Guala rightly emphasizes the difficulties involved in extrapolating from the experimental evidence for reciprocity (whether strong or weak) to the analysis of cooperation and punishment in the “wild.” The experiments, which, as Guala says, are remarkably robust, have significantly increased our understanding of the psychology of trust and retaliation; and their extension to a wider range of subjects in different cultural and social environments has shown how far these may modify or redirect presumptively universal predispositions and preferences. But they do not by themselves generate testable predictions about the costs borne by strong reciprocators in real-world communities, which enhance the probability of the community's survival and reproduction.
Calculation of the costs of punishment in risk or effort is a very different exercise from calculating monetary gains and losses. Among the Bergdama, a decision that an offender is to be thrashed, expelled, or put to death “is reached casually round the camp fire, and if necessary the young men are then told to enforce it” (Schapera Reference Schapera1956, p. 87). There is presumably a cost to the young men so instructed, particularly if the offender may resist. But the young men may welcome the opportunity for the legitimate exercise of violence and anticipate a reward in enhanced prestige. This is not revenge of the kind that is more often triggered in small face-to-face societies by sexual jealousies and antagonisms. But nor is it weak reciprocity. Nor are the young men bounty-hunters rewarded in cash for the risks they incur laying hands on a fugitive. The punishment is (we must assume) effective in enforcing cooperation. But it is not easy to see what the experimental literature contributes to the understanding of what exactly is going on.
If Guala is right, it is plausible to suppose that punishment of defectors, free-riders, and cheats is cheapest either in small egalitarian societies where the sanctions are cultural (i.e., the information affecting behaviour in the phenotype is transmitted by imitation or learning) and weak reciprocity does most of the work, or in fully evolved states where the sanctions are social (i.e., the information is coded in rule-governed practices which define institutional roles) and the punishers are rewarded for punishing. There is very little net cost in punishment by ridicule and ostracism of offenders where behaviour is easily monitored, and very little net cost to punishers in a police state where they are paid both to detect and restrain offenders and to punish fellow-citizens who refuse to inform on them. But what about the intermediate cases?
In the institutions for collective action which Elinor Ostrom has studied to such good effect, Guala argues that the problem of non-cooperation is resolved by removing the obstacles in the way of non-costly punishment and that costs are incurred more in setting up these institutions than in punishing defectors. But “covenants without the sword” (Ostrom et al. Reference Ostrom, Walker and Gardner1992) are not covenants without sanctions. In the inshore fishery at Alanya in Turkey, the fishers are assigned to their locations by lot at the beginning of each season, and the list of locations is deposited with the mayor and gendarme. Cheating is difficult because it is easy for the other fishers to observe it and they have a common interest in ensuring that their own rights will not be usurped. But they have also to be “willing to defend their rights using physical means if necessary” (Ostrom Reference Ostrom1990, p. 220). There is an element of weak reciprocity in the fishers' relationships with one another, and Ostrom reports that disputes are generally handled at the local coffeehouse. There also appears to be a fit with the model presented by Boyd et al. (Reference Boyd, Gintis and Bowles2010), in which the total cost of punishing a free-rider declines as the number of punishers increases. But costs are still costs, and strong reciprocity is waiting in the wings, so to speak, to ensure that the covenant is renewed.
A striking example of strong reciprocity at work is the Mafia, as documented by Gambetta (Reference Gambetta2009). The population under study is as far as it is possible to imagine from a community of Good Samaritans. It consists of adult males unconstrained in their behaviour either by the cultural sanction of conventional moral disapproval or the social sanction of control by agents of the state. Its members depend on sustained and predictable cooperation among themselves. But this involves trusting people of the very kind from whom there is least reason to expect that trust will be forthcoming. The survival and reproduction of the organization is therefore critically dependent on the punishment of non-cooperators against whom the sanction is severe physical injury or assassination. Trust is sustained by a costly signalling system which evolves by variation and selection in the classic Darwinian manner to ensure that threats of punishment of defectors are credible and, where implemented, efficient. Non-punishers are required to demonstrate their trustworthiness by punishing when ordered to do so, and they face the same severe sanctions if they refuse.
Group selection then comes into operation as the “families” in which cooperation is most successfully replicated drive those in which it is less successfully replicated towards extinction. The “families” are impermeable to immigrants carrying different traits, and are relatively stable in composition. Mafiosi sometimes allow their close kin to be recruited by other “families,” but this does not compromise within-group behavioural homogeneity: the incomers are recruited only because they are known to be trustworthy. Differences between groups are partly random, but occasionally arise through the infiltration of an undercover agent of the state. Some groups are in environments where they are under more pressure than others from the state. But those with a larger proportion of more trustworthy punishers have higher survival rates.
