One of the most important questions in the debate over the morality of euthanasia and assisted suicide is whether an important distinction between killing patients and allowing them to die exists. The U.S. Supreme Court, in rejecting challenges to the constitutionality of laws prohibiting physician-assisted suicide (PAS), explicitly invoked this distinction, but did not explicate or defend it.' The Second Circuit of the U.S. Court of Appeals had previously asserted, also without argument, that no meaningful distinction exists between killing and allowing to die.2 That court had reasoned that if this were so, it would be discriminatory to allow persons on life support to end their lives by removing such treatment, while those who are not connected to life support would be denied similar access to death. The amicus curiae brief filed by several philosophers in support of plaintiffs Timothy Quill et al. before the U.S. Supreme Court (the so-called "Philosopher's Brief") also argues vehemently that there is no distinction between killing and allowing to die. The brief contends that belief in such a distinction is simply a "misunderstanding of the pertinent moral principles," and that "Whether a doctor turns off a respirator in accordance with the patient's request or prescribes pills that a patient may take when he is ready to kill himself, the doctor acts with the same intention: to help the patient die."3 These arguments echo those of a series of other philosophers and clinicians who have suggested that this distinction ought to be abandoned.4 They all seem to share a common view that the distinction is "confused and mistaken."5

Likewise, prominent commentaries on the Supreme Court decisions, published in the medical literature, have also suggested forcefully that the distinction between killing and allowing to die, as well as related moral rules that also invoke intentions (such as the rule of double effect), are confused or pointless.6 These commentaries make the point that if a physician can be allowed, under the rule of double effect, to give morphine to a dying patient, even with the belief that this might hasten the patient's death and with the knowledge that the patient might desire death, this cannot reasonably be distinguished from killing the patient.

The distinction between killing and allowing to die has enjoyed a long tradition in medicine and in common morality. Despite deep division in the profession over the issue of legalizing euthanasia and PAS in the present era,7 the Hippocratic tradition has at least implicitly accepted a moral distinction between killing patients8 and allowing them to die9 for many centuries. Many clinicians still intuitively sense a moral difference between turning off a ventilator and giving a lethal injection. If their intuitions are confused or mistaken, however, physicians and other health care professionals will have one less reason for refraining from the practices of euthanasia and assisted suicide.

This question is of more than academic interest. In the Netherlands, where physician assisted death is not prosecuted if physicians follow certain guidelines,' 2 percent of all deaths are due to voluntary euthanasia or assisted suicide, an additional 1 percent are due to nonvoluntary euthanasia, and another 5 percent result from injections that were, at least in part, given with the explicit intention of hastening death." These numbers are certainly not decreasing, and, arguably, may be increasing.12 In the United States, Jack Kevorkian has assisted multiple patients with suicide, has never been convicted, and the new district attorney has pledged to cease prosecution.l3 Euthanasia initiatives were recently defeated by narrow margins in California and Washington State.l4 In 1994, voters in Oregon passed the world's first assisted suicide measure, and affirmed that decision in a second referendum in 1997.15 The constitutional basis of this law has been upheld by the Ninth Circuit of the U.S. Court of Appeals,'6 and it is now in force.

Why another paper on killing and allowing to die? A full philosophical discussion of the distinction between killing and allowing to die is well beyond my scope here." But as detailed above, it has been repeatedly asserted that the distinction is either hopelessly confused or cannot be interpreted in such a way that the physician could not be understood as intending anything different in cases of killing as opposed to cases of allowing to die. I will argue that this is not the case. I will grant critics that the difference has not been clearly and rigorously articulated in the past, but this does not imply that it cannot be clearly and rigorously articulated. Tom Beauchamp has challenged defenders of the traditional view by stating, "if we are to retain the distinction between killing and allowing to die we need to provide clearer, more precise meanings that are useful for medical ethics."" Here, I take up Beauchamp's challenge. I present a novel interpretation of the logic and linguistics of the traditional distinction, in a way that makes the distinction clear. In particular, I will show how the distinction relates to the important concepts of intention, belief, desire, and causation, and how many previous interpretations of the distinction have muddled the intended meaning of those who say there is an important moral difference between killing and allowing to die.

