ECU Logo
Department of Medical Humanities


medical humanities newsletter
The Bioethics Center, University Health Systems of Eastern Carolina
Department of Medical Humanities, The Brody School of Medicine at East Carolina University
Living Wills and Former Decisions
John K. Davis, Ph.D., J.D.

Because incapacitated persons are no longer able to understand or reaffirm the instructions stated in their living wills (or the values behind those instructions), some scholars have argued that living wills no longer represent the wishes of incapacitated patients, and therefore should be disregarded in favor of patients’ current best interests. In other words, what’s stated in a living will is only the patient’s former decision, not his or her current decision. If it’s only a former decision, why should we respect it?

My solution to what I call the ‘Former Decision Problem’ requires a distinction between “highest-order desires” and “latest desires.” A highest-order desire is what you want when you have two conflicting desires, and you must choose between them; your choice between the two desires is a third, highest-order desire. Usually your latest desire is your highest-order desire, for when you form your latest desire you are aware of your earlier desires, but when you formed your earlier desires, you were not aware of your future desires. In cases of declining competence, however, the reverse may be true: if the patient wanted, for example, death with dignity, and knew he would later want to have the simple pleasures available in dementia, his living will probably says that he wants his desire for death with dignity to be fulfilled at the expense of the desires he anticipates having when he becomes demented. When he formed his earlier desire for death with dignity, he was aware of the desires he would later have in dementia. Therefore, his earlier desire, not his later one, is his highest-order desire.

I will argue that normally we respect a person’s latest desire not because it is latest, but because it is usually highest-order. We respect highest-order desires because they express what a person wants, all things considered— literally, after that person has considered all the desires she has that may conflict. Because we respect latest desires on the grounds that they are highest-order, we should respect earlier desires, even former ones, when they are highest order. People use living wills to tell us which of two desires, an earlier one and a later one, they prefer to satisfy. That is why we should
respect living wills [1].

The Former Decision Problem
There are problems with living wills. Many of these problems are practical: living wills are often hard to interpret and apply, patients are often poorly informed about their options when they execute living wills, and a living will may be ignored when the family is set on a different course of treatment. Sometimes, just finding out whether the patient has a living will—and where she put it—is the first problem of all.

Beyond these concerns, many bioethicists and clinicians have deeper, more theoretical misgivings. Some people believe that living wills don’t promote the best interests of mentally incapacitated patients. Sometimes these critics elaborate this concern by saying the demented or unconscious patient is now a “different person,” or that his earlier decision is irrelevant to his current needs.

At first glance, these theoretical concerns do not seem well founded. No one seriously believes the elderly patient there in the ICU is now a different person (how old is she?), and competent patients clearly have the right to make decisions against their best interests, so if the patient was competent when the decision was made, why override it now? As for the “irrelevance” of decisions made before the patient lost mental capacity, if we could override such decisions whenever patients lose mental capacity, then surgeons could ignore patient wishes as soon as the patient is anesthetized, and patients would never have authority over their own surgeries.

However, we should not dismiss these concerns, for they arise from a still deeper theoretical concern: is the patient’s earlier decision still hers? Is she still decided upon that course of action after she can no longer understand or reaffirm it? The answer to this question could fatally undermine the moral authority of living wills, for if becoming mentally incapacitated is like changing one’s mind, then we should not honor living wills any more than we should respect a patient’s decision after she’s changed her mind.

Here is why it matters whether the patient’s earlier decision is “still hers”: if we should respect a patient’s health care choices [2], then presumably we should respect choices made in advance. However, when the patient becomes mentally incapacitated, her earlier decision may seem not to be her decision anymore. Consider a recently reported case concerning a 73-year-old woman on a ventilator and feeding tube, whose living will called for termination of life support if she had no “reasonable expectation for recovery from extreme physical or mental disability:”

The patient remained confused but awake and alert. She interacted with staff and visitors and spent much of the day sitting up in a chair watching television. The psychiatrist who had followed her for manic depression found her cognitively impaired and incompetent for decisionmaking.... The patient’s friends and her primary care physician requested that she be removed from the ventilator. The critical care physician and nursing staff objected.... When asked whether she wanted to be removed from life support she would indicate “no.” The patient’s friends and primary care physician, however, believed she was too confused to understand the question. Citing the psychiatrist’s report, they argued that she was incompetent to make this decision [3].

