Philosophical Studies (2007) 133:257–283 DOI 10.1007/s11098-005-4603-4
Ó Springer 2006
ALISON HILLS
INTENTIONS, FORESEEN CONSEQUENCES AND THE DOCTRINE OF DOUBLE EFFECT
ABSTRACT. The difficulty of distinguishing between the intended and the merely foreseen consequences of actions seems to many to be the most serious problem for the doctrine of double effect. It has led some to reject the doctrine altogether, and has left some of its defenders recasting it in entirely different terms. I argue that these responses are unnecessary. Using BratmanÕs conception of intention, I distinguish the intended consequences of an action from the merely foreseen in a way that can be used to support the doctrine of double effect.
1.
INTRODUCTION
When you formulate a plan and put that plan into action, you do many things intentionally. For example, when you execute your plan to go shopping to buy some food, you intentionally get in your car, drive to the shops, buy the food and return home. It is natural to divide the things you do intentionally into those that you intend, and those that you merely foresee but do not intend.1 The intuitive difference is that you intend what you chose the plan in order to bring about: if you choose a plan in order to bring about E, you intend E as an end; if you decide to bring about M in order to bring about E, you intend M as a means. In carrying out your plan, you may expect to bring about some consequence F, though you did not choose the plan in order to bring about F; then F is merely foreseen, not intended. For example, you devise your shopping plan in order to get food: getting food is your intended end. You foresee, but do not intend, that you will wear down the tyres of your car, for you did not choose to go out in order to wear down your tyres.
258
ALISON HILLS
According to the doctrine of double effect (DDE), there is a distinction between the intended and the foreseen consequences of an action, and that distinction is morally significant: it is morally worse to intend some harm, than merely to foresee bringing it about.2 According to the absolutist version of DDE, it is morally forbidden to intend to kill, but it is sometimes morally permissible to foresee killing.3 In this paper, I am not going to discuss whether there is a morally significant difference between intending and foreseeing harm (I have defended this claim elsewhere).4 Instead, I want to consider a prior issue: is there any distinction at all between what is intended and what is foreseen? This issue is important independently of its connection with DDE. It would be interesting if we could distinguish the intended from the merely foreseen consequences of an action, even if it turned out that the DDE was not correct, that is, even if the distinction proved not to be morally significant. But of course, given that it may carry moral significance, it is even more important to know whether the distinction really is genuine. In this paper I will try to show that we can distinguish the intended from the merely foreseen. This problem is usually considered to be the most serious difficulty for the doctrine of double effect (in either its absolutist or non-absolutist form),5 and has even led some who defend it to recast it in entirely different terms.6 But to reject either version of DDE on these grounds is too hasty. I shall argue that the distinction between what is intended and what is merely foreseen as drawn by Michael Bratman (1999) is adequate to support DDE. 2.
INTENDED AND FORESEEN CONSEQUENCES
We need to see if a clear distinction between the intended and the foreseen consequences of an action can be drawn, so that we can go on to consider whether DDE is correct and the distinction is morally significant. It is worth distinguishing two questions here:
INTENTIONS, FORESEEN CONSEQUENCES, DOCTRINE OF DOUBLE EFFECT
1.
2.
259
Is it possible to draw a clear distinction between what is intended and what is merely foreseen in a way that reflects our intuitive grasp of the distinction?7 In practice is it possible to distinguish what another agent intends from what she merely foresees?
These two questions are quite separate: (1) is a metaphysical question about the difference (if any) between what is intended and what is foreseen; (2) is an epistemological question about whether we can tell what another person intends and what she merely foresees. There might be a perfectly clear intention/foresight distinction, even if it were nearly impossible to tell what another agent intends. Both questions are relevant for DDE, but in very different ways. If there is no real difference between what is intended and what is foreseen, then obviously there cannot be any kind of morally significant difference between them. On the other hand, if there is a clear distinction between intentions and foreseen consequences, then DDE may be correct. An agent who intends to kill another person may be morally worse than another agent who merely foresees killing (if the absolutist version of DDE is correct, it may be morally forbidden to intend to kill).8 This is so even if we cannot distinguish in practice between someone who intends to kill and someone who merely foresees killing. Suppose that it is in fact nearly impossible for us to tell what another agent intends. Then the DDE may be correct, but it is nearly practically useless: we cannot use it to make moral assessments of other agents unless we can tell what they intend to do. If there is no distinction between intentions and foreseen consequences, then DDE cannot be true. If there is no way of telling what other agents intend, then DDE is of little practical significance. In order to defend DDE fully, we need to show first, that there is a clear distinction between what is intended and what is foreseen and, secondly, that we can make reasonable judgements about what other agents intend and what they merely foresee.
260
ALISON HILLS
Bratman (1999) has given an answer to (1) by considering the nature of plans. If you carry out a chosen plan successfully, you will bring about both the intended and the merely foreseen consequences of your action (or at least you will raise the chances that these consequences occur). For example: STRATEGIC BOMBER: The Strategic bomber intends to win the war by destroying strategic military targets (e.g. military barracks and factories making military equipment) by dropping bombs on them. He is aware that he is very likely to kill many civilians who live near the factory.
If you successfully carry out a Strategic bomber plan, you will destroy the factory and kill some civilians. So why do many people say that you intend to destroy the factory but do not intend to kill civilians? The intuitive concept of intention is that you chose your plan in order to bring about the intended consequences.9 Bratman (1999) explicates this intuitive notion in the following way: in carrying out your plan, you are committed to the intended consequences in a way that you are not committed to the merely foreseen consequences. You intend that a consequence C occurs if you 1. 2. 3.
Use means-end reasoning to work out how to make C occur. Constrain your further intentions to ensure that they do not conflict with making C occur. Track your success at making C occur.10
In short, you intend that C if in carrying out your plan, you are committed to making C occur. As a Strategic bomber, you chose your plan in order to destroy the factory, you are committed to destroying the factory. You use means-end reasoning to work out how to destroy it, e.g. you work out which bombs are most likely to destroy it, and which direction to approach the target will be most effective. You constrain your further intentions so they do not conflict with destroying it, e.g. you constrain your intention of not carrying out dangerous daylight raids if you need to do so to hit the factory. And you track your success at destroying it, e.g. if you miss the target with your first attempt, you will try again.
