Individuals in a society do not interact randomly, but rather preferentially interact with a small local subset of the whole population, i.e. their friends and neighbors. Such a picture naturally translates into a network description. Social networks, i.e. the set of relationships in a social group, influence and constrain, often in rather non-trivial ways, the choices of individuals: interaction can change behavior, which in turn can promote proximity, allowing for further interaction. Statistical Physics can prove to be a rather effective tool to describe the interplay between interactions and topology: once the details of the social interactions taking place on the network have been specified, one can often categorize the different states of a given stylized society as the different phases of the corresponding statistical model.

In this talk, I will present two examples of this line of reasoning, one pertaining to the spectacular reduction in social mobility emerging in societies whose individuals highly value social status, the other one related to social learning, i.e. a society’s ability to properly aggregate the noisy information spread across its individuals in order to make the right decisions on relevant matters.