1 Davidson's theory of interpretation is a ramification of his view on what a theory of meaning should be: it is a product of considering how a Tarskian definition of truth, when employed as a theory of meaning, can be tested empirically. But we do not have to go into his theory of meaning in order to understand and appreciate some of the interesting things he says in his theory of interpretation. In this paper I will understand interpretation on the model of discerning facts, that is, facts about a speaker's meanings and beliefs. Accordingly, the idea of the indeterminacy of meaning and the idea that beliefs are theoretical constructs, which are two controversial aspects of Davidson's theory of meaning, will be kept out of the picture. See also note 10.

2 Obviously in saying this I am referring to the situation of radical interpretation in which the interpreter has to start from scratch. But, as Davidson remarks, the problem of interpretation ‘surfaces for speakers of the same language in the form of the question, how can it be determined that the language is the same?', and for this reason it can be said that ‘all understanding of the speech of another involves radical interpretation’ (Davidson, Donald ‘Radical Interpretation,’ reprinted in his Inquiries into Truth and Interpretation [Oxford: Clarendon Press 1984], 125)Google Scholar. It should be noted that even in the literally radical situation, the interpreter does not really start from scratch, for Davidson supposes that the speaker's assent to and dissent from a sentence can be identified independently of the interpretation of his utterances.

3 Cf. ‘My point has always been that understanding can be secured only by interpreting in a way that makes for the right sort of agreement. The ‘right sort,’ however, is no easier to specify than to say what constitutes a good reason for holding a particular belief’ (Inquiries into Truth and Interpretation, xvii).

4 Ibid.

5 Davidson, in his more recent paper ‘Three Varieties of Knowledge,’ in Phillips, A.P. ed. A.J. Ayer: Memorial Essays (Cambridge: Cambridge University Press 1991) 153–66Google Scholar, distinguishes between the principle of coherence, which ‘prompts the interpreter to discover a degree of logical consistency in the thought of the speaker,’ and the principle of correspondence, which ‘prompts the interpreter to take the speaker to be responding to the same features of the world that he (the interpreter) would be responding to under similar circumstances’ (158). As he points out, both principles can be called principles of charity. In understanding the principle of charity in terms of the assumption of massive agreement as I am doing in this paper, I think it is quite clear that the principle of charity is taken to include both of the above-mentioned principles, for to assume massive agreement is to assume not only the truth of the beliefs ascribed, but also the logical relations among them. For an even more fine-grained classification of principles of (radical) interpretation, most of which, if not all, can be broadly called principles of charity, see Lewis, David ‘Radical Interpretation,’ in his Philosophical Papers Vol. 1 (New York: Oxford University Press 1983) 108–18Google Scholar.

6 Inquiries into Truth and Interpretation, xvii

7 ‘Radical Interpretation,’ 137

8 Davidson, ‘On the Very Idea of a Conceptual Scheme,’ reprinted in Inquiries into Truth and Interpretation, 197Google Scholar

9 Davidson, ‘A Coherence Theory of Truth and Knowledge,’ reprinted in LePore, E. ed. Truth and Interpretation: Perspectives on the Philosophy of Donald Davidson (New York: Blackwell 1986), 315Google Scholar

10 Ibid., 317. According to Davidson, in constructing a theory of interpretation for a language one is actually constructing an empirical theory of meaning à la Tarski's theory of truth for that language, so what he calls ‘the general method of interpretation’ may include the fulfilling of the formal requirements of a Tarskian theory. This formal aspect of Davidson's theory of interpretation, however, can reasonably be ignored in the present discussion. By the way, it should be noted that ‘a theory of interpretation’ is used in two different senses here: it can refer to a theory about how an interpretation of a language can be arrived at, or a theory by means of which a language can be interpreted. I think it is clear from the context in which sense the expression is used.

11 ‘On the Very Idea of a Conceptual Scheme,’ 196, italics added.

12 This presupposes holism of beliefs, a thesis Davidson's readers are all familiar with. I believe that the thesis is correct; but this is not the place to argue for it, and I can only take it for granted in this paper.

13 Or, in Davidson's own words, ‘the only, and therefore unimpeachable, method available to the interpreter’ ('A Coherence Theory of Truth and Knowledge,’ 316).

14 Davidson, ‘The Method of Truth in Metaphysics,’ reprinted in Inquiries into Truth and Interpretation, 201Google Scholar; ‘A Coherence Theory of Truth and Knowledge,’ 317 italics added.

15 Davidson's argument requires only that it is possible or conceivable that there is an omniscient interpreter who tries to interpret us, rather than that there is in fact such an interpreter.

16 ‘Epistemology Externalized,’ Dialectica 42 (1991), 195

17 Ibid., 199

18 Davidson, ‘Meaning, Truth and Evidence,’ in Barrett, R.B. and Gibson, R.F. eds. Perspectives on Quine (Oxford: Blackwell 1990), 76Google Scholar

19 Ibid., 78

20 I am indebted to John Heil for making me give more thought to the possible roles of the causal theory of meaning and belief in Davidson's theory of interpretation.

21 Cf. ‘We have the idea of belief only from the role of belief in the interpretation of language’ (‘Thought and Talk,’ 170); and ‘[that] causality plays an indispensable role in determining the content of what we say and believe … is a fact we can be led to recognize by taking up, as we have, the interpreter's point of view’ ('A Coherence Theory of Truth and Knowledge,’ 317).

