Punishment

Since punishment involves pain or deprivation that people wish to avoid, its intentional imposition by the state requires justification. The difficulties of justification cannot be avoided by the view that punishment is an inevitable adjunct of a system of criminal law. If criminal law is defined to include punishment, the central question remains whether society should have a system of mandatory rules enforced by penalties. Relatively small associations of like-minded people may be able to operate with rules that are not backed by sanctions, and a choice by the larger society against authorizing legal punishment is at least theoretically possible. Moreover, actual infliction of penalties is not inextricably tied to authorization. A father who has threatened punishment if two daughters do not stop fighting must decide whether to follow through if the fight continues. Congruence between threat and actual performance on the scene does constitute one good reason for punishing. Future threats will be taken less seriously if past threats are not fulfilled, and parents usually wish to avoid the impression that they will not do what they say. Nevertheless, because he now sees that the punishment threatened is too severe, or understands better the children's reasons for fighting, the father may fail to carry out his threat.

In the broader society also, threatened punishments are not always inflicted on persons who have unquestionably committed crimes. The police or prosecutor may decide not to proceed, a jury may acquit in the face of unmistakable evidence of guilt, or a judge may decide after conviction not to impose punishment. A judge with legal authority to make such a decision must determine if punishment is appropriate; even if punishment is legally required to be imposed, the countervailing reasons may be so powerful that the court will not do so.

If actual punishment never or very rarely followed threatened punishment, the threat would lose significance. Thus, punishment in some cases is a practical necessity for any system in which threats of punishment are to be taken seriously; and to that extent, the justification of punishment is inseparable from the justification of threats of punishment.

The dominant approaches to justification are retributive and utilitarian. Briefly stated, a retributivist claims that punishment is justified because people deserve it; a utilitarian believes that justification lies in the useful purposes that punishment serves (the latter approach is sometimes also referred to as "consequentialist," or "instrumentalist"). Many actual theories of punishment do not fit unambiguously and exclusively into one of these two categories. Satisfying both retributive and utilitarian criteria may be thought necessary to warrant punishment; or utilitarian criteria may be thought crucial for one question (for example, whether there should be a system of punishment) and retributive criteria for another (for example, who should be punished); or the use of retributive sorts of approaches may be thought appropriate on utilitarian grounds. Beginning from rather straightforward versions of retributive and utilitarian theory, the analysis proceeds to positions that are more complex.

Retributive justification. Why should wrongdoers be punished? Most people might respond simply that they deserve it or that they should suffer in return for the harm they have done. Such feelings are deeply ingrained, at least in many cultures, and are often supported by notions of divine punishment for those who disobey God's laws. A simple retributivist justification provides a philosophical account corresponding to these feelings: someone who has violated the rights of others should be penalized, and punishment restores the moral order that has been breached by the original wrongful act. The idea is strikingly captured by Immanuel Kant's claim that an island society about to disband should still execute its last murderer. Society not only has a right to punish a person who deserves punishment, but it has a duty to do so. In Kant's view, a failure to punish those who deserve it leaves guilt upon the society; according to G. W. F. Hegel, punishment honors the criminal as a rational being and gives him what it is his right to have. In simple retributivist theory, practices of punishment are justified because society should render harm to wrongdoers; only those who are guilty of wrongdoing should be punished; and the severity of punishment should be proportional to the degree of wrongdoing, an approach crudely reflected in the idea of "an eye for an eye, a tooth for a tooth."

Close examination of this theory dispels much of its apparent simplicity, reveals some of the tensions between its implications and the practices of actual societies, and exposes its vulnerability to powerful objections. Taken as claiming an intimate connection between moral guilt and justified legal punishment, the retributive theory raises troubling questions about the proper purposes of a state and about any human attempts to equate reward and punishment to moral deserts.

