Wednesday, March 10, 2010

'Significant Possibilities' and Concessive Knowledge Attributions

Suppose you think that it's possible to know that p, even though your epistemic position vis-a-vis p is weak enough for 'it might be that not-p', in its epistemic reading, to be true. I don't really see why you'd want to think this myself, but I guess some people think that (a) this is a good reading of 'fallibilism' and (b) fallibilism is true. If you think this, then you face the problem to explain the infelicity of concessive knowledge attributions. Why's it sound so bad to say "I know that p but p might be false"?

The obvious explanation is that it's a contradiction: according to standard epistemic modal logic, 'might', in its epistemic reading, is just the dual of 'know'. But the fallibilist of this stripe has closed off that response. What's he say instead? Dougherty and Rysiew propose a pragmatic line: "p might be false," they say, implicates but does not entail that there is a significant chance of not-p. And while a chance of not-p is consistent with knowledge that p, a significant chance of not-p is not. Fantl and McGrath supplement the story by suggesting that the significance of various chances can be a stakes-sensitive matter; the same possibility, with the same likelihood, can be significant if the stakes are high, and insignificant if the stakes are low.

Now I get nervous when Gricean pragmatic stories are asked to do work like this. Too often, the data don't generalize the right ways. Here's one problem: the pragmatic effect doesn't seem appropriately cancelable. Consider:

It's possible that it will rain today, but I know it won't rain today.

The badness of this sentence is explained, on the view in question, by suggesting that the first conjunct pragmatically implicates that there is a significant chance that it will rain today. It predicts, then, that if we cancel the implication, we're left with felicity. But this prediction is not borne out; this is still bad:

It's possible that it will rain today, but there's no significant chance that it will rain today, so I know it won't rain today.

Also, there's a point that Derek Ball raised in Jason Stanley's seminar last week, inspired by Seth Yalcin: the infelicity of concessive knowledge attributions persists in non-assertoric contexts. "Suppose that you know it will rain today and it might not rain today." "If you know it will rain today and it might not rain today, then you know something that might not happen." Etc. The Gricean story is peculiar to assertions, and therefore insufficiently general.

I think there's a better view in the same spirit. (Well, maybe in the same spirit; I'm not quite sure what the intuitive motivation behind this project is. My suggestion won't vindicate the coherence of concessive knowledge propositions. But like I said, I'm not sure I see why anyone would want to do that.) The line we've been considering is one in which "there is some possibility of p" pragmatically implicates that there is some significant possibility of p. But the existential quantifier is going to have a context-sensitive domain restriction anyway. We could suppose that in the relevant contexts, we're only quantifying only significant possibilities. Then "there is some possibility of p" would, in the relevant context, entail that there is some significant possibility of p.

On this approach, you can still get a lot of the stuff that Fantl and McGrath want. On this view, whether there is a possibility of p will depend on the stakes, since all possibilities are significant possibilities, and whether a possibility is significant depends on stakes. So their 'impurism' would infect 'possibility' talk too. (This is not a result of the view they actually offer, which I'm criticizing: they have 'pure' possibilities, where talk of them implicates results about 'impure' significant possibilities.) But the concessive knowledge attributions will be genuine contradictions.

7 comments:

  1. Hi Jonathan,

    You say, "this is still bad:
    It’s possible that it will rain today, but there’s no significant chance that it will rain today, so I know it won’t rain today."

    The problem for me is that this seems perfectly natural, as do all of the following:
    "Of course it's possible that not-p. Heck, anything's POSSIBLE. But we know it aint' gonna happen."
    "Yeah, there's a chance that not-p. But c'mon, it's completely insignificant. You know it's not true."
    "I know that the small chance that not-p won't obtain."

    Or, more specifically,
    "Of course it's possible that the large hadron collider will produce black holes. Heck, anything's POSSIBLE. But we know it ain't gonna happen."
    "Yeah, there's a chance that not a single free will be made in the upcoming NBA playoffs. But c'mon, it's completely insignificant. You know it's not true."
    "I know that the small chance that I'll die in the next 2 seconds won't obtain."

    I don't, then, take difficulties with the above to be problems for fallibilism. I take the naturalness of the above to be data on par with the unnaturalness of "I know that p but p might be false."

    I'm not sure that the view you propose allows for the naturalness of all of these. Dougherty and Rysiew's view does. Now, I gather, you don't think that these sound natural. But what sort of theory should someone accept who DOES hear them as perfectly natural? It looks (putting contextualist moves aside) like there is going to have to be a Gricean explanation somewhere -- either for these or for the simpler concessive knowledge attributions. Fallibilists, obviously, should be interested in plumping for the Gricean explanation for the original concessive knowledge attributions.

    ReplyDelete
  2. I'm not sure what you mean when you say that we should be 'putting contextualist moves aside' -- what I've just sketched in this blog post IS a contextualist approach. The epistemic modal takes a context-sensitive base, and a natural reading of it limits that base to significant possibilities.

    And the structure of this approach does end up allowing for quite a lot of flexibility. If you want there to be good readings of concessive knowledge attributions, then I can give them to you: they're readings in contexts in which non-significant possibilities are indeed being considered. I'm explaining the infelicity of such sentences by saying that typically, the contextual parameter is one that excludes insignificant possibilities; this is consistent with this not being mandatory in all contexts.

    ReplyDelete
  3. Thanks, Jonathan. That's helpful. But you do conclude by saying that the concessive knowledge attributions will be genuine contradictions. Are you saying that I shouldn't be interpreting this conclusion as the following?

    1) All concessive knowledge attributions are genuine contradictions.

