It is commendable that the authors of this excellent adaptationist account of an important aspect of human social interaction are concerned that “for some crucial questions about the revenge and forgiveness systems […] data are scant” (McCullough et al., target article, sect. 7, para. 2). However, this is not entirely accurate. The main objective of this commentary is to discuss some very relevant experimental work on revenge that has apparently escaped the attention of McCullough et al., in the hope that these neglected findings and the associated theoretical ideas can be usefully incorporated into their broad view.
Unlike the majority of findings cited by McCullough et al., the work in question is not from the domain of economic games, which is significant given the external-validity doubts that can be raised about games research with regard to the genuineness of participants' motivation and, especially, emotion. Instead, the data come from social-psychological behavioral experiments published in the1960s and 1970s (in top-tier journals), in which ecologically powerful procedures were used that the subsequent human-research regulations made difficult to implement. Furthermore, some of these experiments dealt with issues that may arise in long-term human dyadic relationships; such data may contribute to the authors' complex analysis of repeated “effective updating” (sect. 3.1.1, para. 2).
The key questions are these: What are the behavioral and emotional consequences of revenge? How might these outcomes influence both the avenger's (AV) and the initial offender's (IO) computations of the present and future costs and welfare tradeoff ratios (WTRs)?
Most of the data come from a three-stage research paradigm: (1) IO's offense against AV (such as insults); (2) AV's behavioral retaliation against IO (such as fictitious electric shocks); and (3) obtaining dependent measures of AV's arousal, anger, and additional behavioral aggression against IO. These experimental results are informative about the short- and longer-term, both internal (sympathetic arousal, rated anger) and external (additional aggressive behavior), consequences for AV (and for IO as the target of any additional aggression) of the retaliatory actions previously executed by AV against IO.
To summarize the data which have been obtained as a function of revenge:
1. A sharply reduced amount of immediate (additional – that is, post-revenge) aggression by AV against IO (and also against substitute or “scapegoat” targets) – not only in comparison with the behavior of would-be avengers who did not have a prior opportunity for retaliation (Doob & Wood Reference Doob and Wood1972; Konečni & Doob Reference Konečni and Doob1972; Konečni & Ebbesen Reference Konečni and Ebbesen1976), but also of those who were required to perform tasks (math problems) that minimized the likelihood of anger-producing rumination (Konečni Reference Konečni1975a). In fact, even observing the IO (allegedly) in pain (Bramel et al. Reference Bramel, Taub and Blum1968) or (allegedly) hurt by someone else (Doob & Wood Reference Doob and Wood1972) decreased the amount of retaliatory aggression directed by the offended person at the culprit.
2. A significantly decreased level of AV's physiological arousal (that had been sharply raised by IO) compared to various control groups (Hokanson & Burgess Reference Hokanson and Burgess1962; Hokanson et al. Reference Hokanson, Burgess and Cohen1963; Hokanson & Shetler Reference Hokanson and Shetler1961). Revenge decreases physiological arousal quickly. More generally, because aggressive responses apparently succeed in terminating noxious stimulation emanating from others more effectively than other responses, ceteris paribus, they acquire arousal-reducing properties (Konečni Reference Konečni1975a; Patterson & Cobb Reference Patterson, Cobb and Hill1971).
3. Auxiliary findings that are theoretically congruent with those in point (2) have also been obtained: As a function of behavioral revenge against IO, avengers display a restored affinity for complex stimulation (Konečni et al. Reference Konečni, Crozier and Doob1976) and a reduced level of alcohol consumption (Marlatt et al. Reference Marlatt, Kosturn and Lang1975).
4. A significantly lower level of AV's self-rated anger, compared to participants without a retaliatory opportunity, but, importantly, as high a level of AV's dislike for IO as that observed in appropriate control participants (Konečni Reference Konečni1975a; Konečni & Doob Reference Konečni and Doob1972).
