Following Sell et al. (Reference Sell, Tooby and Cosmides2009), we agree with McCullough et al. that multiple factors shape responses to being wronged (e.g., whether the transgressor is a close ally, kin, or someone likely to exact high costs due to a status or formidability differential), and that this process is intimately related to the motivational profile of anger. McCullough et al. go further, however, by apparently proposing the existence of additional specialized psychological adaptations to enable deterrence. It is most parsimonious to attribute the deterrence-related computations reviewed by the authors to the emotion “anger,” operating in conjunction with (1) mechanisms that transcend the domain of interpersonal conflict (e.g., norm-acquisition, future forecasting, perspective-taking), (2) an attitudinal system that regulates a wide variety of behaviors, and (3) systems related to other motivations, such as reputation management.
Consider the complex case of indirect deterrence. In our view, the computational demands described by McCullough et al. in this regard are met by evolved capacities to categorize events, assume others' perspectives, forecast the future, and weigh costs against benefits. These capacities are directed and organized over short time spans by the emotion of anger (Fessler Reference Fessler, Potegal, Stemmler and Spielberger2010; Tooby & Cosmides Reference Tooby, Cosmides and Buss2005), and over longer time spans by the more enduring attitude of hatred, an evaluative representation that tracks and reacts to the fortunes of an other whose principal relationship with the self is as a source of costs inflicted in zero-sum contexts (Gervais & Fessler, under review). Hence, on the one hand, if by “an evolved cognitive system that implements … deterrence” (target article, Abstract) the authors mean a functionally specialized system that evolved expressly for this purpose, then we would argue that redundant algorithms for deterrence-related event categorization, perspective-taking, cost–benefit analysis, and so on, seem implausible – why engineer new content-dedicated devices when a bricolage of existing devices will satisfy? On the other hand, if the authors concede that there is no uniquely bounded “revenge adaptation,” but contend that, nonetheless, the outputs of this bricolage can be treated as if they are produced by such an adaptation, given that they address a unified domain (i.e., “revenge” is a recurrent adaptive task), then we would argue that the authors have mistaken a folk category (cost infliction motivated by anger and hatred following transgression) for a nonexistent natural kind. There are many kinds of deterrence that do not stem from the anger-hatred nexus (e.g., swatting a dog in order to teach it not to steal food off the table), and hence neither constitute “revenge” in any ordinary sense of the word, nor involve the core motivational components of the bricolage at issue.
The above critique holds for each of the observations adduced by McCullough et al. As further evidence of special design, the authors discuss strategic calibrations made in light of culturally and individually varying exigencies, such as whether the putative adaptation operates in a legalistic society that punishes retaliatory violence, or in a weak soma likely to be injured in combat. We agree that humans adaptively modulate deterrence behavior in light of social and personal contexts, but, again, see no reason to postulate specialized subroutines of a revenge adaptation. Cultural norm acquisition mechanisms (Sripada & Stich Reference Sripada, Stich, Carruthers, Laurence and Stich2007) are sufficient to enable learning of locally accepted ways of resolving conflict. Reputation management mechanisms are also implicated, moderating retributive behavior to the extent that the reputational consequences of how one responds to transgression vary, with some societies valorizing, and others demonizing, violent retribution (Fessler Reference Fessler and Barkow2006). This suggests only the interaction of distinct psychological motives (i.e., to punish, to protect one's reputation, etc.), not, as the authors imply (sect. 3.1.2, paras. 1–4), that the supposed vengeance system contains a customized reputation circuit. This explains why the presence of onlookers can magnify not only violence, but also charitable giving (Harbaugh Reference Harbaugh1998) and shame displays (Fessler Reference Fessler2004) – reputation management systems operate in tandem with, and may potentiate or vitiate, other systems.
