Introduction
There is a consensus in contemporary legal canons about torture: it is prohibited absolutely. Article 2 of the United Nations Convention Against Torture and Other Cruel, Inhuman or Degrading Treatment or Punishment declares that ‘No exceptional circumstances whatsoever, whether a state of war or a threat of war, internal political instability or any other public emergency, may be invoked as a justification of torture’. Article 5 of the Universal Declaration of Human Rights asserts unwaveringly that ‘No one shall be subjected to torture or to cruel, inhuman or degrading treatment or punishment’ – the only enumerated right left unqualified.Footnote 1 Article 3 of the European Convention on Human Rights demands that ‘No one shall be subjected to torture or to inhuman or degrading treatment or punishment’. Title 18 § 2340A of the United States Code categorically prohibits torture and the State Department of the United States has made it clear that ‘No exceptional circumstances may be invoked as a justification for torture’.Footnote 2 Even a nation that has publicly acknowledged its use of torture declares that international prohibitions against torture ‘are absolute’, that ‘There are no exceptions to them and there is no room for balancing’.Footnote 3 But whatever the legal status of torture is, what it ought to be is in dispute. American law professor Alan Dershowitz argues that torture is inevitable and, as such, it would be preferable if its practice were regulated and constrained by legal mechanisms. To that end, Dershowitz defends the ‘torture warrant’: a legal mechanism akin to a search warrant that authorizes the performance of non-lethal interrogative torture on a subject.Footnote 4 But legal officials cannot very well authorize what is illegal, so instituting torture warrants requires legalizing torture.
The novelty of Dershowitz's position should be emphasized: while Dershowitz argues that torture should be legally permitted, he still declares that torture is wrong and absolutely so.Footnote 5 A more typical view holds that torture is sometimes morally permissible but should be legally prohibited.Footnote 6 Another position combines moral and legal absolutism, holding that torture is always morally wrong and should be illegal.Footnote 7 Yet another position rejects both moral and legal absolutism.Footnote 8
I propose to consider what light Kantian ethics can shed on the moral and legal status of torture. The overwhelming consensus among commentators is that Kantians are committed to moral absolutism about torture; one commentator contends that Kantians and only Kantians must be moral absolutists about torture.Footnote 9 I find this consensus surprising for two reasons. First, while Kant has much to say about lying and deception and nonbeneficenceFootnote 10 he has little to say about our darker, violent actions.Footnote 11 In one memorable passage, he insists that some kinds of punitive torture are off limits:
…there can be no disgraceful punishments that dishonor humanity itself (such as quartering a man, having him torn by dogs, cutting off his nose and ears). Not only are such punishments more painful than loss of possessions and life to one who loves honor… they also make a spectator blush with shame at belonging to the species that can be treated this way (MM, 463).
Generally, death sentences must be free from any mistreatment that would make the humanity of the person suffering them ‘something abominable’ (MM, 333). But a Kantian case against torture cannot rest on the claim that torture is more painful then the loss of possessions or life; some people might be more pained by the loss of their treasured irreplaceable possessions than by a brief period of non-lethal torture. The Kantian case similarly cannot rest on torture's tendency to make spectators blush since torturous punishments need not be public spectacles.Footnote 12 In any case, Kant only prohibits a few select varieties of torture in this passage, not torture generally.
Second, whatever Kant says, I doubt that Kantians can defend even a minimalist version of moral absolutism about torture. Still, I doubt that Kantians should be enthusiastic proponents of torture; there is a Kantian case against torture. I close by explaining just what the Kantian case against torture is.
The Many Moral Absolutisms
Typically, moral absolutism implies that some action is prohibited although an absolutist could require, rather than forbid, its performance.Footnote 13 This typical sort of moral absolutism is exemplified by Thomas Nagel's characterization of pacifism as the view that ‘one may not kill another person under any circumstances, no matter what good would be achieved or evil averted thereby’.Footnote 14 This prohibition should not be understood as the material claim that, in the actual world, killing another person is morally impermissible since the pacifist prohibits killing another person ‘under any circumstances’ – that is, in all possible worlds. Pacifism, so understood, entails the modal claim that killing another person is always wrong no matter the circumstances – that is, in all possible worlds. An equally robust absolutism about torture would similarly imply that torture is wrong in all possible worlds.
Some philosophers who self-identify as absolutists about torture defend less robust versions:
There are different kinds of absolutists. Some hold that torture is always wrong in this or any other possible world; some, that torture is always wrong in this world, never mind the others; some that torture, in this world, is wrong in every realistically imaginable case, and could at most be permitted in cases so infinitesimally unlikely that they are not worth talking about. Absolutists are united in opposing the view that torture could be permitted in any realistically imaginable case.Footnote 15
Only this first thesis yields the robust result that torture is always morally wrong, no matter the circumstances.Footnote 16 I have my doubts that anything less than this first robust thesis is absolutism worth the name, but no matter. For present purposes, I understand moral absolutism about torture to be the comparatively minimal view that torture is not morally permissible in any ‘realistically imaginable case’. If Kantian ethics cannot support this minimalist version of absolutism, it cannot support a more robust version either.
Sometimes philosophers who argue for the moral permissibility of torture rest content with invoking the ubiquitous ticking-bomb scenario to demonstrate that interrogational torture is morally permissible.Footnote 17 But interrogational torture is not the only variety of torture whose permissibility needs to be considered. Some philosophers contend that punitive torture in response to grave wrongdoing is sometimes morally permissible.Footnote 18 Other varieties of torture, including terroristic torture in which torture is used to intimidate its victim or others for purposes not limited to deterring dissent or scaring off rivals, are probably not worth considering presently. Since terroristic torture is an especially transparent case of violating the Kantian prohibition against using people merely as a means, there is little utility in considering its permissibility on Kantian grounds.Footnote 19
As a heuristic, I suggest that the Kantian limit her attention to cases of interrogational and punitive torture: if there are no realistically imaginable cases in which interrogational or punitive torture is morally permissible on Kantian grounds, then surely torture is always morally wrong on Kantian grounds. How would a Kantian go about showing that there are no realistically imaginable cases in which interrogational or punitive torture is morally permissible? There are two broad strategies that Kantians have tended to take: the first and decidedly more popular strategy appeals to the humanity formulation of the Categorical Imperative, the second appeals to the Kantian prohibition against cruelty. This first strategy admits of several varieties and deserves attention first.