This essay is therefore not a full defense of the moral proscription against PAS and euthanasia. It is only an argument to suggest that because the distinction between killing and allowing to die can be stated clearly and precisely, the real issues at stake in the debate over PAS and euthanasia are far deeper than the coherence of this distinction. I contend that the distinction is defensible if one holds certain views about the nature of intentions and about the role of intention and causation in the moral evaluation of human acts. If one holds these views about intention, causation, and morality, then the distinction between killing and allowing to die can be rendered meaningful and coherent. The real argument about PAS and euthanasia is thus not about the coherence of the distinction between killing and allowing to die, but about the view of morality that either supports or undermines the meaningfulness of the distinction.

I will refer to the belief that this distinction is coherent and morally meaningful as the traditional view, and will examine several aspects. First, I will present a case to guide the discussion. Second, I will examine some of the distinctions that are all too easily confused with the distinction between killing and allowing to die. Third, I will present precise definitions of the terms and suggest the proper form of the distinction. Fourth, I will discuss the relevance of intention to the distinction. Finally, I will delineate (but not defend) some of the reasons that have been given for holding this traditional view, and what some of its implications may be for public policy.

In essence, I will argue that there is nothing confused about the traditional view that a morally important distinction between killing and allowing to die exists. It is meaningful and coherent, provided one believes that it is immoral for a clinician to act with the specific intention in acting that a patient should die by way of the clinician's intentional act, and one understands intentions to be something different from beliefs and desires.

A case

Mrs. Brown is a thirty-eight-year-old mother of two and is dying of metastatic breast cancer after having failed highdose chemotherapy with autologous bone marrow rescue. She has significant pain from bony metastases. This pain has been imperfectly relieved with epidural analgesics. Her oncologist never spoke to her about advance directives. She arrives in the emergency room short of breath and is intubated. It is determined that the cancer has spread into the lymphatic system of her lungs. She is conscious and alert. After a discussion with her husband, she writes out a very clear note to the attending intensive care unit, stating that she is ready to die and wants assistance with her death. What is the difference between offering this patient a lethal injection (killing) and offering to remove ventilator support (allowing to die)?

What the distinction is not

Active vs. passive

Some have thought that the difference between killing and allowing to die lies in the fact that killing is active and allowing to die is passive. Unfortunately, this omission/ commission account does not take one very far. Most clinicians would readily agree that disconnecting Mrs. Brown from a ventilator is a classic case of allowing to die. But, of course, it is an active act. One actively turns off a switch or actively disconnects a piece of tubing or extubates a patient. So the difference between killing and allowing to die cannot just be the difference between commissions and omissions.

Causing vs. allowing

At second pass, one might suggest that in killing one causes the patient's death, but that in allowing to die it is nature that causes the death. Yet it is hard to see how one could say that in disconnecting a life-support system, one would not be causally involved in the death. If Mrs. Brown needs ventilator support to maintain pulmonary function, and the clinician turns off the ventilator, could the clinician seriously claim to have played no causal role in the death? Certainly that argument cannot carry the weight one might want to give it.

A possible counter-example

Further, suppose one were to say that one could adequately describe the difference between killing and allowing to die. Even so, it is argued, the distinction would not make a moral difference. James Rachels offers the following pair of stories.

Smith stands to gain a large inheritance if anything should happen to his six year old cousin. One evening while the child is taking his bath, Smith sneaks into the bathroom and drowns the child, and then arranges things to make it look like an accident. No one is the wiser, and Smith gets his inheritance. Jones also stands to gain if anything should happen to his six year old cousin. Like Smith, Jones sneaks in planning to drown the child in his bath. However, just as he enters the bathroom Jones sees the child slip, hit his head, and fall face down in the water. Jones is delighted; he stands by, ready to push the child's head back down under if necessary, but it is not necessary. With only a little thrashing about, the child drowns all by himself, 'accidentally', as Jones watches and does nothing. No one is the wiser, and Jones gets his inheritance.19

So, it seems as if Smith killed his cousin, and Jones allowed his cousin to die. But, Rachels asks, does this appear to make a moral difference?

I will argue that there are problems with Rachels's question. First, the debate has been clouded by an inconsistent use of the word killing. Second, anyone who thinks that the story of Smith and Jones disproves the distinction has misunderstood the form of the traditional view.

