This commentary on Zhong et al. begins by addressing the definition of remorse. It then primarily focuses on the relation between remorse and various justifications for punishment commonly accepted in Anglo-American jurisprudence and suggests that remorse cannot be used in a principled way in sentencing. It examines whether forensic psychiatrists have special expertise in evaluating remorse and concludes that they do not. The final section is a pessimistic meditation on sentencing disparities, which is a striking finding of Zhong et al.
Zhong et al.1 have presented the results of a qualitative study of how judges use the concept of remorse in sentencing. Although the study has numerous methodological limitations that the authors admirably recognize, it is an interesting paper and a useful catalyst for further reflections on remorse. This commentary focuses on the relation between remorse and various justifications for punishment that are commonly accepted in Anglo-American jurisprudence. Zhong et al. addressed this relation, but briefly, and I wish to expand their discussion. I will also consider the role forensic psychiatry should play in helping to evaluate whether a defendant is in fact remorseful in cases in which remorse is considered. I begin, however, by addressing the definition of remorse that guides the discussion. I conclude by briefly considering what I consider to be the study's most striking finding: the extensive variability in how judges use remorse at sentencing.
The Definition of Remorse
Zhong et al. use a definition of remorse adopted from Proeve and Tudor2:
Remorse may be defined as a distressing emotion that arises from acceptance of personal responsibility for an act of harm against another person. Often, with further reflection, the remorseful individual may desire that the act had never occurred at all and wish to make restitution toward the victim [Ref. 1, p 41]. It is useful to compare this definition with the Oxford English Dictionary (OED) entry3:
Deep regret or guilt for doing something morally wrong; the fact or state of feeling sorrow for committing a sin; repentance, compunction.
Although similar, they are somewhat different. Consider the similarities. In both, there is a cognitive and an affective component. The agent makes the judgment that he has done something wrong and, explicitly for the former and implicitly for the latter, accepts responsibility for the harm caused. The definition Zhong et al. used is ambiguous about both wrong and the acceptance of responsibility. An agent who causes harm accidentally as a result of intentional bodily movement is causally responsible for harming the victim and may take responsibility because it was his action that caused the harm. But has the agent wronged the victim? It seems that the agent might properly feel what Bernard Williams calls “agent-regret” (Ref. 4, p 27) because the agent has caused harm, but genuine remorse would be inappropriate because no moral wrong has been done and there is no moral responsibility for the harm. Agent-regret would be distressing, but guilty feelings would not be justified. In contrast, the OED makes clear that the judgment must be of a moral wrong. This aspect of the definition would be especially relevant at sentencing because the convicted defendant has been found to be morally culpable. To see this most clearly, consider a crime of strict liability. The convicted agent should not feel guilty, but might properly feel agent-regret if innocent victims were harmed. Remorse is a distinctively moral reaction.
I leave aside cases involving regulatory or malum prohibitum crimes. Even in these, however, there is a point to be made that conviction implies a moral violation to some degree.5 Crimes of strict liability are a clear exception, but they are also notoriously controversial because they condemn and punish without fault.
The definition of remorse in Zhong et al. is also nonspecific about what the distressing emotions should be. Agent-regret might be profoundly distressing, but not because it is a proper result of guilty feelings and repentance would not be appropriate. Moral regret and guilt are the proper response to moral wrongdoing. Moreover, remorseful feelings of regret and guilt need not be distressing for the remorse to be genuine. In some cases, these feelings may be cleansing and help the agent feel that the moral balance is somewhat restored by an appropriate reaction.
Both definitions include a desire that the harm had not occurred and that the pre-harm state can be restored. I think this desire is implicit in the notion of regret. Remorse may also imply a feeling of nostalgia for an earlier, innocent time, a suggestion made to me in a personal communication from Benjamin J. Sadock, MD.
Finally, the definition in Zhong et al. includes a novel criterion: the desire to make restitution. Making restitution would surely be a logical corollary of feeling remorse, but it does not seem to be part of remorse itself.
