Abstract: Previous studies have explored the impact of sanction costs on third-party punishment (TPP) from the perspective of violators and punishers: no statistical difference has been found in the reduction of uncooperative behaviors between conditions when the amount of sanction costs is different, whereas an inverse relationship between amount of costs and frequency of punishing has been documented. However, one shortcoming of previous research is the neglect of the impact of different cost forms on TPP. The cost form was addressed in this paper as another significant determinant of the level of TPP after controlling for the effect of the amount of punishment costs. Three experiments were conducted to examine the influence of cost form on TPP. Experiment 1 analyzed the effect of two different cost forms (payment and retaliation) on the frequency of TPP. 180 university students participated in a five-round Dictator Game, which used a between-subjects design. All participants played the role of third-party and made punishment decisions for each of the five unfair offers presented in random order. If the participants decided to punish, those in the Payment treatment paid 2 monetary units (MU), while the payoff of those in the Retaliation treatment would be reduced by 2MU by the punished. Experiment 2 (with 184 participants) examined the influence of risk on TPP. The participants played a one-shot Dictator Game in which their payoff would be reduced by the punished if they decided to impose a sanction. The participants in the Certainty treatment were presented with the information that the probability of retaliation by the punished was 100% and the amount reduced was 1MU, while the information presented to the Uncertainty treatment was that the probability of retaliation was 50% and the amount reduced was 2MU. Experiment 1 was replicated in the first part of Experiment 3 (with 180 non-student participants), and after that, all the participants were asked to declare the amount of compensation they needed if they were required to withdraw all the punishment decisions. The results of Experiment 1 showed that the punishment frequency of the participants in the Retaliation treatment was significantly lower than that of those in the Payment treatment. Experiment 2 found no statistical difference in punishment frequency between the Uncertainty treatment and the Certainty treatment, suggesting that although the participants in the two treatments in Experiment 1 faced different levels of risk when required to make the sanction decision, difference in the sanction frequency in Experiment 1 was not caused by the risk aversion of the participants. Experiment 3 found that the results of Experiment 1 were robust, and the regression analysis showed that the different subjective values assigned by the participants to the different cost forms had a significant impact on the punishment frequency. This paper found evidence that after controlling for the effect of the amount of punishment costs, the participants had different subjective perceptions of different cost forms, suggesting that it is the individual’s subjective perception of the punishment costs rather than the objective amount that affects the level of TPP. One the other hand, despite being to some extent genuinely nonstrategically motivated, TPP is sensitive not only to the amount of punishment costs, but also to the cost forms, which leads to the belief that the real effect of TPP may have been overestimated.

Key words: third-party punishment, punishment cost, cost sensitivity, cost form

摘要： 第三方惩罚不仅对惩罚成本的数量敏感，也对成本形式敏感。在保持成本数量相等的情况下，不同形式的惩罚成本会对第三方惩罚的频率产生不同影响。相较支付形式，当惩罚成本采取报复形式时，被试的第三方惩罚频率显著下降。进一步的实验表明，惩罚频率的变化并非因为不同的成本形式所隐含的风险水平有所差异，而是因为被试对不同的成本形式赋予了不同的主观价值。进一步探索更多形式的惩罚成本是未来研究的重要方向。

关键词: 第三方惩罚, 惩罚成本, 成本敏感性, 成本形式