by Bernard Chazelle
A signatory to the UN Convention Against Torture, the United States "does not torture." [1] Yet abundant evidence indicates that it does, directly or by proxy—and in fact always has. An old American tradition of state-sponsored torture even has its own lexicon: SOA, Kubark, Phoenix, MK-Ultra, rendition, CIA's "no-touch" paradigm, etc. It is quite popular, too. Torture enjoys more than twice the public support in the US that it does in France, Spain, and the UK. [2] One of the most watched TV dramas, 24, is but an extended ode to the glories of torture. The former director of a prominent human rights center at Harvard writes of the judicious use of sleep deprivation, hooding, and targeted assassinations; he concedes the government's need to "traffic in evils." [3] The nation's most celebrated defense attorney recommends "torture warrants" and "the sterilized needle being shoved under the fingernails" ("sterilized" because he is a liberal). [4] The most cited legal scholar in the land writes: "If the stakes are high enough, torture is permissible. No one who doubts that this is the case should be in a position of responsibility." [5]
Anti-torture voices have been left sounding defensive, insecure, incoherent. Yet, while boasting the world's highest incarceration numbers and supermax prisons characterized by a warden as a "clean version of hell," the US has also begun to question its tolerance of torture. [6] The debate is on, and torture is winning. I intend here to lay the foundation for a strong, cogent anti-torture position. It rests upon three principles:
- Torture is always wrong;
- Torture must be banned by law unconditionally;
- Not all torture decisions should be morally codified.
The first two principles reject torture on moral grounds (it's wrong) and legal ones (it's bad). Unfortunately, they do not imply that one should never torture. If, indeed, our only choice is between two acts that are immoral, these two rules alone won't tell us what to do. This central dilemma arises in principle—we can all imagine ourselves in an extreme situation about which we cannot say with certainty that we would not torture—but does it arise in practice? Many say, with some justification, that it does not. Whatever the case may be, there is a hefty price to pay for dismissing the central dilemma on implausibility grounds, as many liberals are wont to do. Once the improbable is deemed morally irrelevant, torture can no longer claim the status of absolute wrong, for there is no such thing as an "absolute-wrong-in-practice." Any serious condemnation of torture must account for the central dilemma.
Hence my third principle. It stipulates that no ethical code (ie, universal decision procedure) should tell a would-be torturer what to do in all situations. This is to avoid rationalization and, beyond it, the dilution of moral responsibility in the hypothetical case where not to torture is no less an immoral option than to do so (the central dilemma). The third principle is a point of meta-ethics. It is not a moral rule per se, but a statement about the inapplicability of moral rules. It is designed to overcome the justificatory purposes embedded in any ethical code. One may object that the central dilemma arises with any moral wrong, so why single it out? Because it lies at the core of the "torture issue" itself, which, with the wide support it enjoys, is indeed an issue. How to aggregate universal moral principles into decision procedures, a central problem in ethics, is in my view the only interesting aspect of the torture question; the rest is straightforward.
Like many, I feel strongly enough about torture to find the very notion of a "torture debate" distasteful. But sentiment alone means nothing. I feel strongly about racism, too. But racism is not wrong because it offends my sensibilities. It is wrong because it violates reason and human dignity. Likewise, if we cannot offer a reasoned account of the absolute wrongness of torture (especially given the wide public support for it) then our impassioned opposition, indispensable though it may be, will still be, strictly speaking, meaningless. It also matters because one cannot fight effectively for a cause one does not understand. Is it a coincidence that torture has remained so popular in this country amidst such an impoverished public discourse? [7]
What is torture? "I know it when I see it" is a fine answer and rough agreement with common intuition will do. Supermax incarceration and prison rape can be construed as institutionalized forms of torture. For the purpose of this essay, however, I narrow down the definition to the forced exchange of information for the relief of unbearable pain. Much like slavery, torture is coerced trade. To many, its abhorrence requires no empirical evidence: it is a priori, intuitive, and visceral. So much so, in fact, that even asking why seems immoral, as if merely speaking of a ghost might make it appear.
But, if torture is so evil, why is it so hard to explain why? Let's try. Some say a society that allows torture loses its soul and brings shame on its members. This is true, but it explains nothing—at least no more than calling murder wrong because it makes you a bad person. A line often heard is that torture does not work. Never mind the fragility of a proposition that is both unprovable and falsifiable. Even if true, this claim is a gift to the torturers: "Make it work, Mr Inquisitor, and the moral turf is yours." It's like rejecting slavery because "it does not work" or opposing cannibalism on nutritional grounds. Consequentialism is thin gruel against torture. Beware of the sentence that ends with the words, "therefore torture is evil." Better for it to start, "Torture is evil, therefore..."
