Thursday, December 13, 2012

Sam Harris on Torture

Although Sam Harris's PhD is in neuroscience, he never left his undergraduate philosophy degree (from Stanford in 2000) far behind him. This has been clear from the time of his first major bestseller, The End of Faith. In that book--the first of the so-called "new atheist" bestsellers--Harris devotes a chapter to ethics, previewing what he later attempts (and, in my view, fails) to do in The Moral Landscape.

In that ethics chapter from The End of Faith, Harris even does a bit of applied ethics (by which I mean the discipline of thinking about particular moral issues and dilemmas in a rigorous way in the attempt to provide reasoned guidance in decision-making). Specifically, he focuses on the moral status of two controversial questions: the ethics of torture and the ethics of pacifism (spoiler: he thinks torture can be justified while pacifism is "flagrantly immoral").

This past semester I had my introductory ethics students read what Harris has to say on torture. I wasn't surprised to find some students convinced (at least initially) by what Harris says. In a manner characteristic of the new atheist writers, Harris tackles the ethics of torture with the kind of confident authority and eloquence that so effectively obscures the serious defects in his argument. This is not to say that Harris has nothing of value to say about torture. He makes some interesting and important points in the course of developing his unsound argument. But the main argument is nevertheless unsound on several levels.

It struck me as a useful exercise in critical thinking to assess with my students a piece of writing of this sort: writing in which word choices create false equivalences and authorial confidence masks a deep ignorance of relevant literature. And it also seemed that Harris's essay--and a consideration of what he overlooks--could serve as a useful springboard for deeper reflection on the issue of torture and the broader ethical issues it brings to light.

For the very same reasons, it seems worth addressing on this blog. But doing so adequately--especially in a way that affords us a deeper look at torture and surrounding issues with an eye towards broader ethical insights--would take more than a single post. In this post, then, I want to content myself with offering an overview of Harris's core argument and, as I go along, quickly identifying his main oversights. A deeper look into these oversights and their significance will have wait for later.

So, what is Harris's main argument? Harris focuses on the issue of torturing suspected terrorists for the sake of acquiring life-saving information (such as the location of a hidden bomb that will take untold innocent lives unless found and disarmed in time). This way of framing the issue--in terms of the so-called "ticking bomb" case popularized in recent years by Alan Dershowitz--may itself be problematic. David Luban has argued precisely that in the Virgina Law Review, because it treats a highly unusual scenario as if the moral status of torture in that scenario could be considered on its own, isolated from a consideration of the ethics of our broader policy (either secret or open) with respect to torture. Luban offers some quite interesting and useful remarks about the difficulty with such isolated consideration of the extreme-and-rare case. Harris seems utterly oblivious to any such difficulties. If he's aware of them, he ignores them utterly.

Moving on: Having focused in on the ticking bomb scenario, Harris begins his argument with the assumption--which he takes to be rather self-evident--that if the prospective torture victim is known to be guilty (and, presumably, to possess the needed information), torture is clearly justified.

This starting assumption is hardly as uncontroversial as Harris takes it to be. First of all, there are those who will balk at its consequentialist character--taking there to be something intrinsically wrong about the deliberate infliction of pain when such infliction is not merely foreseen as a side-effect of something one is doing but is an intended aim of one's action (even if it is intended not for its own sake but as a means to some more ultimate objective).

Second, there is the whole question of the relative efficacy of torture as compared to other forms of interrogation. If one is sufficiently consequentialist about these matters, what the ticking bomb scenario would do is justify that means of extracting the needed information that is most likely to produce the desired result. When, if ever, is it reasonable for us to suppose that torture is that most-effective means and so act accordingly? Harris doesn't even take up the question. He does concede that the probability of successfully extracting reliable information through torture may be low, but he fails to consider that this may mean there are other interrogation techniques with a higher track record of success (or with which trained interrogators have more training and experience, and so are better equipped to use effectively).

Of course, in order to explore such comparative issues, he'd need to conceptually distinguish between torture and interrogation. Harris never even attempts to do this.

In any event, Harris assumes that torture is justified if you've got a known terrorist in your clutches who has the information you need in order to save countless lives but who refuses to divulge it. The moral problem, for Harris, arises when we suspect but don't know for sure that the person in our custody is indeed a terrorist with knowledge of the whereabouts of the hidden bomb. In that case, Harris says that the moral problem is that we risk torturing an innocent person by mistake.

