19 September 1997
Many thanks for your letter of 8 September with your piece on realism vs. anti-realism...
First, some personal history. I went up to Oxford in 1976 (University College) to start my B.Phil and hoping to work with John McDowell on the philosophy of language. At first he refused (at the time, I think he was the most popular choice of supervisor). The replacement I was offered refused point blank to read essays on the realism/ anti- realism debate as he found it 'too difficult'. Perhaps if McDowell hadn't relented my career might have taken a quite different turn.
To cut a long story short, I become totally immersed in Dummett and realism vs. anti-realism, 'lived it and breathed it'. That remained a main focus of my studies for the next five years until I completed my D.Phil. Imagine banging your head against a brick wall for five years.
I don't think my time was wasted. I learned a lot, took on board a lot of stuff originally intended as ammunition for the debate, but which proved immensely useful elsewhere. And I got to grips with Kant. (I was supervised by Peter Strawson for a term one of my best and most memorable experiences of Oxford.)
Well, you are right on target with the Kantian antinomies as you will discover in due course (I don't think you've got to that bit yet). The question is, though, how exactly does the Kantian-style diagnosis work in the case of realism/ anti-realism? The essence of an antinomy (as Kant explains) is a pair of contradictory metaphysical statements which purport to exhaust all the possibilities. In fact, they share a common presupposition, which turns out to be false. There is a third possibility: viz. to reject that presupposition. In the case of the cosmological antinomies, the presupposition is that it makes sense to make statements about the universe of objects in space and time as a completed totality.
One way to deal with our antinomy would be go anti-realist about metaphysical truths, and say there's no answer to the question whether realism or anti-realism (about factual or non-metaphysical statements) is true. The rejected presupposition would be that irresolvable metaphysical questions nevertheless 'have answers'. I find that move very unsatisfactory, because I a loathe to admit that there are any ultimately irresolvable metaphysical questions!
What about the assumption of totality, then? In the case of the cosmological antinomies Kant's solution essentially involves a partial anti-realism concerning statements of generality: a statement about 'all things in the universe' is in fact about objects potentially encountered in the progress of experience. Could that be applied to the realism/ anti-realism debate? I wonder.
I propose a different solution, which essentially involves criticising the anti-realist for not going far enough. (See units 14 and 15.) Basically, the idea is that the anti-realist is still thinking of the accessible 'truths' in a realist way, as little nuggets of fact surrounded by an ocean of undetermined possibilities. So you get the picture of a 'moth-eaten' reality, where the holes correspond to questions inaccessible to human inquiry. That picture starts to crumble, however, as soon as you consider what it would take to produce an absolutely definitive verification of a statement. Where are these 'hard truths'? If you allow so much as the logical possibility of error or deception then there can be no nuggets of fact. In other words, the truth is not moth-eaten: there is no such thing as truth!
What about Ayer? It is a standard anti-realist move to claim that the entire content of talk of truth is to accept or reject statements. Talk of 'correspondence between statements and the facts' is metaphysical claptrap. We put forward a statement and wait for it to be confirmed or disconfirmed by experience. But what is the truth value of the statement in the meantime? It's a meaningless question, according to Ayer. Yet I think Ayer would have been unhappy to have been labelled an anti-realist. Throughout his life, he stubbornly resisted the idea that his 'verificationist' views amounted to the adoption of a metaphysical theory concerning the nature of reality.
I think Ayer and Dummett would have agreed, meanwhile, that a verificationist criterion of meaning-fulness is not the same thing as an anti-realist theory of meaning in Dummett's sense. Then again, if you take a broad view, there is not that much difference between the two positions. (There may be some relevant stuff concerning Ayer on Dummett or Dummett on Ayer, but I haven't pursued the question...