As Morris Hoffman and Timothy Goldsmith (fellows at the Gruter Institute) point out in their paper The Biological Roots of Punishment (Ohio State Journal of Criminal Law, Vol. 1, No. 2 Spring 2004, 627-641), research in game theory suggests that human beings are endowed with the propensity to punish those who don’t cooperate, and to inflict costs on “free riders” who take more than their fair share in social situations. They argue such a propensity was naturally selected for since those who possessed it were more likely to maintain stable social groups that in turn increased their chances of reproductive success. There is thus an evolutionary rationale for being what economist Herbert Gintis calls an altruistic punisher, someone who goes out of their way, perhaps at significant cost to themselves, to keep free riders in check. Most of us instinctively resent those who shirk their duties, or bend the system to their advantage, and we often seek to inflict comeuppance on cheaters. This is one component of our intuitive sense of justice and fairness.
We’re also strongly predisposed to retaliate against those who hurt us or our loved ones. Again, there’s little mystery about why we have this predisposition, since without it we wouldn’t have made the evolutionary cut. Only those with at least some inclination to protect themselves and their families and tribes by counter-aggression could have produced the lineage that now includes us. This retaliatory inclination too plays a role in our sense of what’s right or just.
We can see that there’s an evolutionary functional logic for having intuitions of justice that target free riders and aggressors for punishment. We can thus explain at least some of our punitive tendencies toward others as more or less hard-wired, ready to be triggered by certain of their behaviors. What’s important to recognize, however, is that this explanation doesn’t in any sense justify acting on such tendencies. Simply because we find ourselves built to retaliate or punish cheaters doesn’t mean that to do so is right in a particular situation, even though such tendencies have an evolutionary rationale. To justify punishment, we have to look beyond our instincts, and consider, in the context of other values we might have, what the consequences of punishment will be, and whether, all things considered, this particular punishment (or any punishment) best serves the purpose of bringing about the sort of society we want. To draw any direct conclusion from the fact of our punitive inclinations, whether altruistic or retaliatory, to their normative claim on us, would be an example of the naturalistic fallacy, the illicit move from is to ought.
But ironically, in a section on the naturalistic fallacy, Hoffman and Goldsmith make just this move, even as they deny making it:
This evolutionary insight—that our evolved behaviors are not simply a set of selfish urges that only civilization can overcome, but in fact include an urge for civilization itself—may be one of the most important contributions of neo-Darwinism, and it has great significance for the law and biology movement. As our cognitive abilities evolved in tandem with the emergence of culture, our urge to punish individuals who break the social contract appeared as an important stabilizing feature. Put this way, the urge to punish is not only natural, it is also a normative explanation of why government has the right to punish. Punishing wrongdoers is a critical part of the glue that holds people together in groups, and ultimately a defining characteristic of all civilizations. Punishment is a duty of civilized society not because all societies embrace it—that would be falling into the naturalistic fallacy—but rather because it is a central part of what being civilized is all about. (p. 633)
In this passage the transition from natural to normative is made on the basis that in order to have civilization (which we all want, of course), we must punish. Punishment is part of the naturally derived recipe for human cooperation – “the glue that holds people together” – but now it’s an instrument of our desire (not evolution’s, of course) for stable societies. The normative force of the state’s right to punish comes from the fact that it serves stability. But although it’s historically true that punishment has played this role, and of course still does, it isn’t a logical or conceptual truth that punishment is “ultimately a defining characteristic of all civilizations.” The normative claim that “punishment is a duty of a civilized society” doesn’t follow from the fact that punishment has thus far played a central role in maintaining civilizations.
It isn’t difficult to conceive of civilizations that count themselves civilized just to the extent to which punishment is minimized or done away with altogether in favor of non-punitive means for achieving social stability. Such a society would categorically deny that punishment “is a central part of what being civilized is all about.” We can see that the government’s right to punish doesn’t serve an intrinsic good independent of the goal of stability. Punishment is not a good we can establish just by consulting our punitive instincts. Rather, it’s an instrument of human flourishing that can conceivably be superseded by other more enlightened means, such as the design of social environments that result in very or vanishingly small needs for deterrence. The fact that evolution stumbled upon a particular means of keeping free riders and aggressors in check doesn’t require us to use it. No matter what evolution cooked up, we don’t have a normative duty to act on our punitive inclinations independent of other considerations.
