When we get evidence that tells us our belief-forming mechanisms may not be reliable this presents a thorny set of questions about whether and how to revise our original belief. This article analyzes aspects of the problem and a variety of approaches to its solution.

David Christensen and others argue that Dutch Strategies are more like peer disagreements than Dutch Books, and should not count against agents‘ conformity to ideal rationality. I review these arguments, then show that Dutch Books, Dutch Strategies, and peer disagreements are only possible in the case of what computer scientists call Byzantine Failures—uncorrected Byzantine Faults which update arbitrary values. Yet such Byzantine Failures make agents equally vulnerable to all three kinds of epistemic inconsistencies, so there is no principled basis for (...) claiming that only avoidance of true Dutch Books characterizes ideally rational agents. Agents without Byzantine Failures can be ideally rational in a very strong sense, but are not normative for humans. (shrink)

Enjoying great popularity in decision theory, epistemology, and philosophy of science, Bayesianism as understood here is fundamentally concerned with epistemically ideal rationality. It assumes a tight connection between evidential probability and ideally rational credence, and usually interprets evidential probability in terms of such credence. Timothy Williamson challenges Bayesianism by arguing that evidential probabilities cannot be adequately interpreted as the credences of an ideal agent. From this and his assumption that evidential probabilities cannot be interpreted as the actual credences of human (...) agents either, he concludes that no interpretation of evidential probabilities in terms of credence is adequate. I argue to the contrary. My overarching aim is to show on behalf of Bayesians how one can still interpret evidential probabilities in terms of ideally rational credence and how one can maintain a tight connection between evidential probabilities and ideally rational credence even if the former cannot be interpreted in terms of the latter. By achieving this aim I illuminate the limits and prospects of Bayesianism. (shrink)

There is a widespread view that in order to be rational we must mostly know what we believe. In the probabilistic tradition this is defended by arguments that a person who failed to have this knowledge would be vulnerable to sure loss, or probabilistically incoherent. I argue that even gross failure to know one's own beliefs need not expose one to sure loss, and does not if we follow a generalization of the standard bridge principle between first-order and second-order beliefs. (...) This makes it possible for a subject to use probabilistic decision theory to manage in a rational way cases of potential failure of this self-knowledge, as we find in implicit bias. Through such cases I argue that it is possible for uncertainty about what our beliefs are to be not only rationally permissible but advantageous. (shrink)

The author uses a series of examples to illustrate two versions of a new, nonprobabilist principle of epistemic rationality, the special and general versions of the metacognitive, expected relative frequency principle. These are used to explain the rationality of revisions to an agent’s degrees of confidence in propositions based on evidence of the reliability or unreliability of the cognitive processes responsible for them—especially reductions in confidence assignments to propositions antecedently regarded as certain—including certainty-reductions to instances of the law of excluded (...) middle or the law of noncontradiction in logic or certainty-reductions to the certainties of probabilist epistemology. The author proposes special and general versions of the MERF principle and uses them to explain the examples, including the reasoning that would lead to thoroughgoing fallibilism—that is, to a state of being certain of nothing. The author responds to the main defenses of probabilism: Dutch Book arguments, Joyce’s potential accuracy defense, and the potential calibration defenses of Shimony and van Fraassen by showing that, even though they do not satisfy the probability axioms, degrees of belief that satisfy the MERF principle minimize expected inaccuracy in Joyce’s sense; they can be externally calibrated in Shimony and van Fraassen’s sense; and they can serve as a basis for rational betting, unlike probabilist degrees of belief, which, in many cases, human beings have no rational way of ascertaining. The author also uses the MERF principle to subsume the various epistemic akrasia principles in the literature. Finally, the author responds to Titelbaum’s argument that epistemic akrasia principles require that we be certain of some epistemological beliefs, if we are rational. (shrink)

Formally-inclined epistemologists often theorize about ideally rational agents--agents who exemplify rational ideals, such as probabilistic coherence, that human beings could never fully realize. This approach can be defended against the well-know worry that abstracting from human cognitive imperfections deprives the approach of interest. But a different worry arises when we ask what an ideal agent should believe about her own cognitive perfection (even an agent who is in fact cognitively perfect might, it would seem, be uncertain of this fact). Consideration (...) of this question reveals an interesting feature of the structure of our epistemic ideals: for agents with limited information, our epistemic ideals turn out to conflict with one another. (shrink)

