Recalcitrant Credences

The following principle strikes me as plausible:

If S is ideally rational, has the concept of justification, and has credence X that P,

then S believes that she is justified in having credence X that P.

In other words, ideally rational agents believe that all their credences are justified. Any thoughts? I’m sure this has been addressed in the literature, so if you know of any relevant citations, please pass them along!

Advertisements

13 Responses to Recalcitrant Credences

  1. Shen-yi Liao says:

    Does S believe that S is ideally rational? It seems that I could be ideally rational and yet not believe I am ideally rational even if I am, for pragmatic reasons. In which case, I might also hesitate to believe that all my credences are justified. So maybe the key step is to show that an ideally rational agent would (always) believe that she is ideally rational?

  2. Dustin says:

    What would an ideally rational agent “hesitate” to believe that her credences are justified just because she doesn’t believe that she is ideally rational? I don’t believe that I am ideally rational, but I don’t stop believing that my credences are justified on those grounds–or, rather, if I do stop believing that my credences are justified, then I stop *having* those credences. That isn’t to say that I *change* my credences–what to?–but, rather, that my credences become indeterminate.

    That’s the overall thought: an ideally rational agent will abandon any credence that she does not believe is justified.

  3. Dustin says:

    Correction: “WHY would an ideally rational agent…”

  4. Suppose that the above principle is true and that S believes P with credence X. Then, by the principle, S believes that she is justified in having credence X that P. Let P’ be the proposition expressed by “she is justified in having credence X that P.” S believes P’ with some credence; let’s say this credence is X’ (I assume this would be 1). Then, by applying the principle, we see that S believes that she is justified in having credence X’ that P’. Then by repeating this procedure, if the principle is correct, then if S believes anything, then S believes infinitely-many (countably-many) things.

    I don’t know if this is significant, and it doesn’t immediately present a problem in my mind; it is just for consideration. For some reason, this kinda bothers me, but I can’t produce an argument that it’s a problem.

    If there is a problem here, maybe a suitable fix would be to change the principle to the following:

    If S is ideally rational, has the concept of justification, and has credence X that P,
    then S does not believe that she is not justified in having credence X that P.

    This formulation does not entail that S has countable-many beliefs if she believes something; it only requires that S does not believe countably-many things, which doesn’t seem like it would be a problem.

  5. Dustin says:

    I thought about that (weaker) formulation. But I thought I’d give it my all on the stronger one first.

    Anyway, I don’t think that having infinitely many beliefs is a problem for ideally rational agents. They will, after all, believe the infinitely many implications of their beliefs.

  6. Suppose externalism about justification is true. BIVs don’t have justified beliefs.

    We’re weakly a priori entitled to treat perception as reliable, so an ideal rational agent will trust her senses, even if she is a brain in a vat.

    Given this, it seems at least an open possibility that an ideal rational agent would be agnostic as to whether she is a BIV, even while trusting her senses (after all, you’ve gotta trust something!). If so, she might be agnostic as to whether she is justified in holding the credences she does.

    This is far from obviously right, but it’s at least a possible view on which the principle is false.

  7. Following up on Jonathan’s comment.

    Here’s a plausible internalist view according to which the view might be false. Suppose evidentialism is true.

    (E) S is justified in believing P iff S is best supported by S’s evidence.

    Doesn’t it seem that an ideally rational agent (if evidentialism is true) would only believe things that are best supported by S’s evidence. Furthermore, doesn’t it seem that they would refrain from believing what was not supported by S’s evidence.

    If that’s all true, then it seems pretty easy to imagine possible scenarios where an ideally rational agent has the concept of justification, believes some simple perceptual proposition, but fails for form the belief that the perceptual belief is justified – because they lack sufficient evidence concerning the conditions on which justification supervenes (I assume they could still have the concept of justification) and any other evidence that would justify the proposition that the perceptual belief is justified.

    Is the world I described metaphysically possible? It seems so to me. So, if evidentialism is true (and the constraints I placed on being ideally rational are true), then the principle is false.

  8. Dustin says:

    Thanks for the comments, guys!

    Two points:

    First, I’m a bit worried about appealing to epistemic principles that are (at least as) controversial as the one I’m offering. If in fact that principle I’ve put forward conflicts with such an epistemic principle, we could simply take that as reason to reject the epistemic principle in question. Or not. The point is, it would be nice to have something like a clear intuitive counterexample, rather than an argument from some other controversial principle.

    Second, I’m not sure that I exactly follow your arguments.

