Moral Cognition in Criminal Punishment

An initially plausible interpretation of Kant's justification for the imposition of criminal punishment is straightforwardly retributive: the government can and must punish criminals because (and only because) they have committed a particular kind of wrong—that is, one which violates the Universal Principle of Right (UPR).

In the Rechtslehre, the section of Kant's Metaphysics of Morals concerned with matters of justice, the UPR is given as the standard against which we can measure our social progress toward justice. It is a “universal criterion” we can use to determine whether an act is just or unjust, and it (similarly to the Categorical Imperative proposed in the more well-known Groundwork of the Metaphysics of Morals) is cognizable “in reason alone.”

Kant, supranote 2, at 23 (6:230).

Kant thinks that if we affirm the UPR (which we must if we are to have a just social order) then we will agree that “whatever is wrong is a hindrance to freedom in accordance with universal laws.”

Id. at 25 (6:231).

To give a simple example, the UPR would require that I refrain from kidnapping a fellow citizen—a very obvious deprivation of that person's freedom. If I were to kidnap someone, however, the State could justifiably imprison me—an equally clear deprivation of my freedom. Imprisoning citizens would normally be unjustified: the ruling party cannot simply imprison opposition leaders on a whim, because this would violate the UPR. But imprisoning me after I have violated the UPR is consistent with the UPR.

On this view, punishment is a moral obligation, not merely a facultative policy option at the state's disposal. The State may not consider another rationale for punishment: it “can never be inflicted merely as a means to promote some other good for the criminal himself or for civil society. It must always be inflicted upon him only because he has committed a crime.”

Id. at 105 (6:331) (emphasis in original).

Furthermore, once punishment has been found to be warranted on retributivist grounds, the “quality and quantity” of punishment must also be determined by the “law of retribution” or lex talionis:

Id. at 105 (6:332). Kant uses the term jus in place of lex, and this may be the most appropriate word: jus properly refers to “law in the abstract” or a “legal right, power, or principle,” while lex is technically reserved for “[p]ositive law,” or even “a statute.” Black’s Law Dictionary, s.v. “jus” and “lex.” However, this technical distinction is not well maintained in Anglo-American legal scholarship, which tends to use “lex” in this context.

One of the common passages used to support the retributivist interpretation is where Kant avers that one who murders another must be put to death in order to satisfy the law of retribution, and any lesser punishment for any reason would be a “public violation of justice.”

Id. at 106 (6:333).

Although facially plausible, there is a rather significant problem with interpreting Kant as a retributivist simpliciter: the passages used to justify the retributivist interpretation are contradicted by other passages, sometimes on the same page of text. For example, although Kant says that punishment must be meted out only “because [the criminal] has committed a crime,” he also seems frequently to refer to deterrence or rehabilitation as goals of criminal punishment. Thus he says that the State may properly “draw[] from [the criminal's] punishment something of use for himself or his fellow citizens,” which sounds suspiciously like rehabilitation (something of use for the criminal) and general deterrence (something of use for fellow citizens).

Id. at 105 (6:331).

Another textual oddity bears particular mention. Kant explicitly approves of non-punishment in cases of what he terms “necessity”—for example, where the drowning man kills another in order to save his own life.

Kant, supranote 2, at 28 (6:235–36).

A retributivist adhering to lex talionis would have to say that the drowning offender should be put to death. We might, more commonly, be inclined to say that the person in this case has an excuse for his wrongdoing: perhaps it was wrong to kill the victim, but we can understand why one might do so in order to save one's own life. Finally, we might think that what the killer did was wrong and inexcusable, but was not murder—perhaps it was manslaughter or some kind of lesser crime. These latter two cases might be compatible with (a weaker form of) retributivism, though not with lex talionis. Still, Kant does not appear to give any of these explanations. Instead, he takes the death penalty off the table for what appear to be purely utilitarian reasons: because the goal of deterrence would be fruitless in such a case. If the retributivist interpretation of Kant's penal theory is correct, then the section on “necessity” is mistaken: Kant neglected to follow his own reasoning to its logical conclusion in this particular case.

In summary, aside from the obvious point that goals such as deterrence and incapacitation are intuitively reasonable ones that Kant likely would have included in his theory (as he explicitly does in the necessity case), the retributivist's main problem is reconciling the clearly retributive-sounding passages in the Rechtslehre with other, equally clearly non-retributive-sounding ones. These do not exhaust the potential worries we might have about retributivism generally, or even about a retributivist interpretation of Kant

For additional considerations, see Scheid, supranote 1; Byrd, supranote 4; Ekow Yankah, Crime, Freedom, and Civic Bonds: Arthur Ripstein's Force and Freedom: Kant's Legal and Political Philosophy, 6 Crim. L. & Phil. 255 (2012).

Legal and moral philosophers have proposed various versions of “mixed” or “hybrid” theories of punishment, not all of which claim to derive from Kantian thought. As a historical matter, one might see mixed theories as an obvious solution to a philosophical problem: that utilitarianism and retributivism are both attractive but ultimately deficient theories with which to justify punishment. As Whitley Kaufman puts it: “[i]n the mid-twentieth century, it was widely believed that the problem [of justifying punishment] 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 to reconcile utilitarianism and retribution . . . .”

Whitley Kaufman, The Rise and Fall of the Mixed Theory of Punishment, 22 Int’l J. of Applied Phil, 38, (2008).

To give a salient example of someone who applies the “mixed” viewpoint to Kant, B. Sharon Byrd argues, based on textual as well as “[h]istorical considerations”

Byrd, supranote 4, at 151. Byrd does not develop her “historical” argument; she cites one journal article for the proposition that “[t]he distinction between punishment as a threat used actively to bind an individual to the rules of conduct and the execution of punishment as a response to the breach of one's obligation was made both before and immediately following Kant's critical philosophy.” Id. at 184. I do not, however, think that this point is crucial for her argument, which is mainly conceptual and textual.

Byrd correctly points out that Kant does not claim that lex talionis does the work that pure retributivists need it to. It is, at most, a limiting principle on the extent of punishment. Scheid phrases the limiting potential of lex talionis this way: “What right does the state have to punish an individual for the purpose of deterring others? Indeed, whether the individual is guilty or not, how can the state ever be justified in using a person in this way, as a mere means?”

Scheid, supranote 1, at 272.

So on this “mixed” interpretation, the role of lex talionis is as a guarantor of “moral acceptability” of the punishment itself. Crime-control considerations (deterrence) give us reason to threaten citizens with punishment. But the “law of retribution” guarantees that, if someone does in fact break the law, she will be punished to the extent she deserves—and no more (but, of course, also no less). This would seem to allay one concern that is commonly raised about purely utilitarian punishment schemas: that they license the punishment of the innocent and the overpunishment of the guilty. Kant would, on this view, permit citizens to be threatened with punishment for violations of the UPR, to whatever extent necessary to deter crime—but would ensure that the actual punishments citizens receive were limited by lex talionis.

This view has some obvious advantages over the pure-retributivist one. It explains the apparent contradiction in Kant's use of deterrence language in some places and retribution language in others: he simply has in mind two different functions of criminal punishment (and perhaps neglects to distinguish clearly between them in his text). The hybrid approach also appeals to common-sense intuitions about the need for societies to deter crime—a theory of criminal justice that sees no role for deterrence would, at best, be one that departs radically from most countries’ penal practices.

There is, however, a problem with the hybrid view, which suggests that Kant would not have endorsed it. Some reflection on how criminal justice systems function will show that bifurcating the criminal law into threat and execution is pragmatically bizarre and possibly incoherent. Assume for a moment that we had a mixed deterrence-retributivist system. Legislators under this system would need to enact legislation that threatens citizens optimally. So, for example, we might find that threatening fifteen years in prison for burglary is the best way to deter people from committing burglary. Of course, it is unrealistic to suppose that nobody will ever commit burglary even when the threatened sentence is this high. Suppose, then, that a judge is now faced with sentencing a burglar—one of the few who were not dissuaded by the severity of the law. We might find that lex talionis demands that the burglar receive five years in prison for his crime. True, we threatened him with fifteen years, but he deserves exactly five years.

We now face a conundrum. If the judge imposes five years, then he has rendered the legislation ineffectual. Citizens would observe that, despite what the law says, the burglar in fact only gets five years. The deterrent effect of the fifteen-year threat is vitiated—indeed, we might expect that the only deterrent effect would be of the actual five-year sentence. On the other hand, if the judge were to impose fifteen years, then the deterrent effect of the legislation would be upheld—but at the impermissible cost of violating lex talionis. So it would appear that the mixed-justification view leads either to illegitimacy (the law is deceptive and untrustworthy) or incoherence (lex talionis, putatively necessary, in fact cannot provide the judge with a reason for deviating from what the legislature has threatened).

It would also be possible for lex talionis to demand a more severe punishment than necessary for deterrence purposes, though this seems less likely in practice. In fairness, my example presupposes that would-be criminals have access to, and in fact obtain, information about sentences handed down by the criminal courts. It might be possible to construct a criminal justice system where all sentences are private—this is the default position in juvenile delinquency cases in the United States, for example, though they are not technically “criminal” matters. It seems unlikely, however, that nobody would catch on to the fact that there was a radical difference between threatened and imposed punishments—surely questions would be raised when Uncle Bill suddenly turned up at the family reunion only five years after his burglary conviction. More importantly, a criminal justice system that was radically private in this way might face significant questions about procedural fairness. In any event, it seems very unlikely that this is the kind of system that Kant had in mind—and it is certainly not one likely to be relevant to readers.

A related question is whether it is reasonable that, as Byrd would have it, our focus should shift entirely away from “instrumental” concerns once an accused criminal has been convicted. Is it really the case that “just treatment of the individual” is the only thing that we ought to be concerned with in determining her punishment? This seems unlikely, for several reasons. Common sense would lead us to assume that, in order for the deterrent effect of a “threat” to be effective, we would at least need to publicize criminals’ sentences—otherwise, the threat of criminal sanctions would be an entirely empty one. Moreover, Kant would surely not say that the only thing that matters once a criminal has been sentenced is how we treat him. This is an important concern—perhaps the most important one—but other considerations merit our attention: whether a dangerous offender will be incapacitated, whether there is a mechanism for publicizing the punishment to the public for deterrence purposes, whether anyone victimized by the crime has been vindicated, and so forth.

For example, suppose that David is convicted of raping Victoria. On the hybrid view, the deterrent purpose of (the threat of) punishment is served by the promulgation of public legislation proclaiming that rape shall be punished by exactly thirty years in prison. This lengthy sentence should, the legislature thinks, dissuade any rational person from committing rape. Once David has been convicted, though, all that matters is discerning the “just” sentence for him. But in order to have a hybrid system be coherent (or something besides a lie), then lex talionis has to play some role at the deterrent stage (the legislature has to figure out what is the appropriate penalty—or range of penalties, perhaps—for a given crime). Conversely, deterrence must play a role in sentencing (it would seem perverse from a deterrence standpoint if we sentenced David to a $100 fine and 20 hours of community service in lieu of prison, even if that were demanded by lex talionis). Surely, if we care about deterrence at all, then we still care about it when we are punishing David. So the “focus-shifting” aspect of the hybrid view is also problematic.