Confusion over language

Some authors have used the word killing as a morally neutral term referring to any act (whether an act of commission or omission) that is causally related to a death.20 This would mean that for the clinician to disconnect Mrs. Brown from her ventilator or to refrain from resuscitating her or to give her a lethal injection, would each be killing, regardless of the morality of these acts. For these authors, then, there are good killings and there are bad killings.

Others use killing to refer only to morally bad acts that result in death.21 This would mean that disconnecting Mrs. Brown from her ventilator would not be killing, while disconnecting a patient who was expected to recover completely would be killing. For these authors, the moral evaluation is made first, then the label "killing" is applied. For them, there are no good killings; there are only bad killings.

To add further complexity, some seem to prefer to use "killing" to refer only to acts that cause death and are not voluntarily authorized22 or involve malice.2 On such accounts, there is a "letting die that is also killing."24 So, for these commentators, either to give Mrs. Brown a lethal injection or to take her off the ventilator without her request would be killing, but to give that same lethal injection or to take her off the ventilator at her specific request would not be killing. All this can be very confusing.

I suggest, therefore, that we can clear up some of the confusion engendered by the claim that "there is a moral difference between killing and allowing to die" if we precisely define what sorts of acts are being distinguished by those who hold the traditional view. For present purposes, the reader should not attempt to argue that he/she prefers some other definition of killing and allowing to die. That is not my point. The point is that under the definitions I will offer, the distinction can be upheld. One could even use neologisms to attach to each of these definitions if one were to insist that the distinction I am describing is not how one prefers to use the terms. For my purposes, therefore, I ask readers' forbearance and acceptance of the following definitions as stipulated.

With this in mind, I offer the following precise definitions for the two types of intentional acts that are referred to by the names used in the phrase "there is a moral difference between killing and allowing to die":

(1) When physicians and others talk of "killing," as it is used in this phrase, I take it that they mean: an act in which an agent creates a new, lethal pathophysiological state with the specific intention in acting of thereby causing a person's death. I will use "Killing*" to refer to this class of acts, distinguishing it from any other way one may use this word.

One should note that this definition excludes (1) accidental deaths and (2) failed therapeutic manipulations in which a lethal state is created with a therapeutic intent (such as in bone marrow transplantation).

(2) When physicians and others talk of "allowing to die," as it is used in the phrase, "there is a distinction between killing and allowing to die," I take it that they mean: an act in which an agent either performs an action to remove an intervention that forestalls or ameliorates a preexisting fatal condition or refrains from action that would forestall or ameliorate a preexisting fatal condition, either with the specific intention of acting that this person should die by way of that act or not so intending. I will use "Allowing to Die*" to refer to this class of acts.

Logic and the distinction

With these definitions, the form of the distinction also needs to be stated precisely. Even with these precise definitions, the story of Smith and Jones could still refute the moral significance of the distinction. However, this is so only if one has misconstrued the form of the distinction. Rachels, for instance, assumes that if the distinction is to be morally meaningful, it must take the form All Killing* is morally wrong, and all Allowing to Die* is morally right. If so, Rachels is correct. Smith Kills* his cousin, and so Smith is clearly in the wrong. Jones's cousin is Allowed to Die*, but Jones is also clearly in the wrong. It would therefore seem that the Killing*/Allowing to Die* distinction could not possibly have moral meaning.

But I argue that the distinction between Killing* and Allowing to Die* actually takes a different form. The defensible form of the traditional view is:

Except in cases of self-defense or rescue, all Killing* is morally wrong. Some Allowing to Die* is also morally wrong, and some is not.

