Donald Davidson (philosopher)

Davidson's most noted work began in 1963 with an essay, "Actions, Reasons, and Causes," which attempted to refute the prevailing orthodox view, widely attributed to Ludwig Wittgenstein but already present in Leo Tolstoy's War and Peace, that an agent's reasons for acting cannot be the causes of his action (Malpas, 2005, §2). Instead, Davidson argued that, "rationalization (the providing of reasons to explain an agent's actions) is a species of ordinary causal explanation" (1963, p. 685). In particular, action A is explained by what Davidson called a primary reason, which involves a pro-attitude (roughly, a desire) toward some goal G and an instrumental belief that performing action A is a means to attaining G. For example, someone's primary reason for taking an umbrella outside on a rainy day might be that wanting to stay dry and believing that taking an umbrella is a means to stay dry today.

This view, which largely conforms to common-sense folk psychology, was held in part on the ground that while causal laws must be strict and deterministic, explanation in terms of reasons need not. Davidson argued that the fact that the expression of a reason was not so precise did not mean that the having of a reason could not itself be a state capable of causally influencing behavior. Several other essays pursue consequences of this view and elaborate Davidson's theory of actions.

In "Mental Events" (1970) Davidson advanced a form of token identity theory about the mind: token mental events are identical to token physical events. One previous difficulty with such a view was that it did not seem feasible to provide laws relating mental states, like believing that the sky is blue or wanting a hamburger, to physical states, such as patterns of neural activity in the brain. Davidson argued that such a reduction would not be necessary to a token identity thesis: it is possible that each individual mental event just is the corresponding physical event, without there being laws relating types (as opposed to tokens) of mental events to types of physical events. Davidson argued that the fact that no such a reduction could be had does not entail that the mind is anything more than the brain. Hence, Davidson called his position anomalous monism: monism, because it claims that only one thing is at issue in questions of mental and physical events; anomalous (from a-, "not," and homalos, "regular", also nomos (law)) because mental and physical event types could not be connected by strict laws (laws without exceptions).

Davidson argued that anomalous monism follows from three plausible theses. Firstly, he assumes the denial of epiphenomenalism, the denial of the view that mental events do not cause physical events. Secondly, he assumes a nomological view of causation, according to which one event causes another if (and only if) there is a strict, exceptionless law governing the relation between the events. Thirdly, he assumes the principle of the anomalism of the mental, according to which there are no strict laws that govern the relationship between mental event types and physical event types. By these three theses, Davidson argued, it follows that the causal relations between the mental and the physical hold only between mental event tokens, but mental events as types are anomalous. That ultimately secures token physicalism and a supervenience relation between the mental and the physical, while respecting the autonomy of the mental (Malpas, 2005, §2).

In 1967 Davidson published "Truth and Meaning," in which he argued that any learnable language must be statable in a finite form even if it is capable of a theoretically infinite number of expressions, as may be assumed that natural human languages are, at least in principle. If it could not be stated in a finite way, it could not be learned through a finite, empirical method such as the way humans learn their languages. It follows that it must be possible to give a theoretical semantics for any natural language that could give the meanings of an infinite number of sentences on the basis of a finite system of axioms. Following, among others, Rudolf Carnap (Introduction to Semantics, Harvard 1942, 22) Davidson also argued that "giving the meaning of a sentence" was equivalent to stating its truth conditions, so stimulating the modern work on truth-conditional semantics. To sum up, he proposed that it must be possible to distinguish a finite number of distinct grammatical features of a language, and for each of them explain its workings in such a way as to generate trivial (obviously correct) statements of the truth conditions of all the (infinitely many) sentences making use of that feature. Thus, a finite theory of meaning can be given for a natural language; the test of its correctness is that it would generate (if applied to the language in which it was formulated) all the sentences of the form "'p' is true if and only if p" ("'Snow is white' is true if and only if snow is white"). (They are called T-sentences: Davidson derives the idea from Alfred Tarski.)

This work was originally delivered in his John Locke Lectures at Oxford and launched a large endeavor by many philosophers to develop Davidsonian semantical theories for natural language. Davidson himself contributed many details to such a theory, in essays on quotation, indirect discourse, and descriptions of action.

After the 1970s Davidson's philosophy of mind picked up influences from the work of Saul Kripke, Hilary Putnam, and Keith Donnellan, all of whom had proposed a number of troubling counterexamples to what can be generally described as descriptivist theories of content. The views, which roughly originate in Bertrand Russell's Theory of Descriptions, held that the referent of a name, which object or person the name refers to, is determined by the beliefs a person holds about that object. Kripke et al. argued that this was not a tenable theory, and that in fact whom or what a person's beliefs were about was in large part (or entirely) a matter of how they had acquired those beliefs, and those names, and how if at all the use of those names could be traced "causally" from their original referents to the current speaker.