To be fair, interpreting Kant via a mixed theory that separates criminal punishment into threat and execution does seem to be an improvement over the pure-retributivist position. And perhaps there is some way to save the hybrid or mixed approach in order to avoid the kind of problem I have identified. I think, however, that we need not make such an attempt—because there is a third way of interpreting Kant's views on punishment which will suggest that whatever roles deterrence and retribution play in Kant's theory, they are at most derivative features of his account of justice as presented in the Rechtslehre.

Kant's account of criminal punishment is, I believe, best understood as one facet of his theory of justice. As explained above, Kant conceives of a just society is one whose basic structure is founded on the Universal Principle of Right. The UPR guarantees that all citizens will be able to pursue their chosen ends, to the extent that those ends are compatible with those of their fellow-citizens. We are free, in the political (rather than moral) sense, when we choose to be governed by just laws, which in turn ensure that our wills are not governed by others’. A requirement of justice, then, is that each of us, as citizens, willingly submits to the just laws of our community.

Willful violations of these just laws—specifically, of those laws which ensure the political conditions of citizens’ civic freedom—are properly referred to as crimes. A criminal is someone who has willfully violated the civic freedom of his fellow citizens; he has not only interfered with others’ freedom, but has also done so in a way that undermines the foundational structures upon which such freedom is based.

Note that if one violates another's civic freedom, but lacks the will to do something incompatible with the UPR (i.e. does not act on such a maxim), then the violation is, by definition, not a criminal act. The state may be justified in using some kind of lesser coercion in order to restore the rightful condition that existed before, but is not justified in violating the actor's civic freedom.

A criminal, then, is someone who has broken the reciprocal bond that is the foundation of a just society. The “rightful condition” of civil society has been upset. According to Kant, though, restoring a state of free, equal, and independent citizenship depends crucially on the availability of state coercion—indeed, this is why he says that such coercion is necessary.

Kant, supranote 2, at 89 (6:311–12). Note that Kant's view is retributive in this limited sense: crime necessitates punishment, and so punishment will always be in “retribution” (a more neutral term might be “response”) for a crime.

This line of reasoning has strong intuitive force. We perceive societies that lack adequate police and judicial powers—or willingness—to prosecute criminals as being less free than societies which have such capacities. We feel that our government does wrong, and we worry about our own rights and those of other citizens, when it fails to punish certain classes of offenders for spurious reasons—racial bias being an obvious example. The State is, then, rightly viewed as obligated to protect us against violations of the conditions of civic freedom, and to hold people who do commit such violations accountable for their actions.

The fact that the state is obligated to punish criminals in order to restore a rightful condition consistent with the UPR is helpful. Still, we may be left wondering how it is that punishment could be justifiable for one who embraces Kantian values, such as respectful treatment of all human beings. After all, “[c]riminal punishment is coercive state power in its most brutal form. . . . If locking human beings in cages or killing them is not a bad way to treat people, it is hard to imagine what would be. Punishment, in short, seems to involve conduct that is in itself wrong.”

Murphy, supranote 1, at 1.

The answer is, in one sense, quite simple. A citizen's civic freedom is, as we have seen, dependent upon his respect of others’ civic freedom. But a crime is, by definition, a willful violation of another's freedom. Therefore, it follows that someone who has committed a crime loses the protection of the UPR. A criminal has, in a sense, forfeited or lost his civic freedom.

This account is probably compatible with the rights-forfeiture view of punishment, at least on some accounts. See, e.g., Christopher Heath Wellman, The Rights Forfeiture Theory of Punishment, 122 Ethics 371 (2012). It does, however, provide a clearer explanation of why we are justified in asserting that criminals forfeit their rights.

Now, this answer is perhaps too simple. For it would appear, upon closer inspection, to engender some questionable results. For one thing, if someone has lost his civic freedom entirely, then the government would appear to be justified in doing anything to him. Someone who is not free could, after all, be made a slave. Another worry is that if someone lacks civic freedom, then he is not a citizen (at least not in the full Kantian sense). But if he is not a citizen, then his fellow-citizens have no obligations toward him as far as justice is concerned. They may retain some moral obligations toward him—but they would still be justified in imposing punishment on him themselves, rather than letting the government do so.

These concerns are unwarranted, however, because Kant places some definite limitations on the coercive power of the state—and these limitations are not merely ad hoc, but are results of the structure of civic freedom. First, Kant asserts that while criminals lose the “dignity of citizenship,” they do not lose their human dignity.

Specifically, Kant asserts that “no human being in a state can be without any dignity, since he at least has the dignity of a citizen. The exception is someone who has lost it by his own crime, because of which, though he is kept alive, he is made a mere tool of another's choice (either of the state or of another citizen).” Kant, supranote 2, at 104 (6:329–30) (emphasis removed). One oddity here is the parenthetical at the end of the passage, which makes it sound as if Kant thinks that criminal punishment can somehow be meted out by private citizens, rather than only by the state. This contradicts his later statements about punishment being the province of the government alone. It is possible that Kant has in mind some kind of lesser (civil) wrong; this may be what he has in mind when he later speaks, without clarification, of “private crime.” Id. at 105 (6:331). I shall proceed under the assumption that this passage is either an aberration, or explicable in another way.

This, then, explains why Kant says that criminals lose their dignity of citizenship: by violating justly enacted laws, they betray themselves as legislators. They remove themselves from the body of those who act according to the “united will,” always upholding the UPR. Note, however, that the criminal retains her dignity of humanity. Human dignity is without “price”

Kant, supranote 38, at 84 (4:434).

So the first limitation on punishment is that the criminal's humanity must not be violated. Precisely what this means is debatable, but it seems clear that some kinds of treatment will be out of the question: torture, rape, and other sadistic practices are obviously incompatible with people's humanity, and even if the criminal has perpetrated such acts on others, we can never be justified in doing so to him. One important question, particularly for American criminal justice, is whether capital punishment is compatible with the respect of human dignity. As we shall see, Kant thinks so—but we might reasonably debate whether this is the case.

