I. Three Views of Retribution

Since this paper sets out to explore the view that there are aims of justice, it must set aside the view that retributive justice is aimless, or that retribution is intrinsically just, requiring no further goals. Kant held that punishment “must always be inflicted upon [a person] only because he has committed a crime. For a human being can never be treated merely as a means to the purposes of another, or be put among the objects of rights to things” (1996: 105). He drew from this view a startling and arguably counterintuitive conclusion: if a society knew that it was about to be dissolved, it would be duty-bound to execute condemned prisoners before the catastrophe struck, so that “each has done to him what his deeds deserve” (1996: 106). Kant's argument has more recently been defended by James Rachels, who argues on the basis that there are widely held notions of desert (1997). Suppose someone has done us a favour in the recent past, such as giving us a ride to work. Suppose a time comes when our benefactor in turn needs a ride to work, and that we have to choose between helping her out and helping out someone who has never been helpful to us in the past, though they had opportunities; we would certainly—unless very special circumstances applied—decide to help our benefactor. Good should be returned for good, no further reason being called for. If this is the case, why then should bad not be returned for bad, for no further reason?

However, the analogy does not hold, because the burden of justification falls differently. In both cases, what we actually need to justify is doing negative or harmful things to people, not doing beneficial things; when we give a benefit to X, our benefactor, what we must justify is the denial of that benefit to Y, the unhelpful person. In Rachels' first case, denying Y of that benefit is simply a natural consequence of conferring the benefit on X—we cannot give rides to both of them at the same time. But when we speak of returning harm for harm, nothing like that justification applies. Punishment of the malefactor is not a natural consequence of conferring a benefit on someone else, and it therefore requires direct justification in a way in which conferring a benefit does not. That bad is being returned for bad certainly tells us that it would be wrong to “punish” someone who had done no wrong, but does not explain why it is right to punish a wrongdoer (Vernon, 2005: 250–52).

Both Kant and Rachels worry that if we do not accept the idea that intrinsic reasons justify retribution and punishment, we will in turn be committed to a consequentialist view that treats punishment as means to a social end, such as deterrence. For Kant, this is unacceptable, because to treat another person only as a means is to deny their humanity. Rachels adds the further worry that if punishment is a means of deterring future crime, then what matters is that punishment falls on people generally believed to be wrongdoers, as opposed to the wrongdoers themselves. However, their distinction between intrinsic and consequentialist justifications is not exhaustive; there is an expressive justification that does not raise the worries outlined above.

Jonathan Allen, in a discussion of the TRC, also argues that “we cannot understand [punishment] solely in terms of retribution” (1999: 327). Indeed, the idea that punishment should serve as moral education is almost intuitive, for we might well ask what purpose punishment serves absent the public space in which it is handed down. The belief that justice must be seen to be done is crucial in this respect, for it demonstrates that while the process of trial and punishment is primarily seen to focus on a particular victim and offender, the practice itself necessarily functions at the societal level, in how we perceive our relations with one another. It would be counterintuitive to suggest that the crime of murder is a matter of concern only to the victim's family and those who committed, or conspired to commit, the act. That it is classified as a crime speaks to its public nature, to its role as a public threat. As such, it must be dealt with in such a way as to communicate to the larger society both that the threat has been addressed, and that similar threats are wrong and will not be tolerated.

The above discussion assumes a common understanding of a set of rules governing social interaction. It is when such rules are broken that dispute arises, and adjudication between parties to the dispute becomes necessary. This commonly accepted set of rules constitutes the moral order of a society. This moral order is illuminated by Allen's discussion of justice as recognition; in a court of law, justice as recognition operates insofar as individuals are subjects of law: “legal institutions … do not recognize everything about the identity and achievement of individuals—simply their equal status as responsible persons capable of making reflective moral decisions” (1999: 329). If status under the law is indeed equal, and understood by the general public to be so, then an individual's calculations concerning his actions would ideally take into account the capacity of others to make equally reflective moral decisions. There is an assumption here that an individual's moral decisions will not be divorced from his understanding of his own moral worth and what that worth entails; consequently, he will treat others as he himself would like to be treated, as the adage goes. A society where every individual operates more or less consistently in this manner gives rise to a set of rules that in turn comes to constitute a moral order to which every member of that society is bound. Insofar as legal institutions recognize the equal status of individuals “as responsible persons capable of making reflective moral decisions,” then these institutions likewise function to uphold the moral order.

