Functional specialization and dynamical interaction in human amygdala subregions support fearful-expression recognition

Fearful-expression recognition is critical for adaptive responses to potential threats and relies on both rapid threat detection and fine-grained face encoding. Yet how human amygdala subregions differentially support these distinct cognitive components remains unclear. Here, we recorded intracranial EEG from lateral and medial amygdala in epilepsy patients performing an emotional face-matching task and combined multivariate decoding, time-frequency and directed-connectivity analyses with intracranial stimulation. The lateral amygdala exhibited early fear-specific responses, characterized by higher decoding accuracy and increased theta/alpha-band (4-12 Hz) power, and transmitted this fear-related information to the medial amygdala, which showed delayed and sustained activation. By contrast, the medial amygdala encoded face-specific information at later stages in the 2-16 Hz band with superior decoding accuracy and then relayed this information back to the lateral amygdala. Intracranial stimulation produced a double dissociation in behavior, with lateral amygdala stimulation disrupting fear detection, whereas medial amygdala stimulation selectively accelerated neutral-face recognition. Together, these findings reveal a temporal hierarchy in the human amygdala, whereby dynamic bidirectional interactions between subregions implement distinct components of fearful-expression recognition, providing a circuit-level framework for understanding social threat processing.