Saturday, November 7, 2009

Williamson on Evidence and Truth

This is from The Philosophy of Philosophy:
Why is it bad for an assertion to be inconsistent with the evidence? A natural answer is: because then it is false. That answer assumes that evidence consists only of true propositions. For if an untrue proposition, p, is evidence, the proposition that p is untrue is true but inconsistent with the evidence.

That's an answer, but it seems there is an equally promising answer that does not assume that evidence consists only of true propositions. Someone might think that:

JE: If S's belief that p is justified, p is part of S's evidence.
R: If S's belief that p is justified, the processes that produced S's belief were reliable.

Whatever is justifiably believed is a part of our evidence and whatever is justifiably believed is likely to be true. This is because whatever is justifiably believed is the result of a reliable method of belief formation. Thus, if an assertion is inconsistent with the evidence, it is likely to be false. In a conversational context where one speaker asserts p and then both speakers discover that p is incompatible with what is justifiably believed by the other speaker, it is natural to retract p until additional reasons for believing p or for discounting the justification that supports believing ~p. We can explain why it is bad for an assertion to be inconsistent with the evidence without having to assume EST.

Adequate response? I want Williamson to be right, but I can't see what's wrong with this response.


Jonathan Ichikawa said...

The notion of 'likely to be true' that's needed is a bit obscure.

On some natural readings, some beliefs are reliably formed but not at all likely to be true. For example, if I know that not-p, and also that the newspaper contains a unusual misprint reporting that p, then your belief that p, based on the newspaper, is justified and reliably formed, but not, from my point of view, at all likely to be true.

So I'm nervous about the proposed explanation for the badness of inconsistency with evidence. (My favorite argument from Williamson, which is related, takes the other side: we shouldn't think that my evidence ever eliminates, in the sense of entailing the negation of, the truth.)

Clayton said...

I think that you're right that 'likely to be true' talk is tricky, but I think that in many contexts if I say 'p' and you say 'But our most reliable methods tell us that ~p', I have a reason to be concerned. It makes it harder for me to assert properly that p is the case and it doesn't seem we need to assume that truth is required for evidence to make sense of this.

Of course, having said that, I'm happy if W's argument works. I agree with him that only true propositions can constitute evidence. I've been reading Goldman's response piece where he seems to offer responses along the lines of the one I offered that are intended to show that someone who takes evidence to consist of whatever we justifiably believe without inference can explain the data that motivates W's truth requirement. Some of the points Goldman makes strike me as plausible, but I think he's trying to save a sinking canoe by flailing around with the paddles.