Moral guilt and social judgment. One fundamental question is whether people are ever morally guilty in the way that basic retributive theory seems to suppose. If all our acts are consequences of preceding causes over which we ultimately have had no control, causes that were set in motion before we were born—if, in other words, philosophical determinism is true—then the thief or murderer is, in the last analysis, more a victim of misfortune than a villain on the cosmic stage. Although he may be evil in some sense and able to control his actions, his character has been formed by forces outside himself, and that ultimately determines the choices he makes. From this perspective, assertions that a vicious person should be punished simply because he deserves to be seem as anomalous as assertions that a vicious dog should be punished simply because he deserves to be. Unless one wishes to take the paradoxical position, analogous to certain religious doctrines of predestination, that people are guilty for qualities and acts they cannot help, the simple retributive theory is incompatible with determinism. It requires some notion of free will that attributes to humans responsibility for doing wrong in a way that is not attributed to other animals.

Acceptance of free will, which is certainly the undergirding for the ordinary sense of morality, does not remove all the obstacles to acceptance of retributivism. One human can rarely judge with confidence the moral guilt of others, and few doubt that among persons who commit similar wrongful acts, vast differences in moral guilt exist. Many of those who commit very serious crimes have suffered extreme psychological or social deprivation, and/or physical or psychological abuse. Moreover, a penalty supposed to redress a moral imbalance should perhaps depend upon an offender's overall moral record and how the good and bad fortunes of his life compare with that record; yet making such an evaluation with any accuracy is even more beyond human capacities than judging the moral guilt attaching to a particular act.

Finally, not all acts that reflect serious moral guilt are the subject of criminal punishment in a liberal society. Personal wrongs that members of families and acquaintances do to one another may be of greater magnitude morally than some petty crimes, even though they do not carry publicly imposed penalties. If the purpose of punishment were truly to redress moral guilt, justifying this variance in treatment would be difficult, but few people believe that a liberal society should make the punishment of all serious moral wrongs its business.

To some, the very idea that pain should be inflicted on a person simply because he has committed an earlier moral wrong may seem indefensible, whether the agent inflicting the pain is human or divine. Even those who believe that a just God would strike some such balance may think that restoring the moral order is not an appropriate human purpose, and is certainly not a proper purpose of the state, limited as the state should be in its capacities to learn about events and to dispose of people's lives.

The retributivist may resist this conclusion and maintain that the infliction of legal penalties for moral transgressions is a legitimate public purpose, one that happens to be outweighed by other values in certain circumstances. The retributivist can argue that the severity of an offense provides at least a rough indication of the magnitude of moral wrong and that a punishment proportioned to the offense, and perhaps tailored to some extent to other factors of moral relevance, can give the offender approximately what he deserves. These responses may save retributive theory from the attack of total irrelevance, but they do not provide a complete justification for practices of legal punishment as they exist or might exist.

Violations of social norms and fairness. A rather different retributive approach is that criminals deserve punishment because they violate norms established by society, the magnitude of the violation being measured by the seriousness with which society treats the offense. In this form, the theory sidesteps the objection that correcting moral wrongs is not the business of the criminal law, and it does not impose upon officials the impossible burden of ascertaining subtle degrees of moral guilt. This version of the theory fits better with existing (and conceivable) practices of criminal punishment, but in doing so, it no longer connects moral guilt so strongly to justifiable punishment and does not resolve the question of why morality demands that society punish those who violate its norms simply for the sake of punishing them.

One answer to this question is that fairness to citizens who make sacrifices by obeying the law requires that violators be punished rather than reap benefits for disregarding legal standards. What is crucial and debatable about this view is the claim that law-abiding members of the community will suffer an actual injustice if the guilty go unpunished. The position is most persuasive in respect to crimes whose commission actually increases the overall burden on those who obey. Given steady revenue needs, a sufficient amount of tax evasion will increase the burdens of those who pay in full. Demanding that the evader pay back taxes does redress an injustice, but whether failure to send him to jail, if that is the only possible penalty, would be unfair to honest taxpayers is less clear. The unjust loss to the honest will not be made up in any event, but the jail term will at least offset the evader's unfair advantage. Some criminal activities, such as speeding and theft, would be engaged in more widely if it were not for the law's prohibition, but their commission does not increase general burdens as directly as does tax evasion. Because the ordinary lawabiding person has foregone some possible gain, the criminal may still be perceived as having attained an unfair advantage that should be offset by punishment. The claim about fairness to lawabiding citizens is least persuasive in respect to criminal activities (such as rape) that very few citizens would wish to undertake, no matter what the law said about them.