Given space considerations, I will not consider in detail the possible exceptions such as self-defense or rescue. But the exceptions are generally considered permissible because they involve situations of forced choice in which the life of one person or another must be forfeited, and under such choices one may choose to save the innocent person's life and end the aggressor's life. Clearly, however, these exceptional conditions are not relevant to cases of euthanasia and assisted suicide.zs

Once one understands the distinction between Killing* and Allowing to Die* as I have formulated it, the stories of Smith and Jones can be seen as illustrations of the distinction, not as counter-examples. According to the proper form of the traditional view, all acts of Killing* are morally wrong, and so Smith, who pushes his cousin's head under the water, is clearly wrong. According to the proper form of the traditional view, some acts of Allowing to Die* are also morally wrong, and some are not. Jones, who stands by idly as his cousin drowns, simply illustrates that some acts of Allowing to Die* are morally wrong. A pathognomonic sign

Clinicians should be quite comfortable with this form of a distinction. The form of the distinction between Killing* and Allowing to Die* is the same as that of the distinction between the presence and the absence of a pathognomonic sign. Pathognomonic signs are clinical clues, such as distinctive rashes or ocular findings, that definitively indicate the presence of a particular disease. All patients who have a pathognomonic sign have the disease it signifies. Some patients who do not have the pathognomonic sign nonetheless do have the disease, and some do not. For example, all patients who have corkscrew hairs on a background of perifollicular hemorrhages have scurvy.26 Some patients who do not have corkscrew hairs on a background of perifollicular hemorrhages do have scurvy. But, fortunately, most who lack this pathognomonic sign do not have scurvy. In the same way, proponents of the traditional view about killing and allowing to die would argue that Killing* is a pathognomonic sign of a moral problem. If a particular act belongs to the class of acts that are Killing*, proponents of the distinction claim that the act can be judged wrong and the question is settled. If a particular act belongs to the class of acts that are Allowing to Die*, proponents argue only that the case is not settled. One still has moral work to do.

So, if a clinician were to Kill* Mrs. Brown by a lethal injection of potassium, that clinician would create a new fatal pathology-hyperkalemia. According to the traditional view, if the clinician does so intending that Mrs. Brown should die as a result, that clinician's act is wrong. One would be hard pressed to find a reason other than to cause death for a clinician intentionally to cause hyperkalemia. If a clinician were to Allow* a patient like Mrs. Brown to die by removing ventilator treatment for a fatal disease called respiratory failure, proponents say only that this might be wrong or it might not be wrong. For example, if the clinician were to intend that Mrs. Brown should die so that the clinician's managed care withhold fund would not be used up, this would be wrong.27 Her death would be part of the clinician's intentional plan. On the other hand, according to the traditional view, if the clinician's intention were that Mrs. Brown should no longer be on a ventilator that she did not want and was merely prolonging her dying, this would not be wrong. The specific intention in acting would be simply that she no longer be on the ventilator.

Authorization

Although deeply concerned with clinicians' intentions and actions, proponents of the traditional view do not discount autonomy and the role of patients in making end-of-life decisions. Proponents hold that, in general, clinicians ought not withhold or withdraw life-sustaining treatments without the authorization of the patient or a morally valid surrogate. However, proponents of the distinction also argue that the patient's autonomous authorization is not the only condition that needs to be met for a clinician to make morally justifiable end-of-life decisions. For example, autonomous demands for biomedically futile treatments need not be honored. But more important, proponents of the distinction hold the deontological view that some acts ought not be done, no matter what the consequences or the preferences of another person. Practically, this requires a belief that the intentions of agents figure importantly in the moral evaluation of human acts.

The belief that the agent's intentions are important in the moral evaluation of human acts is a deeply held moral view of many persons. The law, for instance, recognizes such distinctions as those between manslaughter and various degrees of murder. These distinctions depend completely on the belief that intentions matter in the moral evaluation of human acts. Intending, desiring, and foreseeing death The traditional view about the distinction between Killing* and Allowing to Die* can be illuminated by applying an understanding of intentions that contemporary philosophy of mind is now coming to adopt through a rigorous analysis of concepts not directly related to morality. A salient feature of this view is that desire and belief ought not be confused with intention. To desire that a patient should die so that his/her suffering might end, and to believe that death will likely occur as a result of discontinuing treatment, is not the same as acting with a specific intention in acting that the patient should die by way of one's act.

Many who object to moral arguments based on intentions28 appear to consider intention a matter of desire or belief. But a developing body of recent philosophical work shows that intention is not reducible to any combination of desire and belief.29 I can, for example, intend to do what I do not desire (for example, going to see my patients when I am tired and would rather not). I can also desire to do what I do not intend (for example, I may desire to eat a high cholesterol diet but never form an intention to act that way lest my patients think me a hypocrite).

