Symmetry: Refusal and autonomy

Let us begin by considering the judgment in Refusal. Although we do not believe that being informed is a necessary condition for valid refusal, to make the cases more alike, we are stipulating that the patient’s refusal is sufficiently well-informed about what her refusal entails; she is not subject to any coercion; she does not have any of the conditions that we might think would negatively affect her decision-making capacity (e.g., profound mental illness, severe sleep deprivation, shock, impairment from medication); and her decision flows from her deeply held values. In other words, the decision is autonomous, according to whatever account of autonomy we care to stipulate.Footnote ⁸

Another important element of the case is that accepting the blood transfusion is necessary for the patient’s life to continue. In other words, the transfusion is necessary to promote her well-being, at least medically speaking. This point invokes concerns about beneficence or nonmaleficence—that is, the idea that a core goal of medical treatment should be to improve the patient’s well-being. Surely this feature of the case is what makes it at all controversial or ethically interesting in the first place: If it were not a blood transfusion, but rather a routine and not especially important blood pressure check, then most would not find the patient’s refusal to evoke significant disagreement or concern. The fact that her refusal will result in her death, however, elicits concern about allowing an otherwise healthy patient to refuse lifesaving care when the care involves little risk.Footnote ⁹

So, Refusal involves a tension between two values, autonomy and well-being. Most will agree that, provided that the patient’s refusal is truly autonomous in the way previously detailed, the physician has no basis for resisting her refusal and forcing her to accept a blood transfusion. In other words, between these two values, autonomy trumps well-being. Why should we endorse this judgment of priority?

The basic idea here is that it would be overly paternalistic to override or reject a patient’s autonomous choice in favor of promoting her well-being. We tend to think that we ought to defer to patients’ autonomous choices in the absence of serious countervailing considerations. There may be many such considerations, such as risk or harm to others and public costs; but the patient’s own well-being is not a consideration that is capable of overriding her exercise of autonomy. If the patient lacked decision-making capacity and therefore could not act autonomously, then well-being considerations should play a bigger role in governing the physician’s decisions. In most cases, if a substitute decision-maker is not available, a physician should proceed with a blood transfusion for a patient who is unconscious, incapable of understanding the information relevant to her condition and treatment options, or too young to give meaningful consent or refusal.Footnote ¹⁰ Moreover, physicians must attempt to provide the patient with adequate information about how refusing a treatment will impact their well-being. These points illustrate the fact that physicians are right to aim at promoting well-being; none of the foregoing points count against the relevance of well-being or beneficence. The point, however, is that when autonomy is satisfied, this suffices to override these other conditions.Footnote ¹¹

Despite its broad appeal, the view that autonomy trumps well-being does not enjoy universal acceptance. For one thing, some might worry that this judgment does not apply in all cases. To give an example from a different context, some believe that physicians (or others with the relevant authority) should force-feed prisoners who participate in hunger strikes.Footnote ¹² If this is true, then it casts doubt on the idea that autonomy will always trump well-being. It is worth noting, however, that the majority of bioethicists and philosophers writing on this issue seem to share a broad agreement that force-feeding is morally wrong, since it violates the autonomous choices of the prisoners.Footnote ¹³

Furthermore, most philosophers who deny that physicians ought to respect the patient’s refusal in cases like this believe such a refusal fails to satisfy autonomy. One view is that the patient’s judgment is based on irrational beliefs. On another view, the particular features of the doctrine of certain religious communities seem unduly coercive and restrictive, which calls into question a believer’s ability to make an autonomous decision that conflicts with that doctrine.Footnote ¹⁴ On these views, the patient’s choice is not autonomous, so it does not make sense to discuss autonomy’s priority over well-being. One could accept our argument here—that autonomy trumps well-being—while denying that cases of this sort involve a priority of autonomy, since autonomy is missing altogether. Although autonomy is clearly missing in some cases of this sort, we see no reason for thinking this is true as a conceptual matter. We therefore believe it is fruitful to consider what our judgment should be in cases in which autonomy is present, at least by stipulation.

When autonomy and well-being are both present in a given case of refusal, autonomy trumps well-being. We ought, therefore, to embrace the intuitive judgment behind Refusal. If this is true, then our remaining options are as follows: We can either treat Refusal and Request the same, which requires rejecting the intuitive judgment we saw in Request, or else accept the intuitive judgment in Request and offer an explanation for why these two cases should be treated asymmetrically. We turn now to this latter option. We will canvass what strikes us as the most plausible explanations for this asymmetry and show that none of them succeeds. If this is correct, then we are left with only one remaining option: reject the intuitive judgment in Request.

