Retribution, Crime Reduction and the Justification of Punishment

Transcription

1 Oxford Journal of Legal Studies, Vol. 22, No. 2 (2002), pp Retribution, Crime Reduction and the Justification of Punishment DAVID WOOD Abstract The dualist project in the philosophy of punishment is to show how retributivist and reductivist (utilitarian) considerations can be combined to provide an adequate justification of punishment. Three types of dualist theories can be distinguished split-level, integrated and mere conjunction. Split-level theories (e.g. Hart, Rawls) must be rejected, as they relegate retributivist considerations to a lesser role. An attempted integrated theory is put forward, appealing to the reductivist means of deterrence. However, it cannot explain how the two types of considerations, retributivist and reductivist, are to be genuinely integrated as opposed to merely conjoined. An attempt to find integration at the deeper level of political philosophy is then examined, in the form of Lacey s communitarian theory of punishment. In the end, mere conjunction dualism must be accepted by default. This conclusion points to a deep-lying schizophrenia in our substantive criminal law. Although developing this theme is beyond its scope, the paper ends by saying a little on how a mere conjunction dualist theory of punishment leads on to a similar theory of criminalization. 1. Introduction A major standing concern in the philosophy of punishment is to develop a theory which successfully combines retributivist considerations on the one hand, and utilitarian considerations (or what will be referred to here more broadly as reductivist considerations 1 ) on the other. The central premise of the dualist Law Faculty, University of Melbourne. Earlier versions of this paper were presented at the 2000 Annual Conference of the Australian Association of Legal Philosophy, and at a University of Melbourne Philosophy Department Staff Seminar. I wish to thank Andrew Ashworth, Jim Harris, Sue Harris, David Lyons, Nicola Lacey, Vanessa Mitchell and the Journal s anonymous referee for helpful suggestions 1 The term reductivism appears to have been coined or at least appropriated to this context by Nigel Walker, who confesses to its being rather ugly (Sentencing in a Rational Society (Penguin, Harmondsworth, 1969) at 18). Reductivism holds that the justification for penalising offences is that this reduces their frequency (Nigel Walker, Punishment, Danger and Stigma (Blackwell, Oxford, 1980) at 26). Walker points out that this term is more precise than utilitarianism, since utilitarianism is really concerned with maximising the sum of human happiness, and could conceivably argue that penalties severe enough to make a real impact on the frequency of, say, motoring offences would generate more unhappiness than they would prevent (Punishment, Danger and Stigma at 26 27). Reductivist considerations are broader than utilitarian considerations, as one can have non-utilitarian reasons for reducing crime, for instance, reasons concerning the rights that criminal activity violates or interferes with one does not have to be a utilitarian to be a reductivist. On the other hand, there may be utilitarian benefits which extend beyond reductivist ones, such as greater social cohesion, and an improved sense of community. (See, for instance, Michael Moore, Placing Blame: A General Theory of Criminal Law (Oxford University Press (Clarendon), Oxford, 1997) at 85.) Or consider more generally, improved general living conditions, enabling people to lead happier and more fulfilled lives Oxford University Press

2 302 Oxford Journal of Legal Studies VOL. 22 project (as it could be called 2 ) is that an adequate account of punishment cannot be purely retributivist or purely reductivist, but must find room and appropriate roles for both types of considerations. The question is how this is to be done. This project admittedly no longer occupies centre-stage as it once did. The resurgence of retributivism in the last two or three decades is sure evidence that many philosophers simply reject the above premise. 3 There has been a broadening of interests in the philosophy of punishment, some developments arising out of this trend, others in reaction to it, and others again quite independent of it. Consider, for instance, the numerous attempts to develop expressive, moral educative, and communicative theories of punishment. 4 Nevertheless, this paper takes the central premise of the dualist project as its starting point its working assumption is that it is the basic task in the philosophy of punishment to develop a theory which adequately handles both retributivist and reductivist considerations. If this goes against the trend of recent developments in the philosophy of punishment, no apology is made. It may be necessary to turn the clock back in order to turn it forward again. The paper distinguishes between three types of dualist theories, to be referred to as split-level, integrated and mere conjunction theories respectively. 5 Part 2 points out a major, but not necessarily unfamiliar, objection to theories of the 2 Other terms include pluralist, mixed, hybrid, synthetic, compromise, intermediate and middle-ofthe-road. To take two instances, in Crime, Guilt and Punishment (Oxford University Press (Clarendon Press), Oxford, 1987) at 6, C.L. Ten adopts a pluralist approach to punishment, believing that in the practice of punishment we are often confronted with a number of different considerations, each not reducible to the others, and each having a contribution to make. Ten therefore identifies as his main target, any theory of punishment, whether it be utilitarian or retributive, which seeks to justify punishment in terms of just one ultimate value. Similarly, Nicola Lacey defends a conception of punishment which is pluralistic... in that it identifies punishment as pursuing and respecting a set of different values which cannot necessarily be reduced to each other, or into some overriding common currency such as general utility (State Punishment: Political Principles and Community Values (Routledge, London, 1988) at 199) Lacey