I have bioethics questions that needed to be answered. I will supply with all the material.

Suppose, as I think most people would agree, that it would be permissible to grow a single organ in this way in order to save a person’s life. If it would be permissible to grow one organ this way, it should also be permissible to grow two—for example, if the person needed two organ transplants rather than just one. But, if it would be ac- ceptable to grow two organs separately, it should be acceptable to grow three, or four, or five. Imagine that a person has a rare disease that can be cured only by a series of tissue grafts to most parts of his body. To cure him, it would be necessary to use the techniques of stem-cell-based biosynthesis to grow organic replicas of most of his body parts and then graft a small amount of each replica into the corresponding part of his body. It is hard to see how it could be wrong to grow a set of separate organs and body parts in this way. It would be macabre, certainly, but it is hard to see how the organs themselves could have a moral status that would make it wrong to create them for instrumental purposes, or to use or even to destroy them once they were cre- ated. But if there is no objection to creating each organ or body part separately, it ought equally to be acceptable to grow them all together in the form of an anen- cephalic clone. It would, of course, make a difference if they were grown together as an organism with a functional brain, for that would involve the existence of a further, distinct entity with interests and rights. But if the clone is anencephalic, the only dif- ference is that the organs and parts would be physically integrated and mutually sup- porting rather than maintained separately, each on its own scaffolding of combined organic and inorganic support systems. And it is hard to believe that this could make the difference between permissibility and impermissibility. It seems, rather, that the intrinsic moral status of an anencephalic clone—that is, the moral status it has based on its intrinsic rather than relational properties—is no different from that of a collec- tion of separately grown organs.

2. euthanasia and assisted suicide

2.1. From Suicide to Euthanasia

“Suicide” and “euthanasia” are concepts with blurred edges. It is often unclear whether a certain act counts as suicide or whether an act is an instance of euthanasia. These conceptual problems merit discussion, but only a rather brief discussion here. In the past, some of these questions were thought to have profound significance. For example, when it was widely accepted that suicides were excluded from heaven, it seemed of the utmost importance to be able to determine whether a particular act was an instance of suicide. Here, by contrast, the main point of the conceptual prelimi- naries is simply to delimit the subject matter of this section of the book.

Let us say that an agent commits suicide if he dies as a consequence of acting with the intention of bringing about his own death. This allows that suicide can be ei- ther by act or by omission. It thus treats as an instance of suicide the act of a person who dies as a result of refusing a life-saving medical treatment on the ground that he wished to die rather than to continue to live. Yet the definition’s reliance on the no- tion of intention introduces substantial unclarity. It is clear that a person who acci- dentally or inadvertently causes his own death does not commit suicide; and it is equally clear that a person who deliberately kills himself (for example, by putting a

Endings 455

gun to his head and pulling the trigger) does commit suicide. Whether, in the latter case, the agent intends his own death as an end or whether he intends it as a means to an end (for example, the punishment of his wife or, perhaps, the relief of his own suf- fering) is immaterial. I believe, though others have disputed this, that it is also im- material whether the agent acts under duress or coercion.50 There is nothing para- doxical about the notion of a coerced suicide. It is also reasonably clear that, when a person does something that he merely foresees may or will result in his death (such as playing Russian roulette), he does not necessarily commit suicide. But what about cases in which a person intentionally uses himself in a way that he believes will re- sult in his own death? Examples include the soldier in a foxhole who flings himself on a live grenade to shield his comrades, and the man who deliberately attracts the at- tention of a deadly wild animal to himself in order to allow his child to escape. Have such people, when they are killed, committed suicide?

Our inclination is to deny that they have. And certainly it is reasonable to see these deaths as foreseen but unintended; for, in each case, if the act had accomplished the person’s aim without resulting in the person’s death (for example, if the grenade had merely maimed the soldier or if the animal had mauled but not killed the parent), none of his intentions or plans would thereby have been thwarted. Yet there is another sense, associated with traditional readings of the Doctrine of Double Effect, in which these people’s deaths are intentional. To see this, we need only to imagine variants of the cases in which the agent sacrifices someone else rather than himself. If, for ex- ample, the soldier were to fling someone else on the grenade, most people would see this as an instance of intentional killing—that is, an instance of killing one person as a means of saving others, hence a paradigm of the sort of act that the Doctrine of Double Effect ought to condemn. Perhaps we should say that, even though the death would not be an intended effect in the narrow sense distinguished in section 10.1 of chapter 4, it nevertheless has, because of its relation to the agent’s intentions, the same significance as an intended effect. If so, the question is then whether, in the orig- inal cases, the presence of this same relation between the agent’s death and his inten- tions makes it appropriate to count these cases as instances of self-sacrificial suicide.

I will not pursue these complications here, though I have discussed the analysis of intention elsewhere.51 Because our concern is primarily with euthanasia, we can confine our brief discussion of suicide to cases that are relevantly parallel—that is, cases in which an agent brings about his own death because he believes that his sub- sequent life would not be worth living, and hence that death would be better for him than continued life. In these cases, the death is an intended effect on any account.

Next, a few words about the concept of euthanasia. An act of euthanasia, as I will understand the term, is an act of killing or of letting someone die that satisfies two conditions: first, that death benefits, or is good for, the individual who dies and, sec- ond, that the agent must be motivated to do what is good for that individual and must intend to benefit the individual in bringing about his death. Even if an act that brings about a person’s death is intended thereby to benefit the person, it does not count as euthanasia if it is not in fact good for that person. If it would have been better for the person to continue to live, the act may be an instance of intended euthanasia or at- tempted euthanasia, but it is not actually euthanasia. Similarly, if an act of killing or letting die is not intended to benefit the individual who is killed or allowed to die, it

456 the ethics of killing

does not become an act of euthanasia if, fortuitously, it turns out to have been better for the individual to die rather than to continue to live.

There is an interesting conceptual question that I have never seen discussed in the literature on human euthanasia. This question arises from a problem frequently en- countered in veterinary medicine. Suppose that a man brings his dog to the vet and requests that it be “put to sleep.” When asked why he wants the dog killed, the man replies that he and his family are going on a long trip and cannot afford to have the dog kept in a kennel (or that the dog wets the carpet, that the children have lost inter- est in it, or whatever). Although killing an animal in circumstances such as these is commonly known as “convenience euthanasia,” it may seem that this is just a euphe- mism, for it cannot be good for an animal to be killed when it is perfectly healthy. But there may be more to it than this. The vet may have reason to believe that, if she does not kill the animal painlessly, its owner will take it away and kill it himself, in a man- ner that would not be painless, or that he will turn it loose to starve or be hit by a car. Suppose, in fact, that the man has credibly threatened to do something of this sort. The vet might, of course, offer to take the animal and give it a home herself. But sup- pose that she gets two cases of this sort every week—far more animals than she her- self could care for or even find homes for. In each individual case, then, the situation is that there is an animal whose life could in principle be worth living but no one is willing, nor perhaps obliged, to do what is necessary to make it worth living. The vet therefore has the choice, in practice, between painlessly killing the animal and al- lowing it to live a life that she has good reason to believe will in fact be worth not liv- ing. A parallel case could arise in the case of a human being—for example, if a per- son’s life will be worth not living unless he has a surgical procedure that he cannot afford and that would be so expensive that no one else is willing or obliged to provide it for him. In these circumstances, would it count as euthanasia to kill him and could it be permissible to do so?

Again I raise these questions only to put them aside. I will restrict my attention here to cases in which it is not possible to change an individual’s life in a way that would make it worth living.

I will briefly note a few common distinctions and then turn to substantive matters. Euthanasia is said to be voluntary when a person freely and autonomously requests or consents to be killed or allowed to die for his own good. It is nonvoluntary when it is not possible for the individual who is killed or allowed to die either to give or to with- hold consent. This category, therefore, encompasses all cases of euthanasia in which the individual killed or allowed to die is a fetus, an infant, an animal, a congenitally severely cognitively impaired human being, or a human being who has ceased to be competent to form autonomous preferences about life and death, and who has not previously expressed his will on the matter. Euthanasia is said to be involuntary when an individual who is competent to give or withhold consent is killed or allowed to die either contrary to his expressed will or when his consent has not been sought. (There may be cases in a gray area between nonvoluntary and involuntary euthanasia in which an individual, though capable of expressing a preference, lacks the compe- tence to have an informed, autonomous preference.) Finally, killing an individual for his own good is commonly referred to as active euthanasia, whereas letting an indi- vidual die when that is in his best interests is referred to as passive euthanasia.

Endings 457

It has sometimes been remarked that it is difficult to accept that suicide can be morally permissible without accepting that euthanasia can be as well.52 And indeed there are only a few short steps between the premise that suicide can be permissible and the claim that euthanasia can be.

Consider a case in which a person’s life is worth not living: it is and will remain dominated by pain and suffering that cannot be alleviated and that are not counter- balanced by compensating goods. Some people, of course, claim that life is always worth living, or at least that it always can be made to be worth living. But if it is true that pain and suffering are in themselves bad, it seems that a life that contains little or nothing but pain and suffering—one, moreover, that is neither redeemed by its good effects on others nor elevated, ennobled, or enlightened by the experience of suffer- ing—cannot be worth enduring. I will assume that, when a life is bad in this way for the one whose life it is, and when that individual’s death would not be worse for any- one else, it can be prudentially rational for that individual to commit suicide. Most people accept that, in these circumstances, suicide would also not be immoral. Cer- tainly few people now accept that someone who commits suicide in such circum- stances is guilty of murder.