Asymmetry: What could justify treating Refusal and Request differently?

At this point, one might grant that physicians ought to accede in Refusal and yet deny that this compels us to give autonomy the same weight in Request. And this is because there are morally significant differences between Refusal and Request that justify treating them differently—or so goes the argument. But what are these differences, and why should they justify treating these cases differently?

One possible salient difference in Request appeals to the distinction between doing and allowing, or killing and letting die. Whereas in Refusal, the physician merely allows the patient to die by granting her refusal, in Request, the physician actively kills the patient by acceding to his request. According to this view, active killing (doing) is worse than passively allowing one to die, all else being equal.Footnote ¹⁵ If this is true, then granting the patient’s request might be morally wrong, whereas granting the patient’s refusal is not.

The distinction between doing and allowing is most useful in cases where we must weigh the comparative strength of our duty to avoid harming against our duty to render aid. For example, this distinction is helpful in showing why, all else being equal, we have a greater moral reason to avoid killing an innocent person than to avoid rescuing an innocent person. But although the distinction seems to apply in cases of this sort, it is hard to see why it should make a difference in cases like Request. After all, when a patient autonomously requests that some action be done to her, it no longer seems right to treat it as a harm or at least a wrongful harm. That is, her consent—in this case, via his autonomous request—makes what would have otherwise been a harm in the absence of such a request a (pro tanto) permissible act.Footnote ¹⁶

To see this, consider an analogy. If we ask whether it is worse for A to punch B in the face, or for A to allow B to be punched in the face, we might conclude, with help from the doing–allowing distinction, that the former is worse than the latter, all else being equal. But if B has consented to being punched in the face—for example, if A and B have agreed to a sanctioned boxing match, or if B has autonomously requested a punch—then our reasons for thinking it worse than merely allowing B to be punched in some related scenarios have disappeared. His autonomous request serves to negate the importance of the distinction between these two types of harm, in part because it does not seem right to call it a harm anymore at all. (And even those who believe it is still a harm should agree that it is not a wrongful harm.) No doubt it will still hurt; but it is not a harm, in the morally relevant sense.Footnote ¹⁷

This point is similar to a related point James Rachels makes in his classic paper “Active and Passive Euthanasia.”Footnote ¹⁸ Rachels argues that we cannot appeal to the distinction between doing and allowing to argue that active and passive euthanasia ought to be treated differently. When all else is equal in a given case, if passive euthanasia is justified, so too is active euthanasia. The mere fact that one is passive and the other is active does not by itself make a moral difference.

To be clear, Rachels does not put this point in terms of autonomy’s relative priority to well-being. Although he does consider some cases in which autonomy is a factor, he also considers other cases in which autonomy is irrelevant, such as with newborns. Thus, unlike our present discussion, his argument does not require that a patient has made an autonomous choice. Furthermore, Rachels occasionally invokes the value of well-being in an effort to undermine the doing–allowing distinction: In some cases, it is better to actively hasten death, as opposed to letting someone wither painfully.

In short, there are important differences between what Rachels aims to show and what is relevant to our present argument. And yet, despite these differences, we can nevertheless extend Rachels’ basic point. Refusal and Request seem to share all the same morally relevant features, apart from falling on opposite sides of the doing–allowing distinction. Even if one thinks that the difference between doing and allowing is morally relevant, it is not clear how this distinction alone could justify reordering the priority of autonomy and well-being in these two cases, given that the feature that makes it wrong and a harm—that is, the lack of consent—does not apply. If autonomy trumps well-being in allowing cases, then, in order to preserve the asymmetry, we must identify some element of doing cases that explains why either well-being is more significant or autonomy is less significant. But it is not clear what part of this distinction could explain this. The mere fact that in Request the physician is doing the relevant action does not tell us anything about how these two values ought to be ordered and mutatis mutandis for allowing the outcome in Refusal. We therefore see no reason inherent to the distinction between doing and allowing to think that autonomy is of lesser significance compared to well-being in cases of doing than in cases of allowing. In a discussion of assisted dying, Jeff McMahan makes a similar point:

One might instead appeal to a related distinction—namely, the distinction between negative and positive rights (and duties). In general, negative rights are rights against interference. In the present context, this means that patients have a weighty negative right against others’ interference with their bodily autonomy. This is the type of right the patient exercises in Refusal: The patient has a negative right against the physician (and, indeed, anyone else) interfering with her bodily autonomy. As a corollary, the physician has a stringent duty not to force the patient to accept the blood transfusion. By contrast, positive duties generally concern the securing of important goods, interests, or protections. In the present context, this means that patients also have a positive right to medical assistance from qualified professionals, which is the right the patient attempts to exercise in Request.Footnote ²⁰ (Or, if a right to assistance is too strong, then such assistance is at least permissible.) But positive rights generally impose much less stringent duties on others than do negative rights. To be sure, physicians have a general duty to provide care based on the patient’s positive right, but this right is more easily outweighed than in the case of negative rights.Footnote ²¹ This is in part because, as a general matter, positive rights make greater demands on others than do negative rights: Positive rights require of the duty-bearer that they act or render aid, whereas negative rights require only inaction and non-interference. In other words, positive duties tend to be less stringent than negative duties with similar outcomes because positive duties involve a greater imposition on duty-bearers.Footnote ²² And it is at least partly for this reason that we generally think that it is easier to justify infringing a positive right than a negative right.

This distinction as applied to Refusal and Request appears, at first glance, to avoid the problem that beset the doing–allowing distinction. After all, unlike the more basic doing–allowing distinction, the difference between positive and negative rights (and duties) does seem to pick out a morally relevant difference between the two cases—namely, the fact that Refusal involves a negative right, whereas Request involves only a positive right (or at least a permission). Given that all else is equal, it seems that the right in Request is weaker than the right in Refusal. And this, it might seem, is enough to show why we should treat the two cases differently.

We grant that this distinction helps to ground an important difference between the rights in the two cases, which in turn provides the resources for understanding the extent of the physician’s duties in both cases. In particular, this distinction helps explain the intuitive judgment, which we grant, that physicians have greater moral reason in general to comply with patients’ refusals than patients’ requests, all else being equal. Physicians have negative duties to avoid interfering with the bodily integrity of their patients, and the threshold for justifying forcibly administering a medical procedure to a patient who steadfastly refuses it is incredibly high. By contrast, given that it involves only a positive duty, the threshold for declining to act on a patient’s request for assistance is comparatively low. Moreover, positive duties in general—and this is certainly true in Request—require more than just an agent’s positive action: They also require resources, certain particular kinds of access, skills, individual labor, and so on. Thus, there is a wider range of considerations relevant to determining whether one can or ought to fulfill a positive duty. That is, all else being equal, there are many more reasons that might count against fulfilling a positive duty than would be true of an otherwise similar negative duty.

The differences between the positive right in Request and the negative right in Refusal are important, particularly concerning the extent of the physician’s obligations and our all-things-considered judgments about what he ought to do. But although these are important issues concerning the broader question of what the physician ought to do, neither of them bears on the question we initially sought to answer here—namely, what can explain the difference between the ordering of the values of autonomy and well-being in the two cases?

We grant that positive duties are generally less stringent than negative duties, which is reflected in what physicians can be morally required to do; we grant that positive duties tend to require considering a broader range of relevant factors. But it does not follow from either of these facts that we are justified in reversing the relative priority between autonomy and well-being between the two cases. For one thing, the special factors relevant to positive duties do not make any necessary reference to either autonomy or well-being. It is not as if concerns about the patient’s good suddenly become much more important or salient in cases where the treatment is requested as opposed to refused, nor is it the case that the value of autonomy is weakened—either absolutely or relative to well-being—in cases where the patient invokes a positive right. There is simply no connection between the ordering of these two values and the type of right or duty in question.

Moreover, the fact that the physician’s positive duties are less stringent than her negative duties concerns only the question of what we can demand of the physician in terms of positive action. Whether or not the physician can justifiably decline to comply with the patient’s request does not, however, turn on the independent importance of the patient’s good relative to her autonomous choice. Rather, it concerns the duty’s weight, which is to say, the extent of the demands it can justifiably place on the physician.

