contrastivism and Hawthorne’s principle of practical reasoning

Contrastivism holds that the truth makers for knowledge attributions always involve a contrast, and Hawthorne thinks that if you know something, you are entitled to use it in practical reasoning. So one way to test what it is known is to see what kinds of practical reasoning we’ll allow are acceptable.

Depending on what the contrast is, contrastive knowledge may be easy or hard to have. So, it is easier to know “the train will be on time rather than a day late” than it is to know “the train will be on time rather than 2 minutes late.” One way to put the difference is that one is presupposing more in knowing the first claim that one is in knowing the second.

Consider then a piece of practical reasoning using the following conditional: if you are pointing a gun at me, and if your gun is loaded and if you intend to shoot me, I should shoot you first. Suppose I know that you are pointing a gun at me rather than a twig, and that I know that your gun is loaded rather than having just been disassembled for cleaning, and suppose I know that you intend to shoot me rather than give me a million bucks. Should I shoot you? Maybe this is an anti-gun sentiment coming out, but I think it is far from obvious that I should.

Compare this case to another. In this case, I know that you are pointing a gun at me rather than any non-lethal item, and I know that your gun is loaded rather than merely having the appearance of being loaded from where I stand, and I know that you intend to shoot me rather than anyone else in the universe. Now I think I should shoot you first.

Why the difference? In contrastivist language, I’m presupposing too much in the first case, I think. The knowledge I have is easy knowledge because it presupposes so much. I’m presupposing that the thing in your hand is either a gun or a twig, that it’s either loaded or disassembled, that you intend to shoot me or make me rich. In the context of these assumptions, it’s too easy to come to the conclusion that I should shoot first and ask questions later. In the second, case, however, my presuppositions are much broader, broad enough that my knowledge is no longer easy. And since my knowledge is not easy, I doubt I could be faulted on grounds of rationality for shooting first.

It appears, then, that contrastivists will have to deny Hawthorne’s principle. Moreover, I don’t see any obvious way to qualify the principle for the following reason. If the action is relatively inconsequential, then easy knowledge may be enough to warrant performing the action. But if the action is immensely significant, as it is in the case of taking a life, then easy knowledge doesn’t seem to be enough.

One way to think about such cases is that they may provide a reason for including pragmatic issues in one’s account, either indirectly as contextualists typically do or directly as we find in the invariantist camp. Or maybe a reason for rejecting Hawthorne’s principle…?


contrastivism and Hawthorne’s principle of practical reasoning — 6 Comments

  1. Hi Jon,

    I would agree with Hawthorne that being known entails being available to practical reason. Would you? It seemed so until the very end…

    In any case, I think this connection supports contrastivism. To work with your train case, suppose that I need to make the train, but would really like to get a coffee first, which would only make me 2 minutes late. If I know that the train will be on time rather than 2 minutes late, then I had better not risk it. While if what I know is that the train will be on time rather than 24 hours later, then I might well get that coffee. What action plan my reasoning supports seems to vary with the contrasts.

    In your gun case, if you know that Psycho’s gun is loaded with bullets rather than empty, but don’t know that Psycho’s gun is loaded with bullets rather than loaded with harmless origami cones, then you should weigh the risks. Which seems like the right result. Perhaps Psycho has a history with origami? In any case, knowledge is not all that is available for practical reasoning. Probable opinion is too, suitably weighted.

  2. Jonathan, yes, I think there is something right about Hawthorne’s principle, though I waver on it. Remember that he uses it to defend pragmatic encroachment in epistemology, and I’m opposed to that.

    So, what I was thinking was that the contrastivist examples give a good reason to be very careful about the kind of connection between knowledge and practical reason. I think the two shooting cases show that, at least for contrastivists, you can’t read off what you don’t know from noting what you’re not entitled to do. And my hesitation about Hawthorne’s principle is a generalization of this point, since I view contrastivism as suitably insulated from pragmatic encroachment.

