Background <p>Semantic memory includes knowledge of specific categories and their diverse semantic relations, yet the neural mechanisms underlying thematic and taxonomic semantic processing remain controversial.</p> Methods <p>We explored the neural correlates of thematic and taxonomic processing via ERP in a lexical decision task. In this task, brand names served as primes, and the extension products served as targets. Five experimental conditions were compared: the core–product condition (e.g., Boeing–airliner), taxonomic condition (Boeing–helicopter), thematic condition (Boeing–suitcase), unrelated condition (Boeing–shirt), and the nonword condition (Boeing–dola) for response balance.</p> Results <p>Behavioral results showed significant priming effects (shorter RTs) for core-product, taxonomic, and thematic conditions compared to the unrelated condition. However, N400 analysis revealed that only core-product and thematic conditions showed semantic facilitation (smaller N400), while taxonomic conditions did not differ from unrelated conditions. Critically, frontal negativity (400–550 ms) was smaller for core-product and thematic conditions compared to taxonomic and unrelated conditions.</p> Conclusions <p>These findings indicate that thematic relations are more readily retrieved and contextually integrated, whereas taxonomic relations rely on feature-based categorization. The results support the dual-process model of semantic memory and are consistent with the principle of taxonomic primacy with thematic embedding, in which taxonomic structure provides the primary organizational axis while thematic relations are nested within broader contextual networks.</p>