UNINTENTIONAL PUNISHMENT

Abstract

Criminal law theorists overwhelmingly agree that for some conduct to constitute punishment, it must be imposed intentionally. Some retributivists have argued that because punishment consists only of intentional inflictions, theories of punishment can ignore the merely foreseen hardships of prison, such as the mental and emotional distress inmates experience. Though such distress is foreseen, it is not intended, and so it is technically not punishment. In this essay, I explain why theories of punishment must pay close attention to the unintentional burdens of punishment. In two very important contexts—punishment measurement and justification—we use the term “punishment” to capture not only intentional harsh treatment but certain unintentional harsh treatment as well. This means that the widely accepted view that punishment is an intentional infliction requires substantial caveats. It also means that any purported justification of punishment that addresses only the intentional infliction of punishment is woefully incomplete.

I. INTRODUCTION

In Venezuela's San Antonio prison on Margarita Island, inmates drink whisky and barbecue poolside while visited by their spouses, girlfriends, and children. People on the island voluntarily enter the prison to party, bet on cockfights, and buy drugs. Supervision is so lax at the prison that other than staying on prison grounds, inmates have virtually no restrictions.¹

In the Ohio State Penitentiary, by contrast, prisoners must remain in their small cells, with lights on, for 23 hours per day.² Inmates in this “supermax” prison have few opportunities to communicate with each other or with outside visitors. According to Justice Anthony Kennedy, the “[i]nmates are deprived of almost any environmental or sensory stimuli and of almost all human contact.”³

Though prison conditions vary substantially, we generally ignore the variation when assessing punishment severity. We fetishistically focus on the length of prison terms, even though sentence severity cannot just be a function of time. Surely a five-year sentence served at San Antonio prison is less severe than a five-year sentence served at the Ohio State Penitentiary.

Even identical prison facilities have very different effects on prisoners. One inmate may become extremely distressed, whereas another thrives in the very same facility. Though we do not necessarily intend to cause such distress, bad experiences are clearly foreseen side effects of incarceration. Nevertheless, we generally ignore the varied ways offenders experience punishment when assessing the severity of their sentences.

Punishment theorists overwhelmingly agree that in order for some conduct to constitute punishment, it must be intentional. Nevertheless, even if the unintended side effects of punishment are technically not punishment, the state has a moral obligation to take account of the actual or expected ways in which punishment affects inmates’ lives.⁴

Precisely how the state should take account of these effects depends on one's theory of punishment. But whether one is a retributivist or a consequentialist, the state has what I will call a “measurement obligation.”⁵ Namely, the state must measure the unintentional harms associated with punishment in order to make sure that the punishment is just. I will also suggest that, under some leading views of punishment, we have an even stronger “calibration obligation” that requires us to reduce the purposeful inflictions of punishment by an amount that reflects certain nonpurposeful inflictions.

Some scholars deny these obligations.⁶ They claim that we need not address unintentional aspects of punishment, like the bad experiences associated with incarceration, because these side effects are not imposed intentionally and are therefore not punishment. Even if we knew precisely how sentences would affect prisoners, so the argument goes, as long as conditions are humane and we do not intend to treat sensitive and insensitive prisoners differently, we need not consider how prison affects them.

In two very important contexts, however, I argue that the term “punishment” includes not only intentional harsh treatment but certain unintentional harsh treatment as well. Hence the widely accepted view that punishment is an intentional infliction requires substantial caveats. Moreover, any justification of punishment that purports to address only intentional aspects of punishment is seriously incomplete.

As I argue in Section II, “Measuring Punishment,” our intuitions about the severity of punishment take into account more than just intentionally imposed hardships. Consider two equally blameworthy offenders, Purp and Fore. They are alike in all pertinent respects and receive identical sentences in identical prisons. The only difference between them is that different aspects of their sentences are imposed intentionally. Purp is purposely limited in his liberty to move about, see family, have sex, express himself, possess personal property, vote, and so on. By contrast, Fore is purposely limited in moving about, but all of his other hardships are merely foreseen accoutrements of prison. Because these other hardships are not imposed purposely, they are technically not part of Fore's “punishment” as scholars frequently understand the term.

Despite the different intentions that surround their treatment, we tend to think that Purp and Fore are punished by the same amount. The mental states of their punishers (be they judges, prison personnel, legislators, voters, or some combination of all of these) do not affect the severity of their sentences. So long as the duration of their sentences and the conditions of their confinement are the same, we think that they receive the same amount of punishment. Thus, when assessing amounts of punishment, we consider not only intentional hardships but also certain unintentional hardships as well.

In Section III, “Justifying Punishment,” I note that in addition to measuring amounts of punishment, we must also justify whatever amounts of punishment we impose. Because real-world punishments such as incarceration involve both intentionally inflicted harms as well as others that are merely foreseen, I argue that any justification that refers only to intentional inflictions cannot justify the unintentional inflictions of incarceration that also require justification.

To justify a punishment practice such as incarceration, one must satisfy what I call the justification-symmetry principle. According to this principle, any state actor who harms an offender in the name of just punishment must have a justification for doing so if you or I would need a justification for causing the same kind of harm to nonoffenders. Because you and I must justify all or most of the harms that we purposely, knowingly, recklessly, or negligently cause others, the justification-symmetry principle requires that state actors do the same. A defense of punishment that purports to justify only purposeful inflictions cannot justify the many other harms we forseeably inflict when we incarcerate.

In Section IV, “Splitting Punishment,” I consider a strategy to avoid the measurement obligation that may be particularly appealing to retributivists. Antony Duff has recognized the need to justify a wide range of state coercive practices but claims that state punishment—limited to intentional inflictions—warrants distinctive attention apart from “the actual impact and effects of punishment,” which, he notes, vary “enormously.”⁷ Consistent with Duff's suggestion, retributivists could concede that the side effects of punishment require justification but argue that the justification need not be a justification of punishment. Rather, they might argue, the justification of side effects comes from some other moral or political theory.

I argue that this approach makes retributivism so anemic that it cannot, by itself, justify punishment practices like incarceration. It also leaves retributivists subject to the measurement obligation and, quite likely, the calibration obligation as well. If the shadow theory justifies side-effect harms based on offender desert, it is not at all clear why offenders only repay their desert debt when they receive intentional inflictions. Rather, it would seem, as foreseen inflictions accrue, we must ratchet down intentional inflictions so as not to give offenders more punishment than they deserve. Alternatively, if the shadow theory is consequentialist, retributivists must give up the traditional notion of proportionality between blame and punishment. It would be a cosmic coincidence if consequentialism justified side-effect harms that are precisely equal in duration to the intentional inflictions dictated by retributivism.

