Although he persuasively argues that field evidence is needed to support a “wide” reading of costly punishment experiments, Guala errs in his assessment of the current state of non-laboratory evidence concerning strong reciprocity. First, Guala holds that the dearth of punishment-induced cooperation in the anthropological record undermines a wide reading of costly punishment experiments (target article, Abstract). Second, Guala holds that “there are no natural field experiments on costly punishment” (sect. 7, para. 5). Both of these claims are problematic. Guala's first claim fails to appreciate the strategic logic of costly punishment, which predicts that punishment will occur rarely if it proves effective at fostering cooperation. Guala's second claim results from an incomplete reading of the past literature, which, as I show here, contains at least one natural field experiment concerning costly punishment. Once these problems with Guala's argument are addressed, it becomes clear that field evidence supports a wide reading of costly punishment experiments.
Guala's first claim – which holds that “there is no evidence that cooperation in the small egalitarian societies studied by anthropologists is enforced by means of costly punishment” (Abstract) – seems especially damning to a wide reading of costly punishment experiments. If costly punishment does not exist outside the laboratory, then it cannot support cooperation in the everyday world. Albeit intuitive, that reasoning fails to consider the strategic logic of costly punishment. If punishment makes free-riding more expensive than cooperation, and if a population contains myriad strong reciprocators willing to punish free-riders, then strategic defectors will opt to cooperate and strong reciprocators will have no need to engage in costly punishment. In such equilibrium, observers would not witness any actual costly punishment: cooperation induced by costly punishment and cooperation produced by other mechanisms would appear “observationally equivalent” (for a general discussion, see Weingast & MoranReference Weingast and Moran1983, p. 767, fn. 2). Guala ignores this implication of costly punishment's strategic logic, and in so doing, he fails to realize that the absence of costly punishment from the anthropological record remains consistent with a wide reading of costly punishment experiments.
Also, by ignoring the strategic logic of costly punishment, Guala's point about the importance of natural field experiments carries less force than it should (sect. 6). The strategic logic of costly punishment indicates that true natural field experiments are crucial to test strong reciprocity theory outside the laboratory. Without exogenous variation in either (1) the opportunity to engage in costly punishment (which would allow defectors to demonstrate that they will free-ride if sanctioning is impossible) or (2) the occurrence of free-riding (which would allow strong reciprocators to demonstrate their willingness to punish), it is impossible to assess whether cooperation outside the laboratory results from costly punishment or some other mechanism. In light of this need for experiments, Guala casts a threatening shadow over strong reciprocity theory by claiming that “there are no natural field experiments on costly punishment” (sect. 7, para. 5).
Guala's claim, however, is incorrect. At least one natural field experiment provides evidence concerning the influence of costly punishment on cooperation. In a study of voter turnout, Gerber et al. (Reference Gerber, Green and Larimer2008) told a randomly selected group of citizens that experimenters would send information about their voter turnout record – along with their names and addresses – to other citizens in the aftermath of a forthcoming election. The revelation of this information raised the possibility that fellow citizens could identify and sanction individuals who did not engage in personally costly – yet group benefiting – turnout. In so doing, the experimental manipulation exogenously influenced individuals' awareness of punishment possibilities, thus creating conditions in which experimenters could examine whether the introduction of costly sanctioning opportunities would, in fact, increase cooperation. Indeed, consistent with laboratory evidence showing immediate increases in cooperation when experimenters introduce the opportunity to punish (see Fehr & Gächter Reference Fehr and Gächter2000a, p. 986, Figs. 1A and 1B; Fehr & Gächter Reference Fehr and Gächter2002, p. 138, Figs. 2A and 2B), the mere possibility of costly punishment increased cooperative voter turnout by roughly 8 percentage points, which corresponded to an approximate 27% change from the control group's turnout rate (Gerber et al. Reference Gerber, Green and Larimer2008). Although it was designed to illuminate the mechanisms underlying electoral participation, the field experiment conducted by Gerber et al. (Reference Gerber, Green and Larimer2008) bears directly on costly punishment and it indicates that uncoordinated costly punishment can increase cooperation outside the lab.
Other studies that combine observational and experimental data provide complementary evidence, while also showing an additional means to test the plausibility of laboratory findings concerning costly punishment. For instance, Smirnov et al. (Reference Smirnov, Dawes, Fowler, Johnson and McElreath2010) show that individuals who incur the costs of cooperation and punishment in laboratory public goods games exhibit a greater likelihood of engaging in partisan collective action in their daily lives. Given that maintaining political parties represents a prototypical public goods problem (Aldrich Reference Aldrich1995, p. 31), these findings imply that organizations engaged in non-laboratory cooperative enterprises may succeed – at least in some significant part – precisely because they are populated by strong reciprocators (Smirnov et al. Reference Smirnov, Dawes, Fowler, Johnson and McElreath2010). Not only do those findings offer insight into non-laboratory cooperation, but they also illustrate another means by which researchers can examine whether behavior observed in laboratory costly punishment experiments corresponds with conduct outside the lab.
Ultimately, such concerns about the external validity of costly punishment experiments are important and Guala deserves credit for voicing them. Yet, in the end, those concerns are less pressing than Guala claims. Contrary to his suggestions, scientists should not expect the anthropological record to contain examples of costly punishment; the strategic logic of costly punishment holds that no such examples will exist if punishment truly deters free-riding. Nor should scientists doubt a wide reading of costly punishment experiments due to a lack of natural field experiments; at least one such experiment exists and it supports the hypothesis that costly punishment facilitates cooperation.
