Schedule of Events | Symposia

Time-Course Differences in the Processing of Taxonomic and Thematic Semantic Relations Revealed by EEG Spatiotemporal Cluster Analysis

Poster Session C - Sunday, March 30, 2025, 5:00 – 7:00 pm EDT, Back Bay Ballroom/Republic Ballroom

Stephen J. Ball¹ (stephen.ball-4@postgrad.manchester.ac.uk), Jennifer C. Thompson^1,2, Jason R. Taylor¹; ¹University of Manchester, ²Northern Care Alliance NHS Foundation Trust

Taxonomic relations (items related through shared features, e.g., WOLF-DOG) and thematic relations (items related through context, e.g., KENNEL-DOG) are critical components of semantic knowledge, and thus critical to understanding the world around us. Behavioural studies have reported that thematic relations may be processed more rapidly than taxonomic, yet electrophysiological evidence to support this is inconsistent. To elucidate the neural dynamics underlying semantic relationship processing, we conducted an EEG study using threshold-free spatiotemporal cluster analyses of event-related potential (ERP) and Time-Frequency (TF) data during a semantic relatedness task. 29 healthy young adult participants completed a semantic similarity judgement task on 560 cue-target word pairs while EEG was recorded (64-channel BioSemi ActiveTwo). Cue-target relationships were manipulated between Unrelated/Taxonomic/Thematic conditions. Surface Laplacian-transformed ERP analysis revealed a significant increase in current source density (CSD) in thematically cued targets across left frontal electrodes in an early time window (158-681ms) and left posterior electrodes later in the epoch (455-1000ms). In TF, increases in theta (4-7Hz) power were associated with thematic processing in two temporal clusters across frontocentral channels at 190-280ms and 570-1000ms. By using a robust and unbiased statistical approach to determine significance, we have identified EEG effects which suggest that taxonomic and thematic processing use dissociable neurological systems, supporting existing behavioural and electrophysiological studies. Finally, we discuss the implications of these findings for existing neurocomputational models of semantic representation and control.

Topic Area: LONG-TERM MEMORY: Semantic