According to the traditional understanding, ethical normativity is about what you should do and epistemic normativity is about what you should believe. Singer’s topic in Right Belief and True Belief is the latter. However, though he later rejects this traditional understanding of the distinction (pp. 205–7), he thinks we can learn a great deal from looking at the parallels between these two species of normativity, and his book provides a masterclass in how to do that: this is epistemology as practised by someone very much at home in ethics and well versed in its contemporary literature, its arguments, distinctions, and central positions.

In the first chapter, Singer distinguishes a number of different normative notions to which we appeal when we evaluate beliefs: Is the belief correct? Is it right? Should we believe it? Ought we to? Must we? These he calls ‘deontic notions’, and we use them to evaluate the belief with respect to the believer. But there are also these: Is it praiseworthy or blameworthy to have the belief? Is the believer at fault if they do? Are they rational? Is the belief justified for them? These he calls ‘responsibility notions’, and we use them to evaluate the believer with respect to the belief (pp. 73–74). This distinction he calls bipartite (p. 189).

In the first six chapters, Singer’s interest lies in the deontic notions, and indeed in the notion of right belief. In chapter 1, he argues that what we should believe is what it is epistemically right for us to believe and, in chapter 2, he formulates his account of right belief, which he calls ‘truth-loving epistemic consequentialism’:

A belief that P is right for an agent to have if and only if among the available options for belief, having a belief that P promotes the greatest overall balance of having true beliefs and not having false beliefs. (p. 48)

Truth-loving epistemic consequentialism combines a number of claims. The first is a veritist account of epistemic value. Here’s Singer formulation: ‘all and only true beliefs have final epistemic value, and all and only false beliefs have final epistemic disvalue’ (p. 48). But, added to this, he assumes we can specify the epistemic value not only of individual doxastic attitudes—such as the belief that P or the disbelief that Q—but also of whole doxastic states, which comprise many different beliefs, disbeliefs, and suspensions of judgement; and indeed we can specify the epistemic value of a series of such whole doxastic states, one for each future point in time. It is the overall epistemic value of the series of future whole doxastic states of the believer produced by the doxastic state they adopt at the present time that determines whether a belief in that doxastic state is epistemically right or not. Singer calls this claim ‘diachronic’ (p. 190). A belief is then right if it belongs to a whole doxastic state that produces the future series of whole doxastic states that maximizes overall epistemic value. We might call this claim ‘maximizing’.

So, this is truth-loving epistemic consequentialism, and the remainder of the book is a defence of this position against a series of objections. Notably, truth-loving epistemic consequentialism entails that right belief is not necessarily true belief, though it very often will be. To see this, consider Carrie Jenkins’s ([2007]) wonderfully named example of the truth fairy. Suppose you have very strong perceptual evidence that there is a mug of tea in front of you (call this proposition T). The truth fairy offers you an epistemic deal: if you believe T, she’ll mess around with the world so that nearly all your other beliefs now and in the future are false; but if you disbelieve T, she’ll arrange things so that nearly all your other beliefs now and in the future are true. Singer holds that the correct doxastic attitude to T is belief, since T is true, but claims that the right attitude is disbelief, since that is what produces the greatest veritist epistemic value.

Jenkins’s case in a classic example of what has come to be known as an epistemic ‘trade-off’ in the literature, and many epistemologists, including Jenkins herself, Roderick Firth ([1998]), Selim Berker ([2013]), and Hilary Greaves ([2013]), have objected to veritist versions of epistemic consequentialism on the grounds that they permit or require such trade-offs.

Singer devotes chapters 2 and 3 to addressing these objections. His defence involves a number of moves: First, he notes that the most compelling versions of these objections target veritist versions of epistemic consequentialism about justified or rational belief, not right belief. These theories say you’re rational or justified in disbelieving T in the truth fairy case, and surely that’s absurd. But is it so absurd to say that you’re right to disbelieve T, which is the relevant consequence of truth-loving epistemic consequentialism? Singer thinks not.

