Portal de Periódicos da CAPES

Liberalism and Conservatism in the Epistemology of Perceptual Belief

2009; Routledge; Volume: 88; Issue: 4 Linguagem: Inglês

10.1080/00048400903225734

1471-6828

Ram Neta,

Philosophical Ethics and Theory

Abstract Liberals claim that some perceptual experiences give us immediate justification for certain perceptual beliefs. Conservatives claim that the justification that perceptual experiences give us for those perceptual beliefs is mediated by our background beliefs. In his recent paper 'Basic Justification and the Moorean Response to the Skeptic', Nico Silins successfully argues for a non-Moorean version of Liberalism. But Silins's defence of non-Moorean Liberalism leaves us with a puzzle: why is it that a necessary condition for our perceptual experiences to justify us in holding certain perceptual beliefs is that we have some independent justification for disbelieving various sceptical hypotheses? I argue that the best answer to this question involves commitment to Crispin Wright's version of Conservatism. In short, Wright's Conservatism is consistent with Silins's Liberalism, and the latter helps to give us grounds for accepting the former. Notes 1See Silins [2007: 110–2]. Silins's use of the terms 'Liberal' and 'Conservative' is intended to be at least somewhat faithful to the coinage of Pryor [2004: 354], though, as I note below, there are differences in how these authors use these terms. 2The interpretation of Pryor that I've offered in this paragraph is at odds with parts of the passage that I've just quoted from Pryor's article. In the first quoted paragraph of this passage, Pryor distinguishes conservative treatments of a particular hypothesis from liberal ones by appeal to what they say about whether 'you need some justification to believe H in order to have a given kind of prima facie justification to believe P'. In the second paragraph, he draws the distinction by appeal to what they say about whether 'having prima facie justification to believe P requires you to have antecedent justification to believe H'. Now, these two characterizations are distinct, and can conflict with each other in some logically possible circumstances. Let J be your justification for believing that P, and suppose that a necessary condition of J's being your justification for believing that P is that you have some justification for believing that H. But suppose also that, if you were to lack J, it's possible for you to have some other justification L for believing that P, and that it is not a necessary condition of L's being your justification for believing that P that you have some justification for believing that H. If these suppositions are all true of a particular case, then a theory that treats H Conservatively (on Pryor's first characterization) would be true, but a theory that treats H Conservatively (on Pryor's second characterization) would be false: it's necessary for your having a particular kind of justification (namely, J) for believing that P, that you have some justification for believing H; but it's not a necessary condition of your having any justification at all for believing P that you have some justification for believing H. So the passage quoted from Pryor does not settle how we should understand the distinction between a theory that treats H liberally and one that treats H conservatively. But the rest of Pryor's article gives us very good reason for thinking that Pryor intends to distinguish conservative from liberal treatments of a particular hypothesis in the first of the two ways distinguished above. 3See, for instance, Schiffer [2005: 283–5] on 'mixed' justifications. On the traditional definitions of 'a priori' and 'a posteriori', the former is defined negatively, as a justification that does not contain any sensory or perceptual component, and so Schiffer's 'mixed' justifications will all count as a posteriori on the traditional definitions. 4Could we say that the proof gives us a priori justification for believing the set-theoretic conclusion so long as we lack justification for believing that we did make a mistake (even if we don't have any justification for believing that we did not make a mistake)? Notice that, if we adopt this view, then, so long as we lack justification for believing that we did make a mistake, doing the proof will give us justification both for believing its set-theoretic conclusion, and also for believing that we did that very proof. If justification is closed under conjunction introduction (or at least if it is so closed in this particular instance, even if not in general), then we will have justification for believing that we did not make a mistake in doing the proof. For more on the distinction between what gives us justification (or, as he would call it, 'warrant') for believing a proposition, and the justifiers that enable it to give us justification (or warrant) for believing that proposition, see Burge [1993: 458–63]. It is important for Burge to distinguish between the a priori warrants that testimony can sometimes (on his view) transmit and the a posteriori warrants that must be in place for testimony to transmit these a priori warrants. 5It's important to notice that JB-Closure does not imply that justification boosting is closed under known entailment: something could make you more justified in believing that p and yet not make you more justified in believing that q, even if you know that (p only if q). The kind of justification involved in JB-Closure is not something that comes in degrees. It is rather a form of entitlement. To be justified in believing something, in the sense at issue in JB-Closure, is simply to be entitled to believe it. And entitlement does not come in degrees. (Notice, though, that entitlement is not at all the same as blamelessness: it's possible to blamelessly do or believe something that you're not entitled to do or believe). 