For all these reasons, I fear that the definition given to the judges was not as precise and relevant to the criminal context as it could have been. I might attribute the great variability in the judges' responses to these difficulties, but judicial discretion leads to great variability, even when criteria are clearer, because judges inject the same criteria with different moral and social meaning. I shall return to this theme in a later section, but will use my OED-based interpretation of remorse as the foundation for further reflections on the role of remorse in criminal justice.
Remorse and the Justifications for Blame and Punishment
Blaming and punishing an agent for committing a crime is the most afflictive, awesome exercise of the power of the state over an individual. Punishment involves not only the stigma of blame, but it also most fundamentally includes the intentional infliction of some form of harm or pain on the convicted person. If anything requires moral justification, it is the intentional stigmatizing and inflicting of pain on an agent. In response to such moral concerns, theorists have offered many potential candidates for justifying punishment. These are loosely placed in two categories: nonconsequential or deontological and consequential. The former includes retribution or just deserts. The latter includes general prevention, specific prevention, and some forms of rehabilitation. These various justifications can lead to differing conclusions. For example, a mentally disordered but culpable defendant may deserve less retributive blame and punishment because his rational capacities are compromised, but the same abnormality might make him especially dangerous and thus a good candidate for extended incarceration to prevent him from committing further crimes.
Many jurisdictions list the justifications in the penal code that the legislature has decided should be adopted, and judges explicitly use the justifications in sentencing decisions, but they typically do not explain why they adopt particular justifications and virtually never set forth an algorithm for how they should be balanced. This deficit is not important for our present purposes, but I discuss it in the penultimate section of the paper.
Let us consider the justifications and their relations to remorse.
Retribution is a theory of justice: giving criminal offenders their just deserts for what they have culpably done is justice. A retributive or just-deserts rationale imposes a deserved punishment because it is right in itself to give a culpable offender what he deserves. Although there may be consequential constraints on desert in extreme cases, retributive punishment ignores whether good consequences will follow from giving the offender his just deserts. There are many different versions of retributivism. For example, giving people their just deserts can be either obligatory or permissive. For another example, some retributivists think that the amount of harm a defendant causes is relevant to deciding what blame and punishment the offender deserves; others believe that only the offender's wrong conduct and not the harm it causes is relevant. As a deontological theory, retribution is similar to deontological theories in other areas of law, such as the obligation to keep one's promise in contract law, even if breaching might be efficient. Retribution includes a proportionality principle. No offender should be punished more than he deserves. Thus, retributivists are opposed in principle to criminal liability for strict-liability crimes because defendants convicted of such crimes are not culpable and do not deserve to be punished. Retribution is often confused with, but is most decidedly not, revenge. Revenge is the crude psychological satisfaction that comes from hurting those who have hurt the victim. It is not a justification for legal punishment, which involves doing justice impartially and fairly. Finally, retributively justified punishment need not be harsh. Retributivists can be tender or tough. Few theorists are pure retributivists, and, to the best of my knowledge, no jurisdiction has implicitly or explicitly adopted retributivism as its sole justification for punishment, but virtually all believe that it is a necessary precondition for punishment, even if it is not sufficient.
Is retribution intended to give offenders what they deserve for what they have done or for who they are, for their characters? It is often said that acts are expressive of character, but we know that seemingly good people sometimes do terrible things. Does this mean that the heinous act is less terrible and more justifiable or excusable or does it mean that our previous evaluation of the perpetrator's character was incorrect? I will ignore the debate about whether character or the situation most accounts for the variance in an individual's behavior and will assume that character plays a substantial role and that situational causation is no more of an excuse, per se, than any other kind of causation.