This patient never changed her mind, but she lost the capacity to comprehend or reaffirm her earlier decision to refuse life support. Is she still decided upon refusing life support, or not? Suppose, at age 3, you wanted to become a fireman when you grew up, never changed your mind, but eventually forgot that you ever wanted that. In your late teens, with no idea what you want to do with your life, you no longer want to be a fireman, even though you never changed your mind about it. Moreover, you have clearly outgrown your decision to be a fireman, and your later course of life doesn’t point in that direction. If I remind you that you once wanted to be a fireman, you would renounce that decision as soon as I remind you of it. Perhaps the decisions announced in living wills are former decisions in this sense: if you remind the patient of his earlier decision, he will renounce it.

However, that kind of “former decision” does not clearly fit end-of-life cases like the one above, even though the patient never changed her mind and has forgotten all about her earlier decision. You cannot ask her whether she wants to renounce her but, due to permanent mental incapacity, is no longer capable of reaffirming it either? If you have theoretical misgivings about living wills, there is a good chance that you are troubled by the Former Decision Problem.

Obviously, this problem seems to undermine nearly all advance directives, since most of them come into play when their authors are permanently incapable of understanding their own directives. We are supposed to respect what a patient currently wants, not what a patient used to want. The decisions expressed in advance directives are certainly earlier decisions (in the sense of having been made earlier) by the time the directive is supposed to take effect. If they are also former decisions (in the sense that the patient is no longer decided upon that course of treatment, which is somewhat like a change of mind, but different), then it seems to follow that we should ignore advance directives, and just do whatever seems to be in the patient’s current best interests.

Hypothetical reaffirmations don’t solve the Former Decision Problem
Here is one possible answer to the Former Decision Problem: suppose we can say, with some assurance, that the patient in our case would reaffirm her living will if she had full mental capacity. It’s a hypothetical reaffirmation, for her incapacity is permanent, but in many cases it’s possible to know what a patient would now want if she were not mentally impaired, and in such cases we often treat patients in accord with such hypothetical wants. Perhaps we should respect living wills when we have reason to believe the patient would now reaffirm it (if she could), and ignore living wills otherwise. After all, if the patient would reaffirm the decision (provided she had the mental capacity to understand the question), it’s really a current decision—and we must respect competent current decisions. Right?

Not quite. Even those choices a mentally incapacitated person would now reaffirm are former decisions, because choices, desires, wishes, and decisions are dispositions. If desires (for example) were not dispositions, then Ms. Smith, napping on a couch at home, does not desire a new roof even though she’s been talking about it for months. This is, of course, silly. When someone desires a thing, she is either actively desiring (or choosing, deciding upon, etc.) that thing, or she is disposed to actively desire, choose, or decide upon that thing. That is, if she’s awake and focused on that issue, she will actively want that thing. This is why you don’t want to be a fireman even if you’ve forgotten you ever wanted to be one: you are not disposed to actively want it if the issue comes up.

However, to have a disposition one must be capable of acting out that disposition. Therefore, someone who is incapacitated— incapable of acting out that disposition— has no such disposition. So, in order to say that Ms. Smith wants (or has decided upon, or wishes, or chooses, or values, etc.) a thing requires that she have a disposition to want it whenever she’s thinking about it, and that, in turn, requires that she be mentally capable of thinking about it. If she suffers a stroke during her nap and becomes permanently unconscious, she is then incapable of actively wanting a new roof, and cannot be disposed to actively want a new roof. Because she is no longer disposed to want a new roof, she no longer wants one, even though she never changed her mind.

The same point applies to our 73-year-old patient: because she lacks the mental capacity to actively want life support terminated when she has no reasonable expectation of recovery from extreme disability, she has no disposition to actively want such a thing, and therefore is no longer decided upon that course of action. Thus, we seem to have no reason to honor her living will.

The Substituted Judgment Principle doesn’t solve the Former Decision Problem
So, hypothetical reaffirmations do not support the moral authority of living wills. However, perhaps we can defend them with the Substituted Judgment Principle. This is a familiar part of health law and medical ethics: if a patient cannot make a decision, treat him as he would want to be treated if he could decide. Usually we look at the patient’s past statements and behavior to see what he valued. Families and other surrogate decision makers do this all the time with mentally incapacitated patients.

On this way of thinking, we should not follow a former decision when the patient changed her mind, but we should follow a former decision when she would reaffirm it if she could, even though her inability to reaffirm it makes that decision a former one. What the patient would want is a pretty good indication of what is in her best interests now. Since she knew what she valued, we should respect her values by acting as if she were able to defend them herself. Sometimes those values are announced through living wills, and sometimes they are merely evident in a patient’s life history, but either way, they command our respect. The version of this approach that is based on the patient’s life history is often called the Substituted Judgment Principle. (For convenience, I will use that phrase to denote all versions of this approach).