The entire observed pattern of findings, (1) to (4), can be accommodated by Konečni's (1975a; Reference Konečni, Kaplan, Konečni and Novaco1984) anger-aggression bidirectional-causation model (AABC). The model also predicts, because of the data in the above-mentioned points (2) and (4), that the future execution of aggressive acts by AV against IO would be more likely in long-term dyads (and occur sooner in the offense-revenge sequence): The original angry, righteous avenger may become an anger-free (“cold-blooded”) bully who strikes with little or no provocation. Such pre-emption complicates the computation of long-term WTRs beyond what McCullough et al. have proposed for revenge, possibly with large errors along the long road of adjustment or even a complete breakdown of the relationship (often with dire consequences). Retaliatory pre-emption – an unprovoked attack camouflaged as retaliation for an (imaginary) offense – is also relevant for the computation of “indirect [third-party] deterrence” (sect. 3.1.2).
Another important fact – predicted by the AABC model – that should influence the computations by both AV and IO is that the amount of revenge is strongly affected by the random arousal-related circumstances in which the initial offense occurs. Specifically, the amount of revenge has been observed in experiments to increase as a function of additional (unrelated) stressors that are present concurrently with, or immediately following the initial offense. When AVs do strenuous physical exercise (Zillmann et al. Reference Zillmann, Katcher and Milavsky1972) or listen to loud and complex tones (Konečni Reference Konečni1975b), their retaliation against IOs is more severe than that performed by controls. Therefore, from both AV's and IO's computational perspective, the context of the initial offense is important – as is the perceived intentionality of both the offending and vengeful actions.
The research described above has been largely ignored – for various (bad) reasons. It was pigeonholed as “catharsis” and falsely related to the outmoded “hydraulic” model of Freud and Lorenz, or to Aristotle's “pity and terror” – but, significantly, not to Plato's correct judgment of the benefits of revenge. There was the dubious idea that watching boxing films, fantasy aggression, or children attacking inanimate targets (none of these genuine vengeful activities) should reduce aggression – yet the opposite, and correct, result is predicted by the AABC model. A slew of inadequate experimental procedures has been used to disprove straw “catharsis” hypotheses and reach the socio-politically desirable conclusion that “aggression breeds aggression” (something easily achieved, according to AABC). Fortunately, sound evolutionary thinking (in the target article) has finally imposed a reality check on wishful thinking.
You have
Access
It is commendable that the authors of this excellent adaptationist account of an important aspect of human social interaction are concerned that “for some crucial questions about the revenge and forgiveness systems […] data are scant” (McCullough et al., target article, sect. 7, para. 2). However, this is not entirely accurate. The main objective of this commentary is to discuss some very relevant experimental work on revenge that has apparently escaped the attention of McCullough et al., in the hope that these neglected findings and the associated theoretical ideas can be usefully incorporated into their broad view.
Unlike the majority of findings cited by McCullough et al., the work in question is not from the domain of economic games, which is significant given the external-validity doubts that can be raised about games research with regard to the genuineness of participants' motivation and, especially, emotion. Instead, the data come from social-psychological behavioral experiments published in the1960s and 1970s (in top-tier journals), in which ecologically powerful procedures were used that the subsequent human-research regulations made difficult to implement. Furthermore, some of these experiments dealt with issues that may arise in long-term human dyadic relationships; such data may contribute to the authors' complex analysis of repeated “effective updating” (sect. 3.1.1, para. 2).
The key questions are these: What are the behavioral and emotional consequences of revenge? How might these outcomes influence both the avenger's (AV) and the initial offender's (IO) computations of the present and future costs and welfare tradeoff ratios (WTRs)?
Most of the data come from a three-stage research paradigm: (1) IO's offense against AV (such as insults); (2) AV's behavioral retaliation against IO (such as fictitious electric shocks); and (3) obtaining dependent measures of AV's arousal, anger, and additional behavioral aggression against IO. These experimental results are informative about the short- and longer-term, both internal (sympathetic arousal, rated anger) and external (additional aggressive behavior), consequences for AV (and for IO as the target of any additional aggression) of the retaliatory actions previously executed by AV against IO.
