Strong links promote the emergence of cooperative elites

Abstract

The maintenance of cooperative behavior is fundamental for the prosperity of human societies. Empirical studies show that high cooperation is frequently associated with the presence of strong social ties, but they are silent on whether a causal mechanism exists, how it operates, and what features of the social environment are conducive to its emergence. Here we show experimentally that strong ties increase cooperation and welfare by enabling the emergence of a close-knit and strongly bound cooperative elite. Crucially, this cooperative elite is more prevalent in social environments characterized by a large payoff difference between weak and strong ties, and no gradation in the process of strengthening a tie. These features allow cooperative individuals to adopt an all or nothing strategy to tie strengthening based on the well-known mechanism of direct reciprocity: participants become very selective by forming strong ties only with other cooperative individuals and severing ties with everyone else. Once formed, these strong ties are persistent and enhance cooperation. A dichotomous society emerges with cooperators prospering in a close-knit, strongly bound elite, and defectors earning low payoffs in a weakly connected periphery. Methodologically, our set-up provides a framework to investigate the role of the strength of ties in an experimental setting.

Free neighborhood choice boosts socially optimal outcomes in stag-hunt coordination problem

“…First, many of these studies do not allow for free neighborhood choice as we do but enforce dynamic interaction structures exogenously 4 , 31 , 32 . Second, most of these studies implement, implicitly or explicitly, an infinitely repeated game by either not telling subjects the number of total rounds or implementing a random ending 4 , 31 , 32 , 34 , 35 , 37 , which does not allow to investigate whether dynamic interaction structures also affect behavior when subjects know the end of the game. In fact, the social dilemma study closest to our research that implements a known horizon, finds that in most cases cooperation breaks down completely already well before the game ends 36 .…”

“…Second, the reported effects of fluid interaction structures in social dilemma games are not uniformly and strongly positive. Most studies report mildly positive effects of dynamic interaction structures on cooperation that are, however, far off the fully efficient outcome 31 – 35 or—with one exception—not maintained in the long run 36 , 37 . Thus, direct evidence is needed on whether free choice of the interaction neighborhood helps individuals to coordinate fully and robustly on an efficient outcome in coordination problems.…”

Cooperation in an Uncertain and Dynamic World

“…Previous experimental studies have shown homophily based on cooperativeness, i.e. cooperators tending to have more connections with other cooperators, increases when there is endogenous network formation (Wang et al, 2012), participants have complete information about the network structure (Gallo and Yan, 2015), and there is an opportunity to strengthen a connection (Gallo et al, 2019).…”

“…Previous studies (e.g. Gallo and Yan, 2015;Gallo et al, 2019) have shown that reputational information and a fastchanging social environment increase cooperation by enabling the formation of clusters of cooperators who associate with each other. We find that reputational uncertainty undermines this process in a fast-changing social environment because defectors are more likely to remain embedded in cooperators' neighborhoods thereby creating an obstacle to the formation of the cooperative clusters that would have driven up the cooperation level and average welfare of the whole group.…”

Free Neighborhood Choice Boosts Socially Optimal Outcomes in Stag-Hunt Coordination Problem

“…Second, the reported effects of fluid interaction structures in social dilemma games are not uniformly and strongly positive. Most studies report mildly positive effects of dynamic interaction structures on cooperation that are, however, far off the fully efficient outcome 31,32,33,34,35 or-with one exception-not maintained in the long run 36,37 . Thus, direct evidence is needed on whether free choice of the interaction neighborhood helps individuals to coordinate fully and robustly on an efficient outcome in coordination problems.…”