s theory, and much more briefly Ten s, are discussed in Part 4. See also Igor Primoratz, The Middle Way in the Philosophy of Punishment in Ruth Gavison (ed.), Issues in Contemporary Legal Philosophy: The Influence of H.L.A. Hart (Clarendon, Oxford, 1987) at 193 and Justifying Legal Punishment (Humanities Press International, New Jersey, 1989), Ch. 6 and Wesley Cragg, The Practice of Punishment: Towards a Theory of Restorative Justice (Routledge, London, 1992), Ch See, for instance, Anthony Ellis, Recent Work on Punishment (1995) 45 Philosophical Quarterly 225 at ; R.A. Duff, Penal Communications: Recent Work in the Philosophy of Punishment (1996) 20 Crime and Justice: A Review of Recent Research 1 at 1 3 (ed. Michael Tonry; University of Chicago Press); Moore, above n 1 at 83; Ted Honderich, Postscript: The New Retributivists and Political Philosophy in Punishment: The Supposed Justifications (revised ed.; Penguin, Harmondsworth, 1984) at 208. Neither is the trend confined to academia, extending to sentencing policy and practice, and criminal justice thinking generally. See, for example, D.J. Galligan, The Return to Retribution in Penal Theory in C.F.H. Tapper (ed.), Crime, Proof and Punishment: Essays in Memory of Rupert Cross (Butterworths, London, 1981) at 144. Note, however, that Duff sees the general trend as one of attacks on consequentialist theories, the development of retributivist theory being one of three main responses to this trend, the other two being the development of more sophisticated consequentialist accounts, and of teleological accounts that are neither strictly consequentialist nor purely retributivist (above at 1, 7). 4 To give examples of these three types of theories respectively, see Joel Feinberg, The Expressive Function of Punishment in Doing and Deserving (Princeton University Press, Princeton, 1970) at 95; Jean Hampton The Moral Education Theory of Punishment, (1984) 13 Philosophy and Public Affairs 208; and R.A. Duff, Trials and Punishment (Cambridge University Press, Cambridge, 1986) esp. 233, 238. Duff provides a useful introduction to his communicative theory in his article In Defence of One Type of Retributivism: A Reply to Bagaric and Amarasekara, (2000) 24 Melbourne University L Rev 411 at See also his Desert and Penance in Andrew von Hirsch and Andrew Ashworth (eds), Principled Sentencing: Readings on Theory and Policy 161 (2nd edn, Hart, Oxford, 1998), and more recently, Punishment, Communication and Community (Oxford University Press, New York, 2001). 5 For other ways of classifying dualist or mixed theories, see Moore, above n 1 at 93, and Duff, Penal Communications at 7,

3 SUMMER 2002 Retribution, Crime Reduction & Justification 303 first type, that is, theories which attempt to assign reductivist and retributivist considerations to different levels. Briefly, the objection is that such theories do not take retributivist considerations seriously enough, but relegate them to a lesser role. The rejection of split-level dualism leaves the choice between integrated and mere conjunction dualist theories. Part 3 sketches an attempt at the former, that is, a theory which tries to genuinely combine the two types of considerations, rather than assigning them to different levels. The paper concentrates on the reductivist means of deterrence (whether specific or general), although other reductivist means, such as incapacitation, rehabilitation, and denunciation could also be considered. 6 The point at issue, however, to be explained below (briefly, that of harsh and unpleasant measures which require stronger justification than can be provided by their reductivist value), arises most graphically in the case of deterrence. Part 4 takes up a basic issue with integrated dualism, namely that of explaining how reductivist and retributivist considerations are to be integrated. It considers the suggestion of Nicola Lacey that integration is to be sought at the deeper level of political philosophy. 7 It is argued that, conceived of as an attempted integrated dualist theory, Lacey s theory fails. Moreover, the reason it fails suggests that the prognosis for finding a plausible version of integrated dualism is not good. Given the prior rejection of split-level dualist theories, mere conjunction dualism must be accepted by default the mere conjunction of retributivist and reductivist considerations is the most that reasonably can be achieved. This is scarcely intellectually satisfying, but it is the best that can be done. No progress can be made beyond merely adding the two types of considerations together, relying on one to make up for the justificatory shortfall left by the other. To use a chemical analogy, the result is a mixture, not a compound. This conclusion has ramifications for our understanding of the substantive criminal law, for developing a theory of criminalization, conceived of as a coherent set of criteria for determining what conduct should be criminalized, and how seriously different forms of such conduct should be regarded. 8 It is only to be expected that the body of criminal law doctrine of substantive offences, defences, and other general doctrines which extend criminal liability, for instance, to inchoate offences and accomplices is permeated by the same tension between 6 See further n Lacey, above n2at53,65 6, This is a matter on which very little work has been done, the notable exception being Andrew von Hirsch and Nils Jareborg, Gauging Criminal Harm: A Living Standard Analysis (1991) 11 OJLS 1. See also von Hirsch, Censure and Sanctions (Oxford University Press (Clarendon), Oxford, 1993) at For discussion, see Ashworth, Principles of Criminal Law (3rd. ed.; Oxford University Press, Oxford, 1999) at 39 42; Ashworth, Sentencing and Criminal Justice, Ch. 4 (3rd edn, Butterworths, London, 2000); and Nigel Walker Harms, Probabilities and Precautions, (1997) 17 OJLS 611, See also Martin Wasik, Crime Seriousness and the Offender-Victim Relationship in Sentencing in Andrew Ashworth and Martin Wasik (eds), Fundamentals of Sentencing Theory: Essays in Honour of Andrew von Hirsch (Oxford University Press (Clarendon), Oxford, 1998) at 103.