Second, he responds to Berker’s extreme version of the variety of objection—which says that no epistemic trade-off is ever permissible—by noting a number of cases in which we do seem to permit it. If a seven-year-old child is interested in spaceship design, but is not yet capable of grasping Einstein’s theory of relativity, we think it right to teach them Newtonian mechanics instead, because that will best promote epistemically valuable doxastic states in the future, even giving them the basis on which to build their understanding of relativity when they are capable of grasping it.

Third, he treats the trade-off cases that we do not intuitively permit by, first, offering an error theory for our intuitions in those cases and, second. borrowing from the ethical consequentialists the notion of a sophisticated consequentialism. On the first point: In the vast majority of cases, whether or not we have a belief that P does not affect the truth of P, nor the truth of any other propositions we consider, and so we tend to assume that, in Singer’s words, ‘truth begets truth’ (p. 65). So, we adopt the following heuristic: a belief that P is epistemically right if P is true. And that heuristic drives the intuitions that clash with truth-loving epistemic consequentialism in the truth fairy case. The heuristic is strictly false, but it gets things right in the vast majority of cases.

On the second point: Relatedly, Singer suggests that we take a lead from Peter Railton’s ([1984]) sophisticated ethical consequentialism, according to which, while right belief is belief that produces the greatest veritist epistemic value, we may well do best, epistemically speaking, not by trying to hold right beliefs in each individual case, but by fostering those dispositions to believe that produce the greatest epistemic value for us. After all, we are limited, fallible creatures, and so if we try always to hold right beliefs, we might often fail or we might succeed only by wasting a great deal of time calculating what is the right thing to believe. Better, instead, to foster dispositions, heuristics, and rules of thumb that are more efficient, easier to implement, and, among those available to our limited cognitive capacities, best promote veritist value. Singer calls the claim that we should distinguish the correct theory of right belief from the theory, or heuristic, that individual believers should use when coming to believe ‘sophisticated’ (p. 189); and he calls the claim that we can epistemically evaluate things other than doxastic states, such as these heuristics for forming beliefs, ‘global’ (p. 189).

In chapter 4, Singer moves on from the trade-off objections to consider two further worries about veritist versions of epistemic consequentialism: the problem of the value of uninteresting beliefs, which dates back at least to the debate between Jane Heal ([1988]) and Susan Haack ([1995]), and an objection by Clayton Littlejohn ([2018]) that there is no veritist version of the notion of epistemic good that it is appropriate to promote. In response to the first worry, he agrees that the veritism behind truth-loving epistemic consequentialism entails that uninteresting beliefs contribute to epistemic value, but he notes that it does not follow that people should pursue uninteresting beliefs, since doing so is very rarely the best use of their time. In response to Littlejohn’s worry, he begins to sketch an account of epistemic goodness, but leaves it as a sketch.

In chapter 5, Singer considers epistemic utility theory, an approach to Bayesian epistemology that has grown out of work by Graham Oddie ([1997]), Jim Joyce ([1999]), and Hilary Greaves and David Wallace ([2008]). It appeals to a veritist account of the epistemic value of partial beliefs, or credences, to establish the Bayesian norms that govern them. Singer notes that there appear to be clashes between truth-loving epistemic consequentialism and the approach taken in epistemic utility theory, particularly concerning whether we should have credences that most accurately represent the world, or whether we should have credences that best influence the world to make those credences accurate. But he concludes that since epistemic utility theory has typically considered rational belief and not right belief, the tensions are only apparent.

In chapter 6, Singer asks how the truth-loving epistemic consequentialist would treat the sort of case that Tamar Gendler ([2011]) describes, in which there is some true statistical fact about the distribution of an attribute in a population that suggests a specific belief about a particular member of the population that strikes us as prejudiced and immoral. Gendler’s original example concerns a belief about the status of a particular person in a country club on the basis of their race; this belief is suggested by the true statistical fact that among people of that race in the country club, many more have the lower status than the higher status. And Singer draws on Rima Basu’s ([2018]) example, which concerns a belief about how particular diners will tip in a restaurant on the basis of their race. Singer ends up taking what he calls a ‘dilemmist’ position here: the specific belief about the particular member of the group is both epistemically right and morally wrong.