In fact, Harris's next error is in arguing as if this--the risk of torturing an innocent--is the only risk. He doesn't explore the risk of torturing someone who is indeed guilty of terrorism and involved in the bombing plot, but who lacks the relevant information. Torturing someone to extract information they don't possess is, in the ticking bomb case, not just a dangerous waste of time. The problem runs deeper. What lies behind the idea that torture can be justified in these cases is the assumption that increasing someone's suffering is more likely to break down their resistance to doing what you want. If they don't talk, you move from interrogation to thumb screws. If they still don't talk, you tighten the thumb screws. Follow this pattern with someone who doesn't have the information you want, and you are inflicting ever increasing amounts of suffering to no good effect. Continued escalation can lead to false information or death, but not to anything useful.

But all of this is preliminary. Harris's real argument begins only once he has swept a host of philosophical and practical problems under the rug in order to get to this premise: The main moral problem with torturing a suspected terrorist in order to gain life saving information is that the suspect might be innocent. His aim, then, becomes to challenge the view that "uncertainty about a person's guilt will generally preclude the use of torture." He challenges this view, interestingly enough, by arguing that "such restraint in the use of torture cannot be reconciled with our willingness to wage war in the first place."

Here it is important to pause long enough to remind the reader that Harris's discussion of torture is immediately followed up by an attack on pacifism. Harris's immediate conclusion with respect to torture is this: Either we give up on war, or we accept torture. By making it clear that we should not give up on war (that it would be "flagrantly immoral" to refuse to go to war in the right cases), Harris drives home his own broader conclusion, namely, that torturing people--even those who might be innocent--can be morally justified.

That, then, is his broader argument. But in his focused discussion on torture his aim is the more narrow one of arguing that if we think war can be permissible, then we have to hold that torturing potentially innocent people can be permissible too. And why is that? Because war inevitably brings with it "collateral damage"--the deaths of innocents who are in the wrong place when the military installation is blown up, who end up in the crossfire when enemy soldiers are targeted, etc. Here's how Harris puts the point:

What, after all, is "collateral damage" but the inadvertent torture of innocent men, women, and children? Whenever we consent to drop bombs, we do so with the knowledge that some number of children will be blinded, disemboweled, paralyzed, orphaned, and killed by them. It is curious that while the torture of Osama bin Laden himself could be expected to provoke convulsions of conscience among our leaders, the unintended (though perfectly foreseeable, and therefore accepted) slaughter of children does not.
In other words, Harris's argument for torture purportedly piggy-backs on the justification for allowing collateral damage. If collateral damage is morally justified (given a certain probability of achieving a sufficiently good result), then torture will be justified (assuming a comparable probability of achieving a similarly good result).

But if Harris wants to piggy-back his case for torture on the case for collateral damage, it would make sense to look at how and when collateral damage has, historically, been taken to be justified. It would make sense, in other words, to try to show that the reasons why people have historically justified collateral damage are also reason that would, likewise, justify torture. There is, after all, an enormous body of literature on the ethics of war. There is, in fact, a "just war tradition" that has given rise to the so-called "just war theory," which has through the years served as the basis for much of our national and international policy-making and deliberations with respect to war. And, as a matter of fact, the just war theory has explicit things to say about collateral damage. There is, if you will, a conventional understanding of the conditions under which so-called collateral damage can be justified.

Sam Harris presumably never encountered this just war tradition and the body of literature surrounding it during his years as an undergraduate philosophy major. I say this because he makes no mention of it. None. He argues that torture is justified if collateral damage is, but at no point does he actually take up the classical justifications of collateral damage to see if, in fact, they transfer to torture.

As a matter of fact, they don't. Let me say that again: According to the classical just war conditions for the justifiability of allowing collateral damage, there is no parity between causing collateral damage in the pursuit of military aims and torturing a suspected terrorist to extract information. The conditions that are classically invoked to justify the former DO NOT justify the latter. So Harris' oversight here is a serious one.

It is made even more serious by the fact that Harris goes through the motions of looking for some relevant difference between torture and collateral damage that might be invoked to challenge the moral parity he attributes to them. He considers and nimbly refutes some very poor reasons for denying this moral parity. Perhaps he thinks that the standard just war position on collateral damage--which, if accepted, would undermine the purported moral parity with torture--is just as bad a basis for denying the moral parity as the bases he considers. But if so, he fails to say why. He fails to even take up the question.

In short, he says that A is like B--but fails to consider the standard account of B, which if accepted would render A unlike B. And he fails to consider this standard account, which would refute his thesis if it worked, even though he pretends to go through the motions of looking for considerations that might refute his claim that A is like B.

