McCullough et al. perform several useful services in their target article. They remind us that the familiar phenomenon of revenge-seeking is not a pathology for which forgiveness is a cure – so that the “progress of civilization” might be imagined as leading toward a world where forgiveness abounds and revenge goes the way of foot-binding and cigarette smoking. Forgiveness, McCullough et al. correctly stress, is part of a functional complex with revenge, and the complex as a whole is almost certainly maintained by selection dynamics. Seeking revenge is often welfare-promoting for an agent because it changes others' incentives toward the self (see sect. 2.2.1), ideally inducing an increase in cooperation and/or a decrease in exploitation. Most importantly in my view, McCullough et al. recognize that forgiveness is important, interesting, and indeed theoretically surprising because it seems on the surface to be cheap talk. Once I exact my revenge against you, in equilibrium you should recognize that, if our continued relationship is of any value to me, I have no incentive to continue to follow a vengeful course if you avoid the action that triggered it. And since announcing the words “I forgive you” seems costless to me, what information value could the announcement possibly add to your appreciation of our equilibrium conditions? And yet most people regard the timing and circumstances of forgiveness as matters of solemn significance.
In other writings (Ross Reference Ross2004; Reference Ross2005; Reference Ross2006; Reference Ross2007; Reference Ross2012) I have suggested that a key condition for the existence of equilibrium strategy vectors that include revenge and forgiveness is the human susceptibility to shame – that is, to suffering from highly aversive emotions when one detects that information about one's norm violations may be spreading through social networks by gossip. Gossiping about a transgression is of course not the only, nor always the most efficacious, means by which revenge is carried out. However, it is arguably the most common. And, as I have argued, it is the crucial mechanism that promotes the ubiquitous existence of institutions for legitimizing forgiveness. Members of a species who depend on specialization and exchange of goods and services benefit from mechanisms that allow reputationally damaged members of economic networks to be restored to productive membership once corrective punishment is thought to have triggered policy reform on their part. Institutions for forgiveness allow the social costs of the most common acts of revenge to be massively reduced, at least when revenge is taken through harming a transgressor's reputation as opposed, for example, through harming his kneecaps. This in turn solves the much-discussed problem that arises when punishment of norm violation is costly to punishers, that agents are incentivized to free ride on the public good of norm enforcement (Guala Reference Guala2012).
It is of crucial importance, on this account, that forgiveness is institutionalized and public, and operates mainly in games that involve more than two players. Of course, once institutions for forgiveness exist, and their force is internalized through socialization and enculturation of young people, they can be exapted for use in dyadic interactions. But the shadow of society is always present in such interactions. Use of a phrase such as “I forgive you” is regulated by pragmatic (Gricean) conventions that include in their implicatures the act of promising not to spread blame for the transgression at issue. Forgiveness is a species of promising, and promising is a conventional institution regulated by socially enforced norms.
In light of these considerations, it is doubtful methodology to try to develop a model of the evolutionary function of revenge and forgiveness, as McCullough et al. do, by beginning with the case of the isolated dyad. The authors go to some lengths to analytically distinguish forgiveness – in a technical sense they deliberately construct – from more complicated institutionally governed cognitive/behavioral relatives. They are compelled to go to this trouble because they want to promote the hypothesis that there are specialized evolved cognitive modules for revenge and forgiveness that societies and institutions can then exploit – and sometimes suppress – as networks of interactions become more complex. Such modules would support special senses of revenge and forgiveness that can be isolated from social dynamics.
I do not see that in their target article McCullough et al. produce any evidence for this conjecture. It is plausible that institutional and culturally evolved and stabilized revenge/forgiveness complexes depend on biologically selected dispositions to be emotionally sensitive to perceptions of signals of possible changes in one's reputation among conspecifics. Emotions related to social rejection and reconciliation are probably evolved dispositions present in all normal primates, cetaceans, canids, corvids, et cetera. Such dispositions are likely necessary aspects of the evolution of cognitively mediated sociality in general. However, once social interdependence has evolved and thrown up free rider problems, strategies based on reciprocity are solutions that any cultural dynamics are likely to find and stabilize. I see no evidence that human institutions for forgiveness couldn't be supported by general cognitive processes and should be thought to require dedicated modules. Of course, dogs and elephants (etc.) lack resources for building forgiveness institutions. But what in their behavior should lead us to think that their cognitive processing goes beyond general dispositions to reconcile with group members following non-lethal altercations?
The hypothesis I favor is that the individual psychology of interaction is parasitic, both developmentally and theoretically, on its social psychology, economics, and political anthropology. The science should start at the aggregate level and treat individual and dyadic expressions as, respectively, exaptations (when characterizing development) and abstractions (when building game-theoretic models). For example, avoidance should be modeled as the special dyadic case of the primary phenomenon of ostracism from the group. McCullough et al. deliberately pursue the opposite modeling approach.
Ethics preclude testing between these hypotheses by isolating some human children from opportunities to learn forgiveness conventions. But if there are dedicated modules for computing revenge and forgiveness, we should find people suffering from neural deficits that knock out these capacities while leaving other emotionally motivated and regulated cognition intact. I am aware of no clear reports of such cases in any literature.