Or, consider the paralysis that would follow on encountering equally desirable options if intentions were reducible to belief and desire. For example, if one desired to treat a patient's arthritis and believed that two different brands of ibuprofen were equally efficacious and equivalent in side-effects, and found each brand equally desirable, would one thereby have formed an intention to prescribe one or the other brand? Or both of them? Intention seems to involve something over and above belief and desire. It involves commitment.30

As propositions have truth conditions, so intentions have conditions of fulfillment-that is, what the agent is committed to accomplishing; aiming at. The traditional view holds that aiming at death is morally different from aiming at the abatement of treatment. One may believe (foresee) or desire events that differ from those one is aiming at in one's actions.

This means that clinicians who hold that the traditional view about Killing* and Allowing to Die* can certainly foresee and desire that some of their patients should die quickly after the withdrawal of life support. Mrs. Brown's physician can hope for her quick death; expect it; even pray for it. But this does not mean that her physician has committed herself to bringing about Mrs. Brown's death as the condition that fulfills her intention. Desire and belief are not intention.

It is also important to note that although medicine has traditionally prohibited intending the death of patients, this by no means should be taken as indicating that patients are prisoners of technology. The whole tradition of withholding and withdrawing disproportionate means of treatment first of all presupposes this distinction between Killing* and Allowing to Die*, then helps health care professionals and patients to distinguish when one may legitimately Allow to Die* by withholding or withdrawing treatment. Mrs. Brown is not required to suffer needlessly. If she thinks that the burdens of treatment outweigh the expected benefits, her physician may withdraw treatment, provided that her intention in so doing is that no such treatment be rendered. But defenders of the traditional view would hold that her physician ought never to act with the specific intention that there should be no Mrs. Brown. Her physician can morally aim at not treating her, but cannot morally aim at her not being there to treat. Intention and the distinction

The next question to ask is what explains why all Killing*, as well as some Allowing to Die*, should be wrong. Along with Bonnie Steinbock,3' I argue that those who hold the traditional view hold that the morality of these acts depends importantly on the clinician's intentions. On the traditional view, if the death of the patient is the condition that fulfills the clinician's intention, whether the act is Killing* or Allowing to Die*, the act is wrong. In contrast to Steinbock, however, I argue that the difference between Killing* and Allowing to Die* is in the act descriptions, not in the intentions. The act descriptions pick out two classes of acts. In one class, the intentions are uniform. In the other, the intentions vary.

The distinction between Killing* and Allowing to Die* is therefore morally meaningful for anyone who ascribes moral weight to the intentions of health care professionals. According to the traditional view, except in cases of selfdefense or rescue, all Killing* of patients is morally wrong. In all Killing*, the clinician's intention is uniformly the death of the patient. Thus, according to the traditional view, just as Smith acts wrongly by drowning his cousin, so does the clinician who injects a lethal dose of potassium into Mrs. Brown. Both are Killings*. Both uniformly involve intending death. Nonetheless, some acts of Allowing to Die* are also morally wrong-those Allowings in which the clinician's intention is the death of the patient. So, because Jones intends that his cousin should die in order that he should collect the inheritance, Jones is morally in the wrong. In the same way, on this view, if the clinician disconnects Mrs. Brown from the ventilator with the explicit intention that Mrs. Brown should die, intending that she be annihilated so that his research program, named in her will, can collect the inheritance, that clinician has also acted wrongly. On the other hand, if the clinician disconnects the ventilator and the condition that fulfills her intention is that Mrs. Brown should not be on the ventilator (for example, because it is judged useless in preventing her inevitable death), that clinician has not acted wrongly.

This difference is subtle, but morally crucial. In all medical Killing*, the clinician intends the death of the patient, and, on the traditional view, this is wrong. In some acts of Allowing to Die*, the clinician intends the death of the patient and this is also morally wrong. In other acts of Allowing to Die*, the clinician does not intend the death of the patient, and, under certain conditions, this is morally permissible.