4 304 Oxford Journal of Legal Studies VOL. 22 retributivist and reductivist considerations. 9 If these two types of considerations do not cohere at the level of justification of punishment, there is no reason to suppose that they do at the level of a theory of criminalization. What is to be expected, then, is that embedded in any body of substantive criminal law is a deep-seated schizophrenia, arising from an irresoluble tension between reductivist and retributivist considerations. This conclusion, likewise, has ramifications for the question of the proper ambit of the criminal justice system, as against harm reductivist institutions which serve no retributive purpose, for instance, institutions for the quarantining of suspected carriers of life-threatening diseases, or for the detention of the psychiatrically dangerous. 10 Most importantly, although penal measures require both retributivist and reductivist justification, one should not expect to find coherence between them. Exploring these two overlapping sets of ramifications, however, must be regarded as beyond the scope of this paper, although the paper ends by seeing how a mere conjunction dualist theory of punishment leads on to a similar theory of criminalization. 2. Split-level Dualism The best-known dualist theories are Rawls 11 and Hart s. 12 Rawls is a ruleutilitarian theory, and so involves a distinction between two stages. First, the institution or practice of punishment as a system of rules is justified by appeal to utilitarian considerations. Second, retributivist considerations enter in at the level of particular acts falling under the rules. Rawls theory of punishment has remained prominent, despite the general demise of rule-utilitarianism a trend in which Rawls, of course, in virtue of his subsequent work, was a major player. 13 Hart s theory similarly involves a distinction between two stages. First, there is the question of what he calls the general justifying aim of punishment; second, there is the question of distribution, which is itself divided into two further 9 Note, however, the observation that [i]n recent years, self-consciously philosophical discussion of the proper scope and structure of the criminal law has been treated by many writers as a mere adjunct to discussion of the justification of punishment. (Stephen Shute, John Gardner and Jeremy Horder, Introduction: The Logic of the Criminal Law in Shute, Gardner and Horder (eds), Action and Value in Criminal Law (Oxford University Press (Clarendon), Oxford, 1993) at 1, 2. I certainly am not recommending such a mere adjunct approach, but on the other hand have no qualms about being categorized with the alleged culprits, namely, Lacey, Norrie and Ten. 10 See Walker, Punishment, Danger and Stigma, Ch. 5 and Duff, Trials and Punishments at Harm reductivist institutions measures, practices, agencies, and so on could be considered collectively as comprising a social harm reduction system, although the term system may suggest greater co-ordination and organization than actually exists, is necessary or even desirable. See David Wood, Reductivism, Retributivism, and the Civil Detention of Dangerous Offenders (1997) 9 Utilitas: A Journal of Utilitarian Studies 131, John Rawls Two Concepts of Rules, (1955) 64 Philosophical Review H.L.A. Hart Prolegomenon to the Principles of Punishment, Punishment and Responsibility: Essays in the Philosophy of Law, Ch. 1 (Oxford University Press (Clarendon Press), Oxford, 1968). 13 John Rawls, A Theory of Justice (2nd edn, Oxford University Press, Oxford, 1999); Political Liberalism (Columbia University Press, New York, 1993). The main criticism is that rule-utilitarianism collapses into actutilitarianism. See David Lyons, The Forms and Limits of Utilitarianism (Oxford University Press, Oxford, 1965). As Ten puts the point: [b]ut why would anyone, motivated solely by utilitarian considerations, act in accordance with a rule even on those rare occasions when it is known that obeying the rule will fail to produce the best consequences?, at 70.

