Cognitive processing stages in mental rotation – How can cognitive modelling inform HsMM-EEG models?

Abstract

The aspiration for insight into human cognitive processing has traditionally driven research in cognitive science. With methods such as the Hidden semi-Markov Model-Electroencephalography (HsMM-EEG) method, new approaches have been developed that help to understand the temporal structure of cognition by identifying temporally discrete processing stages. However, it remains challenging to assign concrete functional contributions by specific processing stages to the overall cognitive process. In this paper, we address this challenge by linking HsMM-EEG³ with cognitive modelling, with the aim of further validating the HsMM-EEG³ method and demonstrating the potential of cognitive models to facilitate functional interpretation of processing stages. For this purpose, we applied HsMM-EEG³ to data from a mental rotation task and developed an ACT-R cognitive model that is able to closely replicate human performance in this task. Applying HsMM-EEG³ to the mental rotation experiment data revealed a strong likelihood for 6 distinct stages of cognitive processing during trials, with an additional stage for non-rotated conditions. The cognitive model predicted intra-trial mental activity patterns that project well onto the processing stages, while explaining the additional stage as a marker of non-spatial shortcut use. Thereby, this combined methodology provided substantially more information than either method by itself and suggests conclusions for cognitive processing in general.