Although some scholars have recognized that retributivism does not provide a complete justification of real-world institutions of state-imposed and -financed punishment,⁸ I make a more damaging claim: even if we put aside cost and administrative concerns, principles of retributive proportionality cannot even justify the amount of prison time an offender should serve because they cannot justify the unintentional hardships of prison. I take it that even those retributivists who believe that retributivism fails to justify the allocation of resources in the criminal justice system or fails to provide a general justifying aim of punishment still believe that retributive principles of proportionality can tell us, at least in principle, how long to incarcerate deserving offenders. I show otherwise.

II. MEASURING PUNISHMENT

Criminal law scholars widely agree that in order for some conduct to constitute punishment, it must be intentionally imposed.⁹ H.L.A. Hart, perhaps the most influential punishment theorist of the twentieth century, famously claimed that a central feature of punishment is that it is “intentionally administered”¹⁰ and “deliberate[ly] impos[ed].”¹¹

Courts, too, have given special emphasis to the intentions behind certain conduct when deciding whether it constitutes punishment. Imprisoning an offender is clearly punishment, but some other state-imposed burdens are more debatable. For example, the U.S. Supreme Court has held that sex offenders can be indefinitely confined after they complete a prison term without being punished for purposes of the Ex Post Facto Clause of the Constitution.¹² Similarly, the Court has held that in many instances, property forfeitures, fines, and requirements to give up one's occupation are not punishments for purposes of the Double Jeopardy Clause.¹³ In deciding that these forms of harsh treatment are not punishment, the Court has put particular weight on the presumed lack of punitive intent on the part of the legislature.¹⁴

Judge Richard Posner has also noted the importance of intentions when assessing whether conduct is punitive. According to Posner:

If a guard decided to supplement a prisoner's official punishment by beating him, this would be punishment, and “cruel and unusual” because the Supreme Court has interpreted the term to forbid unauthorized and disproportionate, as well as barbarous, punishments. . . . But if the guard accidentally stepped on the prisoner's toe and broke it, this would not be punishment in anything remotely like the accepted meaning of the word.¹⁵

Talk of “intentional” action is itself ambiguous. If we transfer an inmate to solitary confinement with the conscious goal of making him suffer, the additional harsh treatment constitutes punishment. But what if we transfer the inmate to protect him from others? In such a case, we foresee that the transfer will increase hardship, though we are not transferring him in order to make him suffer. In such a case, the inmate's additional hardship is a nonpurposeful but still known side effect. Discussions of intentional inflictions do not always make clear whether they refer only to purposeful inflictions or to both purposeful inflictions as well as those that are nonpurposeful but still foreseen.¹⁶

The view “almost universally accepted in the literature on punishment,” writes David Boonin, is that only purposeful inflictions can count as punishment.¹⁷ According to Boonin:

When the state punishes someone . . . it inflicts various harmful treatments on him in order to harm him. It is not merely that in sentencing a prisoner to hard labor, for example, we foresee that he will suffer. Rather, a prisoner who is sentenced to hard labor is sentenced to hard labor so that he will suffer.¹⁸

I do not know if Boonin is correct that almost every theorist believes that conduct we call “punishment” must be purposely administered. I address my claims here, however, to those who do. For once you concede that punishment includes foreseen inflictions, then clearly you must measure foreseen inflictions in order to know how much a person is punished. If two offenders are equally blameworthy, yet we foresee that the more sensitive offender will experience substantially more hardship than the other, we must take that additional punishment into account if we seek to dispense proportional punishment.

In this section, I argue that when we consider the severity of a punishment, we consider more than just those aspects of punishment that are intentionally administered.

A. Preliminary Matters

Although criminal law scholars overwhelmingly believe that conduct qualifies as punishment only when it is imposed intentionally, it can be difficult to distinguish harms we cause intentionally from those we merely foresee. The harms of incarceration, for example, result from a complicated series of actions by many people over an extended period of time.

Even if we can distinguish intentional and merely foreseen harms, it is not clear why the distinction has any intrinsic moral salience.¹⁹ Indeed, we often treat purposefully and knowingly caused harms the same way. A person can kill his house guest by setting his home ablaze so that the guest will be engulfed in the flames. Or, he can set his home ablaze with the intent to collect insurance money, while merely knowing that the fire will kill the house guest. The law deems the conduct murder either way.²⁰ For the sake of argument, however, I will assume that there are morally salient differences between intending and foreseeing harm.

Scholars have reached no general agreement about whose intentions are relevant when assessing whether conduct constitutes punishment. In what follows, I first consider the possibility that the intentions of sentencing judges affect whether conduct ought to be considered punishment, at least when the legislature is silent on the matter and gives judges broad discretion. But where I speak of the mental states of judges, we can easily substitute the mental states of prison administrators, legislators, the citizenry, or some combination of all of them. The choice does not affect the substance of my claim that when we consider the severity of a punishment such as incarceration, we consider not only harms to offenders that are intended but also many that are foreseen.

B. Absurdity of Counting Only Purposeful Harms

To see why punishment severity must include certain foreseen inflictions, consider in more detail the thought experiment I mention in Section I. Suppose that Purp and Fore commit crimes for which they are equally blameworthy. They are alike in all pertinent respects except that they are sentenced separately by Judges Purpose and Foresight, respectively. Both judges have the discretion to sentence these offenders to zero to five years in prison. Furthermore, statutes in this jurisdiction, like the statutes in many real-world jurisdictions, say little about the reasons for incarcerating offenders, other than to give judges broad discretion to sentence in accordance with a wide range of possible punishment rationales.²¹ As it happens, both judges issue three-year sentences, and both Purp and Fore will serve their identical three-year terms in identical prison conditions.

It may seem as though Purp and Fore will be punished equally. After all, they will spend the same amount of time in identical prison conditions. And because they are alike in all pertinent respects (except for their sentencing judges), they will even experience the same suffering and deprivations in prison. But if we accept the claim that punishment is limited to intentional inflictions, then I have not told you nearly enough to know whether or not their punishments are equal. It would all depend on the intentions of their punishers.

Judge Purpose is a former corrections officer who has firsthand knowledge of prison life. He sentences offenders to prison when he wants them to be subjected to a wide variety of hardships. He intends not only that inmates be deprived of their liberties of motion but that they also have very limited opportunities to see family, have sex, express themselves, possess personal property, be entertained, vote for elected officials, and so forth. In short, Judge Purpose purposely inflicts many of the hardships associated with prison life.

Judge Foresight, by contrast, was a bankruptcy attorney before joining the judiciary. He has comparatively vague ideas about what life is like in prison. When he sentences offenders, his purpose is merely to limit inmates’ freedom of motion to the confines of a prison. He knows and cares little about the details of prison life: whether prisoners have shared cells, cable television, Internet access, conjugal visits, or opportunities to see friends and family. Either he does not think much about the hardships of prison life or he does think about them but, after careful reflection, decides that they are mere accoutrements of the burdens of prison that are not part of the punishment he intends to mete out. In short, Judge Foresight only foreseeably inflicts most of the hardships associated with prison.

