Tuesday, May 22, 2007

Some Rough Thoughts on Sosa on Epistemic Normativity

From Ernest Sosa's paper on Epistemic Normativity (PDF) at the OPC2:

In calling a belief knowledge, we evaluate it positively by epistemic standards. Within the domain of epistemic assessment, knowledge has a standing higher than that given to its constitutive belief by its mere truth.


I'm always puzzled by this sort of claim. If I say I know that the sun rose today, amy I really evaluating it "positively by epistemic standards"* and giving it a "standing higher than that given to its constitutive belief by its mere truth"? I don't think so. Take the pair:

I know that the sun rose today.
I believe, but do not know, that the sun rose today.


Sosa (and he is not alone) thinks that we hold the former to exemplify more of a some epistemically relevant value, give it a higher standing. This may sometimes be true, but I see no reason to regard it as generally true. After all, it doesn't really matter in most cases whether you know that the sun rose or merely believe it; as far as the thought itself goes, you aren't missing out on anything either way.

I think my doubt on this point is related to my skepticism about another bit of common wisdom, namely, that knowledge (as usually understood in these discussions, namely, as a particular kind of belief) is, as knowledge, more valuable than alone. I see no reason to accept this. The answer to the question, "What makes knowledge more valuable than belief?" is, "Nothing in general, although there certainly are cases of knowledge that are more valuable than belief."

I suspect that the reason people think that knowledge must in itself be more valuable than mere belief in itself is that in inquiry we often try to move from belief to knowledge. That suggests to them that there is something intrinsically valuable about knowledge. But in fact, no one cares about most things that could be known, and everybody implicitly recognizes that it's a waste of time to try to know everything you believe. I believe that there is a clean pair of socks in the drawer not far from me at the moment. I have no need to come to know it; I will have gained not one single thing from the mere fact of having come to know it; I am not in the least inclined to try to know it; when I come to know it, it will be as an entirely incidental result of looking for actual clean socks. I could get something of value from looking in the drawer right now, thus coming to know that there are clean socks in it. I would on its basis be able to formulate a better estimate, for instance, of how long I have before I have to face a choice between doing laundry or going without clean socks, if I needed to do so. But it might not be important to do so. Knowing (again, in the sense usually found in these discussions) adds no 'higher standing' to the belief; it just facilitates doing other things in which certainty is useful.

Thus we can regard truth as the 'fundamental epistemic value' without any worries about the value of knowledge. The answer to the question Sosa asks later,

How can the truth-reliability of an epistemic source give to the beliefs that it yields any additional epistemic worth, over and above any that they already have in virtue of being true?


is that it can't, and we never had particularly good reason to think it could. It can give it other kinds of value, of course -- e.g., practical usefulness.

Likewise, although Sosa thinks that it is central to virtue epistemology that "the value of apt belief is no less epistemically fundamental than that of true belief," I, who certainly am a virtue epistemologist, deny this, since I think the value relevant to apt belief is simply a different sort of value from that relevant to true belief. To speak of them as if they were values in a domain that can be treated with univocal label is simply a mistake. My virtue epistemology has to do not chiefly with value at all, but with what I think knowledge and belief in fact are in the knowing and believing subject (namely, dispositions of a certain sort, differentiated by their potential roles in the fullness of a happy and flourishing human life as structured by certain ends). Any views about value are wholly derivative and presuppose this type of account; they are not central. And I think this is actually true of everyone's account, by nature; you can't seriously talk about the value of apt belief or true belief or knowledge without having any notion of what apt belief, true belief, or knowledge are, because different accounts of what they are will constrain what possible accounts of their value are available to you. Sosa's talk about value presupposes such an account, one in which knowing and believing are performance-like and where justification is a key differentiating factor. Someone who holds such a view will tend to give normativity a key role and the temptation to treat this as something like obligation will hover in the air above him even if he does not succomb. Such a 'virtue epistemology' is in constant danger of turning into a deontological reliabilism. And virtually all modern 'virtue epistemologies' give justification this key role. For someone who holds a broadly Thomistic virtue epistemology, to give something to contrast them with, the emphasis is not put on justification but on cultivation of human potential insofar as we are rational animals; prudence, manifested in some way as an 'illative sense', does most of the work Sosa wants normativity to do; and the preferred way of classifying things as knowledge and belief is different. On such a view, questions about justification, understood as conforming to certain standards, are simply the wrong questions to start with (of course, in particular domains questions about justification may arise as derivative questions), because such justification is not a foundational epistemic notion.

----
* Of course, 'epistemic' here must be taken more loosely than its strict sense of 'pertaining to knowledge'; if it were taken in that sense, and mere belief could not in itself be evaluated in terms of 'epistemic value', not being epistemic at all. But that's not the way the term is used in this context. It's a worthwhile question to ask what it should be understood to mean at all in a context like this, but I won't ask it here.