INTENTIONS, FORESEEN CONSEQUENCES, DOCTRINE OF DOUBLE EFFECT
261
You do not intend that C, if in carrying out your plan you are committed to reducing the chances that C occurs, or if you are neither committed to raising or lowering the chances that C occurs, that is, you are indifferent to whether C occurs. As a Strategic bomber, you are not committed to killing civilians. You do not use means-end reasoning to work out how to kill civilians, e.g. you work out which bombs are more likely to destroy the factory, not which bombs are more likely to kill many civilians. You do not constrain your further intentions so that they do not conflict with killing, e.g. if you could raise the chances that the area around the factory were evacuated, you would not immediately reject this option. And you do not track your success at killing, e.g. you will try again if you fail to destroy the factory, not if you fail to kill civilians.11 You may be indifferent to whether or not you kill civilians, or you may be committed to reducing the chances of killing civilians, though only insofar as doing so is compatible with carrying out your intention of destroying the factory.12 By contrast, the Terror bomber does intend to kill: TERROR BOMBER: The Terror bomber intends to win the war by lowering enemy morale by killing enemy civilians by dropping bombs on them.
The Terror bomber uses means-end reasoning to work out how to kill, e.g. he will work out which bombs are most likely to kill many civilians. He constrains his further intentions so that they do not conflict with his killing, e.g. he would reject an option that would raise the chances of an evacuation of the area. And he tracks his success at killing: if he initially fails to kill as many civilians as he would like, he will try again. Though both the Terror bomber and the Strategic bomber choose to raise the chances that civilians die, and both may even kill the same numbers of civilians, only the Terror bomber is committed to killing: only he intends to kill. If DDE is correct, Terror bombing is morally worse than Strategic bombing.
262
ALISON HILLS
Of course it may not be obvious what any particular agent intends: it may not be clear what her plan is, what means-end reasoning she undertakes, what she would count as success when she monitors what she does. She may have some of the attitudes required of intention without the others in which case it may be impossible to determine exactly what she intends.13 In the same way, it can be difficult to determine precisely the content of another personÕs beliefs. But just as we need not be sceptical that there is a difference between what a person believes and what she does not, we need not be sceptical about the distinction between what a person intends and what she merely foresees. We have now answered question (1): The distinction between the intended and the foreseen consequences of an action can be drawn in a way that fits with our intuitive conception of the difference, though it may be indeterminate what a given agent intends. But there are two concerns about using the distinction as drawn in defence of DDE. First of all, we need to answer question (2). If the distinction is to be used in support of DDE, we need to show that DDE is practically useful. We need to be able to distinguish in practise between someone who intends kill and someone who does not, for example; otherwise we cannot assess the moral permissibility of those agentsÕ plans and actions. Yet it can be extremely difficult to be certain what an agent is committed to when she acts: it can be hard to know whether someone who drops bombs is committed to killing. But if we cannot tell what another agent intends, we cannot use DDE to assess what she does. This problem is not always serious. In many cases, there will be plenty of evidence for what an agent is committed to when she acts. For example, Terror bombers might consistently choose different kinds of bombs to Strategic bombers: Strategic bombers who are committed to hitting a particular site might prefer a bomb that could be accurately targeted, whereas Terror bombers who are committed to killing might prefer a bomb that caused damage over the widest area. This kind of evidence helps us to distinguish Terror bombers and Strategic bombers.14 Even if it is hard for us to be certain
INTENTIONS, FORESEEN CONSEQUENCES, DOCTRINE OF DOUBLE EFFECT
263
what other agents intend and what they merely foresee, we can sometimes tell what their intentions are. But there is a second, even greater worry about using BratmanÕs distinction: is the line between the intended and the foreseen drawn in the right place? 3.
THE PROBLEM OF CLOSENESS
Bratman (1999) has shown that we can distinguish between the foreseen and the intended in the following way: an agent intends some consequence C if and only if she is committed to making C occur. But does his distinction class too little as intended and too much as merely foreseen? Foot (1978) introduces this problem for theories of intention like BratmanÕs, the ‘‘problem of closeness’’, with an example: THE EXPLORERS: A fat man who is stuck in the mouth of the cave has trapped some explorers inside the cave. The only way for the explorers to escape is to blow up the fat man with dynamite. The explorers intend to escape from the cave by blowing up the fat man.
Foot imagines what these explorers might say: Suppose that the trapped explorers were to argue that the death of the fat man might be taken as a merely foreseen consequence of the act of blowing him up. (ÔWe didnÕt want to kill him. . .only to blow him into small piecesÕ or even Ô... only to blast him out of the cave.Õ) I believe that those who use the doctrine of the double effect would rightly reject such a suggestion, though they will, of course, have considerable difficulty in explaining where the line is to be drawn. What is to be the criterion of ÔclosenessÕ if we say that anything very close to what we are literally aiming at counts as if part of our aim? Foot (1978: 21–22).
In FootÕs example, the explorers plainly intend to blow up the fat man as their means of leaving the cave. Assuming that they foresee that he will die, the explorers kill him intentionally. But they claim that they do not intend to kill him.15 Foot thinks that we should reject this claim, but that it is difficult to explain why we are entitled to do so. Bennett raises a similar problem with an example of an unusual kind of Terror bomber, who intends to win the war
264
ALISON HILLS
by making people appear to be dead, with the foreseen consequence that they are permanently dead.16 Call this kind of terror bomber a philosophically sophisticated terror bomber: The PSTB intends to win the war by lowering enemy morale by making it appear that civilians are dead.