Although we would ordinarily think that offenders like Purp and Fore are being punished equally, they are being punished quite unequally according to a literal understanding of the way many theorists understand the term “punishment.” Purp will receive many purposeful inflictions of harm in prison over a three-year period, whereas Fore will receive far fewer. If only purposeful inflictions count when assessing punishment severity, we are led to the surprising conclusion that Fore is dramatically underpunished relative to Purp.

If you find quantitative illustrations helpful, assume that under the conditions described, the following purposeful inflictions impose the corresponding number of units of punishment on offenders like Purp and Fore:

Since Purp and Fore are both purposely limited in their liberty of motion to staying on the prison grounds for three years, they will both receive the corresponding 50 units of punishment by the time their sentences end. If only purposeful inflictions count when assessing punishment severity, then Fore receives only those 50 units of punishment because that is the only aspect of his sentence imposed purposely. Purp, however, will be subject to the additional five purposeful inflictions listed. Hence, Purp will receive 100 units of punishment relative to Fore's 50 units, even though they will both spend three years in prison under what will seem to most observers like identical conditions.

We can make the example even more absurd. Suppose that while they are incarcerated, the state appeals Fore's sentence, arguing that it is far too lenient; his punishment is only half that of offenders like Purp. As the three-year term of their prison sentences nears its end, a higher court finally considers the appeal. In his defense, Fore points out that, intentionality aside, his treatment was identical to Purp's. Moreover, Fore argues, if he must be resentenced, the court should simply redescribe the hardships he has already experienced as purposeful inflictions. Lastly, he argues, the severity difference between his sentence and one like Purp's is so small that it can be treated as trivial.

On the day Fore is scheduled to be freed, the appellate court releases its decision. Accepting the view that punishment is a kind of purposeful infliction, the court sides with the prosecution. The court writes, “Fore's punishment was far too lenient, having received only half the punishment of other similarly situated offenders. Moreover, we cannot simply recharacterize Fore's punishment any more than we can recharacterize payment of a tax as payment of a fine or recharacterize years spent confined in a mental institution as years spent punished in prison.”

The court remands the case with detailed instructions about how Fore should be sentenced. According to this hypothetical appellate court, because Fore's prison sentence has ended, he has already been purposely restricted in his liberty of motion. Having satisfied that aspect of his sentence, he should be released as planned. Although he will now be free in some respects, however, the state will purposely inflict the hardships that Judge Foresight mistakenly treated as merely foreseen. Therefore, for the next three years, Fore will be subjected to “forced deprivation of family,” whereby he will not be permitted to see his family except infrequently and from behind a glass wall. He will be subjected to “forced poverty,” whereby he will be required to give up access to all of his personal property, except for the bare essentials like towels, linens, and a toothbrush. He will be subjected to “forced celibacy,” in which he will be forbidden to have sexual relations with any other person. He will be subjected to “forced deprivation of media,” in which he will be forbidden most access to television, computers, and books, except for the Bible and some others on a short preapproved list. And so on. As for the claim that the difference between Purp's and Fore's sentences is trivial, the court writes, “we believe that Fore will find the purposeful inflictions that await him over the coming years to be anything but trivial.”

Naturally, the appellate court's decision strikes us as ridiculous. We think that the severity of Purp's and Fore's sentences are essentially the same despite the different intentions of their sentencing judges. Although being an intentional infliction may be a requirement for treating some conduct as punishment, when assessing punishment severity, we seem to count far more than just purposeful inflictions. If we treat only purposeful inflictions as punishment, the severity of prison sentences will depend more on the intentions of judges (or legislators or prison personnel or voters) than on prisoners’ actual conditions of confinement.

We pay substantial attention to the duration of prison sentences and very little attention to which aspects of incarceration are imposed purposely. If intentions were so important to establishing the appropriate treatment of offenders, we would expect judges (and legislators) to describe precisely what aspects of an incarcerative sentence are intentional. The fact that they do not and that we do not expect them to casts doubt on the idea that only intentional inflictions contribute to punishment severity.

C. Responses to Possible Objections

I now propose and respond to three objections that aim to show that punishment severity can be understood solely in terms of intentional inflictions.

1. Many-Intentions Objection

As I suggest above, one might argue that the mental states of Judges Purpose and Foresight should not control our understanding of the nature of the treatment of Purp and Fore. According to this objection, there are many actors involved in punishment other than just judges, and so I overstate the differences in Purp's and Fore's punishments by giving such a prominent role to the mental states of their sentencing judges.

Although it is true that judges’ mental states are not the only ones that are relevant, it seems hard to ignore them completely. Indeed, recent revisions to the Model Penal Code reflect the “underlying philosophy . . . that sentencing is, at its core, a judicial function.”²² In the case of Purp and Fore, the judges were granted substantial discretion by the legislature to sentence these offenders to a term of zero to five years to achieve a variety of permissible ends. The judges even had the discretion to provide no punishment at all, making them extremely important causal contributors to Purp's and Fore's treatment. It would be odd for judges to have so much control over Purp's and Fore's punishments and yet have their intentions be entirely irrelevant. If anyone's intentions are relevant here, it seems that those of their sentencing judges must be. Some jurisdictions cabin judicial discretion to greater degrees, thereby giving legislators or sentencing commissions greater control over offenders’ sentences. But no U.S. jurisdiction eliminates judicial discretion entirely.

We would have some difficult determinations to make if a judge considers some burden on offenders merely foreseen whereas prison administrators consider it intentional. But I state that Purp and Fore are alike in all relevant respects except for which judge issued their sentences. So we could assume that they were sentenced under the same legislation and overseen by the same prison personnel. In other words, I hold everything constant but the intentions of the relevant sentencers. So unless judicial intentions are entirely irrelevant, we are implausibly led to believe that there is a substantial difference between the severity of the punishment of Purp and Fore.

More importantly, as I warn before giving the example, we could preserve my point without relying at all on the intentions of particular judges. Imagine that Purp lives in a jurisdiction where legislators, judges, and prison personnel view incarceration the way Judge Purpose does and that Fore lives in a jurisdiction where everyone views incarceration as Judge Foresight does. Nevertheless, we will still think that Purp and Fore are punished equally when sentenced to equal terms at identical facilities, despite the differences in the operative intentions in their jurisdictions.

2. Objective- or Normative-Intentions Objection

A second potential objection says that we should think of the intentionality relevant for analyzing punishment as distinct from the totality of human intentions associated with it. For example, Alice Ristroph writes “[r]arely can a single coherent intent be attributed to the entire institutional apparatus that imposes punishment. The intentions of individual officials within the criminal justice system may be relevant to, but are not dispositive of, the question whether the system is imposing punishment.”²³ If so, perhaps Purp and Fore were punished the same relative to some more objective notion of intention.