For these crimes, as well as others directed at individuals, fairness to victims, rather than (or as well as) fairness to all law-abiding citizens, might be thought to justify punishment. Fairness to victims undoubtedly requires redress of their injuries to whatever extent that can be effected, but whether and to what extent it requires harsh treatment of criminals that does not benefit victims is doubtful.

The general normative question about both fairness arguments above is this: If someone has achieved a comparative advantage over another by an unjust act, does fairness to the person suffering a comparative disadvantage require stripping the offender of his advantage, even when that would do nothing to improve the position of the disadvantaged person? So understood, the fundamental question about the fairness argument is close to the question about the intrinsic value of punishing wrongdoing, although emphasizing a comparative dimension. Many of those who believe that inflicting pain on the morally guilty is not worthwhile for its own sake will also conclude that such pain cannot be supported simply because it nullifies some comparative advantage.

A third fairness argument compares the offender to other, similar offenders. Retributive or just deserts theories of punishment usually seek to promote not only proportionality of punishment relative to culpability, but also uniformity in the treatment of equally culpable offenders. A sentence less severe than that given to other, similar offenders is unfair to those offenders; a more severe sentence is unfair to the defendant. However, since equality norms do not require that offenders receive any particular degree of punishment (or indeed, any punishment at all), such norms are more properly seen as limiting, rather than justifying, the imposition of sanctions.

Utilitarian justification. Utilitarian theories of punishment dominated American jurisprudence during most of the twentieth century. According to Jeremy Bentham's classical utilitarianism, whether an act or social practice is morally desirable depends upon whether it promotes human happiness better than possible alternatives. Since punishment involves pain, it can be justified only if it accomplishes enough good consequences to outweigh this harm. A theory of punishment may make the balance of likely consequences central to justification without asserting, as Bentham did, that all relevant consequences are reducible to happiness and unhappiness. It may even claim that reducing future instances of immoral violations of right is itself an appropriate goal independent of the effect of those violations on the people involved. In modern usage, utilitarianism is often employed to refer broadly to theories that likely consequences determine the morality of action, and this usage is followed here.

The catalogs of beneficial consequences that utilitarians have thought can be realized by punishment have varied, but the following have generally been regarded as most important.

1. General deterrence. Knowledge that punishment will follow crime deters people from committing crimes, thus reducing future violations of right and the unhappiness and insecurity they would cause. The person who has already committed a crime cannot, of course, be deterred from committing that crime, but his punishment may help to deter others. In Bentham's view, general deterrence was very much a matter of affording rational self-interested persons good reasons not to commit crimes. With a properly developed penal code, the benefits to be gained from criminal activity would be outweighed by the harms of punishment, even when those harms were discounted by the probability of avoiding detection. Accordingly, the greater the temptation to commit a particular crime and the smaller the chance of detection, the more severe the penalty should be.

Punishment can also deter in ways more subtle than adding a relevant negative factor for cool calculation. Seeing others punished for certain behavior can create in people a sense of association between punishment and act that may constrain them even when they are sure they will not get caught. Adults, as well as children, may subconsciously fear punishment even though rationally they are confident it will not occur.

2. Norm reinforcement. For young children, the line may be very thin between believing that behavior is wrong and fearing punishment. Adults draw the distinction more plainly, but seeing others punished can still contribute to their sense that actions are wrong, helping them to internalize the norms society has set. Practices of punishment can thus reinforce community norms by affecting the dictates of individual consciences. Serious criminal punishment represents society's strong condemnation of what the offender has done, and performs a significant role in moral education.