Strategy #1: The Humanity Formulation of the Categorical Imperative
According to the second formulation of the Categorical Imperative – the humanity formulation – rational agents must ‘Act in such a way that you always treat humanity, whether in your own person or in the person of any other, never simply as a means, but always at the same time as an end’ (GR, 429). The humanity formulation is a high point for Kantian ethics and any number of scholars contend that it prohibits torture absolutely.Footnote 20
By Kant's own lights, the various formulations of the Categorical Imperative are equivalent, ‘merely so many formulations of precisely the same law’ (GR, 436). If so, then should the humanity formulation fail to secure the absolutist case, no appeal to some other formulation of the Categorical Imperative – like the formula of universal law or the kingdom of ends formula – should be expected to bear fruit. Of course, Kantians might disagree with Kant and deny the equivalence of the various formulations.Footnote 21 I will have more to say about this strategy below. For now, I suppose that if the humanity formulation does not yield absolutist opposition to torture, neither will any other formulation of the Categorical Imperative.
While appealing to the humanity formulation is the dominant Kantian strategy for securing an absolutist case against torture, that strategy is multiform. Indeed, there are at least five different versions of it.
1. The Phenomenology of Torture
One sense in which torturers seem to violate the humanity of their victims is well illustrated by Elaine Scarry's discussion of the phenomenology of torture.Footnote 22 Scarry vividly describes the sense in which the physical pain of torture can ‘obliterate the contents of consciousness’.Footnote 23 A person's body is used against him and used to ‘destroy the self’,Footnote 24 such that the ‘self [is]… lost, or nearly lost, through the intense pain of torture.Footnote 25 In a similar vein, consider Jeffrie Murphy's condemnation:
Sending painful voltage through a man's testicles to which electrodes have been attached, or boiling him in oil, or eviscerating him, or gouging out his eyes – these are not ways of relating to another person. [The victim] could not be expected to understand this while it goes on, have a view about it, enter into discourse about it, or conduct any other characteristically human activities during the process – a process whose very point is to reduce him to a terrified, defecating, urinating, screaming animal… We have here a paradigm of not treating a person as a person – and thus an undermining of that very value (autonomous human personhood) on which any conception of justice must rest.Footnote 26
This is an intuitive way of capturing at least part of what is morally offensive about torture, but it arguably proves too little. Any number of actions similarly ‘obliterate the contents of consciousness’ – say, inducing a coma or administering euphoria-inducing amounts of morphine – but are not necessarily wrong. Obviously, inducing coma or euphoria will fail to produce the most salient wrong-making feature of torture: the experience of intense, blinding pain. Yet even some proponents of the absolutist case have denied that the distinctive wrong-making feature of torture is its tendency to cause such pain.Footnote 27
This suggests a general problem for Kantian absolutists. The humanity formulation entails that lots of actions are wrong but wrong in the same way and for the same reason: both torture and robbery, for example, are wrong because they involve treating another person merely as a means. Yet it is commonly thought that torture is especially wrong – that it is wrong in a way and to a degree that robbery is not – and the appeal to the humanity formulation dissolves this difference between torture and other kinds of wrong action.Footnote 28 So, the Kantian who wishes to advance an absolutist case against torture must proceed carefully: she must both show that there are no realistically imaginable cases of morally permissible torture while also showing what it is about torture that makes it especially and uniquely wrong. Can the Kantian have it both ways?
2. ‘Shocking’ Treatment and Inhumanity
A Kantian argument offered by Michael Davis seems to preserve the intuition that torture is especially wrong. If the humanity formulation prohibits anything, it prohibits ‘shocking’ treatment – that is, treatment that most of us would be unwilling or unable to visit upon non-human animals, much less other human beings – and torture is shocking if anything is.Footnote 29 At least, very many of us would be unwilling or unable to torture other human beings and non-human animals while many of us are willing and able to rob other human beings, especially in desperate circumstances. And that suggests a morally relevant difference between torture and robbery: only the former is ‘shocking’.
While the appeal to the phenomenology of torture proves too little, this current argument proves too much: some morally permissible behavior involves treatment of others that many of us would be unwilling or unable to visit upon our fellow human beings and non-human animals. Many of us, I suspect, are unable or unwilling to help our friends engage in extreme varieties of body modification like tongue-splitting and branding. Some individuals enjoy ‘suspension’, a practice in which an individual is pierced with several large gauge hooks and then lifted in the air or swung about, but the squeamish of us are probably unwilling if not unable to assist. Even if extreme body modification is ‘shocking’ it is far from clear that it is morally wrong, much less that morality absolutely prohibits it.
Of course, tongue-splitting and suspension presumably proceed only with someone's consent and that surely matters. Perhaps torture is shocking, but perhaps it is also inhumane. Being shocking is necessary for action to be inhumane, but not sufficient: inhumane treatment must also be against the victim's will and without benefit to her.Footnote 30 Davis has little doubt that torture is shocking; ‘Nothing’, we are told, ‘could be more universally shocking than what goes on inside the torture chamber’.Footnote 31 And paradigmatic cases of torture are performed neither with the victim's consent nor for her own good. So understood, torture, but not tongue-splitting, is inhumane, exactly the sort of thing the humanity formulation should prohibit absolutely.