Although it is certainly true that the intent of any particular official may not be dispositive of whether the system is imposing punishment, it is hard to see how the intentions of all of those involved in the punishment process are not dispositive of whether the system is punishing, even if we are not sure how to aggregate the data. After all, if the intentions of all human beings are irrelevant, then scholars mischaracterize punishment by saying that it must be intentionally administered. On such a view, scholars are not actually discussing intentions at all.

One might argue that a full justification of punishment will tell us precisely which aspects of incarceration should be imposed purposely and which merely knowingly. According to this view, Purp and Fore happen to have very different punishments only because we have yet to work out in more detail which aspects of incarceration should be purposeful and which merely foreseen.

On the contrary, however, the example is not about the justness of Purp's and Fore's sentences; it is about their severity. Perhaps someone will develop a theory that tells us precisely which aspects of incarceration should be inflicted purposely and which merely knowingly. And perhaps some of the intentional inflictions in Purp's case would be found unconstitutional. That will not change the fact that the hardships in Purp's initial sentence were very much purposeful and those in Fore's initial sentence were mostly just foreseen. Whether these hardships are just or unjust, legal or illegal, we think the severity of their original sentences was essentially the same. Thus we count at least certain foreseen inflictions as part of the severity of a punishment.

3. Public-Perception Objection

A third and final objection says that Purp and Fore received equal punishments (even though their judges intended different hardships) because we should judge the severity of their sentences based on public perceptions of their punishments. Because their sentences will look the same to the public, they are of equal severity.

We can, however, readily distinguish between the actual severity of a sentence and the public perception of its severity. If the public mistakenly believes that Purp is placed in solitary confinement whereas Fore remains in the general prison population, we would not consider Purp's sentence more severe than Fore's. Rather, we would say that the public was simply mistaken in its judgment of the severity of their sentences. Public perceptions of severity are simply irrelevant to measurements of actual severity.²⁴ So we cannot avoid the conclusion: even though punishment theorists traditionally understand punishment as a kind of intentional infliction, our assessments of punishment severity also take into account many of the foreseen consequences associated with incarceration.

III. JUSTIFYING PUNISHMENT

In Section I, I explain why our intuitions about the severity of punishment require us to examine both intentional and merely foreseen aspects of punishment. There is an even more powerful reason, however, for punishment theorists to examine unintentional aspects of punishment. Those who attempt to justify only the purposeful inflictions of punishment cannot possibly justify real-world punishment practices such as incarceration because they fail to justify all of the nonpurposeful inflictions that are necessarily associated with it.

To illustrate the obligation to justify side effects of punishment, suppose that a conjoined twin commits a crime on the Internet without the knowledge or support of his conjoined sibling. We could punish the perpetrator by imprisoning him for a period of time proportional to his blameworthiness, but doing so would, of course, have an unintended side effect. It would foreseeably lead to the unintentional harsh treatment of the innocent twin. Even though the innocent twin would not technically be punished in prison (his harsh treatment would not be intentionally imposed as punishment), we would still need some justification for knowingly harming him. Unlike the perpetrator's confinement, however, the innocent twin's confinement cannot be justified as proportional punishment because he did nothing wrong.

Real cases of punishing conjoined twins are rare.²⁵ The issue raised by the conjoined-twin case, however, is not rare at all. In fact, real-world punishments virtually always have both intended consequences and those that are merely foreseen. For example, prisoners often have spouses, parents, children, friends, and other loved ones from whom they are physically separated during their sentences except for occasional visits often behind glass barriers. These loved ones are not themselves culpable, but they are unintentionally made to suffer by being separated from a beloved prisoner. They are conjoined to the prisoner not by body parts but by emotional attachments that can be just as strong. Justifications of punishment (when understood only as a kind of purposeful infliction) say nothing to justify the foreseen but unintended harms inflicted on family members.

Others have recognized that punishment has negative side effects on nonoffenders, such as the erroneously convicted, the family members of justly convicted inmates, and the taxpayers who finance the criminal justice system.²⁶ What theorists have not adequately recognized is that real-world punishments involve both intentional and unintentional harms to even a single offender. Incarceration purposely causes the loss of an offender's liberty of motion but also causes other suffering and deprivation, such as loss of privacy, sexual autonomy, rights to personal property, and rights to freedom of association and expression. Although some scholars may recognize the need to justify side effects that burden nonoffenders, they have largely ignored the side effects that burden offenders themselves.

In this section, I explain why these unintended hardships of incarceration still require justification. My criticism focuses on retributivist theories of punishment because they are more likely than consequentialist theories to identify a stark difference between purposeful and nonpurposeful inflictions. By focusing on the justification of purposeful punishment rather than the justification of both purposeful and nonpurposeful harsh treatment, retributivists have established a task for themselves that is comparatively too easy. They have carved off the vast majority of our punishment-related conduct from their justification of punishment, meaning they work to justify a set of behaviors that are only a small subset of our real-world punishment practices. After arguing that a justification of punishment restricted to intended harms is inadequate, I describe some minimal requirements that a good justification of punishment will satisfy.

A. The Justification-Symmetry Principle

When we incarcerate offenders, we confine them in small spaces and thereby limit their freedom to move about, associate with others, maintain their privacy, decide what to eat, and so on. Such harsh treatment is ordinarily forbidden outside the punishment context. To justify a punishment practice such as incarceration, we must show why involuntarily confining someone—behavior that would ordinarily be criminal, tortious, and immoral—is actually justified. We need to justify such harsh treatment because it would be wrong to cause distress or restrict people's freedoms in these ways without good reason. Importantly, we seek a moral rather than a legal justification; incarceration is clearly a legally permitted sanction for a wide range of offenses.

Perhaps we should be able to morally justify all of our conduct that causes people harm, unless it was entirely without fault. So even in cases where we cause only minor harm (as when we gently tease), we could be called upon to justify our behavior (citing, for example, autonomy interests in being able to gently tease). Others may have less demanding views about the nature of moral justification, believing that we must justify only certain harms we cause.

What I call the justification-symmetry principle provides a (not necessarily exclusive) test of whether some punishment-related conduct requires justification. The principle says that if you or I must have a justification for risking or causing some harm, then so must any person who risks or causes the same kind of harm in the name of punishment. In other words, a complete justification of punishment will tell us why, by virtue of being just punishment, some ordinarily impermissible behavior is made permissible.

