Monday, May 17, 2010

Knowledge isn't the norm

Suppose there are epistemic norms of assertion, belief, and practical reasoning. According to Williamson (2000), knowledge is the norm of assertion:

KA: You ought not assert p unless you know p but are warranted in asserting p if you know p.

According to Hawthorne and Stanley (2008), knowledge is the norm of practical reasoning:
KPR: You ought not treat p as a reason for acting (for some p-dependent choice) unless you know p but are warranted in treating p as a reason for action if you know p.

According to Sutton (2007), knowledge is the norm of belief:
KB: You ought not believe p unless you know p but are permitted to believe p if you know p.

There is a simple and powerful argument against each of these proposals. These norms are epistemic norms. KA says that if you don’t know p, there is a conclusive epistemic reason to refrain from asserting p or an undefeated epistemic reason not to assert p. KPR and KB tell us there are similar reasons that bear on whether to believe and whether to treat something as a reason for action when deliberating about what to do. These norms tell us nothing about the non-epistemic case for or against asserting, believing, or treating something as a reason for action. Think about KA and KPR. Does the epistemic assessment of assertion or treating something as a reason for action tell us something about the normative standing of the agent’s beliefs? Will situations arise where we say that you believed what you should have concerning p but weren’t in a position to assert that p is true? Will situations arise where we say that you believed what you should have concerning p but weren’t in a position to treat p as a reason for Φ-ing when we acknowledge that you knew that Φ-ing was the thing to do if p? It doesn’t seem that the conditions under which it’s appropriate to believe p and to express your belief by saying that p is so differ in any significant way, whatever epistemic case is to be made against the act of speaking your mind should constitute an epistemic case against forming the belief you did. It doesn’t seem that the conditions under which it’s appropriate to believe are but a proper subset of the conditions under which it’s appropriate to rely on that belief when you know the content of the belief is what matters to the relevant bit of reasoning. So, it is tempting to say that the following links hold:

BPR: If you conform to the epistemic norms that govern belief by believing p, you conform to the epistemic norms governing practical reason by treating p as a reason for action (for some p-dependent choice).
BA: If you conform to the epistemic norms that govern belief by believing p, you conform to the epistemic norms governing assertion by asserting that p is the case.

These knowledge norms tell us whether it is epistemically permissible to believe, assert, or treat something as a reason for action. According to BPR and BA, there aren’t different epistemic standards that apply to beliefs on the one hand and the expressions of belief or reliance on belief in reasoning on the other. Because we don’t rely on some two-stage process of epistemic assessment to determine whether someone who believes has the warrant to speak her mind or reason from what she takes herself to know, I think it’s clear that we assess things epistemically as if we accept BPR and BA. To be justified in believing p requires that you believe p permissibly. It can’t be that you shouldn’t believe p, but you believe p justifiably. Two things of consequence follow. First, a commitment to either KA or KPR comes with a commitment to KB. Second, a commitment to KB comes with a commitment to KJ:

KJ: You can’t justifiably believe p unless you know p.

The problem, of course, is that KJ is false. Gettier cases show that it’s possible for someone to justifiably believe something they don’t know to be true. So, we can see that knowledge isn’t the norm of assertion, practical reasoning, or belief. The problem isn’t just that there are counterexamples to KJ. The real problem is that if we acknowledge that knowledge and justified belief are distinct because we appreciate that there’s more to knowledge than just justified belief, we should see that it is incoherent to say that the additional factors that distinguish mere justified belief from knowledge should figure in epistemic norms. It is incoherent to say this because to say this, BPR and BA imply that someone can be epistemically prohibited from believing what she permissibly believes. So, if knowledge and justified belief are distinct, there’s no question that it’s justification that’s the norm of assertion and practical reasoning because it’s utterly vacuous to say that justification is the norm of belief.

Note that for all I’ve said, it might be that justified belief is a lot more like knowledge than most epistemologists currently believe. I’ve said that JB is utterly trivial:
JB: You ought not believe p unless you justifiably believe p but are permitted to
believe p if you justifiably believe p.

Combine JB with BA and BPR and we have to say that justified beliefs are whatever those things are that warrant assertion and warrant treating something as a reason for action. If you shouldn’t assert false propositions and shouldn’t treat false propositions as reasons for action, it may well be that the only real difference between justified belief and knowledge are the conditions that make for Gettier cases. There’s little that we can say about all the Gettier cases that is informative and true, but this much seems right. The cases that figure in such conditions almost always seem to involve conditions that have little significance for what we should do or what we should say in those kinds of situations, so perhaps these cases bring to light a variety of normatively insignificant aspects of knowledge.

REFERENCES
Hawthorne, J. and J. Stanley. 2008. Knowledge and Action. Journal of Philosophy 105: 571-90.
Sutton, J. 2007. Without Justification. Cambridge, MA: MIT University Press.
Williamson, T. 2000. Knowledge and its Limits. New York: Oxford University Press.

10 comments:

John Turri said...

I wonder about the move from 'ought' to 'conclusive reason'. Could you say more about that?

If morality says that you ought to not lie, does it thereby say that you have conclusive moral reason to not lie?

Clayton said...

Hey John,

I'd say that if morality says you ought not lie, it thereby says that you have a conclusive moral reason not to lie. There's some trickiness about reasons of a kind (e.g., moral, legal, prudential) and just plain 'ought'. I think I'd also be happy to say, as Judith Thomson does, that _if_ X says you ought not lie, X is claiming you have a conclusive reason not to lie, period.

Clayton said...

John,

Just a quick follow up.

I'd be curious to know, do you think that the data about prompting works for belief as well as assertion? I sort of think that it does but that's just off the top of my head.

Anonymous said...

Two comments:

1) I would think that at least some of the defenders of the knowledge norms (e.g. Sutton, Williamson) would be willing to accept KJ. You say that Gettier cases show that KJ is false. However, I'm not sure the knowledge norm defenders would agree. Knowledge norm defenders typically distinguish between when someone justifiably performs some action/holds some belief and when someone blamelessly performs some action/holds some belief. I think they would diagnose Gettier cases as cases where someone blamelessly believes a falsehood, but does not do so justifiably.

2) I'm not sure that Williamson intends KA to be an epistemic norm. He stresses that KA is intended to be the constitutive rule for assertion, employing an analogy with the constitutive rules of various games. The idea is that if we didn't have KA in place, we wouldn't have assertions, since assertions are speech acts that are necessarily governed by KA. One could coherently think that KA is the norm of assertion in this sense without making the further claim that KA tells us what it is epistemically permissible to assert.

Clayton said...

Anon,

Thanks for the comments.

I think you might be right that some who defend the K norms think of Gettiered beliefs as excusably held rather than justified. I didn't get into the issue in the post, but my response to that is essentially that I'm willing to defer to the intuitions of others on this point and I think they tend to go against the K account. My own view on the matter is that in some G cases there's no wrong to excuse or justify and so there's an important difference between reasonably held T beliefs in the land of fake barns and reasonably held F beliefs.

On point 2, you're right that Williamson takes KA to be a constitutive role of assertion, but that seems consistent with further claims about what oughtn't be asserted on epistemic grounds. I thought his remarks about authority and thinks like that in KAIL suggested he thought that there was a decisive epistemic case against any assertion not known to be true _and_ that there's a constitutive rule that enjoins us not to assert what we don't know.

I always thought that the claim that knowledge is an epistemic norm was more plausible of the two and wondered if that might be the direction the literature was going in.

Jeff said...

Suppose I assert to you that "one of the waiters at Denny's is rude". You learn that I formed this belief after a particular waiter at that Denny's was rude to me this morning.

The same afternoon you go to eat at Denny's, and discover that there is indeed a rude waiter there. However, you learn that this waiter was just hired minutes ago to replace the rude waiter I encountered, who died unexpectedly during his smoke break.

The next day, you overhear me telling people, "one of the waiters at Denny's is rude".

Wouldn't BA entail that there's nothing wrong with me going about telling people this? Does this seem right? I guess it seems initially to me that we can think of situations where I believe what I should have concerning p, but my asserting it is criticizable.

Clayton said...

