Traditionally, the notion of experience played two essential but distinguishable roles in the theory of knowledge: Experience functioned as the rock bottom of all empirical justification and as the key to the link between our beliefs and reality. Experience was the ultimate foundation of all empirical knowledge, and it was also our only guarantee that what we think we know has what Kant called “objective validity.” In this way our views of experience affected our views both on foundationalism and on realism.

Around 1930, a number of positivists began to question the adequacy of the foundationalist standpoint that had inspired much of traditional philosophy, and they detected a source of error in a certain doctrine of experience.

In earlier chapters we saw how the domains of mathematics and logic had been taken care of (albeit differently) by Wittgenstein and Carnap. Most positivists were willing to go along with one or the other of them on those topics; but they were far more interested in empirical knowledge. Once again, as in the case of mathematics, the leading questions were semantic and foundational: What does science say, and what kinds of reasons do we have for believing what it says? These two questions were the kernel of what was misleadingly called at the time the “problem of induction.” A variety of ways of looking at it had been taking shape for decades. Their proponents finally confronted one another in Vienna in the early 1930s. The showdown was, as usual, inconclusive, but it greatly sharpened the conflicting positions and helped reveal their respective virtues and vices. Moreover, beneath the wide variety of responses, one may detect further evidence of the methodological shift toward the transcendental approach adumbrated in Chapter 10.

On Christmas Eve, 1984, Alberto declared that a “good penultimate draft” of his book would be finished by the end of the year. The day after Christmas he became ill, and in the early morning hours of December 30, he died. The typescript Alberto left behind was, indeed, almost completed: The research had been finished and the arguments and theses were already in place; all but the introduction and the last chapter had been written in full, and extensive notes for those had been drafted. Some portions of the typescript had already been carefully crafted, and the intended shape of much of the rest was clear. In many places Alberto's dry wit showed through – you could almost see behind the prose the tilt of his head, the lopsided smile, the brief twinkle in his eye. With the help of a number of people, that typescript has been readied for publication. Using the notes Alberto left, I have completed the introduction and final chapter. Repetitions, digressions, and minor errors have been removed, theses and arguments delineated more clearly, grammar corrected, prose smoothed – all, I hope, without disturbing the twinkle. The result is not what Alberto would have produced, but perhaps it is something he would have found acceptable. He had intended to write a conclusion, discussing some of the implications of his study for contemporary philosophy. I have not attempted to write that conclusion.

It isn't easy to decide whether Wittgenstein should be included among the members of the semantic tradition or among its most ferocious enemies. On the surface, at any rate, Wittgenstein's problems and techniques were those of the semanticists; beneath the surface, however, things are less clear.

In the second half of the nineteenth century Kant was regarded with enormous respect by the most reasonable part of the German philosophical community. This attitude was inspired partly by the intrinsic value of Kant's work, but partly also by the dramatic contrast between his sober, reasonable approach and the romantic flights of those who followed him on the German scene.

Early in the twentieth century, that periodic German revolt against reasonableness started once again; and once again Kant compared favorably with the Schelers and the Heideggers that began to capture the imagination of the philosophical masses. Little wonder that those more inclined to take things calmly organized a zurück zu Kant movement of their own, emphasizing the scientific aspect of Kant's work, much in the fashion of Helmholtz, Zeller, Riehl, and also the Marburg school. All of the leaders of Viennese positivism began their philosophical path as neo-Kantians of this kind. Schlick was the first among them.

With Schlick, we are back in the world of Kantian questions and semi-Kantian answers, so different from that of the semantic tradition. The little he knew of that tradition before Vienna was primarily its hypostatic version in the writings of Brentano's students and, to a lesser extent, in Russell's. Like so many others before and after him, Schlick concluded that a nonpsychologistic theory of meaning was incompatible with empiricism.

Logicism and the foundational crisis

In 1900 Russell underwent the one event in his intellectual life that he was willing to characterize as a “revolution”: He met Peano and was struck by the capacity of Peano's work to shed light on the philosophical nature of mathematics. It was at this time that Russell conceived one of his most fruitful ideas, the logicist project.

Peano had identified a notational system, or a cluster of concepts, that seemed to have enormous expressive power. The hope was that it could be used to express all of mathematics, and Peano's school had been working for years at rewriting different fragments of mathematics in their peculiar notation. Russell suggested that its basic concepts might be reduced to purely “logical” notions, in an as yet undisclosed sense of that word, and that perhaps all the assumptions one needed were those of logic, whatever they might be.

We now know what Carnap knew early in 1931 when his thoughts on Wittgensteinian grammar and metamathematics finally converged. The product of that fusion was The Logical Syntax of Language, a book that displayed a new theory of mathematical knowledge and offered it as a model for epistemology.

The single most decisive stimulus to the train of thought that culminated in that book was Gödel's 1931 paper on the incompleteness of formal theories of arithmetic, “Uber unentscheidbare Sätze,” which contains a discovery whose intellectual significance was comparable to Einstein's relativity theory.

