Guala engagingly criticizes the claim that strong reciprocity (SR) explains human cooperation beyond the limitations of classic or weak forms of reciprocity (WR) (direct and indirect reciprocity). For advocates of SR, its unique success comes from two distinctive characteristics: It is biologically altruistic, and it involves normative motivation. Guala shows convincingly that there is no evidence of a biologically altruistic form of human reciprocity outside the lab. I shall argue, furthermore, that normative motivation is present in weak reciprocity as well. In the end, SR is more similar to weak reciprocity than its advocates would admit. Nonetheless, there is some novelty to SR, although of a more modest nature.
SR has four properties as initially introduced: (1) In contrast to withdrawing cooperation, SR entails a net cost for the strong reciprocator, both in the short and the long term and therefore evolves through group selection; (2) it inflicts a direct cost on defectors; (3) repeated Public Goods games (PGGs; i.e., n-person Prisoner's Dilemmas [PDs]) are its proper domain; (4) it involves norms as socio-psychological mechanisms. Of these four properties, Guala shows that the first one is an artefact of lab experiments, with no real counterpart in the field. However, the other three properties are instantiated often enough in human institutions.
Properties (2) and (3) alone justify a distinction between SR and WR. In PGGs (repeated n-person PDs where n ≫ 2) cooperators cannot discipline defectors by withdrawing contributions in subsequent periods, as in direct and indirect reciprocity (Fehr & Gächter Reference Fehr and Gächter2000a; Ledyard Reference Ledyard, Kagel and Roth1995). Therefore, a second stage of the game is required, designed as a two-person game, where cooperators target and inflict costs on individual defectors. This two-stage structure is common to both the lab and the field. However, punishers in the lab pay a fee, usually less than the cost inflicted on the punished. Because retaliation against punishers is possible in the field, there is some reason to view this as unrealistic. But far more common in the real world is that humans device institutions to compensate punishers for their costs. Tax officials fining tax evaders have usually little to fear from them in terms of retaliation. Beyond that, they are paid for their efforts. Guala's convincing argument is that most field examples of SR are like this case. The property listed above as (1) has to be dropped, even if it was originally crucial.
Guala's view is that WR alone is at work in the field whenever punishers are compensated for inflicting a direct cost on defectors, because no net costs are involved in the long term. But the fact remains that in PGGs cooperation is enforced in a second stage where punishers incur short-term costs that have to be institutionally compensated. In some cases it is possible to design a second stage where cooperators withdraw cooperation from those who defected in the PGG (Milinski et al. Reference Milinski, Semmann and Krambeck2002). Defectors experience this as punishment and increase their contributions. But this design is not always possible: Tax evaders have to be fined. There is, therefore, a deep theoretical motivation to introduce a form of reciprocity that differs from direct and indirect reciprocity. It is required in relation to PGGs and leads directly to complex state or state-like institutions like the Carte di Regola. If defectors in PGGs could be punished by withdrawing benefits from them in direct or indirect reciprocity games, humans would not have invented those institutions.
Almost every worker in this field sees reciprocal altruists as “solely concerned about future gains” (target article, sect. 3, para. 3). In contrast, “Strong reciprocators cooperate because they feel it is the right thing to do, and they are ready to punish defectors at a cost” (sect. 3, para. 3). I want to argue, instead, that moral normativity is present in every form of human reciprocity. This claim can be supported historically, simply by showing that Trivers had fairness in mind when he speculated about the psychological mechanism that supports reciprocal altruism. Trivers' reciprocal altruists demand genuine psychological altruism from their partners: “Individuals who initiate altruistic acts out of a calculating rather than a generous-hearted disposition” (Trivers Reference Trivers1971, p. 51) are “subtle cheaters.” Reciprocal altruists demand partners that signal character traits like fairness. Evolutionary game theory designs strategies that are more generous or forgiving than Tit for Tat, and also more successful in the evolutionary dynamics (Kollock Reference Kollock1993). In the real world, these behavior patterns signal unselfish characters, exactly what human players are looking for in the field.
Alternatively, we can reflect about the limitations of WR as pointed out by advocates of SR. If one makes too close an analogy between WR and the “game-theoretic result known as the folk theorem” (target article, sect. 2, para. 7, italics in original), advocates of SR immediately object that, in this case, WR is only applicable when the conditions of the folk theorem hold. In fact, these conditions never hold. But it seems inconsistent for critics of WR to point this out and nonetheless insist that Trivers' reciprocal altruism is the instantiation of pure self-interest leading to cooperation in the – nonexistent – conditions of the folk theorem. Humans display reciprocal altruism only because of a special psychology that substitutes for the absence of those conditions. For example, as Frank (Reference Frank1988) has shown, reciprocal altruism would not be possible without the moral emotions counteracting the effects of temporal discounting. Moreover, as shown in simulations of indirect reciprocity, one-shot encounters in large groups are not an obstacle to cooperation because humans have a real concern for reputation, which circulates in a group through gossip. And despite the lack of complete information regarding reputations, humans extrapolate the bits of information they can get hold of to character traits stored in semantic memory. The emotions that solve the discounting problem, and the selection of prospective partners according to character traits like fairness, make reciprocators in dyadic games psychologically unselfish, though they remain biologically selfish by avoiding the costs of mutual defection. Note that direct reciprocity in PDs is rare in nature; it requires special cognitive and psychological traits that few organisms besides humans can display.