    Or is that you do want to be saying this, but also saying that claims like,

    2) I know that at least one strike will be thrown in the upcoming baseball season, though of course there's a non-zero but insignificant chance there won't be.

    are not genuinely concessive knowledge attributions (because, perhaps, the relevant features of the context change during the course of the utterance)?

    If you're comfortable saying that 2 is both literally true and literally a concessive knowledge attribution, then I think I can get on board. But, if that's the case, then I don't see how possibilities end up inevitably impure (though if contextualism about 'possibility' is true, 'possibility'-talk might be).

    So, clarificatory questions: are you comfortable saying that the proposition I normally express by an utterance of 2 is true? And are you comfortable saying that there is no relevant contextual shift from the assertion of the first conjunct to the assertion of the second?

    ReplyDelete
  4. I don't really care to fuss about what is or is not a genuine concessive knowledge attribution. I'm just taking as paradigms of this theoretical term the sorts of examples of sentences that are intuitively problematic, but which a certain kind of fallibilist might feel pressure to endorse. I say that the reason they're intuitively problematic is that they're contradictory; that's consistent with there being other similar sentences -- maybe even the same sentences in different contexts -- that are not contradictory. Whether you want to call those latter sentences CKAs doesn't much matter to me.

    So, clarificatory questions: are you comfortable saying that the proposition I normally express by an utterance of 2 is true?

    I mean, to be honest, I do think it seems pretty bad. So if you're really asking me to tell you what I think is correct, I do think that (2), in typical contexts, will be a contradiction. But the point of my comment was that the view given in the post doesn't mandate that. Someone who thinks (2) typically expresses a consistent proposition can still have my view about CKAs. So to the end of demonstrating this flexibility, for the purpose of argument, I'm willing to allow that (2) is typically true. (And it is my considered view that there can be contexts in which it is true.)

    And are you comfortable saying that there is no relevant contextual shift from the assertion of the first conjunct to the assertion of the second?

    Again, there are different answers to give for the purpose of telling what my view is, and for demonstrating what the suggestion of this post can handle. My own considered view is a Lewisean infallibilist contextualist approach to 'knows' -- so my considered view is that true readings of (2) need the contextual parameter to shift mid-sentence. But that's not mandatory for the view laid out about CKAs. You could also think -- and this is much closer to what you think -- that 'knows' is invariant, and requires no significant possibilities, regardless of what possibilities are quantified over later in the sentence. Then you could think that in some contexts without any shifts, (2) expresses a truth.

    ReplyDelete
  5. Jonathan,

    Generally, I like the strategy you’re exploring. But I’d like to think (1) is true, or at least consistent:

    (1) Even though it’s strictly speaking possible that all my seventy intro logic students are going to get 100% on their logic final, I know that won’t happen.

    But (1) entails (2)

    (2) It is possible that all my sixty logic students are going to get 100% on their logic final, but I know that won’t happen.

    And if there is this entailment, this would help explain why, if I start by thinking about (1), I feel pressure to say 2 is true – though I feel I want to explain 2 further and to remove misleading implications. One way to do that is to say: yes, it’s possible but it’s outlandish; I know that won’t happen.

    I wonder if there might be a middleground between your view and the one Fantl and I mentioned in our first chapter. One might try to say that without an elaborate conversational background, a bare CKA - one in which ‘possible’ is left bare and unqualified - will be false. This would be because, without such a set-up, ‘it’s possible’ is semantically associated only with significant possibilities. Consistently with that, one might think that with the necessary conversational background – which might include the assertion of a non-bare CKA – ‘possible’ can extend to insignificant possibilities, so that the bare CKAs may be true.

    Some other analogues might be:

    “Do I have a reason to eat this piece of metal? Well, it has iron. But even though I strictly speaking have a reason to eat this piece of metal, it would be completely irrational to do so, because of all the far more powerful reasons against eating it.”

    Here we heavily qualify “I have a reason to…” Without a pretty elaborate set-up like this stressing the overall weakness of the reason, ‘I have a reason to eat this piece of metal’ will seem clearly wrong. Moreover, without an elaborate set-up like this, even when it is in embedded contexts, ‘I have a reason’ will be understood as being about minimally significant reasons. Mark Schroeder uses an example something like this in his book *Slaves of the Passions*.

    Or maybe think about Achinstein on evidence. Achinstein would say that the fact that Michael Phelps is taking his daily swim is not evidence he will drown, even though it raises the probability of his drowning. So, Achinstein concludes probability-raising is not sufficient for evidence. But I like the following reply. Without an extremely elaborate set-up stressing the weakness of the evidence – just how little it affects our overall reason to believe, ‘there is evidence Michael Phelps will drown’ – even in unasserted contexts – will be interpreted as concerning only meaningful or significant evidence. However, suppose we start by saying, “The only way you can drown is by being in the water, and even the best swimmers could always drown, and since Phelps is swimming right now, there is a teeny tiny bit of evidence that he will drown.” This seems right, and it seems to entail ‘there is some evidence Phelps will drown’.

    ReplyDelete
  6. Jonathan, I have no idea why I switched from 70 logic students to 60 in going from (1) to (2). I suppose it might be because I just gave a logic test and a bunch of my students flunked and so probably will drop. But, anyway, it should be the same number in the two statements.

    ReplyDelete
  7. Hi Matt,

    Yes, I take it the 'middle ground' sort of approach you describe is a lot like the one I intended to be making available in my previous clarification. I want to say that the infelicity of (2) is explained by contradiction; this is consistent with there being contexts in which (2) is neither infelicitous nor contradictory.

    Notice that, although (1) entails (2), it's easy to get confused here, on the approach I suggest, since the proposition (1) expresses in typical contexts for (1) does not entail the proposition (2) expresses in typical contexts for (2). So it's easy tacitly to equivocate when thinking about such arguments.

    ReplyDelete