A person's feeling of moral obligation to obey rules may depend considerably on his sense that he is treated fairly under them. If others profit with impunity from violations of the law, a lawabiding person may develop a sense of unfairness, wondering if he too should break the law to obtain similar advantages. Punishment helps assure citizens that the laws as administered deal fairly with their interests. Whether or not the law-abiding citizen actually has some individualized moral claim to have wrongdoers punished, punishment will probably contribute to his willing acceptance of legal constraints. Similarly, if an offender sees that other offenders received less severe punishment, his feeling of having been treated unfairly may diminish his respect for and willingness to conform to the law. All of these considerations constitute the utilitarian side of fairness arguments for punishment.

3. Individual deterrence. The actual imposition of punishment creates fear in the offender that if he repeats his act, he will be punished again. Adults are more able than small children to draw conclusions from the punishment of others, but having a harm befall oneself is almost always a sharper lesson than seeing the same harm occur to others. To deter an offender from repeating his actions, a penalty should be severe enough to outweigh in his mind the benefits of the crime. For the utilitarian, more severe punishment of repeat offenders is warranted partly because the first penalty has shown itself ineffective from the standpoint of individual deterrence.

4. Incapacitation and other forms of risk management. Imprisonment temporarily puts the convicted criminal out of general circulation, and the death penalty does so permanently. These punishments physically prevent persons of dangerous disposition from acting upon their destructive tendencies. Less drastic forms of risk management include probationary or parole supervision, and accompanying requirements (for example, random urine tests to detect use of illegal drugs) and prohibitions (use of alcohol or firearms, association with certain persons, contact with the victim, and so on). As with individual deterrence, more severe risk-management measures are warranted for repeat offenders because such offenders are statistically more likely to commit further crimes. However, incapacitation of a high-risk offender may nevertheless fail to prevent further crimes. This would be the case if such an offender were to be immediately replaced by another offender (for example, on a street corner well-suited for selling drugs). It would also be the case if the offender, when released from prison, had become more dangerous than he was before (so that the crimes he commits after release are more numerous or more serious than those which were prevented while he was imprisoned).

5. Reform. Punishment may help to reform the criminal so that his wish to commit crimes will be lessened, and perhaps so that he can be a happier, more useful person. Conviction and simple imposition of a penalty might themselves be thought to contribute to reform if they help an offender become aware that he has acted wrongly. In that case, punishment acts as a form of norm reinforcement operating at the individual rather than the community level; the importance of promoting the offender's awareness of wrongdoing is also cited by those who stress the "communicative" aspects of punishment. However, reform is usually conceived as involving more positive steps to make offenders less antisocial by altering their basic character, improving their skills, or teaching them how to control their crime-producing urges (for example, their tendency to abuse drugs or alcohol, or to commit sex crimes). Various psychological therapies, medications, and even drastic interventions such as psychosurgery, are designed to curb destructive tendencies. Educational and training programs can render legitimate employment a more attractive alternative to criminal endeavors. These may indirectly help enhance self-respect, but their primary purpose is to alter the options that the released convict will face.

6. Vengeance. The utilitarian, in contrast to the retributivist, does not suppose that wrongful acts intrinsically deserve a harsh response, but utilitarians recognize that victims, their families and friends, and some members of the public will feel frustrated if no such response is forthcoming. Satisfying these desires that punishment be imposed is seen as one legitimate aim in punishing the offender. In part, the point is straightforwardly to increase the happiness, or reduce the unhappiness, of those who want the offender punished, but formal punishment can also help increase their sense of respect for the law and deflect unchanneled acts of private vengeance.

7. Community or victim restoration. Another utilitarian sentencing goal that began to receive much greater emphasis toward the end of the twentieth century is to repair the damage that the offense has caused, by requiring the offender to make restitution or perform compensatory service for the victim or the community. Restorative justice goals are also sometimes defined to include acceptance of responsibility or repentance by the offender, forgiveness by the victim, and victim-offender or community-offender reconciliation, for example, by means of mediation or an apology. Since some of these goals and measures may also benefit the offender, and depend heavily on his cooperation, they might not seem sufficiently unpleasant or stigmatizing to qualify as punishment; however, the same could be said for many measures designed to promote reform or manage the offender's risk of re-offending. Indeed, the broadest goals of restorative justice overlap with several previously mentioned punishment goals, in particular, norm reinforcement, risk-management, and reform. Restorative measures can also be seen as a means of deflecting the desires of victims and the public for vengeance, and providing a more constructive outlet for such feelings.

