Guala reviews a number of interesting field studies that speak against the importance of punishment in maintaining cooperation. This is important because there is an abundance of laboratory research on punishment and cooperation which has outstripped the research in real-world settings. Underlying much of Guala's discussion of reciprocity and punishment, however, there lies confusion over proximate causation and ultimate causation. Confusion over these levels of analysis is not only present in Guala's target article, but is endemic to the entire field of cooperation and is particularly pronounced in the discussion of “strong reciprocity.” This weakens Guala's arguments. In particular, it results in unwarranted statements against so-called weak reciprocity. As such, this topic requires clarification.
Any behaviour, including cooperation and punishment, can be explained at four different levels of analysis (Tinbergen Reference Tinbergen1968). Proximate causes include: (1) the psychological mechanisms that trigger behaviour (e.g., emotions, cognitions); and (2) the developmental processes that cause those psychological mechanisms to arise within an individual's lifetime (e.g., “innate” behaviour, learning, internalization of cultural norms). Ultimate causes include: (3) the evolutionary forces (e.g., reciprocity, mutualism, costly signalling) that result in those psychological mechanisms existing instead of other possible psychologies; and (4) the evolutionary history of those mechanisms and when they arose in our lineage (e.g., unique to humans, shared with other primates). These four levels of analysis – mechanism, development, function, and phylogeny – are complementary, not mutually exclusive. A complete explanation of any phenomenon requires an answer at each level.
An example can help clarify the proximate and ultimate causes of cooperative behaviour. Suppose that I genuinely value your welfare and I help you without any ulterior motives. If my action causes you to genuinely care about me, you will be more likely to help me when I need it, even when you anticipate no benefits for doing so. If I happen to find out, then your actions will cause me to value your welfare more and help you more often, and so on. The reciprocity in this example is not “weak”: both of us unselfishly reciprocate “altruistic” acts. Both of us do benefit from helping each other, but neither one intended to benefit, and neither of us requires any foresight of the consequences. Helping can be altruistic from a proximate psychological perspective, but from an ultimate (functional) perspective it is advantageous to possess such a psychology. Thus, contrary to Guala's assumption, biologists do not assume psychological self-interest. To paraphrase Dawkins (Reference Dawkins1976/2006): The genes are selfish, but this doesn't mean the person is. One can make a similar argument with punishment: I may punish you because I am angry (proximate cause), and this may result in me receiving more future cooperation from you (potential ultimate cause of punitive sentiment), but this does not mean that my punishment was motivated by a desire for your cooperation.
Guala uses terms like “strong” and “weak” reciprocity, which are often misleading because they often conflate the proximate psychological mechanisms with the ultimate functional reasons for why those psychological mechanisms exist (Barclay Reference Barclay2010; West et al. Reference West, Griffin and Gardner2007b). By itself, “strong reciprocity” is merely a description of behaviour, that is, the supposed tendency of people to cooperate, reward cooperators, and punish cooperators, even when there are no apparent benefits for doing so. The goal is to discover – at all levels among levels of analysis – why this tendency exists (if indeed it does). So-called theories of “weak reciprocity” are often theories about the ultimate function of cooperative and punitive sentiment, not theories about what specifically that sentiment is or how it develops. People possess certain emotions and psychological mechanisms which are predicted to be adaptive on average outside the laboratory; for example, if being nice invites reciprocation. People bring these psychological mechanisms with them into the laboratory, where the behaviour produced may or may not still be adaptive on average (Barclay, Reference Barclay and Roberts2011; West et al. Reference West, El and Gardner2011). “Maladaptive” behaviour can persist despite repeated anonymous encounters, as long as the same proximate psychological mechanisms are repeatedly triggered (e.g., anger, desire for fairness, empathy). However, this would say little about the ultimate function that those mechanisms serve outside the laboratory. Too much ink has been spilled by researchers who do not realize that their colleagues are simply addressing a different level of analysis.
On a completely different note, Guala makes a useful distinction between wide and narrow readings of the experimental evidence, and what each reading implies. Wide interpretations can clearly be taken too far: If punishment (or any other phenomenon) supports cooperation in the lab, it does not necessarily mean that this is what supports it outside the lab. However, I would caution against hasty abandonment of such wide interpretations. Sometimes laboratory experiments use controlled conditions to test whether a proposed mechanism could support punishment. At other times, such experiments test the validity of theories of human behaviour (Mook Reference Mook1983): If a predicted phenomenon cannot be found in the lab under ideal controlled conditions, then we must either reject or revise any theory that relies on that phenomenon (see, e.g., the lack of punishment towards non-punishers in Kiyonari & Barclay Reference Kiyonari and Barclay2008). If successful, do these findings need confirmatory non-laboratory observations with real-world phenomena? Absolutely. Convergent evidence is crucial in all scientific enterprises, and the laboratory and the field have their own respective strengths and weaknesses. As such, we should all strongly support the call for collaborations across disciplines and between the lab and the field. Guala's target article has clearly shown that the punishment literature needs more of this, and for that it should be commended.