In chapter 7, Singer summarizes his position by drawing attention to the features that he dubs bipartite, sophisticated, global, veritist, and diachronic. To this we might add also maximizing. He then begins to ask how we should think of the responsibility notions, which he separated out from the deontic ones in the second chapter and has not analysed since. This is of some importance, since it is natural to think that in cases in which all the details of the situation are known to the believer, the epistemically right and the epistemically rational should coincide. And if this is right, then it undermines Singer’s attempt to take the sting from some of the trade-off objections by noting that he defends only a truth-loving epistemic consequentialist account of right belief and not such an account of rational belief, for in those cases the believer is fully aware of the details of the situation. But Singer rejects such an account. For him, the responsibility notions are components of a social practice of giving praise and ascribing blame that we use to try to encourage others to set their beliefs in ways that conduce better to veritist value. Of course, you might wonder why such a practice works, unless the person praised or blamed conceives of those terms as saying something genuinely good or bad about them.

Singer is explicit from the outset that his methodology is roughly the standard methodology of normative theorizing in contemporary analytic philosophy. Intuitive judgements—which might concern particular cases such as Jenkins’s truth fairy, general principles such as veritism, or even features of the normative theory itself such as the ingredients of the dependency basis for claims of right belief—are treated as data and we seek a theory that best explains that data, usually by entailing the intuitive judgements are correct, but sometimes by explaining them away with an error theory. His book is an admirable application of that methodology in the domain of epistemic normativity, and provides a robust defence of one component of a veritist version of epistemic consequentialism of the sort that has proven increasingly popular in recent years—the component that concerns right belief. What’s more, this book acts as a prolegomenon to Singer’s other work, where he uses computational methods to tease out the verdicts of truth-loving epistemic consequentialism in social epistemology, whether it is to understand polarization, epistemic democracy, the epistemic of forgetting, or the epistemic virtues of diversity. Here, the theoretical foundations for that work are laid.

Richard Pettigrew
University of Bristol
richard.pettigrew@bristol.ac.uk

References

Basu, R. [2018]: ‘The Wrongs of Racist Beliefs’, Philosophical Studies, 176, pp. 2497–515.

Berker, S. [2013]: ‘Epistemic Teleology and the Separateness of Propositions’, Philosophical Review, 122, pp. 337–93.

Firth, R. [1998]: ‘Part III: The 1978 Schneck Lectures’, in J. Troyer (ed.), In Defense of Radical Empiricism: Essays and Lectures by Roderick Firth, Lanham, MD: Rowman and Littlefield, pp. 317–70.

Gendler, T. S. [2011]: ‘On the Epistemic Costs of Implicit Bias’, Philosophical Studies, 156, pp. 33–63.

Greaves, H. [2013]: ‘Epistemic Decision Theory’, Mind, 122, pp. 915–52.

Greaves, H. and Wallace, D. [2008]: ‘Justifying Conditionalization: Conditionalization Maximizes Expected Epistemic Utility’, Mind, 115, pp. 607–32.

Haack, S. [1995]: ‘Concern for Truth: What It Means, Why It Matters’, Annals of the New York Academy of Sciences, 775, pp. 57–63.

Heal, J. [1988]: ‘The Disinterested Search for Truth’, Proceedings of the Aristotelian Society, 88, pp. 97–108.

Jenkins, C. S. [2007]: ‘Entitlement and Rationality’, Synthese, 157, pp. 25–45.

Joyce, J. M. [1999]: ‘A Nonpragmatic Vindication of Probabilism’, Philosophy of Science, 65, pp. 575–603.

Littlejohn, C. [2018]: ‘The Right and the Good: A Defense of Teleological Non-consequentialism in Epistemology’, in J. Dunn and K. Ahlstrom-Vij (eds), Epistemic Consequentialism, Oxford: Oxford University Press, pp. 23–47.

Oddie, G. [1997]: ‘Conditionalization, Cogency, and Cognitive Value’, British Journal for the Philosophy of Science, 48, pp. 533–41.

Railton, P. [1984]: ‘Alienation, Consequentialism, and the Demands of Morality’, Philosophy and Public Affairs, 13, pp. 134–71.