Unlike a basic retributive theory, the utilitarian approach to punishment is compatible with philosophical determinism. Whether or not human acts are completely determined by prior causes, punishment can be an efficacious prior cause. A determinist can support even the "condemnation" component of punishment on utilitarian grounds, believing that condemnation and feelings of guilt are useful instruments in guiding human behavior.

From the utilitarian perspective, the acts for which criminal punishment should be authorized are those with respect to which the good consequences of punishment can outweigh the bad; the persons who should be punished are those whom it is useful to punish; and the severity of punishment should be determined not by some abstract notion of deserts but by marginal usefulness. Each extra ingredient of punishment is warranted only if its added benefits outweigh its added harms and costs. (Of course, in real life such a fine scale cannot be developed, but legislators and those administering punishment should be guided by this principle.) The utilitarian does not start with the premise that penalties of equal severity should go to those with equal blame. For general deterrence, roughly equal penalties for the same offenses may be appropriate, but goals relating to individual offenders may support individuation of treatment, leading, for example, to long confinement for those judged irredeemably antisocial, and to rehabilitation and prompt release for those whose character can be positively transformed (or already has been, at the time of sentencing).

Philosophical objections to utilitarianism. Utilitarian programs for systems of punishment are subject to two kinds of objections: those which challenge basic philosophical premises, and those which claim that different systems would better accomplish social aims. When existing practices are attacked, disentangling the theoretical from the practical complaints often is not simple, but the following discussion tries to separate the two, dealing first with basic attacks on utilitarian theory and indicating what modifications may be needed to accommodate valid criticisms.

The most fundamental objection is to treating the criminal as a means to satisfy social purposes rather than as an end in himself. This objection bears on why, and how, guilty offenders may be punished; but the most damaging aspect of the attack is that utilitarianism admits the possibility of justified punishment of the innocent. The retributivist asserts that such punishment is morally wrong even when it would produce a balance of favorable consequences.

Various responses have been made by utilitarians. One is that since the term punishment implies guilt, the innocent cannot logically be punished. The terminological point is highly doubtful in cases in which innocent people are portrayed as guilty and given harsh treatment on that basis. In any event, even if the point is sound, it merely requires the retributivist to restate his worry, now objecting that utilitarian theory countenances subjecting the innocent to harms that have the appearance of punishment. The utilitarian may answer that his theory will certainly not support any announced practice of punishing the innocent. The purposes of punishment would not be served if people knew a person was innocent, and even to establish a general policy that officials would at their discretion occasionally seek punishment of those they know are innocent would cause serious insecurity.

One version of utilitarianism, called "rule" utilitarianism, makes the standard of moral evaluation the rules that would, if publicly announced, accepted, and applied, produce the best consequences. Under this version, punishment of the innocent may cease to be a problem, since no rule authorizing such punishment should be accepted. Suppose, however, that an official or citizen is sure that surreptitiously promoting the punishment of someone he knows to be innocent will be very useful. The rule-utilitarian account avoids the dilemma, but only by presupposing that proper moral decisions must be defensible in terms of rules that can be publicly announced. "Act" utilitarians, who judge the rightness of a particular action by its own likely consequences, do not have this escape. They might, however, also try to foreclose intentionally punishing the innocent as a practical alternative, pointing to the severe insecurities that would be caused by knowledge of such punishment and the difficulties of maintaining secrecy. Alternatively, they might concede that punishing the innocent would be appropriate if the balance of likely consequences were favorable, arguing that such a conclusion conflicts with moral intuitions only because those are developed to deal with ordinary situations.