Guala rightly emphasizes the difficulties involved in extrapolating from the experimental evidence for reciprocity (whether strong or weak) to the analysis of cooperation and punishment in the “wild.” The experiments, which, as Guala says, are remarkably robust, have significantly increased our understanding of the psychology of trust and retaliation; and their extension to a wider range of subjects in different cultural and social environments has shown how far these may modify or redirect presumptively universal predispositions and preferences. But they do not by themselves generate testable predictions about the costs borne by strong reciprocators in real-world communities, which enhance the probability of the community's survival and reproduction.
Calculation of the costs of punishment in risk or effort is a very different exercise from calculating monetary gains and losses. Among the Bergdama, a decision that an offender is to be thrashed, expelled, or put to death “is reached casually round the camp fire, and if necessary the young men are then told to enforce it” (Schapera Reference Schapera1956, p. 87). There is presumably a cost to the young men so instructed, particularly if the offender may resist. But the young men may welcome the opportunity for the legitimate exercise of violence and anticipate a reward in enhanced prestige. This is not revenge of the kind that is more often triggered in small face-to-face societies by sexual jealousies and antagonisms. But nor is it weak reciprocity. Nor are the young men bounty-hunters rewarded in cash for the risks they incur laying hands on a fugitive. The punishment is (we must assume) effective in enforcing cooperation. But it is not easy to see what the experimental literature contributes to the understanding of what exactly is going on.
If Guala is right, it is plausible to suppose that punishment of defectors, free-riders, and cheats is cheapest either in small egalitarian societies where the sanctions are cultural (i.e., the information affecting behaviour in the phenotype is transmitted by imitation or learning) and weak reciprocity does most of the work, or in fully evolved states where the sanctions are social (i.e., the information is coded in rule-governed practices which define institutional roles) and the punishers are rewarded for punishing. There is very little net cost in punishment by ridicule and ostracism of offenders where behaviour is easily monitored, and very little net cost to punishers in a police state where they are paid both to detect and restrain offenders and to punish fellow-citizens who refuse to inform on them. But what about the intermediate cases?
In the institutions for collective action which Elinor Ostrom has studied to such good effect, Guala argues that the problem of non-cooperation is resolved by removing the obstacles in the way of non-costly punishment and that costs are incurred more in setting up these institutions than in punishing defectors. But “covenants without the sword” (Ostrom et al. Reference Ostrom, Walker and Gardner1992) are not covenants without sanctions. In the inshore fishery at Alanya in Turkey, the fishers are assigned to their locations by lot at the beginning of each season, and the list of locations is deposited with the mayor and gendarme. Cheating is difficult because it is easy for the other fishers to observe it and they have a common interest in ensuring that their own rights will not be usurped. But they have also to be “willing to defend their rights using physical means if necessary” (Ostrom Reference Ostrom1990, p. 220). There is an element of weak reciprocity in the fishers' relationships with one another, and Ostrom reports that disputes are generally handled at the local coffeehouse. There also appears to be a fit with the model presented by Boyd et al. (Reference Boyd, Gintis and Bowles2010), in which the total cost of punishing a free-rider declines as the number of punishers increases. But costs are still costs, and strong reciprocity is waiting in the wings, so to speak, to ensure that the covenant is renewed.
A striking example of strong reciprocity at work is the Mafia, as documented by Gambetta (Reference Gambetta2009). The population under study is as far as it is possible to imagine from a community of Good Samaritans. It consists of adult males unconstrained in their behaviour either by the cultural sanction of conventional moral disapproval or the social sanction of control by agents of the state. Its members depend on sustained and predictable cooperation among themselves. But this involves trusting people of the very kind from whom there is least reason to expect that trust will be forthcoming. The survival and reproduction of the organization is therefore critically dependent on the punishment of non-cooperators against whom the sanction is severe physical injury or assassination. Trust is sustained by a costly signalling system which evolves by variation and selection in the classic Darwinian manner to ensure that threats of punishment of defectors are credible and, where implemented, efficient. Non-punishers are required to demonstrate their trustworthiness by punishing when ordered to do so, and they face the same severe sanctions if they refuse.
Group selection then comes into operation as the “families” in which cooperation is most successfully replicated drive those in which it is less successfully replicated towards extinction. The “families” are impermeable to immigrants carrying different traits, and are relatively stable in composition. Mafiosi sometimes allow their close kin to be recruited by other “families,” but this does not compromise within-group behavioural homogeneity: the incomers are recruited only because they are known to be trustworthy. Differences between groups are partly random, but occasionally arise through the infiltration of an undercover agent of the state. Some groups are in environments where they are under more pressure than others from the state. But those with a larger proportion of more trustworthy punishers have higher survival rates.