Replication data: Generosity is a sign of trustworthiness – the punishment of selfishness is not

Peer-punishment is an important determinant of cooperation in human groups. It has been suggested that, at the proximate level of analysis, punitive preferences can explain why humans incur costs to punish their deviant peers. How punitive preferences could have evolved in humans is still not entirely understood. A possible explanation at the ultimate level of analysis comes from signaling theory. It has been argued that the punishment of defectors can be a type-separating signal of the punisher's cooperative intent. As a result, punishers are selected more often as interaction partners in social exchange and are partly compensated for the costs they incur when punishing defectors. A similar argument has been made with regard to acts of generosity. In a laboratory experiment, we investigate whether the punishment of a selfish division of money in a dictator game is a sign of trustworthiness and whether punishers are more trustworthy interaction partners in a trust game than non-punishers. We distinguish between second-party and third-party punishment and compare punitive acts with acts of generosity as signs of trustworthiness. We find that punishers are not more trustworthy than non-punishers and that punishers are not trusted more than non-punishers, both in the second-party and in the third-party punishment condition. To the contrary, second-party punishers are trusted less than their non-punishing counterparts. However, participants who choose a generous division of money are more trustworthy and are trusted more than participants who choose a selfish division or participants about whom no information is available. Our results suggest that, unlike for punitive acts, the signaling benefits of generosity are to be gained in social exchange.

Peer-punishment（同伴惩罚）是人类群体中合作行为的重要决定因素。已有研究表明，从近因分析（proximate analysis）层面来看，惩罚偏好可以解释人类为何愿意付出成本惩罚越轨同伴。但人类的惩罚偏好究竟如何演化，目前仍未完全明晰。一种来自终极分析（ultimate analysis）层面的可行解释源于信号理论（signaling theory）。有观点提出，对背叛者的惩罚可作为传递惩罚者自身合作意向的类型分离信号（type-separating signal）。据此，在社会交换场景中，惩罚者会更频繁地被选为互动伙伴，且能部分弥补其实施惩罚时所付出的成本。针对慷慨行为，学界也提出了类似的论证逻辑。本研究依托实验室实验，探究独裁者博弈（dictator game）中针对自私金钱分配行为的惩罚是否可作为可信任度的信号，同时验证在信任博弈（trust game）中，惩罚者是否比非惩罚者更值得作为互动伙伴。我们区分了第二方惩罚（second-party punishment）与第三方惩罚（third-party punishment），并将惩罚行为与慷慨行为一同作为可信任度的信号展开对比分析。研究结果显示，无论是第二方还是第三方惩罚情境下，惩罚者均未比非惩罚者更值得信任，且惩罚者获得的信任度也未高于非惩罚者。与之相反，第二方惩罚者获得的信任度反而低于未实施惩罚的对应群体。不过，选择慷慨金钱分配方案的参与者，相较于选择自私分配方案的参与者，或是未获得任何相关信息的参与者，更值得信任且会获得更高的信任度。本研究结果表明，与惩罚行为不同，慷慨行为的信号收益能够在社会交换中得到实现。