Dynamic Epistemic Logic Models for Predicting the Cognitive Difficulty of the Deductive Mastermind Game Bonan Zhao Abstract: This thesis studies the cognitive difficulty of the Deductive Mastermind (DMM) game by measuring the complexity of two different logic formalizations of the game. DMM is a version of the board game Mastermind, and it has been implemented in an online educational game system. This system records players’ speed and accuracy data in solving the game, which serves as an empirical indicator of the cognitive difficulty of each DMM game item. In the thesis, we look at an existing formalization of DMM based on analytic tableaux, and we develop a formalization based on dynamic epistemic logic (DEL). The DEL model of DMM performs as well as the tableaux model in predicting the cognitive difficulty of DMM game items, and the DEL model is able to capture more reasoning patterns as self-reported by DMM players. We find that feedback types play an important role in predicting cognitive difficulty of game items, and this result is robust over the two different logic formalizations that we considered.