Stanley on Certainty and Possibility

In his very interesting “Knowledge and Certainty” (Phil Issues, 2008), Jason says the following things to which we might want to refer later.

A.  “A person’s belief satisfies the property expressed by a subjective use of “certain” relative to  a context if and only if that person holds that belief at or above the contextually salient degree of confidence; mutatis mutandis for epistemic certainty and degrees of justification.”

B. “Just as many beliefs may satisfy “certain”, many beliefs may, in context, satisfy “absolutely certain”. The semantic function of “absolutely” is to raise the degree on the scale above that for “certain”. So in any context, it will be harder for a belief or a proposition to satisfy “absolutely certain” than it will be for it to satisfy “certain”. But it still is possible for a belief to satisfy “absolutely certain” in one context, while not satisfying “absolutely certain” (or even “certain”) in another.”

C. “Construed this way, the notions of certainty relevant for the norms of assertion are nowhere near as demanding as willingness to bet on a proposition no matter what the odds, or having Cartesian grounds for its truth.”

D. “In previous work (Stanley, 2005), I have argued that fallibilism does not entail that I can know that p, despite it being possible that ~p. In this paper, I have argued that while fallibilism does entail that I can know that p despite being less than certain that p, it follows from independent facts about norms for assertion that I cannot say that I know that p and am less than certain that p.”

Now, I don’t believe there are any such things as constitutive norms of assertion, or speech acts generally.  I think the only norms for speech acts are the norms for all acts, rational and moral norms (perfectly sufficient to explain all the Gricean and relevance theory truths).

And if I was inclined to believe that there were such critters—and it’s hard for me to even imagine it—I wouldn’t be the least bit inclined to accept an invariant norm of assertion.  And if I *was* inclined to accept the existence of sui generis norms of assertion and invariant ones at that, it would be a reasonable belief norm.  Still, I think I can make my objection to D without running afoul of the assertion stuff Jason likes.  To do so, though, I’ll have to pose a question with a bit of a torturous grammar.

  • Q1  If one were in context C where the contextually salient degree of confidence for a belief to satisfy the property expressed by “certain” was d=.95 and one knew p with .95 certainty, what would prevent it from being the case that ~p was epistemically possible for one?

In my previous post on this, I picked on the idea that knowledge could do the trick when the evidence that secured that knowledge didn’t (still seems incredible to me).  But I think adding the new bit about certainty in 2008 offers a bit more of a foothold to put pressure on the 2005 view.

For if prob(~p)=.05, then there’s a chance that ~p.  And if there’s a chance that ~p, then ~p is an epistemic possibility.  Hawthorne explicitly accepts the equivalence between possibilities and chances (2004: 26), and it seems pretty darned intuitive.

  • (Equiv)  The is an epistemic possibility that p iff there is a(n epistemic) chance that p.

For (Equiv) to be false, one of the following would have to be true.

(Equiv¬→)  There is an epistemic possibility that p but there is no chance that p.

(Equiv¬←)  There’s no epistemic possibility that p but there is a chance that p.

But both these sound weird, even contradictory.  That’s evidence for entailment.

It could be that what Jason said in the last post is supposed to work here too, but I don’t see it yet.  And I have tried and been unable to see how any norm of assertion business could help here (in part, because the way I put the question didn’t involve any assertions at all, it was all “in the abstract”).

So I remain puzzled by a fallibilist stricture against CKA’s, and especially by a fallibilist with varying standards for certainty (or “certainty”).  (After all, one could be a fallibilist in the non-entailing-evidence sense and hold that “certain” expressed the highest possible degree of justification.  J(p) = 1 would seem to be an at least plausible basis for saying ~p was impossible.)

PS – This appears to be the 1001st post, which is pretty cool.  A milestone for CD!


Leave a Reply

Your email address will not be published. Required fields are marked *