Monday, November 26, 2007

The Myth of the "a priori"

After some not inconsiderable pondering, I conclude that Quine was correct and Kant was ... errr... mistaken. Quine's argument based upon the incoherence of synonymy may well be valid, but it is unnecessary. We need simply ask of what is an a priori truth 'a priori'? The usual answer would be that an a priori truth is independent of contingent facts. 1 + 1 = 2. My point is that an a priori truth cannot be a priori of the language in which it is expressed, and languages, including mathematics, are contingent upon the existence of a language community. No people = no language with abstract concepts (such as number); people are a contingent fact; ergo abstract concepts are contingent and therefore not 'a priori'. I am not sure Quine's argument against the analytic/synthetic dogma works quite so well. It seems to me that once a language exists and abstract ideas such as kind or number appear, prompted by survival optimising heuristics, then any argument defined purely in terms of abstract ideas is analytic. I guess I am suggesting that first order abstractions are a posteriori and synthetic, and that subsequent levels of abstraction are a posteriori and analytic. Shoot holes in that..