The Mixed Theory of Punishment
For a brief period in the middle of the twentieth century, it was widely believed that the problem of punishment had finally been solved. The purported solution involved taking elements from both the utilitarian and the retributive theories and creating a “mixed” or hybrid theory. From utilitarianism came the idea that the ultimate goal of punishment was to prevent crime; from retributivism came the idea that punishment must be in response to a prior wrongdoing and be proportionate to that wrongdoing. Hence we could have the best of both worlds; a rational explanation for punishment as well as moral constraints on its use. It however soon became clear that this solution was untenable and even incoherent. It requires arbitrarily separating the two theories without any rational basis; retributivists will object to the idea that the purpose of punishment is utilitarian; utilitarians will object to the arbitrary inclusion of constraints on the maximization of utility. The solution is utterly ad hoc and hence unacceptable.
KeywordsCriminal Justice System Conceptual Analysis Moral Theory Legal Formalism Moral Justification
One of the more tantalizing and frustrating aspects of the problem of punishment is that, while retributivism and utilitarianism each seem deeply problematic on their own, the two theories appear to complement each other perfectly. The utilitarian theory provides us with what retribution seems unable to, a positive and plausible reason to punish people: the prevention of crime. The retributive theory in turn gives us what the utilitarian theory cannot, a set of constraints limiting punishment to the guilty, and to a proportionate amount (one cannot punish the innocent because they do not deserve it, or punish the guilty excessively because they do not deserve such harsh treatment). If we could only find a way to combine these two theories, we would have the best of both worlds. The problem is that the two theories are incompatible and even mutually contradictory. The retributive theory, for example, tells us that one’s motive for punishment may not be future social benefits; that would be to use people as a means to a social good, which violates the Kantian principle of respect for persons. In turn, the utilitarian theory tells us it is irrational to limit the utilitarian principle in advance by ruling out punishing excessively or punishing the innocent; it also holds that retribution as a goal distinct from any resulting social benefits is irrational. Since each theory declares the other immoral and untenable, it seems incoherent to try to combine them.
In the mid-twentieth century, it was widely believed that the problem had finally been solved. In a burst of creativity, a number of different thinkers – most famously H.L.A. Hart and John Rawls – developed an approach that purported finally to reconcile utilitarianism and retribution. The solution, which came to be called the Mixed (or Hybrid) Theory of punishment, was simple and elegant. The mistake of all previous theories, it claimed, was to treat the justification of punishment as if it were a single problem, a mistake which Rawls thought had “blinded” moral philosophers in their treatment of the punishment debate (1955, 30). In fact, according to the Mixed Theory, the justification of punishment must be separated into two distinct “questions”: justification of the institution as a whole, and justification of the individual acts of punishment. The answer to the first question is provided by the utilitarian theory, explaining why we have the institution in the first place. The answer to the second question is provided by the retributive theory, telling us why punishment may be administered only to those who have violated the laws. Thus by clarifying the concept of punishment, one could see that the two distinct moral theories belong in their separate spheres, and thus need not overlap or contradict each other. Was this not the answer we had long been looking for?
Alas, it was not to be. Despite the assurance with which Mixed Theorists assured us that the problem of punishment was “easily solved,”1 it is now evident that the triumphant claims to have solved the problem of punishment were unwarranted. Nonetheless, the Mixed Theory remains influential. A series of reference works including The Encyclopedia of Philosophy, the online Stanford Encyclopedia of Philosophy, and Antony Flew’s Dictionary of Philosophy continue to assert the Two-Question strategy as an unquestioned starting point in the debate.2 Moreover, explicit criticism of the Mixed Theory has been surprisingly restrained.3 The purpose of this essay is to set out briefly and clearly why the Mixed Theory failed, and what it can teach us about the difficulty of combining utilitarian and deontological values in a single coherent system.
This concern is especially important because in practice, the institution of punishment is still governed by what can be called a “pluralist” approach, which incorporates both utilitarian and retributive concerns. Politicians, lawmakers, and judges continue to assert motives of deterrence, rehabilitation, incapacitation, and retribution as among the goals of punishment. It is crucial, however, to distinguish as Hart does between this pre-reflective pluralism, which does not attempt to resolve (or, frequently, even acknowledge) the inherent contradictions between the stated goals, and the Mixed Theory which purports to give a principled basis for how to combine utilitarian and retributive goals (Hart 1968, 3). At present, we are simply getting used to living with the contradiction, but this is an unsatisfactory position. Although this chapter offers a deeply critical perspective on the theory, it is also important to recognize what is to be praised in the Mixed Theory: an honest recognition of the deep incompatibility between the utilitarian and retributive ideas, and the need to find a way to reconcile their contradictions. This is an insight which is all too often forgotten in contemporary “pluralist” or “pragmatic” theories of punishment.
4.1 The Idea of “Separate Questions”
What defines the Mixed Theory above all is its fundamental strategy of distinguishing two separate “questions,” to which utilitarianism and retribution are the respective answers. Actually, proponents of the Mixed Theory disagree on the central terminology; Hart distinguishes separate “questions” but also “issues,” “aspects,” and “features” of punishment (1968, 3, 4, 11); Mabbott divides punishment into two separate “issues” (1972, 165); Scheid sees two distinct “roles” in the “package” of punishment (1997, 505); Benn refers to separate “levels” of justification (1967, 30), Rawls distinguishes between “offices” (1955, 7), and W.D. Ross (who may be the originator of the Two-Level Strategy) distinguishes two “stages” in the process of punishment (1930, 61) (a term which survives in Hart (1968, 7, 8)). It is hard to see what, if anything, all of these terms have in common; but for purposes of argument we will limit our discussion to the most common idea that punishment is composed of separate “questions.”
The Mixed Theory is not merely pointing out the obvious fact that there is an enormous variety of questions one might ask about punishment, but making a much stronger claim that the field of punishment is somehow defined and structured by two key types of question. These questions are said to be “logically distinct” in some strong though obscure sense. Quinton, for example, cites Ross’s distinction between the creating of rules and their application (Quinton 1967, 14), but redefines it as a distinction between logical “modalities” (id., 7); he uses the word “logical” 15 times in this short essay. John Rawls, who also explicitly draws on Ross’s account, also insists on the “logical force” of the distinction between the two questions (1955, 32); Rawls outdoes Quinton by using the word “logical” 16 times. In Hugo Bedau’s recent summary of the idea, the distinction between different questions is said to be “conceptual” and “analytic” (2010, 3).
