How cognitive control is represented in the brain: An EEG Representational Similarity Analysis study

Aims: Cognitive control is a fundamental human ability that allows pursuing specific relevant goals. Several theories postulate that cognitive control relies on neural representations. However, univariate approaches that have been classically used to investigate it are not suitable for studying such representations. Therefore, the aim of our EEG study was to explore how cognitive control representations are encoded at the neural level and to investigate the contribute of different theory-based representations. Methods: To this aim, we used Representational Similarity Analysis (RSA) to model theory-based representations and correlate them to the observed brain patterns. We designed a spatial Stroop task, in which both list-wide and item-specific proportion of congruency were manipulated, respectively, to measure the effects of proactive and reactive control on Stroop interference resolution. We assessed the similarity between control-related representational models and temporal and spatial multivariate patterns of EEG activity, while controlling for low-level confounding effects. Results: RSA revealed specific spatiotemporal EEG correlates not only of low-level sensorial, motor, and cognitive representations, but also of both proactive and reactive control representations. Specifically, significant similarities were found between proactive control representational models and pre-stimulus multivariate patterns of EEG activity, as well as between reactive control representational models and post-stimulus multivariate patterns of EEG activity, in line with theoretical accounts of Stroop interference resolution. Conclusions: Our results suggest that RSA better informs cognitive control theory by revealing the dynamics of its neural representations. Indeed, temporal and spatial RSA patterns provided insights into cognitive control theory-based representations but also into low-level representations.