We have two related critical points concerning the target article. The first is that the evidence surveyed is consistent with a simpler and more general hypothesis. This is clearer in the case of revenge, where the relevant hypothesis is that there is adaptation for reciprocity, which applies to both welfare increasing and welfare reducing actions by others. The second is that the welfare trade-off ratio (WTR) hypothesised to determine acceptable harm/benefit ratios between individuals fails to account for the importance of absolute magnitude of revenge.
The literature in behavioural ecology pays significant attention to reciprocity of cooperative and non-cooperative behaviours, including reciprocation of cheating and aggression with withholding of cooperation (Trivers Reference Trivers1971). Conversely, it pays little or no attention to revenge as a separate topic. Welfare increasing actions, such as grooming, infant access, and resource sharing, are no less suitable candidates for reciprocation than welfare decreasing ones, such as aggressive assault and preventing resource access. So, for example, we see dynamic and context sensitive interactions between rates of welfare-decreasing and -increasing behaviours among female baboons, including aggression against low-ranking ones, and grooming of high-ranking by low-ranking ones (Barrett et al. Reference Barrett, Gaynor and Henzi2002). A relatively unified system is needed for trading off expected costs and benefits of available actions (including actions classifiable as altruistic, antagonistic, reconciliatory, or punitive). This is so even if its implementation is distributed and if some sub-functions – such as cheat detection – are specialised. Hypothesising a single general reciprocity system (or reciprocity sensitivity) instead of a specific revenge system means not having to hypothesise a separate system for cooperation and altruism. In addition, hypothesising a single system is consistent with a large body of established work on the unified neural representation of rewarding and aversive outcomes (Montague & Berns Reference Montague and Berns2002).
The best way to make a case against a general reciprocity system and in favour of a specific revenge adaptation would be to provide evidence of dissociation. For example, if in some cases of local brain damage, genetic intervention, and so on, there were individuals who could reciprocate altruism but not exploitation or attack, or vice versa, the hypothesis of specific revenge adaptations would be on stronger ground. There could be other considerations favouring a specific revenge adaptation that has features not predicted by a general reciprocity disposition, but the target article does not specify them or give evidence that they are satisfied.
It may seem as though a general reciprocity system would struggle to account for forgiveness. But the evidence regarding forgiveness is equivocal. Among nonhuman primates reconciliation behaviours following conflict have been documented in many species, and are sometimes taken to serve the function of repairing valuable relationships. These behaviours are not universal (see, e.g., Kappeler [Reference Kappeler1993] on the absence thereof among ring-tailed lemurs). Among the species in which they are observed, they exhibit properties that are partly at odds with the hypothesised forgiveness among humans. In particular, unlike forgiveness, which is a victim-initiated process, reconciliations are typically initiated by former aggressors (e.g., post-conflict grunting by former aggressor baboons – see Castles & Whiten Reference Castles and Whiten1998a). In addition the same outcomes, in respect of reduced stress by the victim of aggression, and reduced rates of subsequent attack from the previous aggressor, are observed in cases where instead of reconciliation there is redirected aggression by the former victim to an individual lower in the hierarchy (Aureli & van Schaik Reference Aureli and van Schaik1991a). The target article is silent on why redirected aggression might be a substitute for reconciliation.
Our second concern is that the schematic quantitative proposal regarding revenge is, in at least one important respect, incomplete. The basic proposal in the target article is that agents maintain and update welfare tradeoff ratios (WTRs) for other agents, which determine acceptable harm/benefit outcomes. The problem with simple ratios is that they are unable to account for a key feature of revenge or punishment, at least when described by many recorded cultures, which is that it is often quantified in absolute terms. Ratios on the other hand merely fix a range of outcomes, including ones where the revenge is both much larger and much smaller than whatever provoked it. (Put in monetary terms, a hostile WTR of 1 to 10 between me and some other individual says I'd willingly pay a dollar to see them lose ten dollars, or two to see them lose twenty, or a million to see them lose ten million, or…). Absolute specification is much more common. So, various ancient legal systems of which we have surviving records prescribe lists of penalties for recognised offences, which vary in type (death, corporal punishment, specific fines) but are all absolute in the sense of requiring a single death, corporal penalty, or a fine of some magnitude rather than a ratio consistent with a large range of outcomes. Examples include the Babylonian Code of Hammurabi from the 18th century BC, and the older Sumerian Codes of Lipit-Eshtar and Ur-Nammu, as well as more recent but still ancient codes such as the Hittite Laws. At various places in the Old Testament a principle of extracting equal injury in retaliation (at least for some crimes with human victims) is stated, including the specific formulation of “eye for eye” (Exodus 21:23–25; Leviticus 24:19–21; Deuteronomy 19:20–21). When Lady Capulet says “Romeo slew Tybalt, Romeo must not live” (Romeo and Juliet, Act 3, Scene 1), she expresses the view that utterly independent of the energetic cost and inconvenience, a penalty of a specific magnitude must be exacted.
An important critical response here would argue that these codified systems represent attempts to combat the ongoing damage that could ensue in the absence of a way of concluding some conflicts, and that they may even be attempts to solve a problem arising from a simpler WTR system for revenge. Were that the case one might expect that closely related social primates lacking legal systems would sometimes engage in runaway reciprocity and revenge governed by welfare tradeoff ratios and insensitive to absolute magnitudes.