PHILOSOPHICALLY SOPHISTICATED TERROR BOMBER (PSTB):
The PSTB, as described by Bennett, says that he intends to make it appear that civilians are dead. Assuming that he foresees that the civilians will die, he will intentionally kill them, but he claims that he does not intend to kill. Bennett thinks that most people would reject his claim and they would be right to do so. Both the PSTB and the explorers claim that killing is not their aim, that they are not committed to it, and that they do not intend to kill. These cases cast doubt on whether we have drawn a line between the intended and the foreseen in a way that can support our judgement that the explorers and the PSTB in fact intend to kill. This raises a severe problem for defenders of DDE, because it seems that, given BratmanÕs conception of intention, very few consequences of our actions will count as intended. Most of the interesting consequences of our actions will turn out to be merely foreseen, and there will be no grounds for appealing to DDE in order to assess them. DDE will have very few interesting applications. This is not just a difficulty for defenders of DDE, however. It is a problem for anyone who, like Bennett and Foot, believes that the PSTB and the explorers intend to kill. Anyone who agrees with them must try to explain how this commonsense judgement can be correct. There are two possible responses to this problem. First, we might claim that it is impossible in principle for any agent to have the plans that the explorers and the PSTB are described as having without intending to kill, for example, we might claim that anyone who intends to blow someone up into little pieces must also intend to kill him.
INTENTIONS, FORESEEN CONSEQUENCES, DOCTRINE OF DOUBLE EFFECT
265
Second, even if it is possible for someone to intend to blow someone up or to make him seem dead without intending to kill him, we may argue that we have reasons to believe that the explorers and the PSTB in fact intend to kill and that their claims to the contrary are disingenuous. Foot seems to anticipate that we will choose the first response, and reject the claims of the explorers as impossible in principle. In her view, a conception of intention like that specified by Bratman is too thin: too few consequences of our actions are classed as intended and too many are classed as merely foreseen. She thinks that we need to revise our thin conception of intention, and specify a thick conception of intention instead, such that ‘‘anything very close to what we are literally aiming at counts as if part of our aim’’.17 For example, we might modify BratmanÕs account of intention in the following way: any causally necessary consequence of what we are committed to counts as intended. Suppose that in the explorersÕ situation, it is a causally necessary consequence of blowing up the fat man that he will die. Since the explorers are committed to blowing him up, then, given this thick conception of intention, they must intend to kill. If we accepted this modification to BratmanÕs account of intention, we could be right to interpret the explorers as intending to kill. But this cannot be correct. If everything that is a causally necessary consequence of what an agent commits to counts as intended, too many consequences of our actions will be classed as intended. The Strategic bomber aims to hit military targets, and it may be a causally necessary consequence of doing so that he will kill some civilians. So we would have to interpret the Strategic bomber as intending to kill; but he does not. Of course, this might just show that, since it is not the case that any causally necessary consequence of our aim should count as intended, we need to find some other ‘‘thick’’ conception of intention. But in fact, it is a mistake to try to specify a ‘‘thick’’ conception of intention at all. Intentions are intensional: however ‘‘close’’ X is to Y, even if X is identical with Y, it is possible for an agent to intend that X and not to intend that Y. For example, the Strategic bomber may intend
266
ALISON HILLS
to destroy the ball-bearing factory but not intend to destroy the largest factory in the city, even though the ball-bearing factory is the largest factory in the city. Even where an agent correctly believes that X is identical with Y, she may intend that X and not intend that Y; she may simply have not drawn the requisite inference. It is possible for this Strategic bomber to know that the ball-bearing factory is the largest factory in the city, but simply not to have drawn the inference that she intends to destroy the largest factory in the city. It must in principle be possible for an agent to intend to blow someone up or to intend to make him seem dead without intending to kill him.18 We cannot claim that the explorers and the PSTB intend to kill on the grounds that it is impossible for any agent to adopt their plans without intending to kill, because in principle an agent could adopt those plans without that intention. The first strategy to defend DDE fails, and we will have to resort to the second strategy, and find reasons to think that agents in those situations in fact intend to kill. There is a cost to giving up on the claim that there is a thick conception of intention. We are compelled to give up the claim that the explorers and PSTB must intend to kill, given their other intentions. The second strategy cannot prove that, necessarily, these agents intend to kill; all that it can show is that, first, they have reason to intend to kill, and secondly, that we have reason to interpret them as intending to kill. There is inevitably some uncertainty about interpreting other agentsÕ intentions, and therefore about whether it is legitimate to assess their actions using DDE. But we can have good reasons to think that these agents intend to kill that are sufficient to justify using DDE to assess what they do, even when we cannot be absolutely sure of their intentions.
4.