Many people will feel that none of these utilitarian responses adequately accounts for the unacceptability of punishing the innocent, which is regarded as inherently wrongful. Similarly, many regard it as intrinsically unfair and morally wrong to impose severe punishment on those who commit minor crimes, however useful that might be; to give widely variant punishments to those who have committed identical offenses with similar degrees of moral guilt; or to count the interests of an offender as having as much intrinsic weight as the interests of a victim or ordinary law-abiding person.

Mixed or hybrid theories. Given these problems with unalloyed utilitarian theory, some mixture of utilitarian and retributive elements provides the most cogent approach to punishment. The basic reasons for having compulsory legal rules backed by sanctions are utilitarian; these reasons should dominate decisions about the sorts of behavior to be made criminal. Moral wrongs should not be subject to legal punishment unless that is socially useful, and behavior that is initially morally indifferent may be covered by the criminal law if doing so serves social goals. Notions of deserts, however, should impose more stringent constraints on the imposition and severity of punishment than pure utilitarianism acknowledges.

Relevance of deserts. Every practical system of punishment must admit the possibility that mistakes will lead to innocent persons being punished, but knowingly to punish an innocent person is to violate an independent moral norm. Wrongdoing alone may not be a sufficient basis to justify punishment, but the wrongful act creates a right of society to punish that does not exist with innocent persons.

Considerations of deserts should also be relevant to the severity of punishment. One possible position is that someone should never be punished more severely than could be justified both by utilitarian objectives and by the degree of his wrongdoing. Under this principle, a person would not receive more punishment than he deserves, even when that might be useful (a concept sometimes referred to as "limiting" retributivism), and he would not receive unproductive punishment, whatever his degree of guilt (a utilitarian principle sometimes referred to as "parsimony"—punishing agents should impose the least severe sanction necessary to achieve all relevant social purposes). The latter principle, however, might be seen as too rigid in some circumstances. One such circumstance involves violent offenders whose mental condition, while not excusing them altogether, does make them less blameworthy, but also renders them more dangerous and less amenable to being deterred or rehabilitated. Perhaps in an exquisitely precise system such offenders would be given a moderate criminal sentence and an extended form of civil commitment, but in the absence of such fine lines, most observers would support a criminal penalty somewhat greater than the offender really deserves. For a different reason, more severe penalties may also be warranted when those who rationally decide to commit certain crimes are very difficult to apprehend. To have a deterrent effect, the penalties may need to be greater than would be justified by the guilt of the individual offender who happens to be caught. If he has been forewarned and has chosen to take the risk, the punishment may not be unfair to him, but it may be out of proportion to the blameworthiness of his action.

In other kinds of situations, retributive concerns may make it justifiable to inflict punishment even when a balance of favorable consequences is not expected. Under an ordinary utilitarian approach, each person's welfare counts equally, but perhaps the welfare of those who intentionally commit crimes should not be given as much weight in some respects as the welfare of law-abiding citizens. The wrongdoers may, by their acts, have forfeited a right to count equally. Suppose, for example, that every one hundred executions of murderers could save seventy innocent lives. Putting aside all other relevant considerations, one might believe that those who are innocent simply have a greater claim to have their lives protected than those who have knowingly taken the lives of others, and thus, one might accept that saving seventy innocent lives is worth taking a hundred guilty ones.

As noted above, a cardinal principle of the utilitarian approach is that useless punishment should be avoided. Applying that principle may lead on occasion to selective or exemplary punishment—that is, choosing one of a number of offenders for the imposition of penalties. Imposing unequal punishments on similar offenders, however, has an element of unfairness about it, and may violate norms of uniformity or equality in the treatment of equally culpable offenders. Although punishing many people well beyond what is necessary is not warranted solely to achieve equality, when the principle of equality and the principle of parsimony (keeping punishment to the minimum necessary) come into conflict, application of the latter principle may appropriately be tempered to some degree to reduce unfairness.