You have
Access
We have two related critical points concerning the target article. The first is that the evidence surveyed is consistent with a simpler and more general hypothesis. This is clearer in the case of revenge, where the relevant hypothesis is that there is adaptation for reciprocity, which applies to both welfare increasing and welfare reducing actions by others. The second is that the welfare trade-off ratio (WTR) hypothesised to determine acceptable harm/benefit ratios between individuals fails to account for the importance of absolute magnitude of revenge.
The literature in behavioural ecology pays significant attention to reciprocity of cooperative and non-cooperative behaviours, including reciprocation of cheating and aggression with withholding of cooperation (Trivers Reference Trivers1971). Conversely, it pays little or no attention to revenge as a separate topic. Welfare increasing actions, such as grooming, infant access, and resource sharing, are no less suitable candidates for reciprocation than welfare decreasing ones, such as aggressive assault and preventing resource access. So, for example, we see dynamic and context sensitive interactions between rates of welfare-decreasing and -increasing behaviours among female baboons, including aggression against low-ranking ones, and grooming of high-ranking by low-ranking ones (Barrett et al. Reference Barrett, Gaynor and Henzi2002). A relatively unified system is needed for trading off expected costs and benefits of available actions (including actions classifiable as altruistic, antagonistic, reconciliatory, or punitive). This is so even if its implementation is distributed and if some sub-functions – such as cheat detection – are specialised. Hypothesising a single general reciprocity system (or reciprocity sensitivity) instead of a specific revenge system means not having to hypothesise a separate system for cooperation and altruism. In addition, hypothesising a single system is consistent with a large body of established work on the unified neural representation of rewarding and aversive outcomes (Montague & Berns Reference Montague and Berns2002).
The best way to make a case against a general reciprocity system and in favour of a specific revenge adaptation would be to provide evidence of dissociation. For example, if in some cases of local brain damage, genetic intervention, and so on, there were individuals who could reciprocate altruism but not exploitation or attack, or vice versa, the hypothesis of specific revenge adaptations would be on stronger ground. There could be other considerations favouring a specific revenge adaptation that has features not predicted by a general reciprocity disposition, but the target article does not specify them or give evidence that they are satisfied.
It may seem as though a general reciprocity system would struggle to account for forgiveness. But the evidence regarding forgiveness is equivocal. Among nonhuman primates reconciliation behaviours following conflict have been documented in many species, and are sometimes taken to serve the function of repairing valuable relationships. These behaviours are not universal (see, e.g., Kappeler [Reference Kappeler1993] on the absence thereof among ring-tailed lemurs). Among the species in which they are observed, they exhibit properties that are partly at odds with the hypothesised forgiveness among humans. In particular, unlike forgiveness, which is a victim-initiated process, reconciliations are typically initiated by former aggressors (e.g., post-conflict grunting by former aggressor baboons – see Castles & Whiten Reference Castles and Whiten1998a). In addition the same outcomes, in respect of reduced stress by the victim of aggression, and reduced rates of subsequent attack from the previous aggressor, are observed in cases where instead of reconciliation there is redirected aggression by the former victim to an individual lower in the hierarchy (Aureli & van Schaik Reference Aureli and van Schaik1991a). The target article is silent on why redirected aggression might be a substitute for reconciliation.
Our second concern is that the schematic quantitative proposal regarding revenge is, in at least one important respect, incomplete. The basic proposal in the target article is that agents maintain and update welfare tradeoff ratios (WTRs) for other agents, which determine acceptable harm/benefit outcomes. The problem with simple ratios is that they are unable to account for a key feature of revenge or punishment, at least when described by many recorded cultures, which is that it is often quantified in absolute terms. Ratios on the other hand merely fix a range of outcomes, including ones where the revenge is both much larger and much smaller than whatever provoked it. (Put in monetary terms, a hostile WTR of 1 to 10 between me and some other individual says I'd willingly pay a dollar to see them lose ten dollars, or two to see them lose twenty, or a million to see them lose ten million, or…). Absolute specification is much more common. So, various ancient legal systems of which we have surviving records prescribe lists of penalties for recognised offences, which vary in type (death, corporal punishment, specific fines) but are all absolute in the sense of requiring a single death, corporal penalty, or a fine of some magnitude rather than a ratio consistent with a large range of outcomes. Examples include the Babylonian Code of Hammurabi from the 18th century BC, and the older Sumerian Codes of Lipit-Eshtar and Ur-Nammu, as well as more recent but still ancient codes such as the Hittite Laws. At various places in the Old Testament a principle of extracting equal injury in retaliation (at least for some crimes with human victims) is stated, including the specific formulation of “eye for eye” (Exodus 21:23–25; Leviticus 24:19–21; Deuteronomy 19:20–21). When Lady Capulet says “Romeo slew Tybalt, Romeo must not live” (Romeo and Juliet, Act 3, Scene 1), she expresses the view that utterly independent of the energetic cost and inconvenience, a penalty of a specific magnitude must be exacted.
An important critical response here would argue that these codified systems represent attempts to combat the ongoing damage that could ensue in the absence of a way of concluding some conflicts, and that they may even be attempts to solve a problem arising from a simpler WTR system for revenge. Were that the case one might expect that closely related social primates lacking legal systems would sometimes engage in runaway reciprocity and revenge governed by welfare tradeoff ratios and insensitive to absolute magnitudes.