According to the definitions and the logic of the traditional view, classifying an act as Killing* does not explain why the act is morally wrong. It only tells one that the act is morally wrong. The reason defenders of the traditional view hold that Killing* is morally wrong depends on a complex evaluation of the act, the outcome, the circumstances, and the intention. The analogy to a pathognomonic sign is once again useful. People do not have scurvy because they have corkscrew hairs on a background of perifollicular hemorrhages. They have scurvy because they lack vitamin C. The finding of the pathognomonic sign only signifies that the disease is present. It is not an explanation of the pathophysiology of the disease. Recognizing intentions

Intentions, then, seem important in the moral evaluation of the clinician's acts. But intentions are hard to know. Sometimes people do not even understand their own intentions. And it is precisely because intentions are so difficult to know that anyone who believes that intentions are morally important will find the distinction between Killing* and Allowing to Die* to be significant.

Killing* is usually easy to recognize. Injecting 150 mEq of potassium chloride (KCl) directly into Mrs. Brown's right ventricle via a central venous line is a straightforward action. It would be hard to argue that one did not intend her death if one injected 150 mEq of KCI by intravenous push. The burden of proof is overwhelmingly on anyone who creates a new, lethal pathophysiological state to explain how death was not intended-how this was part of a therapeutic maneuver that failed, or that she was deceived and did not know what she was doing, or was drugged so that the action would not really be intentional. If a clinician is rational, she does not intentionally create a new, lethal pathophysiological state that is inconsistent with a therapeutic purpose without intending that the patient should die.

For those who hold the traditional view, it is therefore very useful to distinguish Killing* from Allowing to Die*, because the moral work is easier for Killing*. It is not so with Allowing to Die*. If I unplug a ventilator, how does anyone know what my intentions were? I might intend the patient's death, or I might not. It might even be hard for me to know.

Nonetheless, one can at least apply some simple screening tests to help sort out these intentions. While imperfect, these questions are useful probes of one's intentional state. One can ask, How would one feel and what would one do if the patient were not to die after one's action? Would one feel that one had failed if Mrs. Brown did not die after disconnecting the ventilator? Would one try to figure out how to finish her off? If the honest answer is yes, then the patient's death was probably intended. If the honest answer is no, then, according to the proponents of the distinction, one's intentions are probably morally sound. It is useful in this regard to reflect on the paradigm case of Allowing to Die*-turning off the ventilator for Karen Ann Quinlan. As is well known, she did not die, even though her death was expected. But those involved in the decision did not fail. They were successful in carrying out their intention that she no longer be connected to the ventilator. No one tried to smother her when she started breathing. The condition that fulfilled their intention was simply that she not be on a ventilator. As it turned out, the ventilator was not preventing her death. She lived another decade.

Public policy and the traditional view If, on the traditional view, the underlying moral questions crucially involve intentions, then the distinction makes enormous sense in terms of public policy. All acts in which the intentions are clearly wrong (Killing*) are automatically proscribed. And this is relatively easy to enforce, because acts of Killing* are easy to recognize. In cases of Allowing to Die*, in which intentions are hard to know, the law takes a more circumspect approach. Many, if not most, cases will be morally appropriate. One would not want to police all cases of Allowing to Die* so aggressively that good people with good intentions would be required to suffer needlessly even though life-sustaining treatments had been appropriately refused. But the law is also open to the possibility of prosecuting particular cases of Allowing to Die* in which suspicions might be aroused. There could be cases in which the Allowing to Die* is blatantly part of an intentional plan of nefarious motive. The law leaves open the possibility of prosecuting a clinician who, for instance, allows a patient to die with the intention that the patient should die (so that he could, for example, collect a million dollars in insurance or marry the patient's spouse).

Of further importance for public policy, it makes more sense to distinguish between honoring refusals and honoring requests.32 In general, many persons hold the view that there is a much greater obligation to honor refusals than to honor requests. In rights language, many hold that negative rights (rights of noninterference) are fundamental in a way that positive rights (rights of entitlement) are not. For instance, my duty to respect your freedom of speech does not entail an obligation on my part to publish your first novel. On anyone's view, for either an act of Killing* or of Allowing to Die* to be morally justified, the patient's authorization would be necessary (although not necessarily sufficient). All patient authorizations of Allowing to Die* involve a refusal of an intervention. But all patient authorizations for Killing* involve a request for an intervention. One therefore needs fewer justifying reasons to deny a patient's request to be Killed* than one would need to deny a patient's refusal of treatment so that the patient could be Allowed to Die*. This seems to help justify making policy distinctions between the two.