Possible limits on utilitarian aims. Independent moral standards may be thought to limit not only the absolute and comparative severity of punishments, but also the nature of punishments and the utilitarian purposes that can properly be promoted by them. Torture, for example, may be ruled out on moral grounds no matter how effective it could be in particular instances. A similar position on capital punishment is taken by those who think it absolutely wrong for the state intentionally to take the lives of its members. Such a position can be consistently maintained by a thoroughgoing pacifist, but someone who accepts intentional killing in wartime and intentional killing by the police to stop terrible crimes cannot persuasively argue that execution is an unjustified punishment, however useful it is in saving lives, whatever its side effects, and however fairly it may be administered.

Some writers have urged that imposing penalties on people because of predictions of how they will act in the future is unjust. Given the impossibility of knowing whether a particular individual requires individual deterrence, incapacitation, or reformation, they note that punishment grounded on those rationales will lead to some false positives—that is, people punished unnecessarily. The existence of some false positives in itself, however, does not warrant abandoning those utilitarian aims of punishment unless all utilitarian aims, including general deterrence and norm reinforcement, are indefensible. For each of these aims, one person is suffering punishment to protect the welfare of others. No difference in moral principle exists between punishing a person to deter other potential offenders and punishing him because he is a member of a class many of whose members will commit subsequent crimes, so long as the class is fairly defined and genuinely dangerous members cannot be more precisely determined.

A more plausible attack has been made on reformation as a permissible basis for compulsory measures. The contention is that the state should not force changes in people's character and that to do so violates their right to respect as persons. The offender has, however, already violated the rights of others, and his doing so may give the state more authority to tamper with him than it would otherwise have. Insofar as the offender's difficulty is lack of skills, compulsory efforts to remedy the defect do not represent a fundamental impairment of his personality. Compelled therapy or more extreme measures, such as surgical intervention, may do so. Although one cannot rule out on absolute principle every compulsory technique designed to alter an offender's basic character, measures that would change him radically against his will do violate moral limits on what the state can properly do.

Vengeance has been thought by some to be an unacceptable basis for punishment. Taking the view that people ideally would not seek to hurt those who have done harm simply for the sake of hurting them, this view asserts that morally unworthy human desires should not be satisfied even when that will cause happiness. Whether or not the satisfaction of malicious motives should generally count positively in a utilitarian calculus, the response can be that the state legitimately satisfies feelings of vengeance both because these feelings are linked to the maintenance of healthy moral opinion (a claim discussed below) and because such feelings will find socially damaging outlets if disregarded.

Theories and practices of punishment. During the mid-twentieth century, sentencing practices in the United States were largely consonant with utilitarian premises, although also consistent with important retributive limits on severity. General deterrence and more individually focused aims were given weight, in legislative enactments as well as theoretical analyses. Because individuals committing similar offenses have different characteristics and because circumstances of offenses vary, judges were typically given considerable latitude to set initial sentences. Judicial sentences to prison tended to be indefinite (for example, two to six years), so that the time of actual release could be determined according to a parole board's estimation of the offender's progress toward rehabilitation and of his level of dangerousness prior to release.

Reformation as the keystone. One attack on this system came from those who were highly skeptical about the usefulness of condemnation and imprisonment and who placed hope instead in scientific reformation of the individual criminal. Most extreme were proposals to abolish punishment in favor of a medical model that would consider the antisocial individual as an ill person needing treatment. Appearing at first glance more humane than traditional attitudes and practices, the model's emphasis on treatment could give the state open-ended authority to achieve a cure of the antisocial person—however long that might take, however radical the necessary therapy, and whatever his original wrong. Full acceptance of the model might also lead to compulsory treatment of those identified as socially dangerous before they commit harmful acts, a prospect of social intervention that many people perceive as a denial of human dignity and autonomy. Because of the medical model's potential for sweeping intrusion into citizens' lives, and because few people with political power have been willing to give up the possible benefits of condemnation and fear of penalties, proposals for abolition of punishment have never won wide acceptance.

