Is lucky belief justified?

ABSTRACT

The main lesson from Gettier cases is that while one cannot know a proposition by luck, one can hold a lucky true belief justifiedly. Possibly because the latter is taken for granted, the relationship between epistemic justification and epistemic luck has been less discussed. The paper investigates whether luck can undermine doxastic justification, and if so, how and to what extent. It is argued that, as in the case of knowledge, beliefs can fall short of justification due to luck. Moreover, it is argued that justification-undermining luck is a problem for both internalist and externalist conceptions of justification. Accordingly, it is shown that epistemic luck is a more widespread phenomenon than many in epistemology commonly assume.

KEYWORDS:

• Epistemic justification
• epistemic luck
• internalism
• externalism

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 See also Engel (1992) and Vahid (2001) for discussion of types of k-luck.

2 This methodology can be challenged in a number of ways, such as by doubting the idea that conceivability is a guide to possibility, by criticizing the method of cases in epistemology, or by questioning the reliability of philosophical intuitions in general. Addressing these objections is beyond the scope of this paper. An alternative methodology is to test folk intuitions about widely discussed cases in epistemology and theorize about the results (e.g., Turri, Buckwalter, and Blouw 2015). Of course, this assumes that people are able to reliably individuate intuitions about luck, as opposed to, say, risk, chance, randomness, or similar concepts. Whether people are so reliable is largely an open question.

3 (K3), (K4), and (K5) are different necessary factors for knowing a proposition p. For (K3), what matters is whether the agent adopts a doxastic attitude with content p, i.e., regardless of whether that content is true or false. For (K4), what matters is whether there is a truth-maker for p, i.e., regardless of whether there is, in addition, a doxastic state that has p as content. (K5) is the fact that the agent comes to believe p truly. Preconditions for (K5) are, of course, (K3) and (K4). In other words, (K5) can hold only if (K3) and (K4) hold. And if (K3) and (K4) hold, (K5) also holds. In other words, S comes to believe p truly if and only if S forms the belief that p and there is a truth-maker for p. Given this equivalence, it might be redundant to introduce (K5) as a distinctive necessary factor for possessing propositional knowledge. However, it is useful to do so because it facilitates the identification of a special kind of knowledge-undermining luck, namely veritic luck.

4 Insofar as evidence is typically considered a mental-dependent (internalist) notion, hard-core externalists will reject that posession of evidence is necessary for knowledge. However, since the point here is not to define knowledge but to distinguish interesting varieties of epistemic luck, it is useful to consider factors typically associated with knowledge.

5 See Nozick (1981, 93) for the original case.

6 The case of doxastic luck is somewhat more complicated. We have seen that there are cases of doxastic luck and knowledge, so that doxastic luck is compatible with knowledge. However, there are also cases of doxastic luck and ignorance. Suppose that God hypnotizes Adam and tosses a coin to decide whether to induce in him the belief that p or the belief that not-p, where p is true. As luck would have it, Adam ends up believing p when he wakes up. While this is a clear case of doxastic luck, it is unclear whether Adam has knowledge. This, however, does not necessarily mean that doxastic luck is epistemically dangerous, because the factor that undermines knowledge does not seem to be luck, but the fact that the belief in question is not formed by the exercise of a cognitive faculty. In case (C), on the other hand, the target belief is formed in an epistemically appropriate way, namely by following the available evidence. This is plausibly why it is a case of knowledge.

7 The case is analogous in all relevant aspects to Chisholm’s well-known sheep-in-the-field case (Chisholm 1977, 105).

8 See Baumann (2012) for a dissenting opinion.

9 At this point, theorists of luck whose accounts are inconsistent with folk intuitions can either introduce an error theory to explain why we mistakenly apply or do not apply the concept of luck when we should not, or they can opt for an empirical investigation of folk intuitions and try to derive some general conclusions about how luck actually works to support their theoretical claims—see e.g., Turri, Buckwalter, and Blouw (2015).