However, even assuming that the patient’s interest in something, such as personal dignity, survives her loss of mental capacity [4], her living will does more than announce that she has such an interest. Her living will also announces an intention to promote that interest at the expense of other, more mundane interests she will later have, such as feeling pleasure, avoiding pain, receiving food and water, and so on. We all have many interests, and these interests can conflict. When they conflict, we make decisions about which interests to promote, and which ones to subordinate to other concerns. A living will cannot achieve its purpose unless it announces a decision or preference about how to prioritize the patient’s interests and values. In the case discussed above, the patient’s living will tells us what to do when her interest in personal dignity conflicts with her interest in remaining alive: let her die with dignity. Therefore, she is no longer decided upon promoting personal dignity at the expense of a deeply demented life.

So far, we have rejected two possible solutions to the Former Decision Problem. Hypothetical reaffirmations do not solve the Former Decision Problem, for what the patient would want if she were competent is still a former desire or decision. The Substituted Judgment Principle does not solve the Former Decision Problem, for what the patient would now want if she had full mental capacity may deviate from her best interests. Therefore, if what she would now want commands respect, it must do so because we should respect her autonomy rather than her best interests. However, respect for autonomy requires an autonomous desire or decision that can still be attributed to the patient.

In the last section I’ll offer a solution that I believe succeeds. Before I do, however, I will explain how a clear understanding of this problem helps us to understand two other things: 1) why we are more inclined to interfere with the wishes of mentally incapacitated patients when the wish is shortterm, than when the wish is stable and has been held for a long time; and 2) why many people object to living wills on the grounds that the patient seems like a “different person,” even though no one literally believes the patient has turned into someone else.

The Former Decision Problem explains why we are more inclined to respect stable, long-lasting desires. We have seen that one is decided upon a course of action only if one has the disposition (and thus the capacity) to act upon that decision when actively thinking about it. However, dispositions come in degrees. You might want a new Toyota Camry one week, think about it the following Tuesday with indifference, and think some more that weekend and strongly want one. Some dispositions are stronger than others, and some last longer or occur more frequently than others.

This is important in a medical context because sometimes it is argued that we should honor a patient’s preferences to the extent they are long-standing and stable. Joel Feinberg takes this approach, asking whether a preference at a given time, earlier or later,“ reflects the settled disposition of the chooser as an enduring self over time.”[5] For example, if a man with a drinking problem gets his host to promise to limit him to two drinks, his later drunken request for more drinks does not reflect what he really wants.

Many of us are more inclined to interfere with someone who is temporarily incapacitated than with someone whose mental illness is permanent. If your patient has been mentally ill for many years, you are less likely to try to stop him from sleeping on heating grates, while if his mental illness is episodic, you’re much more likely to take steps to make sure he doesn’t carry out a temporary enthusiasm for outdoor living. It’s easy to think we do this because we know he’ll later regret it, but that’s a matter of overriding his wishes for his own good, and that reason for interfering applies equally well to people most of us are less inclined to interfere with: the permanently mentally ill.

The real reason we are less inclined to interfere with mentally ill people when their preferences are stable and long-standing is that their dispositions to reaffirm those preferences (when they’re consciously thinking about them) are stronger to the degree the preference has lasted longer and with fewer changes of mind. In a contest between the preferences he has in a temporary state of mental incapacity and those he has in his “ normal” state of full mental capacity, the preferences he has in his “normal” state are more truly his. If, however, being demented or mentally ill is his usual state, then preferences formed under those conditions are more truly his.

The ‘different person’ objection is really the Former Decision Problem in disguise
Many people notice that deeply incapacitated patients seem almost to be different persons from their earlier selves. As I mentioned earlier, some people argue that we should discount or disregard living wills for just this reason. After all, such patients are, in a sense, different people from the individuals who drafted their living wills, and no one has a right to impose a decision upon another person, even if that other, later person somehow inhabits what’s left of the earlier person’s body [6]. This is a strange sounding objection to living wills, but a popular one nonetheless [7].

The Former Decision Problem approach helps make sense of this metaphysical issue in more ordinary terms. Few people believe that even profoundly demented patients are literally new persons. Grandmother, there in the ICU, is still Grandmother, even if she cannot recognize her family and her temperament and cognition are drastically different. If she enters a permanent coma, it’s Grandmother who’s in the coma, not someone else.