The principle derives from the very reason we seek to justify our punishment practices. A justification must tell us the moral distinction between a just punishment practice and similar-seeming criminal or immoral behavior. When we can demand a moral justification from you or me for harming someone, then we can make a symmetrical demand of those who cause the same kind of harm in the name of just punishment.²⁷ (For simplicity, I refer to those who purport to inflict just punishment as “state actors,” though it is debatable whether retributive punishment must be inflicted by the state.)²⁸

To illustrate, consider forcible confinement. If you or I confine someone against his will, we clearly need a justification. Indeed, without a justification, doing so would be criminal. The justification-symmetry principle says that because forcible confinement would require a justification if you or I did it, then it also requires a justification when engaged in by state actors such as police and corrections officers. By contrast, because you and I do not ordinarily need a justification for making small talk with another person, the principle does not require police officers to justify the small talk they make with prisoners.²⁹

There is a second feature of the justification-symmetry principle: the justification for the conduct of the state actor must consist of more than just the fact that he is a state actor. In the world around us, we see police and others in the criminal justice system subjecting people to harsh treatment. We can reasonably wonder why police uniforms and other trappings of government authority turn what would ordinarily be an impermissible action into a permissible (and perhaps even laudatory) action. If we are simply told, “prison guards can use coercive force because they are state actors,” then we have not received an adequate justification of punishment. The justification must explain why the guard is acting permissibly.

The justification-symmetry principle helps us identify what a justification of punishment should do. It specifies part of the punishment justificandum, meaning “that which must be justified” by a theory of punishment. In summary, the principle says that we must justify the conduct of state actors at least in cases where we expect a justification of the same kind of conduct by nonstate actors. The justification must provide reasons the state is permitted to engage in actions that we ordinarily consider forbidden.³⁰

Of course, just because some state action requires a justification does not mean that it cannot be justified. It merely means that a proposed justification of punishment must address the matter. To the extent that theorists ignore harms that must be justified, their theories are incomplete.

B. Incarceration Causes Nonpurposeful Harms

Some aspects of incarceration do involve purposeful inflictions. If a judge sentences an offender to four years in prison, it is presumably the judge's purpose to restrict the offender's liberty of movement for four years. But as suggested above, it may or may not be the judge's purpose to limit the offender's food choices, restrict his sexual freedoms, eliminate his right to vote, or make him miss his daughter's first soccer game.

1. Foreseen Infliction of Harm

Foreseen inflictions of harm on offenders frequently require justification whether or not we call these inflictions “punishments.” Three examples follow. First, consider that most prisoners are prohibited from having sex with other people while incarcerated.³¹ They may not have sex with other inmates, prison personnel, or visiting spouses or lovers. The restriction is probably not intended to harm offenders. In fact, a stand-alone punishment of “forced celibacy” outside prison might be unconstitutional.³² But legislators, judges, and prison officials surely foresee harms to prisoners who lose their sexual autonomy. Many of these officials would permit sexual relationships if they could do so without danger or high administrative costs. For such officials, prison involves knowing but unintended restrictions on sexual autonomy.

Yet even if restrictions on prisoner sexual autonomy are unintended, they still require justification. If you or I were to knowingly restrain people from engaging in adult, consensual sexual relationships using force or the threat of force, we would generally be engaged in a crime, a tort, and an immoral action.

As a second example, suppose that a judge who formerly worked as a prison social worker firmly and quite accurately believes that a particular offender he is sentencing will experience severe depression for the first several months that he is incarcerated. When this judge passes sentence, he foresees that the offender will develop severe depression in prison, though it is not his purpose to cause the depression. In fact, if the judge had some miracle cure for depression, he would happily use it.

The foreseeable infliction of severe depression requires some justification. If an evil psychiatrist knowingly causes someone severe depression because he is the only psychiatrist in town and wants to sell more counseling services, he needs a justification and probably does not have one. Because nonpunishers like you and me need a justification for knowingly causing severe depression, so do state actors under the justification-symmetry principle.

The fact that you or I would need to justify knowingly causing someone severe depression does not mean that we are never permitted to do it. A small business owner may knowingly cause an employee severe depression by terminating him because of his consistently subpar job performance. Terminating the employee requires a moral justification, though there may well be one. Perhaps the termination is necessary to save the business. Or perhaps it will maximize profits in a competitive capitalist marketplace. Whatever the asserted justification is for knowingly causing the harm of depression, we can expect a justification and then evaluate whether the asserted justification is successful. Similarly, punishers cannot simply ignore the severe emotional pain that they cause others without reasonably being asked to justify their behavior.

My third example concerns a judge who knowingly separates an offender from his family.³³ Even when state actors only knowingly separate offenders from their families, they must still justify the separation. Suppose a bank manager requires her subordinate to work outside the country for a year. If the manager knows the employee will rarely if ever see his family while abroad but gives it no consideration, we can reasonably fault the manager for her thoughtlessness. She should have a justification for both the harm to the parent from the separation as well as the harm to his children and other loved ones. Perhaps the foreign transfer can be justified for other reasons, but we should doubt that the manager is acting in a justified manner when she fails to consider clearly important data.

Similarly, when a judge sentences an offender to a year in prison knowing that the offender will rarely if ever see his child during this period, the judge should have some justification for the harm to both the parent and the child caused by the separation. Of course, the judge not only makes it difficult for parent and child to see each other but plays a role in physically preventing their interaction. So the separation of parent and child in the prison context is especially in need of justification.

These three examples illustrate knowing inflictions of harm that you or I would have to justify and that the government must also justify under the justification-symmetry principle. You need not be convinced by any one of them, but unless you think that there simply are no knowing inflictions of harm on offenders that require justification, then you should agree that a justification of purposeful inflictions cannot fully justify a punishment practice like incarceration that includes numerous nonpurposeful hardships.

2. Foreseen Infliction of Risks

Although we sometimes foresee harms that are virtually certain to befall a prisoner, other times we merely foresee risks of harm. For example, a judge might reasonably foresee that a particular offender will face a higher risk of physical or sexual violence in prison than outside prison. If so, the state ought to have some justification for increasing the offender's risk of harm.

To see why, consider the manager of a boarding home. He knows that the occupants have limited resources and will be unlikely to leave. One day, the manager has the opportunity to fill a room with a boarder who foreseeably presents a significantly heightened risk of physical or sexual violence to the other boarders. Perhaps the boarder has a history of being repeatedly incarcerated for physical and sexual assaults. Assuming the manager has the discretion to decide, should he accept the new boarder, even though he is consciously aware that doing so will raise the risk of violence to the other boarders? Maybe yes or maybe no. The point again is that he must have some justification for increasing the risk of physical and sexual violence to other people;³⁴ and under the justification-symmetry principle, so must the state when it sentences an offender to a prison with a risk of violence higher than what the offender would face outside of prison.

3. Careless Infliction of Risks

Most people believe that we are morally accountable when we fail to foresee risks of harm that we should have foreseen. In fact, much of the tort system is devoted to failures to take adequate precautions, even when tortfeasors are unaware of the need to do so. Not everyone believes people should be deemed culpable for their negligence, at least in the criminal context.³⁵ I take no stand on the issue, except to say that if you or I can be blameworthy for careless behavior, then so, too, can state actors.