10 See Pritchard (2005) for a seminal defense of this view. The modal account of epistemic luck derives from the more general modal account of luck—see Pritchard (2014) for a recent defense—, which holds that an event is lucky just in case it occurs in the actual world, but would not occur in most nearby possible worlds in which the relevant initial conditions for its occurrence are the same as in the actual world.

11 Perhaps more arguments are needed to prove that the modal view is wrong. For some of its predictions are consistent with our intuitions about luck, namely cases of modally fragile lucky events, i.e., actual events that could easily have failed to materialize, such as winning the lottery.

12 For the converse project of characterizing the distinction between internalism and externalism of epistemic justification in terms of varieties of j-luck, see de Grefte (2018).

13 See Coffman (2009) for arguments for the falsity of this thesis, which he calls the ‘luck infection’ thesis.

14 But see Hazlett (2010).

15 The quote is from Madison (2010), who provides an excellent overview of internalist theories of justification.

16 Empirical research on politically motivated reasoning (e.g., Calvillo, Swan, and Rutchick 2020; Gampa et al. 2019; Aspernäs, Erlandsson, and Nilsson 2022) shows the influence of prior political beliefs on syllogistic reasoning, namely that individuals are more willing to accept conclusions that are consistent with their political beliefs than conclusions that are inconsistent (also known as belief bias in psychology). In some of these studies, participants had undergone logical training (Gampa et al. 2019; Aspernäs, Erlandsson, and Nilsson 2022), so they understood how to reason well, and in one (Aspernäs, Erlandsson, and Nilsson 2022) they were asked to try not to think too long before responding. This gives reason to believe that Adam’s case is not unrealistic after all, and that it is not inconceivable that subjects may be prone to reasoning errors in short intervals, while they think they are reasoning correctly in the intervals in which they actually do so, but also in the intervals in which they manifest belief bias, of which they are unaware (because it is a bias)—further research also shows that rapid responding increases belief bias, which gives further reason to think that short-interval cases are not unrealistic; for example, in a study by Evans and Curtis-Holmes (2005) responding rapidly within 10 s, while did not inhibit all analytic processing of syllogisms, increased the amount of belief bias in participants.

17 I use ‘evidence’ and ‘reasons’ interchangeably (as e.g. Turri 2010)

18 Another possible response on behalf of the accessibilist is this: since Adam does not notice the change from good to bad reasoning, and therefore everything seems to be going well from his point of view, the accessibilist might be willing to ascribe justification even in the intervals when Adam is reasoning badly. If so, the kind of luck at stake does not undermine accessibilist justification. However, this move risks running counter the central internalist assumption presented at the beginning of §2.1. In particular, any internalist view should agree with Cohen (ibid.) that ‘[t]here is a fundamental epistemic difference between the beliefs of A [the good reasoner] and the beliefs of B [the bad reasoner] (…) [which] is marked precisely by the concept of justified belief.’ The proposed version of accessibilism would not be able to distinguish between good and bad reasoners in evil demon worlds, i.e., in scenarios where reliability is irrelevant.

19 There are of course other ways to understand the epistemic support relation (see Conee and Feldman 2008 for a useful overview). I use Conee and Feldman’s recent view for illustration.

20 In their 1985 definition, Feldman and Conee include further specifications of what it takes to base a belief on one’s evidence to accommodate ‘the fact that a well-founded attitude need not be based on a person's whole body of evidence’ (Feldman and Conee 1985, 33), but these are not relevant to the case at hand. I omit them for simplicity.

21 In this regard, Conee and Feldman (2008, 13) claim that "[p]roperly inferring a proposition from others that are justified is evidence that the inferred proposition is true".

22 See Goldman (1979) for the original formulation of this view.

23 This is different from the case we saw in §2.1 of the math student who answers correctly one out of 200 in an exam. In the latter case, the student arrives at the correct answer by a reliable and not fleeting process, which explains why her answer is justified.

24 This condition is inspired by Greco's account of ability possession (Greco 2010, 77).