Cooperation and Contagion in Web-Based, Networked Public Goods Experiments

A longstanding idea in the literature on human cooperation is that cooperation should be reinforced when conditional cooperators are more likely to interact. In the context of social networks, this idea implies that cooperation should fare better in highly clustered networks such as cliques than in networks with low clustering such as random networks. To test this hypothesis, we conducted a series of web-based experiments, in which 24 individuals played a local public goods game arranged on one of five network topologies that varied between disconnected cliques and a random regular graph. In contrast with previous theoretical work, we found that network topology had no significant effect on average contributions. This result implies either that individuals are not conditional cooperators, or else that cooperation does not benefit from positive reinforcement between connected neighbors. We then tested both of these possibilities in two subsequent series of experiments in which artificial seed players were introduced, making either full or zero contributions. First, we found that although players did generally behave like conditional cooperators, they were as likely to decrease their contributions in response to low contributing neighbors as they were to increase their contributions in response to high contributing neighbors. Second, we found that positive effects of cooperation were contagious only to direct neighbors in the network. In total we report on 113 human subjects experiments, highlighting the speed, flexibility, and cost-effectiveness of web-based experiments over those conducted in physical labs.

💡 Research Summary

The paper investigates whether network structure influences cooperative behavior in a local public‑goods game, testing the long‑standing hypothesis that conditional cooperators will reinforce each other more strongly in highly clustered networks (e.g., cliques) than in sparsely clustered ones (e.g., random regular graphs). Using a web‑based platform, the authors recruited 24 participants per session and assigned them to one of five network topologies: three disconnected cliques of eight nodes each, and two random regular graphs where each node has exactly three neighbors. In each round, participants received ten points and chose how many to contribute to a public pool; the total contribution was then equally redistributed among all players, making the game a classic linear public‑goods dilemma.

The first series of experiments (36 sessions) revealed no statistically significant differences in average contributions across the five topologies. This null effect suggests either that participants are not conditional cooperators, or that any conditional cooperation that does exist does not benefit from the positive reinforcement that clustered neighborhoods could provide.

To disentangle these possibilities, the authors conducted two follow‑up series in which “seed” players were introduced. In one condition, seeds contributed the maximum (10 points) each round; in the other, seeds contributed nothing. The remaining participants were naïve and could observe the behavior of their immediate neighbors. The results showed a symmetric conditional response: participants increased their contributions when surrounded by high‑contributing neighbors, but they also decreased contributions when surrounded by low‑contributing neighbors. Importantly, this influence was confined to direct neighbors; the behavior of second‑degree neighbors did not affect a participant’s contribution. Thus, cooperative “contagion” was short‑ranged and did not propagate through the network.

Overall, the study reports data from 113 human‑subject experiments, highlighting the speed, flexibility, and cost‑effectiveness of online experiments compared with traditional laboratory work. The findings challenge theoretical models that predict robust cooperation in clustered networks and suggest that real‑world cooperation may rely more on local, bidirectional adjustments rather than on the amplification of positive behavior through network structure. The authors acknowledge limitations such as the relatively simple network designs, the short horizon of the repeated game, and potential differences in motivation between online and lab participants. They propose future work that incorporates more complex topologies (e.g., scale‑free or small‑world networks), longer interaction horizons, and dynamic network formation to better capture the mechanisms underlying conditional cooperation and its diffusion.