Sunday, June 21, 2009

Allegedly inconsistent knowledge principles

Matt Weiner argues that 'our use of the word "know" is best captured by' an inconsistent set of inference rules. His setup strikes me as strange. He writes:
These are the Knowledge Principles:
(Disquotational Principle)  An utterance of “S knows that p” at time t is true iff at time t S knows-tenseless that p.
(Practical Environment Principle)  S’s evidence concerning p is good enough for knowledge iff S’s evidence for p is good enough to make it epistemically rational for her to act on the assumption that p.
(Parity of Evidence Principle) If the evidence concerning p for S and T is the same, then S’s evidence is good enough for knowledge iff T’s evidence is good enough for knowledge.

The Knowledge Principles are inconsistent, given only the truism that different people can have different practical stakes. Take a Bank Case (DeRose 1992), in which Hanna and Leila each have the same rather good evidence that the bank is open Saturday, but acting on a mistaken belief would harm Hannah much more than Leila.  Hannah is in a high-stakes context, Leila in a low-stakes context.  The Practical Environment Principle, which entails that Leila knows that the bank is open and Hannah does not, here generates an inconsistency with the Parity of Evidence Principle, which entails that Leila knows if and only if Hannah does.

Two things strike me as really strange about this claim, even setting aside the question of whether these principles are plausibly constitutive of the meaning of 'knows'.

First, there are a lot of assumptions at play beyond "the truism that different people can have different practical stakes". The following are all tacit assumptions of Weiner's argument against the consistency of these the knowledge principles:

  • What it's epistemically rational (as opposed to prudentially, etc.) to do depends in constitutive part upon the practical stakes.

  • What evidence one has is invariant across practical stakes.

  • Leila's evidence is good enough for knowledge.

All of these, however, are controversial. Given the extremely radical claim that Weiner is trying to establish, these sorts of loose ends seem troublesome.

The second thing that strikes me as odd about this argument is that the Disquotational Principle seems to be doing no work at all.


  1. Hi Jonathan
    I have read the paper some time ago. Doesn't Weiner have some more cases? it seems your worries about this argument are well motivated.
    I thought the role of the Disquotation Principle was to rule out contextualism, which would be an obvious way to avoid inconsistency.

  2. Hi Jonathan! I'm sorry that it took me ten years to find this post.

    I think one of the issues here is that I didn't define "epistemically rational" clearly. When I say "it is epistemically rational to act on the assumption that p," what I'm trying to exclude is cases like cases where you should trust your friends, or perhaps cases where your evidence that p was learned in an illegitimate way. (See p. 6 of the paper, halfway down the right side.) So that I would claim that, if my friend tells me p even though all the testimony appears to be against them, and I believe them out of the obligations of friendship (in Sarah Stroud's kind of case), then we might still be reluctant to say that I *knew* that p even if it turns out to be true. So that I think there's a sense in which the evidence can be evaluated impersonally, aside from non-epistemic factors that affect whether you ought to act on that evidence, and that's the sense at issue when I say "epistemically rational."

    But whether it's rational to rely on p in this sense is still going to be a question of prudential rationality as well! If you were using a decision theory that depended on degrees of credence and utility functions, then the "epistemically" qualifier would be about what you were using to set the degrees of credence, and the practical stakes would go into the utility function,

    The idea that Leila's evidence is good enough for knowledge isn't supposed to be an assumption here--I'm trying to take the biconditional the other way. I take it as a natural response to the example that it's rational for Leila to wait to go to the bank, relying on the assumption that the bank is open Saturday. Then I think we often (implicitly) rely on the Practical Environment Principle to conclude that Leila knows the bank is open Saturday. Or at least, if we have any inclination to say that Leila knows and Hannah doesn't, I think that's some evidence that we're relying on the PEP. (I do try to make other arguments that we're relying on the PEP, like the one on p. 7 that it sounds odd to say that someone should violate the PEP *unless* we're saying it for reasons that aren't epistemic in the sense I describe earlier in this comment.)

    Daniele is right about the use of the Disquotation Principle. It's not part of the inconsistency there, but gets used later on to explain why certain contextualist-friendly responses don't sound right.

    The point about whether evidence is invariant across practical stakes is a good one. Obviously I need to reject a view on which E = K and K depends on the practical stakes, which I'm fine with doing. I do find it intuitive that there's *something* that Hannah and Leila have in common, and that this is what gets input to the Parity of Evidence Principle. And I'm a little skeptical about whether one could explain everything that I think supports the Principles by adverting to some way in which the evidence changes when our stakes change. It remains true that each of them went to the bank two Saturdays ago and saw that the bank was open, and that that's what they're relying on if they decide to skip the bank Friday.


  3. (part two)

    In fact I think my arguments for the presence of inconsistent principles might be formulable without reference to the notion of evidence. The basic idea is that it's natural to say Leila knows, it's natural to say Hannah doesn't know, and it's weird to say that Leila knows but she wouldn't know if she were in Hannah's situation. I explain that in terms of their having the same evidence, but even if that doesn't work, there's still something to be explained. Footnote 23, where I reject E=K, alludes to a possibility of reformulating the argument in this way. But it'd be really interesting to see a worked-out way of using the stakes-relativity of evidence to explain my data.

    I hope you see and appreciate this comment. I miss when philosophy blogs were a place where philosophy got done instead of a place where we talked about philosophers' awful behavior and got trolled on it. Not that talking about philosophers' awful behavior isn't necessary (and I cringed when I went back to my paper and saw the ratio of women philosophers to philosophers who have been accused of misconduct against women).