Tuesday, January 21, 2014

Truth vs empirical adequacy in a scientific theory

To: Alistair L.
From: Geoffrey Klempner
Subject: Truth vs empirical adequacy in a scientific theory
Date: 9th August 2011 14:15

Dear Alistair,

Thank you for your email of 29 July, with your essay for the University of London Methodology module, in response to the question, 'Explain and assess the view that the acceptance of a scientific theory only involves a commitment to the 'empirical adequacy' of the theory, not to its truth.'

This is a well written essay but it leaves me feeling very puzzled about the focus of the issue between Van Fraasen's anti-realist and realist with regard to scientific theories.

Some background: My D.Phil thesis, 'The Metaphysics of Meaning' looked at the issue of realism vs global anti-realism in the philosophy of language, and specifically Michael Dummett's views about the need for an 'anti-realist theory of meaning'. According to Dummett, an anti-realist theory of meaning rejects truth as the 'central concept' of a theory of meaning and replaces it with something like 'rules of verification', that is to say, the rules that a speaker learns when they learn to use the language, rules which are 'manifest' in linguistic usage. This is what Dummett takes to be the cash value of the later Wittgenstein's assertion that 'meaning is use'.

This commits the anti-realist to doing something rather difficult, namely constructing an adequate anti-realist theory of meaning in terms of rules of verification -- something which no-one has actually done, and there are good reasons why it can't be done (see McDowell 'Truth Conditions, Bivalence and Verification' in Evans and McDowell 'Truth and Meaning' OUP 1976).

To cut a long story short, I came to the conclusion that the case for anti-realism is better stated by assuming from the start a Davidsonian-style truth-conditional semantics, accepting the law of excluded middle (the key axiom which Dummett's anti-realist objects to) and arguing instead that the realist is unable to *explain what they mean* by a statement, e.g., 'A tree stood exactly on this spot 100 million years ago' 'has' a truth value.

The statement is true *or* false. Either a tree stood here or not. But the assertion of the LEM doesn't get you any further. We don't know, we can't know. And simply enunciating the statement doesn't achieve anything either. All you have done is describe two alternate worlds. (I described the realist 'picture' as aiming an arrow at a target which is so far away you can never know whether you scored a bulls-eye or not.)

In an exactly similar way, I would argue, the scientific 'realist' has nothing to add to what scientists already do, which is perform experiments, enunciate theories, test them, revise the theories and so on.

The issue of hidden variables is an interesting case in point, because here it does look as though being a 'realist' commits you to something that merely settling for empirical adequacy doesn't. But that is an illusion. Of course, we'd like to find hidden variables. The Aspect experiment seems to show that there can't be any. But like any experiment, there's always the possibility of discovering new evidence we hadn't considered before. If your interest is empirical adequacy, then you should *consider the possibility* of hidden variables. If you conceive yourself as 'aiming at truth' then you should *consider the possibility* of hidden variables. As to whether there are any hidden variables, we can't be certain. We can only state the best theory we have to date.

I would strongly defend a minimalist view about the *use* of the term 'true'. That doesn't mean there is no metaphysical question about the nature of what is 'out there' or the nature of 'reality'. What it means is that if you say, after performing certain experiments, 'There are no hidden variables,' and your assertion meets the standards for making claims in science (e.g. your paper would be accepted by a respectable Physics journal) then you have asserted what you 'believe' to be a 'truth'. It is 'true' that there are no hidden variables, so far as you have been able to ascertain. That doesn't mean everyone has to agree with you. It doesn't mean you can't change your mind. There are disputes in physics just as there are disputes in history or astrology.

You could have said a bit more about the idea that realism 'explains' the success of science (a la Putnam?). Once again, it seems perfectly possible for the self-styled 'anti-realist' to take on board anything the realist claims about the possibility of a meta-explanation.

But where does this get you? My supervisor, John McDowell took (takes) this to be a clear argument for realism: if you can formulate an anti-realist view without incoherence, then the view can have no practical consequences. You're not saying anything. I took it to be an argument for anti-realism: the self-styled 'realist' is unable to explain what they mean by a statement (or theory) 'really' being true or 'really' having a truth value.

My present view would be agnostic. There is a metaphysical illusion lurking here about the nature of truth, but rejecting that illusion hardly qualifies for the label 'anti-realist'.

Van Fraasen would like to be able to find an issue over which there can be real disagreement between the realist and anti-realist about science. Just making claims about what we 'believe' cuts no ice at all. If you are prepared to defend a proposition, if you are prepared to act on it, then you 'believe' it. There is no further question about *how* you believe it (as a 'truth' or as being 'empirically adequate').

All the best,