22 This is, I believe, how we should read Davidson's remark that ‘we must, in the plainest and methodologically most basic cases, take the objects of a belief to be the causes of that belief. And what we, as interpreters, must take them to be is what they in fact are’ (‘A Coherence Theory of Truth and Knowledge,’ 317-18, italics added), which otherwise appears to be plainly false.

23 ‘Meaning, Truth and Evidence,’ 78

24 Incidentally, some, probably including Davidson himself, believe that the causal theory of belief and meaning by itself is an answer to skepticism. And they may accordingly think the omniscient interpreter argument is not necessary for refuting skepticism. But if the notion of a cause of belief is to be understood in the way I have just suggested, then evidently the cause of a belief guarantees only that the belief is true by the lights of the interpreter who identifies the cause, rather than that the belief is true tout court. If the notion of a cause of belief is understood independently of the notion of interpretation, then even though the truth of our beliefs is guaranteed by the causes of them, we may not know what the causes are (a brain in a vat may not know that its beliefs are caused by the vat-cum-computer environment), and skepticism is then equally untouched. For similar arguments, see, for example, Brueckner, Anthony ‘Charity and Skepticism,’ Pacific Philosophical Quarterly 67 (1986) 264–8CrossRefGoogle Scholar; and Craig, Edward ‘Davidson and the Sceptic: The Thumbnail Version,’ Analysis 50.4 (1990) 213–14CrossRefGoogle Scholar.

25 For a more detailed argument against the omniscient interpreter argument along this line, see my ‘To Interpret, or to be Omniscient,’ Philosophical Papers 32 (1993) 189-98.

26 ‘Radical Interpretation,’ 137

27 Correspondingly, ‘agreement’ also has two senses, one in which it is required that those who have agreement understand one another, and one in which there is no such requirement. But for my purposes here this distinction is unimportant.

28 Let me stress that they are not truly conceivable if conceivability is to be indicative of possibility. Some may want to insist that, say, it is conceivable, though (metaphysically) impossible, that water is XYZ rather than H20. Such insistence may make perfect sense, but then it is irrelevant to the point I am making here, for what is insisted on is based on a sense of conceivability in which conceivability is not indicative of possibility.

29 I discovered after writing this paper that Williams, Bernard in his excellent ‘Imagination and the Self’ (reprinted in his Problem of the Self [Cambridge: Cambridge University Press 1973] 26–45)CrossRefGoogle Scholar has a very interesting and illuminating discussion of some issues related to the distinction I have made. There is a distinction Williams makes in his paper that is particularly worth mentioning here: he distinguishes between ‘imagining (myself) doing, seeing, etc. certain things’ and ‘visualizing from the outside a figure who is myself doing the things in question’ (38). The former is necessarily first-person imagination in my sense, but the latter can also be, depending on whether the visualizing essentially involves me as the central figure. In terms of visualization, if I am having a first-person imagination of, say, my having a piano recital, I can either visualize my hands, the keyboard and some of the audience, that is, the things that I would see when actually playing the piano, or visualize my whole body, the whole piano and the whole audience, as if I am seeing my own recital on television. I am indebted to Jon Ellis for this point.

30 It goes without saying that not every third-person imagination has a possible corresponding first-person imagination.

31 It seems that I can imagine that despite my failure to ascribe any beliefs successfully to the creature I nonetheless insist that it must have some beliefs, for it seems that it is not impossible that I would so insist. Three points are in place here: first, it is not clear how much sense such insistence would make; it is like insisting that a person is ill even though he has no symptoms of any illness and looks perfectly fine. Second, that I can so insist does not mean that I can so insist as an interpreter, or, to put it another way, my insistence is not supported in any way by my attempted interpretation of the creature. Third, I might have some reason for insisting that the creature has beliefs if, say, I have been assured by someone I trust absolutely that this is so; but such reason should be overridden by my repeated failure of ascribing any beliefs to the creature (provided that I am, and know that I am, a normally competent interpreter). Perhaps I can still insist that it has beliefs, but I cannot insist so reasonably. Here I am indebted to Sam Scheffler.

32 Since it is not clear how massive the agreement has to be required by successful interpretation, it may be objected that I can indeed imagine that the interpreter and the interpreted creature are in two-thirds disagreement with each other, and that I am in two-thirds agreement with each of them. But I doubt that two-thirds agreement is massive enough for successful interpretation and understanding, though I don't know how to argue that the agreement must be at least more than two-thirds. Actually, I would rather insist that ‘in massive (dis)agreement’ is a transitive relation than try to settle the matter numerically. I am indebted to Bruce Vermazen for this possible objection.

33 Cf. note 31.

34 See note 25.

35 ‘A Coherence Theory of Truth and Knowledge,’ 315; italics added.

36 Ibid., 317; italics added.

37 Williams, Bernard Ethics and the Limits of Philosophy (Cambridge, MA: Harvard University Press 1985), 148Google Scholar

38 I would like to thank. Jason Bridges, Jon Ellis, John Heil, Sam Scheffler, Barry Stroud, Bruce Vermazen, and an anonymous referee for this journal for their helpful comments on an earlier version of this paper.