    I think your response suggests the following: what you know you’re entitled to use in some practical reasoning or other, but when you’re not entitled to an action, you can’t directly conclude anything about what you know or don’t know.

  3. That sounds right to me, John. When one is not entitled to an action, this does not directly show anything about what one does or doesn’t know.

    Anyway, you have gotten me thinking about how exactly the contrastivist might formalize the knowledge-praxis link. This is off the cuff, but I think it has to run something like this:

    If s knows that p rather than q, then s is entitled to lump all of her credence in the problem space (the union of p-worlds and q-worlds) onto the p-worlds.

    So when s calculates expected utilities, her knowledge that p rather than q entitles her to assign 0 probability to the q-scenarios.

    So for instance, to continue working with the train case above, when s compares the expected utilities for (i) waiting for the train scheduled at 12, versus (ii) grabbing a coffee and getting to the platform at 12:02 , it will matter a great deal whether s is entitled to assign 0 probability to the scenarios in which the train arrives at or after 12:02. So if s knows that the tran will be on time rather than 2 minutes late, she is entitled to assign 0 probability to these scenarios. While is all s knows is that the train will be on time rather than 24 hours later, she isn’t.

  4. Jonathan, I like this approach, and I see an argument of sorts here for contrastivism against contextualism and invariantism. The argument against contextualism wouldn’t be any different from Hawthorne’s, and the argument against invariantism is that the contrastivist solution is preferable because it insulates the purely epistemic character of the cognitive state underlying action.

    One issue, though. The kind of principle Hawthorne wants is one whose instances are conditionals that function in practical deliberation. The reason for this is so that when we judge an action unwarranted, we’ll be able to infer something about what the person doesn’t know. From the above, I take it that your idea would be to let the conditionals result from expected utilities (so that if doing A has the highest expected utility, then the conditional “if you know p rather than q, do A” will be acceptable.) Such an account will negatively affect Hawthorne’s argument, since he wants to be able to use a prior assessment of what is rational to do or not in a situation in order to “read off” what is known or not. If we have to calculate expected utilities before constructing the conditional, we have the order backwards from what Hawthorne wants. Maybe that’s too bad for Hawthorne, but it would be nice to have a knowledge/praxis theory that could be used in this way.

    Maybe we should do something like the following? If it’s irrational to do A, then for any p and q such that Pr(p)=1 and Pr(q)=0 would imply that its rational to do A, it is not known that p rather than q. This isn’t quite right, but maybe something along these lines will let us use judgements about appropriate action to determine what we know (or, more accurately here I think, what contrasts we have and have not ruled out).

  5. I see what you mean, Jon, and I like your final suggestion. Just one minor modification.

    From: (1) Irrational: A
    (2) If (pr(p)=1 & pr(q)=0) then Rational:A
    All that follows is: (3) ~pr(p)=1 v ~pr(q)=0
    And (3) is compossible with knowing that p rather than q, at least on my view, since I would not require absolute certainty of p, but only relative certainty of p rather than q.

    So I would suggest that the modification should run:
    From: (1) Irrational: A
    (2) If pr(q)=0 then Rational:A
    Then: (3) s does not know that p rather than q
    From (1) and (2) we get ~pr(q)=0, which is essentially to say that q is uneliminated, which precludes any knowledge of p rather than q.

    This would formalize a contrastive link from irrationality to ignorance. What do you think?

  6. Oh, yes, I just noticed yesterday afternoon when I was looking at the paper you sent, the section on whether simply ruling out the contrast was sufficient for contrastive knowledge. I see the point, and you’re emendation fits that idea well. Plus, Pr(p)=1 gives you contrastive knowledge for any contrast whatsoever!

    I’m going to do another main entry on contrastivism today, conceived as a theory of understanding instead of a theory of knowledge (not that the two are incompatible, though). It’s motivated in part by reading Stalnaker’s comments on you PS paper, and finding, especially the linguistic parts, to be illuminating but not–how to put it–central? That’s not quite right, but at least not getting to the heart of the theory.

Leave a Reply

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