To see this point more clearly, consider a case in which the physician, having assessed the patient as fully capable and autonomous, is willing to grant the request. In this case, the distinction between positive and negative duties is not nearly as relevant: The physician voluntarily accedes to the request, so there is no longer any concern about the extent of this duty’s imposition. Those who think the positive–negative duty distinction is what explains the reversal in ordering between autonomy and well-being do not have grounds for maintaining that ordering once the physician accedes to the request. Absent some further arguments, they must hold that autonomy trumps well-being in this case. But it is hard to see why the physician’s acceding to the request produces this reordering effect. What about the physician’s choice serves to change the relative values of these two values themselves?

We suspect that even if a physician voluntarily accedes to the request, many will still claim that it is wrong to do so. These objectors’ concerns do not rest entirely with the extent of the demands placed on the physician, but rather with the idea that physicians could reason this way at all—that is, by prioritizing autonomy over well-being—in cases like Request. If it is indeed true that some will hold on to this judgment in this case, then the distinction between positive and negative rights and duties does not offer an explanation for the asymmetry.

To be sure, the weight of this duty is determined in part by the harm (or loss of good) that would thereby be done. Duties that protect agents from more serious harms are, appropriately, more serious duties. But one would need a further argument to explain why the physician would do harm to the patient in a case like Request. After all, the request is fully autonomous. And, as we have already suggested, complying with the request of a fully autonomous patient does not qualify as a harm, at least in the sense relevant here.

Symmetry: Rejecting the intuition in “Request”

If the foregoing arguments are correct, then there is no clear basis for preserving the asymmetry between Refusal and Request. As such, the only remaining option is to reject the judgment in Request, which held that well-being trumps autonomy. The better view is to treat cases like Refusal and Request the same: When looking to the values of autonomy and well-being, the former wins out.

One might object that rejecting this intuition would give rise to several unpalatable conclusions, which might be so serious as to make one of the other claims comparatively easier to accept. For one thing, one might worry that allowing autonomy to trump well-being in cases like Request will give rise to dramatic shifts to physicians’ professional roles: If this were the norm, physicians would turn into professional killers, or, to borrow a phrase from Leon Kass, “scalpels for hire.”Footnote ²³

Since we have addressed this point at length elsewhere, we will offer some briefer responses here.Footnote ²⁴ Our first response is that Kass is mistaken that it is objectionable to use medicine to help people live in accordance with their values. Many types of treatment are already autonomy-centered, including vasectomy, tubal ligation, gender confirmation surgery, and nontherapeutic abortion. Most of the time, these procedures are not medically indicated or proposed by the physician, and they do not involve healing. Rather, people get them because they want them. Of course, some of these treatments are morally controversial, but few people have an issue with, for example, tubal ligation because it is request-based. Similarly, physicians are justified in providing nontherapeutic abortions because having an unwanted baby is a significant disruption to one’s life.Footnote ²⁵ Promoting autonomy is an important goal of medicine.

Second, some might worry that our view is committed to a strong obligation of physicians to fulfill requests and that this might outweigh the right of physicians to decide which treatments to offer. We disagree on the extent that physicians have a right to conscientious objection, but nothing about our view here threatens conscientious objection. As we have said, although our view posits a pro tanto duty to promote autonomy, there are many reasons why the physician need not accede to requests, including, perhaps, matters of conscience.

Finally, there is a general worry that a request-based system will upend healthcare and, perhaps in the extreme, society. This worry holds that, if patients can request treatments, nothing will stop young people from getting assistance in dying while in the throes of a breakup, or people accessing opioids for pleasure, or someone using gene editing to make himself taller so he can be better at basketball. But as we stated above, an absolute right to care does not follow from a request, and there are many reasons why the requests just listed could be refused, including equality, concerns that the request is not autonomous, and the risk of harm to others.

Those who remain unpersuaded by our argument have two options. The first is to offer an explanation for the asymmetry in the value ordering of autonomy and well-being. Such an argument must explain why autonomy trumps well-being in cases of refusal, but the ordering is reversed in cases of request. As we have argued, we doubt that such an argument can be made. The other option is more radical. Some will be convinced that the asymmetry cannot be explained, but, rather than agree with us that autonomy trumps well-being, they will go the other way by claiming that well-being should consistently outweigh autonomy. This is the way medicine used to be done. The physician would determine what he thought was best for the patient and then he would do it. We find this approach indefensible, as do nearly all physicians and bioethicists. Therefore, one of the other options is necessary.