Temporal coding of conceptual distances in the human language area

Abstract

It is largely unknown how the brain encodes semantic distances between conceptual (verbal) knowledge. I presently show that the information on semantic distances is mainly embedded in a temporal measure (oscillation frequency) of human EEG waveforms. Participants judged a semantic relation between two words (synonym, antonym, thematically-related, taxonomically-related, or unrelated) sequentially presented. Brain rhythms (8-30 Hz) over the left language region, especially the anterior temporal lobe (ATL), were accelerated when a word pair with a high conceptual similarity (e.g. synonym) was presented. Importantly, this change in brain rhythm reflected an inter-word closeness as concepts (holistic similarity, typically seen in synonyms or thematically-related pairs), rather than an inter-word overlap of semantic features (partial similarity, typically seen in antonyms or taxonomically-related pairs). These results highlight a role of temporal measures of neural activity in semantic processing, which explains an inconsistency over previous results using hemodynamic measures. Significance statementVerbal concepts are connected with each other in the brain, although neural (electric) signals encoding their semantic distances remain unknown. In the present study, I show that the information about inter-conceptual similarity is embedded in speed (not amplitude) of neural oscillatory signals in the healthy human brain. This approach using a temporal measure further revealed the concept-based (not feature-based) processing of word meanings unique to the left language areas.