Philosophers like to talk about what they call hard cases, of which Tim the terrorist is an example. One reason they like to talk about them is because they think that hard cases sort consequentialists, who think that the only thing that ultimately matters about any action is its consequences, from non-consequentialists, who deny this. It's not disputed by either side that doing the horrendous thing in the hard case is acting badly. What is disputed, often, is whether acting badly in such cases is acting wrongly. (Hence my title.)

The routinely-advertised consequentialist reaction to hard cases, torture cases for instance, is ‘If it produces best consequences then you should do it – so go ahead and torture. And anyone who refuses to torture, like those benighted non-consequentialists, is just not being practical/humane’.

The routinely-advertised non-consequentialist reaction to torture is ‘Torture is on our absolutely-forbidden list, and this is torture – so don't do it, no matter what. And anyone who agrees to torture, like those benighted consequentialists, is someone we do not wish to argue with; he shows a corrupt mind.’ (Here I am echoing the words of the famous Catholic philosopher Elizabeth Anscombe.)

I think these familiar standard reactions on both sides are inadequate. I want to suggest that in fact, clear-thinking and careful consequentialists and non-consequentialists won't generally disagree about what to say about at least many of the hard cases. They will typically reach the same verdicts. ‘But for different reasons?’ Not even that necessarily; in fact, I suggest, the more clearly and carefully they think about the real detail of such cases, the more they will tend to agree not only on their verdicts but on the reasons why they give those verdicts.

The consequentialist wants to move us straight on to saying ‘Yes, go ahead and torture Tim – it's the only way to save the innocent civilians’, and to attack the non-consequentialist for denying this. The non-consequentialist wants to move us straight on to saying ‘You must not torture Tim even then’, and to attack the consequentialist for denying this. But hold on a minute. Must the consequentialist say that?

One familiar reason for doubting it is this: it's very unlikely that we will ever really have quite this much knowledge, quite this securely. To know that a given method of achieving an end is certainly the only method is quite hard to know. It might be the only method you can think of. But maybe that just shows you should think harder – and the worse the alternative you find yourself contemplating, the harder you should think. Consequentialists do not have to ignore this sort of point (even if they often do ignore it in practice). There's nothing in consequentialism to stop a consequentialist from accepting the rule, or rule of thumb, that says ‘If the only way you can think of to prevent a disaster seems to be to act very badly, then think harder’. Indeed it is easy to see how consequentialism itself might recommend such a rule.

Again, to know that torture is the only way of getting Tim to tell the truth is not only hard to know. It is probably impossible to know – since we can't know what isn't true, and there is good evidence that torture is not a reliable method of getting people to tell the truth (as opposed to getting them to say whatever they think will stop the torture). There's nothing in consequentialism to stop the consequentialist from saying ‘Even to achieve the very best ends, we shouldn't use means that are predictably and inherently very unreliable, especially not where those means predictably and inherently will have a lot of nasty side-effects’. Indeed, consequentialism itself familiarly tells the consequentialist to say that.

Thirdly: to know that a given action will have the good consequences it's forecast or meant to have, and no other bad consequences which will outweigh it, can be very hard. (Maybe one of the ‘innocent civilians’ whom Tim's bomb threatens is Hitler, etc.) There's nothing in consequentialism to stop the consequentialist from saying that ‘The actual calculation of consequences is so dizzyingly difficult that we can hardly ever make direct use of consequentialism, true though it is, as a deliberative method; in most, perhaps all, practice, it's much better to follow commonsense morality than to try and calculate for yourself.’ Again, consequentialism itself, in many of its most important and familiar forms, tells the consequentialist to say just this.

But suppose we can fix these three problems. We've somehow found good reason to be confident that we have not missed any important alternative methods of getting Tim to talk. And we've somehow justifiably concluded that in these circumstances torture is just the thing, and the only thing, that will make Tim speak the truth to us. And we've somehow fixed up all the other sorts of uncertainties about what will happen if we torture Tim that might be relevant to the situation. What now? Should we go now ahead and torture Tim, according to the consequentialist?

That still doesn't follow. For when deliberating about how to act badly, we need to deliberate with our imaginations engaged. A good question to apply our imaginations to the torture case is: what sort of ‘torture’ are we talking about here?

Contrast these three possibilities: Tim's arm could be twisted, or Tim could be water-boarded, or Tim could have electrodes attached to delicate parts of his anatomy and have electric shocks put through them. Don't be distracted here by the question whether these are forms of torture (I'll come to that later). Just imagine what it is like for these things to be done.