THICK PLANS AND EXTRA INTENTIONS
According to BratmanÕs thin conception of intention, only the consequences of an action that the agent is committed to
INTENTIONS, FORESEEN CONSEQUENCES, DOCTRINE OF DOUBLE EFFECT
267
count as intended. This seemed to raise a problem, in that too few consequences of our actions would count as intended: it would be possible for the explorers and the PSTB to adopt their plans without intending to kill. But the thin conception of intention only has that result in combination with another assumption, that agents have only one end, and they intend only that end together with (what they take to be) the necessary means to achieve that end. For example, when Foot reports the explorersÕ argument that they do not intend to kill, she portrays them as having only one end (to escape from the cave) and as intending only that end and the necessary means to achieving it (blowing up the fat man). Similarly, Bennett describes the PSTB as intending only one end (to win the war by lowering morale) and the necessary means to achieving it (making people appear dead). Since killing is not their end, nor a necessary means to their end, we are forced to conclude that these agents do not intend to kill. Call any plan that includes only one end and the necessary means to that end, a thin plan. The thin conception of intentions causes problems for DDE in combination with the assumption that agents have only thin plans. It is certainly possible for agents to have thin plans. A thin plan is arguably the minimum that an agent must have in order to qualify as reasonable: it is debatable whether someone who did not intend an end, or who did not intend the necessary means to that end, would be capable of genuinely intentional action at all. But at the same time, it is not very realistic to model agentsÕ intentions by focussing on thin plans. Most agents have thick plans, that is, they intend more than just one end and the necessary means achieve it. Almost any time that we act, almost all of us will have more than one intended end. For example, the explorers, as well as intending to escape from the cave, probably have a number of additional aims too; for instance, it is extremely likely that they also intend to avoid causing themselves unnecessary suffering. They will reason about how to avoid suffering, constrain their intentions and monitor their success at avoiding suffering. Their extra
268
ALISON HILLS
end will affect what they intend as their means. To escape from the cave, all that they need is an explosive that will blow up the fat man. But because they also intend to avoid causing themselves unnecessary suffering, they will try to find an explosive that will not cause a rock fall that would injure everyone in the cave. To defend our commonsense interpretations of the PSTB and the explorers, we have to show that it is likely that these agents do intend to kill. Of course, they do not necessarily have that intention, for they could have thin plans with no commitment to killing. But there are numerous reasons why they might form extra intentions, adopting a thick plan. Here I will discuss two: first, an agent may form the intention to bring about some sufficient (but not necessary) means in order to achieve her ends; second, an agent may form additional intentions to avoid causing additional suffering to others. Ultimately there is a fact of the matter whether an agent in the situation of the explorers has adopted a thin or a thick plan and whether she intends to kill. If DDE is correct, it is morally worse for them to intend to kill, whether or not we know their intentions. But if we are to find out whether or not they are morally bad in this respect, we need to be able to judge what intentions they have. We need to consider not just the reasons why the agents themselves may form extra intentions, but also the reasons that we have for interpreting those agents as having thick plans. This is important for our ascription of intentions to them, and therefore to the practical use of the intention/ foreseen distinction. We need to have a grasp of both kinds of reason if we are to make use of the intention/ foresight distinction (and therefore the DDE) in practice, so I will discuss both kinds of reason. 4.1.
The PSTB: Intending Sufficient Means
When the PSTB chooses his plan, he commits himself to lowering morale by making civilians seem dead in such a way that others believe that they are dead. It is not necessary to
INTENTIONS, FORESEEN CONSEQUENCES, DOCTRINE OF DOUBLE EFFECT
269
achieve this end that they actually do die. But how is the PSTB to succeed in making others believe that some have died? One obvious possibility is to kill them. If he chooses that as his sufficient (but not necessary) means, he commits himself to killing; that is, to using means-end reasoning to work out how to kill; to constraining his further intentions so that they do not conflict with killing; and to tracking his success at killing. In carrying out his PSTB plan, it is very likely that he would in fact commit himself to killing as a means to achieving the appearance of death.19 The PSTB has a good reason to intend to kill, given his intended end, because killing is a salient sufficient means to achieve that end. If he has in fact committed himself to killing, DDE applies to him, and if the traditional absolutist version of DDE is correct, he has acted wrongly. But whether DDE in fact applies to the PSTB is one question. Whether we are entitled to use DDE to assess him is another. This epistemological question is obviously crucial to the use of DDE in making moral judgements about the PSTB. It is reasonable to interpret the PSTB as intending to kill civilians as a way of making them seem dead, because killing is a very obvious means of achieving this aim and we have no evidence that he has adopted an alternative means. So it is reasonable for us to assess him using DDE, on the assumption that, like the Terror bomber, he does intend to kill.20 4.2.
The Explorers: Intending to Avoid Causing others Unnecessary Suffering
There are other reasons why an agent may form additional intentions. Consider the explorers, trapped in the cave by a fat man. There is no way for them to escape from the cave without blasting him out with explosives, and this is what they choose to do. Obviously it is a foreseen consequence of their plan that the fat man will die, but his death is not their end, which is to escape from the cave, nor is it a necessary means to their end: his dying does not contribute to their escaping from the cave.
270
ALISON HILLS
The explorersÕ intended means – blowing up the fat man – is very likely to kill him. But if by some chance their victim does not die, he will be left in an appalling state, in terrible pain, suffering horrible injuries, and very likely to die in the near future. Though it is possible to blow someone up without killing him, given their situation, it is impossible for the explorers to blow up the fat man and leave him with a life that is worth living.21 Before the explorers had adopted the dynamite plan, they did not want or intend to kill. But when they adopt this plan, they must accept that their victim will die or be left in a state worse than death; or they must adopt an intention to kill, to ensure that he is not left in this state. The explorers might be wholly indifferent to the suffering they could cause their victim. But surely most people, given that they had adopted a plan of blowing up the fat man would choose to ensure that he died as quickly as possible; since that they had decided on a plan that would result in either an immediate death, or a death potentially accompanied by prolonged suffering for their victim, they would commit themselves to bringing about the former. The explorers might choose to kill as part of their plan, not for the sake of achieving their original intended end, but to prevent a ‘‘fate worse than death’’ for their victim. Since they would expect their victim to die as a result of their original means, they would not anticipate having to do anything extra to ensure that he did die; but they would nevertheless commit themselves to killing him.22 The explorers can choose plan A or plan B: EXPLORERS, PLAN A:
The explorers intend to blast the fat man out of the cave but are indifferent to whether he suffers or dies.
If the explorers adopt thin plan A, they use means-end reasoning to work out how to remove the fat man from the cave, they constrain their further intentions and track their success at removing him. They are not committed to killing him, so they do not intend to kill. If they carry out this plan
INTENTIONS, FORESEEN CONSEQUENCES, DOCTRINE OF DOUBLE EFFECT
271
successfully, the fat man may die straightaway in the explosion or he may suffer a prolonged and terrible death. Alternatively, the explorers could adopt a more complex thick plan: EXPLORERS, PLAN B:
The explorers intend to blast the fat man out of the cave and intend that they kill him instantly.