Or do we? Retributivists believe we do, and it turns out that Hoffman and Goldsmith are retributivists (or neo-retributivists, as they call themselves). For them, as for other retributivists such as Michael Moore, Stephen Morse, David Hill, Ronald Bailey and Cathy Young, punishment turns out to be an end in itself, and only secondarily a means to an end. Retributive punishment is simply that which it is just to mete out whether or not it has any good, or even bad, consequences. How so, one wonders? Why, if punishment served as an instrument in building cooperative social groups, does it now assume a non-instrumental primary value for us that they think defines civilizations? Why do they suppose punishment is, as they so remarkably put it, “the crux of social living”?
Although they spend a good deal of their paper showing the evolutionary rationale behind the urge to punish (and they do it very well), Hoffman and Goldsmith end up debunking the idea that punishment should be a rational enterprise at all. Instead, we should stick with the retributive impulse. They start with a telling caricature of rehabilitation:
According to the rehabilitationists, in a perfectly enlightened post-Freudian society there can be no punishment at all, because there are no intentional wrongs, only a spasmodic confluence of other causes (our poverty, our schools, our diet, our friends, our mothers, our atoms). (p. 639)
No wonder rehabilitationists are misguided: they don’t believe in persons or intentional agency! It’s good to have that cleared up. If true, this would count as a failure of rationality on the part of rehabilitationists, of not taking into to account commonsense motivational psychology which punishment can engage in attempts to shape behavior. But of course rehabilitationists are perfectly cognizant of motivational psychology and acknowledge the occasional, unfortunate necessity of imposing sanctions as well as rewards in behavior change until utopia arrives.
Having dispatched those iwwational wehabilitationists, Hoffman and Goldsmith find other targets among those who think deterrence and incapacitation should play a role in punishment. But now they take them to task for being too rational in their dispassionate utilitarianism (emphasis added):
According to the utilitarians, the urge to punish has no place in the law, and punishment is moral only if it deters. Vengeance is for uncivilized brutes (or, paradoxically, only for God). Civilized people must therefore impose punishment with a detached and reluctant rationality, aimed simply at making the costs of wrongdoing sufficiently high to deter it. Richard Posner’s modern economic version of that same kind of rarified, detached deterrence—substituting a system of fines for the whole of the correctional system—is equally unsatisfying to those of us who have a hunch that there is something more to punishment than setting game-theoretic payoffs. As we have discussed, we have deep urges to punish free riders, quite apart from whether that punishment accomplishes any utilitarian goals, and in fact those urges are bound up with our nature as social creatures.
Incapacitation enjoys some resonance with our historic and human urge to banish wrongdoers, but it suffers from the same erroneous presumption of rationality as does deterrence: we simply do not banish free riders to prisons as the result of some detached arithmetic calculation about their likelihood of reoffending. We all sense that certain wrongs must be punished, and the severity of that punishment rarely has anything to do with our assessment of the risks of reoffending. (pp.639-40)
Their point is that, although the urge to punish may be explained by its evolutionary rationale in maintaining social cohesion, a rationale that, as the authors show, comes out clearly in game theory analysis, it is now freed from rational constraints. Rehabilitation, deterrence and incapacitation are our rationales for punishment, justified by consequences that might be socially beneficial, but these merely rational considerations now play second fiddle to the “deep urge to punish free riders, quite apart from whether that punishment accomplishes any utilitarian goals.” Why? Because “those urges are bound up with our nature as social creatures” and – the unstated assumption – we should express our natures, we must satisfy our punitive urges, independent of rationally warranted outcomes. This is the essence of retributivism: to give punishment its place at the table of intrinsically good guiding values. But Hoffman and Goldsmith’s justification for this seems a transparent instance of the naturalistic fallacy: that because we have punitive inclinations, it’s good or right to express them.
