We agree with Guala that it can be difficult to draw conclusions about human evolution from highly controlled experimental games. Controlling any and all third variables facilitates replication and repetition, enabling comparison of behavior across experiments (Guala Reference Guala2005). However, striving for internal validity introduces a double-edged sword: Economic games provide insight, but they present only a rough approximation of the real world. Strong reciprocity arguments often strive to connect game play to real life by citing anecdotal evidence. Nevertheless, though we endorse caution in interpreting experiments, Guala himself overlooks the incorporation of real-world aspects into recent field-based economic games. This research allows greater insight into the societies under investigation. Moreover, we take issue with both Guala's homogenizing account of group selection theories and his failure to acknowledge variability across subsistence groups. That said, we believe the present article should spur strong reciprocity theorists to further explore the variable social preferences exhibited by participants.

A narrow interpretation of experimental economic games – an uncontroversial reading of the evidence, as Guala notes – suggests that “punishment mechanisms are useful methodological devices to observe social preferences” (sect. 5, para. 2, italics in original). We agree. These social preferences have been sometimes termed “informal norms,” including norms of fairness and reciprocity (Guala Reference Guala2008). For example, Western participants are annoyed and often angry if another participant has a larger net gain than they do (Dawes et al. Reference Dawes, Fowler, Johnson, McElreath and Smirnov2007; Fehr & Gächter Reference Fehr and Gächter2002). If internal validity is rigorously sought within experiments and maintained across experiments via replication, we can expect any differences in game play to correspond to differences in social preferences applied by participants to the experimental context. The issue with external validity arises because game play, by virtue of experimental control, is far-removed from the real-life situations strong reciprocity theorists seek to explain.

We agree that the simple design of many games necessitates caution in interpretation. We do not mean that all economic games are overly simplistic, but that it is difficult to make inferences without control groups of sorts. Many experiments cited by strong reciprocity theorists do not allow for coalition formation, reputation building, or less expensive punishment options. Rockenbach and Milinski (Reference Rockenbach and Milinski2006) found that reputation formation matters: Public goods contributions were greater when costly punishment and indirect reciprocity (i.e., withholding cooperation) were united with reputation building, and participants preferred to join these groups. Similarly, Jacquet et al. (Reference Jacquet, Hauert, Traulsen and Milinski2011) have demonstrated that both negative and positive reputational consequences external to the game context enhance cooperation within the game context. Egas and Riedl (Reference Egas and Riedl2008) found that low-cost and high-impact punishment best promotes cooperation. These results support the idea that costly punishment is probably not as common when “the full menu of strategies” (target article, sect. 6, para. 3) is available.

Guala's primary concern is the extent to which economic games reflect punishment mechanisms “in the wild.” Though in the past Guala (Reference Guala2008) applauded the MacArthur Foundation–sponsored Economic Man studies, in the present article he emphasizes that the incidental introduction of cultural practice by some researchers (Henrich et al. Reference Henrich, Boyd, Bowles, Camerer, Fehr, Gintis, McElreath, Alvard, Barr, Ensminger, Henrich, Hill, Gil-White, Gurven, Marlowe, Patton and Tracer2005, Table 4) and participants (Henrich et al. Reference Henrich, Boyd, Bowles, Camerer, Fehr, Gintis, McElreath, Alvard, Barr, Ensminger, Henrich, Hill, Gil-White, Gurven, Marlowe, Patton and Tracer2005, sect. 8) is not equivalent to an experiment designed to reflect on the particular population of study. Guala is not the only observer to raise concerns about the external validity of field-based games such as Economic Man (e.g., Gurven & Winking Reference Gurven and Winking2008); however, he overlooks more recent efforts to bring external validity to economic games in the field context. Notable recent field studies have endeavored to match games to context, and to derive clear insights about costly monitoring and punishment within a particular cultural group (see Jack Reference Jack2009; Lamba & Mace Reference Lamba and Mace2010; Rustagi et al. Reference Rustagi, Engel and Kosfeld2010).

Despite the strengths of the present review, Guala risks the same pitfall for which he criticizes others: over-generalization. His failure to differentiate among theories of (what he terms) “group selection” does a disservice to the understanding of this area of study. There is a significant difference between biological group selection and gene-culture coevolution (for discussion, see West et al. Reference West, El and Gardner2011). By describing propensities to internalize norms as an aspect of our innate psychology and explaining the cooperation-enhancement of some norms as the product of cultural group selection, gene-culture coevolution theory affords greater variability across groups than does biological group selection theory. While important, these distinctions are admittedly sometimes obscured in the literature, even though connections can be separately drawn between biological group selection and strong reciprocity, and between gene-culture coevolution and strong reciprocity (e.g., Fehr et al. Reference Fehr, Fischbacher and Gächter2002).

The above distinctions are important because, at the empirical level, Guala provides a simplified view of small-scale societies that minimizes variation among them. Guala distills the variety of punishment behaviors outlined in Boehm's (Reference Boehm1999) research, drawing generalizations about homicide in hunter-gatherers, among other things. Boehm (Reference Boehm1999) himself reports that a dearth of punishment data required him to use unsystematic methods of sampling. Today, better archived ethnographic materials afford more systematic gleaning of examples of punishment (though the cases themselves remain anecdotal). Additionally, by dwelling on fission-fusion as a conflict management strategy, Guala overlooks ecological variation that influences the availability of this strategy. For example, 25% of hunter-gatherers in a sample of 340 societies are actually sedentary (Marlowe Reference Marlowe2005), making fission a less ready solution for conflict.

Guala's review of negative strong reciprocity provides a useful platform for subsequent work. We would like to see more even-handed treatment of both the relevant theories and the available ethnographic data. That said, we agree that, regardless of their cultural group, participants face a contrived social situation in economic game experiments. Investigators need to focus on the preferences participants bring to experimental games, including (1) explaining the origins of these preferences, (2) understanding how they manifest in real-world situations, and (3) accounting for individual- and group-level differences in preferences.