Another clear limitation on punishment has to do with who is doing the punishing. If offenders lose their civic dignity, then why could other citizens not punish them? Kant thinks part of the social contract (wherein we, as members of a just community, agree to act in accordance with the UPR) entails leaving the right of punishment in the hands of the government. Thus, for example, he states that the “right to punish is the right a ruler has against a subject to inflict pain upon him because of his having committed a crime.”

Id. at 104 (6:331) (emphasis added).

In addition, recall that punishment restores the “rightful condition” of civil society. But individuals cannot, on their own, create such a condition, which is the product of the “united will” of a body of citizens. It follows that individuals cannot, by themselves, restore such a condition once it has been disrupted by a criminal act. Only the state—which, of course, is but a representation of the “united will”—can punish criminals in such a way that the reciprocal nature of the UPR can be upheld.

We have, then, good Kantian reasons to assert that punishment must be administered only by the state. We also saw that punishment which violates human dignity cannot be justified. One problem remains, however. Even if we were certain that a particular type of punishment—say, confinement within a safe, well-maintained correctional facility—was permissible, there would seem to be nothing to prevent the government from using it in ways that seem, intuitively, to be unjustifiable. For example, the government could imprison a petty thief for the rest of her life. Even if imprisonment as such is not inhumane, there is something intuitively unjust about imprisoning someone for many years for a minor offense. This is, of course, the oft-discussed problem of proportionality. How are we to ensure that whatever punishment we impose “fits” the crime?

As mentioned in subsection A above, Kant holds that the nature and extent of a punishment should be determined by the principle of lex talionis: punishments should be, as far as possible, identical to the crime committed. Kant thinks that this will answer both the means and extent questions—thus when someone steals, his property is forfeited; when he commits murder, “he must die.”

Kant, supranote 2, at 106 (6:333) (emphasis removed).

Perhaps some insight can be gained by considering why Kant seems to insist on lex talionis. What argument could justify his reliance on this principle as the only proper response to questions about extent and method of punishment? Here is the main passage in the Rechtslehre on this topic:

But what kind and what amount of punishment is it that public justice makes its principle and measure? None other than the principle of equality (in the position of the needle on the scale of justice), to incline no more to one side than to the other. Accordingly, whatever undeserved evil you inflict upon another within the people, that you inflict upon yourself. . . . [O]nly the law of retribution (ius talionis)—it being understood, of course, that this is applied by a court (not by your private judgment)—can specify definitely the quality and the quantity of punishment; all other principles are fluctuating and unsuited for a sentence of pure and strict justice because extraneous considerations are mixed into them.

Id. at 105–06 (6:332) (emphasis in original). Note that Kant's reference to “within the people” is consonant with the view that Kant conceives of crime as a breach of the social contract based on the reciprocal nature of the UPR. Wronging someone outside the state is not a crime (though morally wrong), because one is not legally bound in a reciprocal relationship with that person. The State cannot overstep its authority and punish wrongs that occur to people who are not parties to the social contract. This raises the question of whether Kant's theoretical framework can provide a reasonable account of international criminal law. I cannot hope to broach that topic here, though it should be noted that Kant does discuss matters of international justice, particularly in Toward Perpetual Peace: A Philosophical Project (Mary Gregor trans.) in Immanuel Kant: Practical Philosophy 315–351 (Paul Guyer and Alan W. Wood eds., 1996)

Of note here is that the principle motivating lex talionis is not retribution—it is equality. Although Kant uses the word “retribution,” it is employed only to identify lex talionis, not to justify it. Kant's sole reason for claiming that lex talionis is the only just way to punish is that any other way of meting out punishment would be “fluctuating.” Competing principles would cause the scales of justice to tip to one side or the other—to become unbalanced—resulting in an inequitable distribution of punishments.

In section II, I will argue that lex talionis

For a more detailed argument against lex talionis, see Sarah Holtman, Toward Social Reform: Kant's Penal Theory Reinterpreted, 9 Utilitas 3 (1997). Holtman argues that Kant made some unwarranted empirical assumptions that motivate his discussion of lex talionis; if we do not accept these assumptions, then lex talionis becomes a less compelling principle. Id. at 18. Assessing the strength of each of these assumptions would be complicated; Holtman's main project, though, is not to argue definitively against Kant's insistence on lex talionis, but simply to point out that this commitment does not follow necessarily from his larger theory of justice. Id. My goal in the latter half of this paper is similar.

Similar considerations seem to be at work in another passage, where Kant argues that it would be impermissible to offer to “preserve the life of a criminal sentenced to death if he agrees to let dangerous experiments be made upon him and is lucky enough to survive them, so that in this way physicians learn something new of benefit to the commonwealth.”

Kant, supranote 2, at 105 (6:332).

Kant's objection, however, is probably twofold. First, this kind of experimentation seems to use the criminal's very life as a mere means to an end, which is categorically prohibited by the moral law.

Kant, supranote 38, at 80 (4:429).

The bottom line here is that Kant thinks lex talionis is the proper answer to questions about the mode and extent of punishment because it is conducive to treating criminals equally, and equality is important in a system of Kantian justice. Whatever qualms we might have about Kant's embrace of lex talionis, however, we have seen how punishment can reasonably be viewed as a requirement of Kant's account of civic freedom. And this account is, I hope to have shown, significantly more convincing than the existing retributivist and hybrid ones.