Let us assume, conditionally, that, when suicide is rational and not worse for others, it is also morally permissible. Most people also accept that there are circum- stances in which it is permissible to assist someone to commit suicide. Consider, for example, a case in which a person’s suicide would be rational and not worse for others but in which the person cannot easily bring about his own death because he is tethered to a battery of life-support machines from which it would be difficult to ex- tricate himself. Most of us believe that, in these circumstances, it would be permis- sible for the person’s physician to accede to his request to collaborate with him in withdrawing the life-support systems—for example, by assisting the person to re- move the various pieces of apparatus from his body. Indeed, most of us believe that, provided that the person is competent, the physician ought to enable him to free him- self from the machines. Given that the withdrawal of the life-support systems will lead immediately to the person’s death, we should regard the person’s removal of them as an act of suicide, albeit an instance of “passive suicide,” since he does not create the cause of his death but instead allows himself to die by removing the pro- tections that have thus far been saving him from a preexisting threat. In assisting him to remove these protections, the physician assists him to commit suicide.

I have described a case in which a person chooses to die “passively” by ceasing to resist a threat of death that is not of his own making. In most cases, people com- mit suicide actively, by creating the threat that results in their death. Many people re- gard it as impermissible to assist someone in actively committing suicide, and it is at present illegal in most areas of the United State for physicians to assist their patients in actively killing themselves—for example, by prescribing a lethal dose of some medicine. Yet it is hard to see how it could be permissible to provide assistance to a passive suicide but not to an active suicide. How could the permissibility of assis- tance by a third party depend on whether the person bent on suicide chooses an ac- tive or a passive means, particularly when either would be equally permissible?

It is tempting to appeal here to the general claim that, if it is permissible for a per- son to do something, it must also be permissible for others to assist him. But that

458 the ethics of killing

claim seems false. There are, for example, rare circumstances in which one wholly innocent person will be killed by another wholly innocent person unless the one kills the other first. In these circumstances, it may be permissible for each to try to kill the other but not permissible for a third party to assist either. For while it may be per- missible for each agent to give priority to himself, a neutral third party may be re- quired to treat both principals equally and impartially and may therefore be forbid- den to provide assistance to either at the expense of the other.53 But it seems that this restriction on the permissibility of assistance by third parties applies only in cases in which what an agent is permitted to do may prevent others from doing what they are permitted to do or otherwise have adverse effects on others. In the case of a rational suicide that would not be worse for others, these complications do not arise; there- fore there is no reason to suppose that a third party may not assist the agent in doing what it is permissible for that agent to do.

There are, moreover, positive reasons why, in the case of a rational suicide, as- sistance from others is desirable. People who set out to kill themselves sometimes fail, leaving themselves not only alive but disabled, disfigured, humiliated, and in pain. As Arthur Koestler, “speaking in the name of many . . . who tried and failed,” once observed, “there is only one prospect worse than being chained to an intolerable existence: the nightmare of a botched attempt to end it.”54 And even when people suc- ceed in committing suicide, they are often forced to quit life alone, in terror, without the support or validation of others, and with greater pain and mess than is necessary. If it is prudentially rational and morally permissible for a person to seek death, how much better it would be if he could be assured assistance from others that would en- able him to die with as much comfort, reassurance, and certainty as possible.

Thus far we have passed by short steps from the permissibility of rational suicide to the permissibility of assisted passive suicide and assisted active suicide. But we have also, perhaps without noticing it, taken another step as well: to passive euthana- sia. In the case I cited of assisted passive suicide, the physician assists the patient to disconnect himself from the life-support systems. But by refusing to collaborate, or by actively preventing the patient from disconnecting himself, the physician could have kept the patient alive. Therefore this is a case in which the physician allows the patient to die. If the physician is motivated to do this at least in part by the true belief that death would be in the best interest of the patient, her action constitutes an in- stance of passive euthanasia. It seems, in fact, that this is a case in which the cate- gories of assisted suicide and euthanasia overlap: the patient commits suicide, with the physician’s assistance, by refusing to continue to be saved, or kept alive; yet the physician also allows the patient to die, and does so with the intention of thereby ben- efiting him.

It would, perhaps, be a more obvious case of passive euthanasia if the physician alone turned off or disconnected all the life-support systems. Assuming that the sys- tems she withdrew were all ones that she had herself been providing, this would be an instance of allowing the patient to die and not a case of killing—in short, a case of passive euthanasia, despite the fact that allowing the patient to die required action rather than inaction. Still, in either case—whether the physician assists in the with- drawal of the life-support systems or removes them entirely by herself—most of us acknowledge the permissibility of her action. We accept, in short, the permissibility

Endings 459

of at least some instances of voluntary passive euthanasia. For it is widely accepted that a physician not only may but must comply with a patient’s competent request for the termination of life-supporting medical aid. And it would seem absurd to suppose that, while the physician could permissibly remove the patient from life-support systems, she could not permissibly do so with the intention of benefiting the patient by enabling him to die.

Passive euthanasia is, however, as far as we as a society have so far been willing to go. Many people, at least when they consider the matter in the abstract, think that there is a sharp moral line—one that must not be crossed—between allowing some- one to die when it is in his interest to die and actually killing him. But when one con- siders the matter closely, it seems that the step from passive to active euthanasia is again a very short one. Consider, for example, the activities of the notorious Dr. Jack Kevorkian. (Let us put aside whatever reservations we may have about his character and methods: his publicity seeking, megalomania, insufficient knowledge of his “clients,” and so on.) His earlier cases were instances of assisted suicide: he hooked people up to a device containing a lethal chemical, but the people themselves actu- ally pressed the button that released the chemical into their bloodstream. By contrast, in a more recent case, the person—a man named Youk—suffered from amyotrophic lateral sclerosis and was so disabled that it was difficult for him to push the button. He therefore asked Kevorkian to push it for him, which Kevorkian did, with a video camera rolling all the while. Most of us, on reflection, find it difficult to believe that it could make a momentous moral difference whether Youk pushed the button him- self or whether Kevorkian pushed it for him. Yet that is all that the difference between assisted suicide and killing amounts to.

In addition to the fact that it seems insignificant who pushes the button, provided that the person whose death it causes autonomously wants it to be pushed, there is an- other feature this case that supports the inference from the permissibility of passive euthanasia to the permissibility of active euthanasia. This is Youk’s apparent inabil- ity to press the button himself. Most people who desperately want to die are capable of killing themselves and thus seek the assistance of others only in order to ensure that their own action is painless, minimally shocking to others, and successful. If we accept the permissibility of rational suicide (subject, of course, to certain con- straints), and in particular if we accept the permissibility of assisting in the commis- sion of rational suicide, we are acknowledging that people have a right to die when life has become an intolerable burden and when continued life is not demanded by consideration of others. If, however, we reject the permissibility of active euthanasia, we are effectively denying that right to those who are so disabled that they cannot take their own lives, even with assistance from others. In order for those people to es- cape from an intolerable existence, it is necessary for someone else to kill them. If we forbid others to kill them, when there are others who are willing to do so, we are ex- acerbating the already quite terrible hardships these people have had to bear. Thus some advocates of the rights of the disabled have claimed, with considerable plausi- bility, that a policy that permits suicide but forbids active euthanasia unfairly dis- criminates against the disabled.

Despite these considerations, many people are unable to evade the nagging sense that killing another person is fundamentally different from allowing him to die or

460 the ethics of killing

even assisting him to kill himself. Yet how can they explain why, intuitively, it seems to make no difference whether Youk pressed the button himself or whether Kevorkian pressed it at Youk’s request? Some philosophers have contended that cases of this sort—in which the distinction between killing and letting die seems to have no moral significance—show that the distinction is, in these cases and all others, devoid of sig- nificance. They argue that, because we are wrong to suppose that the distinction has moral significance, we must acknowledge that, in the absence of contingent differ- ences, active euthanasia must be permissible when passive euthanasia would be—par- ticularly in cases in which merely allowing a person to die would prolong his agony.55

I believe that there is a different and more plausible explanation of why it is no worse for Kevorkian to press the button, thereby killing Youk, than it is for Youk to press it himself. This explanation is compatible with, and indeed presupposes, the be- lief that the distinction between killing and letting die is, in general, morally signifi- cant. It assumes that, if the distinction has moral significance, that must be because it is an instance of the more general distinction between doing and allowing. Most people believe that a person’s actively bringing about an outcome that would not have occurred without his intervention has a different moral significance from a person’s allowing that outcome to occur. Perhaps the active causing of the outcome ties the agent more closely to it, making him more responsible for it than he would have been if it had occurred even in his absence. I will not attempt to defend any particular ex- planation of the moral significance of the distinction between doing and allowing. If, however, this broad distinction does have a general significance, it offers an explana- tion of why killing is in general more seriously wrong than letting someone die. If, in general, the mode of agency involved in actively intervening to bring about an out- come is morally more significant than that involved in allowing an outcome to occur, it seems that doing harm must in general be more seriously morally objectionable than allowing harm to occur. Because death is normally harmful to those who die, it follows that killing must in general be more seriously objectionable than letting die.

Notice, however, that the significance of the distinction between doing and allow- ing should be reversed in cases in which the outcome is good rather than bad. If doing harm is worse than allowing harm to occur, actively benefiting someone should, in general, be better than merely allowing someone to be benefited, even when all other things (such as motive, intention, cost to the agent, and so on) are equal. If that is right, it explains why it was not worse for Kevorkian to press the button, thereby killing Youk, than it would have been for him to have allowed Youk to die by allowing him to press the button instead. For Youk’s death was unusual in not being harmful to him. It was instead beneficial; hence, if benefiting someone is better than merely allowing him to be benefited (for example, allowing him to benefit himself), Kevorkian’s ac- tively bringing about that death was, if anything, better or more praiseworthy than merely allowing or enabling Youk to bring it about would have been.

Because, in cases of euthanasia, death is beneficial rather than harmful, active eu- thanasia should in general be better or more imperative than passive euthanasia.56 To suppose otherwise is to divorce the distinction between killing and letting die from its source in the distinction between doing and allowing, and to treat the general pro- scription of killing as a taboo rather than as a rational moral requirement to which there are intelligible exceptions. It is, of course, desirable that each of us should be

Endings 461

profoundly averse to killing other human beings and that this aversion should be sed- ulously inculcated as a fundamental element of each person’s moral education. But the aversion should retain its grounding in the reasons we have for refraining from killing; it should not degenerate into an indiscriminate squeamishness. If, for example, the reason not to kill persons derives from a requirement of respect for persons, we should seek to prevent our aversion to wrongful killing from spilling over into cases in which killing is compatible with, and perhaps even required by, respect for a person. I will consider presently whether voluntary active euthanasia is a case of this sort.

