In the epistemology of disagreement, Conciliationism argues that encountering peer disagreement on controversial matters rationally requires belief suspension, as the disagreement itself defeats justification. This implies an extensive skepticism about philosophical statements. This contribution defends the consistency of the skeptical argument from disagreement against claims of self-defeat by disentangling it from Conciliationism and grounding it instead in the Pyrrhonian dialectic of the criterion. The argument is structured around Thomas Grundmann’s critique that the skeptical argument relies on justified philosophical beliefs, contravening its conclusion. The presentation first challenges Grundmann’s analysis of self-defeat. It then reconstructs a more viable Pyrrhonian disagreement-based skeptical argument independent of Conciliationism. This improved argument avoids the self-defeating charge and is not bound to command suspension of judgment. This makes disagreement skepticism compatible with practical reasons for retaining one’s philosophical beliefs.

The principle of total evidence (PTE) asserts that rational agents should conditionalize their degrees of belief on the total evidence that is accessible to them. In the first part of this talk I will justify the optimality of the PTE from the viewpoint of individual rationality, in the sense of expected utility maximization. My proposed justification of the PTE is based on a seminal proof of Good (1967) which shows that by conditionalizing degrees of beliefs on new pieces of evidence or information, the expected individual utility of the actions of a rational agent can only increase but not decrease.

In the second part I will demonstrate that in spite of its individual optimality, the PTE may become suboptimal for the collective rationality when interests are in conflict. This scenario can arise when the acquiration of a certain piece of information by a rational agent carries a great risk of being "misused" in a way that, in spite of being useful for the individual, causes great harm to others and, thus, decreases the overall utility-sum (or aggregated utility) for the whole collective.

Recent years have seen the emergence of what Martin Smith [2022] called the hardest paradoxes for closure to date. These hybrid paradoxes combine features of the lottery and the preface paradox in a way that (i) puts pressure on the claim that justification is closed under conjunction (henceforth, simply closure) but (ii) resists familiar closure-preserving solutions previously put forward in response to the original paradoxes. This gives rise to the following question: can closure be saved from the new hybrid paradoxes and, if so, how? This paper puts forward what is likely to be the most promising (general) strategy for solving the new paradoxes in a way that's compatible with closure, considers the most prominent attempts to implement this strategy, but argues that they are ultimately unsuccessful. Hence, it's concluded that the case against closure still stands. The paper also corrects some misconceptions about the nature of the hybrid paradoxes and weighs in on the question of what really is the hardest paradox for closure.

We tend to prefer those theories of which we know (or suspect) that they

are closer to truth. The relation of having greater or lesser closeness-to-truth

(verisimilitude, truthlikeness) can therefore be interpreted as a preference rela-

tion and investigated within the context of the rich research about ‘combinative

preferences’ and preference logic. The general concept of verisimilitude has been

studied extensively over the course of the past fifty years. So far, however, this

research programme has rarely examined the logical properties of the relation

‘closer to truth’, and a general logic of truthlikeness is still lacking. Our present

work constitutes a step towards such a logic. We focus on a particular ‘question-

relative’ relation of comparative verisimilitude that is defined in terms of how

well the competing theories address a set of underlying questions. We show

that verisimilitude has logical properties that are very much different from those

of other preference-relations, such as relations of plausibility, or of comparative

possibility. Nonetheless, verisimilitude does also satisfy a number of interesting

axioms and properties that are shared by other preference-relations.