As evidence of a forgiveness adaptation, McCullough et al. observe that transgressors' relatedness, past friendship, or opportunity to injuriously counterattack, mitigate the severity of deterrent responses to transgressions. The competing perspective that we have applied to the revenge adaptation applies here as well. Although humans likely do take fitness-relevant factors such as relatedness, prior cooperation, and relative status/formidability into account during conflicts, it is more parsimonious to ascribe these calibrations to the operation of other systems (e.g., for affiliation in the case of transgressive friends or kin, or fear in the case of formidable adversaries) that moderate anger than to propose new, highly redundant pathways engineered to facilitate strategic détente.
We have argued that the postulated wholes (adaptations for revenge and forgiveness) are not greater than the sums of their parts (perspective-taking, event categorization, norm-acquisition, future forecasting, reputation management, etc.). The proposed adaptations do not appear to possess domain-specific content beyond components that, although useful in calculating deterrence, mostly evolved for other reasons. Anger is indeed considered to have evolved to deter harmful transgressors by inflicting costs or withholding benefits, and has demonstrated unambiguous domain-specificity in this regard (e.g., Fessler & Gervais Reference Fessler, Gervais, Kappeler and Silk2010; Lazarus Reference Lazarus1991; Sell et al. Reference Sell, Tooby and Cosmides2009). McCullough et al. characterize anger as the proximal mediator of the proposed revenge adaptation, but this appears to needlessly multiply entities. The crux of the issue is whether a vengeance adaptation evolved with specialized mechanisms to compute factors such as the likelihood, type, and severity of reprisals, the intentions of the transgressor, social consequences, status differentials between self and transgressor, prior history of cooperation with transgressor, kinship with transgressor, and so forth, or whether these diverse variables are taken into account through the simultaneous operation of multiple domain-specific modules operating within the same mind, perhaps coordinated by anger in the short term, and hatred in the long term. In both scenarios, retaliatory behavior is moderated by personal, cultural, and situational factors; adjudicating the issue is therefore a problem of theory rather than of missing or disputed data. Given these options, we advocate the latter alternative because it is simpler, kludgier, and therefore more evolutionarily plausible.
You have
Access
Following Sell et al. (Reference Sell, Tooby and Cosmides2009), we agree with McCullough et al. that multiple factors shape responses to being wronged (e.g., whether the transgressor is a close ally, kin, or someone likely to exact high costs due to a status or formidability differential), and that this process is intimately related to the motivational profile of anger. McCullough et al. go further, however, by apparently proposing the existence of additional specialized psychological adaptations to enable deterrence. It is most parsimonious to attribute the deterrence-related computations reviewed by the authors to the emotion “anger,” operating in conjunction with (1) mechanisms that transcend the domain of interpersonal conflict (e.g., norm-acquisition, future forecasting, perspective-taking), (2) an attitudinal system that regulates a wide variety of behaviors, and (3) systems related to other motivations, such as reputation management.
Consider the complex case of indirect deterrence. In our view, the computational demands described by McCullough et al. in this regard are met by evolved capacities to categorize events, assume others' perspectives, forecast the future, and weigh costs against benefits. These capacities are directed and organized over short time spans by the emotion of anger (Fessler Reference Fessler, Potegal, Stemmler and Spielberger2010; Tooby & Cosmides Reference Tooby, Cosmides and Buss2005), and over longer time spans by the more enduring attitude of hatred, an evaluative representation that tracks and reacts to the fortunes of an other whose principal relationship with the self is as a source of costs inflicted in zero-sum contexts (Gervais & Fessler, under review). Hence, on the one hand, if by “an evolved cognitive system that implements … deterrence” (target article, Abstract) the authors mean a functionally specialized system that evolved expressly for this purpose, then we would argue that redundant algorithms for deterrence-related event categorization, perspective-taking, cost–benefit analysis, and so on, seem implausible – why engineer new content-dedicated devices when a bricolage of existing devices will satisfy? On the other hand, if the authors concede that there is no uniquely bounded “revenge adaptation,” but contend that, nonetheless, the outputs of this bricolage can be treated as if they are produced by such an adaptation, given that they address a unified domain (i.e., “revenge” is a recurrent adaptive task), then we would argue that the authors have mistaken a folk category (cost infliction motivated by anger and hatred following transgression) for a nonexistent natural kind. There are many kinds of deterrence that do not stem from the anger-hatred nexus (e.g., swatting a dog in order to teach it not to steal food off the table), and hence neither constitute “revenge” in any ordinary sense of the word, nor involve the core motivational components of the bricolage at issue.