I think the real motivation for the “different person” approach is a sense that somehow the wishes expressed in a living will are not the wishes of the patient, even though he never changed his mind. Of course such patients change dramatically from their earlier selves, but it’s a mistake to think the decision in a living will is no longer Grandmother’s because Grandmother is now someone else. The better explanation is that Grandmother is still there but the decision has disappeared—it is a former decision. The change in Grandmother tempts some people to think that Grandmother herself disappeared along with her decision, but that’s an overstatement. She has, however, changed so much that some of her former decisions and preferences are gone.

Solving the Former Decision Problem
To defend the moral authority of living wills, we must solve the Former Decision Problem. I believe the solution lies in distinguishing between a latest decision and a highest-order decision, and understanding which kind of decision is important when it comes to respecting a patient’s treatment decisions.

Consider the 73-year-old woman in the case discussed earlier, and allow me a few extra assumptions for illustration. A few years earlier the woman wanted no life support if she had no “reasonable expectation for recovery from extreme physical or mental disability,” as her living will states. Now she wants to watch Teletubbies on television. When she decided she wanted life support terminated under these conditions, she was aware that she might enjoy television even while extremely demented. Now that she suffers from extreme mental disability, she is aware that she enjoys television, but not aware that she ever cared about death with dignity or any similar value; in fact, she can no longer understand the concept of death with dignity. There are two desires at issue here:

Latest desire: To watch television
Highest-order desire: To terminate life support despite the pleasures of television

A person’s latest desire on a given issue is the one most recently formed, while the agent’s highest-order desire on that issue is the one formed after reflection upon all other preferences and information the agent has had on that issue. Here, the issue is what should be done with this patient in a state of extreme mental and physical disability. Her latest desire is to watch television; her earlier desire was to be allowed to die.

I call that earlier desire a “highest-order desire” because it is a desire formed while she was aware of another desire, while that second desire was not formed while she was aware of the first desire. A few years ago, this woman determined that, in a situation where she is capable of only those experiences a deeply demented person can have, then she wants to have no experiences at all. Her earlier desire discounts the demented desires she anticipates having later on. She formed a desire for death with dignity while aware that she would at some point desire only very simple pleasures. However, when she desires only very simple pleasures, she is no longer aware of her earlier, more complex desire for death with dignity. Because her desire for death with dignity was formed while she was aware that she would (at some point) desire simple pleasures, her desire for death with dignity is a higher-order desire. Her desire for simple pleasures is lower-order, for she formed that desire when she was no longer aware of the desire for death with dignity. Her desire for death with dignity is her “highest-order” desire.

The 73-year-old woman’s latest and highest- order desires conflict: they cannot both be satisfied. Which desire should we satisfy? At first glance, this question seems easy: her earlier, highest-order desire is a former desire (or decision made upon that desire, if you like). Usually we satisfy what a person wants now (her latest desire), not what she used to want (her former desire). Why should we not do so with the 73-year-old patient, and simply ignore her living will?

The answer to this question (and to the Former Decision Problem itself) lies in why we respect a person’s latest desire. Usually an agent’s latest desire is also his highestorder desire. This is because your latest desire about something, e.g., to vacation in Martinique this year, is usually formed after reflecting upon your earlier desire, e.g. to vacation in Paris. Upon reflection, you decided Paris would be too cold during the time you had available for a vacation, so you changed your mind. When you wanted to go to Paris, you were not aware that you would later want to go to Martinique. However, when you desired to go to Martinique, you were aware of your earlier desire to go to Paris. In this case, as in most cases, your latest desire is your highest-order desire on the issue in question (where to spend your vacation). Most of our earlier desires cannot be our highest-order desires, because we formed them before the latest desire, and we could not consider the future desire when considering and forming our earlier desire.

However, in cases of dementia or other mental incapacity, this pattern can become reversed: instead of the latest desire being the highest-order desire, the earlier desire (even a former one) may be the highestorder desire. In such cases, an earlier preference, formed when the person was sufficiently competent to anticipate and reflect on possible later preferences formed during dementia, is his highest-order preference (provided his dementia prevents him from reflecting upon his earlier values). Cases of declining mental competence (and these include many end-of-life cases) are often like this: the person’s latest desire is not his highest-order desire. Rather, some former desire is his highest-order desire on how he should be treated in these circumstances.