But, to be fair, they equivocate about this. They say:
The very instincts that make us human—to survive and excel as individuals but in a social context that often requires us to empathize and care for others—obligate us to punish those whose selfishness is unacceptably antisocial. (p. 640)
The naturalistic fallacy comes out clearly in supposing that we are obligated to our instincts, but note that a consequentialist rationale for the presumptive obligation immediate follows: “to punish those whose selfishness is unacceptably antisocial.” Here we see that punishment is actually instrumental to the governing value of social cooperation, to which we are uncontroversially obligated. But since, to return to an earlier point, cooperation can conceivably be accomplished without punishment, punishment can only be a contingent instrumental good. We all endorse social stability as intrinsic to (or at least necessary for) human flourishing, but we need not, and should not, endorse punishment as an end in itself (the essence of retributivism), even though it might phenomenally present itself to us as an instinctual obligation. As Victorians were about sex, at least in their public personas, so should we be about the urge to punish, except more so. Only do it when absolutely necessary.
It isn’t the case, therefore, that as Hoffman and Goldsmith say, “Only retribution meshes with…evolutionary insights.” Evolutionary insights help explain human nature, while defenses of retribution are normative claims that can’t be justified by elevating one aspect of that nature – the urge to punish – as an intrinsic good, which is simply to commit the naturalistic fallacy. If Hoffman and Goldsmith say that, no, they don’t take punishment as an intrinsic good or an independent governing value, that is, if they mean to count it as instrumental to cooperation, then they aren’t retributivists and their put-downs of rehabilitation, deterrence, and incapacitation quoted above should be retracted, since these are equally instrumental to cooperation.
For at least the near term future, depending on how much we invest in creative social engineering, the prospect of punishment will be necessary to keep cheaters in line, and to effectively deter certain classes of potential offenders (rational corporate executives with lots to lose are far more deterrable than desperate drug addicts). So to deny the evolutionary justification for retribution – non-consequentialist, deontological punishment – is not to fall into a progressive Panglossian swoon. But it does require those enamored of punishment as the sine qua non of civilization to formulate another, valid justification for retribution which avoids the naturalistic fallacy. Failing that, they should abjure retributivism and only punish when it does good that can’t be done in a non-punitive fashion. Seeing that we can’t equate our values with our instincts requires all of us to second guess our urge to punish as a contingent, likely non-optimal contribution to solving the problem of human cooperation, not the essence of civilization.*
TWC, January 22 2006
* As Hoffman and Goldsmith wisely point out, “not all of our urges and desires are well tuned to today’s world” (629), the urge to punish included.
Clark correctly points out that there is an evolutionary “logic for having intuitions of justice that target free riders and aggressors for punishment,” but he has missed the larger picture. We are a social species by nature. Human societies exist because evolution has endowed humans with the interest and capacity to engage in mutually beneficial behavioral exchanges that build complex networks of friendship and cooperation extending far beyond close kin. Civilization arises out of this evolved feature of our species. But there is a problem. Because of sexual reproduction, the genetic interests of every individual are not identical to those of any other. It thus follows that one’s self-interests frequently conflict with those of others, or that perceptions of immediate self-interest conflict with recognition of longer-term self interest that requires bonds of friendship, status, or honor dependant on the friendship or respect of others. At one level, this is common sense and has been recognized for millennia in codes of conduct, but its fundamental nature has been obscured by both religion and philosophy. The seminal paper tethering universal psychological features of humans to evolutionary theory is Robert Trivers’ The Evolution of Reciprocal Altruism, Quarterly Review of Biology 46: 35-57 (1971). Game theory is frosting on the cake.
(For readers new to this literature, altruism in this context refers to behavior in which the cost to self is less than the benefit to the recipient, and where the measure of costs and benefits is the net effect on lifetime reproductive success. With reciprocal exchanges of this kind, both individuals prosper. This is an evolutionary argument showing how altruistic behavior can be the object of positive selection.)