What does it mean to say that an act is out of character if an unjustified and unexcused agent has intentionally committed a crime? How do we decide what the agent's enduring behavioral predispositions are except by observing their actions over time and across situations? Further, as a set of behavioral predispositions, character is a status, and in our constitutional order, it is a violation of the Eighth Amendment to punish for one's status alone, such as being a drug addict.6
A retributive justification based on the offender's past acts is retrospective because it focuses on what the offender has already done in the past. The question is what he deserves for the crime he has committed. Remorse is always expressed after the wrongful deed has been done. Therefore, if one takes an act-oriented view of retribution, remorse is irrelevant, simply because nothing the offender does after the crime is committed affects the crime itself and what the offender deserves for committing it. Before acting, the offender should have brought to mind and been guided by the moral considerations that later caused remorse. In that case, the agent would not have committed the crime, would not deserve punishment, and would have no need for regret.
If one adopts a character-based theory of retribution, it is difficult to understand why the offender's sentence should be altered. The expression of genuine remorse may speak well of the offender's postcrime character, it may help restore the defendant's breach with the victim and society, and recognition that it has occurred may acknowledge the offender's rationality and autonomy, but it is nonetheless hard to understand why the deserved sentence should be affected on grounds of desert. We are punishing the offender for what he has done. Moreover, as discussed in greater detail later in the article, it is difficult to know whether the offender is genuinely remorseful and what his character really is. Perhaps it has morally eroded and it is not clear that remorse, even if apparently genuine, indicates that his character has reverted to the apparent status quo ante. Sometimes we must be judged by our one-offs. As Nick Hornby wrote in his novel, How To Be Good,7 it would not have done for Lee Harvey Oswald to claim that he was not ordinarily the sort of person who killed presidents. Finally, remorse may seem to have more purchase for reducing sentences according to a communicative justification of punishment tied to retributive desert, but its primary proponent denies that this is true, a recent proponent acknowledges that remorse may be best justified consequentially,8 and no jurisdiction has adopted this view.
Deterrence and Incapacitation
If we want to justify the use of remorse in sentencing, we must do so on consequential grounds. General deterrence, specific deterrence, and incapacitation are the primary consequential justifications in Anglo-American penal law. As is well known, general deterrence is a method of crime prevention that seeks to use the potential pain of punishment to give all people reason not to offend. Special deterrence and incapacitation are methods of specific prevention that are used to prevent offenders from committing further offenses by giving them good reason not to or by keeping them incarcerated. These justifications succeed only if a punishment practice in fact achieves the goal of prevention. If it does not, the punishment is unjustified on consequential grounds. Consequentialism also has a proportionality principle. No more pain should be inflicted on the offender than is necessary to maximize social welfare. Any punishment beyond the minimum necessary for that purpose is disproportionate. Consequential justifications identify the good consequence that is desired. Most broadly, the good consequence is considered to be social welfare. This theory then holds that an action is justified if the net result is achieving or maximizing the good. There are theorists who propose a purely consequential justification of punishment, but with limited exceptions, no jurisdiction has adopted such a view. An exception might be enhanced sentences for recidivists, although many retributivists attracted to this practice have tried unsuccessfully to provide a retributivist justification. Punishment for strict-liability crimes is also justified purely consequentially.
The question concerning the effect of remorse on consequential justifications is empirical. Does remorse or its absence validly predict whether the offender is less or more dangerous and thus should receive a lesser or greater sentence (holding every other variable constant)? When remorse is used for mitigation, the consequential goals may lead to inconsistent results. Remorse that leads to lesser sentences because remorseful defendants are less likely to recidivate may also have the effect of undermining general deterrence, because the criminal law may be seen as soft or easy to manipulate by faked emotions. If lack of remorse is associated with enhanced danger of recidivism, then general and special prevention would both be achieved by lengthening the sentences of such offenders.