Davidson picked up this theory, and his work in the 1980s dealt with the problems in relating first-person beliefs to second- and third-person beliefs. It seems that first person beliefs ("I am hungry") are acquired in very different ways from third person beliefs (someone else's belief, of me, that "He is hungry"). How can it be that they have the same content?

Davidson approached the question by connecting it with another one: how can two people have beliefs about the same external object? He offers, in answer, a picture of triangulation: beliefs about oneself, beliefs about other people, and beliefs about the world come into existence jointly.

Many philosophers throughout history had, arguably, been tempted to reduce two of these kinds of belief and knowledge to the other one: René Descartes and David Hume thought that the only knowledge that people start with is self-knowledge. Some logical positivists (and some would say Wittgenstein, or Wilfrid Sellars) held that people start with beliefs only about the external world. (Arguably, Friedrich Schelling and Emmanuel Levinas held that people start with beliefs only about other people.) It is not possible, on Davidson's view, for a person to have only one of the three kinds of mental content; anyone who has beliefs of one of the kinds must have beliefs of the other two kinds.

Davidson's work is well noted for its unity, as he has brought a similar approach to a wide variety of philosophical problems. Radical interpretation is a hypothetical standpoint which Davidson regards as basic to the investigation of language, mind, action, and knowledge. Radical interpretation involves imagining that you are placed into a community which speaks a language you do not understand at all. How could you come to understand the language? One suggestion is that you know a theory that generates a theorem of the form 's means that p' for every sentence of the object language (i.e. the language of the community), where s is the name of a sentence in the object language, and p is that sentence, or a translation of it, in the metalanguage in which the theory is expressed. However, Davidson rejects that suggestion on the grounds that the sentential operator 'means that' is sensitive not only to the extensions of the terms that follow it, but also to their intensions. Hence, Davidson replaces 'means that' with a connective sensitive only to the extensions of sentences; since the extension of a sentence is its truth value, this is a truth functional connective. Davidson elects the biconditional (if and only if) as the connective needed in a theory of meaning. He concludes that a theory of meaning must be such that for each sentence of the object language it generates a theorem of the form 's is true if and only if p'. A theory of truth for a language can serve as a theory of meaning.

The significance of this conclusion is that it allows Davidson to draw on the work of Alfred Tarski in giving the nature of a theory of meaning. Tarski showed how we can give a compositional theory of truth for artificial languages. Thus, Davidson takes three questions to be central to radical interpretation. Firstly, can a theory of truth be given for a natural language? Secondly, given the evidence plausibly available for the radical interpreter, can they construct and verify a theory of truth for the language they wish to interpret? Thirdly, will having a theory of truth suffice for allowing the radical interpreter to understand the language? Davidson has shown, using the work of Tarski, that the first question can be answered affirmatively.

Davidson points out that beliefs and meanings are inseparable. A person holds a sentence true based on what he believes and what he takes the sentence to mean. If the interpreter knew what a person believed when that person held a sentence to be true, the meaning of the sentence could then be inferred. Vice versa, if the interpreter knew what a person took a sentence to mean when that person held it to be true, the belief of the speaker could be inferred. So Davidson doesn't allow the interpreter to have access to beliefs as evidence, since the interpreter would then be begging the question. Instead, Davidson allows that the interpreter can reasonably ascertain when a speaker holds a sentence true, without knowing anything about a particular belief or meaning. That will then allow the interpreter to construct hypotheses relating a speaker and an utterance to a particular state of affairs at a particular time.

Davidson argues that because the language is compositional, it is also holistic: sentences are based on the meanings of words, but the meaning of a word depends on the totality of sentences in which it appears. That holistic constraint, along with the requirement that the theory of truth is law-like, suffices to minimize indeterminacy just enough for successful communication to occur.

In summary, what radical interpretation highlights is what is necessary and sufficient for communication to occur. The conditions are to recognize speakers as speakers, their beliefs must be mostly coherent and correct, according to the principle of charity; indeterminacy of meaning does not undermine communication, but it must be constrained just enough.

I conclude that there is no such thing as a language, not if a language is anything like what many philosophers and linguists have supposed. There is therefore no such thing to be learned, mastered, or born with. We must give up the idea of a clearly defined shared structure which language-users acquire and then apply to cases. And we should try again to say how convention in any important sense is involved in language; or, as I think, we should give up the attempt to illuminate how we communicate by appeal to conventions.

— "A Nice Derangement of Epitaphs," Truth and Interpretation, 446