If the explorers adopt plan B, they commit themselves to working out how much dynamite they need to use to ensure that the fat man will die rather than be left with terrible injuries, to constraining their further intentions and monitoring their success at killing him. They intend to kill him. If they carry out plan B successfully, they blast the man out of the cave and kill him instantly; he suffers only for a very short time, if at all. The explorers have some reason to choose thick plan B and to intend to kill, in order to avoid causing additional suffering.23 Once we recognize that agents can have reasons to adopt thick as well as thin plans, it is obvious that even a thin conception of intention, like BratmanÕs, need not class too few consequences of an action as intended. The explorers plainly have some reason to commit themselves to killing the fat man. Whether or not agents in their situation actually form the intention to kill is a matter of fact that depends on whether they actually chose thin plan A or thick plan B. But there is a separate question that is also very important: how are we to interpret an agent who is in the explorersÕ situation? If the intention/ foresight distinction (and therefore the DDE) is to be of practical use, we need to know when we are justified in attributing extra intentions to agents and when we are not. It is compatible with the description of EXPLORERS that the explorers intend to kill the fat man, and it is compatible with it that they do not. There might be more information that we could gather to help us decide which plan they had chosen. If they used just enough dynamite to blast the fat man out of the cave, but not sufficient to kill him instantly, we would have some evidence in that they had adopted plan A. If they
272
ALISON HILLS
used far more dynamite than necessary to blast him out of the cave, we would have some evidence in favour of plan B.24 But suppose that we have no such extra information: we have no evidence that the explorers do intend to kill, and no evidence that they do not. Nevertheless, according to Foot and many others too, they do intend to kill. I suggest that when we assess EXPLORERS, we judge that only someone unusually callous or cruel would choose a plan that might involve watching their victim die slowly and painfully as a result of injuries that they themselves had inflicted on him, when they could have chosen to kill him instantly. Given that we have no grounds for thinking that they are particularly unusual people, it is reasonable for us to assume that the explorers do not have these traits. That is, we implicitly use the following argument. 4.3. 1.
2. 3.
4.
The Probability Argument Where we have no clear evidence whether or not an agent has an intention to kill, we have reason to interpret him by judging what it is most likely that agents in that situation would intend. In EXPLORERS, we have no clear evidence whether or not any of the explorers has an intention to kill. In EXPLORERS, most people would intend to kill, because in the circumstances, it would be unusually callous or cruel of them not to intend to kill. So, we have reason to judge that it is likely that the explorers intend to kill.
The probability argument concerns the way that we ought to make judgements about what other people intend to do when we do not have full knowledge of what they are committed to do. According to the probability argument, where we have no other relevant evidence, we should base our judgements on what most people would intend to do, in a particular situation. The argument does not rely on claims about what it is right to do.
INTENTIONS, FORESEEN CONSEQUENCES, DOCTRINE OF DOUBLE EFFECT
273
Instead, the key claim is that few people would be so callous as to be prepared to watch people die slowly and painfully from the injuries that they had inflicted on them. The claim is about what most people would choose, not about what they should.25 It is compatible with this argument that the absolutist version of DDE is correct, that though most people would intend to kill, they would be wrong to do so. For many absolutists, the EXPLORERS case would be a very difficult trilemma, where each of the available options carried a considerable cost. The explorers would either have to choose plan B and intend to kill, thereby doing wrong; or choose plan A and act cruelly or callously by being prepared to cause extra suffering to their victim; or choose neither plan, refuse to harm the fat man, and sacrifice their own lives. My claim is that we judge that most people, if faced with this trilemma, would choose plan B, and this is why we assume that the explorers intend to kill. Absolutists can agree with this, whilst adding that the explorers would be wrong to make that choice. If we knew that some character trait was exceptionally unusual, we would be entitled to assume of some agents, about whom we had no other information, that there was a very low probability that they had that trait (whilst keeping in mind the possibility that they were very unusual people who did have it, after all).26 However, the reliability of this kind of assessment of other agents depends on the reliability of our judgements about what people would choose in such circumstances, which may themselves be extremely rare. In cases like EXPLORERS, we may simply not be sure how unusual it would be for anyone to be prepared to watch her victim die slowly and painfully when she could have killed him instantly. Most of us have performed no relevant psychological experiments, and have access to no relevant data. We are relying on ‘‘folk psychological’’ assumptions about character traits. We have some reasons to think that ‘‘folk psychology’’ is reliable, in that we use it everyday with reasonable success to
274
ALISON HILLS
predict what other people will do. Nevertheless, there are at least two qualifications that should lead us to be cautious in interpreting othersÕ intentions by this method. First, it is likely that judgements about character traits will depend a great deal on social and cultural practices. An action that one culture would class as exceptionally cruel or callous, another culture may class as normal.27 So we have to be very careful about using our assumptions about what is unusual behaviour when interpreting other agents. Secondly, the reliability of folk psychological judgements about character traits quite generally has recently come under attack.28 Psychologists have carried out experiments that show that some of our predictions about what agents would do, based on our judgements about their character traits, are mistaken. For example, in MilgramÕs study of obedience, subjects are much more willing than one might expect to cause another person to suffer.29 The willingness of ordinary people to cause each other significant harm, as demonstrated in MilgramÕs experiments, is surprising. But whilst these psychological experiments show that our folk psychological judgements about character traits can be mistaken, they do not prove that they are never accurate, or that we are never justified in using them to interpret other agentsÕ intentions.30 We are justified in interpreting the explorers as intending to kill insofar as we are justified in the assumption that it would be unusual in the circumstances for an agent to be prepared to cause their victim unnecessary suffering. Though there are good reasons to be cautious about such assumptions, at least sometimes they may be reliable. Provided that our judgement about what normal agents would intend in the situation of the explorers is accurate, we do have reasons for interpreting the explorers as intending to kill, but we should keep firmly in mind the possibility that we may be wrong: the explorers may not have committed themselves to killing and they may not intend to kill after all.31
INTENTIONS, FORESEEN CONSEQUENCES, DOCTRINE OF DOUBLE EFFECT
4.4.
