When Do We Punish People Who Don't?

Justin W. Martin; Jillian J. Jordan; David G. Rand; Fiery  Cushman

Article | Cognition | December 2019

When Do We Punish People Who Don't?

by Justin W. Martin, Jillian J. Jordan, David G. Rand and Fiery Cushman

Abstract

People often punish norm violations. In what cases is such punishment viewed as normative—a behavior that we “should”or even“must”engage in? We approach this question by asking when people who fail to punish a norm violator are, themselves, punished. (For instance, a boss who fails to punish transgressive employees might,herself, befired.) We conducted experiments exploring the contexts in which higher-order punishment occurs, using both incentivized economic games and hypothetical vignettes describing everyday situations. We presented participants with cases in which an individual fails to punish a transgressor, either as a victim (second party) or as an observer (third party). Across studies, we consistently observed higher-order punishment of non-punishing observers. Higher-order punishment of non-punishing victims, however, was consistently weaker, and sometimes non-existent. These results demonstrate the selective application of higher-order punishment, provide a new perspective on the psychological mechanisms that support it, and provide some clues regarding itsfunction.

Keywords: punishment; norms; Cooperation; Societal Protocols; Adaptation;

Format: Print

Find at Harvard Read Now

Citation:

Martin, Justin W., Jillian J. Jordan, David G. Rand, and Fiery Cushman. "When Do We Punish People Who Don't?" Cognition 193 (December 2019).

Article | Journal of Personality and Social Psychology

Signaling When Nobody Is Watching: A Reputation Heuristics Account of Outrage and Punishment in One-shot Anonymous Interactions

Jillian J. Jordan and David G. Rand

Moralistic punishment can confer reputation benefits by signaling trustworthiness to observers. However, why do people punish even when nobody is watching? We argue that people often rely on the heuristic that reputation is typically at stake, such that reputation concerns can shape moralistic outrage and punishment even in one-shot anonymous interactions. We then support this account using data from Amazon Mechanical Turk. In anonymous experiments, subjects (total n = 8,440) report more outrage in response to others’ selfishness when they cannot signal their trustworthiness through direct prosociality (sharing with a third party)—such that if the interaction were not anonymous, punishment would have greater signaling value. Furthermore, mediation analyses suggest that sharing opportunities reduce outrage by influencing reputation concerns. Additionally, anonymous experiments measuring costly punishment (total n = 6,076) show the same pattern: subjects punish more when sharing is not possible. Moreover, and importantly, moderation analyses provide some evidence that sharing opportunities do not merely reduce outrage and punishment by inducing empathy toward selfishness or hypocrisy aversion among non-sharers. Finally, we support the specific role of heuristics by investigating individual differences in deliberateness. Less deliberative individuals (who typically rely more on heuristics) are more sensitive to sharing opportunities in our anonymous punishment experiments, but, critically, not in punishment experiments where reputation is at stake (total n = 3,422); and not in our anonymous outrage experiments (where condemning is costless). Together, our results suggest that when nobody is watching, reputation cues nonetheless can shape outrage and—among individuals who rely on heuristics—costly punishment.

Keywords: signaling; Morality; trustworthiness; anger; third-party punishment; Moral Sensibility; Behavior; Trust; Reputation;

Citation:

Jordan, Jillian J., and David G. Rand. "Signaling When Nobody Is Watching: A Reputation Heuristics Account of Outrage and Punishment in One-shot Anonymous Interactions." Journal of Personality and Social Psychology 118, no. 1 (January 2020). View Details

Cite View Details Find at Harvard Purchase

The social function of witchcraft accusations remains opaque. An empirical study of Chinese villagers shows that the label ‘z hu’ influences who interacts across a social network, but appears not to tag defectors in service of promoting cooperation. An open question thus remains: from witchcraft to gossip, which accusations stick?

Why do individuals pay costs to punish selfish behavior, even as third-party observers? A large body of research suggests that reputation plays an important role in motivating such third-party punishment (TPP). Here we focus on a recently proposed reputation-based account (Jordan et al., 2016) that invokes costly signaling. This account proposed that “trustworthy type” individuals (who are incentivized to cooperate with others) typically experience lower costs of TPP, and thus that TPP can function as a costly signal of trustworthiness. Specifically, it was argued that some but not all individuals face incentives to cooperate, making them high-quality and trustworthy interaction partners; and, because the same mechanisms that incentivize cooperation also create benefits for using TPP to deter selfish behavior, these individuals are likely to experience reduced costs of punishing selfishness. Here, we extend this conceptual framework by providing a concrete, “from-the-ground-up” model demonstrating how this process could work in the context of repeated interactions incentivizing both cooperation and punishment. We show how individual differences in the probability of future interaction can create types that vary in whether they find cooperation payoff-maximizing (and thus make high-quality partners), as well as in their net costs of TPP – because a higher continuation probability increases the likelihood of receiving rewards from the victim of the punished transgression (thus offsetting the cost of punishing). We also provide a simple model of dispersal that demonstrates how types that vary in their continuation probabilities can stably coexist, because the payoff from remaining in one’s local environment (i.e. not dispersing) decreases with the number of others who stay. Together, this model demonstrates, from the group up, how TPP can serve as a costly signal of trustworthiness arising from exposure to repeated interactions.

Faculty & Research

When Do We Punish People Who Don't?

Abstract

Citation:

About the Author

More from the Author