The argument from suicide to euthanasia starts from cases in which suicide is both prudentially and morally justified. Often, of course, suicide is either pruden- tially or morally unjustified, or both. A great many people commit suicide in the mis- taken belief that their lives are not or will not be worth living. This is an easy mistake to make: one’s predictions about the character and content of one’s future life may be faulty; and, even if they are accurate, one may inadequately imagine what such a life would be like or fail to appreciate one’s ability to adapt oneself to it. Thus even if one accepts that suicide and euthanasia can in principle be justified, one should recognize that there will be cases in which it will be justified to restrain a person from commit- ting suicide, or to deny a person’s request for assistance in committing suicide or for “euthanasia,” on the ground that the person is mistaken in believing that his life is worth not living. The difficulty of discriminating between these cases and those in which the person’s life is genuinely not worth living is, of course, one of the major problems with accepting the legitimacy of assisted suicide and euthanasia.

Let us, however, confine our attention to cases in which it is true, and all those in- volved can agree that it is true, that a person’s life is worth not living. Even in these cases, suicide or euthanasia may be unjustified because of its effects on others. There are individual cases in which a person’s death would be so harmful to others that the person ought morally to endure a miserable existence for the sake of those others. These cases are, however, comparatively rare. For those who care enough about a person to be devastated by his death are normally motivated by love to give that per- son’s interests priority over their own. Still, there remain some cases in which a pru- dentially rational suicide is morally unjustified because of its effects on others—for example, a suicide by a single parent that would leave small children emotionally maimed and physically and socially vulnerable.

Even if suicide or euthanasia can be prudentially and morally justified in individ- ual cases, there are various objections, based on a consideration of possible side ef- fects, to accepting the legitimacy of either as a matter of social policy. I will consider some of these objections very briefly in section 2.4. First, we should consider whether there are convincing fundamental objections to these practices—objections that claim that either suicide or euthanasia is an instance of wrongful killing for rea- sons that are independent of such contingencies as possible effects on others and the possibility that a person’s evaluation of his own life is mistaken.

I have suggested that, if there are cases in which suicide is prudentially rational and morally permissible, we should conclude that, if other things are equal, voluntary active euthanasia is permissible in such cases as well. Some people, however, believe that, even when suicide is permissible, euthanasia is not. They believe that who brings about the death is fundamentally important. I believe that this is a perversion

462 the ethics of killing

of certain deontological intuitions about agency. Deontologists rightly recognize that there is a sense in which it matters who does a certain act: for example, it may be wrong for me to do a certain act (such as an act of killing) even if this would prevent others from doing more acts of exactly the same sort. But when it is permissible for a person to do a certain act, it should also be permissible for a third party to do that act for him, at his request, provided that in doing the act the third party would not be unfairly favoring the one person over another. Thus if it would be prudentially ra- tional for a person to kill himself and his death would not be worse for others, it should be permissible for a third party to kill him at his request. To believe otherwise is, it seems, to treat the killing of another as a taboo, an act that taints the soul of the agent irrespective of the conditions in which the act is done.

There are two other possible views. One is that, while suicide is wrong, euthana- sia can be permissible. Like the previous view, this view assumes that it is of decisive importance who performs the act of killing. But because it attributes to third parties a prerogative with respect to the lives of others that it denies to the individuals them- selves, it is a view that no one accepts. The final possibility is that there is a funda- mental objection to both suicide and euthanasia that shows that both are wrong. If correct, this view obviously undermines the conditional argument I have given for the permissibility of voluntary active euthanasia. It is therefore important to consider whether there really is a plausible fundamental objection to suicide and euthanasia.

My strategy here will be to argue for the permissibility of suicide and euthanasia by negation or exclusion—that is, I will try to show that, of the various accounts of the wrongness of killing, those that have some plausibility fail to condemn suicide and euthanasia. There are, I will claim, no good reasons for thinking that suicide and euthanasia are fundamentally wrong—that is, wrong for reasons other than those concerned with merely contingent effects. This kind of strategy can never, of course, be decisive, for one can never be certain that one has taken account of all the plau- sible objections or all the plausible accounts of the wrongness of killing. But because there are comparatively few accounts of why killing is in general wrong, the strategy may have considerable force in this particular context.

I will limit my discussion to broadly secular accounts of why killing is wrong. There are various objections to suicide and euthanasia that are necessarily theologi- cal in character—that is, that make sense only in the context of a set of theological beliefs. (Thus I do not include among these arguments those that claim that God pro- hibits suicide or euthanasia for reasons that can be understood and justified inde- pendently of claims about God’s will, preferences, or plans.) Among the arguments that are necessarily theological are those that claim that only God has the right to de- termine when a human life should end, that human suffering serves God’s purposes in ways we cannot understand, that how we respond to the burdensome aspects of life may in part determine our destiny in the afterlife, and so on. I believe that all of these objections fail even in their own terms: they imply, for example, that it is wrong to save people’s lives or to alleviate their suffering, or they attribute motives and inten- tions to the deity that would be shameful in a human being (for example, a parent).57

Still, to do full justice to these arguments, one would have to challenge their theo- logical underpinnings, and that would take us too far afield here. So I will leave these arguments aside.

Endings 463

The more obvious objections to killing that are secular in character seem simply not to apply to rational suicide and voluntary active euthanasia. When most of us begin to reflect about the morality of killing, our initial thought is that killing is wrong because of the dreadful effect it has on the victim: it deprives him of the whole of the good life he would otherwise have had. This intuition is the foundation of the Harm- Based Account of the wrongness of killing and also, when suitably refined, of the Time-Relative Interest Account. I have argued that neither of these is ultimately suc- cessful as an account of the wrongness of killing persons; but the relevant point here is that neither condemns rational suicide or euthanasia. For in these cases death is by hy- pothesis not bad for, or against the time-relative interests of, the person who is killed.

Another common objection to killing is that it overrides the autonomy of the vic- tim, usurping a literally vitally important prerogative that is properly the victim’s own. This, I believe, is a critical part of the explanation of why killing persons nor- mally involves a violation of the requirement of respect for persons: for respect for a person consists in part in a proper deference to the determinations of his autonomous will in matters within his own rightful sphere of control. (What the boundaries of that sphere are is, of course, a controversial question; but, if we put aside theological claims about God’s dominion, choices that determine how a person’s life will go and that do not affect the rights of others must surely lie at the very center of that sphere.) Again, however, neither rational suicide nor euthanasia that is genuinely voluntary can possibly violate the autonomous will of the person killed. Nor, for that matter, can nonvoluntary euthanasia: for, by definition, that involves the killing or allowing to die of an individual who lacks an autonomous will. (Involuntary euthanasia, by contrast, does involve a violation of the autonomous will of the person who is killed or allowed to die, and it is precisely for this reason that it can never, in practice if not also in principle, be justified.)

Another way in which people frequently articulate their opposition to killing is to claim that it violates the victim’s right to life. But, as many rights theorists have pointed out, rights can be waived. A person who kills himself or autonomously re- quests to be killed clearly waives his right not to be killed. Hence suicide and volun- tary euthanasia cannot be objectionable on the ground that they violate the right to life.

Because these familiar objections do not apply, those who have objected to suicide and euthanasia on grounds that are neither theological nor concerned merely with side effects have tended to appeal to one of the other of two notions: the sanctity of life and the notion of respect for the worth of persons. I will begin with the former.

2.2. The Sanctity of Life, Again

It is sometimes held that life—or human life—has a special value or sanctity that is independent of the value that it has either for the individual whose life it is (its per- sonal value, in Ronald Dworkin’s terms) or for others (its instrumental value). Be- cause it is not value for anyone, this value is impersonal. It is sometimes held, there- fore, that this form of value can be present—indeed, is present—in a life even when the continuation of the life is bad for both its possessor and others. When in 1990 the Supreme Court ruled on the Cruzan case, in which the state of Missouri claimed the right to keep the body of a young woman in a PVS alive against the wishes of her par-

464 the ethics of killing

prefers death to continued life, neither suicide nor euthanasia is incompatible with re- spect for that person’s worth. We can concede that persons have a value that is inde- pendent of their good, yet hold that respect for that value is nevertheless manifested in an active concern for their good, at least when that coincides with respect for their autonomous will. There is simply no sense in which a person’s worth is upheld or af- firmed by his mere persistence through suffering.

2.4. Nonvoluntary Euthanasia

The discussion has thus far been confined to voluntary euthanasia. It is generally agreed that nonvoluntary euthanasia is more problematic, and more difficult to jus- tify. There are two types of case in which the question of nonvoluntary euthanasia might arise: first, cases involving individuals that have never been self-conscious and thus have never been able to have or to express a rational preference between death and continued life; and, second, cases involving individuals who were once persons (that is, self-conscious and minimally rational) but have irreversibly lost the capacity to deliberate competently about whether it would be better for them to die or to con- tinue to live. Cases of this second sort raise particular difficulties. It is arguable, and probably true, that former persons remain within the scope of the requirement of re- spect for persons. Their former preferences, at least, continue to exert their moral au- thority over us. An individual who has lost the capacity to deliberate about life and death and whose life now seems, to third parties, to have ceased to be worth living, may formerly, when competent, have been opposed to being euthanized in these cir- cumstances. Or he may have expressed a desire to be euthanized. Or, finally, he may have had or expressed no view at all. If he was formerly set against being euthanized, that seems to constitute a decisive reason not to kill him, though there are limits to what others may be required to do to sustain his life. If, by contrast, he earlier ex- pressed an informed preference to be killed if his condition were to become as it is now, to kill him might be an instance of permissible voluntary euthanasia. Yet a pref- erence formed in advance of the situation may, for various reasons (some of which I will explore in section 3), have less authority than a contemporaneous one.