To be sure, we are stipulating the meaning of ‘inhumane’ here, but so defined inhumanity is pretty clearly unacceptable on Kantian grounds. The question is whether torture is always inhumane. Several paradigmatic examples of torture certainly qualify as inhumane, but as a conceptual matter it is far from clear that every instance of what is rightly regarded as torture must lack any benefit for the victim and must proceed without her consent.
Suggesting that torture might benefit its victim is, well, ‘shocking’. One might point to scenes from Stanley Kubrick's A Clockwork Orange in which something approximating torture is used to correct the anti-social tendencies of the movie's protagonistFootnote 32 but the prospects of using torture as therapy, with all of its tendencies to cause deep and long-lasting physical and psychological harm, are surely dim. Still, torture is sometimes conceived of as beneficial for its victims. Consider how the Catholic Church apparently regarded its use:
…the fasts, vigils, scourging, and every other form of voluntary self-torture that the mystics inflicted on their own persons were turned against the heretics as an instrument of the law. The Church… imagined the Holy Office as an arm of penance, spiritual reconciliation, and the restoration of the wayward back to the community of the faithful. Its tortures were modeled after ascetic practices, and even the burning of the heretic at the stake was based on readings of the scripture, and on the penitential model of purifying fires as a purgatorial blessing.Footnote 33
I doubt this passage settles anything, but it prompts a number of questions worth considering. First, insofar as torture is used as a mechanism for penance and self-punishment, are there any theories of punishment that could justify it? The Church's rationale is at least consistent with a moral education theory of punishment that holds that the goal of punishment is to teach the wrongdoer that what she did is morally wrong and should not be performed for that reason.Footnote 34 Admittedly, if there are going to be cases in which torture is conducive for the education of its victims they will be ‘exceptional cases’,Footnote 35 but if punitive torture is punishment then perhaps it too can educate wrongdoers and show them just how wrong their actions are. Kant himself allows that punishment ‘can serve indirectly as a means of moral training’ and sometimes uses the language of habituation expected from moral education theorists (LE, 287). Since punishment is supposed to benefit the punished on the moral education theory, there is a Kantian argument that torture can benefit its victim. If so, torture need not be inhumane.
Note also the reference to ‘voluntary self-torture’ above. Is voluntary self-torture possible? More generally, can torture proceed with a victim's consent?
3. Torture and Consent
If prospective victims of torture can consent to being treated as such, there is a Kantian argument that their torture is morally permissible. For example, if an agent consents to be tortured in order to pursue some morally permissible end, then, supposing her maxim survives familiar Kantian tests for legitimacy, respect for her rational nature arguably calls for honoring her maxim. Some care is needed in constructing plausible examples of willing one's own torture in the pursuit of an end, however. There is a sense in which someone who undergoes interrogational torture consents, since they presumably would not be tortured if they revealed the desired information. And perhaps there is a sense in which someone consents if she bravely chooses to undergo torture herself to spare her children or loved ones from it. But in both cases, the victim of torture is coerced to an exceptional degree, so exceptional that it probably undermines consent. There is such a thing as ‘act-impelling torture’ which induces a victim to perform some action where the action is undertaken to gain relief from suffering.Footnote 36 But insofar as the infliction of great suffering occurs prior to the victim's decision to perform the requisite action and insofar as the suffering terminates only after the performance of an action, the victim's decision and action are performed while being seriously coerced. So act-impelling torture need not occur with the victim's consent.
That said, coercion-free scenarios in which individuals genuinely consent to torture do seem possible. Couldn't some someone consent to brief non-lethal bouts of torture in the hopes of being richly compensated by a wealthy sadist or a government agency studying torture's effects? Suppose too that once the offer of being tortured for compensation has been fully embraced and the torture has begun, its victim has no resources for bringing his abuse to an end prior to the agreed upon terminus; he cannot withdraw and he cannot bring his abuse to a halt. Unlike instances of act-impelling torture, the willing participant would not be forced to decide what to do while being coerced, and since he cannot bring his abuse to a halt once it has begun there is no sense in which he being coerced into soldering on. Some Kantians will find the prospect of trading torture for profit decidedly un-Kantian given Kant's declaration that our humanity is ‘infinitely above all price’ (GR, 435). But some members of the military consent to waterboarding and other ‘shocking’ treatment as part of their training. Since they have good reason to worry that they will be exposed to such treatment if captured, they have good moral reason to consent to such treatment to help ensure that they do not violate any moral duties. Recall that while Kant opposes suicide (GR, 422 and 429) he allows that ‘If I cannot preserve my life except by violating my duties towards myself, I am bound to sacrifice my life rather than violate these duties’ (LE, 372). A faithful Kantian, worried that she might either lie or betray her cause and comrades if waterboarded, may well consent to such treatment to build up her defenses or learn some coping mechanisms. If we are morally obligated to give up our life to ensure that we do not violate our duties, surely we are obligated to consent to waterboarding to ensure that we do not violate our duties. But that suggests that consenting to torture need not require trading away one's humanity.