In addition, this legal deference regarding cases of Allowing to Die* dovetails nicely with a common law right to be free from interference.33 Even those who do not accept the importance of intentions in the moral evaluation of acts could accept the distinction between Killing* and Allowing to Die* on the basis of legal differences between tenuous rights of entitlement and traditional rights of noninterference. Most of the noninterference involved in Allowing to Die* is appropriate, but, in some cases, this noninterference constitutes grounds for a tort or criminal prosecution.

PAS and intentions

In a putatively rational case of PAS, the patient's intention is clearly to bring about his/her own death. According to the traditional view, the patient's action is morally wrong. Several authors have argued that the clinician's intentions in PAS, however, might be precisely the same as those of clinicians in the morally appropriate cases of Allowing to Die*. Could not the clinician intend only to write the prescription and thereby make the patient feel secure, even though the patient might actually take the pills and die?34 Such a counter-argument would be effective only if intentions were all that mattered in the moral evaluation of acts. This is not the traditional view.35 The traditional view is that morally wrong intentions make an act morally wrong, but there are plenty of other aspects of an act that are part of its moral evaluation.

So, for instance, if one tries to justify this purported division of intentions on the basis of the Rule of Double Effect (RDE), the argument fails. One of the conditions of RDE is that the good one hopes to accomplish cannot be caused by the bad effect one purports to foresee but not to intend. In this case, the bad effect is the possibility of the patient's suicide. But it is precisely the possibility of the patient's suicide (as a "way out") that causes the intended good effect (the patient's sense of relief and security). The conditions of RDE are not met. But more important, RDE is not properly applied to cases involving multiple agents. In such cases, the traditional view would suggest that the principles involving the ethics of cooperation would be more proper. And here again, the argument fails. To provide freely the immediate material means by which a patient carries out an act one considers morally inappropriate is never morally permissible, even if one claims not to share in the intention. So, if I buy a gun for a known bank robber, and he uses it to Kill* someone during a robbery, I share some culpability even if it really is the case that I did not share his intention to Kill*. Omission, causation, authorization, and the distinction

One should notice how the other distinctions I have mentioned sort themselves out once one understands the definitions and the proper form of the traditional view (see Table 1). The understanding of the distinction that I am proposing helps to explain some of the mistaken accounts of the traditional view discussed above. For instance, Killing* Mrs. Brown is always a commission. Allowing her to Die* is sometimes a commission and sometimes an omission. The clinician is causally involved in both Killing* and Allowing to Die*. It is the manner of causation that differs.

And the manner of authorizing differs as well. For Mrs. Brown to authorize her Killing* always involves a request for an intervention. However, all that is required for Mrs. Brown to authorize her being Allowed to Die* is her refusal of an ongoing or proposed treatment. Why should intending death be wrong?

It should be clear that there is nothing confused or illogical about the traditional view that a moral difference exists between Killing* and Allowing to Die* (as I have defined these terms), provided one believes that intentions are morally important. The real question that must be debated, then, is why one should hold it immoral for a clinician to make a patient's death the condition that fulfills his/her intention. Space considerations will not allow a full discussion here, but I outline a few reasons why many people would hold that it is wrong for a clinician to act with the intention of ending a patient's suffering by ending the patient's life. Relationships

Some argue that the value of human life is relational; that humans are inherently communal being. Even a person washed ashore on a desert island has had relationships and knows that further relationships are possible. Those who hold belief systems that place great moral value on human relationships proscribe euthanasia and assisted suicide as violent acts rejecting the value of relationship-violent severings of human beings' connections with fellow human beings.36 To make the destruction of this value the condition of fulfillment of one's intention in acting would always seem, on this view, to be morally wrong. This seems particularly true of the clinician-patient relationship. Perhaps the most powerful healing act available to Mrs. Brown's physician is to remind her that she is still an important part of the human family even as she slips away. Dignity

Some argue that being in control of one's destiny is what gives human life its fundamental meaning and value-that is, its dignity. But one can argue that a human being is no less valuable when afflicted by a disease that takes away freedom and control.37 One may also point out that some of the most morally important human relationships are those in which one party is completely dependent while another is in complete control, like the relationship between mother and child. So, it can be argued that a vitally important aspect of human dignity does not depend on being in control, but belongs to all human beings simply because they are human.