Guala reviews a number of interesting field studies that speak against the importance of punishment in maintaining cooperation. This is important because there is an abundance of laboratory research on punishment and cooperation which has outstripped the research in real-world settings. Underlying much of Guala's discussion of reciprocity and punishment, however, there lies confusion over proximate causation and ultimate causation. Confusion over these levels of analysis is not only present in Guala's target article, but is endemic to the entire field of cooperation and is particularly pronounced in the discussion of “strong reciprocity.” This weakens Guala's arguments. In particular, it results in unwarranted statements against so-called weak reciprocity. As such, this topic requires clarification.
Any behaviour, including cooperation and punishment, can be explained at four different levels of analysis (Tinbergen Reference Tinbergen1968). Proximate causes include: (1) the psychological mechanisms that trigger behaviour (e.g., emotions, cognitions); and (2) the developmental processes that cause those psychological mechanisms to arise within an individual's lifetime (e.g., “innate” behaviour, learning, internalization of cultural norms). Ultimate causes include: (3) the evolutionary forces (e.g., reciprocity, mutualism, costly signalling) that result in those psychological mechanisms existing instead of other possible psychologies; and (4) the evolutionary history of those mechanisms and when they arose in our lineage (e.g., unique to humans, shared with other primates). These four levels of analysis – mechanism, development, function, and phylogeny – are complementary, not mutually exclusive. A complete explanation of any phenomenon requires an answer at each level.
An example can help clarify the proximate and ultimate causes of cooperative behaviour. Suppose that I genuinely value your welfare and I help you without any ulterior motives. If my action causes you to genuinely care about me, you will be more likely to help me when I need it, even when you anticipate no benefits for doing so. If I happen to find out, then your actions will cause me to value your welfare more and help you more often, and so on. The reciprocity in this example is not “weak”: both of us unselfishly reciprocate “altruistic” acts. Both of us do benefit from helping each other, but neither one intended to benefit, and neither of us requires any foresight of the consequences. Helping can be altruistic from a proximate psychological perspective, but from an ultimate (functional) perspective it is advantageous to possess such a psychology. Thus, contrary to Guala's assumption, biologists do not assume psychological self-interest. To paraphrase Dawkins (Reference Dawkins1976/2006): The genes are selfish, but this doesn't mean the person is. One can make a similar argument with punishment: I may punish you because I am angry (proximate cause), and this may result in me receiving more future cooperation from you (potential ultimate cause of punitive sentiment), but this does not mean that my punishment was motivated by a desire for your cooperation.
Guala uses terms like “strong” and “weak” reciprocity, which are often misleading because they often conflate the proximate psychological mechanisms with the ultimate functional reasons for why those psychological mechanisms exist (Barclay Reference Barclay2010; West et al. Reference West, Griffin and Gardner2007b). By itself, “strong reciprocity” is merely a description of behaviour, that is, the supposed tendency of people to cooperate, reward cooperators, and punish cooperators, even when there are no apparent benefits for doing so. The goal is to discover – at all levels among levels of analysis – why this tendency exists (if indeed it does). So-called theories of “weak reciprocity” are often theories about the ultimate function of cooperative and punitive sentiment, not theories about what specifically that sentiment is or how it develops. People possess certain emotions and psychological mechanisms which are predicted to be adaptive on average outside the laboratory; for example, if being nice invites reciprocation. People bring these psychological mechanisms with them into the laboratory, where the behaviour produced may or may not still be adaptive on average (Barclay, Reference Barclay and Roberts2011; West et al. Reference West, El and Gardner2011). “Maladaptive” behaviour can persist despite repeated anonymous encounters, as long as the same proximate psychological mechanisms are repeatedly triggered (e.g., anger, desire for fairness, empathy). However, this would say little about the ultimate function that those mechanisms serve outside the laboratory. Too much ink has been spilled by researchers who do not realize that their colleagues are simply addressing a different level of analysis.
On a completely different note, Guala makes a useful distinction between wide and narrow readings of the experimental evidence, and what each reading implies. Wide interpretations can clearly be taken too far: If punishment (or any other phenomenon) supports cooperation in the lab, it does not necessarily mean that this is what supports it outside the lab. However, I would caution against hasty abandonment of such wide interpretations. Sometimes laboratory experiments use controlled conditions to test whether a proposed mechanism could support punishment. At other times, such experiments test the validity of theories of human behaviour (Mook Reference Mook1983): If a predicted phenomenon cannot be found in the lab under ideal controlled conditions, then we must either reject or revise any theory that relies on that phenomenon (see, e.g., the lack of punishment towards non-punishers in Kiyonari & Barclay Reference Kiyonari and Barclay2008). If successful, do these findings need confirmatory non-laboratory observations with real-world phenomena? Absolutely. Convergent evidence is crucial in all scientific enterprises, and the laboratory and the field have their own respective strengths and weaknesses. As such, we should all strongly support the call for collaborations across disciplines and between the lab and the field. Guala's target article has clearly shown that the punishment literature needs more of this, and for that it should be commended.
ACKNOWLEDGMENTS
I thank Daniel Krupp for discussions and the Social Sciences and Humanities Research Council of Canada (SSHRC) for support.