Consent is relevant to a different sort of argument about the permissibility of torture. Arguably, Kant draws a close connection between violating someone's rights and treating them merely as a means.Footnote 37 Kant insists that ‘it is manifest that a violator of the rights of man intends to use the person of others merely as a means’ (GR, 429). But on some accounts, a terrorist who threatens the life of others is not necessarily wronged if he is tortured because he has abandoned or forfeited his right to be free from torture insofar as he willingly threatens the humanity of others. Arguably, he implicitly consents to be tortured.Footnote 38 Exactly how this argument should be spelled out is an open question, and I don't mean to endorse it, but its plausibility turns on a disputed issue about the irrevocability of human rights. On one conception, being a human being is sufficient to come to have said rights. On another conception, being a human being is sufficient to have and continue to have said rights.Footnote 39 Some rights, like the right not to be killed or the right to free movement, admit of this first reading but not the second since they can plausibly be abandoned or forfeit. If the right to be free from torture admits of the first reading then consenting to torture might be possible since that right can be lost or forfeit. But if the right to be free from torture admits of the second reading, it is inalienable and can be neither lost nor forfeit. I have no idea how to determine whether the right to be from torture should be given this first or second reading. But Kant himself only acknowledges the existence of one innate human right: ‘Freedom (independence from being constrained by another's choice), insofar as it can coexist with the freedom of every other in accordance with a universal law, is the only original right belonging to every man by virtue of his humanity’ (MM, 237).
The Kantian absolutist could take a different tack. She could allow that ‘If [torture] involved the consent of the tortured, it would likely not be wrong at all’ while insisting that torture is just not the sort of thing that we can consent to.Footnote 40 So understood, torture is akin to another sort of atrocity often regarded as absolutely wrong and wrong in a way and to a degree that other morally wrong actions are not. Typically, legal and philosophical conceptions of rape suppose that, necessarily, the victim of rape does not consent to her assault. If, necessarily, the victim of torture does not consent to his assault, then whatever the perpetrator of torturous treatment is guilty of he is not guilty of torture. So understood, the absence of consent is essential to the very nature of torture.
If it is impossible to consent to torture, then there is another Kantian argument for absolutism available. On a ‘possible consent interpretation’ of the humanity formulation, one treats another person merely as a means if one treats them in a way to which they could not possibly consent to being treated.Footnote 41 If no one could possibly consent to being tortured, then it is impossible to will a maxim calling for torture consistent with the humanity formulation.
Still, there are some reasons for skepticism about appealing to consent to secure a Kantian absolutist case against torture. While some of our best Kantian commentators endorse the possible consent interpretation, it has been disputed.Footnote 42 Consent does not clearly justify just anything on Kantian grounds: If performing an action or permitting something to happen would violate some other duty – say, an imperfect duty to develop one's talents or the perfect duty to preserve one's life as a dignified being – then performing that action is still wrong even if consent is given. And while Kant seems to endorse the maxim volenti non fit injuria – ‘to one who has consented no wrong is done’ (MM, 457) – he also insists that ‘[M]an is not his own property, and cannot do as he pleases with his body’ (LE, 157). As such, it is simply unclear whether appealing to consent is the best way of giving content to the humanity formulation.
A further reason to doubt that appealing to consent can secure the Kantian's absolutist case involves the epistemology of torture: there are some cases in which we can be certain that someone has been tortured without knowing whether they failed to consent. Learning that someone has been electrocuted for hours, drawn and quartered, or broken on the rack is enough to infer that he has been tortured, period. There is no need to also determine that he never consented to being treated as such. Quite generally, there are some especially ghoulish treatments and techniques such that count as torture whatever the context. But if torture is essentially tied up with the absence of consent, we could never be confident that this was so; whether some act really is an instance of torture would always be an open question contingent upon establishing that the victim did not consent. Nothing that we could do, no matter how cruel or malicious or painful, would count as torture per se. That result seems to me profoundly wrong.
Further, this Kantian strategy implies that lots of comparatively trivial actions are wrong in the same way and for the same reason that torture is; it is equally essential to robbery that it occurs without the victim's consent, but I find it plausible to suppose that torture is wrong in a way and to a degree that robbery is not.
My conclusion, then, is that Kantian concerns about consent will not secure the absolutist case against torture. Yet concerns about consent are not going away, especially given some conceptions of torture's purpose: ‘to break the will’ of its victim.
4. Breaking the Will
Many acts Kant condemns as immoral are autonomy violations: making a lying promise, for example. But arguably, torture does not merely violate autonomy; it destroys it. Arguably, torture breaks the will.Footnote 43 Talk of ‘breaking wills’ is metaphorical, but we probably understand what it literally amounts to: torture can cause us to betray loved ones and comrades but also ourselves. David Sussman's well-received suggestion that torture forces its victim into a position of colluding against himself through his own affects and emotions so that he experiences his own complicity and powerlessness, a pre-eminent kind of self-betrayal, gives content to the thought that torture breaks the will.Footnote 44 Tying torture to the breaking of wills also tracks the reports of victims of torture.Footnote 45 If the Kantian identifies the will with humanity, then insofar as torture breaks wills it destroys humanity. So, if destroying humanity is absolutely prohibited by Kantian ethics, so is torture.
Evaluating this argument depends upon getting the complaint right. Some kinds of temporary will-breaking are not clearly wrong; if the only way to get some fanatical cult member to abandon his allegiance to a manifestly unjust cause is to temporarily break his will, perhaps we are justified in doing so, especially if we go on to rehabilitate and counsel him. Really, absolutist opponents of torture who complain that torture breaks wills complain that torture tends to break wills permanently – that ‘Torture is ineradicably burned into him, even when no clinically objective traces can be detected’.Footnote 46
This argument too captures an awful lot of what is usually reprehensible about torture but I doubt that it secures the absolutist case. One worry is that Kant himself does not seem to suppose that destroying humanity is always wrong; if he did, we could not explain why Kant thinks that sacrificing one's own life or killing in self-defense or capital punishment are sometimes morally permissible. Further, as a conceptual point, torture need not permanently break the will of its victim nor must a torturer aim to break his victim's will. An especially perverse kind of sadist might enjoy viewing the autonomous struggles of a victim with his will intact. As an empirical point, there do seem to be victims of torture whose wills were not broken: South African anti-apartheid activist Stephen Biko fought his torturers before sustaining injuries that led to his death, for example.
