Beliefs about human agency influence the neural processing of gaze during joint attention

Abstract

The current study measured adults' P350 and N170 ERPs while they interacted with a character in a virtual reality paradigm. Some participants believed the character was controlled by a human ("avatar" condition, n = 19); others believed it was controlled by a computer program ("agent" condition, n = 19). In each trial, participants initiated joint attention in order to direct the character's gaze toward a target. In 50% of trials, the character gazed toward the target (congruent responses), and in 50% of trials the character gazed to a different location (incongruent response). In the avatar condition, the character's incongruent gaze responses generated significantly larger P350 peaks at centro-parietal sites than congruent gaze responses. In the agent condition, the P350 effect was strikingly absent. Left occipitotemporal N170 responses were significantly smaller in the agent condition compared to the avatar condition for both congruent and incongruent gaze shifts. These data suggest that beliefs about human agency may recruit mechanisms that discriminate the social outcome of a gaze shift after approximately 350 ms, and that these mechanisms may modulate the early perceptual processing of gaze. These findings also suggest that the ecologically valid measurement of social cognition may depend upon paradigms that simulate genuine social interactions.

From automata to animate beings: the scope and limits of attributing socialness to artificial agents

“…These findings are corroborated by a number of other previous and subsequent findings . For instance, a recent study using an established gaze cueing paradigm reported increased activation in bilateral TPJ for gaze following when participants thought the gaze had human origins compared to a preprogrammed, computer‐determined origin …”

Attributed social context and emotional content recruit frontal and limbic brain regions during virtual feedback processing

“…This is also found when avatar-gaze is believed to be controlled by a human rather than when it is believed to be fully robot-controlled (Wykowska, Wiese, Prosser, & Müller, 2014). Likewise, joint attention is initiated and responded to differently when an avatar face in a virtual reality environment is believed to be human-controlled (Caruana, Lissa, & McArthur, 2017a;Caruana, Spirou, & Brock, 2017c;Wykowska, Wiese, Prosser, & Müller, 2014).…”

“…This seems to affect brain responses as well, where human-controlled avatars (i.e. supposedly another human participant controls the avatar's eyes) elicit differential ERP responses, but also lead to different behavioral responses to gaze cues (Caruana, Lissa, et al, 2017a;Caruana, Spirou, et al, 2017c;Wykowska et al, 2014). A study by Singer et al (2004) showed that perceived intentionality activated the bilateral insula.…”

How attention gates social interactions

“…demonstrated reduced gaze following when gaze cues were perceived as being delivered by a robot (e.g., a computer program performing predetermined behaviors) rather than a human agent . These effects are reflected in neural activity, with increased activation of the TPJ when gaze is perceived as human controlled, and reduced magnitudes of the event‐related P1 and N170 components when gaze is perceived as computer controlled.…”