Certain situations seem to call for acknowledging the possibility that one’s own beliefs are biased or distorted. On the other hand, certain sorts of epistemic self-doubts (such as ‘I believe it’s raining, but it’s not’) seem paradoxical. And some have put forth epistemic principles requiring rational agents to regard their own credences as so-called ‘expert functions’. This paper examines the question of whether rationality requires agents to respect their own credences in a way in which they need not respect the (...) credences of others. (shrink)

Simon Evnine examines various epistemic aspects of what it is to be a person. Persons are defined as finite beings that have beliefs, including second-order beliefs about their own and others' beliefs, and are agents, capable of making long-term plans. It is argued that for any being meeting these conditions, a number of epistemic consequences obtain. First, all such beings must have certain logical concepts and be able to use them in certain ways. Secondly, there are at least two principles (...) governing belief that it is rational for persons to satisfy and are such that nothing can be a person at all unless it satisfies them to a large extent. These principles are that one believe the conjunction of one's beliefs and that one treat one's future beliefs as, by and large, better than one's current beliefs. Thirdly, persons both occupy epistemic points of view on the world and show up within those views. This makes it impossible for them to be completely objective about their own beliefs. Ideals of rationality that require such objectivity, while not necessarily wrong, are intrinsically problematic for persons. This "aspectual dualism" is characteristic of treatments of persons in the Kantian tradition. In sum, these epistemic consequences support a traditional view of the nature of persons, one in opposition to much recent theorizing. (shrink)

This is surely a bit of Socrates' famous irony. He draws the analogy to explain how his friends should regard poetry as they regretfully banish it from the ideal state. But lovers were no more sensible then than they are now. The advice to banish poetry, undermined already by Plato's own delight and skill in drama, is perhaps undermined still further by this evocation of a 'sensible' lover who counts love so well lost. Yet Socrates' image is one of avowed (...) rationality and prudence. The sensible lover imitates the older literary example of Ulysses' tying himself to the mast. (The example belongs therefore to the class of problems treated in Elster (1979)). Both this lover and Ulysses foresee that under certain possible future conditions, their opinions, values and preferences will or would differ from what they are now, in a very definite fashion. To what extent is such foresight possible? Correspondingly (when we do not claim foreknowledge) to what extent is such opinion reasonable, rational, coherent, or consistent in some suitably broad sense? It is not easy to understand exactly what is possible or even logically permissible in this respect. In an earlier paper, "Belief and the Will", I argued for a principle ("Reflection") to govern such deliberation. Here I will both generalize the treatment of opinion in "Belief and the Will" and respond to criticism. Critical examples mainly resembled the story of Ulysses who foresaw a period of dysfunction (at the sound 2 of the sirens) in his epistemic and/or doxastic future. Other criticism focused on the model of opinion used (precise numerical subjective probability) and on the merits of Dutch Book arguments. The present argument will not rely on Dutch Book arguments and strategies, and the Reflection principle will be formulated so as to apply also to vague opinion. (shrink)

The most immediately appealing model for formal constraints on degrees of belief is provided by probability theory, which tells us, for instance, that the probability of P can never be greater than that of (P v Q). But while this model has much intuitive appeal, many have been concerned to provide arguments showing that ideally rational degrees of belief would conform to the calculus of probabilities. The arguments most frequently used to make this claim plausible are the so-called "Dutch Book" (...) arguments. (shrink)

Conditionalization governs how to reallocate credence in light of new evidence. One prominent argument in favor of Conditionalization holds that an agent who violates it is vulnerable to a diachronic Dutch book: a series of acceptable bets offered at multiple times that inflict a sure loss. van Fraassen argues that an agent who violates the Principle of Reflection is likewise vulnerable to a diachronic Dutch book. He concludes that agents should conform to both Conditionalization and Reflection. Some authors reply that (...) Reflection is implausible and hence that there must be something wrong with diachronic Dutch book arguments. Other authors try to isolate a principled difference between the Dutch book argument for Conditionalization and the Dutch book argument for Reflection, such that the former argument may succeed even though the latter fails. I pursue a version of this strategy. I contend that, once we properly elucidate the notion of sure loss, non‐reflectors are not vulnerable to a sure loss. An agent who violates Reflection is not thereby subject to a diachronic Dutch book. Appearances to the contrary result from an unmotivated focus upon an overly narrow set of gambling scenarios. (shrink)