Interestingly, Wisnewski agrees that torture need not permanently break wills, but insists that tells us nothing about the nature of torture. For even if torture need not actually break a will, it remains true that:
The aim of torture is to destroy the agent – this is not merely a foreseeable but unintended consequence… To point out that torture sometimes fails to achieve the goal postulated in torture manuals, however, is irrelevant to determining what torture aims to do.Footnote 47
This is a notable change in the current argument: what is especially heinous about torture is no longer what it does to its victims but what perpetrators of torture aim to do. But it is far from clear that every perpetrator of torture aims to destroy his victim; recall the sadist who only enjoys watching autonomous struggle against his torturous methods. It is also not beyond hope that some actual practitioners of interrogational torture aim only to extract information and avoid breaking wills as much as they can, that the suffering of their victim is a foreseeable but unintended consequence, and so forth. It is certainly unlikely that military officers tasked with teaching soldiers to withstand torture aim to irrevocably break the will of their trainees.
In sum, torture need not actually break wills and torturers need not aim to break wills. This fourth appeal to the humanity formulation also fails to secure an absolutist case on Kantian grounds.
5. The Prohibition Against Using Someone Merely as a Means
A final strategy appeals more directly to the humanity formulation. Perhaps what is really wrong with torture is that it is an ‘extreme instance of using someone as a mere means to purposes she does not or could not reasonably share’.Footnote 48 This argument correlates well with the thought, popular among contemporary Kantians, that the moral permissibility of an action turns on the possibility of justifying it to others on grounds that they themselves could reasonably accept.Footnote 49 The claim that the victim of torture is being used merely as a means just is, on this view, the claim that there is no justification of her treatment that she could reasonably accept.
Is torture ever justifiable to someone on grounds that she could reasonably accept? Frances Kamm constructs a hypothetical case in which it is supposed to be. We are to imagine that A is about to set off a bomb to kill innocent and non-threatening B, that A will succeed unless he is stopped, and that there are only two ways to stop A: kill him or torture him for an hour in a way that incapacitates him, making it impossible for him to set off the bomb. Kamm argues that torture may well be permissible in this case, partly because A ‘would be tortured for purposes he could reasonably be expected to share, that is, his not being killed’.Footnote 50 But I find it strained to say that sparing him from death is part of our purpose in torturing him; more plausibly, we decline to kill A because there is some side-constraint prohibiting killing an attacker if other available means are sufficient to repel his attack. Generally, acting in light of some side-constraint does not entail acting for the corresponding purpose of avoiding what is prohibited by that side-constraint; my purpose in buying Kamm's newest book is to read some first-rate philosophy, not to refrain from stealing.
Still, Kant's discussion of legal punishment suggests an argument that adopting and executing a policy permitting torture need not treat anyone merely as a means. I have in mind not Kant's remarks about ‘the law of retribution’ (MM, 363) but his complicated discussion of capital punishment.Footnote 51 Suppose that a criminal who has a fair share of what justice requires lives in a state that needs a system of harsh punishment to secure liberty for its citizens. Suppose also that capital punishment is actually accepted by citizens of that state and used, such that it is public knowledge that certain criminals will be executed. Suppose finally that the criminal has the opportunity to avoid being punished. Arguably, condemned criminals cannot complain that they are being treated merely as a means, even if the point of executing them is to deter future crime just because the policy permitting capital punishment can be justified to them on grounds that they can reasonably accept; they too have an interest in deterring violence and know that refrain from committing capital crimes is sufficient to avoid execution.Footnote 52 But then they cannot complain of being treated merely as a means.
Suppose similarly that punitive torture is necessary for deterring some especially heinous criminal behavior and to secure liberty for citizens, that it is public knowledge that certain criminals will be tortured, that criminals have the opportunity to avoid being tortured, and so forth. Arguably, in such conditions, criminals cannot complain that they are being treated merely as a means when they are tortured, even if the point of torture is to deter future crime. Their humanity too is respected since a policy permitting punitive torture can be justified to them on grounds that they can reasonably accept: they too have an interest in deterring violence and know that refrain from committing capital crimes is sufficient to avoid torture. But then they cannot complain of being treated merely as a means.
Alternatively, suppose that interrogational torture is necessary for securing liberty for citizens, that it is public knowledge that certain criminals will be tortured, that criminals have the opportunity to avoid being tortured, and so forth. By analogy, in such conditions, criminals cannot complain that they are being treated merely as a means when they are tortured, even if the point of torture is to secure liberty. Their humanity too is respected since a policy permitting interrogational torture can be justified to them on grounds that they can reasonably accept: they too have an interest in securing liberty and know that refrain from committing grave injustices is sufficient to avoid torture. But then they cannot complain of being treated merely as a means.
Admittedly, whether or not either punitive or interrogational torture is necessary in the actual world to deter crime or secure liberty is an open question; I am confident that punitive torture is not necessary and deeply skeptical that interrogational torture is. But there are realistically imaginable cases in which they are. Accordingly, adopting and executing a policy permitting either interrogational or punitive torture need not amount to using someone as a mere means to purposes she does not or could not reasonably share. But that means that this fifth appeal to the humanity formulation also does not succeed.
So, while it is commonly thought that the humanity formulation of the Categorical Imperative functions as a bulwark against attempts to show that torture is sometimes morally permissible, I contend that no appeal to it will secure the absolutist case. And if the humanity formulation is the most attractive and intuitive component of Kantian ethics, there has to be real doubt that the absolutist case can be secured on Kantian grounds. The Kantian absolutist about torture might try to avoid this result by denying the equivalence of the various formulations of the Categorical Imperative; even if the humanity formulation does yield absolutism about torture, some other formulation might. It is difficult to assess an argument that hasn't actually been advanced; I'm aware of no one who has argued that a maxim calling for torture will run afoul of the kingdom of ends formulation, for example. And whatever is wrong with willing a maxim calling for interrogational or punitive torture, it seems unlikely that it must generate a contradiction in conception.