The above critique holds for each of the observations adduced by McCullough et al. As further evidence of special design, the authors discuss strategic calibrations made in light of culturally and individually varying exigencies, such as whether the putative adaptation operates in a legalistic society that punishes retaliatory violence, or in a weak soma likely to be injured in combat. We agree that humans adaptively modulate deterrence behavior in light of social and personal contexts, but, again, see no reason to postulate specialized subroutines of a revenge adaptation. Cultural norm acquisition mechanisms (Sripada & Stich Reference Sripada, Stich, Carruthers, Laurence and Stich2007) are sufficient to enable learning of locally accepted ways of resolving conflict. Reputation management mechanisms are also implicated, moderating retributive behavior to the extent that the reputational consequences of how one responds to transgression vary, with some societies valorizing, and others demonizing, violent retribution (Fessler Reference Fessler and Barkow2006). This suggests only the interaction of distinct psychological motives (i.e., to punish, to protect one's reputation, etc.), not, as the authors imply (sect. 3.1.2, paras. 1–4), that the supposed vengeance system contains a customized reputation circuit. This explains why the presence of onlookers can magnify not only violence, but also charitable giving (Harbaugh Reference Harbaugh1998) and shame displays (Fessler Reference Fessler2004) – reputation management systems operate in tandem with, and may potentiate or vitiate, other systems.
As evidence of a forgiveness adaptation, McCullough et al. observe that transgressors' relatedness, past friendship, or opportunity to injuriously counterattack, mitigate the severity of deterrent responses to transgressions. The competing perspective that we have applied to the revenge adaptation applies here as well. Although humans likely do take fitness-relevant factors such as relatedness, prior cooperation, and relative status/formidability into account during conflicts, it is more parsimonious to ascribe these calibrations to the operation of other systems (e.g., for affiliation in the case of transgressive friends or kin, or fear in the case of formidable adversaries) that moderate anger than to propose new, highly redundant pathways engineered to facilitate strategic détente.
We have argued that the postulated wholes (adaptations for revenge and forgiveness) are not greater than the sums of their parts (perspective-taking, event categorization, norm-acquisition, future forecasting, reputation management, etc.). The proposed adaptations do not appear to possess domain-specific content beyond components that, although useful in calculating deterrence, mostly evolved for other reasons. Anger is indeed considered to have evolved to deter harmful transgressors by inflicting costs or withholding benefits, and has demonstrated unambiguous domain-specificity in this regard (e.g., Fessler & Gervais Reference Fessler, Gervais, Kappeler and Silk2010; Lazarus Reference Lazarus1991; Sell et al. Reference Sell, Tooby and Cosmides2009). McCullough et al. characterize anger as the proximal mediator of the proposed revenge adaptation, but this appears to needlessly multiply entities. The crux of the issue is whether a vengeance adaptation evolved with specialized mechanisms to compute factors such as the likelihood, type, and severity of reprisals, the intentions of the transgressor, social consequences, status differentials between self and transgressor, prior history of cooperation with transgressor, kinship with transgressor, and so forth, or whether these diverse variables are taken into account through the simultaneous operation of multiple domain-specific modules operating within the same mind, perhaps coordinated by anger in the short term, and hatred in the long term. In both scenarios, retaliatory behavior is moderated by personal, cultural, and situational factors; adjudicating the issue is therefore a problem of theory rather than of missing or disputed data. Given these options, we advocate the latter alternative because it is simpler, kludgier, and therefore more evolutionarily plausible.