The dynamics underlying the representation of observed actions at different taxonomic levels

Previous studies examined the representation of actions across different hierarchical levels in the occipitotemporal and parietal cortex, yet the dynamics underlying this process remain unclear. In this study, we aimed to reveal the latencies associated with the representation of observed actions at different taxonomic levels using time-resolved representational similarity analysis (RSA) of EEG data and EEG-fMRI fusion. In our experimental setup, participants were presented with static images depicting human actions. Each image belonged to one out of three superordinate level categories (e.g. ‘locomotion’), one out of six basic level categories (e.g. ‘swimming’) and one out of twelve subordinate level actions (e.g. ‘swimming backstroke’). We established Representational Dissimilarity Models to capture pairwise differences between actions, separately for the three taxonomic levels. Results of the time-resolved RSA showed consistently highest peak latencies around 170 ms for all three taxonomic level models. The EEG-fMRI fusion provided further insights, indicating that action representations measured via fMRI in the lateral occipitotemporal cortex (LOTC) exhibited the highest similarity with EEG data around 230 ms. In summary, our findings suggest that actions across three taxonomic levels occur simultaneously, with the LOTC playing a crucial role in elucidating the spatiotemporal aspects of hierarchical organization in action representations. This research contributes to a deeper comprehension of the neural processes underlying the recognition and classification of actions in the human brain.