Assuming that we can accurately judge when remorse is genuine and what its depth and quality are, a question to which I return in the next section, as Zhong et al. rightly note, the consequential outcomes of using remorse are almost entirely speculative. Guilt does seem to be protective against further criminal conduct,9 but the data on remorse specifically are sparse. One epidemiological study found a modest association between lack of remorse and antisocial conduct,10 but it was not a prospective study. The major exception to this generally cautious conclusion may be the callous/unemotional (CU) factor in psychopathy, which includes deficient remorse. It has been long known that psychopathy is a significant risk factor for future criminal conduct, but the specific role of CU was unknown. A recent study controlled for the other relevant variables and found that CU was associated with incrementally enhanced recidivism risk,11 but CU includes more than deficient remorse, and it is unclear how much of the variance remorse accounted for. These data are suggestive, but we must still conclude that the association between remorse alone and recidivism is unknown. Moreover, we do not know whether the association, if valid, is smoothly scalar or is distributed differently.
We have no valid data to indicate how using remorse as a mitigating factor that leads to reduce sentences would affect general deterrence. Moreover, if there were inconsistent results between the gain to the offender and the loss to society, can we really calculate the social welfare outcome? Of course, such skeptical concerns about the ability actually to calculate net social welfare are a staple of anti-consequential thought. At present, then, no sentencing judge could use remorse on empirically valid and thus principled consequential grounds.
Nonpaternalistic and Paternalistic Rehabilitation
Rehabilitation is a currently disfavored consequential justification and has been so for about four decades. In the late 1960s and early 1970s there was an unusual bipartisan agreement that in large part blamed rehabilitation theory and its empirical failures for arbitrary sentencing discretion. The bipartisan consensus was largely responsible for the resurgence of the just-deserts model of sentencing and the limitation of judicial and parole discretion. Nonetheless, rehabilitation always retains strong proponents, the efficacy data turn out to be less pessimistic than it first appears, and the justification pendulum can and always does swing in criminal justice.
Rehabilitation can be either nonpaternalistic (or nonparentalistic if one prefers the ungendered locution) or paternalistic. In the former, we impose rehabilitative interventions on the offender to render him less dangerous, whether he likes it or not. We straightforwardly acknowledge that we are violating the offender's autonomy and are manipulating him for our good, not his. A Clockwork Orange,12 in which a psychopathic, violent predator is trained by classical conditioning to become ill when presented with violent stimuli, is a classic example. This is clearly a consequential justification based on the cost savings achieved if an effective treatment is less expensive than continued incarceration. It is a form of specific prevention. The major problems are that effective rehabilitation methods are expensive, they do not work for everyone and we cannot identify in advance the promising candidates, and most methods are not very effective. There are promising exceptions, however, and as data accumulate, this form of consequential justification may again play a larger role in the justification of punishment.
In the case of paternalistic rehabilitation, the goal is to improve the offender so that he can lead a better life and flourish. For several reasons, this justification has no proper role to play as a prima facie justification for punishment. Most fundamentally, paternalistic rehabilitation is not a form of the intentional infliction of pain, and thus it is not a state punishment, even if the offender experiences it as such. Second, in a liberal society, paternalistic interventions are disfavored generally, and most prisoners are not among the classes of individuals, such as minors or people with dementia, for whom paternalism is justified because they are not responsible agents. Third, in a world of limited, scarce resources, it is unfair to provide benefits to those who least deserve them and especially because the offender may not want the benefits. Finally, characterizing punishment as a treatment for the offender's own good promotes a form of moral blindness that allows us to manipulate people in unacceptable ways. This outcome is always a danger of medicalizing a moral or social problem, such as how to respond to so-called mentally abnormal, sexually violent predators.
The problem of the relation of remorse to rehabilitation is the same as besets the relation of remorse to other consequential justifications. There is simply no evidence that a genuinely remorseful offender is a better candidate for nonpaternalistic rehabilitation than an offender who is not remorseful. Once again, remorse can play no principled role in a sentencing decision guided by a rehabilitative justification.