But what justifies this move from an (arguably) useful “expositional device” of separating questions (as Nicola Lacey puts it (1994, 187)), to the far stronger claim of a distinction between logical modalities? The most familiar example of distinct logical modalities is the distinction between normative and factual inquiry, but this is not the distinction invoked by the Mixed Theorists. However, there is little agreement on what they do mean by distinct modalities. For Quinton the distinction is between logical and moral questions, for Rawls it is between the justification of practices versus the justification of particular cases (purportedly corresponding to a distinction between legislative and judicial questions); for Hart there is a distinction between definitional, justificational, and distributive questions; for Lesnoff “teleological justification” versus “entitling justification” (1971, 144); for Dolinko between questions of “rational justification” versus moral justification (1991); and for Hugo Bedau between definition and justification, and also between justification of acts versus justification of the practice as a whole (2010).
It thus seems far from obvious that these “questions” divide up the field of punishment into any obvious, conceptual set of distinctions. But it will be necessary to consider the Mixed Theory in more detail to assess the claim. To do this, we will look at five different approaches to the Mixed Theory: the Conceptual, the Formalist, the Separation of Powers, the Rule-Utilitarian versions, and H.L.A. Hart’s sui generis account. It must be conceded that these distinctions are somewhat artificial, for the different versions overlap, and some theorists adopt more than one (Rawls for instance endorses the rule-utilitarian, the conceptual, and the separation of powers versions). The Mixed Theory encompasses a surprising array of different strategies, not all of them mutually consistent, and we will see how the very variety of strategies reveals the fundamental incoherence of the theory.
4.2 The Conceptual Version of the Mixed Theory
The conceptual version of the Mixed Theory is grounded in the claim that an analysis of the concept of punishment reveals a logical distinction between two different questions. The idea of “conceptual analysis” as a philosophical tool is extremely controversial. Though some sort of analysis of concepts or ideas has always been central to philosophy (see e.g. Plato’s dialogues on justice, virtue, love, etc.) and no doubt will always continue to be so, the so-called “linguistic turn” of the early twentieth century raised the status of the clarification of concepts to an “unprecedented height” (Harris 2002, 2). Conceptual analysis became in the early twentieth century the unique and defining methodology of philosophy, as distinct from the observational method of the empirical sciences. Confidence in the power of this technique raised hopes for the resolution of longstanding philosophical problems (the Mixed Theory of punishment is an example). However, the high status and optimism regarding conceptual analysis did not last, and it is now seen, as Frank Jackson describes, as a “mysterious activity discredited by Quine that seeks the a priori in some hard-to-understand sense” (1998, 7).4 In the contemporary view, Quine’s seminal article “Two Dogmas of Empiricism” refuted conceptual analysis, at least in its pretension to provide access to a priori truth, by undermining the distinction between analytic and synthetic knowledge (though some philosophers, including Frank Jackson, have defended a more modest role for conceptual analysis (Jackson 1998)).
We need not enter into this broader debate about the legitimacy and limits of conceptual analysis, except to note the central role of this method in the formulation of the Mixed Theory of punishment. Leo Zaibert has claimed that the various versions of the Mixed Theory are “creatures of the twentieth century – and mostly creatures conceived by analytic philosophers inebriated with Oxford’s ordinary language philosophy and its concomitant fascination with logic.”5 For the method of the Mixed Theory in regards to the problem of punishment is to bypass the substantive moral debate about the merits of the utilitarian and retributive theories of punishment in favor of an analysis of the logic of the concept of punishment. Such an analysis, it is claimed, could reveal the two logically distinct “questions” involved in punishment, and also the appropriate answers to those questions.
The clearest example of such an approach is found in A.M. Quinton’s influential article “On Punishment” (1972). Quinton argues that the “antinomy” between retribution and utilitarianism can be resolved by recognizing that “retributivism, properly understood, is not a moral but a logical doctrine” (id., 6). That is to say, there is no conflict of moral theories, since while utilitarianism provides us with a moral basis for punishment, retributivism is not a moral doctrine at all, but merely an “elucidation of the use of the word” punishment (id.). Thus even if utilitarianism dictated that we should punish the innocent, Quinton holds that the very idea of punishment logically entails negative retributivism, and hence rules out punishing the innocent. The apparent antinomy, Quinton asserts, arises from a “confusion of modalities,” i.e. logic versus morality. Utilitarianism answers the question when morally ought we to punish; retributivism tells us when as a matter of logic can we punish (id., 7).
Quinton is by no means the only Mixed Theorist to utilize conceptual analysis; elements of conceptual analysis are present in the background of almost all mixed theories. Rawls for example says “It is part of the concept of the criminal law as a system of rules that the application and enforcement of these rules in particular cases should be justifiable by arguments of a retributive character” (1955, 6) while Hugo Bedau insists that “Conceptually, punishment is retributive” (1974, 141). It is important to recognize the essential strategy this approach employs. All of the unpleasant implications of utilitarianism are avoided by the discovery of an intrinsic logical constraint against punishing the innocent or punishing the guilty disproportionately. Nor do we need a moral justification for the retributive idea, as it is found in the very meaning of the word “punishment.” And thus utilitarianism and retribution can be harmoniously combined.
It will, I think, suffice merely to state this argument in order to see its implausibility. It is unlikely that the very logic of the word punishment is retributive; a utilitarian who supports severe punishment to deter crime is not making a “logical” error, or simply misusing the word “punishment.” And even if Quinton were correct that the meaning of “punishment” entails guilt, a utilitarian could evade his objection by declaring he would use the word differently, or by adopting a new concept (say, “telishment”) which did not logically prohibit punishing the innocent.6 Quinton’s theory relies on what Hart labeled the “definitional stop”: attempting to solve moral issues by definitional fiat. Retributivists will also balk at the notion that their theory has no moral content; for Kant the constraint against punishing the innocent is a deep moral principle: punishment “can never be inflicted merely as a means to promote some other good” (1991, 140 (Ak. 331)). Even Quinton cannot avoid hesitantly recognizing that punishing the innocent is more than a mere mistake of logic, but a moral wrong. Guilt, he says, “is a logically necessary condition of punishment, and, with some exceptions, it might be held, a morally necessary condition of the infliction of punishment” (1972, 12).7 It is implausible that the entire debate over the justification of punishment has been based on a mere misunderstanding of the meaning of the word “punishment.” While it is certainly true that the word “punishment” is ordinarily used to mean suffering imposed for a wrongful act, this fact alone cannot resolve the moral debate. Indeed, quite the opposite: it demonstrates how a moral theory is built into our concepts, and that moral theory cannot escape philosophical assessment merely by referring to ordinary usage.