To summarize the data which have been obtained as a function of revenge:
1. A sharply reduced amount of immediate (additional – that is, post-revenge) aggression by AV against IO (and also against substitute or “scapegoat” targets) – not only in comparison with the behavior of would-be avengers who did not have a prior opportunity for retaliation (Doob & Wood Reference Doob and Wood1972; Konečni & Doob Reference Konečni and Doob1972; Konečni & Ebbesen Reference Konečni and Ebbesen1976), but also of those who were required to perform tasks (math problems) that minimized the likelihood of anger-producing rumination (Konečni Reference Konečni1975a). In fact, even observing the IO (allegedly) in pain (Bramel et al. Reference Bramel, Taub and Blum1968) or (allegedly) hurt by someone else (Doob & Wood Reference Doob and Wood1972) decreased the amount of retaliatory aggression directed by the offended person at the culprit.
2. A significantly decreased level of AV's physiological arousal (that had been sharply raised by IO) compared to various control groups (Hokanson & Burgess Reference Hokanson and Burgess1962; Hokanson et al. Reference Hokanson, Burgess and Cohen1963; Hokanson & Shetler Reference Hokanson and Shetler1961). Revenge decreases physiological arousal quickly. More generally, because aggressive responses apparently succeed in terminating noxious stimulation emanating from others more effectively than other responses, ceteris paribus, they acquire arousal-reducing properties (Konečni Reference Konečni1975a; Patterson & Cobb Reference Patterson, Cobb and Hill1971).
3. Auxiliary findings that are theoretically congruent with those in point (2) have also been obtained: As a function of behavioral revenge against IO, avengers display a restored affinity for complex stimulation (Konečni et al. Reference Konečni, Crozier and Doob1976) and a reduced level of alcohol consumption (Marlatt et al. Reference Marlatt, Kosturn and Lang1975).
4. A significantly lower level of AV's self-rated anger, compared to participants without a retaliatory opportunity, but, importantly, as high a level of AV's dislike for IO as that observed in appropriate control participants (Konečni Reference Konečni1975a; Konečni & Doob Reference Konečni and Doob1972).
The entire observed pattern of findings, (1) to (4), can be accommodated by Konečni's (1975a; Reference Konečni, Kaplan, Konečni and Novaco1984) anger-aggression bidirectional-causation model (AABC). The model also predicts, because of the data in the above-mentioned points (2) and (4), that the future execution of aggressive acts by AV against IO would be more likely in long-term dyads (and occur sooner in the offense-revenge sequence): The original angry, righteous avenger may become an anger-free (“cold-blooded”) bully who strikes with little or no provocation. Such pre-emption complicates the computation of long-term WTRs beyond what McCullough et al. have proposed for revenge, possibly with large errors along the long road of adjustment or even a complete breakdown of the relationship (often with dire consequences). Retaliatory pre-emption – an unprovoked attack camouflaged as retaliation for an (imaginary) offense – is also relevant for the computation of “indirect [third-party] deterrence” (sect. 3.1.2).
Another important fact – predicted by the AABC model – that should influence the computations by both AV and IO is that the amount of revenge is strongly affected by the random arousal-related circumstances in which the initial offense occurs. Specifically, the amount of revenge has been observed in experiments to increase as a function of additional (unrelated) stressors that are present concurrently with, or immediately following the initial offense. When AVs do strenuous physical exercise (Zillmann et al. Reference Zillmann, Katcher and Milavsky1972) or listen to loud and complex tones (Konečni Reference Konečni1975b), their retaliation against IOs is more severe than that performed by controls. Therefore, from both AV's and IO's computational perspective, the context of the initial offense is important – as is the perceived intentionality of both the offending and vengeful actions.
The research described above has been largely ignored – for various (bad) reasons. It was pigeonholed as “catharsis” and falsely related to the outmoded “hydraulic” model of Freud and Lorenz, or to Aristotle's “pity and terror” – but, significantly, not to Plato's correct judgment of the benefits of revenge. There was the dubious idea that watching boxing films, fantasy aggression, or children attacking inanimate targets (none of these genuine vengeful activities) should reduce aggression – yet the opposite, and correct, result is predicted by the AABC model. A slew of inadequate experimental procedures has been used to disprove straw “catharsis” hypotheses and reach the socio-politically desirable conclusion that “aggression breeds aggression” (something easily achieved, according to AABC). Fortunately, sound evolutionary thinking (in the target article) has finally imposed a reality check on wishful thinking.