Matters are even more complicated if an incompetent individual whose life now seems to be a burden to him never, when competent, expressed a preference about what should be done to him in his present condition. It is arguable that to kill him could be compatible with the requirement of respect. For his rational nature, having already been subverted, would not be sacrificed, his will is silent, and his good seems to demand that his life should cease. Yet there are nagging worries about the possible consequences of permitting the killing of former persons in these circumstances. The commonest fears are that mistakes would be made, that abuses would occur, and that the acceptance of killing in these cases would erode our sensitivity to the value of life and engender an increasingly callous and promiscuous attitude toward killing.

These same fears are often expressed about voluntary euthanasia. But, when eu- thanasia would be fully voluntary, they seem exaggerated. Doubtless some mistakes would occur: occasionally someone would be killed who was misdiagnosed or would have been among the tiny minority who recover against the odds or for whom a cure is unexpectedly found. Yet rational people must decide for themselves whether to

Endings 485

take this risk—as we recognize when we allow people to decline certain treatments despite the risk that the best predictions about the outcomes of accepting or rejecting treatment may be mistaken. Moreover, even if mistakes are statistically certain, the costs of permitting voluntary euthanasia have to be weighed against the equally cer- tain and probably much greater costs of denying people a release from great suffering.

The risk of abuse is also real but the rational response is the one we have adopted in the case of the even greater risk of abuse in permitting killing in self-defense. Our acceptance of the permissibility of killing in self-defense offers significant opportu- nities for people to perpetrate wrongful killings under the guise of self-defense—for example, by provoking a person to violence and claiming that killing him was neces- sary to save oneself, or by killing, in a secluded spot, a person with a known history of violent aggression and then claiming that one killed in self-defense. Although we are aware of these possibilities of abuse, we do not respond by forbidding killing in self-defense. Instead we erect safeguards against the abuse.

The same response is available to the problem of the “slippery slope.” If we know where the line between permissible and impermissible killings lies, we can take pre- cautions to prevent ourselves from being seduced into crossing it. We seem to have managed this in the case of killing in self-defense, even though the motives and forms of justification for self-defensive killing (for example, “he was at fault and it was him or me”) easily blur into rationalizations for wrongful killing. It seems, more- over, that resistance to euthanasia has its own slippery slope. If those who advocate euthanasia are in danger of becoming less sensitive to killing, it seems equally true that those who seek to deny others the option of euthanasia are in peril of becoming inured and desensitized to the suffering of others and more willing to tolerate it in all areas of life.

The requirement to obtain a person’s unforced consent before he can be eutha- nized constitutes a powerful safeguard against these various problems. When that safeguard is removed, however, as it is in nonvoluntary euthanasia, the scope for these problems is increased. These problems are, however, less serious if the subject of nonvoluntary euthanasia is not a former person but an individual that has never been a person—for example, an animal, a fetus, or a newborn infant. In these cases, the problems are somewhat less likely to arise: for there are fewer incentives for abuse, lines are easier to draw with precision (for example, age limits), and so on. Perhaps more importantly, even if these problems would arise with the same frequency as they would if euthanasia were permitted for former persons, they would nevertheless be less serious when they occurred. This is because individuals that have never been persons are, if my earlier arguments were sound, below the threshold of respect. So, for example, instances of killing that would be abuses of a policy allowing nonvol- untary euthanasia would not involve a violation of the requirement of respect, as they very likely would if the victim had once been a person. For these reasons, it is at best highly speculative to object to permitting nonvoluntary euthanasia for individuals that have never been persons on the ground that it would have pernicious side effects.

If, therefore, there are strong moral reasons for permitting nonvoluntary euthana- sia for individuals that have never been persons, those reasons are likely to outweigh concerns about possible side effects. Yet, if the account I have developed of the morality of killing is correct, there are doubts about whether there can actually be

486 the ethics of killing

strong reasons that favor euthanasia in these cases. In the remainder of this section, I will explain why this is so and discuss the problems it raises for the position I have developed.

In order to explain the problem, I will focus on nonvoluntary euthanasia for ani- mals. I do this because most people accept the permissibility, and desirability, of eu- thanasia for animals. It seems obvious that there can be strong reasons for euthaniz- ing animals in some cases. It is important to realize, however, that most aspects of the following argument also apply to euthanasia for human beings that have never been persons—for example, fetuses and newborn infants. But, because their membership in our species makes us think of fetuses and infants differently, and to doubt the per- missibility of subjecting them to euthanasia, it will be easier to bring out the prob- lems I want to discuss by focusing on the case of animals.

It will help to recall the discussion in section 1.2 of chapter 3 of what I called the view that suffering matters more—that is, the view that animals differ from human beings in that, although their suffering may matter considerably, their lives cannot. I suggested that this view draws support from the fact that an animal’s time-relative in- terests are, in general, strongest with respect to the character of its present experi- ence. Thus its time-relative interest in avoiding present suffering can be quite strong. But, because the prudential unity relations within its life are weak, the strength of its present time-relative interests in its own possible future states declines as the objects of those interests recede into the future. Because its present time-relative interest in having the goods of its own future life is comparatively weak, its present time- relative interest in continuing to live is correspondingly weak. I argued, however, that these facts do not support the conclusion that it can often be justifiable to kill an an- imal as a means of preventing it from suffering. To the extent that euthanasia may be more often rational in the case of an animal than in the case of a person, this is be- cause there is ample scope for suffering within the life of an animal though only lim- ited scope for good, since an animal’s capacity for well-being is limited by its psy- chological nature. Thus it may more often be true of an animal’s future life than of a person’s that the expected good is outweighed by the expected suffering, so that the individual’s present time-relative interest is in avoiding rather than having that future.

To this earlier discussion we can now add a further consideration that emerged in the discussion of prenatal injury in chapter 4. This is that our concern for an individ- ual’s time-relative interests should include a concern for its future time-relative in- terests. Thus, even though an animal’s present time-relative interest in avoiding suf- fering in the distant future may be very weak, the time-relative interest it will have at that later time may be quite strong, and that interest can ground a strong reason now not to do what would cause that suffering to occur, or to do what would prevent it from occurring.

This further consideration yields a rather stronger asymmetry than I acknowl- edged in chapter 3 between one’s reason for concern about an animal’s life and one’s reason for concern about its suffering. One’s reason not to kill an animal (or to save it) for its own sake derives from its present time-relative interest in continuing to live, which is comparatively weak, for the reasons given earlier. But one’s reason not to cause it to suffer (or to prevent its suffering) in the future derives both from its present time-relative interest in avoiding future suffering and from the stronger time-

Endings 487

relative interest it will have, if it lives, at that later time. If the future suffering would be great—that is, intense and protracted—one’s reason not to cause that suffering may actually be stronger than one’s reason not to kill the animal.

This asymmetry is, in part, a result of the fact that, although one’s reason not to cause future suffering is strengthened by the animal’s future time-relative interest in avoiding suffering, one’s reason not to kill it cannot be strengthened by the time- relative interests it might have in the future in having goods at those later times. For to kill an animal is to ensure that it will have no future time-relative interests. Thus one can prevent possible later time-relative interests from constraining one’s present action by preventing them from arising—in this case by killing their potential bearer now. (This is compatible with the recognition that an animal’s future time-relative in- terests in having goods can ground present reasons to ensure the satisfaction of those interests if the animal will live.) The idea that one can avoid being constrained by possible later time-relative interests by preventing them from arising has little appli- cation, of course, in the case of persons—both because persons have strong present time-relative interests in their future lives (including interests in whether they will later have and be able to satisfy further interests) and because there are reasons not to kill persons other than those that derive from respect for their time-relative interests. In the case of animals, however, we are constrained only by their interests, and the only interest that an animal can have independently of whether or not one kills it that is relevant to whether it is permissible to kill it is its present time-relative interest in continuing to live.

With these considerations as background, let us now examine the case in favor of euthanasia for animals. Suppose one discovers a wild animal in the early stages of an incurable disease that, after a few months, will begin to cause it agonizing pain that will then continue unabated for several months until the animal finally dies. The only way to prevent that future suffering is to kill the animal now. If one has the means to kill it painlessly, one would seem to have a strong moral reason to kill it to spare it those months of agony. But notice that the animal’s present time-relative interest in avoiding that future suffering may be comparatively weak because the prudential unity relations between itself now and itself several months from now would be weak. Thus, if one’s only reason to euthanize the animal derives from its present time-relative interest in avoiding future suffering, that reason should be compara- tively weak. Yet it seems that the reason would be quite strong.

In short, the appeal to an animal’s present time-relative interest in avoiding future suffering seems insufficient to explain our sense of the importance of euthanasia in a case of this sort. The problem of explaining and justifying the common view is even more acute for those numerous philosophers who accept what, in section 2 of chap- ter 4, I called the Capacity Condition—that is, the view that something can be in- trinsically good or bad for an individual only if that individual is capable of desiring it or caring about it in some way. These philosophers embrace the implication of the Capacity Condition that death cannot be bad for individuals that are incapable of de- siring either continued life itself or the goods that their lives might contain in the fu- ture. Hence they claim that death cannot be a misfortune for a non-self-conscious an- imal or a fetus. But their claim also has a parallel implication that death cannot be good, or better than continuing to live, for individuals that are incapable of desiring

488 the ethics of killing

to die or to avoid future suffering, even if these individuals can avoid a prospect of great future suffering only by dying. If, therefore, one accepts this view, it seems that one must accept that there can be no reason, for an animal’s own sake, to kill it in order to prevent it from experiencing great suffering in the future.

How can we defend the plausible view that there can be a strong reason to kill an animal painlessly if that is the only way to prevent it from experiencing intense and protracted suffering in the future, suffering that would greatly outweigh any good that its future might also contain? It might be suggested that, just as an animal’s fu- ture time-relative interests in avoiding suffering can ground a reason not to do now what would cause it to suffer later, so those same interests can ground a reason to kill the animal now if that is the only way to prevent those interests from being frustrated. The problem with this suggestion, however, is that if one kills the animal now, it will never have any future interest in avoiding suffering. The interest to which this sug- gestion appeals is not, in other words, a future interest relative to the agent deliberat- ing about whether to kill the animal. It is, instead, a merely possible interest, for whether it will exist depends on the outcome of the agent’s deliberation—that is, on whether or not the agent kills the animal.

It seems, therefore, that if one has a reason to euthanize an animal that is based on considerations other than the animal’s present interests, it must appeal, not to the animal’s future interests, but to its possible future interests. But this raises a different problem. If an individual’s possible future interests can ground present reasons for action that are independent of reasons generated by the individual’s present time- relative interests, its possible interests in having goods should, it seems, matter in the same way that its possible interests in avoiding suffering do. In particular, if one can have a reason to prevent the existence of a possible interest in avoiding suffering if that is the only way to prevent it from being frustrated, it seems that, by parity of rea- soning, one can also have a reason to ensure the existence of a possible interest in having goods in order to enable it to be satisfied. Possible interests in avoiding suf- fering and in having goods should be treated symmetrically. But this means that, if an individual’s possible future interests in avoiding suffering can ground a reason to kill that individual, its possible future interests in having goods can similarly ground a reason not to kill it, or even a reason to save it.

To concede this latter point would, however, be fatal to my earlier argument for the permissibility of late abortion. For that argument hinges on the claim that the only significant reason not to kill a developed fetus for its own sake derives from its pres- ent time-relative interest in continuing to live. If the interests in having goods that a fetus would later have if it is not killed can ground a reason not to kill it (in the same way that the interest in avoiding suffering an animal would later have if it were to live can ground a reason to kill it), there would then be strong reasons not to kill a devel- oped fetus whose future life could be expected to be worth living.

This is not a problem only for my argument in support of abortion. There are wider implications. If it is true that an individual’s possible future interests in having goods can ground a strong reason to preserve its life despite its having only a weak time-relative interest in continuing to live, it seems that we should have a strong rea- son to preserve the lives of fetuses by preventing spontaneous abortions and miscar- riages. We should do this, not just for the sake of the potential parents, but for the

Endings 489

sake of the fetuses themselves. For their possible future interests, according to this view, make demands on us now. Given that more than two-thirds of the conceptions that occur result in spontaneous abortion, those who believe that we begin to exist at conception would have to accept that these demands are quite extensive. If the pos- sible future interests of a fetus make its life matter now, the death prior to birth of two-thirds of those conceived must surely constitute a continuing holocaust.

It seems that we face a dilemma. If interests that an individual will later have only if it continues to live cannot ground present reasons either to preserve or to terminate its life, it seems that one’s only reason to euthanize an animal, for its own sake, must derive from its present time-relative interest in avoiding future suffering. But that in- terest may be insufficiently strong to account for the strong reason it seems there would be to euthanize an animal whose future would otherwise consist mainly of in- tense and protracted suffering. Alternatively, if an individual’s possible interests can ground a reason either to end or to preserve its life, we could appeal to an animal’s possible future interest in avoiding suffering to strengthen the case for euthanasia. But the idea that these interests are relevant seems not only to exaggerate the impor- tance of preserving animal lives but also to oppose the permissibility of abortion and to demand significant efforts to prevent the deaths of fetuses in utero. It seems, there- fore, that the common sense view must reconcile two claims: first, that an individ- ual’s possible future interest in avoiding suffering can ground a reason to end its life now, if that is the only way to prevent the suffering, and, second, that an individual’s possible future interests in having goods cannot ground a reason not to end or to pre- serve its life now.

One possible defense of the commonsense view appeals to the idea that there is an asymmetry between the importance of preventing suffering and promoting the good.92

It would, of course, be implausible to suppose there is no reason to promote an indi- vidual’s good. But it could be maintained that one’s reason to promote an individual’s good is in general substantially weaker than one’s reason to prevent that individual from suffering. If this claim were conjoined with the claim that an individual’s merely possible interests can ground present reasons for action, that would support a view similar to the commonsense view. But, while this conjunction of claims would mitigate our problem, it would not solve it. For the two claims still imply that an in- dividual’s merely possible interests in having goods ground a reason to preserve its life, though not so strong a reason as the reason one has to prevent its suffering.

Another possibility is to claim that, although merely possible interests count, our concern should always be to prevent the frustration of interests rather than to ensure their satisfaction.93 Provided that we treat the elimination of an interest as tantamount to its frustration, this claim implies that we always have reason to satisfy existing in- terests in order thereby to prevent their frustration. But, with regard to merely pos- sible interests, it implies that, while there is a reason to prevent them from arising if they would otherwise be frustrated, there is no reason to cause or to allow them to arise just so they could be satisfied (unless, of course, the individual has an existing interest in having new interests arise).

This may seem promising. It allows that an individual’s present time-relative in- terest in continuing to live grounds a reason not to kill it or to save it. But it denies that this is true of merely possible future interests in having goods, even if they would

490 the ethics of killing

be satisfied if the individual were to live. Finally, it accepts that, if certain interests (whether in avoiding suffering or in having goods) would be frustrated if they were to arise, this grounds a reason to prevent their existence, if necessary (and if other things are equal) by preventing the future existence of the individual whose interests they would be. In short, this view seems to support the commonsense view: it implies that an individual’s possible future interests can strengthen the case for euthanasia but cannot ground a reason not to kill or to save that individual.

This view seems, however, to be vulnerable to a decisive objection. Consider any individual that has never been a person—an animal or fetus, for example. If this in- dividual has a time-relative interest in continuing to live, this grounds a reason not to kill it. But, no matter how many interests it might have in the future and no matter how many would be likely to be satisfied, these possible interests do not, on the view we are considering, add to the case against killing this individual. But any interests it might later have if it were to continue to live that would unavoidably be frustrated— and there are likely to be many of these—do, on this view, count against allowing the individual to continue to live. If the combined strengths of these interests would out- weigh the individual’s present time-relative interest in continuing to live, this view implies that it would be better, for this individual’s own sake, to kill it now, even if its future life would on balance be well worth living. This is clearly unacceptable. (It is worth noting that the claim that there is an asymmetry between the importance of preventing suffering and the importance of promoting an individual’s good may have similar implications.)

One could, of course, stipulate that, while the prospect of satisfying an individ- ual’s possible future interests does not by itself provide a positive reason to cause or allow that individual to continue to exist, it is capable of weighing against and po- tentially canceling out the reason to prevent the individual from continuing to exist that derives from the expectation that many of its possible interests would be frus- trated. This suggestion seems, however, essentially ad hoc. It is hard to imagine how it might be defended. If the prospect of satisfying possible future interests can weigh against the reason to prevent the frustration of possible interests, why can it not also ground a reason not to kill the individual whose interests they would be?

I know of no satisfactory solution to the problem discussed in this section. This should not be surprising. For this problem precisely parallels an intractable problem in the area of ethics concerned with causing people to exist—a problem that has perennially defeated the many attempts to solve it. This problem is also concerned with possible future interests—not those of existing individuals but of individuals who do not now exist but might exist in the future. The question we have considered in this section is whether an individual’s possible future interests can ground a reason to prevent or to cause that individual’s continued existence. The parallel question in population ethics is whether an individual’s possible future interests can ground a reason to prevent or to cause that individual’s existence.

Suppose we are deliberating about whether to cause a person to exist. If we can predict that the person’s life would be filled with suffering that would greatly out- weigh any good the life might contain, it seems that we have a strong reason not to cause that person to exist. This parallels the reason we might have to euthanize an an- imal whose future life would similarly be filled with uncompensated suffering. One

Endings 491

difference, however, is that in the case of the possible person, our reason to prevent the possible suffering cannot be grounded in any present interest (time-relative or otherwise) of the victim—for the victim does not exist. Our reason may, therefore, derive from the possible person’s possible interests. Our reason may be to prevent those interests from arising in order to prevent their frustration. Or our reason may be impersonal in character: for example, a reason not to cause an increase in the amount of suffering in the world. (The reason to euthanize an animal could also be imper- sonal in the same way. The reason that I have not invoked this explanation will soon become obvious.)

The problem here is that, whether our reason derives from the possible person’s possible interests or from impersonal considerations, it seems that the same consid- erations should ground a moral reason to cause the person to exist if his life would be worth living. If, for example, there is an impersonal reason to prevent suffering, there should be a corresponding impersonal reason to promote the good. Thus, if the ex- pectation that a person would have a life that would be bad for him, or worth not liv- ing, grounds a reason not to cause him to exist, it should also be true that the expec- tation that a person would have a life that would be good, or worth living, grounds a reason to cause him to exist. But most of us reject this second claim. It is not that we believe that the reason to cause a person to exist because his life would be good is normally outweighed (for example, by considerations of cost to the agent); rather, we believe that there is no such reason.

In short, we believe that the suffering that a person would experience grounds a reason not to cause, or to prevent, his existence. This reason may be decisive if the suffering would not be outweighed by compensating goods. But we also believe that the good that a person’s life would contain does not ground a reason to cause him to exist. Earlier, in section 5 of chapter 4, I referred to this conjunction of beliefs as the “Asymmetry” and noted that it is very difficult to defend. If we say that a person’s suffering, or the frustration of his possible future interests, counts against causing him to exist, but his possible good, or the satisfaction of his possible interests, does not count in favor of causing him to exist, it seems that there will always be a strong moral presumption against causing people to exist. And that is surely wrong. If we were to say, instead, that a person’s possible good does count in favor of causing him to exist, though less strongly than his possible suffering counts against it, we would have to conclude that it is generally objectionable to cause people to exist, if other things are equal, though it may be morally required to cause people to exist if their lives would be unusually well worth living, again if other things are equal. This too is implausible.