From this line of thought, and also following Jean Hampton (1988), we can view a criminal act as an implied claim to privilege. When a perpetrator harms a victim, he practically implies that his interests count for so much that (at least one of) the victim's interests are simply outweighed. This is inconsistent with their equality of status, as humans or as co-citizens. Punishment, then, expresses the wrong that has been done by striking at the claim to privilege: the damage that is done to the perpetrator is justified as a way of affirming that the victim's interests are of moral account by subtracting from the perpetrator's interests. It is not a case of “returning bad for bad.” Although, in an approximate fashion, more or less serious damage to victims will attract more or less heavy penalties, the damage suffered by the perpetrator is not a kind of two-party exchange for the victim's damage (as in the model of private law; see Nickel, 1976: 379–88). Rather, the relation between victim and perpetrator is mediated by the common need to uphold the primary value of their legal equality, and the schemes of relationship that the recognition of equality of status makes possible. Indeed, it is the societal mediation between the perpetrator's act and the victim's loss that is often taken to be a weakness of retributive systems, which regard criminal offences as affronts to order at the cost of classifying the victim's loss as nothing more than evidence, and marginalizing the victim's subjective experience of it. Legally speaking, criminal acts are offences to the Crown or People or State, though the victim has borne the cost. The point of the judicial process is not to repair the cost to them, but to affirm the importance of the moral order that is violated when the equal status that it mandates is violated.

If we return to the considerations of Kant and Rachels, in favour of an intrinsic model of retribution, we are now in a position to evaluate how the expressive model fares in comparison. If a society knew that it was about to be dissolved, this model would not oblige us to punish prisoners while there was still time to give them “what their deeds deserved,” for an expressive view naturally implies the survival of a continuing public sphere within which the value in question can be expressed to someone. Neither does this model involve using perpetrators (or believed perpetrators) as mere means, for the process is justified as expressing, or truly representing the facts about what was done and suffered, and as declaring its wrongness. Stepping down from a purely conceptual level, one can see this declaratory aspect in the actual practice of trial and punishment. A plea of guilty forgoes the process of determining whether or not evidence points to criminal wrongdoing; indeed, it makes such a process redundant. Conversely, a verdict of guilty counts, of course, as a public acknowledgment of the offender's wrong. Even after a plea of innocence and a verdict of guilty, the offender's own acknowledgment is regularly sought, and “[a] show of remorse by a defendant … routinely has an impact on sentencing and parole” (Gill, 2002: 115). If punishment was of purely intrinsic value, as opposed to expressive value, it would be difficult to understand why acknowledgment and show of remorse would be taken into account: the deed having been done, the offender's just deserts would follow. It would appear, however, that “[t]he power and resources of the system … are reserved for the unapologetic” (Gill, 2002: 115).

With this proposed expressive view of retribution in mind, let us turn to the question of justice in the aftermath of massive atrocity or oppression, and the distinctive issues that are raised in that context.

II. Transitional Societies and Rule of Law: From Disrespect to Denial

In transitional societies, there are many grimly practical reasons for abandoning retributive processes. Acceptably reliable criminal trials are methodical, time consuming and costly, and simply unworkable when (as in Rwanda at one point) over a hundred thousand suspected génocidaires are in jail awaiting trial. There may be no effective and uncompromised judicial system in existence. In the case of oppressive regimes, significant levels of culpability may fall short of the ordinary criteria for criminal offences. And the threat or fear of retribution may deter leaders from giving up power or lead them to negotiate amnesty in advance. If these are the motivating reasons for adopting alternatives to retribution, then the “second-best” justification applies. We can give victims the chance to be heard, we can document what happened for posterity, we can subject perpetrators to the indignity of public confession—regretting, all the while, that we cannot do more.

