Carnap’s Principle of Tolerance and Quine’s Objections

Carnap holds that the role of philosophy is to analyze and clarify the language of science, and to formulate and recommend alternative languages. After 1932, he holds that there are languages which differ in expressive power, not merely as notational variants. The difference between intuitionistic logic and classical logic is an important example here, as is the difference between the language of Newtonian mechanics and the language of relativistic mechanics. (Earlier, he says, he had ‘overlooked the fact that there is a multiplicity of possible languages’, Carnap 1934, 245; cf. also 322 of the same work.) Within a given language, there may be only one correct theory on a given subject. But the choice of a language is in that respect unlike the choice of a theory. (1): In deciding which theory is correct, we will appeal to observations but also to the rules of the language; no such appeal is possible when we are choosing a language, for in this case it is the rules themselves that we are choosing. So Carnap holds that there is no one correct language. This idea has become known as the Principle of Tolerance; from the 1930s on, it is fundamental to Carnap’s view of what philosophy is and how it differs from science. (That there is such a difference is a point which Carnap never questions.) Because there is no one correct language, it is no part of the philosopher’s job to prescribe this or that language, merely to analyze, to clarify, and to suggest alternatives.

(2): The Principle of Tolerance requires the analytic-synthetic distinction. It requires that we can, in all cases, distinguish between theoretical changes which involve a change of meaning, and hence, strictly speaking, a change of language, and those which do not. The former kind of change involves analytic sentences, and is a matter for Tolerance; the latter involves only synthetic sentences, and Tolerance does not apply. The use of the Principle of Tolerance thus presupposes a clear distinction between the analytic sentences of the language and its synthetic sentences.

At least as Quine sees the matter, the use of the Principle of Tolerance puts analytic sentences on an entirely different epistemological footing from synthetic sentences. Synthetic sentences are answerable to evidence; analytic sentences are a matter of the choice of language, which does not require theoretical justification. (3): Quine, however, rejects the idea that there is such an epistemological difference. Even if we can distinguish the analytic sentences from the synthetic sentences, we may still have reasons to reject an analytic sentence. And those reasons may be of the same kind that lead us to reject synthetic sentences. This point is hard to see if one focuses on examples such as “All bachelors are unmarried”. The matter is otherwise if one considers examples such as “Force equals mass times acceleration”. (See Putnam, 1962.) A change of mind about an analytic sentence would be a change in the language. Still, we might have reasons to make such a change, reasons that are of the same sort that lead us to make revisions to synthetic sentences. In that case, we have no more reason to apply Tolerance in the one kind of case than in the other.

This is the view that Quine argues for. On the one hand, he emphasizes the point (which Carnap largely accepts) that choice of language is not theoretically neutral: some choices will make for a better theory than others. On the other hand, he argues that ‘pragmatic’ factors, such as simplicity, which Carnap had accepted as playing a role in choice of language, also play a role in the choice of a theory within a language. Hence, he claims, the two sorts of choice are on the same epistemological footing, and the Principle of Tolerance is unjustified.

(4): Quine’s argument for this position relies on holism. This is the claim that most of our sentences do not have implications for experience when they are taken one-by-one, each in isolation from the others. What has experiential implication is, in most cases, not an individual sentence but a larger chunk of theory. Holism is not a very controversial doctrine. (Carnap accepts it; see Carnap, 1934, 318.) Quine claims that holism shows that most of our sentences are not justified by the relation of the individual sentence, considered in isolation, to experience. Almost always, what matters is the relation to experience of some larger chunk of theory (in principle, although perhaps never in practice, of the theory as a whole). This means that in principle the correctness of a given claim is almost never settled simply by looking at the empirical evidence for that claim alone. Other factors will play a role, in particular the way in which accepting the given claim would contribute to the efficacy and simplicity of the theory as a whole. But these are precisely the ‘pragmatic factors’ which Carnap thought played a role in the choice of language. In arguing that such factors play a role throughout our knowledge, Quine accepts ‘a more thorough pragmatism’ (1951, 46) which puts Carnap’s external changes on the same epistemological footing as his internal changes. (Quine’s early account of these matters is extremely sketchy. For elaboration, see Quine 1991 and 1996; for criticism of the sketchy account, see Sober 2000.)

I suppose I am working up to trying to study “coherentism” from a Quinean perspective, and I’d like a deeper understanding of the above 4 points.

