To: Stuart B.
From: Geoffrey Klempner
Subject: On a criticism of the coherentist theory of knowledge
Date: 26 February 2007 12:05

Dear Stuart,

Thank you for your email of 18 February, with your University of London essay in response to the question, ''My beliefs could form a coherent set even if none of them is true, so the coherence account of knowledge must be wrong.' Discuss.'

I liked the way that you related coherentism and foundationalism to ancient scepticism. However, there are two ways of using the sceptical argument as a way to motivate a particular internalist theory.

The first, most obvious way is to describe the theory of knowledge in a way which implies the refutation of the sceptic's arguments. The sceptic has overlooked a possibility, a possibility which the internalist epistemologist emphasises in his theory. Therefore, the theory of knowledge - whichever one it may be - refutes scepticism.

An alternative way is to leave open the question of how the sceptic is ultimately to be dealt with, but use the sceptic's arguments to determine the form of one's account of knowledge. At some point, it has to be recognized that a 'fully' justified belief - a belief justified to the highest possible standards of justification - can still be false, or that a 'fully' coherent set of beliefs - a set which exhibits the maximum attainable degree of coherence - can likewise still be false.

This implies an implicit element of externalism, as we will see later when we look at the 'tracking truth' question. However, it still seems to me legitimate to view these theories as primarily 'internalist' rather than 'externalist'.

In previous communications, I have given my reasons why I think that a more radical approach needs to be taken in order to specifically address the question of scepticism. This is not the same as the retreat to full externalism which claims, implausibly, that there never was a problem in the first place.

So I am sceptical about any epistemological theory which claims that it can defeat the sceptic, simply by tightening up the conditions for knowledge in such a way that scepticism is ruled out.

As in some previous essays, you have succumbed to the temptation to throw in other objections which are not implied by the objection given in the essay title. It may seem impossible to overlook such a potentially important objection as the claim that it is 'unrealistically demanding' to expect us to be aware of the 'totality of our belief-set', or the 'proposed rules of coherence'. There is room for argument here which would no doubt cast considerable illumination on the coherence theory of knowledge. But it is not logically part of the question. So I will just stress again that you must stick to the question at all costs, otherwise you will be marked down.

(Examiners can be very ruthless. They are looking a specific ability - the ability to respond to the challenge set by the question - and not interested in anything else that you know. Also, it is intrinsic to the evaluation of philosophical ability that one shows a just appreciation of the logic of a question. This is the way the game played. I realize, however, that these essays do double duty for you - as practice answers but also revision aids.)

Earlier, you describe 'a pair of additional challenges - ones that are implied by the essay title'. This is justified, although I would try to be much clearer about how they relate to the objection.

Obviously, it is relevant to the question how one defines 'coherence'. But, once again, the difficulty of defining coherence is not the objection we are looking at. You can talk about the difficulty of defining coherence, but only as a necessary stage in the argument which looks at the objection that a coherent set can still be false.

The question whether a coherent set can track truth is relevant because although this is different from the question whether a coherent set can be false, the two questions are closely related. You need to explain the difference. This is not such an easy thing to do, although we have an intuitive idea of what this would mean. Set A is Tom's coherent set because it is true. That is to say, the best explanation of why Tom holds this set assumes that it is true. However, set A can still be false, if we allow the possibility that one of the sceptical hypotheses might be true.

The sceptic will reply that the sceptical hypotheses render any notion of 'best explanation' empty, since this implies a notion of probability, and probability is relative to evidence. If the Matrix scenario is the truth, then all bets are off so far as 'best explanation' is concerned. But we are not trying to refute the sceptic, merely explaining how the coherence theory can allow room for a notion of 'tracking truth'.

I did get the impression - especially towards the end of your essay - that you forgot that there is a difference between the 'truth' objection and the 'tracking truth' objection. So long as we are not looking to refute scepticism, it is acceptable that a coherent set might conceivably be false if one or other of various far-fetched possibilities turns out to be the case, provided we are satisfied that a coherent set is capable in principle of tracking truth.

Anti-realism about truth is a response to scepticism which brings along its own problems, as you describe. However, a bell should have rung for you when you argued that 'the local set of some one individual may be coherent, yet still inconsistent with the broader vision of coherence truth'. The logically adept paranoid gives excellent reasons for his crazy beliefs. The fact that the beliefs in question are false does not diminish our admiration for the clarity and consistency of his arguments - at least up to a point. However, as a matter of fact (which if one allows the sceptical hypotheses could have been otherwise) we are tracking truth and the paranoid isn't. The paranoid's coherent beliefs are not knowledge because they fail the truth condition.

I haven't mentioned your point that coherentism ignores or underplays important differences between kinds of beliefs, e.g. the special role of perceptual beliefs. You do make an effort to relate this to the question, although once again I would have liked to have seen a bit more argument here.

All the best,

Geoffrey