But there are also conceptual reasons that weigh against punishment, in addition to practical ones. We can consider these reasons in terms of scope and scale. The view of retribution sketched in the previous section implies a set of background beliefs and practices that constitute a society's public moral order. It implies a society that is for the most part law abiding, and made up of schemes of relations that depend on trust and basic mutual respect. Consequently, criminal offences are rightly viewed as disruptions, and criminal punishments can thus be viewed as reaffirmations of the importance of the background order.

In a society where the rule of law is upheld, and where law itself recognizes the equal moral agency of individuals, the rules constituting the moral order can be considered universal in scope. That is to say, they apply equally to all members of that society; moreover, all members have access to the law in order to secure redress for violations that disrespect their moral worth. When moral order and the rule of law coincide in this manner, a society is able to make accommodation for a relatively low incidence of “rule breaking” that will occur in any case. Trial on an individual basis is feasible in such a situation because the society is capable of isolating the rule-breaker from the moral order that he or she sought to undermine through his or her actions. On the isolation of the suspect in a trial, more will be said in the following section. As for the isolating circumstances of punishment, surely imprisonment would have been rejected long ago if the case for it rested on its supposed deterrent or rehabilitative virtues, so weakly supported by evidence. That it survives as a taken-for-granted remedy suggests that it responds to an intuitive idea that offenders must be excluded, even if temporarily, from an order of relations for which their actions are unfitted. Their exclusion is an affirmation of that order's value, as is the understanding that their transgressions must be addressed in some manner before they are permitted to return fully to that order.

The criminal process, then, seeks to isolate, contain and cancel a disrespectful act that violates widely shared norms embedded in a society's practices. But what is wrong with atrocities and oppressive regimes can hardly be captured by this model. Atrocity and oppression play out in an atmosphere wherein respect for the rule of law is lacking, absent, or where the laws themselves are distorted. In cases where respect for rule of law is lacking or absent, the scope of the moral order can no longer be considered universal: although the law may recognize the equal moral agency of individuals, widespread lack of respect operates to nullify such recognition. The wrong to be corrected is not that of disrespect but, still more far-reachingly, that of denial—denial that victims even have a moral claim to be regarded as victims at all. Disrespect is something that occurs against a background of acknowledged practices of respect for other people. But widespread atrocity and systematic oppression constitute practical denials that whole categories of people have any moral weight whatsoever. This is accomplished by denials of the humanity, or membership, or innocence of their victims: they are vermin, cockroaches; they are outsiders, not part of our moral community; they are traitors, they stabbed us in the back, or plan to do so. Or else, this is accomplished by denials of the facts of oppression: there are no death camps; there is no (systematic) torture. It is hardly a mere accident that such denials accompany major evils, which require mass participation (Rwanda) or broadly diffused support (apartheid) and thus depend on the successful mobilization of opinion against their targets. We do not need a particularly sanguine view of human nature to see that, for the most part, people are incapable of sustained participation in causes they themselves regard as straightforwardly criminal in nature. Harm, to be sustained by large numbers of people, must be normalized (see Glover, 2001).