So if a judge sentences an offender to prison when doing so is unjustified given the increased risk of physical harm the offender will face in prison, the judge may be deemed blameworthy for having done so if the judge should have been aware of the risk. Confining people against their will is a serious action. Those who do so should have a good sense of the harms they risk causing. When state actors are unaware of risks that they should have been aware of, they can be faulted for their negligence.

Moreover, because incarceration is a punishment that extends over time, many foreseen or foreseeable risks of incarceration (like the possibility that an offender will be exceptionally distressed by close quarters) will ripen into conditions where the state knows it is permitting more harm than originally anticipated. Just as you or I can develop duties to monitor the consequences of our ongoing activities, so too can the state.

On some views, a lack of justification is embedded into the very meaning of negligence (and recklessness).³⁶ Clearly, one cannot justify behavior that, by definition, requires an absence of justification. Even so, the state still has an obligation to provide compensation or make other efforts to remedy its risk-causing behavior when you or I would have such obligations under similar circumstances.

4. Proximate Causation

Under the law, we sometimes limit people's responsibility for the consequences of their actions to those that are deemed “proximate.” In assessing whether an actor is the proximate cause of some result, for example, the Model Penal Code asks whether the result “is not too remote or accidental in its occurrence to have a just bearing on the actor's liability.”³⁷ One might argue that the state should not be responsible for all harms that it purposely, knowingly, recklessly, or negligently causes unless the result of the state's action can be fairly attributed to the state.³⁸

Though the causation of some bad result is frequently required for legal liability, there is much dispute about whether the causation of a bad result is relevant to moral assessment.³⁹ So it is hardly clear that a moral justification of punishment should have any proximate-cause limitations at all. More importantly, however, the justification-symmetry principle requires us to treat proximate causation the same way for ordinary people as we do for those who purport to inflict just punishment. If an ordinary person must justify the bad experiences he knowingly inflicts when he confines someone, then so must a person who does so in the name of just punishment.

5. Justifying Behavior of All Involved

The justification-symmetry principle also tells us whose behavior must be justified. There are many actors in the criminal justice system, and many of them cause inmates harm of various sorts. They must each be able to justify their purposeful, knowing, reckless, and negligent inflictions of harm or risks of harm where you and I would have to do the same. A term of incarceration cannot be justified if the implementation of the sentence requires a state actor to act unjustly.

Judges are some of the most visible state actors in the state's incarcerative machinery. Although jurisdictions vary considerably in terms of how much sentencing discretion they give to judges, every time a judge has any discretion to set sentence length and decides to sentence for more than the absolute minimum, the judge has contributed to the prisoner's harm. The judge is responsible for any period of confinement above the legally required minimum and must have a justification for imposing any additional confinement, just as you or I would need a justification for forcibly confining someone for a day, an hour, or even a minute.⁴⁰

Judges are not the only ones responsible for sentences, of course. Legislatures frequently set maximum and minimum terms for particular offenses and often create or approve of sentencing guidelines that substantially narrow judges’ sentencing discretion even further. Clearly, sentences are part of a complex web of decisions by judges, legislators, governors, prison employees, voters, and others. All of these actors potentially take steps that require justification.

6. The Substrate of Punishment

Theorists disagree about how to characterize the harms of punishment. Some speak of punishment as an intentional infliction of suffering,⁴¹ some speak of it as an intentional limitation of a person's liberties,⁴² and some speak of punishment as capturing both sorts of actions.⁴³

For purposes of the justification-symmetry principle, it makes no difference how people characterize their own behavior. They must still justify at least those inflictions of harm that would need to be justified if inflicted by you or me, no matter whether their purpose is to inflict harms in experiential terms or nonexperiential terms. So even if one uses the term punishment to refer to the causation of certain objectively-understood deprivations of liberty, any instance of punishment will almost certainly cause experiential harms that still require justification.

7. Time Period to Justify

Criminal law theorists have debated whether incarcerative punishment should be understood to end when an offender's sentence ends or whether offenders are also punished by harsh treatment after incarceration, as when ex-convicts foreseeably have great difficulty getting a job. But even if postincarcerative harms are technically not inflicted as punishment, from the point of view of justification, we must still justify at least some of the postincarcerative harms that are knowingly, recklessly, or negligently caused.⁴⁴ If you or I were to knowingly ruin someone's career, devastate his family life, and deprive him of voting rights, even in the distant future, we would need a justification for doing so. Thus the state must have a justification for foreseeably harming prisoners even when the harms are expected to occur after their incarceration ends.

C. Responses to Possible Objections

I argue that a good justification of punishment will address all of the foregoing. Theories of punishment that address only intentional aspects of punishment are woefully incomplete. I now respond to two possible objections that aim to ease theorists’ justificatory burden.

1. Compensating Harms Objection

At least some of the harms that I identify, one might argue, can be remedied after the fact through compensation to prisoners. Provided one advocates some radical revisions to our current criminal justice system, compensation does indeed provide a method of justifying (or at least remedying) certain harms. But it is worth considering how much these revisions deviate from current law. Today, we compensate prisoners for their conditions of confinement only under very rare circumstances. To violate the constitutional protection against cruel and unusual punishment when challenging prison conditions, inmates must prove that prison officials acted with “deliberate indifference” to substantial needs of prisoners such as their safety and their medical care.⁴⁵ According to the U.S. Supreme Court:

[A] prison official cannot be found liable under the Eighth Amendment for denying an inmate humane conditions of confinement unless the official knows of and disregards an excessive risk to inmate health or safety; the official must both be aware of facts from which the inference could be drawn that a substantial risk of serious harm exists, and he must also draw the inference.⁴⁶

So it appears that prison officials can deliberately impose risks that are not substantial or harms that are not serious. They are also free to act in ways that are indisputably negligent,⁴⁷ so long as the acts do not rise to the level of deliberate indifference.

Clearly, there is a vast gap between the relatively rare situations in which the government can be held accountable as a matter of law for its treatment of offenders and the day-to-day situations in which government must justify its unintended harsh treatment of offenders as a matter of moral principle.

Even if we changed the law to require compensation for all reckless and negligent punishment-related conduct attributable to the state, the state would still be blameworthy for the foreseen harms of incarceration. And it is questionable whether harms we might characterize as merely foreseen (such as reducing offenders’ freedom to express themselves, have sex, and associate with friends and family) can be appropriately compensated with money. We do not generally think it justified to knowingly cause someone injury provided you are willing to pay the amount you will later owe him in tort compensation. But even if the compensation strategy could work, we would have to compensate every prisoner for the knowing inflictions that are part of imprisonment, and it is highly unlikely that retributivists would defend such widespread compensation.