I don't mean by that ‘Imagine what it is like for Tim’ – I'm not making a point about the Golden Rule here, or about putting yourself in Tim's shoes, or empathy, or anything like that (cardinal though those uses of the moral imagination undoubtedly are). My point is rather: what has the context of action got to be like, for any of these three things to happen to Tim.

Notice this: Tim's arm can be twisted pretty spontaneously. That can happen on the spur of the moment, as a desperate interrogator comes to the end of his tether, snaps, and finally assaults Tim physically. The only contextual factor that needs to be in place for this to happen is that the interrogator must either think that, in general, doing things like twisting Tim's arm is morally permissible, or be dispositionally and motivationally close enough to thinking this, that he is not restrained from doing it in the particular case.

By contrast, the interrogator is not going to water-board or electrocute Tim unless either he already has the equipment to do so, or else he is prepared to improvise such equipment on the spot.

But if he already had the equipment – where did it come from? When did he buy it? From whom, and for how much, and out of which budget? Who signed off this purchase? Are we to imagine that the interrogator or some colleague of his calmly filled out an entry in the accounts book that is headed ‘Torture equipment’ (or ‘rendition equipment’ or ‘interrogation techniques’, if you prefer the current euphemisms)? Should we picture the interrogators wondering for months or years if they were ever going to get to use their purchases?

Or if he is now prepared to improvise such equipment – what is that going to be like, as it happens? Is the interrogator going to take off his belt to hold Tim down on the water-board, or get a screwdriver out to turn a kettle-flex into an effective electrode? Should we imagine a hiatus in proceedings, while Tim perhaps watches the means of torturing him being prepared?

To get a proper grip on what we are saying if we say, ‘Yes, the interrogator should go ahead and torture Tim’, we have to apply our imaginations to answering nasty questions like these.

The point here is not to paint an emotive picture of something extremely horrible. That is worth doing too, of course. After all, the emotive picture can be perfectly accurate: Tim's writhing and screaming, his loss of consciousness or physical self-control – all of that does, in point of fact, happen when he is tortured. However, the ethical method I am trying to apply here is not supposed to be an emotivist/expressivist one, about the stoking-up of feelings. It is supposed to be an intuitionist one, i.e. one which takes the idea of moral seeing as seriously as it can be taken.

The main point I want to make here is not about responding to an emotive picture, but about seeing and understanding an intentional structure. Or rather, two intentional structures. In one, the interrogators and others who work with them have been planning this procedure that they are now carrying out on Tim for months. In the other, they are now prepared to leave Tim on his trolley for five or ten minutes while, in front of his eyes, they modify various sorts of artefacts that they have available in order to torture him with them.

Behind the first of these two intentional structures there obviously lies a willingness to do, and to institutionalise doing, things like this not just to Tim, but to lots and lots of people who (the interrogators must have envisaged) are possibly going to be in situations like Tim's. Behind the second of them – the improvising of torture-equipment – there presumably lies a less set and steady will or choice to be torturers. But there still lies what a court of law would recognise as ‘malice aforethought’.

That makes the cases of deliberately buying in torture equipment, and of improvising it when needed, significantly like each other. Since waterboarding and electrocuting captives are both procedures that require such intentional planning, it also makes waterboarding and electrocuting significantly like each other. And it makes all of these procedures significantly unlike spontaneously grabbing a captive and twisting his arm to get him to talk.

The difference in question here, if you ask me, is one of the key differences between torturing captives and not torturing them (even if you are, as the arm-twister perhaps is, putting pressure on them in morally, not to mention legally, inappropriate or impermissible ways). I do not mean to deny that there could be a ‘one-off’ torturer. But I do mean that what is essential to torture, in its most typical forms, is not just that it uses procedures that cause their victims pain, distress, fear, anxiety, indignity, or some or all of these, but also that it institutionalises these procedures.

Thus the consequentialist who assents to torturing Tim – where that means doing something to him like waterboarding or electrocuting him, rather than something like twisting his arm – is not just assenting to Tim's being treated in this way as a one-off. If torture (in this sense) happens, it will happen institutionally. And to assent to torture will be to assent to the existence of this institution; to the intentional structure that consists not simply in torturing Tim, but also in marking out in advance your own preparedness to torture anyone like Tim in any circumstances like these that might come up (in the case where the torture-equipment is pre-bought), or your own preparedness, here and now, to make a plan to torture him ‘with malice aforethought’ (in the case where the torture-equipment is improvised on the spot).