Thus far, our interpretation of Kantian punishment as a facet of his theory of justice sounds reasonable. Can we, though, reconcile this interpretation with textual concerns the hybrid theorist raises? Consider first the references Kant makes to deterrence (and occasionally rehabilitation) as apparent goals of punishment. Threatening citizens with punishment for the violation of the UPR can be seen as a way of preserving the freedom, equality, and independence of all concerned. Such deterrent threats, if effective, reduce the likelihood that a citizen will be victimized by a criminal. Deterrent threats of punishment are applied equally to all citizens—nobody is singled out as a potential criminal—which sends the message that the state takes seriously the rights of all citizens and wishes all to benefit from the freedoms gained by participation in civil society. They also put potential offenders on notice that their freedom will be diminished, and their status as independent citizen-agents jeopardized, should they choose to act in abrogation of their basic duties as citizens to uphold the UPR.

Thus, while deterrence is arguably an important aspect of (the threat of) criminal punishment in a Kantian scheme, this would only be the case insofar as deterring crime actually promoted Kantian justice within civil society—an important point that hybrid theories do not recognize. This is why we are able to evaluate the justness of criminal laws independently of their deterrent efficacy; unjust laws may well serve deterrent purposes admirably but nevertheless be problematic because, say, the act being punished is one compatible with (or even necessary for) civic freedom.

Rehabilitation as a (partial) aim of punishment from a Kantian perspective is a more interesting question. Insofar as the criminal justice system could “rehabilitate” offenders, it is worth asking what this might mean. Perhaps we could conceive of a kind of civic rehabilitation, in which the offender is offered help regaining his literal and figurative citizenship: his place in the community, and his commitment to the civic freedom of his fellow-citizens. This type of rehabilitation seems reasonable, and compatible with Kant's view of civic freedom.

The justice-based interpretation can also make sense of the seemingly strange passage on “necessity” mentioned in above: the drowning man who, “in order to save his own life, shoves another, whose life is equally in danger, off a plank on which he had saved himself.”

Kant, supranote 2, at 28 (6:235).

If I am drowning in the ocean, I am not acting as a citizen within civil society, but as an animal struggling for survival. The circumstances are not such that the demands of justice are relevant—nor can they be met. No action that I take under such circumstances will preserve the aims of the UPR. If I let myself die (which Kant acknowledges may be the more noble and morally worthy act), I fail to preserve my own freedom; whereas appropriating the plank for myself will fail to preserve the freedom of the other drowning man. There can be no equality here, since one of us must die. And our status as independent citizens is hardly at issue at the moment. Since state coercion is justifiable only to preserve the aims of freedom, equality, and independence of citizens, it fails to be relevant in this particular case—even though we privately might judge that I lack some personal virtue because I chose to save my own life at the expense of another's. In legal terms, I am excused, though not justified: what I did might not be morally right, but I should not be punished for having done it.

Thus, while the hybrid theorist can point to the few references to deterrence in order to explain the “necessity” case, the justice-based interpretation provides a fuller and more compelling explanation of that initially abstruse passage. Now, however, we need to confront a particularly well-known section of the Metaphysics of Morals that is commonly used to paint Kant as a hard-nosed retributivist. Can we square Kant's discussion of capital punishment with the justice-based interpretation?

Kant categorically rejects any sort of mitigation or tempering of capital punishment for convicted murderers.

Kant does say that we ought to execute murders humanely. Id. at 106 (6:333). However, this seems a rather minor concession considering that he apparently does not consider the possibility that any murder might be justly punished by anything other than death (except of course in the “necessity” cases).

I believe, however, that this is not an accurate reading of the death-penalty discussion, which begins precisely by distinguishing murder from other crimes. Kant prefaces the murder passage with a discussion of theft, pointing out that, since taking all a thief's possessions would result in a burden on the state to “provide for him free of charge,” the thief can be forced to perform “prison labor” instead.

Id. at 106 (6:333).

For this reason, Kant thinks that, while the thief can be punished by a method other than stealing, if a criminal “has committed murder he must die. Here there is no substitute that will satisfy justice. There is no similarity between life, however wretched it may be, and death, hence no likeness between the crime and the retribution unless death is judicially carried out upon the wrongdoer.”

Id. at 106 (6:333) (emphasis in original).

Many will nonetheless find Kant's argument in favor of capital punishment insufficient. We might think that, while murder is heinous and should be treated accordingly, the government ought to consider other options in punishing even the worst kind of criminals. If we are skeptical about the exacting demands of lex talionis in the first place, then we certainly should question whether homicide is really the only appropriate response to homicide. And while Kant criticizes an anti-death-penalty advocate for being “moved by overly compassionate feelings,”

Id. at 108 (6:334–35).

Kant may simply not have thought out the injustice of executing certain classes of citizens. But we should also remember that Kant's initial attraction to lex talionis arises primarily out of a concern with equality. A common criticism of the death penalty as implemented in the United States is that it is racially biased. Although this is generally taken to be an argument in favor of abolishing capital punishment, one could understand why Kant, given a particular historical context, might insist on executing murderers. While it would be patently unfair to execute only black murderers while sparing the lives of white murderers, there is at least a sense in which equality would be served by executing all murderers, regardless of race. Many contemporary readers (including me) will find this conclusion incomplete, if not perverse—after all, we do not generally find it a good argument for clearly immoral practices (slavery, genocide, and so on) that people are victimized equally. If putting people to death is prima facie wrong, then the fact that we do it equally will not justify it morally. But, as Thomas Hill puts it, “considerations of comparative justice make understandable, even if not defensible, Kant's thought that the long-standing (supposedly) just policy of executing murderers should not be abandoned” until everyone has received the same kind of treatment.

Hill, supranote 18, at 434.

Thus, regardless of one's feelings about capital punishment, Kant's endorsement of the practice can be explained within the context of a prior commitment to the promotion of a just social order—one in which citizens’ civic freedom is of primary importance. We might easily disagree with Kant's conclusions while retaining the general structure of his theory of punishment.