On this view, those who have lost substantial degrees of control and have become dependent because of disease do not lose their dignity in this fundamental, intrinsic sense. To make the destruction of a human being, no matter how dependent, the condition of fulfillment of one's intention in acting would always seem, on this view, to be morally wrong, because one ought not destroy the bearer of such dignity. To do violence of any sort to persons is to attack their dignity in this fundamental sense.

Life is not of in finite value To say that all human lives have intrinsic meaning and value is not to say that this meaning and value is infinite. The point of the traditional view is that human life has a high value, a dignity, and that the core of this value is the same for everyone. A health care professional ought not make the destruction of this value the condition of fulfillment of his/her intention in acting. This is why Allowing to Die* can be appropriate. In Allowing to Die*, one still upholds the value of a human being's life by not acting with the intention to end it. This is part of what respect for persons entails. But one also respects the limits of what it means to be human and the limits of what medicine can offer by no longer resisting death. If one believes that it is humanity itself that gives people dignity, then the distinction between Killing* and Allowing to Die* makes sense and is morally meaningful. On this view, disconnecting Mrs. Brown from her ventilator is right. Giving her a lethal injection is wrong. Subjective standards

Some argue that it is completely up to each patient to define life's value in his/her own way. But this raises serious questions. Can I not know that another patient's life has value even if we have not discussed it? If a patient cannot communicate with me, do I really have no way of knowing that this patient's life has value? As Kant puts it, "humanity itself is a dignity."38 The value of any single individual's life must share something in common with the value of any other individual's life. Otherwise, on this subjective view, there can be no such thing as normative ethics. The slippery slope

Even if one does not believe that human life has any intrinsic value, worries about the consequences of deciding that there is no moral difference between killing and allowing to die lead many to consider it important to retain this distinction. Concerns about abuses of euthanasia and assisted suicide and the extreme difficulty in policing these practices can be cited in favor of maintaining the distinction, independent of the beliefs I have cited regarding intention, morality, and human dignity.39 Conclusion

If what one means by "there's a distinction between killing and allowing to die" is a distinction between what I have called Killing* and Allowing to Die*, then the distinction is clear, coherent, and makes moral sense to anyone who holds that it is wrong for a clinician to act with the intention that a patient should die by way of his/her act. This is, I think, the only proper way to formulate and to defend the traditional view. The traditional view is a commonsense clinical view. And common sense is neither simple nor naive. What common sense grasps can be very difficult to explicate.qo

I have defined Killing* and Allowing to Die* precisely. I have described the proper form of the traditional view: except for self-defense and rescue, all Killing* is morally wrong. Some Allowing to Die* is also morally wrong, and some is not. I have argued that the traditional view holds that it is always morally wrong for a clinician to set the death of a patient as the condition that fulfills his/her intention. I offered some explanations of the importance of the distinction. Because intentions are hard to know, it is convenient to be able to classify an act as Killing*, because the burden of proof is overwhelmingly placed on a Killer to explain how the death was not intended. Further, to authorize a Killing* always involves a request, which is easier to deny than a refusal. The patient's authorizing force in Allowing to Die* is always a refusal. Finally, I offered a sketch of an explanation about why one could hold that intending the death of a human being should always be morally wrong, except in cases of self-defense or rescue. It would be useful if the ethical debate about PAS and euthanasia attended to whether it is wrong for a clinician to act with the explicit intention in acting that a patient should die by way of the clinician's act. Many reasonable people believe this is wrong. They are defenders of the traditional view about killing and allowing to die. Other reasonable people disagree. But because the traditional view itself is neither confused nor incoherent, further ethical discussion about euthanasia and assisted suicide would profitably engage the underlying moral beliefs that render the distinction between killing and allowing to die morally meaningful for those who hold these beliefs. Simply dismissing the distinction as "confused" misses the point.