M.Sc. Chuyang Wu defends his PhD thesis "Towards Machine Understanding of the User: A Study in Interactive Inference of Mental Models" on Friday the 7th of November 2025 at 12 in the University of Helsinki Exactum building, Auditorium B123 (Pietari Kalmin katu 5, 1st floor). His opponent is Assistant Professor Robin Welsch (Aalto University) and custos Professor Antti Honkela (University of Helsinki). The defence will be held in English.
The thesis of Chuyang Wu is a part of research done in the Department of Computer Science at the University of Helsinki and the Faculty of Information Technology at the University of Jyväskylä. His supervisor has been Associate Professor Jussi P. P. Jokinen (University of Jyväskylä).
Towards Machine Understanding of the User: A Study in Interactive Inference of Mental Models
This dissertation investigates how interactive systems can achieve deeper understanding of users by inferring their internal mental models, encompassing goals, expectations, cognitive constraints, and decision-making processes, directly from interaction data. Traditional usability metrics provide limited insights into underlying user cognition, capturing surface-level behaviors but neglecting the motivations and reasoning behind them. To bridge this gap, the research introduces a computational framework combining Partially Observable Markov Decision Processes (POMDP) with the principle of computational rationality. Here, users are modeled as bounded-rational agents, making decisions to optimize rewards under cognitive limitations. User-system interaction is treated as an inverse problem: observed behaviors serve as evidence from which latent cognitive parameters are inferred.
Two methods, Inverse Reinforcement Learning (IRL) and Approximate Bayesian Computation (ABC), are integrated into this framework. IRL uncovers users' reward functions from behavior, while ABC enables tractable inference in complex, real-world scenarios without explicit likelihood calculations. The thesis further explores dynamic and social dimensions, extending the approach to track evolving mental models and to incorporate Theory of Mind for social inference in multi-agent contexts. Validated through theoretical analysis and practical experiments, the framework demonstrates potential for designing adaptive interfaces that truly understand users, continuously aligning with their evolving needs and cognitive states.
Availability of the dissertation
An electronic version of the doctoral dissertation will be available in the University of Helsinki open repository Helda at
Printed copies will be available on request from Chuyan Wu: