EMPATHY ACROSS AGENTS: NEURAL AND BEHAVIORAL MODULATION OF AFFECTIVE INCONGRUENCE PROCESSING IN HUMANS AND A HUMANOID ROBOT

Humans rapidly infer others’ emotional states and update these inferences when confronted with pragmatic emotional incongruence. Individual differences in cognitive empathy are known to modulate such processes, yet it remains unclear whether these mechanisms extend to artificial agents. The present study examined the behavioral and neural correlates of affective incongruence processing in human and humanoid robot facial expressions using an affective priming paradigm and event-related potentials (ERPs). Participants judged whether emotional facial expressions were congruent with preceding emotionally evocative sentences while electroencephalograms (EEG) were recorded. Behavioral measures showed significant effects of congruence and face type, with trends toward slower and less accurate responses in high-empathizing individuals. At the neural level, affective incongruence modulated both the N400 and late positive potential (LPP). Cognitive empathy selectively interacted with congruence in the LPP, indicating that empathy influences later evaluative and integrative stages of emotional processing while early semantic integration remained unaffected. Importantly, empathy-related effects did not differ between human and robot faces, suggesting that the empathic mechanisms underlying emotional incongruence generalize to artificial agents conveying emotional cues.