A series of behavioral and computational experiments examine mechanisms by which cooperation is promoted in networked communities. Social groups are typically built around some form of network structure, and cooperation and efficiency within a group can be strongly affected by the flow of information determined by the structure of the network alone. Furthermore, experimental and observational studies show some real world groups are more successful at sustaining cooperation than others for a given network structure. This research explores the hypothesis that such heterogeneity is due to the interaction between the social preferences of individuals and the network structure that determines the dynamics of cooperation within a social group. Results from experimental economics have shown that there is a great deal of heterogeneity in terms of an individual?s preference to cooperate in group-dilemma settings. Depending on the composition of these cooperative types in the group, socially optimal cooperation may or may not be achieved without costly sanctions. This suggests that, in addition to the network structure, the distribution and the placement of different types within the network can also affect group outcomes. While it has been shown that network structure can facilitate cooperation, this proposal investigates the impact of the distribution of cooperative types and their placement within a given network structure on social outcomes.
This project is an unusual use of both behavioral experiments and the computational simulation methodology known as agent-based modeling. This approach will allow us to systematically investigate the effects of network structure on cooperative behavior in a heterogeneous population. While observational data exists on naturally-occurring networks, the network structure and behavior of the individuals participating in the network have typically co-evolved. Thus, identification of the effect of structure on cooperation would be biased with this data because of individuals choosing the network in which to interact. The strength of this research is to use laboratory experiments to identify the cooperative preferences of individuals and then strategically place them in various network structures to examine how these types interact to enhance cooperation. The research can create counterfactuals of existing networks to examine how individuals with different levels of cooperative preferences would behave in a network they did not choose. The dialogue between the experiments and agent-based modeling will allow scholars to more efficiently investigate the network structures that have the potential to most enhance cooperation and to learn and calibrate the models to better represent behavior in social dilemmas.
