The American philosopher Donald Davidson is famous for his work in the philosophy of language and epistemology, particularly in his development of some important analytic concepts that were prominent in the work of another American philosopher, Willard van Orman Quine. In his 1974 article “On the Very Idea of a Conceptual Scheme”, Davidson famously attacked what he called the third dogma of empiricism (Quine elaborated the first two dogmas in his paper “Two Dogmas of Empiricism”). Quine’s replies to Davidson are interesting though not entirely satisfactory, and Davidson’s arguments are, nonetheless, interesting, at the very least from an historical standpoint. Davidson’s rejection of that third dogma, the distinction between a conceptual scheme and a experiential content is philosophically insightful.

In his now famous article Davidson wrote:
This second dualism of scheme, of organizing system and something waiting to be organized, cannot be made intelligible and defensible. It is itself a dogma of empiricism. The third dogma (pg. 189).
Davidson’s conclusion is directed against theories (including empiricism) that posit the existence of something that is epistemically intermediary between our language and the subject matter of our language (e.g., the world, sensory experience, sense datum, etc.). Davidson’s argument is directed precisely against the idea that there is some third entity involved in communication. Davidson wants to show that there is an unmediated connection between language and the world and that this connection is sufficient for the existence and utility of communication.

Davidson’s argument, to put it very briefly, is that since we cannot make sense of the idea of a difference in conceptual schemes we can neither make sense of a unity in them (there might be connections to the work of Jacques Derrida here in that Davidson is privileging the pair difference/identity as key terms of the philosophical debate). If we can detect no difference in conceptual schemes, we have no reason for asserting that they even exist in the first place. As Davidson says, "If we cannot intelligibly say that schemes are different, neither can we intelligibly say that they are one" (pg. 198). Davidson’s argument to the effect that we can detect no difference in schemes (i.e. no evidence for the existence of schemes in the first place) comes from his examination of the Quinean notion of interpretation. All interpretation always arrives at one of three end-states: 1) the alien language (i.e., the language being interpreted) and the home language (i.e., the interpreter’s language, into which the alien language is being translated) are fully translatable, 2) the alien language is partially translatable into the home language, 3) the alien language is completely untranslatable into the home language. The first case obviously provides us with no evidence for a differentiation in concepts. The second case, Davidson argues, doesn’t necessitate viewing the partial failure in interpretation as a differentiation in concepts rather than a differentiation in beliefs (pg. 197); Quine seems to accept this argument. The third case, when a language presumably cannot be stated using the words of the home language, receives the most focus from both Davidson and Quine in their debate.

Davidson directs most of his attention to the third case because it is in (presumed) cases of interpretation like these that we can most clearly see the ideas at work in the establishment of a scheme and content dichotomy. Davidson writes that the idea behind the scheme and content dichotomy, which he rightly attributes to Quine, is:
that something is a language, and associated with a conceptual scheme, whether we can translate it or not, if it stands in a certain relation (predicting, organizing, facing, or fitting) to experience (nature, reality, sensory promptings). The problem is to say what the relation is, and to be clearer about the entities related (pg. 191).
The idea here is that if something is a language than there must be something else that it conceptualizes or mediates. For Quine, x is a language only if x stands in a certain conceptualizing or mediating relationship to its subject matter. This relationship is one of fitting and that which our language holistically fits is the totality of experience.

Davidson’s problem with this language (scheme) and world (content) dualism is that given this definition of "language" it follows that there is no a priori argument against the idea that there may exist an organizational scheme so different from our own that it fails of any translation whatsoever. If a language is defined not by our (human) recognition of it as such, but by its having a certain relationship to the experience of its speakers, then it follows that there exists the possibility of a complete failure in translation. This alien language, it is essential for Quine’s argument, must be a language that fits the experiences of the creatures that speak it—if it does not fit their experiences and is not shaped according to them, then it is not a language.

The idea that a language stands in any sort of justificatory relationship to either the world itself or the experience of its speakers is, for Davidson, untenable. It is subject to all of the problems of Descartes’ skepticism. Davidson’s most quoted sentence in just about any context forcefully states this point:
Nothing, however, no thing, makes our sentences and theories true: not experience, not surface irritations, not the world, can make a sentence a true. . . . The sentence ‘My skin is warm’ is true if and only if my skin is warm. Here there is no reference to a fact, a world, an experience, or a piece of evidence (pg. 194).
The idea at which Davidson is here driving is that the subject matter of our sentences is related to our sentences in a causal manner, but not in a justificatory manner. There is, Davidson is urging, no epistemic intermediaries between our language and the world. (This notion also ensues from Davidson’s disquotational approach to truth, following on from the work of the Polish mathematician Alfred Tarski (see that node).)

Davidson urges that from Quine’s insistence upon language’s being defined according to its fitting the experience of its speakers, Quine’s characterization of conceptual schemes boils down to the idea that something is a conceptual scheme (i.e., a language) or theory if it is true. As noted above, Quine insists that x is a language only if x fits the experience of its speakers. For Quine, then, the search for something’s being a language comes down to the sole criterion of its being largely true, though not necessarily translatable.

Davidson thinks that Quine’s view is absurd for the simple reason that Quine’s view entails that x can be a language even if we can’t translate x into our own language. But, for Davidson, complete translation failure is unintelligible. The commonsense dictum behind this is: if we can’t (logically) translate it, how do we know it’s a language? We might likewise ask ourselves: if we can’t read it due to logical constraints, how do we know that it’s writing? If the evidence for something’s being writing is other than that thing’s being readable, then it follows that there is no possible argument against the existence of an invisible script. The situation seems almost humorous. We might now ask questions such as: how many invisible scripts are there?; is that a sample of invisible script over there on that wall?; I see you moving your hands—are you writing in invisible script? Likewise, if something is a language by virtue of its being true, then something might be a language but totally unrecognized (and unrecognizable) as such. Quine’s definition of "language" entails that x could be language not just in mere cases of translation failure, but also in those cases in which we don’t even recognize that x is a language, in cases where translation is not even attempted. On Quine’s view, then, no evidence can be produced contrary to the following claims: "Dogs speak a language", "Fish speak a language", "The kitchen faucet speaks a language". That Quine cannot reject the idea that there is a language spoken in the sink follows from two points: first, that the faucet makes sounds and presumably x is a language only if x can be represented acoustically or some other way; the second more essential point is that it is not a priori impossible that the sounds made by the faucet are in fact about the faucet’s experience and do in fact fit this experience. On Davidson’s view, Quine cannot, given his views on what counts as a language, refute this obviously absurd claim. This is most obvious and commonsensical in the case of language-speaking dogs. On Davidson’s view, we can’t intelligibly say that dogs speak a language for the simple reason that we can’t swear their utterances to being about the world. A basic contrast between Davidson and Quine issues from this question; Davidson will not allow us to say that dogs speak a language whereas Quine would probably not say that they do, but could not provide for anybody saying that they do an argument consistent with his writings. Davidson’s refutation of the notion of conceptual schemes springs essentially from his focus on the idea, held by Quine, that evidence for conceptual difference springs from cases of total translation failure.

Quine’s idea of language is that it is defined by its having a fitting relationship to experience. Davidson sets forth the negative thesis that, contrary to Quine’s, language does not fit experience. For Davidson, evidence is not epistemic but causal. Truth is, for Davidson, not a fitting relation that is foundational in nature, but is rather a relation of semantic correspondence and internal coherence.