275
Further Examples
The same ‘‘probability argument’’ can be used in the following case: CRANIOTOMY:
A woman is pregnant, but her foetus is trapped in the birth canal and she will die unless the foetus is removed. The only way to save the mother is to remove the foetus by crushing its skull. The doctor intends to save the mother by removing the foetus.
is a familiar case in the literature of DDE. Defenders of DDE usually interpret the doctor as intending to kill. We can use the probability argument of explain why their interpretation of the case is reasonable. In CRANIOTOMY the doctor intends to remove a foetus from its mother by crushing its head, thereby killing it. The death of the foetus is not the doctorÕs end, which is to save the mother, nor is it a means to that end. However, if the doctor crushes the skull of the foetus, but does not kill it, she is likely to inflict appalling injuries on the foetus that will cause it terrible suffering. The foetus will be left in a state worse than death. Alternatively, the doctor could commit herself to ensuring that the foetus died quickly. Before the doctor had adopted the craniotomy plan, she did not intend or want to kill. When she adopts this plan, she must intend to harm the foetus, with the foreseen consequence that her victim will either die or be left in a state worse than death; or she must intend to kill. The doctor might be indifferent to the suffering she could cause her victim. But surely most people, given that they had adopted a plan of crushing the skull of the foetus, would choose to ensure that their victim died as quickly as possible and was not left in a state worse than death.32 Since they committed themselves to harming their victim, they would also commit themselves to ensuring that it died without suffering terribly. There may be evidence that the doctor intends to kill rather than merely intends to harm. If the doctor has the choice between crushing the foetusÕs skull just enough to remove it from the motherÕs body, or so much that it dies instantly, she may choose to ensure that it dies instantly, even CRANIOTOMY
276
ALISON HILLS
though either method would be sufficient to remove it from the motherÕs body. But even if we have no such evidence, we may have reason to interpret the doctor as intending to kill. Suppose that only an exceptionally callous or cruel individual would be prepared to watch the foetus die slowly and painfully when she could have prevented its suffering. Given that we have no reason to think that the doctor is unusually cruel or callous, it is reasonable to judge it likely that she intends to kill the foetus to avoid causing it unnecessary pain. Once again, we have to bear in mind that we may be wrong: the doctor may not have committed herself to killing and so she may not intend to kill. The ‘‘thick plans’’ here attributed to the doctor and the explorers match the traditional interpretations of those agentsÕ intentions in the literature about DDE. But there might be a problem with this approach in other cases. Might we not find ourselves judging many people as intending to kill, even those who, according to both common sense and traditional interpretations of the cases by defenders of DDE, do not do so? For example, many people interpret the doctor in the following example as not intending to kill: MORPHINE: A patient is suffering from terrible pain. The doctor intends to reduce the pain using morphine. But the doctor knows that, since the patient is in such great pain, she may have to use so much morphine that the patient will die.
In MORPHINE, the doctor plainly could commit herself to reducing her patientÕs pain without committing herself to killing her patient. Do we have reason to interpret her as intending to kill nevertheless, on the grounds that she would be unusually callous or cruel if she did not intend to kill her patient as quickly as possible? No. A doctor who was genuinely committed to relieving her patientÕs suffering (but not to killing) would not be callous or cruel at all, let alone exceptionally or unusually so. We do not have reason to interpret the doctor in this example as intending to kill.33 The Strategic bomber intends to destroy a factory and knows that he is likely to inflict suffering as well as death
INTENTIONS, FORESEEN CONSEQUENCES, DOCTRINE OF DOUBLE EFFECT
277
on civilians who live near the factory, but he is normally interpreted as not intending to kill. Would the Strategic bomber be unusually cruel or callous if he did not intend to kill the civilians instantly in order to save them from suffering? No. It is very unlikely that his intending to kill would make any difference to his victims, because it is difficult, if not impossible, to ensure that people die quickly when you drop bombs on them from an aeroplane. So a Strategic bomber who did not intend to kill would not be significantly more callous than a Strategic bomber who had that intention. We do not have reason to think that Strategic bombers intend to kill. In principle it is possible for the agents in all the examples discussed to carry out their plans without intending to kill. Whether or not they do intend to kill depends on what they actually commit themselves to do. But it is reasonable for us to judge that the PSTB intends to kill as a salient means to his end and that the explorers and the doctor in CRANIOTOMY intend to kill in order to avoid causing additional suffering to their victims. By contrast, we have no reason to interpret the Strategic bomber or the doctor in MORPHINE as intending to kill, as they would not thereby avoid inflicting extra suffering on their victims. By judging what most agents would choose to do in the circumstances, we can interpret the intentions of the explorers, the doctors of MORPHINE, CRANIOTOMY and all the bombers in accordance with both common sense and the traditional accounts of those cases by defenders of DDE. 5.
