Determinación de la relación de los factores predisponentes

I have argued that the collective doxastic ‘should’ is the ‘should’ occurring in epistemic advice.14 This is philosophically noteworthy because it fills a gap in the literature concerning the truth-conditions of epistemic-advice sentences. Furthermore, understanding what it means to givecorrect epistemic advice also helps us to better understand the practice of giving epistemic advice, which has so far been overlooked in the literature.

Apart from these rather big-picture considerations, I think thatCollectivism

also has a more concrete philosophical cash value. It can solve a puzzle that John Turri desribes in his paper ‘A Puzzle about Withholding’, surrounding the case

WM mentioned above. The puzzle arises from three assumptions, which he takes to be common in epistemology:15

• Triad: For every proposition P that a subject S considers, S can adopt three and only three doxastic attitudes towards P: believe it, disbelieve it, or withhold.16

• Optimism: Given any propositionP and any body of evidenceE, E will justify at least one doxastic attitude towards P.

• Evidential Propriety17: The epistemic propriety of a subject’s doxastic attitude is entirely a function of the subject’s evidence.

These three assumptions come into conflict inWM:

Withholding mathematicians (WM). One hundred eminent math- ematicians discuss, outside of your earshot, whether a certain axiom

14_{A longer, self-standing version of the present section has been published under the title}

‘Resolving Turri’s Puzzle about Withholding’ indialectica (vol. 70, issue 2).

15_{In the following, I will refer to the set of these views as ‘the three assumptions’.} 16

One might object to this assumption that it ignores credences, i.e., more fine-grained doxastic attitudes. If one thinks that not every credence is an instance of one of the coarse-grained doxastic attitudes, belief, disbelief or withholding, one will denyTriad and claim that besides the three coarse-grained doxastic attitudes, a subject could also adopt some credence to a proposition. However, this objection does not undermine the following discussion. Turri’s puzzle arises in a similar way if one reads the three assumptions only as claims about coarse-grained doxastic attitudes, and not as claims about all doxastic attitudes in general.

17

Turri (2012) actually calls this assumptionEvidentialism. I have chosen another term to avoid confusion with the different positionEvidentialism, which I introduced in section 3.2.

entails a certain claim. Let’s call the proposition that the axiom entails the claim ‘M’. After their discussion, they all tell you that you shouldn’t withhold on M. You have no other evidence regardingM. Turri (2012: 359) states that the following two claims seem true:

(9) You should not withhold on M inWM.

(10) You should neither believe nor disbelieve M inWM.

Turri’s argument for (9) is concise: “It stands to reason [...] that if all the Mathematicians say that withholding is not the thing to do, then you shouldn’t withhold” (ibid.). His argument for (10) rests on Evidential Propriety. Since your evidence neither supports believing M nor disbelievingM, you should neither believe nor disbelieveM.

Turri argues thatWM creates a puzzle about withholding since it apparently shows the possibility of an epistemic impasse. An epistemic impasse is a situation in which one is not permitted to adopt any doxastic attitude to a proposition

P. According to (9), you should not withhold and according to (10), you should neither believe nor disbelieve. From (9) and (10), Turri concludes that

(11) You are neither permitted to believe or disbelieveM nor permit- ted to withhold on M inWM.18

Since according to Triad, you can only adopt belief, disbelief or withholding towardsM, you are thus not permitted to adopt any doxastic attitude toM. Thus,

WMis an epistemic impasse. This is incompatible withOptimism. According to

Optimism, given the propositionM and your evidence in WM, there is at least one doxastic attitude you are justified to adopt toM.

Turri goes through several possibilities of resolving this puzzle by either giving upTriad,Evidential Propriety or the intuitive verdict that (9) is true.19 He rejects all of these options and concludes: “None of the proposed solutions to the puzzle seems fully satisfactory. I, for one, am left puzzled” (ibid.).

I won’t discuss Turri’s arguments against the possible resolutions he considers. My alternative way of resolving the puzzle should be clear: I claim that (9)

18_{(11) follows from (9) and (10) if one assumes, as Turri obviously does, that if one should}

not doA, one is not permitted to doA. I will challenge this assumption further below.

is true because the ‘should’ in it is collective. When the mathematicians are giving epistemic advice, they talk about what you should believe in light of the collective evidence. I refer to the proposition expressed by (9) when uttered by the mathematicians as (9C):

(9C) In light of the collective evidence, you should not withhold on

M.

The ‘should’ in (10), on the other hand, is subjective. Since (9C) does not entail

that you are not permitted not to withhold, (9) and (10) together do not entail (11).

The reason for this is as follows. That in light of your evidence you should not do Aentails that you are not permitted to do A.20 However, that you should not do A in light of a body of evidence that contains evidence that you don’t possess does not entail that you are not permitted to do A, as I will argue for shortly. That is, the subjective ‘should’ is tied to permission, but the collective ‘should’ isn’t. In other words, (10) entails (11) in combination with (9S), but not (9C):

(9S) In light of your evidence, you should not withhold onM.

