What is Evidence?

Timothy Williamson claims that something is evidence for a hypothesis when it speaks in favor of it (raises the probability of it) and it has some credible standing. The credible standing that Williamson finds necessary for evidence is knowledge. The claim is that one’s total possible evidence is comprised of what one knows and only what one knows constitutes one’s evidence at a time. In other words, S’s evidence is S’s knowledge.
S’s total evidence with regard to a hypothesis then is that knowledge S has which raises the probability of that hypothesis (the hypothesis’s probability is higher when it is conditionalized on S’s knowledge).

Williamson’s account is overly restrictive in what it requires for credible standing. Suppose that Joe has a perceptual experience of a blue book being in front of him and has good reasons to trust his perceptual faculties in this case. Presumably, and intuitively, Joe has evidence that there is a blue book in front of him. However, on Williamson’s account, this is only the case if Joe knows some propositions that support this claim. Though Joe may know a number of propositions that support his perceptual faculties being reliable, he can fail to know that he seems to see a blue book in front of him by failing to form that belief (though it is justified and true). This kind of doxastic failure seems possible, yet if it is possible, then Joe does not know that he seems to see a blue book in front of him, so his perceptual experience does not produce any evidence for him on this score. As such, Joe’s evidence would not support the proposition that there is a blue book in front of him according to Williamson’s account. This is the intuitively wrong result and shows that Williamson’s account is overly restrictive in this way.

Richard Feldman defines a person’s total possible evidence as all and only that information that is stored in that person’s mind at that time. Of this set, a person’s total evidence is that part of their total possible evidence that is available (meets some psychological accessibility constraint) and acceptable (meets some epistemic acceptability constraint). Feldman sees the evidence that passes the accessibility constraint as the evidence that S is currently thinking of (the conscious and perhaps unconscious beliefs, as well as the non-doxastic mental states that one is aware that they are in). Evidence that passes the epistemic accessibility constraint are those available items that are, or could be, justifiably believed.

Feldman’s account is not overly restrictive in requiring knowledge as Williamson’s account is. Applied to the case of Joe above, even if Joe does not form the belief that he seems to see a blue book in front of him he is nonetheless justified in so believing. As such, this is part of his evidence on Feldman’s account. Therefore, Joe’s evidence does support there being a blue book in front of him – the right result. In addition, Feldman’s account is not too lax by allowing mere beliefs to count as evidence. If mere beliefs counted as evidence (as in some coherentist theories), then propositions that one had no business believing would affect what that individual epistemically ought to believe – and this cannot be. Having the epistemic acceptability constraint set to justification avoids being too lax and too strict. Feldman’s account of evidence squares with our intuitions about what evidence a person has at a time.

An objection to this account claims that according to it one’s evidence can support a proposition even if that individual has important counter-evidence stored in his mind but is simply not thinking about it – particularly if it would be easy for him to recall these things. In such a case there is something wrong with the individual believing the proposition, but he is believing in accord with his evidence according to Feldman’s analysis.

The problem with so believing, however, is not that the individual is really failing to believe according to his evidence, but that he failed to act responsibly in forming his belief – he failed to think carefully about the matter and call to mind the relevant information stored in his mind. One can believe according to the evidence and still be blameworthy for the way that they conducted their investigation. This is such a case. Nothing here counts against Feldman’s account of evidence possession.

Another objection to this account of evidence is that according to it, there are many propositions for which we currently do not have any evidence (there is nothing we are currently entertaining that pertains to them) yet intuitively we know some of these propositions to be true. Take for example the proposition that Bush is president. Before it was mentioned, you probably had no thoughts about the matter, yet intuitively you still knew that Bush was president. Such a case seems to go against Feldman’s account of evidence, but the apparent problem can be explained away.

We can distinguish occurrent and dispositional senses of knowledge. Whereas what one occurrently knows is determined by the evidence one possesses, what one dispositionally knows is determined by what evidence that individual would possess were he to think about it. Thus, you dispositionally know that Bush is president since were you to think of it you would possess evidence that supports that proposition – you would recall having heard on the news that Bush is president, having watched his inauguration, experience a feeling of confidence that the proposition is true, etc.. Thus, distinguishing these two kinds of knowledge can account for why we think that you know that Bush is president even though your evidence does not support this proposition – you only dispositionally know it.

1 comment:

Tim Wilkinson said...

Doesn't your account of dispositional knowledge leave out the disposition to draw the conclusion and to do so by the(/a?) justified inference or chain of inferences? (Or substitute for inference your preferred means of deriving the hitherto unknown from the known.)

E.g. Watson had all the same information as Holmes, but didn't know, even dispositionally, some things which, arguendo, we assume Holmes did know.

I suppose I'm suggesting that the account of dispositional knowledge must include psychological capacities or even dispositions as well as merely logical relations between latent evidence and potentially derived knowledge.

Maybe dispositional knowledge would then just end up being defined as satisfaction of a counterfactual conditional with 'if evidence E were brought to mind' as the antecedent?