The idea that there can be a strong reason to euthanize an animal that is inde- pendent of its present time-relative interest in avoiding future suffering parallels the view that there is a strong reason not to cause a person to exist if his life would be dominated by suffering. And the idea that there is not a strong reason to prevent a spontaneous abortion that is independent of the fetus’s time-relative interest in con- tinuing to live (and of the interests of others) parallels the view that there is no rea- son to cause a person to exist just because his life would be worth living. These views run parallel because the prudential unity relations that would bind an animal or a fetus to itself in the future are very weak. Its relation to its own future is, figuratively

492 the ethics of killing

speaking, almost wholly impersonal. For this reason, as I argued in section 6.1 of chapter 2, the death of a fetus (or an animal) is intermediate between the death of a person and the failure of an individual to come into existence. Indeed, the future lives of animals and fetuses seem to matter more in the way the possible futures of possible people matter, and less in the way that the possible futures of existing people matter.

The positive justification for animal euthanasia therefore remains elusive. An an- imal’s possible future suffering clearly matters but exactly how or why it matters is hard to say. For the obvious claims have other implications that are difficult to accept. There is, however, some consolation in the fact that parallel problems beset our ef- forts to defend the view that there is a strong reason not to cause a person to exist if his life would contain so much suffering as to be worth not living. This view, too, simply must be right. But an adequate defense continues to elude us.

3. the withering away of the self

3.1. The Metaphysics of Progressive Dementia

There are various forms and causes of progressive dementia.94 The most common cause is Alzheimer’s disease, which involves progressive degenerative changes in the brain that, over a period of years, result in the gradual erosion of the victim’s mental life and psychological capacities. The disease usually presents with a deterioration of memory, which continues relentlessly through the later stages. As the disease pro- gresses, comprehension, reasoning, and judgment are increasingly impaired, ele- ments of character become unstable, linguistic abilities decline, recognition of other people becomes problematic and eventually impossible, and ultimately even self- awareness is lost. The final stage of the disease has been characterized as follows: “All language skills have been lost, and there seems to be very little left of the patient’s ‘self.’ . . . Motor skills decline until it is no longer possible for the person to walk, sit up, chew and swallow food, or control bowel and bladder. . . . As the brain shuts down, the patient becomes unresponsive, and finally coma ends in death.”95

Let us refer to a person in the very early stages of progressive dementia as the Pa- tient at Onset, and the individual in the later stages of the disease as the Demented Patient. As I noted in chapter 1, the Psychological Account of Identity has the sur- prising and implausible implication that the Patient at Onset and the Demented Pa- tient are not the same individual. For as the day-to-day psychological connections within the victim’s life become progressively fewer, a point is eventually reached at which the Demented Patient is no longer psychologically continuous with the Patient at Onset. I speculated that the point at which psychological continuity ceases to hold coincides roughly with the point at which the individual loses those psychological capacities that are constitutive of personhood. According to this view, then, the De- mented Patient is a post-person who succeeds the Patient at Onset in the latter’s own body. The person who contracts progressive dementia may thus cease to exist well before the disease has run its full course. (There are very few, if any, actual cases in which the Psychological Account implies that a person has been supplanted by a dif- ferent person in his own body. The only kind of case in which this might occur is a case in which brain damage causes total amnesia and a radical transformation of

Endings 493

character, but in which the psychological capacities constitutive of personhood are preserved.)

According to the Embodied Mind Account of Identity, by contrast, there are no cases in which progressive dementia results in a different individual, much less a dif- ferent person. As long as the victim remains a person, he is the same person as the Pa- tient at Onset. And even when the Demented Patient’s psychological capacities have dropped below the level of personhood, he is nevertheless the same individual as the Patient at Onset. As long as the victim’s brain retains the capacity to support con- sciousness, however minimal, the same individual remains in existence.

Perhaps it would be more accurate to say that there is no new or different indi- vidual. For what remains may not in any robust sense be the same individual but merely a fragment of that individual. The losses one suffers in progressive dementia are more than ordinary losses. As the description I cited of the late stages of the dis- ease indicates, progressive dementia gnaws away unrelentingly at the core of the self, eventually stripping it to the vanishing point. One way of understanding this process appeals to the notion introduced in section 2 of chapter 4 that the existence of the self may be partial, so that the self may come into existence or cease to exist gradually, by degrees. I have argued that we are essentially embodied minds. In progressive de- mentia, the elements of the mind and their physical bases in the brain are steadily worn away. It is possible to see this process as the gradual fading from existence of the individual himself.

Consider, for the sake of comparison, a complex physical object such as a car. A car may be dismantled piece by piece. The process of disassembly may be gradual and continuous. At the beginning there is a car; at the end there is no longer a car but merely a collection of parts. But there is no point at which the car suddenly ceases to exist, no point at which the removal of one further part makes the difference between the existence and nonexistence of the car. It is sometimes claimed that, as the car is dismantled, there is a period during which its existence is indeterminate—that is, a period during which it is not true either that the car continues to exist or that it has ceased to exist. As I suggested earlier, however, there is another possibility, which is that during this period the car does not fully exist yet continues partially to exist. On this view, it goes out of existence gradually, by degrees.

Analogous claims could be made about the victim of progressive dementia. Dur- ing the early phases of the condition, the person fully exists. But as the dementia pro- gresses, the individual himself begins to fade from existence. In the late stages, he is barely there at all. If we distinguish, as I have argued we should, between the biolog- ical death of the human organism and the death or ceasing to exist of a person, we could say that progressive dementia is a protracted process of dying—not dying in the traditional sense in which dying precedes and ends in death, but dying understood as a process in which death itself occurs by degrees.

If this is a coherent and accurate conception of progressive dementia, it seems that the Patient at Onset ought rationally to care less, in an egoistic way, about what will happen to him in the later stages of the disease—less, at least, than he ought to care if he were going to be fully preserved. Although the Demented Patient will not be a different individual, he will be only a ghostly remnant of the Patient at Onset. While the Patient at Onset will continue to exist as the Demented Patient, his exis-

494 the ethics of killing

tence then will be only partial; therefore he ought now to discount his egoistic con- cern for his future self for the extent to which his own existence will be diminished.

As I conceded earlier, the notion that existence may be partial is problematic. But the same intuitions that tempt us to see the Demented Patient as slowly fading from existence may be expressed and defended differently by noting that the basis for the Patient at Onset’s egoistic concern about the Demented Patient is radically attenuated. According to the Embodied Mind Account of Egoistic Concern, the prudential unity relations include physical, functional, and organizational continuity of the brain as well as the mental correlate of organizational continuity—namely, psychological con- tinuity. All of these relations may hold to varying degrees, and all hold much less strongly between the Patient at Onset and the Demented Patient than they do over a comparable period of time within the life of a cognitively normal adult person. For, with progressive dementia, areas of the brain involved in the generation of conscious- ness and mental activity gradually atrophy and die: the brain itself shrinks, its capaci- ties diminish, and the neurological bases of the victim’s psychological capacities and mental states, which are the elements of psychological continuity, gradually disap- pear. Between the Patient at Onset and the Demented Patient in the very final stages of the disease, all of the various prudential unity relations are present to only a very weak degree. Because of this, the Patient at Onset may rationally be only minimally egois- tically concerned about what will happen to him in the final stages of dementia.

It is important to be clear about the precise nature of this claim. It should not be confused with the profoundly mistaken claim that the Patient at Onset ought to be only minimally egoistically concerned about the prospect of progressive dementia it- self. About this he should now be intensely egoistically concerned. For the prospect of progressive dementia is figuratively, and perhaps literally, a prospect of gradual extinction. It involves the gradual erosion of the basis of egoistic concern about the future—the very foundations for having anything to care about in the future. In the absence of the prospect of dementia, a great deal of good would lie in prospect for this person, and the person would be related to that future good in the right way. But the prospect of dementia excludes most of these goods that he could otherwise look forward to, and greatly reduces the extent to which he can rationally look forward to those few goods that will be accessible to him in his demented state. Thus the threat that progressive dementia poses to the Patient at Onset’s time-relative interest in hav- ing a good future is almost as grave as the threat posed by death itself. But the prospect of dementia also means that the Patient at Onset should be less fearful of any positive evils that he will later suffer in his demented state. If, for example, he ex- pects to suffer a certain amount of physical pain when he is in the later stages of de- mentia, he should now anticipate that pain with less dread or fear than it would be ra- tional to feel if his basis for egoistic concern about the future were stronger.

Among those who accept that identity is not the basis of egoistic concern about the future and who also believe that the prudential unity relations between the Patient at Onset and the Demented Patient are very weak, most accept only that it is not ir- rational for the Patient at Onset to be egoistically concerned to a weaker degree about his own future than would be rationally required in the absence of dementia. Derek Parfit, for example, claims that, while it is “not irrational” for one to “have a discount rate with respect to the degrees of psychological connectedness” between

Endings 495

oneself now and oneself in the future, one is “not rationally required to have this dis- count rate.”96 I believe, however, that a stronger claim is warranted. We should accept that, insofar as the concern is egoistic in character, it is rationally required for the Pa- tient at Onset to be less concerned about his own future when the relations that ground egoistic concern—the prudential unity relations—are weaker. It is, in other words, irrational for the Patient at Onset to have the same degree of egoistic concern about his own future as he would have in the absence of the prospect of dementia.

Parfit’s claim is weaker and more congruent with common sense. Why insist on the bolder and less plausible claim that discounting for diminished psychological connectedness is rationally mandatory? One reason is that this seems to be demanded by consistency. Recall that progressive dementia is, in a rough way, a mirror-image of our early psychological development. In both cases, the prudential unity relations are weak between a given individual at different times. The difference is that the physical and psychological discontinuities between a fetus or infant and the person it will later become are the products of growth and development, while those between the Patient at Onset and the Demented Patient are the results of deterioration. I have claimed that it is rationally mandatory for third parties to be less concerned about the fetus or infant’s future for its own sake now. Assuming that the fetus or infant is going to continue to live, I believe, as I argued in section 3 of chapter 4, that third parties should be equally concerned about its future time-relative interests. But if their con- cern were only for the fetus or infant as it is now, they would be rationally required to discount the importance to it now of its own future life—that is, they would have to recognize that its time-relative interests in its own future life are greatly weakened by the weakness of the prudential unity relations between itself now and itself in the fu- ture. But if we are rationally required to discount the fetus’s or infant’s time-relative interests in its own future because of the weakened prudential unity relations, it seems that we should maintain this insistence in the case of the Patient at Onset as well— particularly given that the weakness of the prudential unity relations is in this case a result of deterioration rather than development. If discounting for weakened pruden- tial unity relations is mandatory rather than optional in cases involving growth and de- velopment, it seems it should be mandatory as well in cases involving deterioration.