6The labels 'Moorean' and 'Rationalist' strike me as very unfortunate, since Moore himself was not a 'Moorean' in the present sense, and it's possible for a radical empiricist to be a consistent 'Rationalist' in the present sense. But I simply follow Silins in using the labels 'Moorean' and 'Rationalist' to denote the positions just stated. 7One of the other two reasons has to do with the alleged plausibility of the Liberal thesis itself, and the other has to do with the putative fact that there can be no foundational justification unless Liberalism is true. Pryor [2000: 536] appeals solely to the first of these two reasons in defending Liberalism. For reasons that I offer elsewhere [Neta 2004], I believe that this defence of Liberalism has no merit. 9Someone might protest that it is not rational to assign non-zero credence to the BIV hypothesis. But this protest strikes me as obviously wrong. Even if we accept a McDowellian or Williamsonian conception of evidence, according to which some of our evidence is strictly inconsistent with the BIV hypothesis, we still should not accept the claim that one should be willing to bet one's life against a penny on the truth of many of the propositions in one's own evidence set, nor should we accept the claim that one should be willing to bet one's life against a penny on the falsehood of the BIV hypothesis. Since the probabilities in the proof must be understood as rational degrees of credence (in order to justify the transition from step 9 to step 10), these considerations are relevant to determining Prob (BIV). Someone might also protest that, given the infinite variety in our possible experience, it is rationally required to assign a prior probability of 0 to each possible experience. It seems to me that the most plausible response to this protest is to say that the experiences upon which we conditionalize are individuated crudely, so that their variety, for the purposes of assigning prior probabilities, is not infinite. 8Silins's argument actually goes as follows: (1′) If e justifies me in believing that h, then e does not lower my rational degree of confidence in h. (2′) The hypothesis (henceforth, BIV) that I am a brain in a vat being electrochemically stimulated to have this particular visual experience (henceforth, e) entails that e is occurring. (3) Prob (e/h) = Prob (e&h)/Prob (h) (definition of conditional probability) (4) If e is entailed by h, then Prob (e&h) = Prob (h) (follows from axioms of probability) (5) If e is entailed by h, then Prob (e/h) = 1 (from 3, 4) (6) Prob (BIV/e) = [(Prob (BIV))/Prob (e)] x Prob (e/BIV) (Bayes's Theorem) (7) Prob (BIV/e) = [(Prob (BIV))/Prob (e)] (from 2′, 5, 6) (8) Prob (e) > 0, Prob (BIV) > 0 (9) Prob (BIV/e) > Prob (BIV) (from 7, 8) (10) e raises my rational degree of confidence in BIV (from 9) (11) If e raises my rational degree of confidence in BIV, then it lowers my rational degree of confidence in –BIV (requirement of probabilistic coherence) (12) e lowers my rational degree of confidence in –BIV (from 10, 11) (13) e does not justify me in believing –BIV (from 1′, 12) This argument, as rendered here, is defective, because premise (1′) is false. Consider Alice, who takes a pill X to cure her headache. She knows that the pill has a 90% chance of curing her headache within 30 minutes, so she is rational to have degree of confidence 0.9 in the proposition that her headache will be cured in 30 minutes. But immediately after taking X, Alice decides that, rather than suffering the typical side effects of X, she'll take another pill Y that is certain to cancel all the effects of X but has an 85% chance of curing her headache within 30 minutes. Now, Alice is rational to have degree of confidence 0.85 in the proposition that her headache will be cured in 30 minutes. So her taking pill Y lowers her rational degree of confidence that her headache will be cured within 30 minutes, but nonetheless it also justifies Alice in believing that her headache will be cured within 30 minutes. Here, the same bit of evidence both justifies Alice's belief and simultaneously lowers her rational degree of confidence in the proposition believed. Premise (1) is therefore false. See Achinstein [2001: 69–94] for further discussion of such examples. This problem with Silins's argument above is, however, easily fixed. The counterexample that we just gave to (1′) involves a case in which what justifies one in believing a proposition also defeats a previous justification that one had for believing the same proposition. But this distinguishes this counterexample from the case in which the Moorean is interested: the Moorean is interested in a case of normal visual experience, and its justificatory relation to the negation of various sceptical hypotheses. And it cannot plausibly be claimed that normal visual experiences defeat justifications that we already had for believing the negation of various sceptical hypotheses. So this problem with the argument above can be easily fixed by rewriting premises (1′) and (2′), and then giving the argument in the text. 10Thanks to David Christensen, Emily Given, Michael Huemer, Matt Kotzen, Farid Masour, Jim Pryor, Nico Silins, Crispin Wright, and two anonymous referees for the Australasian Journal of Philosophy for helpful questions and comments.