June 15, 2004

Six Reasons to Do Epistemology

Here are six questions that you might want epistemology to help you with. They're arranged in order, though in order of what I'm not sure, and they're obviously not exhaustive.

(1) What actions should we take in order to ensure that we have the best beliefs we can possibly have?

(2) What actions should we take in order to ensure that the things we tell other people are the best things we can tell them (where those actions include telling them or not telling them things)?

(3) What belief-forming processes are good for a given individual to have, or for a species to have, or good in general?

(4) What beliefs are good for a given individual to have?

(5) Given the evidence that an individual has, to what extent does that evidence support certain propositions?

(6) What beliefs are good to have in general, irrespective of the believer's individual position?

You will noticed that I haven't included old chestnuts like "What is knowledge?" or "What is justification?" That's not just because the list isn't exhaustive--it's also because when we've come up with a conception of knowledge or justification, the question may remain: Given that conception, (why) should we care about whether we can attain knowledge/justification? My view is that the answer, with respect to contemporary conceptions of knowledge, is that each of the questions (1)-(6) by itself motivates the importance some concept other than knowledge. In particular, I think the answer to (6) is probably either "Good beliefs are true" or "Good beliefs are beliefs that are true and supported by an understanding of why they are true." This latter is a conception of knowledge, but it's not the contemporary post-Gettier one; it's more like the Platonic conception that Michelle pushed a while back.

(So if I don't think that any of these questions in itself motivates the importance of knowledge, do I think that knowledge is unimportant? No--I think knowledge is important as a one-stop shop for a useful approximation of answers to several of these questions. To say "S knows that P" is to say quickly that S's belief that P does pretty well with respect to at least some of those six concepts.)

My favorite question here tends to be question 5, because that's what I see as the purely epistemic one. If we want to exclude pragmatic factors, it seems to me that we should also be excluding factors such as "How much work will it take to carry out this epistemic procedure?" and "What are the procedures that I can carry out given my cognitive limitations?" And questions (1)-(4) seem to me to depend on these factors--for instance, (1) depends both on what I can manage to carry out and whether it's worth the effort to get the best possible information, (3) depends on our cognitive powers, and (4) may well depend on the natural history of our species. ((6) abstracts not only from our information-processing limitations but from our information-gathering limitations as well! So I think it's pretty well transcended the epistemic.)

You'll notice that this means that, unlike Jonathan Kvanvig, I am happy to say that knowledge isn't purely epistemic. In fact, if it weren't for our information-processing limitations, we wouldn't have any use for the concept of knowledge (on my theory)--if knowledge is a compromise among several independently important desiderata, creatures without cognitive limitations could just keep track of all those desiderata, without shorthand.

This is also what I meant by my gnomic earlier comments that perfect information-processing machines wouldn't have many categorical beliefs. Question (5) abstracts from our information-processing limitations, and it rarely produces the result that our evidence gives perfect support to a belief. Usually, could we get the answer, it would be of the form "On this evidence, P is very very very likely." (I don't want to commit myself to the idea that this could always be expressed by a probability.)

Up till a couple of days ago I would have said that the answer to (5) is the answer to the question "How much does our evidence support a certain belief?" But it is probably better to reserve the word "justified" for question (4), so that a belief that is justified is a belief that the subject should have. And Matt McGrath has convinced me that not to use "justified tout court" in your answer to question (4) and degree of justification in the answer to question (5). Pragmatic factors enter into (4) so that different degrees of evidential support may be required for justification of different beliefs (or the same belief for different people)--the worse the consequences of false belief, the more support is needed to justify false belief. Then one belief can be justified and another not even though the first has less evidential support--to say that the second is "more justified" just confuses the issue.

Well, that's pretty much an explanation of the entire framework in which I think about epistemology. Let me end with a plug for Certain Doubts, a new epistemology group blog with Kvanvig, McGrath, a frightening array of other contributors, and a substantial omission in its blogroll.

Posted by Matt Weiner at June 15, 2004 11:49 AM