“Oughts” are not simply the work of a moral cortex overruling an “is” that, to use Clark’s phrase, is “more or less hardwired.” The inclination to punish or retaliate is “hard wired” to the same extent as a need for friends, the capacity to forgive, the desire for respect, or empathy for another’s misfortune. Clark’s view that punishment is “an instrument of our desire (not evolution’s, of course) for stable societies” does not acknowledge the evolutionary origins of that desire. The evolutionary outcome, however, is imperfect, and the fundamental “oughts” of law and religion exist because self-interest readily motivates people to manipulate or even cheat. Cultural proscriptions and contracts – indeed moral systems – are thus stabilizing agents serving to regulate behaviors in ways that provide a good for society by preserving the rights and the expectations of individuals. If the rules were totally out of step with our evolved desires they would not be effective. In economically stratified societies they may work to the advantage of those who have the power to make the rules, but that is another story. Whether any particular outcome is good depends on how society chooses to balance the impacts of rules on all parties concerned. That is the arena in which moral decisions transcend our evolutionary heritage.
An evolved “is” can be accompanied by a supporting normative “ought.” For example, we are evolved to care solicitously about the needs of close kin. For people everywhere, inclusive fitness and kin selection have made caring about children an “is,” strongly motivated by an evolved brain. The normative “ought” that children not be abused reinforces our inner self. Society finds reinforcement necessary because a bit of the serpent accompanies the dove in each of us, in some more than others.
A similar argument is helpful in understanding punishment. Normative decisions about punishment are ordinarily utilitarian (at least in the limited sense of having practical consequences) even though they map onto evolved motivators. Although Clark “can conceive of a world in which punishment is minimized or done away with altogether in favor of non-punitive means for achieving social stability,” no such society (other than, perhaps, small groups of closely related individuals) has ever existed. Such a society would have to overcome some important features of nature, starting with the universal non-identity of self-interests. When violations of fairness occur, societies respond in ways the culture deems effective, ranging from execution to forgiveness. That societies characteristically settle on some form of punishment is thus pragmatic.
Although punishment can be colored with emotion (retaliation, victim’s rights), measured punishment is inherently defensive and protective of society, thus utilitarian in the sense of distributing good to the greatest number. To be sure, societies do not always choose wisely, and punishments can reflect extremes of religious doctrine or the whims of powerful leaders.
Evolution can help us to understand our deeper motivations, but both the formulation and application of rules requires rational analysis. Judge Hoffman and I were quite specific in pointing out that punishment should fit the crime, and that there is positive value to society if a transgressor can become a productive member of society. Society is, in general, very poor in achieving this latter goal on any but a modest scale. The effort raises other issues as well as being, arguably, difficult.
I certainly accept the evolutionary origins of our desires and made that clear in the first part of my commentary; indeed I opened with a quote from Hoffman and Goldsmith to that effect. And of course it’s the case that although we are to some extent hardwired in our responses (which is another way of saying they have their origins in evolution), culture mediates them such that “moral decisions transcend our evolutionary heritage.” I also agree that “If [moral systems] were totally out of step with our evolved desires they would not be effective.” But my critique of retributivism is consistent with these claims. Questioning the retributive impulse is precisely one instance of culture transcending the innate, evolutionarily bequeathed urge to punish as a moral benchmark. And the humanitarian goal of minimizing punishment in achieving human cooperation is certainly not out of step with our evolved human nature, which includes irenic as well as punitive dispositions.
But to get to the main point, what’s most striking about Goldsmith’s response is that he accepts the pragmatic, utilitarian function of punishment: “Normative decisions about punishment are ordinarily utilitarian (at least in the limited sense of having practical consequences) even though they map onto evolved motivators.” And: “When violations of fairness occur, societies respond in ways the culture deems effective, ranging from execution to forgiveness. That societies characteristically settle on some form of punishment is thus pragmatic.” And: “…measured punishment is inherently defensive and protective of society, thus utilitarian in the sense of distributing good to the greatest number.” And: “Judge Hoffman and I were quite specific in pointing out that punishment should fit the crime, and that there is positive value to society if a transgressor can become a productive member of society. Society is, in general, very poor in achieving this latter goal on any but a modest scale.”
