What Drives Second- and Third-Party Punishment?

The Bigger Picture: Conceptual Replications of the “Intuitive Retributivism” Hypothesis

Some of the studies presented here closely followed or adapted procedures that were used in prior research to test the “intuitive retributivism” hypothesis (e.g., Molho et al., 2022; Rehren & Zisman, 2022), while others focused on replicating the “intuitive retributivism” hypothesis from a more theoretical perspective and tested the hypothesis without referring to methods from a specific original study (e.g., Funk & Mischkowski, 2022; Hechler & Kessler, 2022). In the latter case, it is more challenging to describe to what extent the new study can be considered a “conceptual replication”. In general, the studies in this topical issue (i) cover a wide range of study designs, (ii) include various populations, and (iii) scrutinize different boundary conditions.

Most studies used the “indirect approach” (see above): they systematically varied motive-congruent information about a misbehavior and then tested the effect of this variation on people’s punishment tendencies or behavior (Aharoni et al., 2022; De Cristofaro & Giacomantonio, 2022; Hechler & Kessler, 2022; Molho et al., 2022; Nockur et al., 2022). One study investigated whether people’s prospect of receiving information about the effect of their punishment on offenders changed their punishment decisions (Funk & Mischkowski, 2022). Moreover, while some of these experiments manipulated information in vignettes and measured punishment tendencies or suggestions (Aharoni et al., 2022; De Cristofaro & Giacomantonio, 2022; Molho et al., 2022), others used economic games and, thus, measured actual (incentivized) punishment behavior (Funk & Mischkowski, 2022; Hechler & Kessler, 2022; Nockur et al., 2022). Some studies did not measure punishment tendencies (or behavior) as a proxy for individuals’ punishment motives but investigated their information search behavior (Rehren & Zisman, 2022), their support for different punishment reactions (Strauß & Bondü, 2022), or direct self-reported support for various punishment motives (Fousiani & Van Prooijen, 2022).

Furthermore, the studies presented in this topical issue differ not only in the methodological approaches used to investigate the motivational basis of laypeople’s punishment but also in the populations studied. Most prominently, the majority of data were collected in English-speaking countries such as the UK (Fousiani & Van Prooijen, 2022; Molho et al., 2022; Rehren & Zisman, 2022) or the US (Aharoni et al., 2022; Hechler & Kessler, 2022) but some studies were conducted in Germany (Funk & Mischkowski, 2022; Strauß & Bondü, 2022) and Italy (De Cristofaro & Giacomantonio, 2022). Samples not only differ regarding their country of origin, but also the age group that was studied. While most of the studies consider an adult population, Strauß and Bondü (2022) aimed at replicating the “intuitive retributivism” hypothesis in children – which is a rare exception in the literature on punishment motives.

Importantly, the perspective of the punisher is typically of great relevance in the punishment literature. This also applies to the research presented in this topical issue, as some studies investigated punishment motives in third-party punishment (De Cristofaro & Giacomantonio, 2022; Fousiani & Van Prooijen, 2022; Molho et al., 2022; Rehren & Zisman, 2022; Strauß & Bondü, 2022), whereas others tested punishment motives in second-party punishment (Funk & Mischkowski, 2022; Nockur et al., 2022) or even compared the two perspectives as a potential boundary condition (Aharoni et al., 2022; Hechler & Kessler, 2022).

Lastly, several other boundary conditions of the “intuitive retributivism” hypothesis were explored, such as punisher status (Strauß & Bondü, 2022), transgression type or magnitude (Aharoni et al., 2022; Strauß & Bondü, 2022), the centrality of punishment (Nockur et al., 2022), thinking style (Rehren & Zisman, 2022), direct versus indirect punishment (Molho et al., 2022), transgressor’s power (Fousiani & Van Prooijen, 2022), and interindividual differences in punishers (De Cristofaro & Giacomantonio, 2022; Funk & Mischkowski, 2022; Strauß & Bondü, 2022).