This paper offers two new arguments for a version of Reflection, the principle that says, roughly, that if one knew now what one would believe in the future, one ought to believe it now. The most prominent existing argument for the principle is the coherence-based Dutch Strategy argument advanced by Bas van Fraassen (and others). My two arguments are quite different. The first is a truth-based argument. On the basis of two substantive premises, that people’s beliefs generally get better over (...) time and that being a person requires having knowledge of this fact, it concludes that it is rational to treat your future selves as experts. The second argument is a transcendental one. Being a person requires being able to engage in plans and projects. But these cannot be meaningfully undertaken unless one has Reflection-like expectations about one’s future beliefs. Hence, satisfaction of Reflection is necessary for being a person. Together, the arguments show that satisfaction of Reflection is both rational and necessary for persons. (shrink)

I bring together social ontology and social epistemology by consideringsocial entities (``epistemic unities'') that are constituted by the holdingof epistemic relations between their members. In particular, I focus onthe relation of taking someone as an expert. Among the types of structuresexamined are ones with a single expert and one or more non-experts whomay or may not know of each other's situation; and ones with more thanone expert, including cases in which the relation between the experts ishierarchical and cases in which (...) it is symmetrical. These structures modela variety of social situations which can thus be given a unified treatment.Among the cases I discuss are persons, which I argue are multiple-expertunities of persons at times. Taking a person as a social unity like this offersa clear sense in which some groups can also be person-like. (shrink)

The Preface Paradox has led many philosophers to believe that, if it isassumed that high probability is necessary for rational acceptability, the principleaccording to which rational acceptability is closed under conjunction (CP)must be abandoned. In this paper we argue that the paradox is far less damaging to CP than is generally believed. We describe how, given certain plausibleassumptions, in a large class of cases in which CP seems to lead tocontradiction, it does not do so after all. A restricted version (...) of CP canthus be maintained. (shrink)

It is plausible that Newcomb problems in which causal maximizers and evidential maximizers would do different things would not be possible for ideal maximizers who are attentive to metatickles. An objection to Eells’s first argument for this makes welcome a second. Against it I argue that even ideal evidential and causal maximizers would do different things in some non-dominance Newcomb problems; and that they would hope for different things in some third-person and non-action problems, which is relevant if a good (...) theory of rational choices of acts should fit smoothly into a good theory of rational desires for facts. (shrink)

Rational actions reflect beliefs and preferences in certain orderly ways. The problem of theory is to explain which beliefs and preferences are relevant to the rationality of particular actions, and exactly how they are relevant. One distinction of interest here is between an agent's beliefs and preferences just before an action's time, and his beliefs and preferences at its time. Theorists do not agree about the times of beliefs and desires that are relevant to the rationality of action. Another distinction (...) is between actions that would, in one sense or another, maximize expected utilities given relevant beliefs and desires, and actions, decisions for which would, in one sense or another, be stable. There is disagreement about the relevance of these things to the rationality of actions. Here, in the first part below, several possible positions on these issues are explained and compared. In the second part, I contrast perspectives on these issues, and comment on arguments that might be brought against the position I favor. In the third part, I restate and elaborate on this position. (shrink)

Horgan and Mahtani (Erkenntnis 78: 333–351, 2013) present a new argument for the 1/3 answer to the Sleeping Beauty problem resting on a principle for updating probabilities which they call “generalized conditionalization.” They allege that this new argument is immune to two attacks which have been recently leveled at other arguments for thirdism. I argue that their new argument rests on a probability distribution which is (a) no more justified than an alternative distribution favoring a different answer to the problem, (...) and (b) ultimately unjustified. I go on to show that generalized conditionalization cannot be applied in the manner suggested, given the cogency of the aforementioned attacks on thirder arguments. Hence, the new argument fails to advance the case for the 1/3 answer. (shrink)

There are narrowest bounds for P(h) when P(e) = y and P(h/e) = x, which bounds collapse to x as y goes to 1. A theorem for these bounds -- bounds for probable modus ponens -- entails a principle for updating on possibly uncertain evidence subject to these bounds that is a generalization of the principle for updating by conditioning on certain evidence. This way of updating on possibly uncertain evidence is appropriate when updating by ’probability kinematics’ or ’Jeffrey-conditioning’ is, (...) and apparently in countless other cases as well. A more complicated theorem due to Karl Wagner -- bounds for probable modus tollens -- registers narrowest bounds for P(not h) when P(not e) = y and P(e/h) = x. This theorem serves another principle for updating on possibly uncertain evidence that might be termed ’contraditioning’, though it is for a way of updating that seems in practice to be frequently not appropriate. It is definitely not a way of putting down a theory -- for example, a random-chanc. (shrink)

Fitch’s paradox shows, from fairly innocent-looking assumptions, that if there are any unknown truths, then there are unknowable truths. This is generally thought to deliver a blow to antirealist positions that imply that all truths are knowable. The present paper argues that a probabilistic version of antirealism escapes Fitch’s result while still offering all that antirealists should care for.

