Relationships between classifier-quantified priming effects in ERPs and face cognition abilities: Contributions of...

Previous research has consistently shown that individual differences in face cognition abilities correlate with repetition priming-induced amplitude changes in event-related potentials, known as the early repetition effect (or N250r). However, the association with subsequent priming effects (e.g., N400) remains unclear, although this is crucial for understanding the cognitive significance of these different effects. This gap in knowledge may be due to factors such as different paradigms or latency variability. In our recently published classifier-based analysis, we described the impact of latency variability across trials, conditions, and participants on priming effects. Building on these findings, the present analysis used the classification performance of deep neural networks for each participant as an indicator in structural equation models to explore the relationships between priming effects and face cognition abilities. We investigated how these relationships were affected by task difficulty and latency variability. Through our RIDE-based stepwise latency correction method, we found a substantial association between the N250r and face cognition speed, while the N400 was more closely associated with face memory accuracy. Notably, these relationships were significantly stronger in difficult than in easy ERP tasks. Correction for latency shifts between primed and unprimed conditions eliminated the associations between ERP amplitudes and face cognition abilities, indicating that latency shift is a major factor driving brain-behavior relationships. Our results suggest that classifier-quantified priming effects provide an advanced and useful measure for modeling brain-behavior relationships in face cognition.