To sum up, the authors made use of the wide spectrum of methodological decisions present in the literature on people’s punishment motives for replicating studies of the “intuitive retributivism” hypothesis. The variety of the studies presented herein allows for testing under which conditions results regarding the “intuitive retributivism” hypothesis can be replicated.

A Brief Summary of the Papers in This Issue

The first paper in this topical issue (Molho et al., 2022) closely replicates a seminal study on laypeople’s punishment motives by Carlsmith and colleagues (2002, Study 1). In this study, the effects of manipulating retribution-relevant information (i.e., offense severity) and utility-relevant information (i.e., observability of punishment) in everyday misbehavior on punishment tendencies are investigated. Extending the original study, the authors differentiated between specific punishment responses by measuring direct, confrontational punishment tendencies versus indirect punishment tendencies (i.e., gossip and social exclusion). Overall, Molho and colleagues (2022) successfully conceptually replicated the results of the original study, showing that manipulating offense severity shifted people’s punishment behavior (irrespective of the type of punishment), whereas punishment observability did not.

A similar methodological approach was taken in the second paper by De Cristofaro and Giacomantonio (2022). Again, retribution-relevant information (i.e., intentionality of misbehavior) and utility-relevant information (i.e., difficulty to detect the misbehavior) were manipulated orthogonally, while focusing on a specific type of misbehavior, that is, tax evasion. In addition, the authors measured a potential moderator: economic system justification, that is, the tendency to justify the economic status quo in a society. In line with the “intuitive retributivism” hypothesis, retribution-relevant information affected punitive decisions more strongly than utility-related information, and the effect of retribution-relevant information was stronger among participants scoring low on economic system justification. Respondents with high system justification scores were less punitive in general.

A different methodological approach was pursued by Nockur and colleagues (2022) who applied an economic game to study people’s punishment behavior (following Crockett et al., 2014). Participants played this game under three conditions (within-subjects): a baseline condition (“unintentional offense”), a “hidden punishment” condition (allowing for retribution, but not for deterrence), and an “open punishment” condition (allowing for both retribution and deterrence). In line with the “intuitive retributivism” hypothesis, participants were most punitive in the “hidden punishment” condition, especially when participants were the only players in their group who were able to punish (“centralized punishment”).

An economic game approach was also used by Hechler and Kessler (2022), again investigating the effects of the intentionality of misbehavior on costly punishment. Interestingly, while the studies outlined above investigated the motivational basis of third-party punishment, Hechler and Kessler (2022) also studied second-party punishment and showed that retribution drives punishment in both third and second parties.

Also, using an economic game, Funk and Mischkowski (2022) conceptually replicated previous findings that challenged the “intuitive retributivism” hypothesis. Specifically, they studied the effects of the availability of information on the consequences of one’s punishment (i.e., whether or not punishers would be informed about the effect of their punishment on the transgressor) on punishment behavior. Again, just as the “intuitive retributivism” hypothesis would predict, there were no effects of future notification about the effect of punishment on people’s actual punishment behavior.

Correspondingly, Aharoni and colleagues (2022) manipulated the transgressors’ understanding of punishment (whether or not the transgressor regretted their actions and would like to apologize) as utility-relevant information and transgressors’ suffering (whether or not they suffered from being punished) as retribution-relevant information and tested whether these two manipulations affected participants’ punitive decisions. The authors found an interaction effect between the two factors: when the transgressor understood that their behavior was wrong, participants preferred lighter sentencing decisions, irrespective of suffering. However, when the transgressor did not understand the wrongness of their actions, participants preferred harsher sanctions, especially if transgressor suffering was low. These findings suggest that making the offender understand (“sending a message”) was what participants primarily care about when deciding upon fair punishment, while offender suffering is only of secondary importance.

