Cortical categorization of facial expressions: Threat priority or local configural saliency?

Evolutionary theories propose a "threat superiority effect", whereby the brain prioritizes threat signals like fear and anger via a fast subcortical pathway. It remains unclear, however, whether this prioritization extends to cortical categorization. Using sweep steady-state visual evoked potential electroencephalography, in which images of a specific facial expression were embedded periodically among neutral faces, we measured implicit cortical responses to facial expressions with progressively increasing exposure durations (from 30 Hz to 3 Hz) in two experiments. Results revealed that fear and happiness​ elicited robust, comparable cortical responses, while anger​ was processed less efficiently. In Experiment 1 (when images were stimulated from 12 Hz to 3 Hz in 6 steps) with 15 participants, significant response differences between anger and fear emerged across all stimulation conditions (ps < 0.05, FDR-corrected), except at 4 Hz; notably, significant categorization responses were observed in all conditions except for anger at 12 Hz. In Experiment 2 (30 Hz to 15 Hz; 16 participants), anger-fear response differences persisted at higher temporal frequencies (30 Hz, 24 Hz, 15 Hz; ps < 0.05). There were also significant response differences between anger and happiness, reflected as weaker responses to anger (30 Hz, 24 Hz, 15 Hz; ps < 0.05). Critically, a broadly similar pattern was also observed in a post-recording explicit expression detection task. These findings challenge the notion of a unified threat-processing advantage in the human brain. Instead, they suggest that early cortical categorization is shaped by local configural saliency, favoring high-contrast, open-feature expressions (e.g., fear, happiness) over perceptually complex, closed-feature ones such as anger. These findings have important implications for understanding how visual features guide emotion perception in both typical and atypical human brains.