But there are obvious consequentialist reasons to be against intentional structures of either sort. The consequentialist can say that if intentional structures of either kind have come about, then the consequences of these structures are likely to be very seriously evil – so evil that it will take far more in the way of countervailing reasons to out-balance this evil than you might expect. And the consequentialist is quite right about this.

There are also, of course, obvious non-consequentialist reasons to be against intentional structures of either sort. The non-consequentialist can say that if intentional structures of either kind come about, then that expresses a chilling violation of justice and human dignity which constitutes a serious evil – so evil that it will take far more in the way of countervailing reasons to out-balance this evil than you might expect. And the non-consequentialist is quite right about this.

The consequentialist and non-consequentialist can thus agree that torturing Tim is a serious evil. All they disagree on is the reasons why it is a serious evil. But they wouldn't even disagree on that if the non-consequentialist were prepared to allow the consequentialist's appeal to the consequences of torture, and if the consequentialist were prepared to allow the non-consequentialist's appeal to the expressive value of torture.

What they disagree about is just what forms of explanation of wrongness (and rightness) are permitted. But they only disagree about that because each of them insists on only permitting his own proprietary form of explanation of rightness and wrongness. And this insistence is only necessary if you believe, as systematic moral theorists like consequentialists and Kantians characteristically do believe, that there has to be (ultimately at any rate) just one form of explanation (consequentialist, or Kantian, or whatever) for everything in ethics.

When we engage our imaginations in thinking about the hard case of torture, and ask exactly what we should imagine is involved in torture, we see that (whether we are consequentialists or non-consequentialists) there is a whole range of distinctions and differences that we might want to observe in the moral phenomena. We might assent to Tim's having his arm twisted by his interrogators – acting spontaneously, and going beyond normal interrogation procedures under the stress of a dire emergency – where we would not assent to Tim's having his arm twisted by his interrogators where this action is part of their normal routine. And we might assent to Tim's having his arm twisted in either way, where we would not assent to his being electrocuted or waterboarded by an interrogator who improvises the means to inflict electrocution or waterboarding on him. And again, we might assent to improvised electrocution or waterboarding where we would not assent to fully institutionalised electrocution or waterboarding. Those who have acted as apologists for the Bush presidency's practices of torture in Abu Ghraib and other places are at least right about this: torture is not a unitary moral phenomenon, and there is more than one way for it to be a bad thing to do, as well as being more than one way for it to be a wrong thing to do.

Saying that any of the above forms of procedure is morally all right is different from saying this about any of the others.

Saying that any of them is morally all right is also different from thinking that it is morally all right, but (as we might) refusing to say this. And thinking it all right but not being prepared to say so is different again from neither saying nor thinking that a given practice is morally all right – yet finding yourself performing that practice spontaneously, or under the stress of circumstances, and not thinking afterwards that you acted wrongly.

I have been concerned with how, typically, practices like torture, strictly so called, necessarily involve forms of intention that imply institutionalisation. Notice also from this discussion how speech about torture, or in fact any practice, can also be a subtle kind of institutionalisation of it. So even can thinking about it (as I am doing here).

This point presumably is what Anscombe, as quoted above, was getting at when she insisted that even to consider certain possible courses of action is morally corrupting. The person who spontaneously twists Tim's arm is perhaps a case in point. To do that in response to a crisis, without any kind of prior thought or planning, is morally questionable. But it is not as morally questionable as doing it after planning it or contemplating doing it. And to contemplate doing it as a philosophical example is, after all, one way of contemplating doing it.

I don't say that this rules out even thinking about such courses of action. I do say that there is a paradox in thinking about them or giving them imaginary assent. And there is a related paradox about publicising our willingness to do such things. It may be that, for some kinds of action, the good person who is only considering philosophical examples will always indignantly protest his refusal to do them – yet in sufficiently dire circumstances, will nonetheless end up doing them.

My main point is that none of these thoughts about intention and institutionalisation are thoughts of a kind that only a non-consequentialist or a consequentialist can access. They are thoughts that are available to anyone who does not block them out because of moral-theoretical or other preconceptions. That tends, I think, to show that what distinguishes good ethical thinking about torture from bad ethical thinking about torture is not adherence to any particular moral theory. Rather, the difference is (at least partly) the deployment of the moral imagination in such thinking. The answer, or one answer, to the title's question how we should deliberate about acting badly, is with our imaginations engaged; and this answer is equally available to non-consequentialists and consequentialists.

