My research interests focus primarily on topics in social epistemology. My doctoral dissertation is concerned with how dependencies between sources of evidence affect the epistemic success of deliberating groups. I’m working analytically and formally, sometimes using simulations/agent based models. Recurring themes in my research are mis- and disinformation, the relationship between beliefs, preferences and decisions and rational reasoning. I’m interested in extending my research towards the epistemology of AI.
Recent Papers
Dependence Falsification in Group Deliberation (2025) [Working Paper, available upon request]
Abstract: With the goal of manipulating beliefs and collective decision-making outcomes, strategic actors often claim that epistemic sources such as scientists or traditional media outlets are problematically dependent, copying falsehoods from each other or a third party. Correspondingly, manipulators present own claims as supported by a number of independent sources to inspire credibility. Both strategies abuse the rational principle that repeated evidence from dependent sources should be rationally discounted, while additional independent evidence can corroborate beliefs. I provide a formal model to analyse this strategy in the context of group deliberation and collective decision-making. The model assumes that individuals have limited information about the relationships between networked sources of evidence and can misrepresent these relationships to the other deliberators. I show how such misrepresentations can alter individual beliefs and preferences for a collective decision. In particular, I prove that there can always be an opportunity and an incentive for individuals to act in this way strategically for all meaningful collective decision-making set-ups. As the likelihood that such manipulation is possible is significant, I provide a game-theoretic analysis that suggests that if a rational individual interested in the truth is faced with a potential misrepresentation of source dependencies, she is best off making her decision entirely independent of the available evidence. This can lead to fact-insensitive reasoning and polarisation. I therefore discuss how to minimise the probability that such manipulation is possible, considering three escape routes that supplement each other. The first one describes how an increase of independent sources while keeping dependent sources fixed can, at the limit, prevent opportunities for misrepresentations. The second discusses how to reduce incentives for manipulation by deliberatively homogenising underlying preferences. The third aims at fostering awareness of dependency relationships between sources.
Group Deliberation in Epistemic Source Networks (2025)
Abstract: Group deliberation is claimed to possess epistemic features that improve collective decision-making. The assumption is that deliberation neutralises negative epistemic effects of overcounting of unequally distributed evidence. The main result of this paper is that this is true only under specific circumstances. For deliberation to avoid overcounting, it is necessary that deliberators are aware of the dependency structures between the pieces of evidence they make use of. This condition is not generally satisfied. Most deliberators become informed in networked epistemic systems, the structures of which are often opaque and untraceable. To study these effects I introduce a model of deliberation that assumes epistemic source networks. Within this model, theoretical and simulation results indicate that while deliberation tends to be epistemically successful, simple majority voting without deliberation more often than not can perform equally well. Additionally, deliberation under source dependencies can distort individual credences. The model suggests that on average, individuals tend to hold credences closer to the evidentially justified credence before group deliberation than after.
Group Deliberation, Informational Cascades and the Reason-Giving Requirement (2025)
Abstract: Political scientists and philosophers often promote group deliberation as epistemically beneficial for collective decision-making. But, theoretical and empirical evidence also suggests that it can polarise and radicalise judgements. These effects are often presented as the result of informational cascades that lead to path-dependency. In reply, proponents of deliberation claim that such evidence and models do not apply to genuinely deliberative communication. If this were true, deliberative communication must possess certain features that immunise it against informational cascades. The mechanisms by which this is supposed to happen remain, however, unclear. This paper offers a systematic analysis of informational cascades over different degrees of deliberative communication to identify this feature. The main contributions are that it can be shown that the problem of informational cascades and resulting path-dependency is just a special case of a more fundamental problem of evidential dependencies. A way out of this problem is to take seriously the reason-giving requirement of deliberation. Thus, a new justification for requiring deliberators to present reasons to each other is developed.