This brings us to the deontological perspective. Do we recoil from torture because it treats a person only as a means to an end? It is a principled view that might account for our rational rejection of torture, but Kant's Categorical Imperative is too much at variance with Anglo-American norms to explain the instinctive revulsion the practice commonly elicits. (As the death penalty illustrates, note that popularity does not contradict abhorrence.) In his paeans to torture, Dershowitz is merely echoing Bentham and, beyond it, the reigning utilitarianism of our time, which, from conditional welfare to advertising, routinely flouts Kantian ethics. And yet, is there a doubt that the wrongness of torture finds its source, not in a holy book or in the final link of a chain of observations, but deep in humanity's moral intuition? On this we all agree.
Or do we? Few would argue that waterboarding Khalid Sheikh Mohammed was worse than shooting him in the head. Yet killing does not make us wince the way torture does. Why? Could it be the excruciating pain? Doubtful. Baby Mohammed lost both legs during Shock-and-Awe and, over a 10-hour period, bled to death stuck in the debris of his home, a horror entirely foreseen in its outline, if not its particulars, by the architects of the war. The baby's pain vastly exceeded that of his namesake. Yet if Rumsfeld must one day cross Europe off his travel plans, it will be because of Khalid Mohammed, not baby Mohammed—despite the former SecDef's direct responsibility in the latter's agony. Pain and death do not explain why torture feels so evil.
Then what does? Perhaps the deadly mix of fear, humiliation, abandonment, and open-ended sadism that the practice connotes. The torturer never says, "I go home at 5." Torture stirs in all of us the age-old anxiety of a cruel deity that keeps us forever conscious to suffer an endless agony. Pain, like relativity, distorts time. (A root-canal patient can tell you all about eternity.) Past a certain point, the victim's fear is no longer that he will die but that he won't. Torture is a window into hell, with a satanic god cast as a human sadist. I believe one cannot grasp the role of torture in the imagination without integrating its metaphysical resonance. Torture rehearses eternal damnation. And that's not a good thing, because hell scares the hell out of everyone, even those who don't believe in it.
To add insult to injury, the torturer reflects back to us a magnified image of that repressed speck of sadism buried in all of us. This did not always bother us. God gave Moses not one but two commandments against lust, and not a single one against cruelty; likewise, Augustine deemed cupidity a more serious offense. It was not until Montaigne and Montesquieu that cruelty acquired a special status in moral philosophy. [8] Our revulsion toward torture is hardly universal—children can be astonishingly cruel to animals—but, rather, the sign of a certain liberal disposition. Torture offends us through its frontal assault on human dignity. Beyond subverting free will into "anti-will"—your being tortured does not simply violate you: it makes you violate yourself—it denies something even more fundamental than freedom: personhood. It dehumanizes not only the victim and the torturer, but society as a whole. Or so our modern liberal sensibilities tell us.
Should torture be legalized in exceptional circumstances? The answer is an unequivocal no. The ban must be unconditional. Why? Because grotesquely evil behavior must be criminalized? Pleasing though it may be, this simple answer won't do. We must first examine whether there might not be a utilitarian reason to make legal exceptions. (Even the most committed deontologist will recognize the need to test laws against their consequences.) I will show that there is no room for exceptions by revisiting the three arguments central to the issue: TBS, self-defense, and torture creep. I'll also discuss the criminal prosecution of torturers.
The ticking bomb scenario (TBS) would appear to beg for an exception—see [9] for a definition. (I'll assume the usual conditions of imminence, gravity, proportionality, and certainty, without which TBS is not worthy of consideration.) The first issue to address is consistency. TBS advocates often lack the courtesy to grant the same rights to their enemies. They remain oddly silent on whether, say, the Taliban would be entitled to torture captured American soldiers thought to know about imminent drone attacks. There might appear to be a normative basis for the double standard. After all, we're the good guys and they're not, so why should we grant them the same moral latitude? That's nonsense. Our own code of warfare, such as it is, dictates that it apply equally to both sides—as do the Geneva Conventions. Whether it should be so or not is an interesting philosophical question, but in practice this point is already settled.
