Department of Psychology, University of Michigan;
Felix Warneken
Department of Psychology, University of Michigan
Acknowledgement: This research was funded by a National Science Foundation CAREER Grant (1760238) to Felix Warneken. We thank the research assistants for their help with data collection. We also thank Sebastian Grüneisen for helpful comments on the previous version of this article, Amy Nowack for proofreading the article, and CSCAR at the University of Michigan for statistical advice.
We pre-registered our hypotheses and analyses (
Fairness norms guide how resources should be divided between individuals (
Developmental studies have begun to trace the developmental trajectory of TPP in children. Children’s TPP against unfairness has been found in children aged 6 and older (
Although these earlier studies provide insight into the development of TPP, one critical question remains. In previous studies, children’s punishment was binary (
Here we examine whether children punish with the aim to create equality. We start with the notion that at least in adults, TPP has been identified as a potential mechanism to enforce norms. Therefore, if children’s TPP is motivated to create equality, children should punish in a way that reduces inequality among third parties. Alternatively, if their TPP is driven by a self-centered motive such as spite, competition (
To examine whether children intervene with the goal to create equality, we presented participants with three (preprogrammed) allocations between two peers that were represented as two avatars on a computer screen: fair allocations (2:2), mildly unfair allocations (3:1) and extremely unfair allocations (4:0). Critically, our participants were in the role of a third-party and were free to choose how many coins they wanted to take away from which individual. Children therefore decided not only whether to punish but also on the degree of punishment.
We tested the hypothesis that children use punishment to establish equality against several other possible outcomes. Here we focus on how children would respond to 3:1 allocations which are the critical trials that allow us to disentangle different hypotheses.
Our final sample were N = 60 five- to 9-year-old children (M = 88.47 months, range = 61–119 months, n = 12 in each age group, 30 male, 30 female). Children were tested at a museum in the Midwest of the US. Demographic information such as race, education and income could not be obtained as per the rules of the museum. Four additional children were excluded because of failure to correctly answer at least one of the comprehension checks (2), parental interference (1), or parental report of their child having autism (1).
With a final sample of N = 60, we conducted n = 1,000 simulations using the package simr (
This study was approved by the institutional review board at the University of Michigan (IRB protocol number: HUM00139220, IRB protocol title: Third-Party Interventions).
After parents gave written consent, children sat at a table with the study apparatus while the parents watched passively from a few steps away. A female experimenter introduced the computer game referred to as the “coin game” and explained that players could collect virtual coins to later exchange for prizes. During a prize introduction, children learned that the more coins they have during the coin game, the more and the better prizes they would be able to choose afterward (Section 1.1. on p. 1 in online supplemental materials).
In the subsequent practice phase, the experimenter introduced the two other players in the game by stating that they were children of the same age and gender at another museum, who are currently connected online. In reality, the decisions of the two other players were computer-programmed. The experimenter introduced the role of the divider and the recipient. Children were told that the divider has four coins and can give any number of coins to the recipient. The divider then made one of three allocations: (a) two for the self and two for the recipient, (b) three for the self and one for the recipient, and (c) four for the self and none for the recipient. The recipient was a passive player who could only accept the divider’s allocation.
After introducing the roles, children watched on the screen how the divider produced different allocations and practiced their role as a third-party punisher. After the divider made an allocation, children could press either the gray button or their own coin above their basket (see
Alternatively, if children pushed their own coin above their own basket (rejection), they could decide which other coins they wanted to take away from either the divider or the recipient or both players. When children pushed a coin, a vacuum appeared at the top of the screen and sucked up the coin, such that no one could keep it. Any remaining coins on the screen that children choose not to take away from players went into each player’s basket. Therefore, in our game, children could be flexible about who to punish (divider, recipient or both) and the number of coins they want to take away. Critically, regardless of allocation, children had to pay their one coin first to take coins away from others.
Rejection rate refers to how often children paid their cost to take coins away from other players, which is the binary measure used in prior studies as well (
There were four practice trials in total. Children practiced four possible outcomes of each button (accept vs. reject) in each allocation (fair vs. unfair). The experimenter asked comprehension checks about the consequence of each button and whether each button required the payment of the child participant’s coin or not. All children included in the data analysis passed these comprehension checks.
