Automatic vs. controlled processes in semantic priming — differentiation by event-related potentials

Abstract

Semantic network models propose that automatic (e.g. spreading activation) and controlled processes are involved in semantic priming. Behavioural studies propose that the influence of each of these processes depends on the stimulus onset asynchrony (SOA). To investigate this hypothesis with a more sensitive method, we applied high-resolution event-related potential (ERP) measures to a word–pseudoword lexical decision task that contained direct, indirect, and non-related prime–target pairs. SOAs consisted of 150 or 700 ms. The results showed that independently of SOA, increasing semantic distance prolonged reaction times and enlarged N400 amplitudes. Furthermore, the word–pseudoword decision evoked a parieto-central late positive complex (LPC respectively delayed P300), which was sensitive for semantic relatedness in the short SOA only. In addition, we found two early frontal components: a P250 in the short SOA only and a N310 sensitive to semantic relatedness more prominent in the short SOA. We conclude that ERP-differences between both SOAs indicate two separate processes: (1) an access to semantic memory, which is facilitated by spreading activation in the short SOA only; and (2) an SOA-independent, controlled process, which integrates prime and target words into a semantic context.