It is important to bear in mind, however, that discounting for weakened pruden- tial unity relations is rationally mandatory in cases of progressive dementia only in- sofar as the form of concern is egoistic. My claim is only that, if he is thinking ego- istically from his present point of view, the Patient at Onset ought rationally to be less concerned about the particular goods and evils in his own future life. It does not fol- low that he must be less concerned simpliciter. If, for example, the basis of his con- cern is moral rather than egoistic, he may rationally care as much about his future, demented self as he cares about himself now, in the same way that he may rationally care as much about another person as he cares about himself.

3.2. The Moral Authority of Advance Directives

This understanding of the prudential significance of progressive dementia has certain implications for an unusually puzzling problem, which emerges in the following case:

496 the ethics of killing

The Advance Directive. A woman whose life has been devoted to creative intellec- tual work finds herself in the very early stages of Alzheimer’s disease. Given her na- ture, her values, and the character of her previous life, she believes that for her to continue to live in a demented state would be horribly degrading, worse than simply ceasing to exist. She accepts, however, that her life will continue to be worth living as long as she remains competent; therefore to commit suicide now would be bad for her. But, once she becomes incompetent, she will no longer be able to recognize and act on her reasons for preferring death to dementia. She therefore settles on a com- promise. She signs an advance directive stipulating that, if she contracts a potentially fatal condition once she has ceased to be competent, no life-supporting treatment is to be administered. (She would prefer to stipulate that, on passing a certain thresh- old of dementia, she should be painlessly killed. But given the present state of the law, such an advance directive could not be honored.) Later, when she has in fact be- come incompetent, she develops pneumonia, which is almost certain to prove fatal unless she is administered penicillin. At this point, however, she is quite contented, cannot remember the advance directive, and, when asked, says that she wants to live.

In this section, let us use our generic labels—“Patient at Onset” and “Demented Pa- tient”—to refer to this woman in the early and late stages, respectively, of Alz- heimer’s disease. The question is what her physicians should do when the Demented Patient develops pneumonia. The advance directive she signed when competent de- mands that she be allowed to die. Yet it may seem unconscionable to allow a patient to die when she is contented and expresses a desire to continue to live.

We should distinguish two questions. What, if anything, is required by respect for the person? And what would be best for, or in the best interests of, the Demented Pa- tient? If respect for a person involves both a deference to certain of her preferences and an appropriate concern for her good, the two questions may not be clearly sepa- rable. But before determining whether the questions can be answered independently, there is an antecedent question that must be considered. This is whether the De- mented Patient, who has ceased to be a person, is an appropriate object of respect at all. If a human being does not come within the scope of the morality of respect prior to becoming a person with an autonomous will, should we not conclude that an indi- vidual who irreversibly ceases to be a person thereby also ceases to be within the scope of the morality of respect?

This inference is unwarranted. There is an important asymmetry between indi- viduals who have not yet become persons and individuals who once were but have ceased to be persons. For the latter once had autonomous preferences that may still make a claim on us. Most people have preferences that extend beyond the boundaries of their lives—for example, preferences concerning the posthumous disposition of their property or the treatment of their dead bodies. If we regard it as a failure of re- spect for the person to disregard these preferences—and most of us do—then we should also accept that respect for a person can require that we honor the autonomous preferences she previously expressed concerning how she is to be treated after she ceases to be competent.

In the case of the Advance Directive, the Patient at Onset articulates an au- tonomous preference to be allowed to die once she becomes demented. It is true that, as the Demented Patient, she later has a different and conflicting preference. But the

Endings 497

preferences of the demented are notoriously arbitrary, whimsical, and ephemeral. And respect for an individual does not demand the honoring of preferences that are not autonomously formed. Thus, because the new, nonautonomous preference does not countermand or supersede the earlier autonomous preference, the Demented Pa- tient’s physicians have a reason grounded in respect for the preference of the person the Demented Patient once was to allow her now to die.

This reason will be decisive if there is no conflict with the Demented Patient’s good, or best interest. Thus an earlier advance directive will be decisively authorita- tive if what it demands is that life-saving treatment not be administered if the person becomes permanently unconscious in what I have called a deep coma. It will also be decisive if it demands the withholding of treatment if the person both becomes in- competent and has a life that is clearly intolerable from any perspective. But what if we conclude that it is in the best interest of the Demented Patient to continue to live? In that case, we believe that there is a conflict between this individual’s earlier au- tonomous preference and her present good. Sometimes when a person’s good con- flicts with her autonomous will, respect requires deference to the will—for example, when a Christian Scientist whose life would be worth living refuses life-saving med- ical treatment. But in other cases we accept that it is compatible with respect for a per- son to override her autonomous will—for example, when we restrain a teenager who, after serious reflection, attempts to commit suicide. In short, it is not obvious what re- spect for a person requires when it is in her interest to live, but her autonomous pref- erence is to die. And the uncertainty is exacerbated in the case of the Advance Direc- tive by the Demented Patient’s inability to reconsider her earlier preference.

Is it in the Demented Patient’s best interest to continue to live? Ronald Dworkin claims that it is not. He claims that her best interest coincides with her earlier au- tonomous preference. This is because what he calls her “critical interests” are in large measure determined by her autonomous preferences. Dworkin distinguishes our crit- ical interests from our “experiential interests.” We have experiential interests in those things that we find “pleasurable or exciting as experiences.”97 Our critical interests, by contrast, “represent critical judgments rather than just experiential preferences.”98

(Critical interests are not, however, entirely subjectively determined, for people “would be mistaken, and genuinely worse off, if they did not recognize” their own critical interests.)99 Critical interests are, moreover, closely tied to the value of a life as a whole. “Convictions about what helps to make a life good on the whole,” Dworkin writes, “are convictions about those more important interests.”100 Because the Patient at Onset’s life has been devoted to creative intellectual activities rather than passive pleasures, so that a period of dementia at the end would be jarringly in- harmonious with what has gone before, and because she autonomously desired to maintain the integrity of her life by avoiding a protracted period of dementia at the end, her critical interest—what would be better for her life as a whole—is to be al- lowed to die. And her critical interest, Dworkin assumes, has priority over the De- mented Patient’s experiential interest in continuing to live.

Is Dworkin right? Is the present best interest of the Demented Patient determined by what would be best for her life as a whole? Or is it possible that her present good diverges from what would be best for her life as a whole? Normally, of course, what is best for a person at a time coincides with what is best for her life as a whole. The

498 the ethics of killing

obvious exception is when what would be best for a person now, considering the mo- ment in isolation from the remainder of the life, would exclude some greater good in the future, so that to have the greater good—and thus the better life as a whole—the person must accept a lesser good now. But this is not an issue in the case of the De- mented Patient. There is, however, another possible form of divergence. In the case of the Advance Directive, the period of dementia is isolated from the rest of the life by the dramatically weakened prudential unity relations. Because of this, how good the period is for the Demented Patient—that is, how strong a time-relative interest she has in having it—is only weakly affected by its lack of coherence with the rest of the life. It may be good in its own terms even if it is worse for the life as a whole. (By contrast, when the prudential unity relations are strong, a person’s time-relative in- terests coincide with her interests, so that what would be best for her life as a whole will also be in her present time-relative interest.)

What is good for an individual at a time, or in his best time-relative interest, may vary with a number of factors, among which is the individual’s nature at the time. Thus what is good for an individual as a child is different from what is good for her when she becomes an adult. What is good for a child is not determined by reference to the nature she will later have as an adult; nor is it much affected by how it fits into the narrative structure of the life that is established primarily by choices made in adult life. What is good for the child is instead determined primarily by the child’s nature at the time. What is good for the Demented Patient might be similarly de- pendent on her nature at the time.

The comparison with childhood is, however, intended only to be suggestive. There are important asymmetries between a childish period at the beginning of a life and a childish period at the end. By the nature of the case, there are no prior au- tonomous preferences that might partially determine the nature of a child’s good. Nor do people typically regret that they once had a childish nature or believe that their childhood, so altogether alien in character from their adult life, mars the overall unity or integrity of their life as a whole. The natural progression from childhood to adult- hood is regarded as an acceptable narrative or normative structure for a life. Yet while a life may appropriately begin with a childish phase, it is a tragic decline if a mirror- image childish phase follows maturity. (It is difficult to say to what extent these eval- uations of the narrative patterns that a life may follow simply track our perception of what is normal or natural. One does not regret the childish bit at the beginning of one’s life although one may dread a comparable childish bit at the end. But would one take the same view if a period of dementia were universal after a certain age and were generally pleasant and contented the way childhood normally is? Some people, of course, do not regard a contented phase of dementia as degrading for them and, if contented passivity is consistent with the way they have lived hitherto, there is no reason to suppose that they are wrong.)

What is good for a person depends, to some extent, on what kind of person he is, what he cares about, and what his personal ideals or values are. Consequently, what is good for a person can change over time along with his character and values. If, for example, a person is committed to egalitarian ideals and as a consequence wishes to lead an ascetic life, the acquisition of expensive possessions might not be good for him at all. But if, after time, his ideals change and he becomes materialistic and self-

Endings 499

interested, it may then be in his interest to accumulate these same expensive posses- sions. As this example suggests, what is good for a person may change with his char- acter even when the change of character is a corruption, or a change for the worse. This suggests that it is necessary to distinguish between forms or levels of personal good. We must distinguish between, first, what is better for a person given, or condi- tional upon, his actual nature and values, and second, what nature and values it would be better for a person to have. But I will not pursue this complication here.