2. Doctrine-of-Double-Effect Objection

According to the so-called “doctrine of double effect,” the good that a person intends to achieve can sometimes justify foreseen harms that go along with it.⁴⁸ To take a familiar example, suppose an air force pilot intends to bomb an enemy target knowing that, as an unfortunate side effect, innocent civilians will be killed. Under the doctrine of double effect, the death of the innocent civilians might be justified by the positive aim of destroying a legitimate military target. So, retributivists might argue, many of the unintended harms of incarceration can be justified by the state's legitimate retributive intentions.

The job of justifying the side-effect harms of incarceration, however, is hardly so easy. First, there is much dispute over how to formulate the doctrine of double effect and whether it is sound.⁴⁹ Second, even if we agree that the causation of some negative side effects can be justified by some worthy aims, it is not at all clear that deserved punishment is sufficiently worthy to justify the negative side effects of incarceration. Surely any instance of punishment cannot justify any amount of side-effect harm. Recall the conjoined-twin case. It is doubtful that the purported positive effect of delivering deserved punishment to the offending twin justifies depriving the innocent twin of his liberty. This is especially true if one accepts Blackstone's assertion that it is “better that ten guilty persons escape, than that one innocent suffer.”⁵⁰

Influential versions of the doctrine of double effect always include some requirement that side-effect harms be not too large relative to the benefits a person intends.⁵¹ Thus there are limits on the magnitude of foreseen harm that can be justified by an intended positive aim. Consider again the air force pilot. If the pilot has an alternative target that is like the original target in all pertinent respects except that the alternative target will lead to fewer side-effect casualties, then the pilot is morally obligated to pursue the alternative. Even if there is no better alternative, if the expected number of civilian casualties gets high enough and the importance of the military target gets low enough, causing the side effect is no longer justified. So, even armed with the doctrine of double effect, retributivists must explain why the good aspects of delivering deserved punishment warrant the bad aspects of causing foreseen but unintended harms. And there is no way to do that without taking the magnitude of foreseen and foreseeable harms into account.

IV. SPLITTING PUNISHMENT

Retributivists may concede that they fail to justify the wide-ranging side effects of punishment that require justification. Nevertheless, they may argue, these side effects need not be justified by a theory of punishment. What justifies a punishment practice like incarceration, they could claim, is the joint action of a theory of punishment (which justifies purposeful inflictions) and some other moral or political theory that justifies nonpurposeful inflictions. I call this other theory a “shadow” theory because it follows along with a theory of punishment. In this section, I explain why a shadow strategy makes retributivism extraordinarily anemic and fails to avoid the measurement and calibration obligations.

A. Too Anemic to Justify Terms of Incarceration

Some retributivists argue that I overstate their justificatory obligation. According to Ken Simons, “[t]he state's responsibility is simply to ensure that the punishment that it directly inflicts is proportionate to desert,” and Simons does not consider the foreseen bad experiences of prison to be directly inflicted by the state.⁵² Dan Markel and Chad Flanders emphasize, in particular, that the state need not consider the harmful effects of imprisonment after offenders are released: “As for what people might contingently and speculatively suffer after the state has appropriately punished them, retributive justice has very little to say. . . . Those harms are simply not state punishment and thus are not part of what retributive justice theories have to justify.”⁵³

These theorists could be read to deny the measurement obligation entirely. They may, however, be making the more modest claim that while the side effects of incarceration need to be justified, they need not be justified as punishment. For example, David Gray claims that “no theory of criminal punishment is obliged to justify . . . the unintended suffering that may incidentally result from punishment.”⁵⁴ Nevertheless, he seems to adopt a shadow strategy, noting that “incidental suffering secondary to objectively justified punishment may raise independent moral, constitutional, legal, or institutional questions; incidental suffering may even rise to the level that amelioration or adaptation of penal technology is required.”⁵⁵

The shadow strategy leads to a surprisingly anemic version of retributivism. By claiming that a theory of punishment need justify only certain purposeful inflictions, this strategy implicitly concedes that retributivism cannot, on its own, justify punishment practices like incarceration, because the incarceration of any particular offender will inevitably have foreseen and foreseeable side effects. In an effort to avoid the measurement obligation, retributivism severs its ties to real-world methods of punishment.

By contrast, most theorists seem to have more ambitious views of the power of retributive justification. Although some retributivists acknowledge that a broader political theory is required to justify setting up institutions to deliver retributive justice,⁵⁶ retributivists typically speak as though they are capable, at least in principle, of setting appropriate sentence lengths. Doug Husak, for example, suggests that the retributive tradition is generally thought to tell us how much we can punish individual offenders:

Following the lead of H.L.A Hart, the dominant tendency has been to consign the relevance of consequences to the general justifying aim of the institution of punishment. Consequentialist considerations are thought to play no further role once questions of distribution are raised, to answer the questions of who should be punished and to what extent. These latter issues are regarded as more properly resolved within the retributive tradition, to which consequences are foreign.⁵⁷

Gray himself states that the “core challenge to any theory of criminal law” is “to justify punishment generally and to rationalize the punishments inflicted in particular cases more specifically.”⁵⁸ Though Husak and Gray refer to justifying the “punishment” in particular cases, they seem to imply that retributivism is expected, at least in principle, to provide for and justify the incarcerative sentences of particular individuals.⁵⁹

We sensibly expect a theory of punishment to justify terms of incarceration. It is hard to see why we would be interested in a purported justification of punishment that is fundamentally incapable of justifying the incarceration of particular individuals. Consider again the difficult ethical issues raised when an air force pilot is tasked with destroying a legitimate military target but knows that doing so will have the unfortunate side effect of killing several civilians. Suppose a theorist purports to have solved the problem as follows: the pilot can justifiably attack the target because his purpose in acting is legitimate. “But what about the civilians killed as a foreseen side effect?” we ask. Surely the theorist cannot adequately respond by claiming that some other moral theory must justify the side-effect harms. Such a response reveals that the theorist has offered no solution at all. We clearly cannot send pilots on a combat mission without knowing how to evaluate the side-effect harms we expect them to cause. Similarly, we cannot incarcerate offenders without knowing what justifies the side-effect harms we expect incarceration to cause.

Moreover, as I explain in more detail, the shadow strategy leaves retributivism unable to measure punishment properly. If only purposeful inflictions constitute punishment, then we are led to believe, contrary to our intuitions, that Purp and Fore receive very different amounts of punishment even though they are alike in all respects except for the intentionality of their treatment. The traditional notion of punishment proportionality at the heart of retributivism is rendered obscure when assessments of amount of punishment differ so dramatically from our own.

B. Shadow Theory Based on Desert

Retributivists would be wise, then, to present a particular shadow theory that justifies the side effects of punishment. But the task of finding a satisfactory theory may be more difficult than they realize.