The legal issue hangs on the "rarity principle." We all see the need for a law against murder. But do we need a law for a bad act that happens at one millionth the rate of murder? Probably not. Legality should offer only a blurry reflection of morality, not its mirror image. Whereas morals delineate complex fractal lines, laws should follow smooth contours free of singularities. As the saying goes, "Hard cases make bad laws." This is not a weakness of the law but a strength: that's how it can be both universal and enforceable. TBS theorists will agree but say: "Look, 1,000,000/1,000,000 = 1, so an action likely to cause one million deaths at one millionth the rate of murder matches the expected harm of murder, and hence merits its own law; ergo, legalize torture. QED." A three-word refutation: Break the law. No one's yet suggested a new speed limit sign: "55 — Unless You're Taking Your Dying Uncle To The Hospital." Speed up if you must, and pay the price later. Tucking exceptions into law is courting the same trouble as overfitting a machine learning classifier, ie, loss of generalizing power and diminished appeal to universality.
Self-defense was invoked in the infamous Bybee "torture memos." [10] On the face of it, this is preposterous. A torture victim is not a threat. A captive terrorist such as Khalid Sheikh Mohammed is a culpable bystander, not a culpable aggressor. Therefore, first, the argument would need to appeal to self-preservation and not self-defense; second, this in turn would crash against the accepted legal doctrine that bystanders, even guilty ones, may not be hurt intentionally. [11] Perhaps a legal argument of "distributive" justice could be made that, if harm is unavoidable, it is preferable to inflict it on the guilty party if there is the option. But, by that logic, a gunman who shoots you might be forced to give you his organs to save your life: if one life has to go, hey, guess which one? Trouble is, this constitutes a brand of justice far too alien to our own to be acceptable. To that normative consideration, I would also guard against the slippery notion of "collective self." Most aggressive wars in history have been fought in the name of self-defense. This might then justify the torture of war prisoners when one's country is under attack, thus losing the classical distinction between jus ad bellum (why one may go to war) and jus in bello (how one may fight a war). Even if all other options have been tried, the torture of terrorists cannot be called self-defense.
The case of individual, non-state sponsored TBS is not as clear-cut. You're entitled to stab a man on self-defense grounds if you see him break into your house and try to strangle your daughter. (No one will dispute that your "self" may extend to her.) So why can't you do the same if he refuses to divulge her location after he's kidnapped her and buried her alive with 20 minutes left to live? Simple: because the man is not attacking your daughter. But isn't his silence every bit as much a weapon as his hands? After all, he can wield either one at will to decide her fate. One can draw two distinctions, neither of which resists scrutiny.
The first one is epistemic: your belief that the man knows the burying location of your daughter and that she is still alive cannot match the certainty of your witnessing her strangulation. This can be postulated away—certainty is an accepted part of any serious TBS narrative. (No need to assume here that what you believe is true: only that you have no reason to doubt it.) The other distinction, silence vs strangling, ie, omission vs commission, concerns neither causality nor intentionality—in both cases the man acts willfully to kill. It rests solely on timing, a consideration of no discernable normative relevance. One can, likewise, torture by omission. If the captive were diabetic, it would be torture to withhold his insulin until he talks, since this would fit our characterization of torture as a form of coerced trade. In sum, tying a stand on torture to a distinction between omission and commission is dicey. And even a plausible self-defense plea (which, it is fair to say, would never happen in practice) must give precedence to the rarity principle: break the law if you must.
Torture creep is yet another reason to make the legal ban watertight. The historical record indicates that the slightest legal opening to torture will metastasize into widespread institutional abuse. This "cancerous" spread affects intention, which leads to intimidation, submission, and extraction of false confessions. Even a state that allows torture only in rare cases will soon insist on competent torturers; hence torture schools, torture experts, torture research, and, given the gravity of the matter, an administrative state structure to oversee it all. In other words, it will build itself a "School of the Americas." [12] The evidence is overwhelming: torture intended as a security tool will always morph into an instrument of power. That alone justifies a total ban on torture.
Finally, how much leniency should a judge extend in hypothetical cases of torture that demonstrably save lives? It would seem wise to grant judges enough sentencing discretion to keep would-be torturers in the dark and induce them to proceed on worst-case assumptions. Just as torturers may not invoke the Nuremberg defense, so anyone who orders torture, directly or by proxy, must be held legally responsible. Wartime torture admittedly poses a conundrum. There is empirical evidence that it is an inevitable by-product of aggressive warfare. If so, it may thus fall under the rubric of war crime (like executing children) and lose its categorical singularity. The issue of punishment becomes more complex. In theory, the jurisprudence on war crime should provide the relevant legal authority. Right, but in theory war criminals are dragged before a judge even when their side wins the war. In theory.
