SCIENCE@home

Updating for externalists

J. Gallow

Published 2019 in Noûs

ABSTRACT

The externalist says that your evidence could fail to tell you what evidence you do or not do have. In that case, it could be rational for you to be uncertain about what your evidence is. This is a kind of uncertainty which orthodox Bayesian epistemology has difficulty modeling. For, if externalism is correct, then the orthodox Bayesian learning norms of conditionalization and reflection are inconsistent with each other. I recommend that an externalist Bayesian reject conditionalization. In its stead, I provide a new theory of rational learning for the externalist. I defend this theory by arguing that its advice will be followed by anyone whose learning dispositions maximize expected accuracy. I then explore some of this theory’s consequences for the rationality of epistemic akrasia, peer disagreement, undercutting defeat, and uncertain evidence. O rthodox Bayesian epistemology is designed to model rational uncertainty. 1 When you lack relevant evidence, its norms permit uncertainty about various and sundry matters: the weather, the victor, the price of tea in China. But there is a certain kind of uncertainty which orthodox Bayesianism has more difficulty modeling: uncertainty about what evidence you do or do not possess. Influential arguments in support of orthodox Bayesianism take for granted that your evidence will always tell you what your total evidence is, so that there may be no uncertainty about what your evidence says. Let’s call this thesis ‘internalism’, Final draft; forthcoming in Noûs. † For helpful conversations and feedback on this material, I am indebted to Sara Aronowitz, Adam Bjorndahl, Catrin Campbell-Moore, Nilanjan Das, Kevin Dorst, Daniel Drucker, Julien Dutant, Adam Elga, Jeremy Goodman, Harvey Lederman, Stephen Mackereth, Alexander Meehan, Jim Joyce, Jim Pryor, Daniel Rothschild, Teddy Seidenfeld, Julia Staffel, Robert Steel, Pablo Zendejas Medina, Snow Zhang, and two anonymous reviewers. Thanks also to audiences at Princeton University, University College London, the Updating and Experience conference at Ruhr University, Bochum, the Formal Epistemology Seminar at Carnegie Mellon University, and the 2019 Pitt-CMU Graduate conference. 1 As I’ll understand the position, orthodox Bayesianism is committed to at least the following theses: probabilism, which says that rational credences are (at least finitely additive) probabilities; conditionalization, which says that you should be disposed to learn from your evidence by conditioning on it (see §2), and van Fraassen’s principle of reflection, which says that your current credences should equal your expectation of your future credences (see §2.1). For each of these assumptions, there are epistemologists justly called ‘Bayesian’ who deny it; but these assumptions compose a familiar, ‘off-the-shelf ’ Bayesian theory of rationality. updating for externalists 2 of 30 and let’s call its negation ‘externalism’.2 The externalist thinks that your evidence could fail to tell you what evidence you have or don’t have, in which case, it could be rational for you to be uncertain about what your evidence is. Both internalism and externalism have able defenders. Myself, I’m undecided. I can feel the force of arguments on both sides. So I won’t be defending either position here. Instead, I will be asking: what becomes of orthodox Bayesianism if externalism is correct? And on this question, I am decided. The externalist Bayesian should reject the orthodox learning norm (or updating rule) of conditionalization. This is a lesson I’ve learned from Salow (2018), who teaches that, if an externalist Bayesian follows conditionalization, then they will be capable of engaging in acts of deliberate self-delusion—repeatedly ‘learning’ from experience in such a way as to raise their rational credence in some proposition as high as they like, even when the proposition is false. It is difficult to see this as rational inquiry. So I recommend that the externalist reject conditionalization. In its stead, I will advance a new theory of rational learning for the externalist. I’ll defend this theory by arguing that its advice will be followed by anyone whose learning dispositions maximize expected accuracy. And I’ll show that those who follow this theory’s advice will be incapable of engaging in deliberate self-delusion. Assuming evidentialism—that is, assuming that the rationality of your doxastic states is determined by the evidence you possess—externalism entails that your evidence could fail to tell you whether your doxastic states are rational or not. For this reason, externalism has played a starring role in recent debates about the rationality of epistemic akrasia. Some externalists have held that your evidence could make it likely both that it will rain and that your evidence doesn’t make it likely that it will rain. In that case, they have proposed that it is rational for you to be epistemically akratic, believing both that it will rain and that it’s irrational to believe that it will rain. Relatedly, some externalists have held that the disagreement of an epistemic peer—who has all the evidence that you do, and is equally good at evaluating it as you are—may give you reason to think that your belief in rain was irrationally formed; nevertheless, this need not give you any reason to revise your views about the weather.3 The externalist theory of learning I’ll develop here yields a distinctive form of externalism, according to which certain kinds of epistemic akrasia are always irra2 Internalism is presupposed by the Lewis-Teller diachronic Dutch book argument for conditionalization (for more, see Gallow 2017), as well as the more recent accuracy (or ‘epistemic utility’) arguments for conditionalization from Greaves & Wallace and Briggs & Pettigrew (forthcoming) (for more, see Schoenfield 2017a). For justifications of conditionalization which do not presuppose internalism, see (for instance) van Fraassen (1989, ch. 13), Lange (1999), Leitgeb & Pettigrew (2010b), Titelbaum (2013, ch. 7), Gallow (2019), and Zendejas Medina (ms). 3 For more, see Elga (2013), Weatherson (ms, 2013), Horowitz (2014), Greco (2014), Lasonen-Aarnio (2014, 2015, forthcoming), and §3. §1. internalism and externalism 3 of 30 tional; and, if the disagreement of an epistemic peer gives you reason to think that your beliefs were irrationally formed, then it can be rational for you to ‘conciliate’ with that peer after learning of the disagreement. It additionally gives guidance in cases of undercutting defeat, and learning experiences in which certainty in no proposition has been rationalized. The latter cases are usually treated with Jeffrey conditionalization (see Jeffrey, 1965). My past self treated the former with a norm I called holistic conditionalization (Gallow, 2014). I will show that, in the paradigm cases, the theory defended here agrees with both of these learning norms. 1 Internalism and Externalism In general, to be an internalist is to think that some condition must always lie within your epistemic reach. Given some condition c , an internalist says: if you satisfy c , then you must have access to the fact that you satisfy c . An externalist, in contrast, says that you may satisfy c without having access to the fact that you satisfy c . Different conditions and different kinds of access yield different forms of internalism and externalism. For instance: let the condition be being in pain, and say that you have access to a fact when you know it. We then get the internalist thesis that, if you are in pain, you must know that you are in pain, and the corresponding externalist thesis that you may be in pain without knowing that you are in pain. To get the form of internalism that I’ll be interested in, let the condition be possessing the total evidence e , and say that you have access to a fact when it is part of your evidence. Then, the internalist says: whenever e is your total evidence, your evidence must say that e is your total evidence. The externalist: on the contrary, sometimes e can be your total evidence without your evidence telling you that e is your total evidence.4 Internalism If e is your total evidence, then your evidence must tell you that e is your total evidence. ƒ (Te → ETe ) Externalism You may have the total evidence e without your evidence telling you that e is your total evidence.

PUBLICATION RECORD

CITATION MAP

EXTRACTION MAP

CLAIMS

  • No claims are published for this paper.

CONCEPTS

  • No concepts are published for this paper.

REFERENCES

Showing 1-65 of 65 references · Page 1 of 1

CITED BY

Showing 1-21 of 21 citing papers · Page 1 of 1