Brain Mechanisms engaged in social interactions

The brain computations engaged in social learning have mostly been investigated in dyadic interactions and in small groups. Yet, it is still unclear exactly what are the different types of computational mechanisms used by the brain for making inferences about others in different types of social interactions. First, I will present a taxonomy of distinct computations used by the brain for learning and inferences made during social interactions (reward learning, belief learning, mental-state learning and distributed learning). I will illustrate this taxonomy with examples from experiments performed in our lab. In particular, I will present recent model-based functional MRI results concerning mental-state learning in humans, children and baboons. In human adults, I will show how the brain adapts to fluctuating intentions of others when the nature of the interactions (to cooperate or compete) is not explicitly and truthfully signaled. In Guinea baboons living in a social colony, I will present recent data addressing whether the ability to mentalise is confined to humans. These baboons freely came to play a 2-players coordination game with any other baboon, or alone. Moreover, I will present recent models (RL vs DeGroot) and experimental work shedding light on the computations people use to integrate and transmit information in social networks. Finally, I will present a fMRI study investigating how the brain decides whether to share extra information with others, depending upon one’s own confidence about the reliability of information and upon one’s beliefs concerning the preferences of receivers. Together, these results pave the way towards developing a mechanistic understanding of social learning mechanisms used by networked individuals interacting dynamically in real time, that shape collective performance and consensus.