Guala engagingly criticizes the claim that strong reciprocity (SR) explains human cooperation beyond the limitations of classic or weak forms of reciprocity (WR) (direct and indirect reciprocity). For advocates of SR, its unique success comes from two distinctive characteristics: It is biologically altruistic, and it involves normative motivation. Guala shows convincingly that there is no evidence of a biologically altruistic form of human reciprocity outside the lab. I shall argue, furthermore, that normative motivation is present in weak reciprocity as well. In the end, SR is more similar to weak reciprocity than its advocates would admit. Nonetheless, there is some novelty to SR, although of a more modest nature.
SR has four properties as initially introduced: (1) In contrast to withdrawing cooperation, SR entails a net cost for the strong reciprocator, both in the short and the long term and therefore evolves through group selection; (2) it inflicts a direct cost on defectors; (3) repeated Public Goods games (PGGs; i.e., n-person Prisoner's Dilemmas [PDs]) are its proper domain; (4) it involves norms as socio-psychological mechanisms. Of these four properties, Guala shows that the first one is an artefact of lab experiments, with no real counterpart in the field. However, the other three properties are instantiated often enough in human institutions.
Properties (2) and (3) alone justify a distinction between SR and WR. In PGGs (repeated n-person PDs where n ≫ 2) cooperators cannot discipline defectors by withdrawing contributions in subsequent periods, as in direct and indirect reciprocity (Fehr & Gächter Reference Fehr and Gächter2000a; Ledyard Reference Ledyard, Kagel and Roth1995). Therefore, a second stage of the game is required, designed as a two-person game, where cooperators target and inflict costs on individual defectors. This two-stage structure is common to both the lab and the field. However, punishers in the lab pay a fee, usually less than the cost inflicted on the punished. Because retaliation against punishers is possible in the field, there is some reason to view this as unrealistic. But far more common in the real world is that humans device institutions to compensate punishers for their costs. Tax officials fining tax evaders have usually little to fear from them in terms of retaliation. Beyond that, they are paid for their efforts. Guala's convincing argument is that most field examples of SR are like this case. The property listed above as (1) has to be dropped, even if it was originally crucial.
Guala's view is that WR alone is at work in the field whenever punishers are compensated for inflicting a direct cost on defectors, because no net costs are involved in the long term. But the fact remains that in PGGs cooperation is enforced in a second stage where punishers incur short-term costs that have to be institutionally compensated. In some cases it is possible to design a second stage where cooperators withdraw cooperation from those who defected in the PGG (Milinski et al. Reference Milinski, Semmann and Krambeck2002). Defectors experience this as punishment and increase their contributions. But this design is not always possible: Tax evaders have to be fined. There is, therefore, a deep theoretical motivation to introduce a form of reciprocity that differs from direct and indirect reciprocity. It is required in relation to PGGs and leads directly to complex state or state-like institutions like the Carte di Regola. If defectors in PGGs could be punished by withdrawing benefits from them in direct or indirect reciprocity games, humans would not have invented those institutions.
Almost every worker in this field sees reciprocal altruists as “solely concerned about future gains” (target article, sect. 3, para. 3). In contrast, “Strong reciprocators cooperate because they feel it is the right thing to do, and they are ready to punish defectors at a cost” (sect. 3, para. 3). I want to argue, instead, that moral normativity is present in every form of human reciprocity. This claim can be supported historically, simply by showing that Trivers had fairness in mind when he speculated about the psychological mechanism that supports reciprocal altruism. Trivers' reciprocal altruists demand genuine psychological altruism from their partners: “Individuals who initiate altruistic acts out of a calculating rather than a generous-hearted disposition” (Trivers Reference Trivers1971, p. 51) are “subtle cheaters.” Reciprocal altruists demand partners that signal character traits like fairness. Evolutionary game theory designs strategies that are more generous or forgiving than Tit for Tat, and also more successful in the evolutionary dynamics (Kollock Reference Kollock1993). In the real world, these behavior patterns signal unselfish characters, exactly what human players are looking for in the field.
Alternatively, we can reflect about the limitations of WR as pointed out by advocates of SR. If one makes too close an analogy between WR and the “game-theoretic result known as the folk theorem” (target article, sect. 2, para. 7, italics in original), advocates of SR immediately object that, in this case, WR is only applicable when the conditions of the folk theorem hold. In fact, these conditions never hold. But it seems inconsistent for critics of WR to point this out and nonetheless insist that Trivers' reciprocal altruism is the instantiation of pure self-interest leading to cooperation in the – nonexistent – conditions of the folk theorem. Humans display reciprocal altruism only because of a special psychology that substitutes for the absence of those conditions. For example, as Frank (Reference Frank1988) has shown, reciprocal altruism would not be possible without the moral emotions counteracting the effects of temporal discounting. Moreover, as shown in simulations of indirect reciprocity, one-shot encounters in large groups are not an obstacle to cooperation because humans have a real concern for reputation, which circulates in a group through gossip. And despite the lack of complete information regarding reputations, humans extrapolate the bits of information they can get hold of to character traits stored in semantic memory. The emotions that solve the discounting problem, and the selection of prospective partners according to character traits like fairness, make reciprocators in dyadic games psychologically unselfish, though they remain biologically selfish by avoiding the costs of mutual defection. Note that direct reciprocity in PDs is rare in nature; it requires special cognitive and psychological traits that few organisms besides humans can display.