The foregoing constitutes an overview of three competing interpretations of Kantian punishment. I have argued that an interpretation based on Kant's account of justice—wherein the purpose of punishment is to preserve the civic freedom of all citizens—is superior to the standard retributivist and hybrid accounts. Further work remains to be done in order to show what practical implications this interpretation has for a criminal justice system aspiring to fulfill a Kantian model of justice. First, however, we must address a salient issue I mentioned and left unresolved: whether we should accept lex talionis as the ideal way of determining the nature and extent of criminal punishment.

In the Tugendlehre, the part of the Metaphysics of Morals concerned with personal morality, Kant describes two duties that one owes to oneself as an autonomous moral agent: the duty to be one's “own innate judge,”

Kant, supranote 2, at 188 (6:437).

One might question whether this conception of conscience-as-courtroom is satisfying. For one thing, why is it that there is no “reward” for morally right actions? We have a tendency to praise others for morally laudable actions that is perhaps as strong as our penchant for condemning wrongdoing. We think good parents praise their children for right behavior as well as punish them for wrong behavior. It may be that children learn to act at least in part based on rewards and punishments, and Kant would say that they have for that reason not attained full moral development. We do, though, also reward adults who (for example) perform acts of charity in the community, or who sacrifice time and money on projects that benefit others. Perhaps these are supererogatory acts: public rewards are often bestowed upon those who seem to go above and beyond the requirements of morality. But what about the person who is overall an unsavory character but manages to rise above his animalistic inclinations to perform an act of “merely moral” quality? I am thinking, for example, of the addict who celebrates a few months free of her addiction, or the lifelong Scroogeish miser who finds joy in deciding, for once, to bestow his largesse on others. While Kant would find these people to have behaved in a morally worthy fashion, he seems to deny that they should be able to “reward” themselves for such laudable behavior.

One possible response, compatible with the one I will explore shortly, is that “rewards” of this latter type are not for acting morally in the fullest sense, but because someone like the addict is, in part, morally childlike. Rewards in this type of case function as encouragement toward fuller moral development, which includes the capacity to act free of the influence of addictive substances. Importantly, though, Kant would likely point out that all of us are morally deficient in some respect—we are all subject to influences that interfere with our moral autonomy. Incentives toward goodness (publicly bestowed or granted by oneself) are perhaps a permissible way to encourage moral development, so long as we do not make the mistake of thinking of the reward as the right reason for acting morally. For further discussion of Kant's views on moral training or education, see Barbara Herman, Training to Autonomy: Kant and the Question of Moral Education, in Philosophers on Education: New Historical Perspectives 255 (Amélie Oksenberg Rorty ed., 1998).

A second potential objection is that it seems at first glance far too simplistic in at least some cases to be able to make a binary judgment about the moral worth of one's actions. If your inner prosecutor calls you to task for having stolen something, it seems obvious that your inner judge will condemn you. Likewise, if you try to prosecute yourself for a momentary lapse of forethought (you unwittingly neglect to hold a door open for someone), it seems equally clear that a well-functioning inner judge will acquit you. In many cases where there is an ostensible moral violation, though, it is not at all clear that the correct judgment is simply condemnation or acquittal. It seems that in many cases something more is required: “No, I didn’t steal anything, but I had morally unworthy thoughts about doing so”; or “Yes, I failed to treat others with polite respect but I had a lot on my mind and had every intention to behave otherwise.” How can the binary “courtroom” model account for the seeming necessity of making these more nuanced judgments?

One way of responding to these concerns is by appealing to the second duty that Kant presents in this section of the Tugendlehre: that of self-knowledge.

Kant, supranote 2, at 191 (6:441).

In other words, the particular self-judgments rendered by our conscience ought to be balanced by a more generalized view of our moral selves, which requires “impartiality” and “sincerity” about our “moral worth or lack of [moral] worth.”

Id.

The tempering effect of this type of self-knowledge—which allows us to avoid being both too hard and too easy on ourselves—supplies the apparently missing elements from Kant's description of the conscience. For although our inner judge might not allow us to experience “joy” simply because we are acquitted of putative wrongdoing, perhaps Kant would allow us to feel this sort of “reward” when considering ourselves objectively. Thus the aforementioned addict who has found the strength of will to remain free from her vice for a period of time might genuinely experience joy at having come this far since, all things considered, that is rather an impressive accomplishment for her. Similarly, although one's conscience renders a simple “guilty” or “not guilty” when considering the moral worth of a specific act, self-knowledge may be what supplies the “but” that we often attach to actions with moral content. Thus I might as an act of conscience correctly acquit myself of, say, a particular instance of lying, yet also as an act of moral cognition recognize that honesty is unfortunately not representative of my actions as a general matter and, therefore, I deserve very little self-satisfaction for not having lied on this particular occasion.

One question we might pose is whether Kant intends for the self-knowledge aspect of our moral duty to diminish or augment in any way the penalty our conscience-qua-judge imposes on us for violation of the moral law. Kant does not address this explicitly. Indeed, he does not say much at all about what the penalty is for a guilty verdict rendered by one's conscience—only that it will be “happiness or misery.” Presumably, though, not all moral failings ought to make us feel equally guilty: murdering and overindulgence in food hardly seem to warrant the same degree of self-imposed misery. It would seem, then, that there must be some means by which we determine how much misery or happiness we ought to allot ourselves based on our moral status. Since Kant does not supply such a mechanism in the section on conscience, but follows this up with the section on self-knowledge, it would be reasonable at least to consider whether this latter component of the moral being can do this work.