Worse, denying the equivalence of the various formulations of the Categorical Imperative might be part of an argument that torture is, at least sometimes, morally permissible. Consider Christine Korsgaard ‘double-level theory’. Korsgaard's well-known position is that the various formulations of the Categorical Imperative are not equivalent and that the humanity formula and the kingdom of ends formula suggest ideals that may be impossible to live up to in non-ideal circumstances.Footnote 53 In particular, if living up to such an ideal ‘would make you a tool of evil, you should not do so’.Footnote 54 But even in non-ideal circumstances, the formula of universal law is still available to guide our conduct, if only by telling us what we absolutely may not do.Footnote 55 But in any case in which we would even contemplate the use of punitive or interrogational torture, it is likely that we are working in non-ideal circumstances in which we may well be permitted to depart from the ideal advanced by the formula of humanity in favor of the less strict formula of universal law. But if the stricter formula of humanity doesn't necessarily rule out torture as morally impermissible, it is hard to see why a more lax formulation that governs non-ideal circumstances must rule it out.
I have argued that the humanity formulation of the Categorical Imperative does not clearly yield an absolutist case against torture. Yet the faithful Kantian need not give up just yet; a different strategy is available.
Strategy #2: The Kantian Prohibition Against Cruelty
Recall that while Kant denies that we have direct moral duties to non-human animals, he allows that we have an indirect duty to show care and concern for them ‘for a person who already displays such cruelty to animals is also no less hardened towards men’ (LE, 459). And that indirect duty rules out all manner of cruelty on Kantian grounds. If we have reason to believe that torturing will similarly harden our hearts then torture should similarly be prohibited. And we should at least be concerned that torturing will tend to make one crueler. If torturers are going to be able to effectively and reliably extract information in a sufficiently brief period, then torture techniques must be refined and skills honed; torturers must become less responsive to the cries of victims, less receptive to pleas for mercy, and generally habituated in ways a virtuous person is not habituated.Footnote 56 The problem is not that torturers will have the opportunity to become disposed to cruelty. The problem is that training them to torture effectively ensures they will be so disposed; unskilled torturers moved by pleas for mercy are not likely to be very good at what they do. Further, the concern that torturers will become hard in their dealings with men is all too real: the abuse of prisoners at Abu Ghraib was committed against the backdrop of situational influences that dampened responses to the pain and suffering of others and were committed by those trained and encouraged to roughly handle prisoners in the first place. So, if encouraging cruelty is wrong on a Kantian perspective, torture too is wrong.
This powerful argument admits of several limitations. First, it is an argument against institutionalizing torture, not an argument against torture per se: even if institutionalizing torture demands torturers practiced and habituated in the ways of cruelty, it would not follow that torturing is wrong. At best, this argument implies that states willing to practice torture must be content with having untrained and unpracticed novices doing the nasty business of torturing, individuals not yet disposed to cruelty. Virtue ethicists stress the importance of habituation and education in the development of good character and typically note that the single or occasional performance of some morally praiseworthy action does not guarantee virtue. Parity of reasoning suggests that a single or occasional performance of some blameworthy action need not result in vice. So even if concerns about cruelty speak decisively against institutionalizing torture, that argument will not secure the absolutist case.
Further, the appeal to habituation and cruelty depends upon the plausibility of certain contingent empirical claims; in particular, that the consequences of being trained as a torturer cannot be mitigated. My understanding is that United States military officers responsible for training soldiers to withstand torture must themselves undergo significant therapy to mitigate the psychological consequences associated with inflicting such grave suffering on others. Whether these counterbalancing measures are successful is, admittedly, an open question but it is surely realistically imaginable that they can mitigate cruelty.
Finally, this Kantian complaint about torture is prone to an objection similar to one that plagues Kant's position concerning non-human animals. On Kant's view, the wrongness of animal cruelty consists primarily in its contingent relationship to the corruption of our character: the mistreatment of non-human animals can lead us to become like butchers who become accustomed to the sight of death and hardened as a result (LE, 459).Footnote 57 Surprisingly, the wrongness of beating a dog or torturing kittens for fun is not a function of the harm done to the animal being abused, but in the damage done to the abuser. Similarly, torture's essential wrong-making feature on this strategy is not a function of the harm done to persons tortured, but to torturer. And that seems wrong: any plausible account of the wrongness of torture had best identify its wrong-making feature primarily in terms of what torture does to its victims.
What Kantians Can Say About Torture
Neither appealing to the humanity formulation nor the duty to refrain from cruelty will secure an absolutist case against torture on Kantian grounds. But it hardly follows that faithful Kantians should be enthusiastic proponents of torture either. Kant is already on record as ruling out some particular kinds of torture absolutely. And Kantians can defend any number of moral side-constraints governing torture.
By way of argument, return to some of Kant's remarks about cruelty and non-human animals. Again, Kant denies that we have any direct moral duties to non-human animals. But he also holds that:
The human being is authorized to kill animals quickly (without pain) and to put them to work that does not strain them beyond their capacities (such work as he himself must submit to). But agonizing physical experiments for the sake of mere speculation, when the end could also be achieved without these, are to be abhorred (MM, 443).
Thus, it is plausible to think that using non-human animals in painful medical experiments when other alternatives are available is impermissible on Kantian grounds, as is using them for purposes of, for example, testing cosmetics that have even less moral worth than experiments performed ‘for the sake of mere speculation’. And while Kant does not clearly state whether we can consume animals as food, his remarks certainly suggest that factory farming, with all of its tendencies to cause unnecessary pain, should be abhorred.Footnote 58 So while Kant allows that we can use non-human animals in experiments and for labor, there is a limit to how much we can make them suffer: causing suffering that is excessive and unnecessary amounts to cruelty and is thus condemnable on Kantian grounds.