After the practice trials, to make children believe that the other players were real, the experimenter pretended to call the other players on speakerphone and checked if they were ready to play the game. In reality, a confederate answered the phone call. Upon the completion of the study, the experimenter left and a secondary experimenter asked children whether they thought the players were real or pretend. We found that 78% of children (47 out of 60) said the players were real.
During the subsequent test phase, children played 6 rounds as a third-party observer in total. Children received 20 coins as their initial endowment (coins that dropped into their basket on the screen). Children’s initial endowment was chosen based on prior research that used a costly third-party punishment task (
After the test phase, to assess whether children’s numeric understanding affected their performances in the coin game, we administered a nonsocial numeric task. Children saw four triangles on the computer screen that were identical to the three allocations in the coin game and were asked to make both sides equal (see Section 1.2. on p. 2 in online supplementary materials). Children were able to create equal numbers of triangles in 96% of the trials. Thus, any potential age-related sensitivity to equality in the coin game cannot be attributed to their numeric inability to make both sides equal.
We counterbalanced the order of test trials, practice trials, deception check questions, numerical task trials, the other player’s identity, and the type of unfair allocation during practice.
Children’s responses were recorded by GameMaker Studio (
We preregistered our hypotheses and analyses before data collection (
We analyzed the rejection rate and the rate of creating exact equality with Generalized Linear Mixed Models (GLMM) using an R package “glmmTMB” (
We assessed whether children’s decision to intervene was influenced by age and allocation. A full GLMM on children’s rejection (0 = acceptance, 1 = rejection) provided a better fit to the data than the null model (LRT, χ
One could speculate that 4:0 allocations would provoke more punishment than 3:1 allocations. However, we found no significant difference in the rejection rate between 3:1 and 4:0 allocations, suggesting that both types of unfair allocations elicit similar responses in children. Thus, it seems that the relatively smaller deviation from equality is sufficient to induce punishment.
Next, we examined the rate at which children used punishment to establish perfect equality such that both players ended up with exactly the same number of coins. Here we look at the final outcomes across all trials, aggregating across trials with and without punishment. A full GLMM on children’s creation of perfect equality between two other players (0 = inequality, 1 = perfect equality) provided a significantly better fit to the data than the null model (LRT, χ
Importantly, we found a main effect of age, suggesting that across allocations, children became more likely to establish exact equality with age (b = .05, SE = .01, p < .001). As shown in
One question is how children achieved exact equality. For example, exact equality could be established by for example, turning 3:1 into 1:1 (by only taking away coins from the unfair divider) or 0:0 (by taking coins away from everyone). To assess this, we calculated the average number of coins taken away from each player across all trials, aggregating across trials with and without punishment. In terms of the number of coins children took away from the divider, with age, children became more likely to take coins away from the unfair divider (
Additionally, the distributional outcomes in 3:1 allocations provide converging evidence (see
Taken together, with age, children take more coins from unfair dividers, whereas they take fewer coins from fair dividers and recipients, suggesting that they establish equality by directing punishment toward unfair dividers specifically.
Our findings demonstrate the development of punishment in which children use punishment to create equality. This was possible because in our task, children were able to express their distributional preference by deciding not only whether to punish, but how much to punish. We showed that with age, (a) children were willing to pay a cost to punish unfairness, and (b) became increasingly interested in creating exact equality. These findings support our hypothesis that with age, children use punishment to restore equality. Our study shows that children use punishment not only to prevent unequal outcomes, but specifically to create equality.
