Collective action demands that individuals efficiently coordinate how much, where, and when to cooperate. Laboratory experiments have extensively explored the first part of this process, demonstrating that a variety of social-cognitive mechanisms influence how much individuals choose to invest in group efforts. However, experimental research has been unable to shed light on how social cognitive mechanisms contribute to the where and when of collective action. We build and test a computational model of human behavior in Clean Up, a social dilemma task popular in multi-agent reinforcement learning research. We show that human groups effectively cooperate in Clean Up when they can identify group members and track reputations over time, but fail to organize under conditions of anonymity. A multi-agent reinforcement learning model of reputation demonstrates the same difference in cooperation under conditions of identifiability and anonymity. In addition, the model accurately predicts spatial and temporal patterns of group behavior: in this public goods dilemma, the intrinsic motivation for reputation catalyzes the development of a non-territorial, turn-taking strategy to coordinate collective action.

翻译：集体行动要求个体高效协调合作的数量、地点与时机。实验室实验已深入探究该过程的前半部分，证明多种社会认知机制会影响个体在集体努力中投入的资源数量。然而，实验研究未能阐明社会认知机制如何促进集体行动中地点与时机的协调。我们构建并测试了人类在"清洁任务"（Clean Up）这一多智能体强化学习研究领域常见的社会困境任务中的行为计算模型。研究表明，当人类群体能够识别成员身份并随时间追踪声誉时，他们能在此任务中有效合作；但匿名条件下则无法组织起来。基于声誉的多智能体强化学习模型在可识别与匿名条件下呈现出相同的合作差异。此外，该模型准确预测了群体行为的时空模式：在此公共品困境中，声誉的内在动机催化了非领地性轮换策略的形成，以协调集体行动。