If torture is illegal at home, subcontract it overseas. Extraordinary rendition is the process of handing suspects over to third-world dictators with the promise that they won't be tortured and the certainty that they will. If only because of the rank hypocrisy behind it, one should not extend to rendition the customary distinction between directly causing harm and merely allowing it. If anti-torture activism results only in increased rendition, what is being morally gained? Torture is barbaric; rendition is barbaric and hypocritical. It must be an integral part of the fight against torture.
Most opponents of torture would declare the matter settled: always immoral; always illegal. They would be wrong, for neither morality nor the law can answer the question: were you to face your daughter's kidnapper, may you torture him? The answer, whatever it is, must remain unjustifiable, so as to impose upon you, and you alone, the full moral weight of your action. This is the only subtle point of this essay, so I'll begin with a gentle introduction.
There are two common objections to the TBS question itself, neither of them wise. One is that it is a trap set up by torture lovers to force a small concession from their opponents, shift the debate to "settling the price," and then gloat: "See, we agree!" Of course, it's used as a trap. But the question is legitimate, in fact necessary, and the dreaded shift is easy to avoid. The second objection is that "it never happens," so why even discuss it? This collapses on three grounds. First, the claim is unprovable. Second, if it never happens, why should you care about the moral outcome? (Do you worry about my ethical take on green Martians?) Perhaps you care out of concern for torture creep: if so, address it on its own merits. Third, the philosophical value of hypotheticals is undeniable. Yes, the field of ethics has been hard on fat men being shoved in front of trolleys, but as in mathematics these "singular" points can be the lampposts that light up the dark street. Just as the corners of a triangle tell us all there is to know about its shape, so extreme cases help us read our moral compass.
The unease that accompanies the discussion comes from elsewhere. It stems from a confusion between morals and ethical codes, ie, between moral principles and the decision processes by which we make moral choices. Morality is about right and wrong. A moral (deontological) system may include specific injunctions, "Do not kill," as well as abstract precepts, "The Golden Rule." It is preferably universal in a Kantian (negative) sense, in that it should, at the very least, not contradict its adoption by everyone else: "I may leave a restaurant without paying for my meal" cannot be a moral maxim, for its wide adoption would quickly cause all restaurants to close shop, which in turn would contradict my desire to eat out. Moral systems consist mostly of intuitive, a priori judgments, unshackled from the need for empirical validation. Lying is wrong and so is tormenting a child. These are truisms. But what do we tell a terminally ill child who asks: "Am I going to die?" We lie, of course. Are we thereby signifying that honesty is less important than compassion? Not at all. For example, most of us would agree that empathy may not always excuse perjury. Honesty and compassion are both universal moral values, but their relative ranking may vary depending on the context. Unfortunately, there is no simple rule to tell us which should prevail when. We can always follow our moral intuition. But that can be dangerous: hate, self-interest, prejudice, biases can be all too "intuitive." And we are all so good at lying to ourselves.
Far better for us to turn to an ethical code, ie, a decision procedure to convert our moral beliefs into action. To guard against egoism, we'll try to make it universal so others might want to adopt it, too. We're all familiar with ethical codes: etiquette; chivalry; just-war theory; political ideologies, etc. The conservative code, for example, tells us that the best way for government to help the poor is not to help them at all. This is so weird we might never have come up with anything like it on our own. But it meets the three criteria of an ethical code: it seeks to match our actions with our morals; it is not self-evident (if it were we wouldn't need it); and it is effective (it helps us identify whom to vote for).
The ideal ethical code would be a big handbook—infinitely long, to be precise—with, next to each possible situation, a list of moral actions to choose from. This being somewhat unwieldy, a code will look more like an "algorithm," ie, a coherent set of interconnected generalization and abstraction rules based on representative cases that mesh with our intuitions. A perfect code would have to be complete, meaning that it covers all cases, but that is unrealistic. We might hope for it to be sound, ie, never to prescribe actions that violate our moral intuitions. To fix such violations, Rawls suggested tweaking rules and intuitions back and forth until we reach some sort of stable, "reflective" equilibrium. Inevitably, an ethical code will be on occasion intractable, meaning that it may actually tell us not to do X in situation Y but we're just too dumb to figure that out by the time Y has passed. The complexity of a code must be honored. It is in fact unethical to gerrymander moral boundaries to make it easier to lead a moral life—Bush tried to do just that with his "You're with us or against us." Naturally, to adopt an ethical code is itself a moral decision. (The vigilant reader will immediately spot the self-referential implication but I'll leave that one for another day.)
