In 1933, while in his mid-twenties and having just completed his doc- toral thesis, Quine travelled to Prague to visit Rudolf Carnap.2 Quine
knew of Carnap’s Der Logische Aufbau der Welt (1928; translated as The Logical Structure of the World in 1967), and knew that Carnap was working on his Logische Syntax der Sprache (1934; translated as The Logical Syntax of Language in 1937). In science, these were the early years of Relativity Theory, Quantum Theory, and the breakthroughs in mathematical logic due to Frege, Russell and others.3 New dawns
Vienna Circle, a group including Carnap, known for proposing Logical Positivism (Logical Empiricism, as Carnap preferred). It is difficult to recapture the excitement that all of this occasioned in those involved. In the Foreword to the Logical Syntax, Carnap wrote:
Philosophy is to be replaced by the logic of science – that is to say, by the logical analysis of the concepts and sentences of the science of the sciences, for the logic of science is nothing other than the logical syntax of the language of science. (Carnap, 1937, p. xiii)
Yes – philosophy is to be replaced by the logic of science. (Carnap was well-aware of the hackles this would raise, applying his ideas to none other than Heidegger, for one, purporting to show that some of Heide- gger’s sentences were literally nonsense.) For Carnap, then, the key question for philosophy is just this: What does one have to assume in order to do science, to explain the ways of the world?
In particular, Carnap aimed to show that there is no need for the synthetic a priori, for Kantian a priori intuition. The idea of what Carnap called phenomenological reductionism was that sensory experience plus logic is all that required to account for genuine knowledge, namely sci- ence. It’s an old idea, but the new and much more powerful symbolic logic of Russell and Frege promised that such a thesis might be demon- strated precisely rather than merely sketched in the manner of Hume. But – leaving sensory experience aside for the moment – of what exactly does knowledge of logic consist? Neither Russell nor Frege could answer that question satisfactorily; both held that the basic principles of logic must simply be assumed by science. Carnap’s Logical Syntax provided a more sophisticated answer: by developing an exact method of con- structing logic in the form of ‘sentences about sentences’, Carnap held that knowledge of logic is ultimately based upon our acceptance of certain conventions for the use of language. Just as one plays chess only if one follows the rules of chess, to speak a language is to be bound by certain rules, the analytic truths of the language. Our knowledge, for example, that If it’s raining and it’s night time then it’s raining, depends simply on our decision to speak the language we do. The domain of the a priori is decided pragmatically, not theoretically.
Quine’s doubts over Carnap’s scheme surfaced as early as 1935 in ‘Truth by Convention’ (WP, pp. 77–106). First, especially in view of
Carnap’s toleration for different languages with different conventions, what exactly is the basis for distinguishing the conventional from the non-conventional? Why not declare the principles of thermodynamics, say, to be conventions? Second, it is hard to see how the truths of logic can themselves be matters of convention. If we say: ‘It is a convention of language governing “if-then” that if we are given “if p then q” and “p” as premises, then we may infer “q”’, we are still using a statement which involves if-then in order to state the supposed convention. One might say with Carnap that such things are stated from outside the language, in a ‘metalanguage’, but that only raises the same question with respect to the metalanguage.
After World War Two, Quine generalized and sharpened his attack on Carnap’s philosophy in his ‘Two Dogmas of Empiricism’. It’s vital to keep the foregoing background in mind in reading Quine’s discussion; Quine’s central criticism of the notion of analyticity was not that no such notion could be coherent – in fact he would later define his own notion of analyticity – but analyticity has nothing like the epistemological importance claimed for it.4 Quine discusses various definitions, explan-
ations or explications of the notion that might be advanced, making use of such notions as synonymy, definitions, necessity, empirical content, verification conditions and so on. We will pass over the details and instead highlight two fundamental points. First, Quine’s most general claim as regards the list of possible explanations and analyticity itself is simply that the situation is ‘like a closed curve is space’ (FLPV, p. 30) – if one is dissatisfied with any of them, then that dissatisfaction will carry over to the rest. Carnap and others were trying to account for the a priori in terms of some such notions; they assumed that such a frame- work must be presupposed before any empirical discoveries can be made and interpreted. It is because of this that Quine’s second main point is vital to the force of the first: Quine’s crucial contention is that there is no need for such a presupposition.
In two pages at the middle of section 5 of ‘Two Dogmas’, Quine points out that in Carnap’s scheme, at a critical stage of the reduction of state- ments about physical objects to statements about sensation, it is neces- sary to define the term ‘is at’ as it appears in statements of the form q is at <x, y, z, t> where q is a sensory quality, x, y, z, are spatial coordinates, and t a time. Such a definition is needed to effect the transition from inner sensory experience to statements about the external world as a matter of
meaning. The trouble is that to be at all plausible, there has to be a certain looseness in that notion, for among other things we sometimes make mistakes in perception – about where q ‘is at’, for instance. Carnap’s solu- tion is to impose certain principles whereby sensory qualities are to be assigned to space-time in such a way as to maximize simplicity, continuity and other features. This presumably will take care of such cases as a ref- eree’s finding upon reflection that the ball could not possibly have crossed the line despite its having appeared to; the constraints tell him to rescind his original judgement. But those constraints do not tell us how to reduce a single statement of the form q is at <x, y, z, t>; at most they tell us how, given a whole collection of statements about immediate experience, we should assign truth-values to those statements so as to yield the right sort of overall world. In the case just described, an assignment of sense-quali- ties to space-time points may have accorded well with experience up to a point, but must now be revised in order best to satisfy the principles gov- erning such assignments.
