Reflections on Davidson and the omniscient interpreter

Because I have added separate pages on Davidson and radical interpretation and Davidson and Traksi, I will ignore those here and concentrate instead on Davidson’s views on the nature of truth and knowledge and then Davidson and the sceptic.

Davidson’s views on the nature of truth and knowledge

It is worth asking why Davidson calls the paper a coherence theory of truth and knowledge. If truth is a matter of coherence, and knowledge is true, then knowledge will at least include an element of coherence, so why put both in the title? What, also, should we make of the comment ‘I do not hope to define truth in terms of coherence and belief. Truth is beautifully transparent compared to belief and coherence’ [155]?

I think that the answers to both questions are related. To take the latter first, Davidson does not try to reduce truth to coherence. Rather he aims to show how a number of concepts are interdependent. Second, he aims to do this by arguing that the ascription of beliefs and meanings is essentially a holistic enterprise (that is, the totality of beliefs and meanings are themselves interdependent: they have to cohere) governed by the principle of charity which says that they have to be largely true, rational and coherent. This counts as a coherence theory on his account of coherence which is: ‘What distinguishes coherence theory is simply the claim that nothing can count as a reason for holding a belief except another belief.’

(By contrast with the second half of the term, Davidson does not bother to analyse knowledge. He seems to assume that his account of true beliefs that are rationally structured and cohere is pretty much what we mean by knowledge. It is!)

A third preliminary question is what Davidson means when he says that his slogan is ‘correspondence without confrontation’. Two points on that.

First, the implication is that normally correspondence does imply confrontation. Why? Take a statement of a traditional correspondence theory:

• • ‘S’ is true iff ‘S’ corresponds to the fact that S

For each true statement or belief there is a fact it fits or confronts which makes it true. That’s the traditional picture of correspondence which Davidson rejects.

He does not reject it for this reason but we might object that as a theory of what truth is it cannot work because we have no independent hold on what 'correspondence' means, except the relation of being true of, nor what a 'fact' is, except what a true statement states, and we cannot appeal to either of those without begging the question. (By contrast, it is fine to say that ‘S’ is true iff ‘S’ corresponds to the fact that S as a truism to reflect our prior understanding of the whole of this package of ideas. See this blog entry.)

Davidson’s own objection to this stems from the holism of belief and meaning: because beliefs and meanings are holistic there is no determinate relation between true statements / beliefs and facts only a relation between an entire world-view and the world. But in this paper the point he largely makes is epistemological: we can make nothing of the idea of bits of the world compelling our beliefs because facts could only compel beliefs via experiences (we 'drink in' the the world only via experiences) but experiences have the wrong logical properties to justify beliefs. Hence, again, ‘What distinguishes coherence theory is simply the claim that nothing can count as a reason for holding a belief except another belief’.

Second, despite advocating a coherence theory, Davidson does not reject a correspondence theory. He says: ‘truth is correspondence with the way things are... So if a coherence theory of truth is acceptable, it must be consistent with a correspondence theory’. Further, he worries that that coherence alone by itself might not yield truth. That’s why he wants to turbo-charge coherence with correspondence and thus searches for a reason – though not evidence – to say that that coherence does yield truth. The reason is provided by his account of radical interpretation. Interpretation has to be true.

Davidson and the sceptic

Outside the context of scepticism, we might think that, if it were correct, Davidson’s account of radical interpretation is enough to show that beliefs have to be generally true. That would already be an interesting claim. We cannot generally interpret as false.

In the context of scepticism, the worry is that all that radical interpretation shows is that an interpreter has to assume that alien tribes / the rest of one’s own linguistic community have true beliefs by his/her own standards. But those might be false. This is not a simple empirical possibility: it could not easily be the case that a tribe has one set of beliefs and the interpreter another, inconsistent, set and yet through systematic mismatch they do not notice. So the only plausible account is that the tribe and the interpreter are collectively, systematically, wrong about the world.

Davidson’s response is to suggest that, if that further worry is a worry, it can be blocked by the idea of an omniscient interpreter. He argues that if a fallible interpreter has to take the tribe to be true by his/her standards so equally the omniscient interpreter has to take the fallible interpreter’s beliefs (including beliefs about the alien tribe’s beliefs) are true by the omniscient interpreter’s lights. But, being omniscient, those are absolutely, objectively true. So everyone’s beliefs (the fallible interpreter and the tribe) are largely true.

Note that the omniscient interpreter does not have to exist. Rather, if it did, it would find our beliefs to be true. So our beliefs must be true, even if it does not actually exist.

Does this anti-sceptical argument work?

Grant the whole argument. Grant that if any being interprets another’s beliefs, it must do so in such a way as to make them in the main true, by the interpreter ‘s own lights. Assume that there is an omniscient being who interprets us, and conclude that the beliefs it ascribes to us really are, in the main, t rue ones. Allow that the beliefs which such a being would ascribe to us are the beliefs which we really have. Conclude, finally, that our beliefs cannot involve massive error.

Where does all this granting, assuming and allowing get us in relation to scepticism? Nowhere at all —unless we know what the beliefs are that the omniscient interpreter ascribes to us. Otherwise all we know, everything being granted and allowed, is that the belief we evince when we assent to the sentence ‘Here is a hand’ is true, whatever it may be. And since that may, for all that has been shown, be something about the intentions of Cartesian demons, or computer-input to artificially sustained brains, or something else quite beyond our imagination, this “concession” should hardly worry a sceptic. If anything, it broadens the range of his scepticism, since it puts our knowledge of what it is that we believe at risk along with our knowledge of what is the case. (Earlier versions of scepticism did at least tend to have the former alone.) [Craig 1990: 213]

The problem is this. Davidson, like Putnam, is a semantic externalist. Facts about meaning and content are dependent on and partly constituted by the state external world. Hence he tries to glue together beliefs and the world against the sceptic’s attempt to separate them. (According to the sceptic all our beliefs could be false. According to Davidson, they could not all be. In fact, most must be true.) But that does not help unless we know what our beliefs are. And it is not clear, in the context of the possibility of external world scepticism that we do. A new sceptical gap opens up not between our beliefs and the world but now between, on the one hand, the glued together beliefs and world, and on the other, what we take our beliefs to be (what we take them to be about, 'saying' etc.).

It is all very well being told that, if we were brains in a vat, our beliefs would be largely true about the electrical signals we receive. But if – qua such brains – we wrongly take our beliefs to be about what they seem to be about in our virtual reality (they seem to be about trees and chairs and grass and sky) we will be wrong about the contents of our beliefs. That’s even worse!

Two slight complications. It is not entirely clear what this amounts to since we normally take ourselves to have – if not Cartesian style complete transparent access to our minds – then at least an epistemic authority. We know what we are thinking directly and generally successfully. So this possibility (ie Craig’s reading of Davidson) is a bit weird and suggests that something has already gone wrong. But it is not clear that it is Craig who is wrong about Davidson.

Second, Davidson himself does attempt to explain how we can still have first person authority even given his view of beliefs. But the picture is not very clear and takes us too far from the anti-sceptical argument. If interested see:

Davidson, D. (1984) ‘First Person Authority’, Dialectica 38: 101-111

Davidson, D. (1994) ‘Knowing One’s Own Mind’ in Q. Cassam (ed.) Self-Knowledge Oxford: Oxford University Press pp. 43-64