It is beyond dispute that an ethical code that advises us to lie to the dying child can be sound. But can all moral dilemmas be resolved by sound ethical codes? No code can be expected to be complete and always tractable, so the only reasonable answer is no. Fine, but what about torture? A ruling cannot be derived a priori, unless perhaps one considers torture a wrong universally greater than all others, a proposition clearly untenable. So what do we do? David Luban argues on incompleteness grounds that moral systems (hence ethical codes) should not apply to TBS. [13] I agree, but for somewhat different reasons. I plant a flag right in the middle of the TBS swamp with the sign, "No Ethical Code Here." By that, I am not simply stating the impossibility or intractability of always reaching a decision via a universal code, something of which I cannot be sure. Rather, I am decreeing it. I disallow any code for TBS not because I have to (Luban's position), but because I choose to. My rejection of an ethical code appeals to an existentialist intuition. If morality is going to be incapable of helping us decide, then our choice should engage us fully so as to avoid any rationalization. It should be the ultimate act of free will. The would-be torturer must accept full moral responsibility and be denied both alibis: "My code made me do it" and "I was confused." The latter says that even incompleteness is no excuse. Here is a quick explanation.
Considering the real-life story of a young Frenchman in the 1940s torn between his urge to fight the Germans to avenge the execution of his brother and his desire to stay with his heartbroken mother, Sartre reviewed various moral systems to highlight the difficulty of teasing ethical guidance out of them. His point was that ethical codes are dressed up as advisory devices when in fact they serve only justificatory purposes. In other words, how do we know that ethical codes aren't rationalization engines in disguise, mechanisms for evading responsibility, sophisticated dodges? This points to the exculpatory nature of an ethical code. The neat thing about being advised, you see, is that we can always blame the advisor.
Sophie is given a choice: to kill one of her two children or have both of them killed. She has no choice but to act immorally. This is not my judgment but hers; more accurately, it is an a posteriori inference from the knowledge that she'll be racked with guilt for the rest of her life. Sophie's choice falls within the world of morals but beyond the human reach of moral guidelines. The injunction, "Spare the child who..." is morally impermissible. Sophie is thus able to act in a way that, though necessarily immoral, is not ethically mistaken. No one can ever tell her, "You saved the wrong child." In fact, we so believe that no ethical code should apply that we'd be shocked to hear Sophie justify her choice by appealing to some holy book claiming that God prefers, say, older siblings. "My religion made me do it" would reek of bad faith (in the Sartrean sense of self-lying) and suggest that perhaps Sophie preferred her older child but blamed her faith instead. It is morally imperative for her to renounce any ethical code and take full responsibility for her choice—or, as an existentialist would put it, to admit that she is condemned to be free.
The kidnapping story is identical except in one key aspect: the temptation of a code. Faced with the kidnapper and the mental picture of your daughter imploring you to do everything in your power to save her, your intuition is likely to whisper in your ear, "Torture the bastard!" Unlike Sophie's, your choice may even seem entirely obvious. This intuition may help you decide but, even after integrating the moral relevance of family, it still violates the deontological constraint of treating torture as a universal moral wrong: after all, your intuition might be quite different if the "bastard" were your son and the captive girl the past killer of your daughter. To give up on an ethical code altogether may be quite difficult, in fact. But this is the only way to respect the absolute wrongness of torture. If you're going to do it, you'd better be ready to "own it" and take all the blame. In this instance, free will implies carrying out, and accepting to carry out, your own decision in the belief that you would do it again in the future. In other words, neither of these excuses is acceptable: "I was foolish"; or "I did what I thought was best but that was against my will."
Consider a variant of TBS where torturing one person prevents the torturing of two other people elsewhere. Should you do it? Basic utilitarianism of rights says yes. The danger is that moral calculus is nothing but the exercise of an ethical code, hence a rationalization. I am not saying this is not allowed to influence your decision—one cannot shield oneself from all moral calculus. I am saying that in the end you must be the only owner of your decision. You must accept the guilt for your action as the necessary consequence of your freedom and you must reject any attempt to justify your choice. What if you are ordered to torture? Assuming a moral choice is possible, ie, disobedience is not punishable by death, you should refuse to torture unless you are confident that you would give the same order—were you in a position to do so—if you also knew that you had to carry it out yourself. (Merely believing that you would give the order is not enough.)
The final verdict on torture: always wrong; always illegal; always unjustifiable.
— Bernard Chazelle, July 2009