A more moderate reformist position was that within a system of punishment, heavier concentration should be placed on reform, with the length of sentence to depend even more on the rate of rehabilitative progress. The movement to emphasize reform had already influenced American sentencing practices, promoting more flexibility in prison terms than had previously existed and more attention to the quality of programs within prison; but some believed that genuine change had not gone nearly far enough.

Just deserts. During the 1970s there was a sharp reaction against the emphasis on rehabilitation. Despairing over achievement of earlier reformers' goals, critics of existing practices argued that rehabilitation had largely proved a failure, that prison was more likely to harden criminals than to cure them of antisocial tendencies, and that parole boards were almost wholly unable to judge which prisoners were fit to be set loose on society. Flexible sentences, it was said, caused prisoners acute anxiety over their future, encouraged them to feign attitudes and emotions they did not feel, made them prey to the arbitrary dictates of prison officials and parole boards, and engendered in them a sense that the system was unfair in fundamental respects. Critics also contended that in practice, reliance on individual predictions to imprison persons was unjust, since many of those considered dangerous would not commit crimes if released, and since many of those viewed as apt candidates for individual deterrence or reformation are not rendered more law-abiding by confinement. Broad discretion and unequal treatment of similar offenders were challenged because they were intrinsically unfair and because disparities failed to contribute to utilitarian objectives and caused deep resentment among those convicted.

Although not rejecting general deterrence as a proper aim of punishment, proponents of the "just deserts" model urged that sentencing should seek to achieve the goals of proportionality and uniformity: penalties for particular offenses should depend mainly on the severity of the offenses, and those who commit similar offenses should be given equal, or nearly equal, treatment. Terms of imprisonment would be squarely fixed at the time of sentence. In-prison efforts to help convicts should continue, but participation in therapy or job training would no longer be relevant to the timing of release.

In certain respects the just deserts model seems obviously overdrawn. Whatever the uncertainties of individual prediction, confining those whose dangerousness is attested by repeated violent crimes must be acceptable, and the plausibility of rejecting this ground rests on the implicit assumption that some other basis for imprisonment will be available for these persons. (One such alternative in fact began to be used with increasing frequency at the close of the twentieth century: indefinite civil commitment of sexually dangerous persons.) Moreover, making penalties equal for similar offenders would disregard both the significant differences between those who commit the same offense, and the value of avoiding useless punishment (sentencing "parsimony").

Nevertheless, in their positive emphasis on the values of equality and perceived fairness, in their distaste for arbitrary discretion, in their skepticism about rehabilitation induced by the implicit promise of release, and in their realism about the effects of prison life, just deserts theorists and other critics of highly discretionary sentencing systems have made a substantial contribution to thought about punishment. They have also encouraged reforms in many states (in particular, the enactment of sentencing guidelines or other determinate sentencing laws) that have reduced judicial discretion and reduced or eliminated parole board discretion to determine date of release. However, even in those states, such as Minnesota, whose reformed sentencing systems were strongly influenced by the just deserts model, individualized sentencing based on the goals of rehabilitation and risk management continues to play a very substantial role in the design of formal rules and, especially, in actual practice.

Utilitarian bases for retributive perspectives. The modem debate over sentencing practices raises a more general theoretical question: May official decisions based on retributive premises be socially useful? The idea is that since people naturally think in retributive terms, they will be disenchanted and eventually less law-abiding if the law does not recognize that offenders should receive the punishment they "deserve." Although love for one's enemies may be a moral ideal, perhaps most people cannot feel strongly committed to a moral code without also wanting to see those who break that code punished. If the complex psychological and sociological assumptions that underlie this view are accurate, utilitarianism and retributivism may subtly blend. The ultimate philosophical justification offered for punishment would be promotion of human good, a utilitarian justification; but a retributive outlook among citizens would be welcomed and the operating official standard for punishment would be retributive. This apparent paradox is but an example in the context of punishment of an idea that has often been discussed in connection with utilitarian theory—the possibility that human welfare will be best advanced if people subscribe to a more absolutist morality than one which makes the promotion of good consequences the test of an act's rightness.