What Should We Believe?

1 What Should We Believe? Thomas Kelly, University of Notre Dame James Pryor, Princeton University Blackwell Publishers Consider the following question: What should I believe? This question is a normative question. It is, of course, notoriously obscure exactly what one is claiming when one claims that a given question is a normative question. (Indeed, this issue is among those which we will explore.) But intuitively, the question What should I believe? differs from purely descriptive questions such as What do I believe? or What will I believe? in a way in which it resembles other paradigmatic normative questions such as What should I do? It is this question which serves as both the starting point and guiding focus of our book, an exploration of selected issues in contemporary epistemology. Our choice of guiding question affects the content of the book in various ways. For example, unlike many epistemology books, we do not discuss the project of analyzing the concept of knowledge, in the sense of attempting to provide necessary and sufficient conditions for the correctness of its application. Indeed, issues concerning the concept of knowledge are in general less central to the book than issues which bear more directly on

2 questions about what one is justified in believing. 1 In keeping with this emphasis on the distinctively normative aspects of epistemology, we will devote considerable attention to issues that are sometimes discussed under the rubric the ethics of belief questions about the nature of epistemic norms, what reason we have to conform to such norms, the extent to which what we believe is under our voluntary control, and so on. In addition to this strong emphasis on the concept of justification and concern with the normative aspects of epistemology, various other topics will receive more extended treatment than is usual. These include: the role of probability in epistemology, questions about rationality over time, the concept of coherence, relevant alternatives theories, and contextualist approaches to skepticism. (A more detailed description of the book s projected contents is given below.) We will also devote more attention than is usual to certain key areas of intersection between epistemology and neighboring fields, e.g., ethics and the philosophy of mind. In treating these and other topics, our primary concern is not to present or defend our own views but rather to enable the reader to better understand the contemporary debate. However, our approach does not consist of providing a simple explication of competing views about a given issue. Rather, throughout the book, we will offer proposals about what we believe is ultimately driving the debate. Thus, our aspiration is not to provide a mere survey of a given debate but rather to identify and to begin to explore the underlying issues that we believe lie at its core, issues that need to be brought more fully 1 Of course, the choice of?what should I believe? as guiding question (as opposed to the question?what is knowledge? or something similar) leaves open the possibility that the concept of knowledge will figure prominently in the answer to that question. Thus, one might hold (e.g.) that one should believe all and only those propositions that one is in a

3 into the center of the contemporary discussion. This general approach influences not only the content but also the structure of the book. This is seen, for example, in our treatment of skepticism. While it is typical in epistemology books to begin with a discussion of skepticism, our own discussion of the topic occurs in the last two chapters of the book. We believe that this inversion of the usual ordering is useful inasmuch as at the end of the book the reader will be better equipped to appreciate important subtleties in both the skeptic s argument and in the most interesting responses to that argument. In particular, by the end of the book the reader will already be familiar with the notions of epistemic priority and defeasibility, notions which we believe are absolutely crucial to fully understanding the dialectic between the skeptic and his opponent. (Moreover, we also believe that an early focus on skepticism tends to distort the reader s approach to various other topics.) We will not presuppose that the reader is already familiar with the terrain of current views in epistemology. With respect to content then, we will start at the very beginning. Moreover, the book will be written in a provocative manner that is designed to stimulate interest in the subject. In these respects, the book will be introductory in nature. However, we assume that the typical reader of the book will possess a certain level of philosophical sophistication. (In this respect then, the book is not a gentle introduction, of a kind that might be suitable for one who has no experience with philosophy at all.) Indeed, we anticipate that many of our readers will have already encountered and grappled with epistemological issues. Because it does not presuppose specific knowledge of the field, the book will be position to know, or all and only those propositions that are supported by one s evidence

4 especially suitable for students: primarily intermediate and advanced undergraduate philosophy majors, as well as beginning graduate students. As we have emphasized, however, the book will frequently address considerations and issues which we believe are either wholly absent or unduly neglected in the contemporary literature. For this reason, the book should be of considerable interest to the professional philosopher as well. The book is thus intended for a relatively wide readership. Comparable Recent Books In purporting to be both suitable as an introduction to the field as well as a substantive contribution to various contemporary debates, perhaps the most comparable recent book is Contemporary Theories of Knowledge, by John Pollock and Joseph Cruz (Rowman and Littlefield, 1999). Also comparable are Laurence Bonjour s Epistemology: Classic Problems and Contemporary Responses (Rowman and Littlefield, 2002), Robert Audi s Epistemology (Routledge, 1998), Keith Lehrer s Theory of Knowledge (Ridgeview, 1990), and Richard Feldman s Epistemology (Prentice-Hall, 2002). When used as an introductory or general survey text, our book will cohere well with Blackwell s collection of articles and excerpts, Epistemology: An Anthology (2000), edited by Ernest Sosa and Jaegwon Kim. The Length of the Book (where one s evidence is a function of one s knowledge).

