The autobiography is interesting about Lewis’s early life. His father worked in a shoe factory and got into difficulties for his membership in the Knights of Labor. The young Lewis was brought up on Fabianism and Bellamy’s Looking Backward and heard and revered Gompers and Debs. He got to and through Harvard only by the skin of his teeth (in 1902 40 percent of Harvard students depended on what they could earn for themselves). By dint of waiting on tables Lewis got his A.B. in three years. By 1912, after various vicissitudes, he had obtained his Ph.D. And so on to the study of Principia Mathematica, the devising of his own alternative logical system, and work in the great chaotic accumulation of the papers of Charles Sanders Peirce.
The characteristic stoicism of his mode of self-description cannot conceal the enormous amount of work he got through in his early years. He picks out Josiah Royce as the teacher who most influenced him but from that point on has little to say about influences, apart from Peirce’s papers. In philosophy, as in life, he was a model of old-fashioned New England self-reliance.
All sides of his work receive full treatment in the essays. A strikingly dull piece on his contributions to the history of logic reminds us that for fifty years his Survey of Symbolic Logic was the best history of logic in English, indeed it was the only serious one until the publication of William and Martha Kneale’s more comprehensive Development of Logic. There is a very thorough and detailed essay on Lewis’s work as a creative logician by W. T. Parry which, by its references to a mass of logical work somewhat at the margin of current interest, draws attention to the considerable influence Lewis has exercised in this area.
Roderick Firth and Roderick Chisholm are good on Lewis’s epistemology. Chisholm has done more than anyone else to develop Lewis’s conception of the theory of knowledge as a normative discipline, an ethics of belief. He points out here that Lewis is a strict moralist in this region: for him there are no indifferent beliefs, any belief that is not justifed ought to be rejected. Firth, with customary exquisiteness, questions Lewis’s assumption that the ultimate foundation of empirical knowledge must be indubitably certain. E. M. Adams neatly points out the uncomfortable relationship, verging on incompatibility, that holds between, on the one hand, Lewis’s empiricist assumptions that the basis of empirical knowledge is subjective appearances and that the meaning of words is determined by the experiences with which they are connected and, on the other, his insistence that his view of our beliefs about material things is a realistic one. Realism implies that there is more to the material world than the appearances it presents to us; but how, on Lewis’s assumptions about meaning and experience, can the belief that this is so be expressed?
Lewis’s work in formal logic has suffered a curious fate. It was undertaken in order to provide an alternative to the system of Principia Mathematica which would accommodate within itself, as that system does not, the essential connectedness that we ordinarily suppose to obtain between the antecedent and the consequent, the if-clause and the then-clause, of a conditional statement. Everyone agrees that “if p then q” is false when p is true and q is false. In Russell’s extensional system that is all there is to “if p then q.” Provided that p is false or q is true, or both, it is true in Russell’s system that if p then q. On this view the truth or falsity of a conditional statement can be established simply by considering the truth or falsity of its constituents. It is not necessary to consider the specific meaning or content of the constituents.
It turns out that this rather meager interpretation of the type of statement that is essential to all reasoning confirms all the rules of inference which we accept in our more reflective moments and which we should want a logical system to preserve, even if it also underwrites as laws certain patterns of inference that are not intuitively acceptable. In particular it is a law of Russell’s logic that if p is false then it follows that if p then q is true, whatever p and q may be and however unrelated they are to each other as regards their content. It also follows that if p then q is true if q is true. whatever p and q may be. These “paradoxes of material implication” impelled Lewis to construct a system which excluded them. But most logicians have accepted them as a reasonable price to pay in oddity for the immense manipulative benefits that accrue from the extensional interpretation of “if—then.” What such an interpretation makes possible is a purely mechanical way of deciding on the logical validity of patterns of inference in which conditional statements occur and on the consistency of sets of statements some of which are conditional.
So, despite Lewis, extensionalism generally prevails in logic. The result may be a bit of a caricature of our ordinary ways of reasoning but it contains what is essential and is vastly fertile from the point of view of formal construction, a liberating simplification. But Lewis’s approach (one should not say “system,” since he devised several, overlapping ones) has not been abandoned. It has found a new employment of an interesting and important kind: as modal logic, concerned with the rules governing the relations of statements containing such words as “must,” “may” and “can,” “necessarily” and “possibly,” “ought” and “oughtn’t.” For example, if p is necessary then p is possible; but if p is not impossible it does not follow that p is necessary. Lewis’s logic has not been accorded the position its creator designed for it, that of representing the rules of ordinary assertion, about what is or is not the case. But a more specialized task has been found for it, that of specifying the rules governing assertions qualified by the pervasive range of modal terms, assertions about what must or should or can be the case.
