A Cognitive Modeling Approach to Surprise
Master's thesis of Stefan Lindner 
The goal of this thesis is to explore the nature and structure of surprise. To this end we created a surprise-inducing reaction time experiment and also devised three computational model of surprise that were specified and implemented in the cognitive architecture of ACT-R. In two of these models we introduce the new idea of expectation chunks that could help to better model the reactions to uncertain and surprising events. The predicted reaction times of the three models are compared to the actual data of the reaction time experiment. RESULTS: The simple activation model best captured absolute reaction times while the model containing a sole schema-update mechanism best captured the relative trend. The full-blown surprise-model on the basis of Meier et al.’s (1997) structural model of surprise proved to be inadequate to predict reaction times.