Where the laws themselves are distorted and do not afford all members of society equal moral status, then the model of isolation becomes inappropriate and even radically misleading. The scope of the moral order becomes restricted and particular, and this narrowness of scope allows for wide-scale rule breaking. In contrast to the rule breaking in societies where rule of law is respected, “rule breaking” in this second context may not be acknowledged as such. Indeed, what would be considered as rule breaking where rule of law is the norm, may actually be justified on the basis of the narrow moral order: it is a wrong that has entered into the actions or omissions of thousands, perhaps millions, of people, becoming, in effect, a counter-norm of its own. As long as the moral order confirms their actions, the place of these perpetrators remains within that order. And insofar as the laws conform to that order, then there will be no need for isolation of perpetrators because their actions are condoned, even encouraged by the ethos of the law.

It is for this reason that recourse to a court of law, while not without value and a role to play, does not suffice to meet the challenge facing transitional societies. If retribution, as argued earlier, seeks to encourage acknowledgement of wrongdoing at the individual level, and moral education at the societal level, it cannot achieve these goals if acknowledgement of a primary moral order that dictates rules is not first secured. The victims suffered what they suffered because a wrong assumed a systematic, typical and normative character, and it is that character that needs to be uncovered and condemned. This is not inconsistent with calling individual perpetrators to account, whatever practical or political obstacles may obstruct that—the process of uncovering and condemning the larger pattern of wrongdoing may, for example, coexist with a criminal process that takes account of levels of degree and responsibility. To the extent that the truth-recovery process affirms a status that in the victim's case was violently denied, its aim is identical to that of retributive justice, whether it replaces or supplements punitive practice itself. It takes a different form because of the difference—a morally significant difference, not merely an empirical difference—between abnormal disrespect within a system of law and normal denial in a context in which the operation of law itself has become systematically exclusive.

III. Responsibility in Context

Above, we noted that the circumstances of a criminal trial serve to isolate the suspect: her communications with others are curtailed, her actions are subjected to a level of scrutiny to an unparalleled degree, and at the same time entirely exceptional safeguards are extended, in terms of the admissibility of the evidence that may be brought against her. Trials are concerned with facts as they pertain to a particular and negative interaction between particular individuals. Minow points out how trials “interrupt and truncate victim testimony with direct and cross examination and conceptions of relevance framed by the elements of the charges. Judges and juries listen to victims with skepticism tied to the presumption of the defendants' innocence” (1998: 238). The suspected offender is taken seriously as an agent; this point is particularly stressed in Rachels' defence of retributivism: in ways that ignore familiar kinds of restraint on agency, she is taken to be the sole authentic author of the actions that the trial process allows to be attributed to her. Any larger context of events is excluded from the attribution of responsibility, even if, as with remorse, it may mitigate sentencing. One person is on trial—not the group that she belonged to, the incentive structure that she was exposed to, or the influences, however powerful, that entered into her motives. The suspected offender is, so to speak, de-contextualized in the name of establishing individual agency, as a prelude to individual acquittal or punishment. All sorts of considerations that, from a purely moral or historical point of view, might be relevant to understanding the offender's acts, are rigorously excluded. All that is included is evidence about what, within narrowly circumscribed parameters, can reliably be attributed to the offender's individual agency.

Clearly, there is a lack of fit between such narrow parameters and what is called for in the case of massive atrocities or oppressive regimes. What is called for may be described, in fact, as recontextualization. For in the situations that we are considering here, the place of decontextualized individual agency is extraordinarily problematic. The final report of the South African TRC argued that the narrow focus of the Nuremberg and Tokyo tribunals meant that “many perpetrators and co-conspirators remained in obscurity. The structures of society and its most formative institutions remained unchallenged” (quoted in Dyzenhaus, 2000: 480). Clearly, such a situation would be unacceptable in cases like South Africa, where the apparatus of apartheid was propped up by numerous mechanisms at the disposal of the state.