In this paper, a problem for utility theory - that it would have an agent who was compelled to play “Russian Roulette’ with one revolver or another, to pay as much to have a six-shooter with four bullets relieved of one bullet before playing with it, as he would be willing to pay to have a six-shooter with two bullets emptied - is reviewed. A less demanding Bayesian theory is described, that would have an agent maximize expected values of possible (...) total consequence of his actions. And utility theory is located within that theory as valid for agents who satisfy certain formal conditions, that is, for agents who are, in terms of that more general theory, indifferent to certain dimensions of ‘risk’. Raiffa- and Savage-style arguments for its more general validity are then resisted. Addenda are concerned with implications for game theory, and relations between ‘utilities’ and ‘values’. (shrink)

In Belief and the Will, van Fraassen employed a diachronic Dutch Book argument to support a counterintuitive principle called Reflection. There and subsequently van Fraassen has put forth Reflection as a linchpin for his views in epistemology and the philosophy of science, and for the voluntarism (first-person reports of subjective probability are undertakings of commitments) that he espouses as an alternative to descriptivism (first-person reports of subjective probability are merely self-descriptions). Christensen and others have attacked Reflection, taking it to have (...) unpalatable consequences. We prescind from the question of the cogency of diachronic Dutch Book arguments, and focus on Reflection's proper interpretation. We argue that Reflection is not as counterintuitive as it appears — that once interpreted properly the status of the counterexamples given by Christensen and others is left open. We show also that descriptivism can make sense of Reflection, while voluntarism is not especially well suited to do so. (shrink)

In a recent article, David Christensen casts aspersions on a restricted version of van Fraassen's Reflection principle, which he dubs ‘Self-Respect’(sr). Rejecting two possible arguments for sr, he concludes that the principle does not constitute a requirement of rationality. In this paper we argue that not only has Christensen failed to make a case against the aforementioned arguments, but that considerations pertaining to Moore's paradox indicate that sr, or at the very least a mild weakening thereof, is indeed a plausible (...) normative principle. (shrink)

The orthodoxy that conditional probabilities reflect what are for a subject evidential bearings is seconded. This significance suggests that there should be principles equating rationally revised probabilities on new information with probabilities reached by conditionalizing on this information. Several principles, two of which are endorsed, are considered. A book is made against a violator of these, and it is argued that there must be something wrong with a person against whom such books can be made. Appendices comment on Popper-functions, elaborate (...) on bets and odds, and relate dutch books and strategies to conditions of inconsistency (Ramsey's idea) and imperfection. (shrink)

Relations between conditional probabilities, revisions of probabilities in the light of new information, and conditions of ideal rationality are discussed herein. The formal character of conditional probabilities, and their significance for epistemic states of agents is taken up. Then principles are considered that would, under certain conditions, equate rationally revised probabilities on new information with probabilities reached by conditionalizing on this information. And lastly the possibility of kinds of ‘books’ against known non-conditionalizers is explored, and the question is taken up, (...) What, if anything, would be wrong with a person against whom such a book could be made?1. The standard mathematical treatment takes probability functions as basic, and conditional probabilities defined. If P(-) is a probability function, then is its conditional probability function if and only if P(h/e) is defined only for e such that P(e) > 0, and is then equal to P(e & h)/P(e). (shrink)

Bradley has argued that a truth-conditional semantics for conditionals is incompatible with an allegedly very weak and intuitively compelling constraint on the interpretation of conditionals. I argue that the example Bradley offers to motivate this constraint can be explained along pragmatic lines that are compatible with the correctness of at least one popular truth-conditional semantics for conditionals.

According to the Dutch Book Argument (DBA), if an agent's subjective probabilities fail to satisfy the axioms of the probability calculus and so make the agent vulnerable to a Dutch Book, the agent's subjective probabilities are incoherent and the agent is therefore irrational. Critics of DBA have argued, however, that probabilistic incoherence is compatible with various kinds of rationality – logico-semantic, epistemic, instrumental and prudential. In this paper, I provide an interpretation of DBA on which it is true that probabilistic (...) incoherence entails agent irrationality. Articulating this interpretation requires the specification of some of the connections among the varieties of rationality. Once this is done, it becomes possible to vindicate a modest version of DBA. (shrink)