Further evidence for the “intuitive retributivism” hypothesis is provided by a study by Rehren and Zisman (2022) that replicated another seminal study by Carlsmith (2006). In this study, people’s information search behavior was investigated using behavioral process tracing. Rehren and Zisman (2022) further investigated the cognitive underpinnings of people’s punishment motives by applying a dual-process model of punitive reactions. In sum, they show the robustness of the “intuitive retributivism” hypothesis by replicating Carlsmith (2006). Interestingly, there was no evidence for differences between deliberative and more reflexive decision-making.

The study by Fousiani and Van Prooijen (2022) is the only one in this topical issue that measured people’s punishment motives directly using self-reports. More precisely, participants indicated their motives for punishing an individual suspected of immoralities whose guilt is uncertain. The effect of a suspect’s power and the mediating role of guilt likelihood and a suspect’s recidivism were also investigated. In line with the “intuitive retributivism” hypothesis, individuals indicated that they were motivated to punish the person suspected of immoralities mostly on retributive rather than utilitarian or restorative grounds.

Lastly, Strauß and Bondü (2022) provide rare data on the replicability of the “intuitive retributivism” hypothesis in children. In this study, a sample of elementary school children indicated which responses by teachers they considered appropriate in the case of a norm transgression by a pupil. Crucially, punishment responses were designed to satisfy only one specific punishment motive. Strikingly, results showed that appropriateness ratings of overall retributive and overall consequentialist punishment responses did not differ, contrasting the “intuitive retributivism” hypothesis.

All the studies in this issue were preregistered before data collection. Of course, it is sometimes inevitable that researchers have to deviate from the preregistered procedure. For example, one research group had to change the population from German to US-American adults (Hechler & Kessler, 2022); another group had to switch to a different sample provider (Fousiani & Van Prooijen, 2022); and in one case, data collection could not take place in the lab, as preregistered, due to the COVID-19 pandemic – eventually, children were interviewed over the phone (Strauß & Bondü, 2022). In addition, some analysis plans were modified slightly after the data were collected (e.g., Rehren & Zisman, 2022), for instance, because preregistered follow-up tests were pointless given the results (Molho et al., 2022). Importantly, the preregistered protocols are referred to in each paper, and deviations from the preregistered procedure are reported.

Why Do People Punish? Summarizing Evidence on the “Intuitive Retributivism” Hypothesis

To summarize the effects reported herein not only qualitatively but also quantitatively, we conducted a meta-analysis on the most central effects reflecting the “intuitive retributivism” hypothesis in all studies. More precisely, we extracted the most central effects from all studies that, from our perspective, tested the “intuitive retributivism” hypothesis most critically.¹ The effects extracted from the articles as well as the technical details of the meta-analysis are provided at PsychArchives at https://dx.doi.org/10.23668/psycharchives.4999. Figure 1 shows a forest plot of the observed outcomes and the estimate based on a random-effects model.

As much as the methodological diversity of the studies published in this topical issue reflects the breadth of the approaches reported in the literature, the reported effects reflect the general heterogeneity of two decades of research on the “intuitive retributivism” hypothesis. Still, most studies yielded small but positive effects in favor of the hypothesis, with one study yielding a noticeably small (or negligible) and two studies yielding noticeably large effects.

The negligible effect found by Strauß and Bondü (2022) may be due to the special sample investigated. That is, while most of the existing literature has studied the motivational basis of punishment in adults, Strauß and Bondü (2022) provide rare data on punishment motives in children of approximately nine years of age. Consequently, the results revealed in this research may suggest that, in children, punishment is not predominantly driven by retributivism, as the “intuitive retributivism” hypothesis would suggest. This is much in line with other very recent literature (Marshall et al., 2021; Twardawski & Hilbig, 2020).