4.3 Legal Formalism
A second version of the Mixed Theory, of which there are suggestions in Quinton’s essay, is one we might characterize as legal formalism. Legal formalism, in this context, is the idea that a set of penal rules, once created, carry within them the terms of their own application, so that administering punishment does not require further moral justification. Thus the decision whether to institute a rule-based system of punishment is one question, to be determined by utilitarian values such as social protection, while the question of whether to punish one who breaks rules is wholly determined by the rules. In particular, the point is that the imposition of punishment in a particular case must not be influenced by utilitarian (or any moral) considerations. As Quinton states, “Utility is irrelevant to the application of rules, for once the rules have been set up, punishment is determined by guilt; once they [sic] are seen to apply, the rule makes a sentence of punishment necessarily follow” (1972, 15). Among those who have formalist leanings we might include Rawls, Ross, and also Bedau (2010, 3, 8), but especially Mabbott. For Mabbott, “one may consider the merits of a legal system or of a credit system, but the acceptance of either involves the surrender of utilitarian considerations in particular cases as they arise” (1972, 176); the choice of a “legal system as one of our instruments for the establishment of the conditions of the good life…is logically prior to and independent of the actual punishment of any particular persons or the passing of any particular laws” (id., 177). Punishment is not a moral choice but simply a “corollary of lawbreaking” (p. 174), nor do legislatures choose to punish, it is the criminal who brings it on himself by choosing to break the rules (id., 175). For John Rawls, the “practice” of punishment is defined by rules, and is adopted on “utilitarian” grounds. But once one has adopted the practice, this “necessarily involves the abdication of full liberty to act on utilitarian and prudential grounds” in individual cases, for it is essential to the idea of a practice that the rules are treated as “definitive” (1955, 24). W.D. Ross endorses the formalist conception as well, though on grounds that a law is a “promise to the injured persons and his friends, and to society” that the punishment will be carried out (1930, 63), and hence there is no choice as to whether to punish or not in individual cases. Hugo Bedau insists that justification of the practice of punishment is based on social purposes, whereas justification of any individual act of punishment is “to be done by reference to the norms (rules, standards, principles) defining the institutional practice” (2010, 3). For the formalists, retribution for wrong is an automatic result of the decision to adopt a rule-based system of punishment.
Whatever the validity of formalism as a legal doctrine, it is not a plausible account of retributive punishment,8 for it rests ultimately on the dubious premise that punishment is logically connected to the adoption of rules of criminal law. But there is no logical requirement that criminal laws impose a punishment for their breach (to say that “penal laws” must include a penalty would not be conceptual analysis, but a circular argument).9 The commandment “Thou shalt not kill” is no less a legal prohibition in that it does not prescribe a penalty. Whatever the connection between rules and penalties is, it is not a “logical” one. It may be necessary to impose a penalty in order to secure obedience to the rule, but this is not a logical necessity. While it is true that criminal laws apparently universally do provide a penalty for their violation, this fact does not provide evidence for a logical or conceptual connection between the two. In fact, we often do not punish lawbreakers – prosecutors decide not to prosecute, or accept plea bargains, juries acquit by nullification, judges suspend sentences, governors pardon criminals. It is a moral choice whether to have a penalty, what level to set it, and whether to apply it in particular cases. Further, formalism cannot resolve the contradiction between utilitarian and retributive goals, for several reasons. For one thing, it cannot logically preclude a rule that requires punishing the innocent. For another, there is no reason to deny the relevance of utilitarian concerns at the level of sentencing; a judge or parole board might be directed to take into consideration the likelihood of the person committing more crimes. Retributivism cannot be reduced to the idea of rule-following, for it is a moral claim, not a claim about the logic of rules or the concept of a criminal law.10 People who deeply believe that criminals ought to be punished do not base this claim on their knowledge of logic.
4.4 The Separation of Powers Principle
One distinguishes two offices, that of the judge and that of the legislator, and one distinguishes their different stations with respect to the system of rules which makes up the law; and then one notes their different stations with respect to the system of rules which make up the law; and then one notes that the different sorts of considerations which would usually be offered as reasons for what is done under the cover of these offices can be paired off with the competing justifications of punishments. One reconciles the two views by the time-honored device of making them apply to different situations (1955, 7).12
However, it is unlikely that the principle of separation of powers can provide establish the logical distinction between utilitarian and retributive aspects of punishment. For one thing, the analogy does not fit. Rawls and others use the distinction between the legislative and judicial role to support a two-level theory, but overlook the fact that the separation of powers doctrine divides government into three branches: legislative, judicial, and executive, and would therefore suggest a three-level theory not two. Indeed, the executive plays a substantial role in administering punishment: policing the laws, arresting wrongdoers, prosecuting them, supervising parole and probation, etc. The Mixed Theory would have to determine whether the executive branch is governed by utilitarian, retributive, or both principles. In short, the utilitarian/retributive dichotomy does not line up with the trichotomy of branches of government.13
Indeed, the sentencing of criminals has long been recognized as one area that does not divide neatly into the separation of powers, as all three branches have important and overlapping roles to play in the administration of punishment: the United States Supreme Court explained in Mistretta that “the sentencing function long has been a peculiarly shared responsibility among the Branches of Government and has never been thought of as the exclusive constitutional province of any one Branch.”14 Sentencing is divided among the legislature, which may choose to give wide discretion or narrow limits for the judge; the executive, which has substantial discretion to accept plea bargains, or to drop charges, or to grant parole; and the judiciary, which at least traditionally has been given wide discretion in sentencing. Nor is there any good reason to think that concern for consequences is limited to any one branch, or that the legislature cannot be concerned with retribution. In any case, the separation of powers doctrine provides at most a pragmatic, prudential basis for separating distinct goals, and thus cannot support any “necessary” or conceptual distinction between modes of justification (Bedau 2010, 3; Rawls 1955, 19 and passim). The separation of powers doctrine, in short, provides no good basis on which to build a two-level theory of punishment.
4.5 The Rule-Utilitarian Theory
The most logically coherent but also the least convincing of all the mixed theories is the rule-utilitarian version. Rawls’ famous essay, despite its repeated insistence on conceptual or logical distinctions, in the end endorses a rule-utilitarian theory, as do Benn’s and Mabbott’s essays.15 The rule-utilitarian theory is grounded in the idea that, human nature being what it is, the best way to maximize good consequences in the long run is to require people to follow rules that in themselves tend to maximize utility overall. Thus, rather than have people decide in each case whether telling the truth or lying would maximize utility, it is better to have them take truth-telling as a universal rule. This is so because first, people do not have the time or ability to calculate the consequences of every single action, and second, because if we allowed them to do so their biases would inevitably creep in, and people would rationalize lying when it serves their interests. The same reasoning can apply to punishment: we are better off having judges simply apply the rules as given and not try to determine in each case whether punishing this or that defendant would maximize social utility. Thus, on this view, the system of criminal law is justified on utilitarian grounds (crime prevention), but individual sentences are decided on “retributive” grounds (i.e. punishment must be for a violation of the laws) irrespective of social utility concerns.16
It is important to distinguish the rule-utilitarian principle from legal formalism. The latter implies that it is a logical property of rules that they must be followed without exception. Rule-utilitarianism however is not a logical doctrine but a pragmatic one: if humans were perfect calculators and could act disinterestedly, then we would not have any reason for rules at all, or at least we would be free to break rules whenever it was socially useful. This point is important, for it undercuts any claim to a logical or conceptual distinction between separate “questions”: the reliance on rules in rule-utilitarianism is a contingently-useful method, not a necessary logical process. An act-utilitarian would reject the idea of separate rules as contradicting the goal of maximizing utility.