The moral limits on animal cruelty that Kant recognizes surely have analogues in the form of side-constraints governing torture. If causing non-human animals to suffer excessively and unnecessarily constitutes impermissible cruelty, then causing human beings to suffer excessively and unnecessarily is similarly impermissible. If torture is to be practiced at all, comparatively slight methods of torture must be used – say, the use of stress positions or waterboarding – and especially gruesome and painful methods of torture, like beating and burning and amputation, must be avoided entirely. Also, if we may not engage in agonizing experiments on non-human animals when alternatives are available, we may not use torture when alternatives are available. This side-constraint is especially relevant in debates about interrogational torture: interrogational torture is impermissible if alternative means of interrogation are available, thus making it permissible only as a last resort. And if some alternative to punitive torture is available and sufficient to ensure that a criminal is punished to an adequate degree, then punitive torture too must not be used.
Kant's remarks about punishment strongly suggest another position concerning torture that, while falling short of moral absolutism, has significant practical import. Speaking of punishment generally, Kant holds that:
[Legal punishment] 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… He must previously have been found punishable before any thought can be given to drawing from his punishment something of use for him or his fellow citizens (MM, 331).
Despite the retributivist tone in the above remark, there is reason to doubt that Kant endorses deeply retributivist principles at the level of legal practice.Footnote 59 After all, Kant suggests that the punishment for rape and pederasty is castration and that the penalty for bestiality is permanent expulsion from civil society; in none of these cases is the guilty party forced to suffer in the same manner in which he forced others to suffer (MM, 363). While Kant's overall philosophy of punishment is complicated, there are reasons to doubt that Kant supports punitive torture at the level of legal practice, reasons that suggest that Kantians should oppose institutionalizing torture in any form.
One basis for this doubt is manifest in Kant's skepticism about the ability of legal officials to discern whether prospective victims of torture deserve to be punished as such. Familiarly, Kant worries that it is impossible to judge based on experience and with certainty the maxim that led an agent to act (GR, 407, 419). Indeed, ‘The morality proper of actions (merit or guilt), even the morality of our own conduct, remains entirely hidden to us’ and ‘Our imputations can be referred only to the empirical character’ (PR, A551n/B579n). Speaking of capital punishment in particular, Kant holds:
This fitting of punishment to the crime, which can occur only by a judge imposing the death sentence in accordance with the strict law of retribution, is shown by the fact that only by this is a sentence of death produced on every criminal in proportion to his inner wickedness… (MM, 333).
But if a judge cannot impose a death sentence in accordance with the strict law of retribution, because we cannot impute guilt to the accused, then arguably we should not adopt capital punishment at the level of legal practice even if capital punishment is morally permissible.Footnote 60 Kant's skepticism about our ability to determine a criminal's guilt implies skepticism about our ability to determine whether prospective victims of punitive torture really deserve such a harsh penalty in response to their crimes. We might, for all we know, be causing someone to suffer in excess of what they deserve if we subject them to punitive torture. And since institutionalizing punitive torture would make it that much more likely that we will, in fact, punish someone in excess of what they deserve, we ought to be deeply skeptical about putting punitive torture into actual legal practice.Footnote 61
For similar reasons, we ought to be deeply skeptical about putting interrogational torture into legal practice. Above, I sketched an argument that a policy permitting interrogational torture can arguably be justified to victims of interrogational on grounds that they can reasonably accept, but that argument supposed that criminals have the opportunity to avoid being tortured such that it is only the guilty who will suffer interrogational torture – namely, those who played some culpable role in bringing it about that interrogational torture is necessary to secure liberty. Certainly, torturing an innocent is a case of using someone merely as a means if anything is. But if we are unable to determine the desert of a particular agent then we must be incapable of determining whether they did or did not play a culpable role in bringing it about that interrogational torture is necessary to secure liberty. And since institutionalizing interrogational torture would make it that much more likely that we will, in fact, torture an innocent, we ought to be deeply skeptical about putting interrogational torture into actual legal practice.
In sum: an ability to reliably impute guilt is necessary to practice morally permissible punitive and interrogational torture, either to avoid punishing in excess of what is deserved or to refrain from using innocents as mere means. But that means we cannot put into practice morally permissible punitive and interrogational torture on Kantian grounds. This is not an argument that we must refrain from institutionalizing interrogational torture or punitive torture; it is an argument that we cannot practice them, period.
The position that I am sketching is not moral absolutism about torture; just as some murderers might deserve to die on Kantian grounds, some criminals might deserve torture. But my argument does demand that states refrain from empowering their agents to issue and execute torture warrants; it does demand that they put an end to clandestine programs that enable torture; it does demand that states sanction the practice of torture and punish transgressors, and so forth. These results are consistent with Strong Legal Absolutism.Footnote 62 To explain what Strong Legal Absolutism is and what it demands, consider a rival view.
While some commentators reject Dershowitz's call for torture warrants, their opposition to legalizing torture is fairly weak. Having called for us ‘to leave in place the customary legal prohibitions’ against torture, philosopher-judge Richard Posner commends doing so ‘with the understanding that of course they will not be enforced in extreme circumstances’.Footnote 63 Oren Gross similarly notes that while those who torture ‘must assume the risks involved in acting extralegally’ it is nonetheless ‘up to society as a whole… to decide how to respond ex post to such extralegal actions’, adding that society ‘may act to approve her actions retrospectively’.Footnote 64 Posner's knowing nod and Gross' appeal to the people suggest a weak form of legal absolutism, one that does not require punishing torturers. Advocates of weak varieties of legal absolutism hold that torture should be illegal but nonetheless allow that familiar criminal defenses – say, necessity or exigency – will be permitted at trial. Thus, weak legal absolutism is consistent with refusing to punish practitioners of torture and with happily tolerating the existence of a formally ‘illegal’ practice.
