Cooperative behavior cascades in human social networks

Abstract

Theoretical models suggest that social networks influence the evolution of cooperation, but to date there have been few experimental studies. Observational data suggest that a wide variety of behaviors may spread in human social networks, but subjects in such studies can choose to befriend people with similar behaviors, posing difficulty for causal inference. Here, we exploit a seminal set of laboratory experiments that originally showed that voluntary costly punishment can help sustain cooperation. In these experiments, subjects were randomly assigned to a sequence of different groups to play a series of single-shot public goods games with strangers; this feature allowed us to draw networks of interactions to explore how cooperative and uncooperative behaviors spread from person to person to person. We show that, in both an ordinary public goods game and in a public goods game with punishment, focal individuals are influenced by fellow group members' contribution behavior in future interactions with other individuals who were not a party to the initial interaction. Furthermore, this influence persists for multiple periods and spreads up to three degrees of separation (from person to person to person to person). The results suggest that each additional contribution a subject makes to the public good in the first period is tripled over the course of the experiment by other subjects who are directly or indirectly influenced to contribute more as a consequence. These results show experimentally that cooperative behavior cascades in human social networks.behavioral economics | cooperation | public goods | social influence | pay-it-forward

Cooperation and contagion in web-based, networked public goods experiments

“…Combining contributions over many rounds therefore artificially amplifies the differences, leading to the appearance of statistical significance where none may exist. In fact, as Table 5 in [27] itself makes clear, the final (and also average) difference between topologies is roughly the same as the initial difference (period [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20]; thus essentially all of the difference can be explained in term of initial contributions, which are by construction unrelated to the network topology. Second, the significance of the NetworkClustering and NetworkLength coefficients in the PD1 logit model (Table 10 in [27]) is marginal and disappeared when other factors, such as the % cooperation in the previous experiment (PD2) or dummy variables for the session (PD3) were included.…”

“…Specifically, Fowler and Christakis reanalyzed data from Fehr and Gachter [32] (the same results that we replicated in our preliminary experiments described above) in which groups of n~4 players were randomly reassigned to new groups after each round. Whereas in our networks, all individuals appear just once and play with same set of neighbors each turn, in [20] each individual appears r times (where r is the number of rounds of the experiment) and plays with a different set of neighbors each time. As a result, the measure of network distance in [20] does not map precisely to the conventional meaning of network distance, which is the meaning that we have adopted here, but rather refers at least in part to the relation between an individual's present and past states.…”

“…These neighbors may in turn cause their neighbors to cooperate more as well, and so on, leading to a cascade of cooperation that sustains itself over multiple steps. In fact, recently it has been claimed that cooperation is characterized by a ''three degrees of influence'' rule [20], meaning that an individual who increases his or her level of cooperation can positively impact the contribution of an individual who is three steps removed from them in the network. Because the number of individuals who can be reached within three degrees of a cooperating individual will in general depend on the non-local structure of the network [18], the presence of social contagion would imply that network features other than local clustering should also impact aggregate cooperation levels.…”

Clustering by well-being in workplace social networks: Homophily and social contagion.

“…Researchers have examined the social contagion effects of prosociality in experimental economic games, and found that generous allocations of resources could indeed spread from person to person (DeSteno, Bartlett, Baumann, Williams, & Dickens, 2010;Fowler & Christakis, 2010;Gray, Ward, & Norton, 2014). In an economic exchange game, for example, participants who had been helped by another gave more money to a stranger than those who had not been helped-a pay-it-forward effect (DeSteno et al, 2010).…”

“…In an economic exchange game, for example, participants who had been helped by another gave more money to a stranger than those who had not been helped-a pay-it-forward effect (DeSteno et al, 2010). In a multi-round economic game in which participants were constantly changing partners, giving more money to a partner instead of keeping it increased the partner's voluntary donations to others in subsequent rounds (Fowler & Christakis, 2010). Experimental studies that include both givers and receivers can be difficult to design without the use of economic games.…”