Evolution of cooperation in multichannel games on multiplex networks

Humans navigate diverse social relationships and concurrently interact across multiple social contexts. An individual’s behavior in one context can influence behavior in other contexts. Different payoffs associated with interactions in the different domains have motivated recent studies of the evolution of cooperation through the analysis of multichannel games where each individual is simultaneously engaged in multiple repeated games. However, previous investigations have ignored the potential role of network structure in each domain and the effect of playing against distinct interacting partners in different domains. Multiplex networks provide a useful framework to represent social interactions between the same set of agents across different social contexts. We investigate the role of multiplex network structure and strategy linking in multichannel games on the spread of cooperative behavior in all layers of the multiplex. We find that multiplex structure along with strategy linking enhances the cooperation rate in all layers of the multiplex compared to a well-mixed population in Prisoners’ Dilemma games, provided the network structure is identical across layers. The effectiveness of strategy linking in enhancing cooperation depends on the degree of similarity of the network structure across the layers and perception errors due to imperfect memory. Higher cooperation rates are achieved when the degree of structural overlap of the different layers is sufficiently large, and the probability of perception error is relatively low. Our work reveals how the social network structure in different layers of a multiplex can affect the spread of cooperation by limiting the ability of individuals to link strategies across different social domains.Author summary: People hold simultaneous membership in different social networks where they are engaged in interactions with neighbors, some of whom may be common across all networks while others may be unique to a specific social network. The distinct nature of interactions in different social networks can be modeled as games with distinct benefits. How does the nature of the interactions in different social networks as well as the underlying structure of each network affect cooperation levels in the population? Can linking behaviors across different social networks aid in enhancing cooperation levels across all such networks? We use the framework of repeated games on multiplex network-structured populations to address these questions. We find that linking behaviors can be favorable for increasing cooperation levels in all social networks if there is significant structural overlap between the different networks and perception errors in recalling past actions of interacting neighbors, are low.