From the lab to the “wild”: Why we need theory
Social interactions during lab-based studies differ substantially from interactions in the “wild.” According to interdependence theory (Kelley et al. Reference Kelley, Holmes, Kerr, Reis, Rusbult and Van Lange2003; Van Lange & Rusbult Reference Van Lange, Rusbult, Van Lange, Kruglanski and Higgins2012), such differences in social interactions are likely rooted in profound differences in the situations that people face. What might be key differences in interdependence between situations in the lab and the wild? One important difference is the one between single-trial versus repeated interaction (i.e., temporal structure). In small communities, people typically know each other, have a history of social interaction experiences, and may have formed strong attachments to (some) other members of their community. In contrast, in the lab, researchers have almost exclusively relied on strangers. Moreover, there is almost always an interdependent future ahead of people in the wild, which is not typical for lab research.
Another situational feature of interdependence theory is information availability. There is little doubt that the “rules of the game” are clearer in the lab than in the wild. For example, in the lab, it is often specified how costly an action is for a person, but in the real life such costs are often less tangible or known. Psychologically, the lack of information about others' preferences, magnitude of costs, and the like, challenge important processes such as those linked to risk, uncertainty, and interpersonal trust.
The next question is, of course, this: In what ways might temporal structure and information availability be important to our understanding whether costly punishments are effective, or why they are used only sporadically in the field? Temporal structure is important because it is linked to the history of social interaction experiences as well as the anticipated future of social interaction. The history is important to relationship development, including growth or decline in feelings of attachment, and trust, along with its manifestations and embodiments (e.g., IJzerman & Koole Reference IJzerman and Koole2011). The future is important because it may trigger the “shadow of the future” (Axelrod Reference Axelrod1984); that is, an implicit or explicit mindset that tends to promote cooperation, so that punishment is less often called for (Van Lange et al. Reference Van Lange, Klapwijk and Van Munster2011). In light of the future, what might people do? It seems reasonable to expect that people first start to communicate, or to gossip, as Guala notes, before seeking punishment in any more material, or tangible sense.
Information availability may also affect punishments of non-cooperators. Costly punishment may be less likely to be used if one is not completely confident that another person intentionally acted as a non-cooperator. For example, in everyday life, unintended errors (called noise) are quite commonplace, in that external barriers prevent a person from translating his or her cooperative intentions into cooperative action (e.g., Van Lange et al. Reference Van Lange, Ouwerkerk and Tazelaar2002). Given that people at some level are likely to realize that particular outcomes are not intentional, it is likely that people might give each other the benefit of the doubt. And in light of incomplete information, people may seek information about the intentionality of an action instead of immediately punishing a non-cooperator.
From the lab to the “wild”: Why we need meta-analysis
While theory provides the concepts and logic, a meta-analysis provides the comprehensive databank – from the lab and the field – that should allow for rigorous conclusions. In this regard, it is interesting that Guala's narrative review of the lab studies reaches the conclusion that “‘costly’ punishment works only if it costs relatively little” (sect. 12, para. 3). However, in a recent meta-analysis of incentives and cooperation including 187 effect sizes from both psychology and economic studies, Balliet et al. (Reference Balliet, Moulder and Van Lange2011) found that costly punishments (and rewards) were more effective at encouraging cooperation, compared to when punishments were free to administer. They interpreted their findings in terms of interdependence theory (Kelley et al. Reference Kelley, Holmes, Kerr, Reis, Rusbult and Van Lange2003; Van Lange & Rusbult Reference Van Lange, Rusbult, Van Lange, Kruglanski and Higgins2012), noting that people translate the costs of administering punishment as a strong indicator of benign intent. Hence, the conclusion about the cost of punishment reached by Guala may be inaccurate because of a restricted survey of the literature.
Concluding remarks
Interdependence theory provides a taxonomy of situations that can be fruitfully used to explicate the key differences in situational structure between the lab and the field. This theory provides much needed insight into the key elements of situations that affect social interaction processes in dyads and groups. We have suggested the relevance of temporal structure and information availability for understanding the discrepancy between lab and field research on costly punishment, although other features (e.g., asymmetries in dependence) might be crucial as well.
We hope that field data about the topic of punishment (and reward) and cooperation grow, and that their operationalizations match those of the lab. Likewise, we hope that researchers in the lab focus more strongly on key features of interdependence that characterize social dilemmas outside of the laboratory. Such efforts should make future meta-analyses even more informative. As Kurt Lewin noted, there is nothing more practical than a good theory (Lewin Reference Lewin1952, p. 169). Perhaps we may add: We need good data as well, as good data ultimately determine the contribution of a meta-analysis.
