Perceived facial happiness during conversation correlates with insular and hypothalamus activity for humans, not robots. An investigation of the somatic marker hypothesis applied to social interactions.

The somatic marker hypothesis posits that perceiving emotions entails reenacting markers of self emotions, in particular in the autonomous nervous system. Well studied in decision-making tasks, it has not been tested in a social cognitive neuroscience framework, and in particular for the automatic processing of positive emotions during natural interactions. Here, we address this question using a unique corpus of brain activity recorded during unconstrained conversations between participants and a human or a humanoid robot. fMRI recordings are used to test whether activity in the most important brain regions in relation to the autonomic system, the amygdala, hypothalamus and insula, is affected by the level of happiness expressed by the human and robot agents. Results indicate that for the hypothalamus and the insula, in particular the anterior agranular region strongly involved in processing social emotions, activity in the right hemisphere increases with the level of happiness expressed by the human, but not the robot. Results indicate that perceiving positive emotions in social interactions induces the local brain responses predicted by the somatic marker hypothesis, but only when the interacting agent is a real human.