5 The book will be relatively short: its projected length is approximately 80,000 words. Completion Date We expect to complete the book by the academic year of 2005-2006. Contents 1. The Ethics of Belief The attempt to think systematically about the question?what should I believe? has given rise to a number of philosophical projects. One such project is that of specifying epistemic norms. In Chapter 1, we discuss different conceptions of epistemic norms. We distinguish a number of different states and activities which epistemic norms might be thought to govern (e.g., one s present choices about what to believe, how one should revise one s beliefs over time, how one should gather evidence) and examine the relations between them. Here, we will look at the debate about evidentialism, explore questions about the extent to which our beliefs are under our control, as well as other issues that are sometimes discussed under the general heading of?the ethics of belief. 2. Probability and Changing One s Beliefs One natural answer to the question of?what should I believe? is the following:?you

6 should believe what is likely to be true or?you should believe what is probably true. In chapter 2, we explain the distinction between probability talk as describing objective chances and probability talk as describing how justified one is. We introduce and explain Bayesianism as an example of a well-worked out theory of epistemic probability that lays down normative constraints both on one s beliefs at a given time and on how one should revise one s beliefs over time. We also take up a number of issues surrounding the key concept of defeasibility. 3. Foundationalism and its Critics For a number of reasons, Bayesianism is best understood as a species of foundationalism. In Chapter 3, we explore the motivations for foundationalism, the view that some beliefs are immediately justified. We explore the crucial concept of immediate justification at some length. In particular, we examine issues surrounding the extent to which immediately justified beliefs can be defeasible, and we look in some depth at various coherentist arguments that purport to show that no beliefs are or could be immediately justified. We then examine foundationalist responses to such arguments. 4. Coherence and Inductive Reasoning In Chapter 4, we examine differing views about the role that considerations of coherence play in determining facts about what one should believe. We examine the concept of coherence in some detail and distinguish various roles which have been claimed for it

7 within epistemology. Here, we explain the distinction between linear and holistic conceptions of justification. We consider various challenges for the coherentist (e.g., concerns that the coherentist cannot account for perceptual or introspective knowledge) and discuss various coherentist replies to such challenges. Finally, we examine the role that coherence plays in explaining what distinguishes good inductive inferences from bad ones. 5. Naturalism and Normativity In Chapter 5, we examine competing views about the role that the natural sciences play in answering the question?what should I believe? We distinguish carefully between two different roles that the natural sciences might be thought to play: (i) as providing evidence for and against various propositions, and (ii) as determining the content of the epistemic norms themselves. We discuss various versions of epistemological naturalism, understood as the view that the natural sciences are highly relevant to (ii) as well as (i). We examine in some detail the traditional objection that naturalism cannot do justice to the distinctively normative aspect of epistemology and we consider various possible naturalist replies to this charge. 6. Externalism versus Internalism about Justification In attempting to provide a general answer to the question?what should I believe? it is natural to appeal to the concept of reliability. (Compare the following view about

8 testimony:?one should believe the testimony of reliable sources and not believe the testimony of unreliable sources.) In Chapter 6, we discuss reliabilism as the paradigm example of an externalist theory of justification. In our examination of reliabilism, we draw heavily on earlier discussions in the course of explaining (e.g.) why reliabilism has proven so appealing to many naturalists (Chapter 5) and the kind of probability which the reliabilist will invoke in attempting to explicate the concept of a reliable process (Chapter 2). We also examine the more general distinction between externalist and internalist theories of justification and make several suggestions about how the contemporary debate is best understood. 7. The Skeptic s Argument The final two chapters of the book are devoted to an extended discussion of skeptical challenges to our claim to have knowledge of the external world. In Chapter 7, we distinguish different kinds of skeptical scenarios and lay out in considerable detail what we take to be the most formidable argument available to the skeptic. (Again, our discussion here will draw heavily on distinctions and concepts that have been introduced in previous chapters.) We emphasize that the skeptic is a realist about the external world and take up the question of why it matters whether we are in one of the skeptic s scenarios. 8. Responding to the Skeptic

9 In Chapter 8, we survey and critically examine a number of the most interesting and important responses to skepticism. Special attention is given to relevant alternatives theories and contextualist responses to skepticism. We distinguish competing views about the nature of perceptual justification and explain how one s choice among them will influence one s estimation of the skeptic s argument. We also discuss certain more general, methodological issues that naturally arise in this context. These include: what is it for a given response to skepticism to be question-begging, and what is it reasonable for us to expect from a response to skepticism?