As shown in experimental research in laboratories, introducing a system whereby people can punish non-cooperators at a cost to self effectively promotes cooperation (e.g., Fehr & Gächter Reference Fehr and Gächter2002; for a recent meta-analysis, see Balliet et al. Reference Balliet, Moulder and Van Lange2011). Such lab-based research findings are now being challenged in Guala's provocative target article, which reviews evidence from anthropology (and some other disciplines) to reach the conclusion that people may not engage in costly punishment to encourage cooperation outside the laboratory. Although Guala identifies an important discrepancy between lab-based and field data, we argue that this discrepancy, at least as it currently exists, is less problematic than the target article suggests. We emphasize two solutions to understanding this discrepancy: namely, theory and meta-analysis.
From the lab to the “wild”: Why we need theory
Social interactions during lab-based studies differ substantially from interactions in the “wild.” According to interdependence theory (Kelley et al. Reference Kelley, Holmes, Kerr, Reis, Rusbult and Van Lange2003; Van Lange & Rusbult Reference Van Lange, Rusbult, Van Lange, Kruglanski and Higgins2012), such differences in social interactions are likely rooted in profound differences in the situations that people face. What might be key differences in interdependence between situations in the lab and the wild? One important difference is the one between single-trial versus repeated interaction (i.e., temporal structure). In small communities, people typically know each other, have a history of social interaction experiences, and may have formed strong attachments to (some) other members of their community. In contrast, in the lab, researchers have almost exclusively relied on strangers. Moreover, there is almost always an interdependent future ahead of people in the wild, which is not typical for lab research.
Another situational feature of interdependence theory is information availability. There is little doubt that the “rules of the game” are clearer in the lab than in the wild. For example, in the lab, it is often specified how costly an action is for a person, but in the real life such costs are often less tangible or known. Psychologically, the lack of information about others' preferences, magnitude of costs, and the like, challenge important processes such as those linked to risk, uncertainty, and interpersonal trust.
The next question is, of course, this: In what ways might temporal structure and information availability be important to our understanding whether costly punishments are effective, or why they are used only sporadically in the field? Temporal structure is important because it is linked to the history of social interaction experiences as well as the anticipated future of social interaction. The history is important to relationship development, including growth or decline in feelings of attachment, and trust, along with its manifestations and embodiments (e.g., IJzerman & Koole Reference IJzerman and Koole2011). The future is important because it may trigger the “shadow of the future” (Axelrod Reference Axelrod1984); that is, an implicit or explicit mindset that tends to promote cooperation, so that punishment is less often called for (Van Lange et al. Reference Van Lange, Klapwijk and Van Munster2011). In light of the future, what might people do? It seems reasonable to expect that people first start to communicate, or to gossip, as Guala notes, before seeking punishment in any more material, or tangible sense.
Information availability may also affect punishments of non-cooperators. Costly punishment may be less likely to be used if one is not completely confident that another person intentionally acted as a non-cooperator. For example, in everyday life, unintended errors (called noise) are quite commonplace, in that external barriers prevent a person from translating his or her cooperative intentions into cooperative action (e.g., Van Lange et al. Reference Van Lange, Ouwerkerk and Tazelaar2002). Given that people at some level are likely to realize that particular outcomes are not intentional, it is likely that people might give each other the benefit of the doubt. And in light of incomplete information, people may seek information about the intentionality of an action instead of immediately punishing a non-cooperator.
From the lab to the “wild”: Why we need meta-analysis
While theory provides the concepts and logic, a meta-analysis provides the comprehensive databank – from the lab and the field – that should allow for rigorous conclusions. In this regard, it is interesting that Guala's narrative review of the lab studies reaches the conclusion that “‘costly’ punishment works only if it costs relatively little” (sect. 12, para. 3). However, in a recent meta-analysis of incentives and cooperation including 187 effect sizes from both psychology and economic studies, Balliet et al. (Reference Balliet, Moulder and Van Lange2011) found that costly punishments (and rewards) were more effective at encouraging cooperation, compared to when punishments were free to administer. They interpreted their findings in terms of interdependence theory (Kelley et al. Reference Kelley, Holmes, Kerr, Reis, Rusbult and Van Lange2003; Van Lange & Rusbult Reference Van Lange, Rusbult, Van Lange, Kruglanski and Higgins2012), noting that people translate the costs of administering punishment as a strong indicator of benign intent. Hence, the conclusion about the cost of punishment reached by Guala may be inaccurate because of a restricted survey of the literature.
Concluding remarks
Interdependence theory provides a taxonomy of situations that can be fruitfully used to explicate the key differences in situational structure between the lab and the field. This theory provides much needed insight into the key elements of situations that affect social interaction processes in dyads and groups. We have suggested the relevance of temporal structure and information availability for understanding the discrepancy between lab and field research on costly punishment, although other features (e.g., asymmetries in dependence) might be crucial as well.
We hope that field data about the topic of punishment (and reward) and cooperation grow, and that their operationalizations match those of the lab. Likewise, we hope that researchers in the lab focus more strongly on key features of interdependence that characterize social dilemmas outside of the laboratory. Such efforts should make future meta-analyses even more informative. As Kurt Lewin noted, there is nothing more practical than a good theory (Lewin Reference Lewin1952, p. 169). Perhaps we may add: We need good data as well, as good data ultimately determine the contribution of a meta-analysis.