The Evolution of Other-Regarding Preferences in Networked Populations

We expand upon previous theoretical treatments of cooperation on networks by considering the effects of conditional strategies and continuous strategy space. We model agents as utility-maximizers, with utility functions over their own and their partner’s payoffs, weighted by a parameter denoting otherregard. The other-regard parameter evolves over time through learning from nearest neighbors based on payoffs. In simulation experiments, we find that agents using a multiplicative form of the utility function, which allows conditional behavior, support higher levels of cooperation than those endowed with an additive utility function, which does not allow conditionality. We also find a positive effect on cooperation of the continuous strategy space, though only in the presence of the multiplicative utility function. Our results suggest that comparisons of the evolutionary viability of different utility functions can contribute to economic theories of social preferences.