The large effect found in the study by Rehren and Zisman (2022) may be explained by the methodological approach chosen. In this study, people’s punishment motives were investigated by analyzing their information search behavior using behavioral process tracing methods. Importantly, this approach forces participants to choose between punishment motives (or at least information associated with these motives) rather than measuring these motives independently. As a result, such an approach juxtaposes punishment motives and, therefore, increases potential differences.

The large effect found in the study by Fousiani and Van Prooijen (2022) is somewhat more surprising, as punishment motives were measured directly. Strikingly, such direct measures yielded rather smaller differences between punishment motives in the past (e.g., Twardawski et al., 2020). One reason for this surprisingly large effect may be that this study measured people’s motives for punishing an individual suspected of immoralities whose guilt is uncertain. Although speculative, it may be that retribution predominates other motives even more strongly when thinking about the punishment of those suspected of immoralities as compared to proven wrongdoers. Or put differently, individuals may endorse the utilitarian value of punishment more strongly in proven wrongdoers, but less so when guilt is uncertain. However, future research is needed to test this hypothesis directly.

In sum, the results of the meta-analysis clearly support the “intuitive retributivism” hypothesis. Although the overall correlation coefficient can only be described as small to medium (Cohen, 1988), it is still larger than the average effect size in social psychological research (r = .21; Richard et al., 2003). Crucially, the majority of papers in this topical issue provide evidence for the notion that punishment is predominantly driven by retribution rather than other punishment motives.

Conclusion

For almost two decades, social psychological literature has suggested that laypeople’s punishment is primarily driven by retributive rather than other (e.g., utilitarian) motives (e.g., Carlsmith, 2006; Carlsmith et al., 2002; Keller et al., 2010). This body of research has led to the consensus that “people are intuitive retributivists” (Carlsmith & Darley, 2008, p. 211). In this topical issue, nine papers report studies that tested this “intuitive retributivism” hypothesis and its boundaries in a diverse set of samples and using a variety of methodological approaches. In general, most studies yielded (partial) support for the hypothesis that laypeople’s punishment is predominantly driven by retribution. Importantly, however, the present studies not only tested the robustness of past results but also extended the literature in several fundamental ways. For example, although not expressed explicitly, most of the previous literature provided evidence for the “intuitive retributivism” hypothesis in third-party punishment. Here, several studies showed that, much like third-party punishment, second-party punishment is also largely driven by a retributive motive. Importantly, this does not mean that people consider punishment to be fair only to the extent that it makes the offender suffer (the proverbial eye-for-an-eye principle; the “lex talionis”). Rather, the findings reported here and elsewhere (e.g., Fischer et al., in press; Funk et al., 2014) suggest that punishment helps close the injustice gap (in the eyes of the punisher) even in the absence of suffering on the part of the offender (Aharoni et al., 2022). Thus, rebalancing the scales of justice can be achieved in many different ways. But it is exactly this rebalancing that people care about when thinking about “punishment”.

That said, this topical issue may not be perceived as the endpoint of research on the motivational basis of punishment but rather as a starting point for new paths of research. The pattern of results shows that study design decisions impact effect estimates. However, as different design aspects were varied simultaneously, we cannot judge which design decisions affect the results, e.g., whether it is the different operationalization of motives, different measurement of punishment, or differences in the sample. In further replication studies, design aspects may be systematically varied to tackle this question. This allows us to identify boundary conditions of our theories in general and the “intuitive retributivism” hypothesis in particular. Furthermore, all samples reported here were collected in so-called WEIRD countries (Western Educated Industrialized Rich Democratic; Henrich et al., 2010). In fact, to the best of our knowledge, there is no research testing the replicability of the “intuitive retributivism” hypothesis in other countries or cultures. Likewise, more contemporary methodological approaches (e.g., interactions in virtual reality labs, analyses of big data from social media) may be used to further test the replicability of the “intuitive retributivism” hypothesis and, thus, broaden our understanding of laypeople’s punishment motives. These and related questions may be subject to future research, and we are confident that this topical issue provides a fruitful basis for this endeavor.