Decoding and Characterizing the Intracranial Representation of Semantic Information

High-gamma power from intracranial sEEG during semantic language tasks enables decoding of semantic category membership across 15 categories, reaching 29.8% mean accuracy (chance = 6.7%).

• Why it matters: It addresses whether higher-level semantic representations, not just motor or articulatory speech signals, are decodable from human cortical activity for both neuroscience and future language BCIs.
• What they did: Patients undergoing stereotactic electroencephalography for epilepsy monitoring performed semantic processing language tasks; high-gamma power from local field potentials fed supervised machine-learning classification evaluated with cross-validation.
• The result: Semantic category information was decoded significantly above chance, showing that conceptual category membership is extractable on individual trials and supporting semantic decoding for future language BCIs.

AI summary of the abstract · may contain errors · not medical advice · read the paper.