5 SUMMER 2002 Retribution, Crime Reduction & Justification 305 questions, of liability (who to punish?) and amount (how much to punish?). Utilitarian or reductivist considerations are relevant to the first question, of why have an institution of punishment at all, whereas appeal is made to retributivist considerations and considerations of justice and fairness more generally at the level of the second question in order to restrict punishment to those who deserve it, 14 and the amount of punishment to the extent of their culpability. In distinguishing between their respective two stages, Rawls and Hart could be described as split-level dualists utilitarian or reductivist considerations are accounted for at the first stage or level, and retributivist considerations at the second. 15 As such, they face the criticism that, in confining retribution to the second level, and so restricting it to grounding constraints on the pursuit of utilitarian or reductivist goals, they do not treat it seriously enough. Rather than merely being another crime reductivist institution or practice, punishment is special in having retribution as its pivotal notion. As D.J. Galligan points out: [t]o many people there is a deeply embedded intuition that part at least of the general purpose of criminal justice is the correction of wrongs, quite apart from any consequential contribution to crime control Namely, it might be suggested, those found guilty of a criminal offence. However, it may be questioned both whether this is a necessary condition and a sufficient condition of justified punishment. Concerning the former, there are those who have obviously done morally wrong acts, but have not been convicted of a legal offence. Consider war crimes, in particular crimes against humanity. On the question of limiting punishment to the legally as opposed to morally guilty, see Ten, above n 2 at 71ff. Concerning the latter, there are at least two reasons for guilt not being a sufficient condition. First, there may be a moral gap between guilt and responsibility. A may clearly have performed the criminal act, but whether he is fully or even partially responsible is another matter. Consider criminal defences such as insanity, automatism, provocation and diminished responsibility. Second, and conversely, a legal offence may involve little or no moral wrongfulness. Consider some strict liability offences, although some critics of strict liability try to make too much of this point. (For discussion, see Wojciech Sadurski, Distributive Justice and the Theory of Punishment (1985) 5 OJLS 47, 49 50; Giving Desert Its Due (Reidel, Dordrecht, 1985) at ) Indeed, a legal wrong may be morally praiseworthy. Consider a popular rebellion against a tyrannical regime. 15 For instance, as Ten points out in relation to Rawls, he thus reconciles the utilitarian theory with the retributive theory by assigning them to different levels of justification and thus avoiding conflicts between them at 68. Split-level dualism provides what could be called a structured hybrid justification of punishment. (See John Gardner, Crime in Proportion and in Perspective in Ashworth and Wasik (eds), above n 8 at 31, 32n.) 16 Above n 3 at 151. C.L. Ten says similarly, if less emphatically, that retributive considerations can provide positive reasons in support of punishment, above n 2 at 84. A related criticism is that Hart does not establish the link or connection between the general aim and the principle of distribution (see, for instance, Galligan, above n 3 at 150, although it is certainly made by others). According to this criticism, it is arbitrary to appeal to utilitarian or reductivist considerations only at the first level of the general justifying aim, and to retributivist considerations only at the second level of distribution. To so move from one type of consideration to the other is like changing horses mid-stream. As Galligan puts it, [s]o the reasons given for punishing at all, and the guiltiness of the offender float, as it were, side by side, without being able to make contact. The split-level dualist appears to be opportunistic, appealing alternatively to reductivist or retributivist considerations as convenient. He must show how the questions which are to be answered appealing to reductivist and retributivist considerations respectively can be genuinely kept separate. Indeed, it could be argued (I owe this point to Sue Harris) that the difficulty with split-level theory may be precisely this, that it arbitrarily assigns retributivist and reductivist justifications to different levels, rather than that it assigns a lesser role to retributivism. The split-level theorist, particularly Hart, could argue that his theory places the punishment of criminal behaviour in a sub-set of a social harm reduction system (see n 12). The criminal justice system likewise has crime reduction as an aim, the factor which distinguishes punishment from other sub-sets of the reduction system being its retributive nature. In this sense retribution is the distinguishing feature of punishment, but it does not follow that it is necessarily the most important factor. According retribution a lesser role is not incompatible with recognising its central role. A brief response is that what distinguishes punishment from other social harm reduction institutions is not just at the level of distribution (although this could be sufficient), but at the level of general aim or purpose. We are, then, making the stronger claim, that punishment is distinguished both at the level of general aim and distribution, although the latter is, as this criticism rightly points out, sufficient to distinguish punishment from other harm reduction institutions.

6 306 Oxford Journal of Legal Studies VOL. 22 Moreover, there seems no way of accommodating this intuition within splitlevel dualism. Having separate levels requires keeping the materials to the questions at each level separate, and this means confining retributivist considerations to the second level, and so giving them (according to this criticism) a lesser role. This intuition requires an alternative type of dualist theory, which could be referred to as integrated dualism. The integrated dualist realizes, in contrast to retributivists, that although retribution requires a more significant role than that accorded to it by split-level dualism, reductivist considerations cannot be ignored or marginalized. 17 Such considerations are necessary to justifying a system of punishment. Retribution does not have sufficient weight to justify punishment as a social institution that is, to do so completely or at least sufficiently. Retribution may do so partly, but it is unduly moralistic to create a system of punishment for the sole purpose of retribution. 18 There is insufficient reason to systematically pursue retribution as an end in its own right. 19 Note also that in addition to split-level dualism, there is what could be termed reverse split-level dualism, on the grounds that it changes around the order of retributivist and reductivist considerations. For instance, as Lacey states, in this type of theory, desert is the central justifying factor, whilst utilitarian justification is required for the infliction of punishment in individual cases. Utilitarianism thus acts as a limitation on the desert principle, above at 55. Such a theory is put forward by A.H. Goldman in The Paradox of Punishment (1979) 9 Philosophy and Public Affairs 42, see further n 26. From our viewpoint, reverse split-level dualism is no improvement on ordinary or obverse split-level dualism: it merely replaces the problem of not taking retribution seriously enough with that of not taking crime reduction seriously enough. As pointed out in the text shortly below, we need both crime reduction and retribution in the general justifying aim. 17 Neither, it may be added, can they be relegated to a lower level, as does reverse split-level dualism. 