So the single greatest attempt to make out phenomenological reduc- tionism failed. But now, memorably, Quine claims that there is no need for the key idea on which such an idea depends. Considerations of sim- plicity, continuity and the like are appealed to in generating theories of the world – even the most mundane ones – but they apply only holistic- ally, that is, only to whole theories, or otherwise large collections of statements. Here then is the famous Quinean web of belief: rather than absolute rules telling us how to respond to particular experiences, we have various weightings of various factors leaving room for choice, and which may themselves be re-evaluated:
The totality of our so-called knowledge or beliefs, from the most casual matters of geography and history to the profoundest laws of atomic physics or even of pure mathematics and logic, is a man-made fabric which impinges on experience only along the edges. Or, to change the figure, total science is like a field of force whose boundary conditions are experience. A conflict with experience at the periphery occasions readjustments in the interior of the field. Truth values have to redistributed over some of our statements . . . [T]he total field is so underdeter- mined by its boundary conditions, experience, that there is much latitude of choice as to what statements to reëvaluate in the light of any single contrary experience. No particular experiences are linked with any particular statements in the interior of the field, except indirectly through considerations of equilibrium affecting the field as a whole. (FLVP, pp. 42–3)
Crucially, the web includes the statements ‘of pure mathematics and logic’, hence obviating the clearest motivation for analyticity, conven- tion, or the a priori. All statements are in that sense ‘revisable’. Quine himself came later to stress that the point is only ‘legalistic’: there is little real prospect of discarding what are the most central and certain prin- ciples of logic and mathematics. A supposition, say, of 1+1=5, would reverberate intolerably throughout science, simply because it is logically related to the rest of arithmetic and other accepted statements. But this thoroughgoing entanglement with science explains not only why it has seemed tempting to suppose that their necessity explains the apparent irrevisability of statements of logic and mathematics, but why there is no such need for such a supposition. Our mastery of these statements need not be explained in terms of the cognition of special features not found to some degree in any other statement.
2. Naturalism
The web of belief metaphor needed, as Quine remarked later, consid- erable ‘unpacking’ (CE, p. 398). If sensory reductionism is not viable, the challenge now is to give a positive, detailed and explicit explan- ation of human knowledge without the bald assumption of the con- cept of analyticity or concepts in terms of which it might be defined, such as the concept of meaning. There are three very general points we should have before us in advance of seeing how Quine’s scheme works.
First, Quine assumes that the main questions of epistemology are questions of language. Language is the very medium of theories, which collectively constitute the whole of human knowledge. Of course, some of what one knows doesn’t seem to amount to a theory; one’s know- ledge that one has a neck, for example. All the same, one is disposed to assent to the sentence ‘You have a neck’; although his focus is often on the higher flights of theory – physics, mathematics – Quine simply extends such words as ‘theory’ to cover the whole of a person’s beliefs.
Secondly, Quine intends to remove from circulation what Carnap had identified as the last remaining realm of irreducibly a priori, philo- sophical concepts. Only genuinely scientific concepts – especially those
pertaining to causal relations – remain. And as we shall see shortly, this means that an explanation of knowledge or language should not pre- suppose such ideas as immaterial referential ties between mind and things or word and object. And it will not be justifiable to suppose, as Frege did, that language is significant by virtue of the mind’s ‘grasping’ those abstract entities known as thoughts or propositions; that psycho- logical idea is from Quine’s perspective unintelligible – how can abstract, immaterial objects explain human cognition, the stuff of axons and syn- apses? As Frege famously put it (Frege, 1984 [1919], p. 371), how can thoughts act? They can’t, is Quine’s answer.
Thirdly, Quine envisages a certain re-ordering of epistemology. Quine’s principal aim is not to prove or justify our claims to know- ledge, but to explain them (see RR, pp. 1–4; SS, pp. 15–16). Given that we have such and such knowledge, what precisely does it involve? What are its most general presuppositions? To take a striking example, Quine agrees with Hume about the principle of induction – the prin- ciple that if a certain proportion of a randomly chosen, large enough sample of a kind of thing has a certain feature, then that proportion may be generalized to the whole population of the kind. It is not a truth of logic – its negation is not self-contradictory – and it would be circular to try to establish it empirically: ‘the Humean predicament’, as Quine puts it, ‘is the human predicament’ (OR, p. 72). But for Quine, to show that the principle is presupposed by and systematically embed- ded in empirical science is all the justification that may be asked for, just as those things showed why we must accept the truths of arithmetic.
Naturalism sums all this up. There is, Quine proclaims, no ‘first philoso- phy; it is only within science itself, and not in some prior philosophy, that reality is to be identified and described’ (TT, p. 21) – meaning not that philosophy is doomed, but that it operates in the more abstract and gen- eral end of things, not in a totally separate sphere of concepts and prin- ciples. It is as if, as Quine quotes from Otto Neurath, we are aboard a ship at sea, having to make running repairs; philosophers look after the more abstract and fundamental things, but cannot deconstruct the whole thing on pain of sinking (WO, p. 3). The essential thing for our purposes is toler- ably clear: Our aim is to explain the core use of language – linguistic competence as it is sometimes called – strictly within the causal realm of scientific psychology.