THE DOCTRINE OF DOUBLE EFFECT
According to Bratman, an agent intends that C if she is committed to C: if she uses means-end reasoning to work out how to make C occur, she constrains her further intentions to ensure that they do not conflict with making C occur and she tracks her success at making C occur. She merely foresees C if she is aware that she is either certain or likely to bring C about but she is not committed to it. In principle this distinction is clear, though it can be indeterminate and it is often
278
ALISON HILLS
extremely difficult to be sure to what any agent (even oneself) is committed. Though BratmanÕs account of intention is ‘‘thin’’, it will not class too few consequences as intended, since, for a great variety of reasons, many agents adopt thick rather than thin plans, intending extra ends and sufficient (but unnecessary) means to their ends. BratmanÕs theory draws a clear distinction between what is intended and what is merely foreseen in a way that reflects our intuitive grasp of the distinction. The second question discussed in this paper concerns the practical use of the distinction. In real life cases, can we tell what agents intend and what they merely foresee? Sometimes we will have evidence of what an agent is committed to on the basis of what she does. But sometimes we have no such evidence. I have suggested that even so, we sometimes have reason to ascribe extra intentions to her, on the grounds that most agents in her circumstances would have such intentions (provided that our judgement about what normal agents would intend in the circumstances is reliable, and we have no reason to think that she is unusual). Though there will always be some uncertainty over exactly what other agents are committed to, and so what they intend, we can draw the distinction between the intended and the foreseen, using BratmanÕs account of intention, in a way that supports DDE (in either its absolutist or the non-absolutist version). But it still remains for a defender of DDE to explain in what way, if any, the resulting intended/ foreseen distinction has moral significance.34 NOTES 1
Harman (1976), Bratman (1999). McIntyre (2001: 221–223), Kamm (1991: 571–572). 3 Provided that the agentÕs intention is good, that the agent does not want or intend that the bad consequence occur and that the foreseen harm is proportionate to the intended good (McIntyre, 2001: 221–223). 4 Hills (2003). 5 McIntyre (2001: 233–237), Quinn (1993: 176–186), Fischer et al. (1993), Davis (1984: 107–121), Hart (1968: 122–125). 2
INTENTIONS, FORESEEN CONSEQUENCES, DOCTRINE OF DOUBLE EFFECT 6
279
Quinn (1993). In this paper, I rely on intuitions about what is intended in certain problem cases in order to develop an account of intention. The account of intention I defend fits with intuitions about intentions that are widely held amongst philosophers (as reported in the philosophical literature on DDE). It would be interesting to have empirical data about how widely these intuitions are shared amongst non-philosophers and amongst people of different cultures, similar to the studies of Knobe (2003, 2004) on intuitions about which actions are performed intentionally, but as far as I know, no such data has yet been collected. 8 Obviously, even if there is a clear intention/ foresight distinction, it does not follow that it is a morally significant distinction: DDE may be false for other reasons. 9 Anscombe (1957: 41f.). 10 Bratman (1999: 140–143). An agent has these attitudes towards an outcome under a particular description, so that using means-end reasoning to work out how to achieve C and tracking your success at C is not the same as using means-end reasoning to work out how to achieve D and tracking your success at D, even where C and D are identical. 11 Even if it is certain that the Strategic bomber will kill some people, he is not committed to killing in these ways. 12 Bratman (1999: 142). The Strategic bomber who is committed to lowering the chances of killing civilians (who for example would encourage the evacuation of the area) could be said to intend not to kill, even though in fact he is likely to raise the chances of killing, with the proviso that this intention is constrained by his intention of destroying the factory: if it turned out to be impossible for him to destroy the factory without killing some civilians, he would be prepared to kill. 13 For example, an agent might treat S as a sign of C and monitor whether she successfully brings about S, but use means-end reasoning about how to make C, not S, occur: does this agent intend that S? In most circumstances, the agent herself will be best placed to know what her intentions are. But she herself may not be certain to what she is committed, and it is certainly possible for her to deceive herself about her commitments. 14 Moreover, the fact that it is difficult to know what oneself or others intended is a problem only on the assumption that the primary purpose of DDE is retrospective, used to make judgements about what oneself and others have done in the past: when you bombed your enemy, did you or did you not intend to kill civilians? But this is not the only use of DDE: it is also important in the agentÕs own prospective assessment of plans that she might adopt. If you are working out how to fight your enemy, you may consider a Terror bomber or a Strategic bomber plan. You know 7
280
ALISON HILLS
that if you adopted the Terror bomber plan you would commit yourself to killing in a way that you would not if you chose Strategic bombing. If DDE is correct, other things being equal, it is morally worse for you to choose a Terror bomber plan. DDE can be extremely important for the assessment of future plans. 15 What you do intentionally includes both what you intend to do and what you merely foresee doing (Bratman, 1999). 16 Bennett (1995: 210–211) 17 Foot (1978: 22) 18 My claim is that it is in principle possible for an agent to intend to blow someone up without intending to kill him and it is in principle possible for an agent to intend to make some people appear dead without intending to kill them. It is compatible with this claim that any real human being in the situation of the PSTB or the explorers who had the former intentions would as a matter of fact also have the latter intentions. In what follows, I suggest that most people, in the situations described, would in fact intend to kill. 19 Kamm (1992: 377–378). Are there any other ways that the Terror bomber could make people seem dead? Perhaps he could formulate an elaborate plan in which he dropped bombs on highly realistic dummies that made it seem that civilians had died. It is possible that he might formulate and carry out such a plan, so it is possible that he might not intend to kill. But in the absence of such a plan, it is likely that he will in fact commit himself to killing. 20 Whether our assumption about his intentions is correct depends on what in fact he is committed to. So we must always bear in mind that our assumption may be wrong, that he may not have chosen the most salient means to his end, and that he may not intend to kill. 21 Of course, if it were possible to remove the fat man without significantly harming him, we can assume that this is what the explorers would do, but in EXPLORERS it is stipulated that this option is not available. 22 Kamm denies that you need always intend some means (M) to your end (E): you need not aim at M if you are confident that you will bring it about as a side-effect of some other intended end (E*) at which you are aiming (Kamm, 2000, 1996: 179–181). This suggests that if the explorers were confident that the fat man would die instantly as a consequence of being blowing up, they would not need to commit to killing him, even if they were committed to avoiding causing him unnecessary suffering. But this does not seem correct: if in fact you brought about E* without M (and therefore without E), you would have to try to bring about M after all. You must intend the means to your end, even if you anticipate bringing them about as a side-effect of something else. If the explorers are genuinely committed to avoiding causing their victim unnecessary suffering, they ought also to commit themselves to killing him.
