The Dynamics of Two Cognitive Heuristics for Coordination on Networks

This paper is about how groups solve global coordination problems such as the distributed graph coloring problem. We focused on scenarios in which agents are not able to communicate explicitly, but can rely on observing the momentary choices of their immediate neighbors in a social network. It has been reported that humans use two cognitive heuristics when solving such problems: (i) the frequency-based heuristic, where people make choices that minimize conflicts with neighbors; and (ii) the degree-based heuristic, where people avoid conflict with well-connected neighbors. In this paper, we present a model capable of capturing these cognitive heuristics to varying intensities. Then, through simulation, we shed light on the behavior of these heuristics under different classes of social networks. Our analysis generally speaks in favor of both heuristics, provided they are used in moderation, and illustrates the utility of taking social status (connectivity) into account.