Although he persuasively argues that field evidence is needed to support a “wide” reading of costly punishment experiments, Guala errs in his assessment of the current state of non-laboratory evidence concerning strong reciprocity. First, Guala holds that the dearth of punishment-induced cooperation in the anthropological record undermines a wide reading of costly punishment experiments (target article, Abstract). Second, Guala holds that “there are no natural field experiments on costly punishment” (sect. 7, para. 5). Both of these claims are problematic. Guala's first claim fails to appreciate the strategic logic of costly punishment, which predicts that punishment will occur rarely if it proves effective at fostering cooperation. Guala's second claim results from an incomplete reading of the past literature, which, as I show here, contains at least one natural field experiment concerning costly punishment. Once these problems with Guala's argument are addressed, it becomes clear that field evidence supports a wide reading of costly punishment experiments.
Guala's first claim – which holds that “there is no evidence that cooperation in the small egalitarian societies studied by anthropologists is enforced by means of costly punishment” (Abstract) – seems especially damning to a wide reading of costly punishment experiments. If costly punishment does not exist outside the laboratory, then it cannot support cooperation in the everyday world. Albeit intuitive, that reasoning fails to consider the strategic logic of costly punishment. If punishment makes free-riding more expensive than cooperation, and if a population contains myriad strong reciprocators willing to punish free-riders, then strategic defectors will opt to cooperate and strong reciprocators will have no need to engage in costly punishment. In such equilibrium, observers would not witness any actual costly punishment: cooperation induced by costly punishment and cooperation produced by other mechanisms would appear “observationally equivalent” (for a general discussion, see Weingast & MoranReference Weingast and Moran1983, p. 767, fn. 2). Guala ignores this implication of costly punishment's strategic logic, and in so doing, he fails to realize that the absence of costly punishment from the anthropological record remains consistent with a wide reading of costly punishment experiments.
Also, by ignoring the strategic logic of costly punishment, Guala's point about the importance of natural field experiments carries less force than it should (sect. 6). The strategic logic of costly punishment indicates that true natural field experiments are crucial to test strong reciprocity theory outside the laboratory. Without exogenous variation in either (1) the opportunity to engage in costly punishment (which would allow defectors to demonstrate that they will free-ride if sanctioning is impossible) or (2) the occurrence of free-riding (which would allow strong reciprocators to demonstrate their willingness to punish), it is impossible to assess whether cooperation outside the laboratory results from costly punishment or some other mechanism. In light of this need for experiments, Guala casts a threatening shadow over strong reciprocity theory by claiming that “there are no natural field experiments on costly punishment” (sect. 7, para. 5).
Guala's claim, however, is incorrect. At least one natural field experiment provides evidence concerning the influence of costly punishment on cooperation. In a study of voter turnout, Gerber et al. (Reference Gerber, Green and Larimer2008) told a randomly selected group of citizens that experimenters would send information about their voter turnout record – along with their names and addresses – to other citizens in the aftermath of a forthcoming election. The revelation of this information raised the possibility that fellow citizens could identify and sanction individuals who did not engage in personally costly – yet group benefiting – turnout. In so doing, the experimental manipulation exogenously influenced individuals' awareness of punishment possibilities, thus creating conditions in which experimenters could examine whether the introduction of costly sanctioning opportunities would, in fact, increase cooperation. Indeed, consistent with laboratory evidence showing immediate increases in cooperation when experimenters introduce the opportunity to punish (see Fehr & Gächter Reference Fehr and Gächter2000a, p. 986, Figs. 1A and 1B; Fehr & Gächter Reference Fehr and Gächter2002, p. 138, Figs. 2A and 2B), the mere possibility of costly punishment increased cooperative voter turnout by roughly 8 percentage points, which corresponded to an approximate 27% change from the control group's turnout rate (Gerber et al. Reference Gerber, Green and Larimer2008). Although it was designed to illuminate the mechanisms underlying electoral participation, the field experiment conducted by Gerber et al. (Reference Gerber, Green and Larimer2008) bears directly on costly punishment and it indicates that uncoordinated costly punishment can increase cooperation outside the lab.
Other studies that combine observational and experimental data provide complementary evidence, while also showing an additional means to test the plausibility of laboratory findings concerning costly punishment. For instance, Smirnov et al. (Reference Smirnov, Dawes, Fowler, Johnson and McElreath2010) show that individuals who incur the costs of cooperation and punishment in laboratory public goods games exhibit a greater likelihood of engaging in partisan collective action in their daily lives. Given that maintaining political parties represents a prototypical public goods problem (Aldrich Reference Aldrich1995, p. 31), these findings imply that organizations engaged in non-laboratory cooperative enterprises may succeed – at least in some significant part – precisely because they are populated by strong reciprocators (Smirnov et al. Reference Smirnov, Dawes, Fowler, Johnson and McElreath2010). Not only do those findings offer insight into non-laboratory cooperation, but they also illustrate another means by which researchers can examine whether behavior observed in laboratory costly punishment experiments corresponds with conduct outside the lab.
Ultimately, such concerns about the external validity of costly punishment experiments are important and Guala deserves credit for voicing them. Yet, in the end, those concerns are less pressing than Guala claims. Contrary to his suggestions, scientists should not expect the anthropological record to contain examples of costly punishment; the strategic logic of costly punishment holds that no such examples will exist if punishment truly deters free-riding. Nor should scientists doubt a wide reading of costly punishment experiments due to a lack of natural field experiments; at least one such experiment exists and it supports the hypothesis that costly punishment facilitates cooperation.