A Short Refutation of Strict Normative Evidentialism

Abstract

This paper shows that strict evidentialism about normative reasons for belief is inconsistent with taking truth to be the source of normative reasons for belief. It does so by showing that there are circumstances in which one can know what truth requires one to believe, yet still lack evidence for the contents of that belief. IntroductionStrict normative evidentialism is the view that all normative reasons for belief are evidential reasons.¹ In this paper, I show that the falsity of strict normative evidentialism-just 'evidentialism' from here forward-follows from weak and uncontroversial premises. I argue that none of the premises may be plausibly rejected, and thus that evidentialism is false.² Unstable beliefs with many fixed pointsEvidentialism fails straightforwardly, at least on common views about evidence, for certain classes of truths. Reasons to believe logical truths, mathematical truths, or certain a priori truths are generally non-evidential. However, it is often touted as the right view for 1 empirical truths. For at least one class of empirical truths, however, it can also fail. A thought experiment illustrates the category.Alice is the subject of a psychology experiment. She knows all of its rules and parameters. The experiment goes as follows: Alice is attached to a computer that can read her beliefs accurately and reliably. In front of her is a screen that is connected to the computer.Alice is asked to predict what number will appear on the screen. The rules (of which Alice is fully aware) for generating this number are as follows:The computer will scan for Alice's belief about what number will appear on the screen, and then after a one minute delay, will put up a number on the screen based on Alice's belief about what number will appear on the screen. The computer determines what number will appear on the screen in the following way: Let n be the number that Alice believes will appear on the screen. If n ≥ 0, then the number that will appear on the screen is ½n + 1 . If n < 0, then the number that will appear on the screen is ½n -1. IfAlice changes her belief during the delay, the computer will take ½n + 1 or ½n -1, as appropriate, of the new number and display it instead one minute from the change of belief.³ If Alice suspends judgement or has no view about what number will appear on the screen, has multiple beliefs about what number will appear on the screen, believes that ³ This procedure applies only in the case where Alice has a single belief about what number will appear on the screen, and her belief concerns a specific, single number (i.e. not a disjunctive belief or a belief just that some unidentified number will appear).2 multiple numbers will appear on the screen, has a disjunctive belief about what number will appear on the screen, or has a belief that is a category error (e.g. that the number 'Herbert' will appear on the screen), then the number 16 will appear on the screen.Two final remarks should be made about Alice. She has no particular dispositions concerning what to...

P, but you don’t know that P

“…It's raining over here, but you don't believe that it's raining. 36 35 See Reisner (2015, p. 482), Raleigh (2017, and discussion in Willard-Kyle (2021, p. 765) for defenses of the epistemic impermissibility of believing when one knows one would believe falsely thereby. 36 There is, perhaps, a cheeky reading of this sentence where, by the second conjunct, I indicate that the fact that it is raining over here is a secret and you should act as though you don't know (or as though I haven't told you).…”

On the epistemic rationality and significance of self-fulfilling beliefs

“…First, what's rational to believe is-in a certain sense-2 Raleigh (2015Raleigh ( , 2017, and Joyce (2007) argue that belief-fulfillable propositions are rational to believe, even when they're improbable. Dahlback (forthcoming), Drake (2017), Kopec (2015), Peels (2015), Reisner (2007Reisner ( , 2013Reisner ( , 2015, and Velleman (1989) defend the view that belief-fulfillable propositions are rational to believe in cases where their probabilities are unspecified (rather than low). But the principles they employ imply that BFI propositions are rational to believe.…”

“…Pettigrew (2018), Joyce (2018), andCaie (2013) endorse consequentialist versions of epistemic utility theory that imply that high credences in HCFI propositions are rational to have. Drake (2017), Foley (1991), Reisner (2007Reisner ( , 2013Reisner ( , 2015, Sharadin (2016), andTalbot (2014) all argue that a reason to believe that a proposition would be true if believed, is an epistemic reason to believe the given proposition. This view plausibly implies that if you have a great reason of this kind concerning an improbable proposition, then this proposition is epistemically rational for you to believe.…”

Justifying the principle of indifference

“…Thus we must take E to be the set of those chance functions compatible with evidence to which belief functions can be calibrated. This qualification avoids a worry about Evidentialism of Reisner (2015).…”