One might think that determining how much guilt to punish oneself with does not require self-knowledge in most cases. Obviously one who commits murder ought to feel much worse about himself than even the most reprehensible glutton. But the issue is not simply one of comparing two types of moral failings, but one type under different circumstances. Thus, for example, a person who has told a small lie one time to avoid a stressful confrontation might reasonably impose less misery on herself than ought to be entailed by the average lie; conversely, one who engages in systemic deception of a spouse in order to cover up an affair ought with full self-knowledge to impose a greater degree of conscience-ordered misery than for an average lie.

Kant is notoriously opposed to all forms of lying, even in cases where doing so would seem to be morally permissible, if not obligatory—as in the infamous “murderer at the door” scenario. See Immanuel Kant, On a Supposed Right to Lie from Philanthropy in Immanuel Kant: Practical Philosophy 605 (Paul Guyer & Alan W. Wood eds., Mary Gregor trans., Cambridge Univ. Press, 1996). Part of the explanation for this may be Kant's concern about the moral wellbeing of the liar. Indeed, his discussion in the Tugendlehre on lying (supranote 2 at 183–83 (6:429–30)) falls under the general heading of “a human being's duty to himself merely as a moral being.” Id. at 182 (6:428) (capitalization altered). By lying, even about small matters, one opens oneself to the possibility of self-deception. But proper moral judgment requires, above all, honesty with oneself, for moral cognition cannot function properly in the absence of such truth. It might be possible to agree with this general sentiment without taking the seemingly extreme position that Kant does with respect to lying to the murderer. But there is undoubtedly something right, even noble, about his unwavering commitment to the truth.

So far, I have argued that Kant intends the duty of self-knowledge or moral cognition to temper or refine the judgment we pass on ourselves through our conscience, based on the totality of our moral life circumstances. This is not the usual interpretation of Kant's view of conscience, which is commonly characterized as being “far from a gentle whisper of moral encouragement. It places us on trial for (perceived) moral failings, accuses us, passes sentence, and makes us suffer.”

Hill, supranote 18, at 408.

My point in this section has been to show that, while Kant's conception of conscience can seem just as unyielding as his discussion of punishment is sometimes taken to be, this is at best a superficial reading. While self-judgment is an important to our moral life, so is the duty of self-knowledge or moral cognition. Only when we attend to both of these duties do we treat ourselves in a way that is respectful both of our moral agency, but also of our position as imperfect human beings subject to many influences other than the self-legislated moral law. Assuming this interpretation is correct, we might then ask whether a Kantian should countenance some analogue to self-knowledge or moral cognition in our relationships with others—a proposition for which I argue in the following section.

Kant does not discuss the capacity for moral cognition as applicable beyond the duty of self-knowledge. I think, however, that reflecting on the ways that we interact with other people will convince us that moral cognition is, in fact, a natural and essential part of social life—and that Kant would himself willingly endorse such an extension of this principle. Consider, for example, the types of judgments we must regularly make about the motives behind someone's actions. For example, imagine that a friend fails to follow through on a promise, or that a child misbehaves, or that a spouse deceives. What are we to do under such circumstances?

I have chosen negative actions because, as should be obvious, I will ultimately be drawing an analogy with criminal actions. Still, this is clearly the case with positive actions as well. We must determine how to respond appropriately to praise, gifts, and other indicia of social approbation, esteem, or love.

I should think it clear that our response in such situations is highly dependent on the motives of the actor and background conditions of the action. If the friend is ill, or the child very young, or the spouse under intense stress, then we are likely to react in different—specifically, more merciful—ways than if the friend turns out to be selfish, the child old enough to know better, or the spouse systematically dishonest. But how do we make such determinations? Quite naturally, and without always being aware of it, we engage in moral cognition.

For example, suppose that Joan's friend, Kevin, promises to watch her children one afternoon so that she can go to a job interview. Kevin does not show up, and fails to answer her phone calls. Joan, having relied on Kevin's promise, cannot find anyone else to watch her children on such short notice, and so she misses the job interview. This is a significant setback for her, since she has been unemployed for months, this job would have been ideal for her, and the interview cannot be rescheduled.

How should Joan react in this situation? That is, what should she do with respect to Kevin and their friendship? She has many options, such as telling Kevin what a great friend he is, pretending that nothing happened, writing an angry letter to him, refusing to ever talk to him again, explaining to him that she is upset, or hiring someone to kill him. Some of these options will be morally inappropriate due to their nature: killing someone without a sufficiently good reason is obviously immoral, and obsequiously praising someone who has harmed us may be as well.

See Kant's discussion of “servility,” supranote 2, at 186–88 (6:434–37).

What Joan ought to do is, of course, try to learn more. Ideally, Joan will determine the facts of the situation with “impartiality” and “sincerity” in order to determine Kevin's “moral worth” with respect to this incident.

Id. at 191 (6:441).

One might wonder whether expecting Joan to “morally cognize” Kevin in such a situation is unrealistic. Given the harm caused to her in this case, can we reasonably expect Joan to react in such a rational manner, when her initial inclinations will likely be anger at Kevin? Kant's moral theory is again helpful here. He proposes that personal virtue involves the “capacity and considered resolve to withstand a strong but unjust opponent.”

Id. at 146 (6:380). See also the discussion of virtue, id. at 156–57 (6:394–95).

In reality, of course, engaging in moral cognition will be particularly difficult when we have been seriously wronged by others. In some cases it may simply be psychologically impossible under the circumstances, and we must be wary of judging Joan if she fails to fulfill this moral duty in this case. Still, it seems reasonable to assert that the use of moral cognition in social circumstances such as Joan's is the proper moral ideal, however difficult it may be to attain in practice.

A more significant worry here, though, is that the purpose of moral cognition—proper judgment of others—is misguided. While we might be tempted to pass judgment on other people, and while we might in fact do this regularly, we really should not do so. One might appeal to religious aphorisms, such as “judge not, that ye be not judged” in support of this proposition.