18 Kant notoriously held that if there were only two people left in civil society, a condemned man and a hangman, the latter should still carry out the sentence (The Metaphysical Elements of Justice (John Ladd trans.; Bobbs-Merrill, Indianapolis, 1965) at 102). For discussion, see, for instance, Ten, above at 75; Moore, above n 1 at 99n. The integrated dualist can say, quite apart from any reservations about capital punishment, that the sentence should not be carried out on the grounds that no useful social purpose is thereby served crime reduction is just not at issue. 19 At least, this is the case at the social if not private level. As Galligan points out, [w]e need to draw an essential distinction between retribution as a moral principle which we might employ in our personal lives, and retribution as a social principle which justifies institutionalized state coercion of individuals above n 3 at 153. Retribution may be acceptable as a private moral principle, but it is another question whether it is justified as a principle of social action. This is not just because of the vast allocation of resources required to sustain the network of criminal justice institutions (courts, prisons, remand centres, police forces, and so on), but even more so because of the human and other social costs and perhaps most importantly the moral costs of the risk of wrongful convictions and their consequences of unjustified punishment and stigma. On the other hand, retributivists obviously see no difficulty in retribution being a sufficient social end in its own right. Von Hirsch says that [d]esert may be viewed as reason in itself for creating a social institution (Doing Justice: The Choice of Punishments (Report of the Committee for the Study of Incarceration) (Hill and Wang, New York, 1976) 52 53). And Moore says that [r]etributivism... is truly a theory of justice such that, if it is true, we have an obligation to set up institutions so retribution is achieved above n 3 at 91. Note also Douglas Husak s view. He recognizes that the value of realizing retributive justice, by itself, is insufficient to justify the creation of an institution of punishment with [its] formidable drawbacks ( Retribution in Criminal Theory, 37 San Diego L Rev 959, (2000)). In addition to the two drawbacks mentioned above, of economic costs and the moral costs of wrongful convictions, Husak adds a third, of the inevitability of abuse of authority. According to Husak, some additional value must be found to offset these drawbacks, and he submits that this value is crime reduction, above at However, overcoming the economic costs of punishment is one thing, but overcoming the moral costs of wrongful convictions, and furthermore of abuse of authority, is quite another. These moral costs may constitute not just an insuperable argument against the most severe and irreversible possible forms of punishment, such as capital punishment. They may ultimately render any system of social or public punishment unjustified. The more general issue, however, is whether crime reduction is seen as having independent value in its own right, and not just having a negative function of neutralising objections to punishment, in a generally retributivist theory of punishment. As mentioned in the text below, we see crime reduction as compensating for the shortfall

7 SUMMER 2002 Retribution, Crime Reduction & Justification 307 On the one hand, then, it is insufficient to relegate retribution to the secondary level of constraints on the pursuit of the goal of crime reduction, as does splitlevel dualism; on the other hand (however much retributivists may seek to deny this) retribution is not by itself capable of justifying punishment as a social institution. The guiding thought behind integrated dualism is that, even though retribution does not by itself offer a sufficient justification of punishment as a social institution, it does once reductivist benefits are taken into account such benefits compensate for the shortfall in the justificatory force of retribution. 3. An Attempted Integrated Dualist Theory: Crime Reduction through Deterrence According to integrated dualism, then, retribution is pivotal to punishment. In relegating retributivist considerations to the secondary level, split-level theories fail to properly account for them. Punishment is not a purely harm reductivist institution in contrast with the institutions already mentioned for the quarantining of suspected carriers of life-threatening diseases and the incarceration of the psychiatrically dangerous. Being at the heart of a system of punishment, retribution cannot be restricted to acting merely as a constraint on punishment, limiting the conditions under which a person can be punished (for instance, to those guilty of an offence), and the extent to which he or she can be punished (for instance, to no more than is commensurate with or proportionate to the offence). With all due respect to Hart, retribution is integral to the general justifying aim of punishment. However, although retribution is necessary to the justification of punishment as a social institution, it is not sufficient. As just pointed out, it is overly moralistic to suggest that retribution is weighty enough to be the sole justification. Retribution only gets us part of the way reductivist justification is required as well. 20 One could try to sketch an integrated dualist theory as follows. 21 Retributivist measures are necessarily harsh or unpleasant. (This harshness or unpleasantness could take different forms, for instance corporal or even capital punishment, and obviously extends to the deprivation of liberty. 22 ) If they were not harsh or in the justificatory force of retribution. In contrast, Husak sees retribution as having sufficient justificatory force in its own right, but crime reduction as required to offset drawbacks of a system of punishment. (Why suppose, however, that retribution and retributive justice are that important?) Unfortunately, the matter cannot be pursued further here neither can von Hirsch s views on the role of crime reduction in a basically retributivist theory of punishment in Censure and Sanction, which Husak also refers to (at 977, n 73). 20 Note that the shortfall from retributivist justification could be made up by some third type of justification, for instance, rehabilitation and denunciation, considered as ends in their own right, rather than means to the end of crime reduction. Or consider denunciation as a means to a further, or different end, for instance helping to maintain social cohesion or provide a sense of community. This could be considered as an end in itself, irrespective of any crime-reductive consequences (see for instance, Moore, above n 1 at 84 85). The basic contrast is between retributivist and non-retributivist justification generally, and not just between retribution and crime reduction. 21 For an earlier, partial statement, see Wood, above n 10 at And also, presumably, to financial penalties. However, this raises the question of how harsh or unpleasant they are to those who are sufficiently well-off to pay them without any effect on their welfare or well-being. Consider, however, proposals for unit or day fines. See, for instance, von Hirsch, Censure and Sanctions at 4.