(9S) is false. As Turri remarks himself, your evidence neither supports believing

M nor¬M. Consequently, you should withhold in light of it. You might think that (9S) is actually true since after the mathematicians uttered (9), you gathered

new evidence supporting that you should not withhold: the experts told you not to withhold. What speaks against this line of reasoning is that on the natural picture I am assuming throughout this thesis, which doxastic attitude one should have to a propositionP in light of a body of evidenceE depends on the evidential support relation between E and P. That the mathematicians uttered (9) is part of your evidence, but this additional evidence neither supports nor speaks against

M, as can be seen by the fact that we are at a loss to say whether it supports M 20_{This needs qualification. As explained in chapter 2, ‘should’ is a weak deontic necessity}

modals in comparison to the strong deontic modal ‘must’. It is agreed that<Smust doA>(in its subjective sense) entails<S is obliged to doA>and therefore, since obligation is the dual of permission, that<S must not doA>entails<S is not permitted to doA>. However, ‘should’ is thought to be weaker and not to entail obligation, from which follows that<S should not do

A>might not entail<S is not permitted to doA>. (Harman 1993 is one among many to have made this point.) I ignore this subtlety here. If ‘should’ is not the dual of ‘is permitted’, Turri could simply restate the problem in terms of ‘must’.

or speaks against it. Therefore, even once you have gathered the evidence that the mathematicians uttered (9), you should withhold onM in light of your evidence.

Now, since (9) inWM is only true in the sense of (9C), but not (9S),WM

does not constitute an epistemic impasse because (9C) and (10) do not entail

(11). To see that such an entailment relation between the collective ‘should’ and permission does not hold, imagine that you simply formed the belief thatM (or

¬M; which is irrelevant). It seems fairly clear to me that you are not permitted to do this. You have no evidence supportingM, so you’re arbitrarily choosing a belief on whetherM holds. But arbitrarily choosing beliefs is hardly epistemically permissible. If it were, it would be OK for you to simply form the belief that the numbers of stars in the universe is even or that the fair coin you are about to flip will land heads up. This is so, even if someone else happens to have excellent evidence that the numbers of stars is in fact even or that the coin will land heads up. Thus, whether you are permitted or not to adopt a doxastic attitude seems to be merely a function of your own evidence, but not of that of others. We have accordingly no reason to assume that the fact that you should not, in the collective sense, withhold onM entails that you are not permitted to withhold on

M.

A worry one might have about my solution is whether (9) can really be called advice and whether it is hence plausible to claim that the ‘should’ in it must therefore be collective. As pointed out in section 5.2, it is widely assumed that advice requires that advisors intend to give advisees a pro tanto reason to do whatever they advise them to do. If the mathematicians utter (9), this doesn’t seem to give you any kind of reason not to withhold on M. It’s not like the utterance of (9) gives you a pro tanto reason not to withhold, which is then overridden by another reason. Since it doesn’t change your evidence with respect toM at all, it is hardly a reason for you to stop withholding onM at all.

These considerations do not only make it doubtful whether (9) is a form of epistemic advice, but also raise the question of why the mathematicians utter it in the first place. If they don’t want to advise you, what else are they trying to achieve? Turri doesn’t tell us more about the case than what is said inWM, so it is underspecified with respect to the advisors’ intentions. I see two possible reasons for the mathematicians to utter (8) when it carries its collective meaning (9C). First, as I explained in section 5.3, the mathematicians can use (9) to simply

imply that they have evidence that determines whetherM is true. Where making this implication is their sole intention, they do not try to give you a pro tanto reason not to withhold on M. Under this specification of WM, (9) would not be epistemic advice, but nonetheless contain the collective doxastic ‘should’.

Second, things in WM could also be as follows. Imagine that what the mathematicians are, and only are, discussing outside your earshot are the merits of a recently suggested proof for M and that, so far,M has neither been proved nor disproved. They can therefore only come to two possible conclusions after discussing this proof: that you should (collectively) believeM or that you should keep on withholding on M. They come to conclude that the proof works and that you should believeM. The crucial point is that they reasonably, but mistakenly, believe that you know that they have been discussing this proof. Furthermore, they reasonably, but mistakenly, believe that you have been told by a rival, evil mathematician that they cannot be trusted and that whatever they say, you should maintain withholding on M. In such a situation, the (good) mathematicians can reasonably intend to give you a pro tanto reason not to withhold onM by uttering (9). They assume that on hearing (9) you get a reason to believe that the collective evidence supports M because they think that you know that the only options are that the evidence supports believing or withholding onM and that you therefore will infer from (9) that they believe that the collective evidence supportsM. They don’t think that you have an all-things-considered reason not to withhold on M

after hearing (9), as they believe that you also have evidence from their rival, evil mathematician that supports withholding on M and weighs against the reason they have provided. To conclude, under this second specification of WM, the good mathematicians can reasonably intend to give you a pro tanto reason not to withhold on M by uttering (9), and (9) will therefore be reasonable epistemic advice under these circumstances.21

It is important to notice that my view is compatible withEvidential Propriety.

