The evolution of cooperation in situations where selfish behavior would lead to defection is at the root of the formation of human societies and has attracted a lot of attention as a result. In structured populations, both spatial clustering of cooperators in lattice-like topologies, as well as heterogeneous contact networks, have been shown to favor cooperation in social dilemmas. Here, we present a unified framework that can describe and quantify the formation of spatial clusters of cooperators in a metric space, and also represent heterogeneous contact networks, in particular scale-free topologies as observed in most real networks. We find that the system quickly self-organizes into a long-living clustered state and show the importance of this state for the evolution of cooperation in social dilemmas on scale-free networks. We compare the efficiency of the hubs (most connected nodes) and local clusters to drive the system towards cooperation and show that, under certain conditions, hubs lose the ability to control the system and local clusters are more effective in sustaining cooperation. Strikingly, this implies that heterogeneity does not always favor—but can even hinder—cooperation in social dilemmas. The predictions of our model topologies are confirmed by empirical networks.
Supplementary Information: here
Supplementary Datasets: here