Non-Binding Suggestions, Self-Enforcing Norms and the Determinants of Cooperation in Repeated Games

面对机会主义动机，实现共同利益往往是困难的。重复互动可以缓解这种困难：未来互动中的奖励和惩罚可能会克服短视的考虑。这样，合作就变成了一种可能性，但不是必须的。该项目通过实验研究了这种合作规范的自我可持续性，并解决了如何使更好的规范成为焦点的问题。社会通常期望领导者（如首席执行官、政治家和学校校长）发表声明，希望这种“软推动”能影响制度文化和规范。这个项目调查了在重复的关系和社区的背景下，通过建议的力量来影响和维持期望的结果是否可行。它将揭示领导层将行为转向理想结果的能力的决定因素；调查哪些建议的行为会产生最大的坚持，哪些没有希望出现；检查建议是否有持续的影响；并考虑用道德语言表达的建议是否比用中性语言表达的建议更有效。社会困境是普遍存在的，但在重复互动的背景下，理论上是可以解决的。然而，理论只是揭示了可能结果的集合在扩大。它并没有阐明如何实现一种平衡，以实现所期望的社会结果。最近关于重复游戏的实验文献考虑了在过滤掉短暂行为后，哪些规范会在实验室中自发结晶。然而，在实验室中产生的平衡可能与在更丰富的环境中产生的平衡不太接近，在更丰富的环境中，历史背景、战略的共享意义和游戏前沟通发挥了作用。因此，在评估合作的极限时，当前的实验可能过于苛刻：没有看到合作自发出现可能无法反映出现实问题中合作的全部潜力。实验室环境无法重现平衡可能成为焦点的许多方式。这项研究采用了一种更为温和的方法，即使用非约束性建议（类似于领导人的声明）来测试不同策略的自我执行性质。这并不需要目睹策略的自我出现，也不需要引入违背实验室设置目的的不受控制的元素。如果你不能自愿地遵循一些非约束性的合作均衡策略，这就证明了它不太可能在现实生活中自发地结晶并持续下去。该方法包括改变延续概率，作为对第三方（在本例中是实验者）所作陈述引起的需求效应的自然处理控制。这项研究将调查许多问题，包括提出不具约束力的建议（如“严厉触发”或“以牙还牙”的策略）是否比让约定自发形成时效率更高；以及非约束性建议如何与先前研究的合作决定因素相互作用（例如，风险支配，风险厌恶和利他主义的作用）。除了标准的重复游戏之外，本项目还将探讨社区执行的既定理论概念的实证有效性。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

It is often difficult to achieve the common good in the face of opportunistic incentives. Repeated interaction can alleviate this difficulty: rewards and punishments in future interactions can possibly overcome shortsighted considerations. Cooperation then becomes a possibility, but not a necessity. The project experimentally investigates the self-sustainability of such cooperative norms, and addresses the question of how to make better norms focal. Society often expects leaders (such as CEOs, politicians, and school principals) to make pronouncements in the hope that such 'soft nudges' impact institutional culture and norms. This project investigates whether it is feasible, through the power of suggestion, to effect and sustain desired outcomes in the context of repeated relationships and communities. It will shed light on the determinants of leadership’s ability to shift behavior towards desirable outcomes; investigate which suggested behaviors yield the greatest adherence, and which have little hope of emerging; examine whether suggestions have persistent effects; and consider whether suggestions couched in moral language are more effective than those phrased in neutral language.The social dilemma is pervasive but theoretically solvable in the context of repeated interactions. Yet theory only reveals that the set of possible outcomes expands. It does not shed light on how to bring about an equilibrium achieving the desired societal outcomes. The recent experimental literature on repeated games considers what norms will spontaneously crystallize in the lab, after filtering out transitory behavior. However, the equilibrium that arises in the lab may not come close to those arising in richer settings, where historical context, strategies' shared meaning, and pre-play communication play a role. As such, current experiments may be overly harsh when it comes to assessing the limits of cooperation: not seeing cooperation spontaneously emerge may not reflect the full potential for cooperation in real-life problems. A laboratory environment cannot recreate the many ways in which an equilibrium might become focal. This study takes the more modest approach of testing the self-enforcing nature of different strategies using non-binding suggestions, akin to the pronouncements made by leaders. This does not require having to witness the self-emergence of strategies, nor does it require introducing uncontrolled elements that defy the purposes of a lab setting. Failing to voluntarily follow a non-binding suggestion of some cooperative equilibrium strategies is evidence that it is unlikely to spontaneously crystallize and persist in real life. The methodology involves varying the continuation probability, as a natural treatment-control for demand effects arising from statements made by third parties (in this case, the experimenter). This study will investigate many questions, including whether efficiency is higher when making a non-binding suggestion such as grim-trigger or tit-for-tat strategies, as opposed to when letting a convention emerge spontaneously; and how non-binding suggestions interact with previously studied determinants of cooperation (e.g., the role of risk dominance, risk aversion and altruism). In addition to standard repeated games, this project will also explore the empirical validity of established theoretical notions of community enforcement.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.