You have
Access
McCullough et al. perform several useful services in their target article. They remind us that the familiar phenomenon of revenge-seeking is not a pathology for which forgiveness is a cure – so that the “progress of civilization” might be imagined as leading toward a world where forgiveness abounds and revenge goes the way of foot-binding and cigarette smoking. Forgiveness, McCullough et al. correctly stress, is part of a functional complex with revenge, and the complex as a whole is almost certainly maintained by selection dynamics. Seeking revenge is often welfare-promoting for an agent because it changes others' incentives toward the self (see sect. 2.2.1), ideally inducing an increase in cooperation and/or a decrease in exploitation. Most importantly in my view, McCullough et al. recognize that forgiveness is important, interesting, and indeed theoretically surprising because it seems on the surface to be cheap talk. Once I exact my revenge against you, in equilibrium you should recognize that, if our continued relationship is of any value to me, I have no incentive to continue to follow a vengeful course if you avoid the action that triggered it. And since announcing the words “I forgive you” seems costless to me, what information value could the announcement possibly add to your appreciation of our equilibrium conditions? And yet most people regard the timing and circumstances of forgiveness as matters of solemn significance.
In other writings (Ross Reference Ross2004; Reference Ross2005; Reference Ross2006; Reference Ross2007; Reference Ross2012) I have suggested that a key condition for the existence of equilibrium strategy vectors that include revenge and forgiveness is the human susceptibility to shame – that is, to suffering from highly aversive emotions when one detects that information about one's norm violations may be spreading through social networks by gossip. Gossiping about a transgression is of course not the only, nor always the most efficacious, means by which revenge is carried out. However, it is arguably the most common. And, as I have argued, it is the crucial mechanism that promotes the ubiquitous existence of institutions for legitimizing forgiveness. Members of a species who depend on specialization and exchange of goods and services benefit from mechanisms that allow reputationally damaged members of economic networks to be restored to productive membership once corrective punishment is thought to have triggered policy reform on their part. Institutions for forgiveness allow the social costs of the most common acts of revenge to be massively reduced, at least when revenge is taken through harming a transgressor's reputation as opposed, for example, through harming his kneecaps. This in turn solves the much-discussed problem that arises when punishment of norm violation is costly to punishers, that agents are incentivized to free ride on the public good of norm enforcement (Guala Reference Guala2012).
It is of crucial importance, on this account, that forgiveness is institutionalized and public, and operates mainly in games that involve more than two players. Of course, once institutions for forgiveness exist, and their force is internalized through socialization and enculturation of young people, they can be exapted for use in dyadic interactions. But the shadow of society is always present in such interactions. Use of a phrase such as “I forgive you” is regulated by pragmatic (Gricean) conventions that include in their implicatures the act of promising not to spread blame for the transgression at issue. Forgiveness is a species of promising, and promising is a conventional institution regulated by socially enforced norms.
In light of these considerations, it is doubtful methodology to try to develop a model of the evolutionary function of revenge and forgiveness, as McCullough et al. do, by beginning with the case of the isolated dyad. The authors go to some lengths to analytically distinguish forgiveness – in a technical sense they deliberately construct – from more complicated institutionally governed cognitive/behavioral relatives. They are compelled to go to this trouble because they want to promote the hypothesis that there are specialized evolved cognitive modules for revenge and forgiveness that societies and institutions can then exploit – and sometimes suppress – as networks of interactions become more complex. Such modules would support special senses of revenge and forgiveness that can be isolated from social dynamics.
I do not see that in their target article McCullough et al. produce any evidence for this conjecture. It is plausible that institutional and culturally evolved and stabilized revenge/forgiveness complexes depend on biologically selected dispositions to be emotionally sensitive to perceptions of signals of possible changes in one's reputation among conspecifics. Emotions related to social rejection and reconciliation are probably evolved dispositions present in all normal primates, cetaceans, canids, corvids, et cetera. Such dispositions are likely necessary aspects of the evolution of cognitively mediated sociality in general. However, once social interdependence has evolved and thrown up free rider problems, strategies based on reciprocity are solutions that any cultural dynamics are likely to find and stabilize. I see no evidence that human institutions for forgiveness couldn't be supported by general cognitive processes and should be thought to require dedicated modules. Of course, dogs and elephants (etc.) lack resources for building forgiveness institutions. But what in their behavior should lead us to think that their cognitive processing goes beyond general dispositions to reconcile with group members following non-lethal altercations?
The hypothesis I favor is that the individual psychology of interaction is parasitic, both developmentally and theoretically, on its social psychology, economics, and political anthropology. The science should start at the aggregate level and treat individual and dyadic expressions as, respectively, exaptations (when characterizing development) and abstractions (when building game-theoretic models). For example, avoidance should be modeled as the special dyadic case of the primary phenomenon of ostracism from the group. McCullough et al. deliberately pursue the opposite modeling approach.
Ethics preclude testing between these hypotheses by isolating some human children from opportunities to learn forgiveness conventions. But if there are dedicated modules for computing revenge and forgiveness, we should find people suffering from neural deficits that knock out these capacities while leaving other emotionally motivated and regulated cognition intact. I am aware of no clear reports of such cases in any literature.