A neurocognitive interactive activation model of semantic priming in lexical decisions

This study introduces the sequential read-out model (SROM) to investigate the strategic and decision-making contributions to the semantic priming effect in a lexical-decision task (LDT). We use behavioral and fMRI data from two experiments (n = 32 and 31), which factorially manipulated the association strength, semantic similarity, and stimulus-onset-asynchrony. Using a leaky competing accumulator layer for lexical decisions, the SROM successfully accounted for behavioral data and showed that switching from short to long stimulus-onset-asynchrony changed the decision-making mode from competitive to independent race regime. We then used the individually estimated SROM parameters to predict the interindividual differences in BOLD responses to targets. We found that lexical-orthographic evidence modulated BOLD amplitudes in the left lingual gyrus, suggesting that it maintains lexico-orthographic evidence in working memory, which is then used for lexical decisions. The decision threshold, inhibition parameters and top-down semantic excitation predicted activation differences in the left inferior frontal gyrus, suggesting it regulates the decision process. Decision noise and top-down semantic excitation were associated with the left angular gyrus, supporting its involvement in evidence accumulation. Overall, the SROM provided mechanistic interpretations of brain activations in the regions involved in the semantically primed LDT, while accounting for associative semantic and strategic factors.