By contrast, strong legal absolutism requires punishing those who transgress legal prohibitions against torture: even in ticking-bomb scenarios, criminal charges must be brought, necessity defenses will not be recognized nor will pleas of self-defense or provocation, and transgressors will be held strictly liable for their transgressions – roughly, what is called for by Article 2 of the United Nations Convention Against Torture. I contend that Kantians should endorse strong legal absolutism.
Against my suggestion that Kantians should embrace Strong Legal Absolutism, note that Kant is not generally hostile to necessity defenses: he denies that a death sentence is appropriate for someone who ‘shoves another, whose life is equally in danger, off a plank on which he had save himself’ judging that such an offense is ‘unpunishable’ (MM, 235). Uwe Steinhoff argues that Kant would similarly reject the thesis that punishing torturers is morally obligatory and thus would reject Strong Legal Absolutism. Steinhoff observes that the Categorical Imperative is compatible with violent self-defense generally.Footnote 65 But it hardly follows that just any sort of violent response to unjust aggression is permissible on Kantian grounds. Consider Steinhoff's suggestion that ‘Kant makes it very clear that the unjustly attacked person owes his attacker nothing’.Footnote 66 But Kant pretty clearly recognizes that there are limits to the method and manner in which we may respond to unjust aggressors. Even if we may respond to an aggressor with violence on Kantian grounds, we cannot ignore the fact that we are responding to an aggressing agent, and insofar as we are obligated to respect the humanity of other agents, we are still obligated to respect his humanity. As Barbara Herman suggests:
The [Kantian] justification of self-defense does not devalue the aggressor because he is guilty of aggression. He forfeits no moral title; I have no claim of moral superiority. If I may act with violence against aggression, I must do so without ignoring the fact that the object of my action is an aggressing agent.Footnote 67
So while we might be able to will that a maxim calling for torturing unjust aggressors should become universal law, other tests for Kantian legitimacy might preclude defensive torture and the various Kantian side-constraints still govern our response to them. Pace Steinhoff, we do owe our attacker something.
In any case, there are at least two reasons for thinking that the Kantian case against torture is Strong Legal Absolutism. First, declaring that torture is legally prohibited but with a nod to non-enforcement does little to deter would-be torturers. There would be considerable doubt as to whether standing prohibitions really would be enforced, especially given that the sort of emergency scenario that is most likely to tempt potential torturers – namely, the ticking bomb scenario – is one in which extreme circumstances are obvious and concerns about expediency most pressing. Given that any realistically imaginable ticking bomb scenario is likely to be the sort of case where torturers can appeal to some familiar affirmative defense – say, necessity or exigency – how likely is it that a jury wouldn't be convinced? What judge wouldn't be inclined to suspend a torturer's sentence, especially if the trial is a public one and the decision to punish is unpopular? Prospective torturers could reasonably lack fear of punishment if weak legal absolutism were practiced and legal prohibitions of torture should not be expected to have much deterrent effect. If concerns about deterrence have any role to play in Kant's justification of legal punishment – and they do (LE, 286) – then prohibitions of torture must be believed to have teeth, a belief undermined by anything less than Strong Legal Absolutism.
Second, there are costs incurred by refusing to punish torturers. It is a truism that engaging in torture harmed the reputation of the United States abroad by flouting international law. As such, it is legitimate to worry that ‘international law may be of no account if even the most powerful regime – the one that can most afford to sustain damage – is willing to dispense with legal restraint for the sake of a tactical advantage’.Footnote 68 To the extent that refusing to enforce standing laws undermines trust among states, such refusals are objectionable on Kantian grounds. The sixth of Kant's Preliminary Articles for Perpetual Peace Among Nations demands that:
No state at war with another shall permit such acts of hostility as would make mutual confidence impossible during a future time of peace. Such acts would include the employment of assassins or poisoners, breach of agreements, the instigation of treason within the enemy state, etc. (PP, 346).
This article is relevant whether or not one thinks that the United States is engaged in a literal war against terrorism. If the United States is at war, then Kant's concerns about mutual confidence are clearly on point. But even if the United States is not at war, his remarks are still relevant. If acts that make mutual confidence during wartime impossible are not permissible because they make mutual confidence impossible during peacetime, then acts that make mutual confidence impossible during peacetime should similarly be impermissible. After all, they too make mutual confidence impossible during peacetime. And breaching agreements, including agreements concerning the prohibition of torture, threatens mutual confidence. Since international law does presently legally prohibit torture and does demand punishing torturers, states are obligated to comply with those standing arrangements for the sake of preserving mutual trust. Thus, they are required to abide by nothing less than Strong Legal Absolutism.
Conclusion
It might be objected that the Kantian position that I advocate seems to give priority to legal and political matters rather than ethical matters: if current domestic and international law were different then we might not have a moral duty to legally prohibit torture. But there is no reason to think that Kantian political philosophy cannot give some content to Kantian moral philosophy. Indeed, it may well be that the content of many of our moral duties depends on the results of political decision-making and philosophizing.Footnote 69 And what I call the Kantian case against on torture better coheres with Kantian scholarship than some rival view that makes the Kant a moral absolutist about torture. For while Kant does sometimes talk like a moral absolutist – say, in his remarks about suicide and lying promises – few Kantian scholars are willing to suppose that Kant is a moral absolutist per se. If he were, then it would be impossible to make sense of, among other things, Kant's division of moral duties into perfect and imperfect varieties: only the former are properly regarded as giving rise to exceptionless duties. Once the tired image of Kant as a dogmatic absolutist is shed, there is little reason to demand that Kantians be absolutists across the board, even about a topic as prima facie morally repellent as torture.
