Detection of human agents in interaction: early ERPs evidences

In order to interact effectively, we need to swiftly detect other human beings and recognize them as potential interagents. According to direct access theories, the first steps of visual information processing offer us an informed direct grasp of the situation. Biological system detection may be the gateway of such smart processes and then may influence initial stages of perception fostering adaptive social behaviour. Neural markers of human features have been found but classic paradigms used contextual individual stimuli. To overcome that limitation and investigate early neural correlates of human agents detection in ecological situations, we compared scenes showing a human vs. artificial agent interacting with a human interagent. 20 volunteers participated to the study. Participants were asked to observe dynamic visual stimuli showing realistic interactions. EEG was recorded and then ERP for selected ROI were computed. stimulus depicts an arm executing a gesture addressed to a human interagent. Visual features of the arm were manipulated: in half of trials it was real; in other trials it was deprived of some details and transformed in a statue-like arm. The dynamic stimuli was composed by two frames presented in succession (SOA = 200 ms) and ERP has been time locked to the second frame in order to explore the specific processing phase of intention attribution. Morphological analysis revealed an early negative deflection peaking at about 155 ms and two positive deflections peaking respectively at 250 and 350 ms. Peak amplitude data have been statistically analysed by repeated measures ANOVA. No statistically significant differences due to Agent (human vs. artificial) have been revealed by the analyses applied to the positive peaks. The analysis of the N150 peak amplitude, instead, showed a significant simple effect for Agent and significant interaction effects (Agent x Side and Agent x Region x Side). The peak was ampler in the left anterior fronto-temporal region, in particular when the gesturing arm was human. Two negative deflections have been associated to perception of human features: N170 and N190. They are respectively sensitive to faces and body parts and are ampler in parieto-occipital regions. The different scalp localization and the earlier time frame of the negative component we found suggest that it may be a different potential and may have a different source. The early negative deflection, N150, that we found to be different between the human and artificial conditions, may not depend on luminance or contrast values, it being presumably associated to human agency detection. In fact, the cortical localization of this peak, that is more left fronto-temporally distributed, suggests the hypothesis of an early agent-related categoryeffect mediated by cortical anterior structures.