INTENTIONS, FORESEEN CONSEQUENCES, DOCTRINE OF DOUBLE EFFECT 23
281
If the traditional absolutist version of DDE is correct, however, it is wrong for anyone to intend to kill, i.e. even if the explorers have a reason not to cause unnecessary suffering, they have a stronger reason not to intend to kill. 24 Although this evidence is defeasible: the explorers might not know how much explosive to use; in addition, they might have other reasons for using a lot or a little dynamite. 25 The claim in premise 1 of the probability argument should therefore be distinguished from a strong version of the principle of charity (Davidson, 1980: 222), according to which we should interpret others as intending to do what is morally right (provided there is no conflicting evidence against this interpretation). Instead, according to the probability argument, we should interpret people according to the traits that we think are common, whether they are morally good (kindness), morally bad (selfishness) or morally indifferent (squeamishness). For example, we might claim that most people in the position of the explorers would intend to kill, because most people would be too squeamish to watch someone slowly die as a result of their actions. 26 When we do have extra information about the agentÕs character (e.g. that she is unusually cruel or unusually kind) or about what she intends in a particular situation (e.g. she has used so much dynamite that she almost certainly intends to kill), we should obviously use this information to judge what the agent intends, rather than base our judgement on what a normal agent would intend in the circumstances. 27 Watching public hangings and watching men fight to the death as a form of entertainment have been considered normal by some cultures at some times, but most modern Western societies would consider them to be highly unusual. 28 For example, Doris (2002) and Harman (2000). 29 All of the 40 subjects were prepared to give their co-participant an electric shock of at least 300 V and 26 were prepared to administer the highest shock available to them, 450 V (the experiment was set up so that the co-participant pretended to have been given the shock, though in fact he was not). Previous predictions were that only 0–3% of the subjects would be prepared to administer such a large shock (Milgram, 1974, reported in Kamtekar, 2004: 462–463). 30 Doris (2002) and Harman (2000) sometimes suggest that these experiments show that our folk psychological judgements about character traits are entirely worthless, and furthermore, that there are no such things as character traits as we ordinarily understand them. But these conclusions are much too strong, as Kamtekar (2004) and Sreenivasan (2002) demonstrate. Though the experiments show that we need to revise our conception of character traits, they do not prove that there are no such things as virtues or vices or that our predictions on the basis of those traits are always wrong. Indeed, such experiments can help us to improve the reli-
282
ALISON HILLS
ability of our judgements about what people in situations like the explorers would intend to do, and so help us to make more accurate assessments of what other agents intend and whether DDE applies to them. 31 It is important to emphasise that whether the explorers in fact intend to kill depends on whether they are actually committed to killing, even if we are right to think that most agents would intend to kill in that situation. 32 Nicholson claims: ‘‘That the narrowing of the head and not the death of the foetus is the means to the end of saving the woman is demonstrated by the fact that the foetus would not be killed should it somehow survive the force applied to its skull and be removed alive from the birth canal.’’ (Nicholson, 1978: 26, quoted in Davis, 1984: 116). Nicholson is right that the death of the foetus is not a means to the doctorÕs ends, but surely most people, given the alternatives open to them, would ensure that the foetus died as quickly as possible during the operation, rather than that it suffered prolonged agony with a crushed skull. 33 There might even be some chance that the patient could survive the course of treatment and afterwards enjoy a life worth living, in which case it would obviously be wrong for a doctor to intend to kill. 34 I offer such an explanation in Hills (2003). Alternative defences of the moral significance of intentions are provided by Chappell (2002), Quinn (1993: 175–193), Nagel (1986: 175–88). I am grateful to Hallvard Lillehammer, Tim Chappell and an anonymous referee for helpful comments on earlier drafts of this paper.
REFERENCES Anscombe, G.E.M. (1957): Intention, Oxford: Blackwell. Bennett, J. (1995): The Act Itself, Oxford: Oxford University Press. Bratman, M. (1999): Intentions, Plans and Practical Reason, CSLI. Originally published: Cambridge, MA: Harvard University Press, 1987. Chappell, T. (2002): Two Distinctions that Do Make a Difference: the Action/Omission Distinction and the Principle of Double Effect, Philosophy 77, 211–233. Davidson, D. (1980): Essays on Actions and Events, Oxford: Oxford University Press. Davis, N. (1984): The doctrine of double effect: problems of interpretation, Pacific Philosophical Quarterly 65, 107–123. Doris, J. (2002): Lack of Character: Personality and Moral Behaviour, Cambridge: Cambridge University Press. Fischer, J.M., Ravizza, M. and Copp, D. (1993): Quinn on Double Effect: The Problem of Closeness, Ethics 103, 707–725. Foot, P. (1978): Virtues and Vices, Oxford: Blackwell.
INTENTIONS, FORESEEN CONSEQUENCES, DOCTRINE OF DOUBLE EFFECT
283
Harman, G. (1976): Practical Reasoning, Review of Metaphysics 29, 431–463. Harman, G. (2000): Explaining Value and Other Essays in Moral Philosophy, Oxford: Oxford University Press. Hart, H.L.A. (1968): Punishment and Responsibility. Oxford: Oxford University Press. Hills, A.E. (2003): Defending Double Effect, Philosophical Studies, 116, 133–152. Quinn, W. (1993): Morality and Action. Cambridge: Cambridge University Press. Kamm, F.M. (1991): The Doctrine of Double Effect: Reflections on Theoretical and Practical Issues, Journal of Medicine and Philosophy 16, 571–585. Kamm, F.M. (1992): Non-Consequentialism, Person-as-an-End-in-Itself, and the Significance of Status, Philosophy and Public Affairs 21, 354–389. Kamm, F.M. (1996): Morality, Mortality II. Oxford: Oxford University Press. Kamm, F.M. (2000): The Doctrine of Triple Effect and why a Rational Agent Need not Intend the Means to his End, I, Proceedings of the Aristotelian Society Supplementary Volume 74, 21–39. Kamtekar, R. (2004): Situationism and Virtue Ethics on the Content of Our Character, Ethics 114, 458–491. Knobe, J. (2003): Intentional Action and Side-Effects in Ordinary Language, Analysis 63, 190–193. Knobe, J. (2004): Intention, Intentional Action and Moral Considerations, Analysis 64, 181–187. McIntyre, A. (2001): Doing Away With Double Effect, Ethics 111, 219–255. Milgram, S. (1974): Obedience to Authority, New York: Harper and Row. Nagel, T. (1986): The View from Nowhere, Oxford: Oxford University Press. Nicholson, S.T. (1978): Abortion and the Roman Catholic Church, Knoxville, Tenn: Journal of Religious Ethics Monograph. Sreenivasan, G. (2002): Errors about Errors: Virtue Theory and Trait Attribution, Mind 111, 47–68.
Department of Philosophy University of Bristol UK E-mail:
[email protected]