Michael Cohen

Bootstrapping is a suspicious form of reasoning that seems to allow agents to gain knowledge about the reliability of their sources of information ‘from thin air’, without gathering independent evidence about those sources. The bootstrapping problem is the problem of explaining what is wrong with bootstrapping reasoning. I identify one defect in bootstrapping reasoning by making explicit an implicit dynamic assumption which is made in such reasoning, involving the perfect-recall principle. The defect stems from a fallacious reasoning pattern that arises by misapplication of perfect-recall. I argue that understanding this defect can explain away some aspects of bootstrapping reasoning.

The geometry of the environment can affect numerous psychological, social, and ecological processes. But its roles in social learning and the dynamics of descriptive norms remain unclear. Here we use agent-based modeling to explore how environments with different geometric shapes can constrain social learning to produce universally shared descriptive norms. Our simulations show that an environment with an irregular layout facilitates the emergence of multiple descriptive norms in a population, whereas an environment with a regular grid plan constrains social learning to produce a behaviorally homogeneous population.

Williamson (2000) argues that the KK principle is inconsistent with knowledge of margin for error in cases of inexact perceptual observations. This paper argues, primarily by analogy to a different scenario, that Williamson's argument is fallacious. Margin for error principles describe the agent's knowledge as a result of an inexact perceptual event, not the agent's knowledge state in general. Therefore, epistemic agents can use their knowledge of margin for error at most once after a perceptual event, but not more. This insight blocks a crucial step in Williamson's original argument. Along the way, the value of standard epistemic logic for analyzing margin for error reasoning is challenged.

Epistemic closure refers to the assumption that humans are able to recognize what entails or contradicts what they believe and know, or more accurately, that humans’ epistemic states are closed under logical inferences. Epistemic closure is part of a larger theory of mind ability, which is arguably crucial for downstream NLU tasks, such as inference, QA and conversation. In this project, we introduce a new automatically constructed natural language inference dataset that tests inferences related to epistemic closure. We test and further fine tune the model RoBERTa-large-mnli on the new dataset, with limited positive results.

Within psychology, philosophy, and cognitive science, theory of mind refers to the cognitive ability to reason about the mental states of other people, thus recognizing them as having beliefs, knowledge, intentions and emotions of their own. In this project, we construct a natural language inference (NLD) dataset that tests the ability of a state of the art language model, RoBERTa-large finetuned on the MNLI dataset, to make theory of mind inferences related to knowledge and belief. Experimental results suggest that the model struggles with such inferences, including after attempts for further finetuning.

Cases of inexact observations have been used extensively in the recent literature on higher-order evidence and higher-order knowledge. I argue that the received understanding of inexact observations is mistaken. Although it is convenient to assume that such cases can be modeled statically, they should be analyzed as dynamic cases that involve change of knowledge. Consequently, the underlying logic should be dynamic epistemic logic, not its static counterpart. When reasoning about inexact knowledge, it is easy to confuse the initial situation, the observation process, and the result of the observation; I analyze the three separately. This dynamic approach has far reaching implications: Williamson’s influential argument against the KK principle loses its force, and new insights can be gained regarding synchronic and diachronic introspection principles.

If updating with E has the same result across all epistemically possible worlds, then the agent has no uncertainty as to the behavior of the update, and we may call it a transparent update. If an agent is uncertain about the behavior of an update, we may call it opaque. In order to model the uncertainty an agent has about the result of an update, the same update must behave differently across different possible worlds. In this paper, I study opaque updates using a simple system of dynamic epistemic logic suitably modified for that purpose. The paper highlights the connection between opaque updates and the dynamic-epistemic principles Perfect-Recall and No-Miracles. I argue that opaque updates are central to contemporary discussions in epistemology, in particular to externalist theories of knowledge and to the related problem of epistemic bootstrapping, or easy knowledge. Opaque updates allow us to explicitly investigate a dynamic form of uncertainty, using simple and precise logical tools.

The problem of perception is the problem of explaining how perceptual knowledge is possible. The skeptic has a simple solution: it is not possible. I analyze the weaknesses of one type of skeptical reasoning by making explicit a dynamic epistemic principle from dynamic epistemic logic that is implicitly used in debating the problem, with the aim of offering a novel diagnosis to this skeptical argument. I argue that prominent modest foundationalist responses to perceptual skepticism can be understood as rejecting the dynamic assumption made by the skeptic, that there are independent reasons to doubt the truth of such a principle in the context of skeptical reasoning, and that making the dynamic principle explicit allows for a better understanding of at least one objection to modest foundationalism.

According to one form of epistemic contrastivism, due to Jonathan Schaffer, knowledge is not a binary relation between an agent and a proposition, but a ternary relation between an agent, a proposition, and a context-basing question. In a slogan: to know is to know the answer to a question. I argue, first, that Schaffer-style epistemic contrastivism can be semantically represented in inquisitive dynamic epistemic logic, a recent implementation of inquisitive semantics in the framework of dynamic epistemic logic; second, that within inquisitive dynamic epistemic logic, the contrastive ternary knowledge operator is reducible to the standard binary one. The reduction shows, I argue, that Schaffer’s argument in favor of contrastivism is compatible with a binary picture of knowledge. This undercuts the force of the argument in favor of contrastivism.

Recent years have witnessed an effort to explicate the concept of explanatory power in a Bayesian framework by constructing explanatory measures. It has been argued that those measures should not violate the principle of explanatory justice, which states that explanatory power cannot be extended “for free.” I argue, by formal means, that one recent measure claiming to be immune from explanatory injustice fails to be so. I end by concluding that the explanatory justice criticism can be dissolved, given a natural interpretation of the concept of negative explanatory power.