the past. In short, an empirical judgment is justified by its relation to past experiences of like cases. The warrant producing character of those experiences for a particular judgment depends upon the recognition of the presentation as classifiable with other qualitatively similar appearances as significant of future experience, and the character of the passages of experience attending past instances of the judgment. Epistemic warrant at its bottom level is the animal's recognition of future objectivity lodged in present experience; present experience is a sign of experience to come. A multi-storied interpretive structure of concepts is built upon this adaptive responsiveness. Concepts become criteria of classification because they allow us to make empirically valid judgments, and because they fit usefully in the larger structure of our concepts. The structure, viewed apart from experience, is an a priori system of concepts, but looked at in terms of experience it is a network of sense meanings. The concept of probability plays a more prominent role in AKV than it does in MWO, but it is not a role of a different kind.
Perceptual knowledge has two aspects: the givenness of the experience and the objective interpretation which, in light of past experience, we put on it. But these are both abstractions and only distinguishable by analysis. What is given in experience as spontaneously arising expectancies is already conceptually structured, to recognize the given is to classify it with qualitatively similar cases and that recognition, although spontaneous, has the logical character as a generalization. The system of concepts within which our judgments are formulated and the pyramidal structure of empirical beliefs which intend a set of possible worlds of which ours is but one, by themselves suggest a coherence theory of justification. But here, as in MWO, Lewis resists this idealist alternative. Lewis takes the given to be essential for a series of interrelated reasons. Mere coherence of a system of statements does not even give meaning; the student of Arabic mentioned earlier does not know what any of the terms mean and cannot even use a statement to express a judgment. The given thus plays the role of fixing what beliefs mean because it lodges the actual world among the various possible worlds which are compatible with my knowledge: whichever world I am in it is this one. A merely hypothetical system of congruent and consistent statements could be fabricated out of whole cloth, as a novelist does, but however richly developed, the congruence and coherence of the system would be no evidence of fact at all. Independently given facts are indispensable and they are the actually given expectancies whose objective intent we then can evaluate for their mutual congruence and coherence.
Lewis's emphasis on the given has been taken by many contemporary philosophers to be an instance of classical foundationalism. As we saw in the discussion of MWO Lewis considered the very idea of sense data to be incoherent. There is, however, a debate about whether his views changed between that book and AKV. Christopher Gowans (in "Two Concepts of the Given in C.I. Lewis, Realism and Foundationalism") has argued that Lewis had two different conceptions of the given but failed to recognize the difference between them. On this view, while Lewis was an anti-foundationalist in MWO he embraced foundationalism in AKV and his later thinking. Determining Lewis's position is, of course, a matter of interpretation. I think that a non-foundationalist position is dictated by the larger structure of his thought. He was certainly not a foundationalist in the British empiricist sense of the word.