One concern might be that children’s punishment reflects a competitive motive to win against others. Specifically, if children’s punishment is driven by competition, (a) children should punish fair as well as unfair allocations indiscriminately and (b) their punishment should result in 0:0, taking coins away from the recipient as well as the divider. By contrast, if children’s punishment is motivated by a concern to establish equality, (a) children should punish unfair allocations more often than fair allocations and (b) their punishment should aim at unfair dividers, not disadvantaged recipients. Our results do not support the competition account. Specifically, in both 2:2 and 3:1 allocations, creating 0:0 was observed in a minority of trials across ages and became less than 10% of trials at age 9 (see
Another possibility was that children are focused solely on the unfair dividers without considering how their punishment impacts the balance between the two individuals. In fact, they might try to avenge the victim and overpunish in terms of taking away more from the divider than is necessary to create equality (e.g., turning 3:1 into 0:1). Prior work using punishment as a binary measure could not directly address this hypothesis. However, our study using a continuous measure showed that this account is unlikely to explain children’s punishment. Children rarely took away all the coins from the unfair divider: When we look at trials in which children punished 3:1 allocations, only 15% of punishments resulted in 0:1 across ages. Especially among 9-year-olds who rejected 3:1 allocations, a vast majority of trials (71%) resulted in 1:1, whereas only 7% and 14% of the trials were turned into 0:0 and 0:1, respectively. Together, the results do not support the notion that children punish to avenge victims and not caring about an egalitarian outcome. In summary, with age, children’s punishment reflects their fairness concern to move allocations toward equality rather than to outcompete others or simply punish unfair individuals.
Previous research also supports the notion that older children’s punishment reflects their fairness concern. In
Our study found an interesting developmental change. Five-year-olds intervened in both fair and unfair allocations indiscriminately. By age 9, however, children became highly selective: nine-year-olds punished unfair allocations in a way that could establish equality and rarely punished fair allocations. Converging evidence comes from studies using different paradigms. For instance,
One remaining question is why older children did not always punish unfair allocations. Although 9-year-olds were most likely to create equality, they were far from perfect (establishing equality in about 50% of all unfair trials in which children either accepted or rejected). It is likely that the costs of punishment played a role: Neither children nor adults always punish in third-party or second-party punishment paradigms when costs are involved (
We want to emphasize that our findings do not necessarily imply that 5-year-olds lack a sense of fairness per se. In contexts in which there is no cost to the self, even infants and young children expect equal allocations (
We would like to address some potential limitations of our paradigm. Although the computer game allows for better experimental control and anonymity, it obviously only simulates social interactions. It was encouraging to learn that the vast majority of children said that they interacted with real children, although 13 of 60 did not. To address this, we confirmed that the results were the same when these children were excluded from analyses, suggesting that they did not bias our findings (see Section 2.7. on p. 13 in the online supplemental materials). Another feature is that because children interacted with the same divider and recipient throughout, they might not have focused on individual allocations but the whole sequence of events. However, supplementary analyses indicated that there were no effects involving trial number or children’s intervention decision in a previous trial (see Section 2.8. on p. 14 in the online supplemental materials). Lastly, our task required a basic understanding of numerical equivalence. The nonsocial numeric task showed that even our youngest children had no problem making both sides equal, suggesting that the lack of equality restoration in 5-year-olds cannot be attributed to a lack of numerical understanding (see Section 1.2. on p. 2 in the online supplemental material).
One question for future research is why older children showed equality-oriented TPP. It is possible that this reflects a genuine sense of fairness that emerges at this age. However, this might be partly influenced by older children wanting to signal how much they care about fairness. Future research should examine how children’s reputational concerns and norms of fairness might interact in the emergence of TPP.
Another question for future research is how an alternative intervention option might affect children’s punishment decisions. For example, if children have an alternative option to redistribute coins (turning 3:1 into 2:2), even young children might show an increased rate of creating equality. Thus, it would be important to investigate how having an alternative option such as redistribution or compensation compares to children’s use of punishment.
Another critical question is individual differences in children’s punishment. Although the current study was not designed to assess individual differences, exploratory analyses looking at individual patterns of responding indicate that some young children showed an advanced sense of fairness earlier compared with their peers (see Section 2.11. on p. 22 in the online supplemental material for details). Future research could use our paradigm to systematically test individual differences in fairness development.