8 308 Oxford Journal of Legal Studies VOL. 22 unpleasant, they would be incapable of serving a retributivist purpose or function. This would negate the distinction between punishment and reward, which is clear in theory, whatever difficulties may arise in practice. (The standard example is the masochist what is harsh or unpleasant for the overwhelming majority is not necessarily harsh or unpleasant for all. 23 ) However, the same harshness or unpleasantness gives retributivist measures reductivist value as well, through their capacity to serve as deterrents (whether specific or general). 24 Deterrence measures likewise are necessarily harsh or unpleasant (at least for all practical purposes 25 ). The distinction between deterrents and incentives or forms of encouragement (between sticks and carrots) is clear in theory, just as that between punishments and rewards again, whatever difficulties may arise in practice. Harsh or unpleasant measures, then, are capable of serving both retributivist and reductivist (or more exactly, crime-reduction-through-deterrence) functions. 26 However, a measure required on crime-reduction grounds may not be justified on such grounds alone, because its harshness or unpleasantness may not be warranted by its reductivist value. The measure may rate highly on any 23 It may be asked, however, whether this is a case of the exception proving the rule: it is because there is general agreement as to what measures are harsh or unpleasant that they are capable of serving a retributive function (and, as seen in the text immediately following, a deterrent function as well). To say, as we have, that retributivist measures are necessarily harsh or unpleasant does not mean that they must always be felt as such. See von Hirsch, Censure and Sanctions at Of course, they may have reductivist value through some other reductivist means, such as incapacitation, rehabilitation, or denunciation. To take the most obvious case, imprisonment may not just incapacitate, but have a deterrent function, and possibly even some denunciatory or moral educative effect. However, deterrence is the most obvious reductivist means for our purposes, which is why we are concentrating on it. In the case of effectively serving a general deterrent function, it is not claimed that it is sufficient that a measure is harsh or unpleasant. Of course, the existence and imposition of the measure must be known, and generally accepted to be harsh or unpleasant (whatever masochists may think). Some causal, epistemological account of the relation between the deterrent and the conduct to be deterred is required (however, see further n 25). The parallel with retribution (insofar as there is one) is that the measure must not simply be harsh or unpleasant, but imposed for the purpose of retribution. 25 Of course, all that is strictly necessary is that the measures are perceived as harsh or unpleasant, and this might be achieved as the result of deception. As Lacey points out specifically in relation to general deterrence, its effect is achieved by means of the threat of punishment: all, therefore, that is needed is a general belief in the reality of the threat; in fact the whole system could be a sham, if it were feasible to maintain the presence over a long period at 29. It is an intriguing point about the logic of deterrence (and indeed its morality too), that the most effective deterrent is the one that is so severe that it never has to be carried out. The threat to execute bicycle thieves (to take our example below) may be sufficient to eliminate the crime. 26 Von Hirsch also discusses the interdependence of desert and deterrence (Doing Justice at 49 55). He suggests a route from deterrence to desert (49 52), and then a converse route from desert to deterrence (52 55). Our argument could be seen as a development of the first route. Note also von Hirsch s more recent attempt to clarify the relation between the deterrent and retributive (or censure) elements of punishment. This could be viewed as an attempt at an integrated dualist theory of punishment, and so warrants closer attention than we have time to give it here. (See Censure and Sanction, 9 14, and especially 13 14, on The Relation between the Two Elements of punishment, namely the reprobative and preventive, (13) and his suggestion regarding the intertwining (14) of censure and hard treatment. See also Uma Narayan, Adequate Responses and Preventive Benefits: Justifying Censure and Hard Treatment in Legal Punishment (1993) 13 OJLS 166. For discussion of both von Hirsch and Narayan, see Duff, Desert and Penance at 164, and Penal Communications at 41 44, 52n.) However, our argument must be clearly distinguished from Goldman s argument (Goldman, above n 16 at 48 49) that effective deterrence requires punishment far more severe than can be permitted on retributivist grounds, in order to compensate for low rates of detection and apprehension. Although this may be the case with some deterrents, it is not necessarily the case with all. In terms of our three categories of measures distinguished below, Goldman fails to realize that many effective deterrent measures could fall into the second category, of measures justified on combined reductivist and retributivist grounds, rather than the third category, of measures which are simply unjustified, beyond justification (see further Duff, Penal Communications at 44).