A mixed justification of punishment attempts to blend retributivism and consequentialism. Although many claim that mixed justifications are incoherent,13 it is probably the dominant form of justification used today. I confess to being a muzzy-headed, mixed theorist, myself. The most usual form acknowledges that retribution has no theory to provide a single fixed amount of punishment for each crime type. Rather, it sets a range of proportionate punishment for crimes. An offender should receive at least the minimum but not more than the maximum of the deserved range. Within the range, consequential justifications can be used to adjust the sentence up or down within the range For example, more dangerous offenders may receive sentences toward the higher end of the range and vice versa.
Remorse fares no better as a principled ground for sentencing under the mixed justification than it did under either of the two pure categories, for the obvious reason that if it cannot be properly used by either alone, no moral alchemy makes it viable if one tries to blend justifications.
Remorse and Forensic Practice
Mock studies with college student subjects, which may be of limited ecological validity,14 show that the study subjects take remorse into account in making judgments about criminal responsibility and punishment,15,16 and many sentencing judges take remorse into account. These facts are positive, not normative, however. I have suggested that, at present, the law should not consider remorse when making decisions about an offender's fate because there is apparently no good justification for doing so. Nonetheless, in the discussion that follows, I shall assume that, as a practical matter, remorse will be considered. It seems that there would then be two roles for forensic psychiatrists. The first would be to help identify whether an offender is in fact remorseful. The second would be to do research that would create a reliable and valid instrument for identifying remorse and then to determine whether remorse is a valid indicator of one or more of the consequential justifications of punishment. Let us consider these two roles.
The question the first role raises is whether forensic psychiatrists and psychologists have special expertise, qua forensic psychiatrists and psychologists, in identifying genuine remorse and its depth and character. I know of no study that addresses this question specifically, and a search of PubMed disclosed none. For various reasons, I believe we should be cautious before claiming that we have special expertise or before offering our services for the purpose of evaluating remorse.
Remorse is mostly not a mental health concept. It is what is often termed a moral emotion. In the entire Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5), it receives only one mention, as the diagnostically unnecessary (because the diagnosis can be made without this criterion being satisfied) seventh criterion for antisocial personality disorder (APD) (Ref. 17, pp 659–63). Lack of remorse is indicated by “being indifferent to or rationalizing having hurt, mistreated, or stolen….” With all respect, this criterion is vague. For example, how indifferent must the subject be and what will count as a sufficient rationalization? The relatively good reliability of APD does not obviate this problem, and it does not address remorse specifically. There is no entry for remorse in the American Psychiatric Publishing volume, The Language of Mental Health: A Glossary of Psychiatric Terms.18 Although I have no study to prove this point, I am quite certain that the evaluation of remorse plays a trivial role in most psychiatry and psychology clinical training, including specialty training in forensic practice; but my supposition may be inaccurate. Finally, because lack of remorse is not really a psychiatric or psychological symptom, studies concerning the identification of malingering of psychopathology are inapposite.
Psychopathy, which is a relatively well-characterized condition and best measured by the Hare Psychopathy Checklist-Revised (PCL-R),19 is not a recognized DSM disorder, but it may seem to provide an exception to the cautious concern about lack of special expertise. One of its 20 items is “lack of remorse or guilt,” which is itself part of what is called the affective facet, the same facet that includes callousness and lack of empathy. The PCL-R has excellent overall interrater reliability among trained users, but there is reason to question whether the reliability is as good in applied forensic settings as in research studies, and few field trials of reliability have provided scores for the subscales or individual items.20 In the most recent report of an applied field trial with relatively few subjects, the overall reliability was moderately good, but the reliability of the lack of remorse item among highly trained clinicians was only 0.51, with a large confidence interval, and the reliability of the affective facet was only 0.59 overall.21 Indeed, the only facet of the instrument that has genuinely good reliability, the antisocial facet, measures essentially antisocial conduct, a more objectively identifiable set of variables that do not measure remorse.
I thus conclude, on the basis of the currently available data, that forensic professionals possess no special expertise in evaluating remorse and that we should proceed with caution in this domain.
The second role for forensic mental health professionals would be to develop an instrument that would reliably and validly measure remorse and then to determine whether it is a valid indicator of future dangerousness and amenability to rehabilitation. That task is for the future, however.