Jeff,

That's an interesting example, but here's what I might say. Is there something wrong with saying it? Sort of. It might create the false appearance that the waiter still alive is rude (or one waiter still alive is rude). But, that's not what you said, it's just something someone might pick up from the assertion. I'd say that it's akin to a case where there's false implicature generated by a true assertion. So, maybe what's really wrong is not that the thing was asserted but that it was asserted and there's this unfortunate side-effect that hasn't been tidied up. Something similar goes for the belief. There's nothing wrong with the belief, per se, but there's an unfortunate side-effect that hasn't been tidied up, someone might have something against the wrong waiter.

Jeff said...

Hi Clayton,

Thanks for the response. So, let's suppose my utterance "one of the waiters is rude", while true, has some secondary uptake like (which is false, because he is dead).

However that pragmatic element is cashed out, I'd assume it involves the presumption that the speaker's assertion of p follows a certain rule -- that his reasons for believing p ought to have the right sort of explanatory/causal connection to p. If that's the case, then there is a kind of anti-Gettier "norm" for assertions. Violating that norm makes an assertion defective in the way that an assertion which violates one of the conversational Maxims is defective.

However, I see now I've likely gotten off on the wrong track. The sort of "knowledge norm" for assertions which I had in mind is more like the norm that you ought to wear fashionable clothes to expensive events. It doesn't follow that you ought not to go to expensive events if you don't wear fashionable clothes, since you might have important reasons for going, or you might be blamelessly mistaken about what's fashionable. What Williamson needs is that lack of knowledge be a conclusive reason not to assert something, not just a defect or flaw in assertions. And that's implausible for the reasons you give, that there are aspects of knowledge which aren't any use in deciding what to do since we can't be aware of them. Have I followed you right?

Still, I would resist calling the non-JTB aspects of knowledge "normatively insignificant". I think a norm can be significant for judgment, criticism, praise, etc., even if it doesn't change what an agent should have done. It's a defect in an ice skater's performance if she falls during her routine, even if the fall was unavoidable due to a crack in the ice. Does that sound right to you?

Clayton said...

"However that pragmatic element is cashed out, I'd assume it involves the presumption that the speaker's assertion of p follows a certain rule -- that his reasons for believing p ought to have the right sort of explanatory/causal connection to p. If that's the case, then there is a kind of anti-Gettier "norm" for assertions. Violating that norm makes an assertion defective in the way that an assertion which violates one of the conversational Maxims is defective."

I think a lot rides on how we read "right sort of explanatory/causal connection". Right for knowledge? Right for permissibility? To the extent that we think of the assertion as expressing a justified belief, a belief that stands in the right explanatory/causal connection to be permissibly held, the question will be why we shouldn't think this connection suffices for permissible assertion.

"And that's implausible for the reasons you give, that there are aspects of knowledge which aren't any use in deciding what to do since we can't be aware of them. Have I followed you right?"
That's the idea, but I wonder if someone would want to concede this but say that there's just a pro tanto reason not to assert what you don't know.

"It's a defect in an ice skater's performance if she falls during her routine, even if the fall was unavoidable due to a crack in the ice. Does that sound right to you?"

That might be right and it can be cashed out in terms that aren't deontic. We could cash it out in evaluative terms, perhaps.

Btw, if I want to credit you for these points in a written version, should I just refer to you as "Jeff"?

Jeff said...

Thanks again for the response... Hmmm. Why should the conditions for permissibly believing p not also suffice for asserting p, insofar as asserting p involves expressing a belief that p? I guess my answer would be something like this: asserting p does something in addition to expressing a belief that p. It also produces testimony which sometimes entitles my audience to assert that p to third parties.

You are often entitled to tell others "Smith killed Jones" if I asserted to you "Smith killed Jones". But if I tell you "I believe Smith killed Jones" (expressing a belief), then you aren't entitled tell others "Smith killed Jones" without further evidence, you can at most tell them "Jeff believes Smith killed Jones".

So, even if the conditions sufficient for permissible belief are met, it seems plausible to me that there are further conditions which assertions should meet in order to produce better testimony, or better overall chains of testimony from person to person within a community (which mere expressions of belief don't have to meet). I wonder sometimes if the Gettier condition fulfills a role of this sort... but I guess it's hard to say much more than that...

Jeff Watson

[grad student/blog reader
Arizona State University]