Wednesday, December 17, 2008

On balance?

Another delayed flight. I'm now in San Diego waiting for the pilots to arrive. I'm reading Conee and Feldman's new piece, "Evidence". It answers some of the questions I've had about their view, but it still leaves loads unanswered. They seem to hedge a bit about whether evidence is propositional. They don't come out and say whether they think evidence must consist of true propositions. They say that they reject E = K, but for reasons that strike me as obscure. Anyway, here's their latest statement of their view:

E: S is justified in believing p at t iff S's evidence at t on balance supports p.

I'm pretty sure I don't think this is right because I'm pretty sure that on every reading of "on balance", the right side could be true while the left is false. But, what does it mean to say that someone's evidence "on balance" supports a proposition? Any suggestions? If the conditional probability of p on someone's evidence is greater than .5, does that mean that their evidence "on balance" supports p? If that's not sufficient, what is?

4 comments:

John Turri said...

I've not yet seen this article. Where's it published?

Utterly abstracted from context, it seems like there's a reading that doesn't succumb to your worry.

Suppose 'on balance' modifies 'evidence', so that it's understood basically as a total evidence condition. (Misplaced modifier--it happens.)

Whether E supports Q is kind of like whether someone is tall. It's vague. You need to be tall enough to be tall simpliciter. E must support Q enough for E to support Q simpliciter.

Clayton said...

Hey John,

It's in the Quentin Smith collection, my friend Philippe brought it to my attention.

I'm not certain I understand your suggestion because I had intended "total evidence" in talking about conditional probability on the evidence. To my ear, to say that the evidence "on balance" supports some hypothesis, h, suggests that (a) there is evidence that supports h and (b) whatever evidence there is that supports ~h is 'balanced' (matched and outweighed) by evidence that supports h.

But, this strikes me as highly implausible. Anyway, like I said, I'm not certain I've understood your proposal. Unfortunately, I don't think I understand C&F's proposal either! But, they don't do any of us any favors because they don't give any gloss on what they mean.

John Turri said...

It sounds like you got half the proposal right. Understand 'on balance' to gesture toward the total evidence condition. Then understand 'supports' in such a way that passing some vague threshold clearly beyond .5 is required to support a claim.

Clayton said...

Okay, I see that now. That makes some sense. I've thought about emailing C&F and asking them. I'm sort of tempted to think that the best we can do is give a normative account of 'sufficient' evidence. Our sense of whether S has sufficient evidence reflects our sense of whether we think S has done everything that can reasonably be expected of them to determine whether the relevant proposition is true. I doubt that they'd be happy with that, however.