But wait, you say: why respect the 73-yearold woman’s former desire, even if it is her highest-order desire? She doesn’t want that anymore. My answer begins with another question: why do we believe we should respect a person’s latest desire, and not former desires? Latest desires have two properties: they are latest, and they are (usually) highestorder. Do we respect latest desires because they are latest, or because they are highestorder? In normal cases, we don’t need to worry about this; they are both. In cases of declining mental competence, however, the situation can become reversed: the earlier, former desire can be the highest-order, and the latest desire can be lower-order.

I contend that, when a latest desire conflicts with a higher-order former desire, we should respect the former desire and not the latest one, even though the latest one is what she wants now. There are two reasons for this claim.

First, respect for autonomy requires respect for what the person prefers when he is most informed, and thus requires respect for his highest-order desire. If the principle of respect for autonomy requires respect not for the person’s latest but for his highest-order preferences, and if in cases of mental incapacity the latest preference is not the highest-order preference, then the principle of respect for autonomy would require respect for the earlier, highest-order preference, even though it is a former preference. Therefore, we must respect even former preferences, provided they are the highest-order preferences the person has ever had on the issue in question.

The second reason why we should respect highest-order desires over latest, lower-order desires (when they conflict, of course) is that a highest-order desire is essentially a choice between two desires. The patient in our case was aware that she wanted a death with dignity, and that she would later want to watch television, and have no conception of dignity. In effect, she had to choose between a desire for dignity and a desire to watch Teletubbies once again. Given that choice, she chose to satisfy the desire for dignity at the expense of Teletubbies. If we are to respect what she desires, we must respect what she desires all things considered. When she considered all things, she desired death with dignity. There is no way to satisfy both desires, for they conflict. When a patient’s desires conflict, the only way to respect the patient’s autonomy is to respect her choice concerning how to resolve that conflict. The woman in our case earlier chose to resolve it in favor of her desire for dignity. Therefore, to respect her autonomy, we must respect her former desire—her highest-order desire on that issue, and carry out her living will.

I don’t contend that all living wills must be respected. For my argument to apply, the patient must, at the time the living will was executed, have anticipated and rejected the kind of later desires the living will would eventually conflict with. If she did not anticipate her later situation on at least a general level, then the earlier desires expressed in the living will are not highest-order, for she did not form a desire about her later desires. She failed to form a desire about her later desires for the simple reason that she didn’t anticipate them.

In the defense of living wills I propose, the key practical question in ascertaining the moral authority of an advance directive is whether the agent anticipated and understood the preferences she would later have if she were to become mentally incapacitated. If that understanding was not sufficient, then arguably she cannot have autonomously rejected the future desires, for she did not know what she was rejecting. There are, of course, interesting questions about the nature and degree of understanding that is necessary for such decisions to command our respect, but those questions are large enough to require a paper of their own.


1. This article is a shortened and revised version of “The Concept of Precedent Autonomy,” which appeared in Bioethics 2002; 16: 114-133. It includes some new materials.

2. Beauchamp, T, and Childress, J; Principles of Biomedical Ethics 4th ed. (New York: Oxford University Press, 1994), pp. 125-127.

3. Dresser, R, and Astrow, A; “Commentary” on “An Alert and Incompetent Self: The Irrelevance of Advance Directives.” The Hastings Center Report 1998; 28: 28-30. Eventually the advance directive was followed, the ventilator was removed, and the patient died. That the patient said “no” to removing the ventilator is disturbing. It is not clear from the case description whether she truly understood the question, or simply responded to subtle behavioral cues. For purposes of discussion, I will assume the psychiatrist was right, and that she lacked the mental capacity to understand the question.

4. I contend that interests, unlike decisions and preferences, do survive a loss of mental capacity. However, my argument for that claim requires another paper.

5. Feinberg, J; Harm to Self (New York: Oxford University Press, 1986), p. 83. This may also be what Thomas Schelling has in mind when he suggests deferring to the “ authentic self” when past and future preferences for the agent’s future treatment conflict, in “Ethics, Law, and the Exercise of Self-Command, in McMurrin,” S (ed.); The Tanner Lectures on Human Values IV (Salt Lake City: University of Utah Press, 1982), pp. 74-75.

6. For a general discussion, see Buchanan, A, and Brock, D; Deciding for Others: The Ethics of Surrogate Decision Making (Cambridge: Cambridge University Press, 1990), pp. 152-189. Buchanan and Brock do not endorse this objection.

7. Buchanan and Brock discuss and respond to this objection in Deciding for Others, Id.