Social norm dynamics in network-structured populations

Slimane Dridi

March 12, 2019, 12:45–13:45


Room MS001

IAST Lunch Seminar


Social norms govern many aspects of human societies. The adoption of a language to communicate, the acceptance of technological innovations, the rules of driving can all be seen as social norms. In the current era of globalisation, some argue that local norms and customs are threatened by a growing dominant culture that spreads over the internet. Hence, an important contemporary question is that of the dynamics of norm acceptance and how population structure affects the diffusion of such norms. Previous theoretical work considered these types of dynamics but failed to consider how learning with long memory at the individual level translates into global population dynamics. I analyse fictitious play in coordination games that take place in a population of fixed size. Every agent in the population uses simple assumptions about how others behave. Each agent then learns how others behave and tries to play a best response to her learned estimation. If the population has no social structure, sufficiently high levels of noise in decision-making lead to an inefficient situation where every player chooses a mixed strategy that puts greater weight on the risk-dominant norm. For lower levels of noise a perturbed Nash equilibrium is reached in the population. In coordination games played on a network, I study the conditions leading to agreement within a subset of players. In the presence of norm influencers who do not learn, I show that a risk-dominant norm spreads more easily than a payoff-dominant one. These results and predictions suggest that it is useful to model how the cognition of agents affect global population dynamics in order to understand cultural diversity in an interconnected world.