In any case, the fundamental objection to this version of the Mixed Theory is that, as we argued previously, rule-utilitarianism is not a plausible moral theory, and has few adherents nowadays.17 Its promise of finally reconciling deontology and utilitarianism is now seen as unfulfilled, for it has proven to satisfy neither side. Utilitarians accuse it of irrational rule-worship; if breaking a rule will maximize utility, then one should do so.18 Deontologists reject the very idea that fundamental moral rules such as keeping promises, telling the truth, and not punishing the innocent could be based on considerations of utility, as opposed to the principle of respecting human autonomy. Rule-utilitarianism is moreover not a “mixed theory” at all, but a whole-hearted acceptance of the utilitarian moral philosophy, with the relegation of retributivism to the mere pragmatic value of following rules. It is surely no accident that the rise of the Mixed Theory of punishment coincided with a period of great enthusiasm for the idea of rule-utilitarianism as providing a Grand Unified Theory of ethics. It is also relevant that Rawls himself came to abandon both act- and rule-utilitarianism in favor of a Kantian, contractualist moral theory. Indeed, the persistent influence of the Mixed Theory today seems to have become separated from the rule-utilitarianism version, and become far more based on H.L.A. Hart’s version of the theory, to which we now turn.
4.6 H.L.A. Hart’s Two-Level Theory
Hart’s account is the most sophisticated and also influential of all the varieties of the Mixed Theory, though no more satisfactory. To his credit, Hart saw his account only as a “prolegomenon” to a theory of punishment: “Till we have developed this sense of the complexity of punishment (and this prolegomenon aims only to do this), we shall be in no fit state to assess” the institution (1968, 3). However, no such further development of this theory has gained widespread acceptance, and so we will focus on Hart’s initial account. His reference to the “complexity of punishment” is meant to reject the idea that punishment is a single or simple institution requiring a single moral justification. Hart thus points us to the basic strategy of the Mixed Theory: that punishment must be analyzed into multiple different “questions” or “issues” or “stages” involved in the institution, each of which requires “separate consideration” (id.).
However many “questions” there may be for Hart, his central concern in the Prolegomenon is with just two of them: the “General Justifying Aim” of an institution and its “Distribution” (id., 4).19 This distinction has become remarkably influential despite its obscurity; Hart presents the terms (and even capitalizes them) as if they were already obvious and in need of little explication or defense. But it is far from obvious what Hart means by these distinctions; indeed the very awkwardness of the neologism “General Justifying Aim” raises serious doubts about its validity (as does the fact that nobody before Hart ever recognized the existence of such a thing). Just what is a “General Justifying Aim”? In what sense is it “general” (is it to contrast with a specific aim?)? Must there be only a single general justifying aim? And in what sense is the general aim “justifying?” Does he mean morally justifying, or justifying in some other sense?20 Does he mean the putative justification or the actual? If the latter, what is it precisely that justifies it morally? Can there be a general aim that is not justifying?
It is even harder to see why the “General Justifying Aim” is contrasted with a principle of “distribution.” Indeed, the term distribution seems quite inapt, both in respect to property law (in what sense are rules of title and amount “distributional” principles? (id., 4)) and to punishment, where Hart coins the even more awkward phrase “retribution in Distribution” (id., 9). The decision regarding whom to punish and how much is hardly a question of “distribution” (as if there were a certain fixed amount of suffering that had to be distributed amongst the populace).21 Hart tells us that the distinction can be conceived more generally as one between the “general aim” of an institution and the “principles limiting” that aim, i.e. the principles that constrain the “unqualified pursuit of that aim or value” (id., 10). Applied to the institution of punishment, Hart’s conclusion is that the most plausible account will hold that the general justifying aim of criminal law is social protection, while retribution serves as the limiting principle, ensuring that we do not go so far as to punish the innocent or punish the guilty disproportionately in pursuit of the social good.
It is puzzling to call the General Aim “justifying,” if it is clearly not justified without the limits imposed by the Distributive principles. But this is presumably Hart’s point: that there are two separate justificational questions, one for the system as a whole and one for the individual sentences. However, the challenge for Hart (and for the Mixed Theory) is to provide a compelling reason for insisting on such a counterintuitive bifurcation of the moral justification of punishment. Hart’s strategy is to prove to us that we must distinguish between these two levels by demonstrating that the principle of “Distribution” cannot be deduced as a “consequence” from the General Justifying Aim, no matter whether that Aim be utilitarian or retributive (id., 11, 25). This argument is critical, for it bears the burden of showing why we should accept the bifurcated notion of justification of punishment (or apparently of any institution). The principle of “retribution in Distribution,” Hart tells us, demands that we limit punishment to those who have broken the law; this is the idea usually called “negative retribution,” to distinguish it from the positive idea that we should punish the guilty (Hart calls this the Retributive “General Aim.”). Since the principle of Distribution of punishment does not derive from the Aim, he argues, the two must be understood as distinct questions. Let us consider this argument carefully.
The first part of the claim asserts that negative retribution does not follow when Utilitarianism is taken as the General Justifying Aim of punishment. This result should come as no surprise: Hart is merely reaffirming the idea that utilitarianism account for our intuitions regarding punishing the innocent. But this observation provides no support for the claimed two-level distinction; all it shows is that neither act- nor rule-utilitarianism can provide a plausible theory of punishment. Hence the entire case for Hart’s theory seems to rest on Hart’s surprising claim that even if retribution were the General Aim of punishment, it would not follow that one may not punish the innocent (or punish the guilty excessively). That is to say, Hart claims that the retributivist needs separate arguments for negative and positive retributivism; one who believes that the goal of punishment is to punish the guilty needs a further argument for not punishing the innocent or not punishing the guilty disproportionately. This is a remarkable claim, and one that virtually no retributivist would accept.22 To the contrary, the retributivist principle, that punishment should be based on desert, would seem to entail both that the guilty should be punished and that it is wrong to punish the innocent or to punish the guilty disproportionately (since their punishment would not be deserved). As C.L. Ten explains: “The retributivist confines punishment to those who deserve it. This obviously rules out the punishment of those who have not committed any offense” (1987, 95). Hart thus owes us a compelling argument for his claimed distinction.
Hart’s central argument for the purported logical distinction between positive and negative retributivism (punishing the guilty versus not punishing the innocent) is this: “we attach importance to the restrictive principle that only offenders may be punished, even where breach of this law might not be thought immoral. Indeed, even where the laws themselves are hideously immoral as in Nazi Germany…the absence of the principle restricting punishment to the offender would be a further special iniquity” (1968, 12; cf. also 18). Hart’s point here is that if a law is unjust, we should think it a wrong to punish anyone under it, regardless of whether they had violated it or not. But why then should it seem a further wrong to punish the innocent? For those “guilty” of violating the law no more deserve punishment than the innocent. Thus the principle of negative retributivism, it seems, must be based on some further principle of justice, not on the retributive principle itself.