9 SUMMER 2002 Retribution, Crime Reduction & Justification 309 reasonable scale of harshness or unpleasantness (quite simply, it may be very harsh or unpleasant 27 ), but its harm reductivist value may be minimal, or at least, difficult to establish beyond some minimal level. Of course, it is notorious that there is no straightforward way of determining reductivist value, that the enterprise of trying to assess, for instance, the deterrent or rehabilitative value of a measure is fraught with difficulties. However, the reductivist value may be minimal for an obvious reason, concerning the harmfulness of the criminal conduct itself rather than any question of how, and by how much, it is to be reduced this value may be minimal simply because the conduct the measure seeks to reduce is not particularly harmful, does not rank highly on any reasonable scale of harm. To illustrate with a purposely absurd example, executing bicycle thieves may effectively deter and hence significantly reduce bicycle theft, perhaps to the point of elimination. However, even if the legitimacy of capital punishment in principle is granted for the sake of the argument, the harshness of the measure is out of all proportion to the harm of the criminal conduct to be reduced whatever some aggrieved victims of bicycle theft might feel. The loss to the individual subjected to the measure (i.e. being executed) is too great to warrant its imposition on reductivist grounds (or more precisely, crime-reduction-through-deterrence grounds), even if it can be established as highly effective as a deterrent. What is of concern is not reducing the harm caused by the criminal conduct alone, but the overall harm caused by both this and the deterrent measure. 28 In this case, there is no such overall reduction of harm. The cure is worse than the disease. 29 This is not to suggest, however, that matters could not change if retributivist justification is added to reductivist justification (although it is not claimed that the cause of would-be executioners of bicycle thieves is anything but lost). For instance, some might argue that capital punishment is justified in certain cases (typically murder 30 ) on combined reductivist and retributivist grounds. Reductivist grounds alone are insufficient for such a measure, however, according to this argument, capital punishment is justified in such cases (whatever they are) if one takes into account the element of desert. Desert has a distinct 27 We do not propose such a scale here. However, von Hirsch and Jareborg, above n 8, put forward a scale of crime seriousness, based on different living standard levels, and von Hirsch points out that, with modifications, this scale can be used to gauge the comparative severity of penalties (Censure and Sanctions at 34). 28 As Duff points out, [i]t is not enough to show that punishment prevents crime, by deterring, rehabilitating, or incapacitating actual or potential criminals: we must also show both that the costs of such prevention do not outweigh its benefits and that no other, more cost-effective, techniques of crime-prevention are available ( Penal Communications at 5). 29 Of course, although the example is being used as a case of a penal measure which is absurd on reductivist grounds, the measure is absurd on other grounds as well, distributive and retributive. Even if somehow this measure did bring about an overall reduction of harm, doing so would still be absurd as this would amount to a grossly unfair distribution of the relevant harms or risks. There is no reason why one person should suffer execution in order to save others (no matter how many) from being victims of bicycle theft. Executing a bicycle thief is also absurd in that it is simply not deserved. Even the strongest believer in capital punishment would not suggest that a bicycle thief deserves such a fate. (Indeed, a further ground of absurdity lies in the failure of the proposed measure to fit into any coherent sentencing policy or framework. Why, for instance, execute bicycle thieves but not car thieves or tricycle thieves?) 30 Or more narrowly, the murder of particular persons, for instance, police officers in the performance of their duties, or murder in particular contexts, for instance, in the course of a terrorist attack.

10 310 Oxford Journal of Legal Studies VOL. 22 justificatory force some measures, which are not justified on reductivist grounds alone, are justified if they are deserved by their recipients. 31 Desert is an inherently stronger type of justification as it is individualized to the recipient, rather than depending on considerations of the general good. As Sadurski puts it, desert considerations are always person-oriented... Furthermore, the grounds of desert must have something to do with this concrete person directly. 32 A similar point is made concerning consent and fairness or fair play arguments for an obligation to obey the law, 33 in contrast to utilitarian arguments, namely that the former offer individualized arguments for such an obligation, whereas the latter do not. Utilitarian arguments do not establish why a particular individual should obey the law if the benefits that supposedly flow from general obedience are enjoyed only, or disproportionately, by others. With consent and fairness arguments, however, there is no implication that the person who respectively has not consented, or has not benefited, is under any such obligation. Likewise, it is dubious to seek to justify imposing a particularly harsh or unpleasant penalty on someone by referring to the consequent crime-reductive benefits, where such benefits are enjoyed only, or disproportionately, by others. It is, of course, a well-known criticism of utilitarianism that, as Rawls puts it, it does not take seriously the distinction between persons. 34 Briefly, then, the claim is that there are some things which can be done to people which can be justified on harm-reductive grounds alone. 35 (Let s call these first category measures.) The candidates mentioned above are the quarantining of those reasonably suspected of carrying infectious life-threatening diseases, and the detention of the psychiatrically dangerous. 