Proceedings of the Annual Meeting of the Cognitive Science Society

Learning constitutes an essential part of human experienceover the life course. Independent of the domain, it ischaracterized by changes in performance. But what cognitivemechanisms are responsible for these changes and how dosituational features affect the dynamics? To inspect that inmore detail, this paper introduces a cognitive modelingapproach that investigates performance-related changes inlearning situations. It leverages the cognitive architectureACT-R to model learner behavior in an interrupted learningtask in two conditions of task complexity. Comparisons withthe original human dataset indicate a good fit in terms of bothaccuracy and reaction times. Although interruption effects aremore obvious in the human data, they are prevalent as well inthe model. Furthermore, the model can map the learningeffects, particularly in the easy task condition. Based on theexisting mapping of ACT-R module activity with fMRI data,simulated neural activity is computed to investigate underlyingcognitive mechanisms in more detail. The resulting evidenceconnects learning and interruption effects in both taskconditions with activation-related patterns to explain changesin performance.