Last, future research should examine children’s punishment of norm-violations in contexts other than egalitarian fairness norms. This might include free-riding (
Bernhard, R. M., Martin, J. W., & Warneken, F. (2020). Why do children punish? Fair outcomes matter more than intent in children’s second- and third-party punishment. Journal of Experimental Child Psychology, 200, 104909. 10.1016/j.jecp.2020.104909
Blake, P. R., & McAuliffe, K. (2011). I had so much it didn’t seem fair”: Eight-year-olds reject two forms of inequity. Cognition, 120(2), 215–224. 10.1016/j.cognition.2011.04.006
Brooks, M. E., Kristensen, K., van Benthem, K. J., Magnusson, A., Berg, C. W., Nielsen, A., Skaug, H. J., Machler, M., Bolker, B. M. (2017). glmmTMB balances speed and flexibility among packages for zero-inflated generalized linear mixed modeling. The R Journal, 9(2), 378–400. 10.32614/RJ-2017-066
Dawes, C. T., Fowler, J. H., Johnson, T., McElreath, R., & Smirnov, O. (2007). Egalitarian motives in humans. Nature, 446(7137), 794–796. 10.1038/nature05651
Deutsch, M. (1975). Equity, equality, and need: What determines which value will be used as the basis of distributive justice?Journal of Social Issues, 31(3), 137–149. 10.1111/j.1540-4560.1975.tb01000.x
Fehr, E., & Fischbacher, U. (2003). The nature of human altruism. Nature, 425(6960), 785–791. 10.1038/nature02043
Fehr, E., & Fischbacher, U. (2004). Third party punishment and social norms. Evolution and Human Behavior, 25(2), 63–87. 10.1016/S1090-5138(04)00005-4
Fehr, E., & Schmidt, K. M. (1999). A Theory of fairness, competition, and cooperation. The Quarterly Journal of Economics, 114(3), 817–868. 10.1162/003355399556151
Fischbacher, U., Gächter, S., & Fehr, E. (2001). Are people conditionally cooperative? Evidence from a public goods experiment. Economics Letters, 71(3), 397–404. 10.1016/S0165-1765(01)00394-9
Geraci, A., & Surian, L. (2011). The developmental roots of fairness: Infants’ reactions to equal and unequal distributions of resources. Developmental Science, 14(5), 1012–1020. 10.1111/j.1467-7687.2011.01048.x
Green, P., & MacLeod, C. J. (2016). simr: An R package for power analysis of generalised linear mixed models by simulation. Methods in Ecology and Evolution, 7(4), 493–498. 10.1111/2041-210X.12504
Gummerum, M., & Chu, M. T. (2014). Outcomes and intentions in children’s, adolescents’, and adults’ second- and third-party punishment behavior. Cognition, 133(1), 97–103. 10.1016/j.cognition.2014.06.001
Gürerk, O., Irlenbusch, B., & Rockenbach, B. (2006). The competitive advantage of sanctioning institutions. Science, 312(5770), 108–111. 10.1126/science.1123633
Hamann, K., Bender, J., & Tomasello, M. (2014). Meritocratic sharing is based on collaboration in 3-year-olds. Developmental Psychology, 50(1), 121–128. 10.1037/a0032965
Hamann, K., Warneken, F., Greenberg, J. R., & Tomasello, M. (2011). Collaboration encourages equal sharing in children but not in chimpanzees. Nature, 476(7360), 328–331. 10.1038/nature10278
Henrich, J., McElreath, R., Barr, A., Ensminger, J., Barrett, C., Bolyanatz, A., & Ziker, J., (2006). Costly punishment across human societies. Science, 312(5781), 1767–1770. 10.1126/science.1127333
House, B. R., Kanngiesser, P., Barrett, H. C., Yilmaz, S., Smith, A. M., Sebastian-Enesco, C., Erut, A., & Silk, J. B. (2020). Social norms and cultural diversity in the development of third-party punishment. Proceedings of the Royal Society, B: Biological Sciences, 287(1925), 20192794. 10.1098/rspb.2019.2794
Johnson, T., Dawes, C. T., Fowler, J. H., McElreath, R., & Smirnov, O. (2009). The role of egalitarian motives in altruistic punishment. Economics Letters, 102(3), 192–194. 10.1016/j.econlet.2009.01.003
Jordan, J. J., McAuliffe, K., & Warneken, F. (2014). Development of in-group favoritism in children’s third-party punishment of selfishness. Proceedings of the National Academy of Sciences of the United States of America, 111(35), 12710–12715. 10.1073/pnas.1402280111
Lee, Y. E., & Warneken, F. (2020). Children’s evaluations of third-party responses to unfairness: Children prefer helping over punishment. Cognition, 205, 104374. 10.1016/j.cognition.2020.104374
Lee, Y., & Warneken, F. (2019). Is punishment for equality? [https://aspredicted.org/up4qS.pdf]
Lee, Y., & Warneken, F. (2021). Does third-party punishment in children aim at equality?