Hannah Arendt contributed a classical discussion of this issue (1965). Great evils may not necessarily express the aggregate of greatly evil individual acts. They may express an aggregate of banal acts, especially in the context of bureaucratized systems in which power is diffused and accountability rendered impersonal and opaque, in such a way that few individuals can be described as intending the overall result: what they intended, or had in mind as a deliberate purpose, may have been quite mundane. Rather different considerations are advanced by David Cooper, who discusses a case in which participants can be considered to have intended the overall result, but in which their agency was diminished by what he terms a “cognitive model” (2001: 211). Cooper asks what difference it should make to our judgments, as observers, that we were not in the participants' shoes, and have been lucky enough never to have faced the choices that they faced. The question we must consider, he says, is not whether we, with our existing beliefs and commitments, would have done as they did: the question is, rather, “whether I, if brought up in a very different climate of beliefs and values … would have acted as they did” (2001: 211). He concludes that the cognitive distortions of perpetrators should make a difference to our judgments, even though we may hold the perpetrators themselves morally liable for acknowledging responsibility.

Arendt's and Cooper's approaches bring out different ideas of diminished agency, which we may call causal and cognitive. In the first idea, agency is diminished because causal connections between act and result are so fractured, while in the latter idea agents are submerged in a climate of fear and prejudice that systematically clouds reality. But in both cases, what needs to be done is to recover the social, organizational and ideational context that surrounded agency, and to record its origins and results. This is just what a trial does not do. As before, however, the objective is entirely continuous with that of a trial, in terms of moral education and equilibrium. A social context, it is true, cannot be prosecuted or punished in the way that an individual perpetrator can be, but if we consider the matter in relation to victims, rather than perpetrators, then truth processes and trials more nearly converge. If the aim is to uncover and condemn the violent loss of status inflicted on victims, then in cases of mass atrocity and systematic oppression a contextualized account will accomplish this much better.

We can see a further or related aim of recontextualization. Allen poses the question thus: “What kind of disposition must be widespread if evil and unjust social arrangements are to be identified and resisted?” (1999: 335). The social, organizational and ideational context in which atrocity and systematic oppression occur is not only a matter of diminished agency; it is also, and importantly, a matter of a greatly diminished sense of injustice. Where atrocity and oppression has become systematic, recognition of the consequences as injustice has important implications: “a sense [of injustice] is important if government during a transition is to preserve legitimacy and a commitment to the constraints of the rule of law, constraints which make it possible for citizens to call government to account for injustice” (quoted in Dyzenhaus, 2000: 484).

In this way, then, recontextualization will more accurately represent, and hold up for judgment, the processes from which victims suffered. It will establish that those processes were not, as it were, aggregates of crimes that happened to occur in parallel, but processes directed against people identified and deprived of moral status, on the basis of their group membership, and that fact has to be central to the recovery and condemnation of the wrong that was done. In accomplishing that, truth commissions, unlike trials, can draw in the context in which perpetrators acted, draw out the common elements in victims' experiences, and in some cases, investigate the involvement of whole institutions.

IV. On Truth

While there is an obvious sense in which criminal trials are concerned with the truth of the matter before them, they can be regarded as truth-seeking processes only in a qualified way. First, they can consider, not all the evidence, but only admissible evidence: since the overriding consideration is the just application of the law to the suspected offender, procedural safeguards stand in the way of constructing what for other purposes would be a true account of the matter, such as a historian's account. So, for example, the accused's previous record—which historians would regard as particularly rich material, indispensable in fact, for constructing an explanation of his acts—is inadmissible, as is evidence gathered by means that are forbidden for reasons that have nothing to do (directly) with the truth-value of what it may reveal. A second qualification, different but related, concerns what may be termed “personal truth,” that is, the victim's own subjective experience of events. As Minow points out, the adversarial feature of trials—itself justified as a procedural safeguard of the rights of the accused—does not lend itself at all well to the recounting of narrative that victims often seek (1998: 238). Cross-questioning and procedural challenges interrupt story-telling, reducing evidence to an atomic registry of carefully sifted factual claims; and yet, facts “do not stand alone but have to be understood in order to function in discourse—which is always understanding in a certain way, one that is not necessarily shared by others” (Parlevliet, 1998: 145). In the process both of writing history and of the trial, “[t]he narrator—single or collective—arranges testimonies in an order that seems self-evident, but is necessarily artful” (Maier, 2003: 271). Especially in trial proceedings, although witnesses have stories to tell, their stories are not valued as stories of experienced loss but for their contribution to the larger narrative, which either the defense or prosecution is attempting to produce, in order to establish innocence or guilt respectively.