But if anything should be taken up in a philosophically serious attempt to show that A and B are morally equivant, it would presumably be the standard understanding of B, at least if, according that standard understanding, A and B are not equivalent at all. Failing to do this is either a serious philosophical failing in its own right (completely ignoring the clearest and most historically important basis for objecting to your thesis) or evidence of a profound ignorance of the scholarly knowledge that anyone making Harris's argument ought to have.

I should note that, even though I think the standard just war view here would, if acceptable, undermine Harris's purported equivalence between collateral damage and torture, I'm not at all committed to the standard just war position. I think it may very well be one we shouldn't accept. And if we don't accept it, the equivance between torture and collateral damage that Harris asserts would be restored--but for reasons that, I suspect, would make it much harder for Harris to make the case that war is justified even when it produces collateral damage. In other words, I think the strongest basis for challenging the just war tradition on this point is more likely to establish an equivance that tips the scales against BOTH collateral damage and torture than one that tips the scales in favor of both. But that is an issue for another time.

For now, I want to say this: In this post I've boldly asserted that Harris's claimed equivalence between collateral damage and torture would be undermined if we accepted the standard just war position on collateral damage. I have yet to explain why this is so. A full explanation will have to wait for a later post, but a brief account is warranted here. In brief, then, the just war tradition works out, in relation to collateral damage, the implications of something called the doctrine of double effect. This is a doctrine that lays out precise conditions under which it can be morally permissible to pursue an action that has both a good effect and a bad one. Among other things, the bad effect cannot be a means to the good effect (another crucial requirement is that the good effect outweigh the bad). In other words, the bad effect has to be an unintended, if foreseen, side-effect--that is, an effect the elimination of which would not prevent the good effect from being achieved.

This is what distinguishes bomb attacks with "collateral damage" from bomb attacks that directly target civilians in order to terrorize the population and thus weaken their resolve to continue the war. The latter is strictly ruled out on traditional just war grounds because the bad effect is in this case intended as a means to the good. Fire bombing of Dresden? Ruled out by traditional just war theory. Nazi bombing raids over London? Ruled out. Atomic attacks on Hiroshima and Nagasaki? Ruled out.

What characterizes a legitimate case of "collateral damage," on just war theory, is this: If the innocent civilians could be evacuated from the site of a bomb attack on a military target, it would still be possible to perform the very same bomb attack for the sake of achieving the very same end result (taking out the military target). You don't need the suffering and death of the civilians in order for the bomb attack to achieve the military objective it is intended to achieve. The same cannot be said of the suffering of a torture victim that is pursued for the sake of extracting information. You can't torture information out of someone without causing them pain. Put bluntly, if you give them enough sedatives that the torture doesn't hurt them, it's not torture and the whole method of pursuing the goal has been abandoned. In the case of torture, you are pursuing the bad effect in order to achieve the good one.

You might think this distinction makes no moral difference. In other words, you might disagree with the weight of the tradition and the majority of scholars in the just war tradition. But if so, it is not a sign of impeccable scholarship to refuse to say why--to refuse to even acknowledge that there is such a basis for rejecting your thesis.


  1. Eric-

    I agree that this ticking time bomb scenario is poorly set up. Many of the "trolley" problems are bad in a similar way, because one is missing a great deal of information. The chances that torture is going to get what you want are very low. The chances that it will get the opposite of what you want- bad information- are very high. One's certainty of accomplishing the task at hand (by, say, throwing a person off a bridge) are, realistically, very low, mitigating any temptation to violate what are extremely important moral boundaries. Boundaries that themselves define the meaning of the very conflict/predicament one is in.

    OK, so on to the question of whether torture is justified against a certain perpetrator with certain knowledge of the intended harm, ignoring the issues above. I'd agree again that this is no license to torture at all. The point of (human) life is not simply to exist. So a calculus of one life vs two, or two thousand, etc. is, on any but the very simplest of conceptions, not a matter of arithmetic. I personally accept a system of justice where we take reasonable but limited precautions against various harms in a preventative sense, and take up investigation, remediation, and justice after the fact.

    This system becomes a train wreck if the purposefully limited scope of preventitive activities, which safeguard so much of our freedoms, is cast aside in haste by what are typically among the worst people to make such judgements, as case after case of police malfeasance shows. I thought the whole post-9/11 torture-fest that Hollywood went through (even infesting my favorite Star trek shows) was philosophical garbage, and highly repulsive.

    It is weird to hear that Harris draws on a war analogy here, since it fatally undercuts his case. I think the idea of pre-emptive war has been pretty well discredited of late, and the same logic applies to torture, which has a simliarly pre-emptive purpose. It is easy enough to impute all sorts of bad intentions and dangers to others. But attacking them pre-emptively is both practically disasterous, as it plunges the world into perpetual war in principle, and also loses the moral high ground. One reason WW2 was such a great war, (so to speak) was that our moral position was unimpeachable, having been attacked in the most unprovoked manner.

    As you note, collateral damage is never justified on its own terms as some kind of direct action. It is only "acceptable" by virtue of being inadvertant and minimized. That is what gets me about those who criticize our drone programs- they reduce collateral damage tremendously. So even if they are not perfect, they sure beat invading or carpet-bombing Pakistan, on any untilitarian or moral spectrum. They are a better option (though that in turn encourages over-use, another conundrum...)

    There.. I agree completely, so I guess I am not adding much. Sounds like a wretched book, though I will be interested in your defense of pacifism. Best wishes!

  2. Good post, and I don't have anything to add, except to point out that, interestingly, this just popped up in the news today:

    Not really news, since these reports have been circulating for years. But I found the timing of this to be interesting.

  3. I came across your post whilst researching for a post I did on the subject ( I think I would accept a similar conclusion to yours on this - that Sam Harris really hasn't done the necessary legwork of engaging with the literature on the subject. It's therefore extremely frustrating to see him engage on the subject of morality both generally (The Moral Landscape) and in this instance when it is patently obvious that he either regards most academic moral philosophy with contempt (probable) or doesn't feel it necessary to engage with academic literature as he's writing for a popular audience. Both scenarios are troubling because they completely undermine his thesis.

    1. Alex,

      Thanks for the link. I appreciate your contribution to this issue. I doubt any such efforts as ours will inspire Harris to admit to sloppy thinking, concede he may have been wrong, and lead him to deliberately rechart his course on the subject--but it may inspire successive qualifications that ultimately amount to the same thing (as he resists admitting so). And even if it doesn't even do that, it's important to cut through Harris's rhetoric to expose the inadequacies of his argument, so others aren't sucked in by poor reasoning.

  4. This comment has been removed by the author.

  5. Peter Chetkins concludes: "Theology obfuscates, Philosophy equivocates, Science elucidates". So very well demonstrated by this blog and the likes. Known to every body except writers like this one and Andrew Brown who envy his success, Sam Harris is all for compassion and how to maximize everybody's well being. So quit quoting him out of context and work hard on your own fame Eric.

  6. Peter Atkins concludes: "Theology obfuscates, Philosophy equivocates, Science elucidates". So very well demonstrated by this blog and the likes. Known to every body except writers like this one and Andrew Brown who envy his success, Sam Harris is all for compassion and how to maximize everybody's well being. So quit quoting him out of context and work hard on your own fame Eric.

  7. "You might think this distinction makes no moral difference."

    Correct. The use of an analogy whereby the people could be evacuated is just as "set up" and hence irrelevant as using the ticking bomb scenario. Both are implausible, extreme cases. At some scale of violence, "collateral damage" is unavoidable and hence "intended." The only thing not intended is the particular identities of the victims. Using the "intent" argument, we could say that the atomic bombing of Hiroshima was justified as an attack on infrastructure. Evacuate the people and we still destroyed a valuable harbor city and military installation. So all the people are collateral damage. I want to make an omelett, but don't want to break eggs. So breaking the eggs isn't my fault, even if I intentionally make an omelett. That can't be the best argument out there. I hope it isn't. I hope I'm straw-man-ing it.

    "In other words, you might disagree with the weight of the tradition and the majority of scholars in the just war tradition."

    Perhaps. It's been 25 years since I read any of that particular scholarship. I'll have to revisit it. But as portrayed here, in broad strokes, I don't find it convincing.

    In torture, the victim is far more likely to be the real target. The harm done (in scale and kind) is far more controllable than when exploding ordnance in an urban setting. This side of pacifism - which I reject, for the time being at least, for other reasons - I see little reason to reject torture. Like Harris, I don't like having this view and hope that someone will soon argue it away.

    I see the intuitive power of "intent" in other situations (the difference between murder and self-defense is, in essence, _only_ intent - or the lack of intent is the difference between evil and neglegance), but my mind doesn't make the jump to war in this case.