There is a second answer to the title's question, and this too is also available alike to both non-consequentialists and consequentialists. This time the answer is: we should deliberate about acting badly with due attention to the first-personality of deliberation and action.

This is, this time, an exercise of putting yourself in someone else's shoes: it's about putting yourself in the torturer's shoes. Don't just ask whether he, ‘x’, some vague indefinite place-holder, should (be prepared to) torture Tim. Ask whether you, the real and concrete you, would (be prepared to) torture Tim. And as before, engage your imagination with the concrete realities involved; not in order to work up a froth of righteous emotion, but in order to see clearly just what those realities are. Also, keep still in mind the points about what you or those around you must have been institutionally committed to for months, if, for instance, a procedure of waterboarding or electrocuting Tim with specially-designed equipment is available to you.

When you do this, it may be that you will find that there are some things that you can imagine yourself doing to Tim without feeling at all bad about them; putting him under heavy psychological pressure to say what he knows, for instance. That is, after all, an interrogator's job, and it is (unfortunately) hard to imagine a world in which we never need interrogators. There may be other things you can imagine doing to him which you know you would do in certain circumstances, but which you would feel very uncomfortable about after the event: twisting his arm, perhaps, under the stress of crisis. Other things again you may well think you simply couldn't live with yourself for doing to him, because, though spontaneous, they are just too horrible.

And then there will be kinds of procedure – those which by my definition above count as torture properly so called – that you could inflict on him where inflicting them would commit you to assenting to a whole institutional choice, not just to torture Tim, but to be torturers in general, as occasion demands. What conclusion will you draw about these, at the level of imaginary assent that is involved in thinking about such philosophical scenarios?

You might well conclude that, if the external threat is bad enough, then perhaps your assent to some horrors is demanded; yet – you might say – giving your assent to those horrors is not just the breaking of the victim of the torture; it is the breaking of you, the torturer, as well.

Or else, you might say that such a situation is a situation where there is, quite literally, nothing to be done: every option presents itself as not-to-be-done, and no option presents itself as to-be-done. Of course it is the job of ethics to find the practicable, to compromise with hard circumstances, to make tough choices in a tough world. That doesn't mean that we are obliged to think that there is, in every conceivable situation, even one ethically practicable option; any more than thinking that the point of the activity of chess-playing is to make moves commits us to thinking that there can be no chess position in which no further move is legally playable.

Or you might say one of these things about some horrible choices, and the other thing about some others.

And note well: you can make either response whether or not you are a consequentialist. A non-consequentialist will say something like: ‘These responses represent the morally-required responses of a virtuous agent to certain terrible dilemmas’. A consequentialist will say something like: ‘Things go best if agents have the sort of dispositions and reactions that these responses represent’. Here too, the choice between the consequentialist justification of responses like these, and the non-consequentialist justification of them, is a false choice. There is no reason why both can't be true – beyond the consequentialist's insistence on his own form of explanation to the exclusion of the non-consequentialist's, and vice versa.

This is why I say that, to deliberate well about acting badly, we need to engage our imaginations to think through what the reality of acting badly will actually involve; and we need to do this first-personally – I as deliberator need to understand that the reality of acting badly will be, necessarily, the reality of me acting badly.

One moral of the argument is that certain kinds of acting badly – torture, for example – should not be thought of as controversial as between consequentialists and non-consequentialists, or a good way of sorting them from each other, or of criticising consequentialism (if you're a non-consequentialist) or of criticising non-consequentialism (if you're a consequentialist). Non-consequentialists don't have an automatic monopoly on principled opposition to torture and similar bad ways of acting, and consequentialists don't have an automatic monopoly on pragmatic flexibility about torture and similar bad ways of acting. It is only likely to seem that way if we are not engaging our moral imaginations fully. When we do engage them, one set of differences – the differences between particular instances that fall or might fall under the label ‘torture’ – will stand out, while another set of differences – the differences between systematic moral theories – will come to look much less important.

And so a second moral is that there may well less than there often seems to be to separate honest, serious, and clear-thinking consequentialists from honest, serious, and clear-thinking non-consequentialists. The differences may well come down to one side's insisting that all badness and goodness, rightness and wrongness, has to be ultimately explicable in terms of welfare-promotion or the like, while the other insists that all badness and goodness, rightness and wrongness, has to be ultimately explicable in terms of fairness or universalisability or the like. This insistence, on both sides, on the uniformity of moral explanations seems to have no real rationale other than the urge towards theoretical tidiness. It stands in the way of progress in ethics, and we would do much better to drop it.Footnote ¹