36 However, there 31 Of course, that a measure is deserved does not imply that it is, all things considered, justified. For instance, the deserved sentence may not on balance be the justified sentence because the recipient offers good prospects for rehabilitation. Or it may be better all things considered that a sentence is merciful than that it is just. Sometimes it is morally permissible and obligatory to override the dictates of justice (H.J. McCloskey, A Non-Utilitarian Approach to Punishment in Gertrude Ezorsky (ed.), Philosophical Perspectives on Punishment (State University of New York Press, New York, 1972), as quoted in Sadurski, Giving Desert its Due, above n 14 at 8). See also David Dolinko, Three Mistakes of Retributivism, 39 UCLA L Rev 1623, (1992). 32 Giving Desert its Due, above n14 at 117 (emphasis in original). Or as von Hirsch says: While deterrence explains why most people benefit from the existence of punishment, the benefit of the many is not by itself a just basis for depriving the offender of his liberty and reputation. Some other reason, then, is needed to explain the suffering inflicted on the offender that reason is desert (Doing Justice at 51). For a more historical and sociotheoretical approach to questions of criminal responsibility, see Lacey In Search of the Responsible Subject: History, Philosophy and Social Sciences in Criminal Law Theory, (2001) 64 MLR And indeed, gratitude arguments as well. See, for instance, M.B.E. Smith, Is There a Prima Facie Obligation to Obey the Law? 82 Yale Law Journal 950 (1973); J.W. Harris, Legal Philosophies, Ch. 16 (2nd edn; Butterworths, London, 1997). 34 A Theory of Justice at This should perhaps be read subject to some proviso concerning a fair distribution of the relevant harms or risks (see the distributive point about executing bicycle thieves, in n 29, and the Floud and Young point in n 36 about moral choice between risks). However, with measures such as quarantine, there is no tension between the crime reductive and fair distribution issues. Reducing harm overall also results in a fairer distribution. It is better on both counts that some lose their liberty for a limited time, rather than that others run the risk of dying from a highly infectious disease. 36 See the final paragraph of Part 1. Consider also preventive detention for dangerous offenders. See Jean Floud and Warren Young, Dangerousness and Criminal Justice (Heinemann, London, 1981), esp. Ch. 4. As they put it, [w]e are talking of two risks and the problem of making a moral choice between them: the risk that an offender may be unnecessarily deprived of his rights and his liberty; and the risk that innocent, unknown persons may suffer harm in the future (55). See also David Wood, Dangerous Offenders and the Morality of Protective Sentencing [1988] Criminal Law Review 424, reprinted in Andrew von Hirsch and Andrew Ashworth (eds),

11 SUMMER 2002 Retribution, Crime Reduction & Justification 311 are more serious, more harsh or unpleasant, second category measures, which require retributivist justification as well. In short, there are more serious things which can only be done to people if they deserve them. As has been seen, measures required for deterrence purposes or more to the point, required if sufficient deterrence value (whatever that is, however it is to be assessed) is to be achieved may well fall into this category. The justification of second category measures is two-fold: (1) first category measures offer insufficient deterrent value, or more generally, reductivist value (even where other reductivist means, such as incapacitation, rehabilitation and denunciation are relevant, and their reductivist value, indeed, non-retributivist value generally, 37 is taken into account); (2) even if measures in the second category cannot be justified on reductivist grounds alone (because they are too harsh or unpleasant in relation to their reductivist value), they can be justified if retributivist considerations are taken into account as well. ((1) justifies second category measures as against first category measures. And (2) justifies them as against, as will be seen, third category measures, which are beyond justification.) Imprisonment, understood in the sense of penal incarceration, is an example of a second category measure. If justified at all, imprisonment cannot be justified on reductivist grounds alone, as say, quarantine, or a system of civil incarceration, at least in principle can be justified. Imprisonment offers insufficient reductivist value if attention is limited to incapacitation and rehabilitation. Concerning the former, imprisonment offers insignificant crime-reductive effect, because of the epistemic difficulties in determining who to incapacitate and for how long, resulting in an unacceptably high rate of false positives. 38 As for rehabilitation, apart from some specialized programmes directed at specific offender subgroups, which seem to have had some success, imprisonment has if anything a dehabilitative effect. 39 One has to turn to deterrence, then, to find any significant reductivist value. 40 However, with deterrence there is not just the epistemic problem of determining what deters? a problem which is still significant, if not of the same order as the parallel problem of what rehabilitates? and in particular, the problem of making comparative marginal judgments between Principled Sentencing (Edinburgh University Press, Edinburgh, 1992) at 129; and in Jill Peay (ed.), Criminal Justice and the Mentally Disordered (Ashgate, Aldershot, 1998) at 419; and Frederick D. Schoeman, On Incapacitating the Dangerous in Hyman Gross and Andrew von Hirsch (eds), Sentencing (Oxford University Press, New York, 1981) at Consider, for instance, rehabilitation not as a reductivist means, but as an end in itself. 38 See, for instance, Floud and Young at Ch. 2, and the readings at von Hirsch and Ashworth (eds), Principled Sentencing (2nd edn) at Ch See, for instance, Sue Rex, A New Form of Rehabilitation, in von Hirsch and Ashworth, above, at 34, and the readings at Ch. 1 generally. For questioning of the supposed corrupting effects of imprisonment, see Mirko Bagaric Incapacitation, Deterrence and Rehabilitation: Flawed Ideals or Appropriate Sentencing Goals?, (2000) 24 Criminal Law Journal 21, See Bagaric, above n 39 esp This is, it could be objected, hardly a comprehensive survey of reductivist means. We have said nothing about moral education as such a means, or about how punishment may have a reductivist effect through its expressive, symbolic, denunciatory, or communicative value or meaning. Again, we are concentrating on deterrence as the reductivist means which most obviously requires harsh or unpleasant measures.