Hart is no doubt correct that punishing the innocent under an unjust law strikes us as an additional wrong that does not seem to be accounted for by retributivism. However, this intuition does not entail a distinction between positive and negative retributivism. What Hart’s example demonstrates is the importance of a set of values known as the “rule of law,” demanding that rulers satisfy demands of legality such as fair notice, capacity to obey, and so forth.23 Thus Hart could equally have pointed out that a person guilty of violating a just law would still not be punished justly if the law was kept secret, or applied ex post facto, etc. But as important as the rule of law is, it is simply irrelevant to the question of the relation between positive and negative retributivism. Indeed, the rule of law and retributivism are often at odds (as when a guilty person deserving punishment gets off on a legal “technicality”). Hart’s error is to assume that retributivism is the only principle governing the punishment of individuals. But the retributivist has no trouble accepting that the rule of law provides an additional set of norms governing punishment, involving procedural justice in addition to substantive justice. Hence even if the law is unjust, convicting the innocent under it commits a further procedural wrong on top of the substantive wrong. To claim that this demonstrates a divide between positive and negative retributivism is, to use Hart’s phrase, a “spectacular non sequitur.”24
Thus Hart fails to prove there is a fundamental distinction within the institution of punishment between two different “questions,” the general aim and the limiting principle. In fact, it is worth remarking on just how little Hart tells us about the nature of this distinction. What sort of distinction is it: Logical? Conceptual? Sociological? Pragmatic? Moral? Is it the only such distinction, or merely one of many? How strong a distinction is it? It is clearly not absolute, since Hart believes it can be overridden in some circumstances (see 1968, 12). To be sure, it is trivially true that in any human institution there will be aims or goals and the recognition of limits. But to reify this distinction into a fundamental defining principle of justification would require powerful arguments; Hart has not met the substantial burden of proof for his extraordinary claim.25
Indeed, the distinction violates common sense; preventing crime and carrying out justice (by punishing the guilty and protecting the innocent) are both part of the General Aim of the criminal justice system, not answers to different questions. It is highly doubtful that we can even make a clear distinction as regards most parts of the criminal justice system: is the criminal trial governed by the general aim (to uncover the truth, and ensure conformity to law), or by the principle of distribution/limits (to ensure that only the guilty are punished)? David Dolinko calls it “preposterous” to separate the question of who may be punished from both the definition and justification of the institution, and notes that Hart himself cannot consistently maintain it, building into his definition of punishment the idea that it must be for an offense.26 Lacey points out that it is inaccurate to separate the utilitarian goal from the distributive principle, as utilitarianism already contains within it a principle of distribution (i.e., maximize utility overall; the problem is that it is a faulty principle) (1994, 52). Equally, for retributivists the “general aim” and “distribution” are both based on the Desert Principle. Perhaps all theorists, lawmakers, and judges are simply deeply confused, but the more charitable explanation is that the distinction Hart proposes simply does not exist.
In the end, Hart’s claim cannot be that a General Justifying Aim can never entail a principle of distribution. Both utilitarianism and retributivism provide counterexamples; each claim to provide us with both a positive goal and limiting principle, and Hart has not demonstrated that either is clearly wrong. The only logical gap Hart has demonstrated, then, is between a utilitarian General Justifying Aim and a retributive Distribution, or between a retributive General Justifying Aim and a utilitarian principle of Distribution. In other words, the logical gap is not between some purported “General Justifying Aim” and a principle of “Distribution,” but between utilitarianism and retribution. That is to say, Hart has not solved the problem of punishment, but merely restated it.
4.7 The “Practice Conception” Argument
There is one more argument in support of the Mixed Theory to be addressed, however briefly. It is perhaps implicit in all or most versions of the theory, though it is made explicit by Rawls and Bedau. Rawls calls it the “practice conception” (1955, 24). It holds that individual acts of punishment (particular sentences) can exist only within a system of punishment governed by rules; the practice of punishment is logically independent of the particular cases.27 Rawls explains that the rules of practices must be “logically prior to particular cases. This is so because there cannot be a particular case of an action falling under a rule of a practice unless there is the practice” (id., 11). Hugo Bedau defends the distinction on the grounds that “it is possible to have a practice of punishment” without there being “any occasion to inflict its threatened punishment on anyone” (2010, 3). Further, Bedau notes, the practice of punishment might be justified even though a “given act of punishment” is not (id.). There is therefore a “logical” distinction between the institution of punishment and the application to particular individuals, and this opens up a logical space for segregating the different goals of utilitarianism and retribution. On this theory, the emphasis is no longer on the conceptual basis for punishment, but on social practice as a source of “logical” distinctions.
Assessment of this position is difficult given the ambiguous terminology adopted by its proponents.28 Furthermore, the argument raises enormously complex and controversial issues in epistemology and ontology. The central idea of the argument, as developed by John Searle, is that “institutional facts”, such as the fact that a person is punished, have a very different ontology from ordinary physical facts, for institutional facts gain their reality only from the prior existence of rule-governed social institutions that constitute them as facts.29 For example, “second base” has no independent reality apart from the social institution of baseball. Similarly, locking a person in a prison counts as an act of punishment only when it is defined as such by the social institution of punishing (indeed, the building only is a “prison” at all in virtue of this social institution). This is in contrast to natural (or “brute”) facts such as the fact that water consists of hydrogen plus oxygen; this claim is true independent of the existence of any social institutions. Institutional facts thus it seems have a distinct ontology from ordinary physical facts.
For Rawls and Bedau, this observation has tremendous significance for the debate on punishment. An individual application of punishment is logically dependent on the existence of the institution of punishment. It follows, they claim, that the justification of the institutional fact will have a very different form from the justification of the institution itself. When one asks whether the institution of punishment as a whole is morally justified, one is looking for a very different answer from when one asks whether an individual sentence is justified, for in the latter case the issue is whether the sentence is justified relative to the institution itself. That is, is it a fair application of the rules as defined by the institution of punishment? In contrast, in order to justify the institution, one needs to ask a very different sort of question: what is the purpose of the institution, what are its goals, etc.
That a certain sound or shape acquires the status of a word of symbol does not require any institutional authority. Indeed language could not have evolved if it were necessary to have a set of institutional facts before language ever got going.30
One might equally wonder whether punishment requires a social institution to get going. It would seem that the parental punishment of children is closer to a “natural” form of punishment not requiring any rule-governed social institution to constitute it as punishment. Indeed, some ethologists have claimed to find punishment occurring among social animals, including both primates and canids.31 If so, this casts doubt on the relevance of the “practice conception” to punishment, or even on the claimed dichotomy between institutional and natural facts. And most fundamentally, the theory seems to espouse a form of cultural relativism that evades the problem of moral justification of punishment; surely there are grounds for questioning the morality of a given set of practices of a community, and such challenges cannot be estopped by claims that the internal logic of the practice precludes any such questioning. We cannot pursue this very difficult and controversial question here, but can note that the prospects for solving the problem of punishment are dim indeed, if it requires a prior resolution of the problem of the ontological status of institutional facts.
Indeed, even if the Practice Conception Argument were correct, it is far from clear that it would resolve the problem of the moral status of punishment. For the ill-defined notion of logical priority does not obviously entail a two-level theory of punishment. Rawls claims that the practice is governed by utilitarian considerations, while “particular cases” are governed by “retributive” concerns (1955, 7). But he gives us no reason why particular cases cannot be governed by utilitarian considerations. If the rules direct the judges to take utilitarian concerns about future consequences into account (as they often do) in sentencing, then the applications are utilitarian in nature, not retributive. Nor does he tell us why the raison d’être of the criminal justice system as a whole cannot be to give wrongdoers what they deserve, as many people think it is. Indeed, the argument only makes sense if, implausibly, we take “retributive” to mean nothing more than “following the rules as written.” Flew concedes that this is retribution in a “rather thin sense” (2002, 294), but in fact it is not retribution at all but rather a version of formalism. And if the rules permit the judges discretion in sentencing, as they often do, then it is not even “retributivism” in this implausibly narrow sense. The reductio ad absurdum of this view is revealed in Bedau’s claim that, since the entire notion of “desert” reduces simply to a punishment “being the product of a system of law,” it follows that “the system of punishment that emerges under this argument leaves punishment in any actual individual case something of a ritual – in some cases an empty ritual – and in any case a highly formalized act…” (2010, 9). It is perhaps needless to point out the implausible and disturbing view that the sentencing of individual criminals should be viewed as a mere ritual.32
Nor, even if we agreed there were two separate “questions,” would it follow that we need a distinct sort of justification for each of them. Hart tells us that once we have decided on the general justification for the institution of punishment, we still need to ask “what more is required” in order to justify its application to an individual case (1968, 82). The proper answer, contra Hart, is nothing. If the rule is justified, and a person violates it, then it is unnecessary to require a further justification (let alone one of an entirely different logical type) for applying it. Consider in comparison the example of the baseball game: if the rules declare that a runner will be declared out when he runs out of the basepath, then an umpire does not need a further justification for calling the runner out if he leaves the basepath. To be sure, a failure on the part of officials to follow the rules provides independent grounds for complaint. But we discussed this issue above in relation to Hart’s essay; all it shows is that there is an additional value to having clear, regularly applied rules. It does not prove that the moral justification of the application of a rule is of a different logical modality from the moral justification of the institution as a whole.
4.8 Utilitarianism, Retribution, and the Two Levels
Our critique of the Mixed Theory so far has been that it has given us no reason to believe in a logical distinction between two types of justification of punishment. But in this final section we demonstrate that, even if such a distinction were real, it would not solve the very problem for which it was designed. Consider for example Hart’s version. Even if we were to grant the cogency of Hart’s distinction between “General Justifying Aim” and “Distribution,” the most remarkable feature of his theory that he appears simply to assume that the two aspects of punishment must be justified by two separate theories, utilitarianism and retribution.33 This presupposition is even clearer in his discussion of the “immediate aim,” which he simply states must be understood in terms of the “socially beneficial consequences” being sought. But it also applies to the “General Justifying Aim”; while Hart does not rule out the possibility that this Aim might be retributive, he makes it quite clear that he can make no coherent sense of retribution as a General Justifying Aim, and relegates retributive concerns to the role of side-constraints on the utilitarian aim.
Not only does Hart assume that the Aim must be utilitarian and the Distribution retributive, but remarkably he does not seem to consider the possibility that both utilitarian and retributive values might apply at each level. For Hart, the aim of the criminal law must be either to secure socially beneficial consequences, or to inflict retribution for its own sake, but not both.34 Thus consider his objection to retribution as a General Justifying Aim on the grounds that it is “paradoxical” to see the law as “designed not to discourage murder at all,” but simply to extract a penalty – as if one could not say it has both purposes, retributive and deterrent (1968, 8). But there is no reason to exclude the possibility that both theories may provide answers to both “questions.” It is of course obvious why Hart would desire to avoid this result, for it would mean that rather than resolve the conflict between utilitarianism and retribution, it would simply reproduce the problem, but now at two different levels.35
The problem is a general one for the Mixed Theory, including for the Practice Conception argument. Hugo Bedau flatly asserts that the justification of the practice and of the individual sentences “necessarily” refer to “different considerations” (2010, 3). But he gives us no argument for this extraordinary claim of a “necessary” distinction. Thus even if we believed that there are two separate levels of justification (itself a problematic claim), it would not follow either that the utilitarian theory must apply to the one and the retributive theory to the other, or that both could not apply to each level. Indeed, as a matter of empirical fact both utilitarian and retributive motivations have always entered into decisions at every level in the criminal justice process. The criminal justice system is designed both for social protection and to ensure the punishment of the guilty; the rules of evidence are designed to balance the values of efficiency at trial with the rights of the defendant; prison sentences are determined both by the culpability of the defendant and his likelihood of reform (the parole system, for instance, is a good example of how sentencing must balance these concerns). It is a sign of how detached from reality the Mixed Theory has become that its proponents simply ignore the fact that it does not fit punishment as an actual institution. Quinton is one of the rare Mixed Theorists who acknowledges that the theory is “not an accurate description of the very complex penal systems actually employed by states, institutions, and parents” (1972, 15), but he nevertheless defends his account as a “schema, a possible limiting case” (id.). However, if he means it is a normative guide or an ideal to which our system aspires, he is equally mistaken. Our criminal justice system does not aspire to limit utilitarian concerns to the level of the rules, and retributive concerns to the level of cases, nor has there ever been a serious reform movement to change it in that direction.
The theory, in the end, will satisfy no one. For utilitarians, it is irrational to adopt utilitarianism as the overall aim of the system, but then to deny the relevance of utilitarian values at the level of sentencing (again, this concern does not apply to rule-utilitarians, who are the only people likely to be happy with the Two-Level Strategy). For retributivists, the objections go even deeper. The very idea that utilitarianism is the General Aim of the institution of punishment itself is morally objectionable, for it begins with the premise that people may be used as a mere means to promote the social welfare. To add in retributive side-constraints in a subordinate role at the level of sentencing is hardly sufficient to overcome this objection. Moreover, to the retributivist, the Mixed Theory is in no way a “compromise” between utilitarianism and retribution, but rather utilitarianism with a few ad hoc restrictions added in to answer its critics.36 This is most obvious in the case of Rawls and other rule-utilitarians, but it is implicit in Quinton (who sees retribution as merely a logical principle and not even a moral doctrine, so that all the moral work is done by utilitarianism), and also in Hart, whose disdain for retributivism is barely concealed (he implies that there is no such thing as retribution, for positive retributivism is merely a form of utilitarianism or is empty, and that negative retributivism derives from some unspecified further principles of justice). Hence even if the Mixed Theory could establish the existence of separate “questions” about punishment (assuming it could even agree on what those questions are!), it would not follow that they had solved the problem of justifying punishment, for they would have to face the same problem all over again, this time at each separate level. Rather than solve the antinomy of punishment, the Mixed Theory threatens to make it twice as bad.
The Mixed Theory of punishment promised a logical reconciliation of utilitarian and retributive concerns that would satisfy both camps, but instead attempted to achieve by conceptual fiat a certain predetermined result: a “logical” distinction within punishment between two aspects or “questions”, one of which is governed by utilitarianism and the other by retribution. It is thus no coincidence that virtually all Mixed Theories end up discovering that the goal of punishment is utilitarian, while retribution is retained as a negative constraint on that goal. The desire to find a way of reconciling the two great theories of punishment has made philosophers all too susceptible to claims to have found a “logical” basis for achieving this goal (and would account for the remarkable extent to which the Mixed Theory has escaped serious criticism, despite its flaws). The Mixed Theory, rather than uncover complexity in the concept of punishment, artificially introduces complexity into it. The distinction it insists on are groundless and even positively pernicious in that they can lead us to falsely segregate forward-looking and backward-looking concerns in the designs of our institutions. The problem of punishment cannot be solved by inquiry into the logic of the concept of punishment. There is no avoiding moral and rational inquiry into the purpose of retribution. We begin this quest by returning to the question of revenge and its relation to retribution.
See, e.g., Rawls (1955, 7): “The answer, then, to the confusion engendered by the two views of punishment is quite simple”; Hare (1981, 163): “The well-worn ‘problem of the justification of punishment’ is easily solved”; Hugo Bedau (2010): attention to these conceptual distinctions show that “the practice of punishment is clearly justified.” Also noteworthy is the remarkable range of philosophers to whom were attributed mixed theories, including Kant, Hume, and Plato.
Benn (1967): “The question of justification arises at two levels”; Bedau (2010): “The general form of any possible justification of punishment involves several steps”; Antony Flew (2002): “The essential first step is to distinguish between justifying the punishment of particular persons and justifying the maintaining of systems of laws with penalties for infractions.”
Zaibert (2006). Actually, “ordinary language” philosophy is closely related to but not identical with conceptual analysis; the former emphasizes close consideration of the way words are used in ordinary language. Most defenders of the Mixed Theory do not rely heavily on ordinary language analysis.
Thus Quinton’s argument seems to turn on the question of whether conceptual analysis can provide a priori necessary truths.
Cf. also his odd assertion that punishing the innocent is a kind of “lying” imputation of guilt (id., 7), as if the wrong involved in punishing the innocent were morally equivalent to telling a lie.
Legal formalism is widely rejected among legal philosophers. But see Ernest Weinrib’s defense of the doctrine in “Legal Formalism” (1999).
See e.g. Walker (1999, 598).
Note that formalists would have to make the further arguments that laws are conceptually connected to a proportionate punishment, and that not breaking the law is conceptually connected to not being punished.
In. Acton (1969, 127).
See also Mabbott (1972, 175), Mabbott (1969, 127–128), Hart (1968, 7, 39), Hare (1981, 163), Bedau (2010, 3) (“The values and considerations appropriate to justifying acts are often assimilated to those that define judicial responsibility, whereas the values that bear on justifying the punitive institution are akin to those that govern statutory enactments by a legislator.”).
The idea that legislatures only take into account utilitarian motives in the debate on criminal justice and never retributive concerns is also bizarre and clearly falsified, as any examination of Congressional debates on punishment will quickly demonstrate.
Mistretta v. US, 488 U.S. 361 at 391 (1989).
Whether this account really counts as “retributive” is a question I leave until later in this essay.
For a recent discussion of the fundamental objections to rule-utilitarianism, see Arneson (2005).
See e.g. Smart, in Smart and Williams (1963, 9–12).
Hart in fact begins with the puzzling distinction between the “immediate aim” of the criminal law, which is to secure “socially beneficial consequences,” versus the “justification” of punishment (id., 11). This distinction raises multiple further problems: what is an “immediate aim” – is it to be contrasted with long-term aims? Is it another name for “general justifying aim”? Is there an immediate aim for criminal law but not for punishment? Is there a justification for criminal law as well as for punishment? Why does Hart attach the word “justifying” to the general aim, but not the immediate aim? Why does he separate the criminal law from punishment? And why must the immediate aim be utilitarian?
Dolinko thinks Hart is “confused” here in failing to distinguish the rational justification from the moral justification (though it is unclear what Dolinko means by “rational justification”) (1991, 591).
If Hart is implying that a principle of distributive justice is at work here, he does not explain himself, though some have attempted to develop a theory of punishment based on distributive justice (see, e.g., Scheid 1997, 483 ff). For a discussion of the difficulties of squaring punishment with distributive justice, see Murphy (1985), Alexander (2002, 818).
A rare example of this view is found in Barton (1999, 46).
The classic treatment is Fuller (1996).
1968, 19. Hart makes two other arguments, both equally unsatisfactory. First, Hart asserts that retributivism as a General Aim would have to hold there is a “natural” penalty for every crime (id., 25). This is a caricature of the retributivist position; the retributivist could agree with Hart that proportionality is at best a rough, commonsense judgment (id.). Second, Hart suggests the distinction is evidenced by the fact that the concept of “justification” (of which self-defense is paradigmatic) belongs on the side of General Aim, while “excuse” belongs on the side of “distribution” (id., 13–14). But most theorists believe the distinction between justification and excuse is far less clear. See e.g. Kent Greenawalt (1984). Indeed, Hart’s key example of justification, self-defense, was in fact considered an excuse, not a justification, until the mid-nineteenth century! See Rosen (1986).
Some other commentators have tried to defend Hart’s definition by demonstrating its applicability to other institutions. Joel Feinberg for example considers the game of football, distinguishing its general justifying aim from its rules of procedure and fair play. It is hard to be confident that there is a “general justifying aim” of the game of football. But even if it does have such a thing, surely the rules are part of it, indeed constitutive of the game, not an external limit imposed from without (much less a principle of “distribution”) (1990, 147). George Fletcher thinks Hart’s distinction fits perfectly on the tax system, the general aim of which is to raise revenue, while the distributional question is determined by one’s ability to pay. But it is simply false that the tax system is designed with the single aim of raising revenue. Congress pursues all sorts of aims in the tax code, including creating incentives for certain behavior (promoting energy efficiency, home ownership, business investment, etc.), and redistribution of wealth (a “distributional” question which is also a General Aim). Nor would it even make sense to talk of raising revenue in the abstract, as if that question could be separated from the “distributional” question of who will be taxed and at what rates; the question of individual ability to pay is essentially connected to the goal of raising revenue, not a separate question (2000, 419).
Dolinko (1991, 541).
It is perhaps the unstated idea behind Hart’s otherwise puzzling distinction between General Aim and Distribution: Hart seems to have chosen the modifier “general” to suggest it concerned with the system as a whole, as distinct from the specific application to individual persons (cf. 39, distinguishing general aim from particular case).
Nowhere are we given a satisfactory definition of the key terms of institution, practice, system, application, rules; nor is the concept of “logical priority” carefully defined. Rawls says (in a footnote) that he means to use “practice” as a “sort of technical term,” meaning “any form of activity specified by a system of rules which…gives the activity its structure” (1955, n. 1). But this is not particularly helpful, especially given the lack of definition of “activity” or “rules” or “system.” Are there activities not specified by rules? Are some activities more structured than others?
See Searle (1979).
Searle (1997, 456).
See, e.g., de Waal (1997, 92, 94).
The claimed logical priority of the institution to the individual cases is also belied by the fact of continual interplay between the two in our system. For example, the various “landmark” criminal justice decisions in the Supreme Court involved situations where the decision as to the individual case required a radical restructuring of the criminal justice system itself: e.g. Gideon v. Wainwright, in which defendants were guaranteed a right of counsel, Miranda v. Arizona which reshaped police interrogation, and Furman v. Georgia which resulted in overturning capital punishment as currently practiced.
Strictly speaking, for Hart the question of liability to punishment is a retributive question, while the question of amount of punishment involves both retributive and utilitarian concerns (1968, 14, 24). Note also that Hart characterizes distribution as a “limiting” principle, thus begging the question in advance in favor of negative retributivism (if utilitarianism were the distributive principle, it would be odd to call it a “limit”).
Cf. the same error in Scheid (1997, 452).
Note that even for Hart, the conflict between utilitarianism and retributivism remains unresolved with respect to setting the amount of punishment.
Indeed, Rawls seems to have used the two-question strategy as a general strategy for rescuing utilitarianism from its critics (see Rawls (1955, 30), explaining that the key problem cases for the theory, promises and punishments, can both be handled by the practice/application distinction).
- Acton, H.B. (ed.). 1969. The philosophy of punishment. Macmillan: St. Martin’s Press.Google Scholar
- Alexander, Larry. 2002. The philosophy of criminal law. In The Oxford handbook of jurisprudence and philosophy of law, ed. Jules L. Coleman and Shapiro Scott. Oxford/New York: Oxford University Press.Google Scholar
- Arneson, Richard. 2005. Sophisticated rule consequentialism: Some simple objections. Philosophical Issues (Normativity) 15: 235–251.Google Scholar
- Barton, Charles. 1999. Getting even: Revenge as a form of justice. Chicago: Open Court.Google Scholar
- Bedau, Hugo. 1974. A World without punishment. In Punishment and human rights, ed. Goldinger Milton, 141–162. Cambridge, MA: Schenkman Pub. Co.Google Scholar
- Bedau, Hugo and Kelly, Erin. 2010. Punishment. In The stanford Encyclopedia of philosophy, ed. Edward N. Zalta, URL = <http://plato.stanford.edu/archives/spr2010/entries/punishment/,
- Benn, Stanley. 1967. Punishment. In The Encyclopedia of philosophy, ed. Edwards Paul. New York: Macmillan.Google Scholar
- De Waal, Frans. 1997. Good natured: The origins of right and wrong in humans and other animals. Cambridge: Harvard University Press.Google Scholar
- Fletcher, George. 2000. Rethinking criminal law. New York: Oxford University Press.Google Scholar
- Flew, Antony. 2002. Punishment. In Dictionary of philosophy, 3rd ed, ed. Flew Antony. New York: Macmillan.Google Scholar
- Fuller, Lon. 1996. The morality of law. New Haven: Yale University Press.Google Scholar
- Harris, James. 2002. Analytic philosophy of religion. NewYork: Springer.Google Scholar
- Hart, H.L.A. 1968. Punishment and responsibility. Oxford: Oxford University Press.Google Scholar
- Jackson, Frank. 1998. From metaphysics to ethics: A defense of conceptual analysis. Oxford: Oxford University Press.Google Scholar
- Kant, Immanuel. 1991. The Metaphysics of morals. Trans. Mary Gregor. New York: Cambridge University Press.Google Scholar
- Lacey, Nicola. 1994. State punishment. London/New York: Routledge.Google Scholar
- Lacey, Nicola. 2004. A life of H.L.A. Hart: The nightmare and the noble dream. New York: Oxford University Press.Google Scholar
- Mabbott, J.D. 1969. Punishment. In The philosophy of punishment, ed. H.B. Acton. London: Macmillan.Google Scholar
- Mabbott, J.D. 1972. Punishment. In Philosophical perspectives on punishment, ed. G. Ezorsky, 165–181. Albany: SUNY Press.Google Scholar
- Murphy, Jeffrie (ed.). 1985a. Punishment and rehabilitation, 2nd ed. Belmont: Wadsworth.Google Scholar
- Murphy, Jeffrie. 1985b. Marxism and retribution. In Punishment and rehabilitation, ed. Jeffrie Murphy, 74–92. Belmont: Wadsworth.Google Scholar
- Primoratz, Igor. 1989b. Justifying legal punishment. Atlantic Highlands: Humanities Press.Google Scholar
- Quinton, A.M. 1967. On punishment. In Philosophical perspectives on punishment, ed. Gertrude Ezorsky, 6–15. Albany: SUNY Press.Google Scholar
- Ramsey, William, and Michael DePaul (eds.). 1999. Rethinking intuition. Maryland: Rowman and Littlefield.Google Scholar
- Rosen, Cathryn Jo. 1986. The excuse of self-defense. The American University Law Review 36: 11–56.Google Scholar
- Ross, W.D. 1930. The right and the good. New York: Oxford University Press.Google Scholar
- Scheid, Don. 1997. Constructing a theory of punishment, desert, and the distribution of punishments. Canadian Journal of Law and Jurisprudence X(2): 441–506.Google Scholar
- Searle, John. 1979. The social construction of reality. New York: Free Press.Google Scholar
- Smart, J.J.C., and B. Williams. 1963. Utilitarianism: For and against. Cambridge, UK: Cambridge University Press.Google Scholar
- Ten, C.L. 1987. Crime, guilt, and punishment. Oxford: Clarendon Press.Google Scholar
- Weinrib, Ernest. 1999. Legal formalism. In A companion to law and legal theory, ed. Patterson Dennis. Cambridge: Blackwell.Google Scholar
- Zaibert, Leo. 2006. Punishment and retribution. Aldershot: Ashgate.Google Scholar