Background

In evolutionary biology, large-scale cooperation among unrelated individuals has long been a puzzle. W.D. Hamilton's inclusive fitness theory provided a clear explanation for altruistic behavior among genetic relatives. The persistent cooperation with others who have low genetic similarity was a puzzle that could not be fully resolved by traditional kin selection theory alone. For decades, attempts have been made to mathematically establish the dynamics that encourage selfish individuals to forgo short-term gains and contribute to public goods.

Previous researchers often proposed direct reciprocity based on repeated interactions or indirect reciprocity relying on reputation as solutions. However, these theoretical models revealed vulnerabilities, such as failing to effectively suppress free-riders when group size increases or information asymmetry intensifies. The problem of secondary free-riding, where punishers themselves are eliminated if punishment costs are not distributed across the group, also remained unresolved. This created an urgent need for an integrated framework to bridge the gap between theoretical mathematical models and large-scale decision-making experimental data from actual human groups.

Key Findings

Researchers constructed an integrated dynamics equation within evolutionary game theory that combines multi-level selection pressure with variable network dynamics. To overcome the limitations of existing fixed-network-based Public Goods Games (PGG), we designed a model in which players consider not only their opponents' past contributions but also the costs of network connectivity. In essence, they established an adaptive network model where the interaction structure itself evolves.

Mathematical simulation results showed that when the cost of disconnection exceeds a specific threshold (c* = 0.28), a stable equilibrium point emerges where free-riders are pushed to the periphery of the network, and the proportion of cooperators converges to over 82%. To verify the theory, 24 sessions of an online behavioral economics experiment were conducted with 1,440 non-kin participants randomly assigned. Participants determined their contribution levels and reselected interaction partners in each round.

The data analysis results aligned remarkably well with the model's predictions. While the cooperation maintenance rate remained at approximately 41% when a simple punishment system was applied, the introduction of a selective connection blocking protocol linked to reputation scores caused the average cooperation rate to soar to 84.6%. This result proves that a large-scale cooperative order can be stabilized through individual participants' partner selection rights alone, without the need for external enforcement agencies.

Significance and Outlook

This discovery holds academic value in that it quantifies the evolutionary link that allowed human society to transcend genetic limitations and form a hyper-cooperative society. It revealed that topological changes in complex networks, rather than remaining within kin-centered evolutionary models, are the core engine for preserving altruism. This provides guidelines for establishing various digital interaction rules, such as malicious user blocking algorithms in social media platforms or collaboration designs in Decentralized Autonomous Organizations (DAOs).

The tasks are also clearly defined. Unlike controlled laboratory environments, real human societies are characterized by information distortion, biased rumors, and severe resource asymmetry. It is difficult to rule out the risk of distortion in which the weak are excluded when asymmetric power structures intervene. The researchers stated that their follow-up study will investigate the impact of complex cognitive biases and information incompleteness on network dynamics.