Matthew 7:1.

To see this, take a more serious case. Suppose that Vivian discovers that Wayne, her spouse, has had an affair; furthermore, Wayne refuses to admit to the affair, apologize, or even discuss the matter with Vivian. After engaging in the process of moral cognition, Vivian determines that the appropriate response is to seek a divorce. Vivian might reasonably say that she still cares about Wayne, wants the best for him, and respects him as a fellow human being; she has, however, determined that divorce is the morally appropriate response to Wayne's actions. Admittedly, maintaining such a positive attitude toward Wayne might be difficult, but the point is simply that Vivian's “judgment” of the appropriate response to Wayne need not entail a condemnation of Wayne. Moreover, this type of judgment is fully compatible with the notion that making an ultimate determination about whether Wayne is, all things considered, a “good” person or not is not one that mere mortals are equipped to make. What Vivian can, and should, make a judgment about is what her relationship to Wayne should be, and what response his actions and motives merit.

Finally, as a practical matter, it is hard to see how we can avoid making these kinds of judgments, nor would it be healthy in many cases to do so. We cannot, and should not, expect Joan, much less Vivian, to simply go about their lives as if nothing at all had happened. This may be possible (and desirable) in cases of very minor social conflicts, as when an inconsiderate driver cuts one off in traffic. But in cases where we suffer a cognizable harm at the hands of those we associate with, then we must determine what response is appropriate under the circumstances—and failing to do so, at least consistently, amounts to a failure to respect oneself.

Kant memorably declares that “[b]owing and scraping before a human being seems in any case to be unworthy of a human being. . . . [And] one who makes himself a worm cannot complain afterwards if people step on him.” Id. at 188 (6:437).

So far, I have explained Kant's view that humans are capable of engaging in moral cognition after condemning themselves by the operation of their consciences—and that doing so is morally required in order to respect themselves as moral agents. I then proposed that we can and should conceive of moral cognition as possible and desirable in our interactions with other people, particularly in cases where others wrong us. In this section, I suggest that moral cognition is required of good Kantian citizens who act as decision-makers in the area of criminal punishment. My comments in this section will be mostly general, and are intended to motivate the proposition that moral cognition makes sense to discuss in this context; I turn to more specific proposals in subsection D. The main goal here is to show that moral cognition does better than lex talionis alone as a principle guiding the nature and extent of punishment.

As in the previous subsection, this analysis is intended as an extension of Kant's thought, rather than a direct interpretation of it. Indeed, Kant limits his discussion of moral cognition to the context of self-knowledge. My contention, however, is that introducing this notion into the criminal justice system will be in keeping with Kant's more general commitments to justice and morality.

Moral cognition in the realm of criminal punishment may be fruitfully compared to moral cognition in the two circumstances we have already covered:

(1) In its mode of self-knowledge, moral cognition complements the operation of the conscience. Fulfilling our duties of conscience and self-knowledge are both required in order to properly respect ourselves as human beings. When we submit to the judgment of our conscience, we respect ourselves as moral agents capable of choosing in accordance with the demands of morality. When we introspectively seek self-knowledge, however, we respect ourselves as mortal beings subject to factors external to our will. Both conscience and self-knowledge are required in order to properly fulfill our duties of personal virtue with respect to ourselves.

(2) In its mode of making social judgments, moral cognition complements the operation of practical reason. Respecting other human beings requires (in certain cases) passing judgment on their actions—in doing so, we respect them as moral agents. Yet respecting others also requires cognition of their circumstances, including the most general circumstance of being subject to the conditions of mortality. A judgment tempered by moral cognition is required in order to fulfill duties of personal virtue with respect to other people.

Moral cognition in criminal punishment is both similar and different from moral cognition in the areas above. For one thing, moral cognition of the self is required by all competent moral agents—it is an inescapable duty of beings that have the capacity for moral agency. Moral cognition of other people, while not strictly necessitated by virtue of being a moral agent, is inescapable as a practical matter due to the social nature of human beings. Judgments made through the operation of the criminal law, however, are rarer. Human beings are not called upon to make such judgments except when required to serve as jurors or, perhaps, when making decisions about what kinds of criminal legislation to support.

Nevertheless, while the circumstances in which the moral cognition of criminals is relevant will be more limited than the previous categories, the duty will look quite similar:

(3) In its mode of judging convicted criminals, moral cognition moderates the binary judgment of the criminal law. Respecting people who have committed crimes requires punishing them, for this treats them as moral agents who could have chosen not to violate others’ civic freedom. But it also requires attending to the circumstances and background conditions that contributed to the act in question. Both administering punishment and engaging in moral cognition of offenders are therefore required in order to fulfill our duties of virtue as citizens.

Virtuous Kantian citizens will, then, support policies and procedures which impose reasonable punishments that attend both to the nature of the criminal act (via lex talionis), but also to the motives and background of the criminal (via moral cognition).

It is worth thinking about the special difficulties citizens engaging in this kind of moral cognition will face. It will often require citizens to set aside their prejudices in order to reason about the needs of their community in pursuit of the ideal of justice. Most of us have no doubt experienced visceral negative responses in the face of serious criminal acts—either directed at us or at fellow human beings. Some theorists have argued that these sentiments are themselves indicia of the direction that criminal justice ought to take.

See, e.g., Michael Moore, The Moral Worth of Retribution, in Punishment and Rehabilitation 94 (Jeffrie Murphy ed., 1995); and Jeffrie Murphy, Getting Even: The Role of the Victim, id. at 132.

Moreover, despite his reputation as a retributivist, Kant believes that we have the duty to “[d]o good to other human beings insofar as we can . . . whether [we] love[] them or not . . . [and] even toward a misanthropist.”

Kant, supranote 2, at 162 (6:402).