  1. I feel like I need to see a strong example of why Carnap felt “all languages are equivalent”. Didn’t he think there had to be at least some constraints on languages, languages which we hope meaningfully describe the world? Obviously, he probably wasn’t referring to gibberish. Then what was he referring to? How would Carnap defend his view?
  2. This sounds sort of like that “meaning” has two axes, the intensional aspect of logic, where the structures have some kind of “significance” in relation to each other, and the extensional aspect of logic, where those structures correspond to real-world phenomena. So, Carnap believed “devoutly” in the analytic-synthetic divide (like Descartes). But in advocating “tolerance”, what is a concrete repercussion of such an assumption? (For example, does it buttress some kind of epistemological relativism in the social sciences, or “hard” logical pluralism in the philosophy of mathematics?)
  3. Quine rejects the analytic-synthetic divide, but I need far more details why. For example, one might consider logical reasoning itself “empirical”, in that, for any “principles” of reason we hold to be true, we are forced to reject and revise them when they lead to unsound conclusions. Russell’s paradox would be a great example of this: it was not known that there could not be a “set of all sets” until someone discovered a problem with that idea. Thus, “analytic knowledge” is by no means a priori - and since, in a way, we are “observers” of our own thoughts in our own minds, you could claim that even “logical discovery” is a kind of a posteriori, experiential knowledge. Is that why Quine rejects this divide, or something different?
  4. I do not understand (4) at all. Carnap thinks that since that analytic meaning and synthetic meaning are completely orthogonal, it follows that they are independent. Thus, you can choose which “frame” to look at “the world” through, and there is no meaningful way to say which one is “better”; they are just “options”. Quine thinks that the “meaning” of sentences is not just how the language is designed to map to “the real world”, but also very influenced by the intensional characteristics of the language itself. So… since Quine thinks languages “take meaning” in relation to one another, he basically does believe in “one true language” - and that’s “pragmatism”?


  • Carnap didn't hold that all languages are "equivalent" (what would that even mean--equivalent with respect to what? Obviously he didn't think all languages were equivalent with respect to their scientific utility, for the entire point of philosophy is to analyze and formulate languages which are better suited for a given scientific purpose). Commented Apr 13 at 2:31
  • Rather, his point is that there could be no meaningful propositions which could give a theoretical justification for the use of a language, because all such theoretical/epistemic content presupposes the rules of the language in which the proposition is formulated. So you can't have propositions that justify the choice of linguistic rules. But you can have other pragmatic considerations which justify the choice of a language for a given purpose. Commented Apr 13 at 2:33
  • With respect to point (2), no, you're confusing the analytic-synthetic distinction with the difference between intensional and extensional. On point (3), no that's not why Quine rejected it. Quine held that no philosopher has ever succeeded in giving an intelligible account of analyticity, and to the extent that Carnap's conception of analyticity makes sense, it is empty: it has no instances. So either the notion is incoherent or empty. Commented Apr 13 at 2:38

1 Answer 1


Your question asks many things and would need a long essay to answer them all. I'll focus on just one.

Quine rejects the concept of analyticity on several grounds. One is that attempts to establish that some sentences are true by virtue of their meanings is either vacuously true or viciously circular. Another is that even sentences identified as analytic by its defenders are not immune to revision in the light of new data. This is perhaps not fatal to the concept of analyticity, but it undercuts the logical positivist project of explaining necessity and a priori knowledge by appeal to analyticity. A third is based on his holism principle. We do not verify or falsify our beliefs individually on a sentence-by-sentence basis, but by testing how entire chunks of beliefs, or theories, fit with our experience. A fourth claim is that the very concept of 'meaning' is unclear and lacks explanatory value. The latter is part and parcel of Quine's doctrine of the indeterminacy of translation, and follows from his behaviourist approach to linguistics.

To expand the point about holism a little... Suppose we have a theory. We may designate some of the sentences within that theory as axioms, some as definitions and the rest as theorems. But labelling some sentences definitions does not confer any special epistemological status on them. Axioms and definitions are not automatically necessarily true or a priori knowable just because of their status within the theory. The choice of axioms and definitions is made for pragmatic reasons, e.g. for the benefit of clarity and simplicity. The theory as a whole is revisable.

Now suppose we acquire evidence that conflicts with our theory, and a better theory is available. When we replace the old theory with the new one, we are not obliged to retain the old definitions. We might choose to do so, but in some cases it is better to revise the definitions too. Quine and Putnam give an example of this. In classical mechanics momentum is defined to be mass x velocity. Then along comes relativistic mechanics and momentum is redefined to be mass x velocity x Lorentz factor. The change in definition is not arbitrary: the new definition makes a better fit with our superior new theory.

The point is that even definitions themselves are not immune to revision in the light of empirical data, the same as any other propositions are. So if your understanding of 'analytic' is anything like Frege's definition: a proposition that is derivable from a logical truth by substituting definitions, then nothing is immune to revision. Quine is content to allow that sentences like "all bachelors are unmarried men" qualify as analytic in the sense that for a native speaker of a language, that may well be how they learned the meaning of the word 'bachelor'. But how we learned the meaning of words, and what words may come to mean under future scrutiny and criticism can be different things.

Quine's position is developed in a series of papers: Truth by Convention. Two Dogmas of Empiricism. Carnap and Logical Truth. The first few chapters of Word and Object. Also useful is his "Two Dogmas in Retrospect", Canadian Journal of Philosophy, Vol. 21, No. 3, pp 265–74. (1991).

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .