Monday, September 19, 2011

The Ethics of Torture (Part One)

Several years ago — several more than I care to remember — I had to write an assignment on the legality of torture. It was for a class in international human rights law. As a result, it focused largely on the legal side of things, but even back then I was trending towards the dark side and so I included some discussion of the philosophical arguments as well. I can’t remember exactly what I wrote, or what my precise conclusion was, but I think it was down-the-line absolutely anti-torture, rejecting the legitimacy of torture even in ticking bomb style scenarios.

I wrote that essay at a time when my political and philosophical views were decidedly more naive and ideologically driven than they are now, and at a time when the torture debates were to the forefront of the public consciousness. As a result, I had a tendency to rush toward the conclusion that other members of my ideological clique seemed to adopt. Sad, but true.

Anyway, since I feel more philosophically mature these days (whether I actually am is a separate question), and since I have to teach a class about arguments in law and ethics pretty soon, I thought I might try to revisit the whole torture debate. As is typical on this blog, I’ll use some articles to guide my analysis, feeling free to speculate and expand upon their contents when I see fit.

First-up for consideration is an article by Catherine McDonald entitled “Deconstructing Ticking Bomb Arguments”, which seemed about as good a place as any to start.

1. The Structure of the Ticking Bomb Argument
Torture is the inhuman and degrading infliction of severe pain and suffering on another human being. It is for many the epitome of the morally bad act. This is a proposition which can apparently be justified on any number of grounds, e.g. “pain is intrinsically bad” “it is contrary to human dignity” and so forth. But is it always and everywhere morally wrong? This is what the ticking-bomb scenario tries to disprove.

The ticking bomb scenario will, no doubt, be familiar to many readers of this blog. Its essential details are as follows:

Suppose there is a large bomb that has been planted in a highly-populated location. Suppose this bomb is due to go off in the near future. Suppose you have in your custody someone who knows where the bomb is planted, but who will not share this information. Suppose there is no other source for this information. Suppose receiving this information could help you prevent the bomb from going off. Now ask yourself, would it be legitimate to torture the person?

The suggestion is that it would be. The reasoning here is based on two key ideas: (i) the quasi-dilemmatic nature of the decision problem posed by the ticking bomb scenario; and (ii) the correctness of moral consequentialism, i.e. the view that one ought to act so at to achieve the best possible consequences. The ticking-bomb scenario is dilemmatic because no matter what one does (torture, or not-torture) there is a morally undesirable outcome. (Illustrated below)

But it is only quasi-dilemmatic because in a true dilemma both outcomes would carry equal moral weight; in the ticking bomb scenario the whole point that one outcome is less bad (the torture) than the other (the deaths of the bomb victims).

This is where the consequentialist principle kicks-in: if one outcome is less bad than the other, then one ought to pick the lesser of the two evils. This leads naturally to the conclusion that torture is justified in this kind of scenario. But we need to be careful about this. Probability estimates need to taken into consideration: only if we think that torturing the person has a reasonable chance of leading us to the required information, would it really be justified.

To put all this in argumentative terms:

  • (1) For any moral decision problem with possible choices C1…Cn, one ought to pick the C that achieves the maximum balance of good over evil.
  • (2) In the ticking bomb scenario one has two possible choices, either one: (a) tortures the person in custody or (b) one does not torture the person in custody.
  • (3) If one tortures the person in custody, one has a reasonable chance of obtaining information that could prevent a large number of deaths; if one does not torture the person in custody, a large number of people will die.
  • (4) Preventing a large number of deaths while torturing one person is less evil than allowing a large number of people to die.
  • (5) Therefore, in the ticking bomb scenario, one ought to torture the person in custody.

No doubt hardcore non-consequentialists will reject premise (1). They will balk at the notion that you should perform a prima facie immoral act merely because it leads to better consequences. Some of them will stick with the absolute prohibition on torture as a result. But because of the quasi-dilemmatic structure of the scenario, these individuals must make their peace with the other morally undesirable outcome. Whether they can really do so is debatable.

Value pluralists are likely to reject premise (4) by arguing that such a comparative exercise cannot be carried out. While I have some sympathy for this position, I don’t think it carries much weight in the present context: if the outcomes can’t be compared in the manner suggested by the argument, then surely we are faced with a tragic choice, i.e. one in which either outcome is morally permissible?

Others will reject premises (2) and (3) of the argument, the ones setting out the factual conditions of the ticking bomb scenario. And it is with these others that McDonald makes her stand.

2. False Assumptions
McDonald’s attempt to “deconstruct” the ticking bomb argument is based largely on the empirical naivety of its proponents. Such proponents recklessly presume that, in the real world, the assumptions of the ticking bomb argument could hold true. McDonald thinks that the real world tells a different tale. Consider the following real-world incidents:

The Shooting of Jean Charles de Menezes : The ticking bomb scenario presumes that we can reliably identify those who know where the bomb is, in advance of the bomb going off. In reality, in preemptive operations of this sort, law enforcement officials c err in the judgment of who is and is not a threat. For example, the London police shot Jean Charles de Menezes multiple times in the back of the head as he was boarding an underground train. They did so because they were convinced he was involved in a terrorist plot (this was soon after the 7/7 bombings). As it turns out, he was innocent. If we were to allow torture, similar errors would take place. Are these errors an acceptable price to pay?
The German Kidnapping Case : This case is often cited by proponents of interrogative torture. It involved a kidnapper named Magnus Gafgen who disclosed the location of a child he had kidnapped after being threatened with torture by the German police. Unfortunately, the child was already dead, but this hasn’t deterred those who wish to argue that torture may occasionally provide important information. The problem with those people, as McDonald points out, is that the example does nothing to prove the informational utility of torture: torture was never actually used in this case, it was merely threatened.
The Khalid Sheikh Mohammed Case: It is claimed that many months of torture in Guantanamo bay led Khalid Sheikh Mohammed to confess to involvement in numerous bomb plots. This again suggests the informational utility of torture, right? Wrong says McDonald. For one thing, the confessions were of dubious veracity (she cites no source for this but I have no reason to disbelieve her), and for another the torture took place over many months. This is not the kind of scenario envisaged by proponents of the ticking bomb argument.

Other cases are discussed by McDonald, but I’m going to cut things short here. What do cases like these really prove? How do they help to undermine the ticking bomb argument? The answer comes in three parts.

The first part is that, contrary to the ticking bomb argument, torture is not an effective means of obtaining information. Thus, premise (3), which claims that torture has a reasonable chance of providing you with the relevant information, is false. So:

  • (6) The information obtained through torture is often dubious, thus torture would not provide you with a reasonable chance of obtaining information that could help prevent a large number of deaths.

There are serious questions to be asked about this challenge to premise (3). For starters, in a situation like the ticking bomb scenario, it’s not clear that you really need to have even a reasonable chance of success: any chance at all may be enough to justify the use of torture. I’m going to leave these kinds of criticisms to the side for now since I hope to cover them in more detail in a future lecture.

The second part is that, contrary to the ticking bomb argument, you are unlikely to face a simple choice between torture and no-torture. Other, arguably more effective options, will be present. This is proved by the German kidnapping case — in which threats were enough — and by other analyses which suggest that traditional interrogation methods are just as effective (if not more so) than torture. This gives us the following direct challenge to premise (2):

  • (7) In any real-world ticking bomb scenario, one will have more than two choices. Indeed, one will likely have ways of obtaining the requisite information that would be more effective than (and at least no worse than) torture.

The third part is that the consequentialist calculation — stated in premise (4) — is flawed. It focuses on the short term benefits of torture; it neglects the long term costs. If torture is ever thought to be morally justifiable, and this thought gets a foothold amongst those in authority, the long-term costs could be great. In every society in which it has been practiced, torture has led to a breakdown in trust, and an increase in fear and intimidation. Thus:

  • (8) The long-term costs of torturing the one person outweigh the short-term benefits of preventing deaths.

Proponents of the ticking bomb argument will respond by saying that it’s ridiculous to assume that their argument would lead to the widespread use of torture: they are only talking about torture in very limited circumstances. McDonald describes this response as being “risible”. She says that “once instituted in one circumstance, the practice of torture invariably expands.

McDonald goes on to ask why, despite it obvious flaws, the ticking bomb argument has proved so seductive. Although this is an interesting topic, I’m not going to cover it here. Instead, I’m going to move on in the next part to consider alternative “pro”-torture arguments. In particular, the arguments of Uwe Steinhoff. Stay tuned.


  1. As usual, this looks like it's going to be an interesting series. One suggestion I have (in case there are papers exploring this) is to look at a justification of the use of torture in the ticking time bomb scenario based on double-effect considerations.

    A comment about McDonald's argument: there is a tension in the points McDonald makes -- if it's true that torture is a slippery slope and will be widely adopted once used successfully, then it seems the same should be said about the proposed alternative of merely threatening torture. If McDonald is right that this is even more effective than torture, then it seems that the slipperiness of adopting threats of torture as a policy would be higher than that of actual torture (unless there is a sufficiently strong human bias in favor of torture to counteract this). But once the practice of merely threatening torture becomes widespread, the threats will cease to be credible and the effectiveness of the policy will deteriorate. If we are to take into account long-term effects of torture, we'd have to do the same for any proposed alternatives.

  2. Unthinkable (2010) presents this scenario on a movie if you want feel more "inside it" then I would recommend it.

    Well my take on it is that I would approve the torture if there is no other way arround ... it's sad but in that case have to be done. I don't think (8) is true, in cases where (7) is true then of course we will try to do otherwise but in cases like Unthinkable I don't think we have another feasible alternatives and about (6) well it deppends, if the torturer 'knows what he's doing' I think valid information can be obtained, and as I said if there is no other alternative only.

  3. I think I'm missing something basic here - is the argument about whether torture in practice is a poor moral choice, or is it whether philosophically it is always and without exception bad? It appears to me that essentially the above objections merely meant adding more caveats to the scenario, distancing it further from reality. But they did not undermine the point that as a matter of principle, one cannot say that torture is always to be prohibited. So the argument becomes one about practical matters, not principles.

    And when it comes to practical matters, I'm sorry but I'm very doubtful of the anti-toture arguments. I strongly suspect that if you'll look at the records of people who actually did use torture regularly, but not dogmatically, as part of their intelligence gathering apparatus you will find that it was quite effective. The ticking bomb scenario is cute, but look into more realistic scenarios like uncovering terrorist cells and insurgent organizations or their plans and I'm quite certain that you'll find that taking hostages, kidnapping, threats, and yes also torture are all effective tools in the arsenal. I suspect the main reason the West finds it isn't useful is that it isn't willing to be as systematic and to combine it with other modes of intelligence gathering and pressure; the West uses torture "amatourishly", so gets poor results.

    I do suspect, however, that we're heading towards more scientific interrogation. At the first stages you would augment normal interrogation with real-time neural scans, and possibly drugs, to lower psychological barriers, create trust where it shouldn't be, detect efforts to lie, and so on. Ultimately, you would simply "upload" the person's brain into a computer database, crunch some numbers, and draw out any information you would want. That will make torture rather redundant. I'm less worried about the latter option, as I suspect it's centuries into the future (although I'm fearful that post-death scanning might be doable in a few decades, leading to, ehm, another option...), but I suspect something like real-time neuronal imaging and advanced drug use may already be feasible if not in actual use.


  4. Interesting comments all round here.


    I agree on the credibility issue and I've mentioned this in part two (crediting you for the observation). But I think I may have misled you as to McDonald's argument. As far as I can make out, she is not claiming that threatening torture is more effective than torture in all cases, she is merely pointing out that the Gafgen case doesn't prove what it is thought to prove. She does, however, claim that standard interrogation methods are likely more effective than torture - that's what premise 7 is saying. She does have some references for this, but I excluded them from my discussion.


    Thanks for the recommendation. I'll look into it. As for the truth of (8), that's a really tough issue to decide. It's a classic problem facing all consequentialists: how far into the future do you look when assessing consequences? I'm pretty sceptical about long-term assessments myself, but I just don't know what the best consequentialist response to this kind of problem really is. That's why Steinhoff's argument (discussed in part two) is interesting: it tries to avoid the consequentialist calculus.


    I think McDonald's argument is really an objection to the use of torture in practice, as opposed to in principle. But as you point out, her argument isn't necessarily effective anyway.

    You may also be interested to know that the forensic and counter-terrorist uses of brain-scanning is one of my research interests. It's also the only thing I've managed to publish on to date. I have a couple of pieces out on the use of brain-based lie detection evidence in court, and I'm currently writing a new one on guilty knowledge tests, which I'll be looking to publish next year sometime.

  5. John -

    That's cool; I'll look up those publications.