They could try to justify the side-effect harms of punishment based on the value of giving offenders what they deserve. As a preliminary matter, this approach makes the effort to avoid the measurement obligation very superficial. Under this approach, both the intentional and unintentional aspects of punishment are justified by the value of giving offenders what they deserve, but only the intentional hardships are considered part of a theory of punishment. If the unintentional hardships are also justified on the basis of criminal desert, it seems more straightforward to describe the entire apparatus as part of a justification of punishment.

There are two more substantive problems with using a shadow theory based on desert. First, these conjoined theories still have to measure unintentional hardships of prison and take them into account at sentencing for the reasons I give in the discussion of the doctrine of double effect. Namely, we cannot be confident that the allegedly valuable act of purposely giving an offender what he deserves justifies the bad side effects unless we know something about the magnitude of the side-effect harms. There must be some proportionality between the severity of unintended hardships we impose and the offender's desert. Otherwise we could knowingly inflict tremendous side-effect harms with abandon.⁶⁰ So, at sentencing, we must make sure not only that the intentional harms of punishment are appropriate but also that the unintentional harms are appropriate as well. Merely ensuring that conditions of confinement are not cruel and unusual is insufficient; the conditions must be justified by the value of giving offenders what they deserve.

Second, if criminal desert justifies both the intended and unintended aspects of punishment, it is not clear why offenders repay their desert debt only when they receive intentional hardships. The conventional view seems to be that an offender deserves punishment and that as he receives his punishment, he subsequently deserves less of it. Similarly, it seems, if the unintentional harms of punishment are justified by an offender's desert, the offender should deserve less of them as his sentence proceeds. If offenders repay their desert debt with both purposeful and nonpurposeful inflictions, then accurate sentencing would require us to calibrate sentences in a manner that considers both kinds of inflictions. Rather than having to measure side-effect harms just to make sure they are justified by punishers’ primary intentions, they have to go further and actually adjust sentence length (or other aspects of confinement) in order to punish proportionally.

The strongest argument for a calibration obligation returns us to the Purp and Fore example. Purp and Fore did not receive proportional punishments when we considered only the intentional aspects of their incarceration. They received proportional punishments only when punishment severity included both their intended and foreseen harms. Both purposeful and nonpurposeful inflictions can contribute to our assessments of punishment severity. So ignoring nonpurposeful inflictions risks increasing punishment severity beyond what is deserved.

In the air force pilot scenario, we recognize that we must sometimes give up our first-choice intended military target when foreseen harms to civilians are too high. The side effects are sufficiently weighty that we must sometimes settle on less ambitious goals. Similarly, when a sensitive prisoner complains that he suffers much more in prison than the average offender, retributivists have an obligation to reduce the amount of intended hardship he is forced to undergo.

Failing to adjust the sentences of those who suffer harms greater than those expected at sentencing means knowingly harming them in a manner that is insensitive to their desert. It should call out to retributivists for reparation. Moreover, the pilot scenario concerns intended harms to one entity and foreseen harms to another. Punishment, by contrast, requires intentional and foreseen harms to the very same people. Claims that we cannot offset intentional harms by foreseen harms in the punishment context seem particularly suspect.

C. Shadow Theory Based on Consequentialism

Alternatively, retributivists could endorse a consequentialist shadow theory. But then they would be subject to the demanding obligations faced by consequentialists. Consequentialists must measure the side-effect harms of punishment-related action to make sure that they are outweighed by the expected good consequences of the action, such as crime prevention and rehabilitation.⁶¹ They must also calibrate punishments to optimally deter potential offenders who will vary in how aversive they anticipate prison life to be, and both purposeful and nonpurposeful aspects of incarceration are aversive.⁶² Consequentialists are no less obligated to consider the unintentional aspects of punishment than retributivists, though consequentialists are less likely to recognize a stark difference between the purposeful and nonpurposeful aspects of punishment in the first place.

A consequentialist shadow theory may appeal to those already sympathetic to hybrid or mixed theories of punishment, but it sounds the death knell for traditional notions of retributive proportionality. Based on an offender's blameworthiness, retributivism may dictate a punishment (meaning the intentional infliction of a hardship) for a particular period of time. But there is no reason to think that a consequentialist shadow theory will justify foreseen hardships for the same duration.

To illustrate, assume that retributivists can justify one year in which an offender is purposely restricted in his freedom of motion to the confines of the prison. A shadow theory must still explain why we are justified in imposing foreseen restrictions, like his reduced freedoms to express himself, see family, and have sex, for one year. Why would the foreseen restrictions justified by consequentialism have the same duration as the purposeful restrictions justified by retributivism?

While retributivism may justify one year of purposeful inflictions, it would be a cosmic coincidence if consequentialist principles happen to settle on the same duration. Some offenders may deserve to be punished for a long time, but for various reasons, such as quadriplegia, pose little risk of future danger such that there are only weak consequentialist grounds for confining them. So even if retributivists can offer a good justification for the inflictions they purposefully impose, they cannot punish in proportion to desert if they must justify the duration of foreseen inflictions using a consequentialist shadow theory.

True, some foreseen hardships associated with incarceration do not have the same duration as the intended liberty restrictions associated with prison sentences. For example, supervised release restrictions and various restrictions on rights to vote and possess firearms frequently continue after an inmate is released from incarceration. Nevertheless, retributivists must explain why so many foreseen hardships of prison have a duration that just happens to coincide with the duration of retributively determined purposeful hardships.

Some retributivists believe that sentences should be no longer than what is permitted by traditional retributivist principles of proportionality. Such retributivists can adopt a policy that advocates sentencing offenders to the shortest term that is justified by both retributivist and consequentialist principles. But as shown here, doing so does not square with pure retributive notions of proportionality and does not avoid obligations to measure and calibrate punishment. Thus, I am arguing, whether you adopt a shadow theory rooted in desert or in consequentialism, you cannot avoid the obligation to sentence in a manner that takes account of the many unintended hardships associated with prison, including the varied ways in which prisoners experience confinement.

V. CONCLUSION

Under the justification-symmetry principle, we must justify the purposeful, knowing, reckless, and negligent inflictions of harm by those who purport to inflict just punishment when we would expect a justification if a layperson inflicted the same kind of harm.

As noted in the introduction, the harms of incarceration vary considerably by facility. A five-year term in a supermax prison is more harmful than a five-year term in San Antonio prison. Such differences in harm are easily foreseen and should be taken into account at sentencing.⁶³ Even under identical conditions, prisoners will have very different subjective experiences. When we know about prisoners’ sensitivities to punishment, we should try to take them into account at sentencing.

Retributivists cannot avoid considering prisoner sensitivities by focusing only on the intentional aspects of punishment. Merely justifying the purposeful aspects of incarceration will never justify the practice of incarceration as a whole. Vague references to a shadow theory meant to justify unintended punishment-related inflictions cannot be accepted uncritically. While punishment theorists have long noted that punishment must be intentional, punishment severity consists of both intentional and unintentional hardships, and it is ultimately the severity of punishments that we strive to justify.