So truth commissions may differ from trials in those two ways, at least. The former difference is most easily accommodated: where no one is on trial, demands for procedural safeguards can be relaxed. Those safeguards are not based on any view that truth is unimportant: they are based on a principle of over-protection that reflects a moral view that the punishment of the innocent must be avoided even at a high cost in terms of truth. There is, then, a difference of aim, but not a conflict of aims, for the principle of over-protection does not in any way conflict with belief in the importance of truth-recovery. It is the second difference that may seem harder to accommodate, for, as we have already seen, it is the virtual exclusion of the victim's perspective that is often taken to be among the greatest weaknesses of criminal trials. Conceived of as encounters between the accused and the sovereign, they marginalize the victim, whose own subjective suffering is implicitly conceptualized as nothing more than a by-product of the offence itself. “Victims' rights” initiatives attempt to remedy this by seeking to give victims the right to have their perspectives made directly present to the court. Such initiatives are, however, very much an add-on to a process which is still at heart Sovereign vs. the Accused; and if truth commissions give a central privilege to victims' narratives, they may well seem to diverge quite radically from the ideal type of a criminal trial.

At a level of abstraction that is only one step higher, however, the difference tends to dissolve. Within a functioning, rights-respecting system of law, victims suffer wrongs that the system takes it upon itself to acknowledge by coercively insisting on the importance to the public of the violated right. The victim can call upon the system to act, and her moral standing is acknowledged in her access to the police and to legal advice. Her subsequently marginal role is a procedural artifact that has nothing to do with a denial of her moral standing. But as we have seen, in post-atrocity contexts it is just that standing that needs to be reaffirmed. It needs to be authoritatively and publicly established that classes of people were wrongly denied standing. And the importance of “personal truth,” or the recounting of narrative, in such circumstances, can readily be understood: it expresses the fact that those whose humanity or citizenship was denied were agents and patients with a subjectively experienced view of the world that atrocity or oppression sought to obliterate. As in the case of agency discussed earlier, causality is crucial, for implied in a chain of causality is a judgment as to why the events played out in the fashion they did. We accept this in productions of history and in trials, both of which “presuppose a narrative; that is, a coherent account in which earlier events are cited to account for later ones” (Maier, 2003: 271). Truth commissions also presuppose and seek out such a coherent account, and in this way the purpose of a commission, of a political trial, and of the historian converge: each, in its own ordering of events and rules for admission of evidence, seeks to convey “not just violence or repression or conflict, but a reason for even the most brutal confrontation” (Maier, 2003: 272). In the process of a commission, the narrative of the victim works to dismantle the myths and denials that made widespread atrocity or systematic oppression possible. This very attempt to provide a different narrative, a different reason, is crucial to the idea of reestablishing a moral equilibrium among citizens of a transitional society.

This justification, which builds on the notion of recognition of standing that is implicit in criminal trials, may have advantages that other justifications lack. In particular, it is immune to the problems faced by the once-popular appeal to cathartic effects. That those who have suffered wrongs experience release by re-telling—the “revealing is healing” hypothesis—has a priori elements (or perhaps religious assumptions) that experience may not bear out (Shaw, 2005). People may be too different for any such psychological generalization to apply and, in any event, the process of personal recovery is likely far too extended to be accomplished in a single institutional event. The claim advanced here is that the opportunity for personal narration is the exact and appropriately recast equivalent to the enjoyment of legal standing within a functioning system. It looks different, but the affirmative aim of restoring equality is no different.