The relevant point here is that an individual’s good may change along with his nature, character, values, and preferences. And this is the basis for claiming that it would be better for the Demented Patient now to continue to live. The dementia has caused a profound alteration in her nature. She was once devoted to creative intellec- tual endeavors. For a person like that, the plunge into dementia could be deeply de- grading. Yet now that the dementia has already occurred, the elements of her nature that opposed or were hostile to a state of contented dementia have been eradicated. Now that she is demented, the good that seems appropriate to her present nature is contentment.

It remains true, however, that the Demented Patient now is one and the same in- dividual as the person whose personal values condemned a condition of bovine con- tentment as degrading and unworthy. (I say “personal values” to allow for the possi- bility that she recognized that a state of contented dementia would not necessarily be degrading for everyone.) And given that her life was shaped by those values, we should accept, as Dworkin insists, that her life as a whole would be better without a protracted period of dementia at the end. This is true even though she is and would continue to be contented in her demented state. For part of what she found repellent about dementia, we may suppose, was precisely the prospect of becoming capable of being content with personal vacancy.

It seems, in short, that there are conflicting criteria for determining what is best for the Demented Patient now. For an individual with her present nature, it is better to continue to live. But continued life with her present nature would make her life as a whole worse. Can these two claims be combined to yield a single answer to the question whether it would be better for her to be treated or to be allowed to die? Which has priority in determining her present good or best interest—what would be best given her present nature, or what would contribute most to the value of her life as a whole?

Recall that the destruction wrought by Alzheimer’s disease has radically weak- ened the various prudential unity relations between the Patient at Onset and the De- mented Patient. My suggestion is that, when the prudential unity relations are weak between an individual now and herself during most of the rest of her life, what is good for her now is determined less by the way in which her present life contributes to the value of her life as a whole and more by her nature and preferences at the time. Her present good is less a matter of how the present moment fits into the overall pat- tern of her life and more a function of the intrinsic features of the moment. This is be- cause the less prudential unity there is within the life, the less the life matters as a whole. For it is the prudential unity relations that make the life a significant whole rather than just a series of “discreet and mutually irrelevant episodes” (Huxley’s de- scription, quoted earlier, of the life of a lower animal). In the case of the Demented

500 the ethics of killing

Patient, it may indeed be appropriate to regard her prudentially unified life up to the onset of dementia as forming something of a whole, with the period of dementia being rather an excrescence dangling at the end—a period alien to and sufficiently distinct from the earlier unified life to have its own independent good.

An alternative way of making essentially the same point is to claim that the De- mented Patient’s present time-relative interests are more authoritative in determining her present and future good than the time-relative interests she previously had, when competent, in what would happen to her now. For the latter were time-relative inter- ests that she had during a period when she was only distantly connected via the pru- dential unity relations to the individual she is now.

For these reasons, I believe we should conclude that what is best for the De- mented Patient now is determined primarily by her present nature and preferences rather than by what would make her life as a whole better. In ascertaining her present good, we should give more weight to the intrinsic rather than the relational features of her present life. This means that her overall present good is more a function of what Dworkin calls her experiential interests and less a function of her critical inter- ests. In short, what is best for her as she is now is to be treated and to continue to live.

What is best for the Demented Patient now thus diverges from what would be best for her life as a whole. This raises new questions. Should her physicians do what would be best for her now, or ought they instead to do what would give her the best life as a whole? Assuming that her present good diverges from the previous au- tonomous preference that she has never actually renounced (and which itself gives priority to the value of her life as a whole), what does respect for her require? Al- though he does not accept that there is a conflict between the Demented Patient’s present good and the value of her life as a whole, Dworkin does accept that there is a conflict between her experiential and critical interests. And he believes that this must be resolved in favor of the critical interests. He writes: “when we consider how the fate of a demented person can affect the character of his life, we consider the patient’s whole life, not just its sad final stages, and we consider his future in terms of how it affects the character of the whole.”101 If we disregard the earlier autonomous prefer- ence for the sake of the present experiential interest, “that would not be compassion- ate toward the whole person, the person who tragically became demented.”102

Dworkin is presumably assuming that identity is what matters. If identity were the basis of egoistic concern, there could presumably be no conflict between what is best for the Demented Patient now and what is best for her life as a whole. What would be best for her, as a temporally extended individual, would necessarily be what would be best over time, or for her life as a whole. But I have argued, following Parfit, that identity is not the basis of egoistic concern. It is instead the prudential unity relations that ground egoistic concern within a life and give the life significance as a whole. When these relations are weak, as they are between the Patient at Onset and the Demented Patient, the life as a whole matters less. For when the basis of sig- nificant unity within a life is weak, the life, considered as a unit or whole, has less significance, and the different moments or periods that together constitute the life have greater significance, independently of their relation to the whole.

This last claim has most force when all the parts of a life are relatively weakly connected to one another via the prudential unity relations, as in the life of a lower

Endings 501

animal. The life of the Demented Patient is not, however, like that. It divides into two major parts, the predementia period and the period of dementia. While the prudential unity relations hold only weakly between these two parts of the life, so that they to- gether do not constitute a tightly unified whole, the earlier predementia period was deeply unified in the ways that matter. Because it is overwhelmingly the dominant part of the life, there is a genuine unity to most of the life. The life does, after all, con- stitute a significant whole, though the Demented Patient now is, as it were, dangling outside the unified part of that whole, and the longer she continues to exist, the more the integrity of the whole is compromised or degraded.

One way to conceive of the problem, then, is as a conflict between the good of two distinct parts of the life—or, alternatively, between the good of the earlier com- petent self and that of the later demented self. The later segment, considered inde- pendently of its relation to the whole, will be better if it is extended. And given that the Demented Patient’s physicians must make their choice now, there is a presump- tion that they should give priority to the present part of the life, or the present self. On the other hand, this same individual, when she was competent, had a time-relative in- terest in preventing her life as a whole from being marred by a period of dementia at the end. It might be argued, however, that the strength of this earlier time-relative in- terest must be heavily discounted for the weakened prudential unity relations be- tween herself then and herself now. If, for example, she had less reason, when com- petent, to care about her future pain, it should also be true that she had less reason to care about the degrading nature of her future dementia.

This claim misses an important point. The concern that the Patient at Onset had about her future dementia is different from her concern about her future pain. It was, as I have argued, rational for her to be less concerned in an egoistic way about her fu- ture pain. Her future pain would, she knew, affect only her future self, to whom she would be only rather distantly related in the ways that ground egoistic concern. But her future dementia mattered to her more directly. For whether and how long she would live in a demented state would in part determine what kind of life as a whole the earlier segment would be a part of. And that could affect the meaning and value of her life at the earlier time. For the meaning and value of a certain period within a life can be affected by the nature of the life as a whole of which it is a part. So whether the Demented Patient continues to live can retroactively affect the meaning and value of her life prior to the onset of dementia.

I believe that, in this situation, the Demented Patient’s physicians ought to give priority to the earlier part of her life, or to her earlier self. They ought, in other words, to allow her to die. So my practical conclusion coincides with Dworkin’s, though my reasons for accepting this conclusion are different. Unlike Dworkin, I do not believe that the Demented Patient’s present good is determined by what would be best for her life as a whole. It is, on the contrary, in her present time-relative interest to continue to live. But her continuing to live would, as Dworkin rightly emphasizes, be worse for her life as a whole, which in turn would be worse for the earlier life, making it a component or constituent of a lesser whole. Because the earlier part of the life is overwhelmingly the dominant part, its good should have priority. The earlier part was, in itself, a reasonably full and complete life with its own deep prudential unity. It was the life of the individual in her higher state, when she was a rational and au-

502 the ethics of killing

tonomous person. Its good—which is the good of that earlier, higher self—is there- fore more significant than the good of the shallow and necessarily rather brief period of dementia dangling at the end of the life. And its good—the good of the dominant part of the life—coincides with what would be best for the life as a whole. So the Demented Patient’s present good ought to be sacrificed for the greater good of her earlier self, which is also the greater good of her life as a whole.

It is true that to allow the Demented Patient to die would be bad for her in a way that is more direct and immediate than the way in which her death would be better for her former self. But it is important to appreciate how comparatively weak the De- mented Patient’s time-relative interest in continuing to live actually is. There is very little good—even of the lower sort that remains accessible to her—in prospect. And the prudential unity relations that would bind her to herself in the future are ex- tremely weak and growing weaker all the time. Her present and prospective future life are unhappily but aptly describable as a series of “discreet and mutually irrele- vant episodes.” Just as the prospect of this sort of life does not normally make a strong claim on us in the case of an animal, so it does not offer a strong reason for sustaining the Demented Patient’s life. It would, of course, be a different matter if, when she was competent, this woman had autonomously preferred that her life should continue in this way. But, in the Advance Directive, her present time-relative interest in sustaining her sadly diminished life is actually opposed by her former au- tonomous will.

There is a further consideration, contingently present in many cases, that rein- forces the conclusion that the Demented Patient ought to be allowed to die. It is often the case that, when a person stipulates in an advance directive that she should not be given life-supporting treatment in the event of dementia, her reasons are only partly self-interested. She may also be motivated by a desire to ensure that her life in a de- mented condition should not be prolonged at the expense of her family members. Her values may be such that, however she evaluates the prospect of dementia from an egoistic point of view, she is powerfully averse to becoming an emotional and finan- cial burden on those she cares about. There is no reason to suppose that this reason for allowing her to die is in any way weakened if it comes to be in her time-relative interest, when demented, to continue to live. There is, indeed, often reason to believe that many of an individual’s values persist well into dementia, even when the indi- vidual loses the capacity to be guided by those values in her behavior.103 In such a case, one may honor a demented individual’s present values by allowing her to die, even if she professes to want to live. For it may be that the values remain and what she has lost is the ability to understand that her continued existence is incompatible with them.

Endings 503