Zhong et al.1 strikingly confirmed that judges use widely different sentencing criteria in very different ways and impose markedly different sentences in similar cases, even if the sentencing criteria have been identified in advance. Although the virtues and vices of judicial sentencing discretion are not a central concern in forensic psychiatry, it is worth a brief discussion because it affects how our work is treated in criminal cases and because forensic psychiatrists and psychologists are often asked for their input concerning the criminal process.
Unequal treatment was a large part of the motivation of the bipartisan movement in the 1960s and 1970s to limit judicial discretion in various ways. Disparate judicial treatment of like cases was one of the foundations for the imposition of the Federal Sentencing Guidelines in federal criminal cases. The mandatory nature of the guidelines was declared unconstitutional for complicated reasons in 2005 in United States v. Booker,22 but the guidelines were retained as advisory, and federal judges are expected to operate within them, although they are not bound by them. The result has been guideline creep, to wit, the reintroduction of substantial sentencing disparities in like cases. In Zhong et al., the judges ranged remarkably widely in how they applied the same definition of remorse. In light of the history of judicial discretion about sentencing, this is entirely unsurprising and entirely unsettling. Experienced criminal lawyers know that most criminal defendants care more about whether and how long they will be imprisoned than about whether they will be convicted. Unjustified disparities in sentencing are, quite simply, unjust.
Judges are human beings who have their own perceptions and value systems. Sentencing criteria, when they exist, are often vague, and there is seldom an algorithm for how to weigh and balance the various factors. It is thus no surprise that sentencing judges are shaped by their implicit or explicit underlying justifications for punishment that are in turn shaped by the judges' unique histories. They will wittingly or unwittingly create their own criteria or infuse existing criteria with their subjective meanings and values. Only the most constraining legislative limits can prevent such exercises of discretion. In the worst cases, judges may be swayed unconsciously by unacknowledged prejudice. I am not suggesting that judges act in bad faith when they sentence. Virtually none does, I believe. Virtually all state and federal trial judges who I have talked to over the years recognize the extraordinary power that they possess over the liberty and well-being of convicted offenders, and they try to sentence in a principled and fair manner. For the reasons given, however, even if every judge behaved that way, the disparities would still be broad.
The argument to the contrary is that each crime is unique and so many factors may be considered at sentencing that no set of criteria that impose severe constraints could conceivably do proper justice. Unless one had a theory about how all of the potential sentencing factors should affect sentences and how they should all be weighed and balanced, this account of justice is, with all respect, a pipe dream. We would all like to believe that there are people who by dint of rigorous training and professional experience can be trusted to make wise, fair decisions. Unfortunately, as noted, legislatures seldom make clear what theory of justice is guiding the adoption, weighing, and balancing of possible sentencing factors. Thus, in most cases, each judge applies an individualized theory of justice and account of punishment.
Even if my diagnosis of the ills of sentencing is accurate, and many would disagree, the solution remains unclear. It is possible to wring virtually all discretion out of sentencing, but that will not prevent law enforcement officials and prosecutors from exercising their discretion. Charging decisions are a notorious example of a situation in which discretion and disparities abound. Perhaps, however, we should put our jurisprudential pants on one leg at a time and reduce unjustified discretion when and where we can.
Remorse is a familiar concept and a moral emotion that judges and laypeople alike use in judging others. Zhong et al. have provided an impetus to reflect on the use of remorse by judges at sentencing and to conclude that, at present, this practice is almost impossible to defend normatively. I have also suggested that the evidence to date does not support the conclusion that forensic mental health professionals have special expertise concerning the evaluation of remorse. Finally, I bemoaned the sentencing disparities resulting from judicial discretion that Zhong et al. have so manifestly confirmed, but observed that, as a practical matter, judicial discretion will continue.
Disclosures of financial or other potential conflicts of interest: None.
- © 2014 American Academy of Psychiatry and the Law