As an epistemologist Lewis’s position always rested on a fundamental distinction between knowledge that is necessary and a priori, independent of what actually happens in the world, and contingent, empirical knowledge. For Lewis, necessary truth is determined by the relations between the meanings of the terms we use, while all contingent statements must be somehow reducible to the incorrigible deliverances of immediate experience. A full exposition of this logical dualism, which has been fundamental to much of the analytic philosophy of the past forty years, is to be found in Mind and the World-Order (1929), Lewis’s first great book on the theory of knowledge. It anticipated many of the doctrines of the logical positivism of the 1930s but presented them in a more perceptive, perhaps because less polemically combative, way. There is still no better defense of the view that all necessary propositions are analytic than the one to be found in that book.
His account of empirical knowledge received its full development in his long Analysis of Knowledge and Valuation in 1946. It is the culmination of a long tradition of more or less radical empiricism in English-speaking philosophy; prepared for by Locke and Berkeley, salvageable from the skeptical desperation of Hume, set out plainly in outline form by Mill, broken into a number of brilliant, hasty fragments by Russell. Lewis goes at the problem of accounting for our knowledge of, or at any rate justified belief in, a world of material objects existing independently of us, on the basis of the momentary and discontinuous impressions of our senses with a cool head, in great detail, and with a solid readiness to follow the argument wherever it may lead him.
It leads him a long way, into investigations of induction (since our beliefs about the material world are in his view derived by generalizing from the regularities disclosed in our sense experience), probability (since no such belief is ever certain or finally established), and the nature of counterfactual conjectures (since all beliefs about what we are not here and now observing assert, he contends, what we should be observing if certain conditions were to obtain which in fact do not).
The problem of our knowledge of the external world is perhaps more calculated than any other to inspire the non-philosopher with an amused conviction of the essentially dotty and fantastic character of epistemology. This results from incautious formulation of the problem as that of justifying our belief in a world that exists independently of us. As Reid too mildly remarked, that belief “is older and of more authority than any argument of philosophy.” Serious doubt of it is more a sign of mental derangement than of conspicuous intellectual penetration. In fact, the point of the problem is not whether the belief is justified; it is, rather, what the justification actually is.
Lewis’s handling of the complicated network of abstractions involved is an exemplary display of patient and systematic rationality. It amounts to a demonstration that, if the initial presumption that empirical knowledge must have infallibly certain foundations is correct (and, of the large majority of epistemologists who have supposed this, Lewis is almost alone in arguing for it instead of taking it for granted), our belief in a material world must be justified in his way. For it is plain that no beliefs about material objects have the required kind of certainty and such beliefs must, therefore, be inferred from beliefs about private experience which are certain in the required way.
In the last twenty years the theory of empirical knowledge has been considerably transformed by the abandonment of the presumption about empirical knowledge I have mentioned. Various things have contributed to the change. One was G. E. Moore’s persistent efforts to distinguish certainty, properly so called, from the much narrower concept of incorrigibility with which Lewis and those of his persuasion identified it. Another is the recognition that it does not follow from the fact that a belief is less than certain, in either sense, that it must therefore be the outcome of an inference. Since this is so our less precarious beliefs about material things, the ones we should unreflectively regard as direct reports of perception, are fit to serve as the foundations of our knowledge of empirical fact.
Lewis’s reason for thinking that only the certain is uninferred is that probability is always relative to evidence. But that does not, as Lewis thought, entail that all less than certain beliefs must be inferred. Any reason there is for thinking that probability is relative to evidence is a reason for thinking that certainty is relative to evidence too. The type of “direct realism” propounded by many current theorists of perception sees no logical difficulty in the admission that fallible beliefs about material things can be direct and uninferred reports of our perceptions just as well as the infallible beliefs about sense impressions which theorists of knowledge from Descartes to Lewis have held to be the only possible occupants of this epistemologically basic status.