The relevance of Gödel's conclusions to Hilbert's program are widely acknowledged. But its relevance to what one might call “Wittgenstein's program“ was no less decisive.

Wittgenstein said that “there are no true a priori propositions” (Lectures, 1930–32, p. 13; Tractatus, 2.225), Carnap tirelessly denied the synthetic a priori, and Schlick went so far as to define empiricism as the rejection of synthetic a priori knowledge. In spite of all this there can be no doubt that the major contribution of Wittgenstein's and Carnap's epistemologies in the early 1930s was their interpretations of all a priori knowledge, both analytic and synthetic. Their theories of philosophical grammar and of logical syntax may well be regarded as the first genuine alternatives to Kant's conception of the a priori.

By the beginning of the twentieth century, no philosopher of consequence was satisfied with Kant's solution to the problem of the a priori. Many had come to understand far better than Kant what was involved in particular instances of a priori knowledge; but efforts to build general accounts of what that form of knowledge was and the way it was grounded were far less successful. In Part I we examined some of the alternatives to Kant's theory that had been put forth by Kantians or anti-Kantians.

Quite apart from its contributions to logic and in uneasy alliance with them, Carnap's LSL contains a radically new approach to the philosophy of mathematics that he and others would soon take as a model for epistemology as a whole. Carnap's attitude toward philosophical considerations was roughly that of the scalded cat toward boiling water. He was second to none in his ability to state clearly and argue cogently formallevel philosophical issues; but the deeper and less obviously formal those issues become, the harder it is to find either a clear statement or an argument for Carnap's position.

The primary topic of this book is a decade in the philosophical life of what might loosely be called Vienna. Between 1925 and 1935 in the neighborhood of Vienna, the usually sluggish step of the Spirit suddenly quickened as some of his most enlightened voices started talking to one another. Wittgenstein, Tarski, Carnap, Schlick, Popper, and Reichenbach were, perhaps, no wiser than some of their contemporaries, but circumstances led them to interact during that decade, and the result of that dialogue deserves our attention.

When I started writing this book, I intended to explain in the preface that this was the history of epistemology since Kant, the way Carnap would have written it had he been Hegel. Since then I have come to think that while the Spirit may not be malicious, he is certainly forgetful. In Vienna he took a decisive step forward on the subject of the a priori; but he also moved sideways and backward on other crucial matters. Most of his erratic behavior could have been avoided had he been aware of some of his achievements in the preceding century. He may, perhaps, be excused since his best insights were due to the least noticeable of his voices.

Within the field of epistemology one may discern three major currents of thought in the nineteenth century: positivism, Kantianism, and what I propose to call the semantic tradition. What distinguished their proponents primarily was their attitude toward the a priori.

Modern Continental philosophy had always maintained close ties with scientific developments. In Kant the link became so close that the whole doctrine of the a priori had been motivated largely by a datum that had emerged from the sciences – an allegedly transparent feature of geometry, arithmetic, and the calculus that demanded philosophical explanation. Kant's successors in the nineteenth century were of two types: those who wanted to check whether what he said about the a priori sciences was true and those who didn't really care. The latter embraced his Copernican turn for “metaphysical” reasons. The former, by and large, devoted a great deal of time to an analysis of mathematical knowledge. As a result, their more gullible colleagues tended to look on them as lowlevel mathematicians trying to make a reputation in philosophy. “Mathematica sunt, non leguntur” is what Frege once guessed most philosophers would say about his writings. He was right. The same could have been said of the major writings of the semantic tradition.

The semantic tradition may be defined by its problem, its enemy, its goal, and its strategy. Its problem was the a priori; its enemy, Kant's pure intuition; its purpose, to develop a conception of the a priori in which pure intuition played no role; its strategy, to base that theory on a development of semantics.

In the early stages of their intellectual development, the founding fathers of logical positivism were at least as close to Kant as they were to classical positivism. We have just seen that Schlick tried to argue for the essential unity of the epistemological approaches of Kant and Avenarius, and we have also seen the strong Kantian inspiration of his Allgemeine Erkenntnislehre. Around 1920 Reichenbach judged Kantianism to be the most appropriate standpoint for interpreting the theory of relativity. At about the same time, Carnap was writing a Kantian-style meditation on the nature of space, and he would devote most of the remainder of the decade to that most Kantian of projects, the development of a theory of constitution (Chapter 11). There can be little doubt that logical positivism started as a branch of neo-Kantianism differing from its rivals in that movement only in its concern for clarity and its appreciation of science as a model for epistemology. Yet these differences would, in the end, make all the difference.

The 1920s were a soul-searching decade for the logical positivists. During those years they struggled with the Kantian language in which they had chosen to couch their views of science and knowledge. Very slowly they came to realize how inadequate that language was to the message they were trying to convey.