The human tendency to impose costs on those who have behaved antisocially towards third parties (third-party punishment) has a formative influence on societies, yet very few studies of the development of this tendency exist. In most studies where young children have punished, participants have imposed costs on puppets, leaving open the question as to whether young children punish in real third-party situations. Here, five-year-olds were given the opportunity to allocate desirable or unpleasant items to antisocial and neutral adults, who were presented as real and shown on video. Neutral individuals were almost always allocated only desirable items. Antisocial individuals were instead usually allocated unpleasant items, as long as participants were told they would give anonymously. Most participants who were instead told they would give in person did not allocate unpleasant items, although a minority did so. This indicates that the children interpreted the situation as real, and that whereas they genuinely desired to punish antisocial adults, they did not usually dare do so in person. Boys punished more frequently than girls. The willingness of preschoolers to spontaneously engage in third-party punishment, occasionally even risking the social costs of antagonizing an anti-social adult, demonstrates a deep-seated early-developing punitive sentiment in humans.

When prompted, preschoolers advocate punishment for moral transgressions against third parties, but little is known about whether and how they might act out such punishment. In this study, adult demonstrators enacted doll stories in which a perpetrator child doll made an unprovoked attack on a victim child doll, after which an adult doll punished either the perpetrator (consistent punishment) or victim (inconsistent punishment). When asked to help retell the story, given free choice of their own preferred actions for the adult doll, 4-year-olds (N = 32) were influenced by the demonstrated choice of target when selecting a target for punishment or admonishment. This influence was weak following inconsistent punishment, however, because the participants tended to change the story by punishing or admonishing the perpetrator when the demonstrator had punished the victim. Four-year-olds’ tendency to select a moral rule violator as a target for punishment is therefore stronger than their tendency to copy the specific actions of adults, which itself is known to be very strong. The evidence suggests that 4-year-olds’ enactment of punishment is at least partially based on a belief that antisocial actions deserve to be punished.