21_{You might wonder why I have added an extra complication to this specification by stipulating}

that you have been misinformed by a rival, evil mathematician. The reason is that otherwise the good mathematicians are giving you an all-things-considered, not just a pro tanto, reason not to withhold by uttering (9). An opponent of my solution to Turri’s puzzle could therefore claim that the (9) in ‘should’ is in fact subjective, and not collective. Don’t get me wrong: I do believe that in such a case the ‘should’ is also collective, since (9) is still advice, but I wanted to show that there is a specification of WMwhere it is uncontroversial that (9) is both epistemic advice and contains a collective ‘should’.

While I claim that which doxastic attitude a subject should adopt, in the collective sense, is partly a function of evidence that is not the subject’s, I do not deny that the epistemic propriety of a subject’s doxastic attitude is entirely a function of the subject’s evidence. I assume that by ‘epistemic propriety’ Turri means something along the lines of ‘justification’; and whether you are justified in adopting a doxastic attitude might very well be the same issue as whether you should adopt itin light of your evidence. However, whether you should adopt it in thecollective

sense is another issue and neither entails nor is entailed by whether you are justified to adopt it. Since my view is furthermore compatible withTriad and

Optimism, it thus resolves the puzzle while accepting both (9) and the three assumptions.

There are two more reasons why it matters that justification and epistemic propriety come apart from what one should believe in light of the collective evidence. First, it allows me to acknowledge the main point Comesa˜na (2013) makes in his solution to Turri’s puzzle, while maintaining that my solution is superior to his. Comesa˜na argues that (9) is false since you are not justified in believing or disbelievingM. He concludes that you are and you are only justified to withhold onM.

I agree with this last point, but it does not follow that (9) is false, at least if it means (9C). It is tempting to derive from the fact that you are not justified

to believe or disbelieveM that you should not believe or disbelieve M. This is indeed a plausible principle if ‘should’ is read subjectively, and we can correctly derive from it that (9S) is false. However, this principle does not apply where

‘should’ is collective. Thus, Comesa˜na has only shown that (9S), but not that

(9C) is false.

The important difference between Comesa˜na’s view and mine is that Comesa˜na claims that (9) is wrong. He does so because he only considers a subjective reading of ‘should’, as his focus on whether you are justified to not withhold on M in

WM indicates. The advantage of my view over Comesa˜na’s is that I can explain why we intuitively judge (9) to be true: we correctly read it as expressing (9C)

when the mathematicians utter it. On behalf of Comesa˜na, one could suggest the following alternative, error-theoretic explanation of our judgement that (9) is true: we read it as expressing (9S) and mistakenly judge it to be true since we wrongly

relative to our evidence. My solution has the advantage that it does not have to posit that we’re making such a relatively simple mistake and is thus more in line with the assumption that speakers are competent. If this advantage came at the cost of having to make implausible claims about the meaning of (9), the error theory might be preferable. However, my arguments in support of Collectivism

have given us good reason to assume that the ‘should’ in (9) is collective.

Second, the distinction between epistemic propriety and the collective ‘should’ also prevents my solution from being subject to the following objection. One might wonder whether the evidence-sensitivity of ‘should’ does not also imply that ‘is permitted’ is evidence-sensitive. In this case, ‘is permitted’ can carry a collective sense, too. This would allow us to infer (11) from (9C) and (10) if (11)

meant:

(11*) In light of the collective evidence, you are not permitted to with- hold on M, and in light of your evidence, you are not permitted to believe or disbelieve M.

Personally, I have difficulties hearing ‘is permitted’ as being relativizable to collective evidence. As suggested above, if it were, someone who knows that a coin about to be flipped will land heads up could correctly say about a person who does not know this:

(12) She is permitted to believe that the coin will land heads up.

(12) sounds off to me. In any case, even if (11) could mean (11*), this wouldn’t make WM an epistemic impasse. Remember that what is problematic about an epistemic impasse is that it is a situation where there is no doxastic attitude to a proposition P that would be epistemically proper for the subject. However, the truth of (11*) does not entail that there is no epistemically proper doxastic attitude for you inWM. Whether one is justified in adopting a doxastic attitude might be identical with whether one should, in the subjective sense, adopt it, but is distinct from whether one should, in the collective sense, adopt it. Analogously, the fact that in light of some collective evidence, which contains evidence you do not possess, you are not permitted to withhold on M does not mean that it is not epistemically proper or justifiable for you to withhold on M.

5.9

Conclusion

I have argued forCollectivism, the view that epistemic advice of the form ‘You should adopt doxastic attitude D to proposition P’ is true iff D to P reflects how much the combined evidence of the advisor and the advisee supports P. This reveals a further instance of the information-sensitivity of the doxastic ‘should’: its collective sense. Since no one has so far put forward an account of the truth-conditions of epistemic-advice sentences,Collectivism fills a gap in the literature. It also helps us to better understand the so far under-researched practice of epistemic advice. On top of this, it can solve Turri’s (2012) puzzle