ABSTRACT. According to a probabilistic conception of rationality, for any case that a thinker can be in, the degree to which the belief-system that the thinker has in that case counts as rational is given by that belief-system’s distance from a certain probability function – call it the “rational probability function” for that case. Are there diachronic constraints on this rational probability function—constraints that the thinker’s past beliefs impose on which probability function counts as this rational probability function? Some philosophers—such as Brian Hedden—object that the idea of such diachronic constraints is inconsistent with internalism about rationality. These objections are answered here. The thinker’s past beliefs continue to exert an influence over the present. This lingering influence of the thinker’s past beliefs will be referred to as the “doxastic shadow of the past”. It is suggested that this rational probability function is constrained by this shadow in the following way: first, consider the “shadow priors”—the probability functions that, of all those that meet all the relevant synchronic constraints, are closest to this doxastic shadow; then take the functions that result from conditionalizing these shadow priors on all the facts that are in the relevant way certain for the thinker in the case. The rational probability functions for the case are the closest of these probability functions to the belief-system that the thinker has in this case. Finally, a proposal is given about what explains why this view of diachronic rationality is correct.

ABSTRACT. Contrastivism about epistemic reasons is the claim that reasons to believe are relative to sets of alternatives. A serious problem for contrastivism is that it fails to validate perfectly good inferences across different sets of alternatives. I explore two ways of remedying the situation. The first one provides the required constraints via a principle linking reasons to promotion of relevant objectives. The second does it by defining the notion of epistemic reason at the expense of conditional probability. I argue that the second account is to be preferred. I conclude by raising an issue with this account and exploring a few ways to resolve it.

ABSTRACT. Uniqueness theorists and permissivists generally agree that the ultimate epistemic goal is to maximize expected accuracy. Against the claim that accuracy is the ultimate epistemic value, I argue that understanding is another contributory value. If so, to what extent a doxastic state is justified may depend not only on its expected accuracy value, but also on its expected understanding value. Regardless of whether the two sources of contributory value can be aggregated, epistemic value pluralism has ramifications for the (im)permissivism debate. In situations where each epistemic goal recommends a different credence, more than one doxastic state may be rationally permissible.

ABSTRACT. According to Alvin Goldman's process reliabilism, a belief is justified if it is acquired through a reliable process – a process that leads to predominantly true beliefs or a high ratio of true to false beliefs. One of the challenges for the theory is the generality problem – the issue of individuating the process types relevant for forming a particular belief token. I will review a group of proposed solutions for the problem, the so-called psychological approaches, and argue that combining philosophical methods with the empirical disciplines that investigate human cognition is, in the spirit of moderate naturalism, crucial for addressing the generality problem.

ABSTRACT. We argue for a belief norm of assertability: you are allowed to assert a proposition just in case you are justified to believe the proposition. Along the way, we offer a suitable notion of justified belief and compare the norm favourably to the knowledge norm of assertion. We conclude that assertion requires no more than justified belief.

ABSTRACT. Credences have mostly been studied in formal epistemology, whereas belief (and its potential to qualify as knowledge) takes centre stage in traditional epistemology. In my talk, I will investigate one natural point of connection between these ways of doing epistemology, namely the relation between credences and probability beliefs. I will defend what I call the Interdoxastic Coherence Principle (ICP), which says that for any subject S who has a credence and a probability belief about p: S is epistemically rational only if (S has a credence x in p iff S has a belief that the probability of p is x). I will provide two arguments for thinking that this coherence requirement for credences and probability belief holds. I will first argue that ICP explains a new class of puzzling cases involving probabilistic incoherence. I will then provide a Dutch book argument for ICP that shows that violating the latter makes on susceptible to sure loss.

ABSTRACT. Leitgeb (2017) proposes a theory for linking categorical beliefs and degrees of credence that does not face the lottery paradox. According to his theory, one can rationally believe a proposition, if and only if they can maintain a stable high credence towards it. I will use his theory to argue that, as laymen, we cannot rationally believe stereotypical statistical propositions, because they are always under the threat of being undermined by the report of a serious bias.

ABSTRACT. Pragmatic responses to skepticism treat it as a decision and argue that it is a bad one. Standard versions use decision theories that vindicate implausible belief-forming strategies (e.g. “believe everything that seems false”). I offer a better version, based in subjective epistemic value, using an obscure but well-founded decision theory. It only vindicates plausible belief-forming strategies. An objection is that I’m discussing rational belief-forming strategies, not justified beliefs. But I have proven to the skeptic that, in the pursuit of truth, we can rationally form non-accidentally true beliefs. These either are justified beliefs or we should not care about justification.

ABSTRACT. There are several competing probabilistic measures of coherence, but all major contenders face difficulties in evaluating coherence relations among known facts. I argue that taking inspiration from information theory can help us to solve this issue, and to develop a more thorough understanding of coherence in general. To demonstrate the potential benefits of applying information theory, I will tentatively propose a measure based on conditional mutual information, which is also related to correlation-based measures of coherence. A central feature of this proposal is that probability assignments are relative to a background hypothesis, which encodes information about prior distributions and the target system.

ABSTRACT. The development of Formal Epistemology resembles the development of logic during the second half of nineteenth century. Then logic, like now epistemology, turned from a philosophical into a formal discipline. The focus is on the epistemic subject in formal epistemology resembles Mill’s position, who wanted to turn logic into a rigorous discipline using psychology. The Fregean approach in logic replaced reasoning understood as a mental process by reasoning as relation between formulas. The Fregean approach in epistemology aims to do something similar with justification—to see it as a relation between theories.

ABSTRACT. The hallmark of knowledge is that the fact that p, by itself, does a better job of explaining S’s truly believing about whether p, given method m, in circumstances c, than a more complex explanans which mentions not only the fact that p, but also additional facts needed to make sense of S’s epistemic success. If there is the need to cite additional facts, the true belief does not qualify as knowledge. If S knows, however, then <p>, by itself, suffices to explain her epistemic success.