So there’s no longer anything about retribution per se that figures in this account. Rather, punishment serves goals, for instance getting transgressors to be productive members of society. Judging by his reply, for Goldsmith punishment isn’t any longer a retributive imperative, such that, as he suggests in his original paper, we should heed our “deep urges to punish free riders, quite apart from whether that punishment accomplishes any utilitarian goals.” Instead, it’s an instrument of human flourishing that can conceivably be improved upon. For instance, as Goldsmith points out, our society is very poor in achieving the goal of getting transgressors to be productive members of society. Might we do a better job at this if we assiduously explore non-punitive alternatives to punishment, or even more broadly, seek to prevent transgressors in the first place by means of a morally enlightened mentor state?
These are empirical questions, of course, but in seeing that punishment is a means to an end, not an end in itself, Goldsmith isn’t committed to punishment whether or not it achieves agreed upon goals, the essence of retributivism. So he isn’t a retributivist in any standard sense of the term, and I don’t see that he’s still defending the normative claim that punishment “is the crux of social living.” True, any viable society “would have to overcome some important features of nature, starting with the universal non-identity of self-interests.” But there’s no good reason to claim that punishment, although a pragmatic necessity at the moment, is essential to such a society. Not unless, that is, one thinks that we are somehow obligated to the urge to punish.
I am pleased to see that Clark’s rejoinder makes clear there is less separating us than may have appeared. What remains I take to be either a lingering misunderstanding of the original article or a conviction that there is an effective alternative to the entire concept of punishment (as distinct from creating social conditions in which there are many fewer instances of antisocial behavior). In what follows I consider the possibility of the first option. The second remains a hypothetical, of which only a bit more below.
Judge Hoffman and I write at one point, “As we have discussed, we have deep urges to punish free riders, quite apart from whether that punishment accomplishes any utilitarian goals, and in fact those urges are bound up with our nature as social creatures.” That, I submit, is a description of reality. Simply consider how frequently people follow those urges, “getting even,” taking vengeance, upholding honor or status, showing that you are not to be trifled with. Note that in many instances such behavior actually proves to be counter-productive—devoid of utility, natural but hardly good, censored by society—simply because the consequences were not adequately considered. More to the point, Judge Hoffman and I did not say anywhere that society must “heed” such urges in its search for justice. Quite the opposite: punishment should be an act of collective reason. In the interests of society, however, there is an obligation to act when agreed-upon rules are infringed.
In the words of the article: “Punishment is not simply a social aggregate of vengeance, something enlightened societies should strive to avoid. On the contrary, it is the crux of social living. The very instincts that make us human—to survive and excel as individuals but in a social context that often requires us to empathize and care for others—obligates us to punish those whose selfishness is unacceptably antisocial. Those same instincts obligate us to be judicious in the severity of the punishment we impose, to forgive transgressors after appropriate punishment, and to accept, and express our willingness to accept, our own proportionate punishment when we deserve it.”
As for the “crux of social living,” I infer that some are looking forward to an alternative to any form of punishment, achieved, as Clark puts it, through “creative social engineering.” As I suggested earlier, that topic raises other, difficult issues. It lies outside the scope of the article.
In the wider chat room discussion there is a suggestion that retribution and punishment have been confused. This is not a valid criticism of the article, which includes a discussion of how retribution means different things to different people.
More important, and in summary, Judge Hoffman and I neither engaged in the naturalistic fallacy nor suggested that we are prisoners of our instincts (another unfortunate word meaning different things to different people). Contemporary evolutionary theory and neurobiology have much richer palettes of understanding than implied by those charges. It is disconcerting to see (in the chat room, not so clearly in Clark’s rejoinder) evolutionary biology substituted for theism in an argument about punishment that is made to hang on the issue of free will. It is also frustrating beyond measure to see a carefully nuanced argument relating psychology and evolution reduced to the label “retributivist.” If I were to offer a comment in a constructive spirit, it would be to suggest that some of the philosophy in the forum would benefit from a little more science and a little less labeling. But then, it’s your forum, so having blunted a few lances I shall withdraw from the field.