    Jonathan, could you tell me what P would be in your argument? On the one hand, you say that she “trusts her sense” but on the other you say that she is “agnostic as to whether she is a BIV”. On the straightforward reading of the former, it seems to be that the latter is false.

    Andrew, your argument seems (but I’m not sure) to assume something like the following: if C* is the state of affairs upon which C supervenes, then if an ideally rational agent S does not have sufficient evidence that C* is the case, then S will not believe that C is the case. That principles seems false to me: my young nephew lacks sufficient evidence that there is H2O in his sippy-cup, but he believes that there is water in his sippy-cup. And although he is less than ideally rational, he’s not less than ideally rational for *that* reason!

  9. Hi Dustin,

    The argument doesn’t assume that principle. It only assumes the following:

    (1) Evidentialism
    (2) An ideally rational agent would believe whatever is best supported by his evidence
    (3) It is possible for a perceptual belief to be best supported by one’s evidence without the proposition that the perceptual belief is justified being best supported by the evidence.

    One would only need those three propositions to make the argument go through. They are independently plausible , and none of them assume the principle you suggest.

  10. Dustin says:

    Thanks, Andrew. So your mention of “supervenience” above was beside the point?

    Also, I think that you need to strengthen (2) above to

    (2*) An ideally rational agent would believe whatever is best supported by his evidence *and nothing else*.

    in order to conclude the negation of my principle.

    [Moreover, it seems that once you’ve got (2) or (2*), (1) has become idle. Is this correct?]

    Finally, it seems to me that (as all good deductive arguments are) the argument from (2*) and (3) to the negation of my principle is pretty plainly question begging. You say that these principles are “independently plausible” but I contend that so is my principle. Not surprisingly, I actually think that my principle is *more* independently plausible than yours! But of course that won’t convince you or anyone else. Like I said before, it would be nice to have a clear intuitive counterexample–that is, a case in which an agent violates my principle and isn’t thereby irrational (or rather, wouldn’t be if he had unlimited computational powers, the time to apply them, etc and was still in violation of my principle).

  11. I agree that (1) is idle…

    Originally I was thinking
    (1) Evidentialism
    (2) An ideally rationally agent will believe all and only those things that satisfy the conditions for epistemic justification (which I think meets your strengthening requirement)
    (3) It’s possible to have evidence support P without supporting the proposition that P is justified.

    Regarding the Question Begging Charge
    (1), (2), and (3) can be used to construct a valid argument against your principle – So it’s question begging in the way that all valid arguments are question begging.

    It’s not plainly question begging in the two senses that seem illegitimate. None of the premises explicitly assumes the denial of your principle, nor does the reasonableness of accepting any of the premises antecedent to considering the argument as a whole depend on denying your principle. When people talk about question begging they usually have one of these two things in mind. It’s not clear that this does either or those.

    Regarding the Plausibility Comparisons
    This is just clarificatory. Which of (1), (2), and (3) do you think is antecedently less plausible than your principle? Is it one of them in particular or all three?

  12. Dustin says:

    “it’s question begging in the way that all valid arguments are question begging.”

    Hey, that sounds familiar!

    “It’s not plainly question begging in the two senses that seem illegitimate.”

    Here’s an argument against ~(P&Q) that also passes your requirements.

    1. P
    2. Q
    3. Therefore, P&Q.

    “Which of (1), (2), and (3) do you think is antecedently less plausible than your principle?”

    Well, definitely (2), since it seems to assume justification internalism (ideal rationality being an internal constraint), which, although plausible, is not as plausible as my principle.

    But perhaps (3) as well, since you need strengthen it to:

    (3*) It is possible *for an ideally rational agent* to have evidence in support of P but lack evidence in support of the claim that believing that P is justified.

    I don’t really have much of a (direct) intuition about (3*).

    Then again, (1) also seems fairly implausible to me as well, but that shouldn’t be a surprise. Many (most?) people do, after all, reject evidentialism.

    So I guess I’ll say “all three”.

  13. (yeah, I was borrowing that phrase from the literature discussing the problems with rejecting an argument with the stated reason being that the argument is question begging)

    Again, on the question begging – conjunction introduction certainly seems like a case where the conclusion is very explicitly assumed in the premises of the argument. So we only need to tweak the the requirements and conjunction introduction would fall under them.

    So – it’s still unclear to me how the argument is plainly question begging And it seems to me that conjunction introduction would be an instance where the conclusion is very explicitly contained in the premises.

    I’ll have to think more about the independent plausibility of your principle against (1), (2), and (3). (I should probably sit back and just think about your principle for a bit…)

%d bloggers like this: