ERP correlates of associative priming and subsequent memory effects: Evidence from sentence-embedded paired-associate learning.

The aim of this study was to examine the fine-grained interplay of episodic and associative-semantic long-term knowledge in a sentence-embedded paired-associate paradigm and to probe separable contributions to encoding-related event-related potential (ERP) amplitudes. Participants were instructed to learn sentences of the structure: article, subject-prime, auxiliary verb, closed-class word, and target-object (e.g., "the car has an engine"). Later, they had to recall the target-object, given the preceding sentence context. Words were displayed using rapid serial visual presentation. Corpus-based co-occurrence statistics were used to orthogonally manipulate direct association (strong vs. none) and semantic feature overlap (high vs. low) between the prime and the target word (cf. Hofmann et al., 2018). Strong direct associations led to more correct responses, whereas semantic feature overlap showed no reliable main effect and no interaction was observed. However, descriptive trends suggested a potential supportive contribution in the presence of direct associations. This aligns with the largely additive pattern of associative and semantic influences reported for lexical processing (Rölke et al., 2018). Response times were faster for strong direct associations, with no effect for semantic feature overlap. ERPs were recorded during target-object encoding, with condition-related analyses restricted to correctly recalled trials. Further, ERPs were contrasted between subsequently correct vs. incorrect recall to quantify subsequent memory effects. Strong direct associations elicited a greater Late Positive Component (LPC; 460-800 ms) with an occipito-parietal scalp distribution, whereas semantic feature overlap showed no LPC modulation. Correctly recalled targets showed more positive-going ERP amplitudes, reflected in an attenuated N400 (250-500 ms) and an enhanced P600 (500-800 ms), both with a centro-parietal scalp distribution and regardless of associative-semantic priming manipulations. This study demonstrates that classic priming patterns from lexical processing (Rölke et al., 2018) can extend to paired-associate learning outcomes and further identifies dissociable ERP signatures during encoding that track associative priming effects (LPC).