Marshall, J., Yudkin, D. A., & Crockett, M. J. (2021). Children punish third parties to satisfy both consequentialist and retributive motives. Nature Human Behaviour, 5, 361–368. 10.1038/s41562-020-00975-9
McAuliffe, K., Blake, P. R., Steinbeis, N., & Warneken, F. (2017). The developmental foundations of human fairness. Nature Human Behaviour, 1(2), 0042. 10.1038/s41562-016-0042
McAuliffe, K., Jordan, J. J., & Warneken, F. (2015). Costly third-party punishment in young children. Cognition, 134, 1–10. 10.1016/j.cognition.2014.08.013
Mendes, N., Steinbeis, N., Bueno-Guerra, N., Call, J., & Singer, T. (2018). Preschool children and chimpanzees incur costs to watch punishment of antisocial others. Nature Human Behaviour, 2(1), 45–51. 10.1038/s41562-017-0264-5
Nelissen, R. M. A., & Zeelenberg, M. (2009). Moral emotions as determinants of third-party punishment: Anger, guilt, and the functions of altruistic sanctions. Judgment and Decision Making, 4(7), 543–553.
R Core Team. (2018). R: A language and environment for statistical computing. R Foundation for Statistical Computing.
Raihani, N. J., & Bshary, R. (2019). Punishment: One tool, many uses. Evolutionary Human Sciences, 1(e12), 1–26. 10.1017/ehs.2019.12
Riedl, K., Jensen, K., Call, J., & Tomasello, M. (2015). Restorative justice in children. Current Biology, 25(13), 1731–1735. 10.1016/j.cub.2015.05.014
Shaw, A., & Olson, K. R. (2012). Children discard a resource to avoid inequity. Journal of Experimental Psychology: General, 141(2), 382–395. 10.1037/a0025907
Sheskin, M., Bloom, P., & Wynn, K. (2014). Anti-equality: Social comparison in young children. Cognition, 130(2), 152–156. 10.1016/j.cognition.2013.10.008
Sloane, S., Baillargeon, R., & Premack, D. (2012). Do infants have a sense of fairness?Psychological Science, 23(2), 196–204. 10.1177/0956797611422072
Smith, C. E., & Warneken, F. (2016). Children’s reasoning about distributive and retributive justice across development. Developmental Psychology, 52(4), 613–628. 10.1037/a0040069
Smith, C. E., Blake, P. R., & Harris, P. L. (2013). I should but I won’t: Why young children endorse norms of fair sharing but do not follow them. PLoS ONE, 8(3), e59510. 10.1371/journal.pone.0059510
Trivers, R. L. (1971). The Evolution of reciprocal altruism. The Quarterly Review of Biology, 46(1), 35–57. 10.1086/406755
Twardawski, M., & Hilbig, B. E. (2020). The motivational basis of third-party punishment in children. PLoS ONE, 15, e0241919. 10.1371/journal.pone.0241919
Warneken, F., Lohse, K., Melis, A. P., & Tomasello, M. (2011). Young children share the spoils after collaboration. Psychological Science, 22(2), 267–273. 10.1177/0956797610395392
Yamagishi, T., Li, Y., Fermin, A. S. R., Kanai, R., Takagishi, H., Matsumoto, Y., Kiyonari, T., Sakagami, M. (2017). Behavioural differences and neural substrates of altruistic and spiteful punishment. Scientific Reports, 7, 14654. 10.1038/s41598-017-15188-w
Yang, F., Choi, Y. J., Misch, A., Yang, X., & Dunham, Y. (2018). In defense of the commons: Young children negatively evaluate and sanction free riders. Psychological Science, 29(10), 1598–1611. 10.1177/0956797618779061
Yudkin, D. A., Van Bavel, J. J., & Rhodes, M. (2020). Young children police group members at personal cost. Journal of Experimental Psychology: General, 149(1), 182–191. 10.1037/xge0000613
Submitted: October 20, 2020 Revised: November 24, 2021 Accepted: December 3, 2021