Respect for autonomy is in trouble. In recent work in this journal¹ and elsewhere,² O’Neill has forcefully argued that respect for autonomy, as it has come to be used in medical ethics, is philosophically indefensible. If her arguments are sound, then, contrary to the standard view, respect for autonomy cannot be the source of the ethical requirement to seek informed consent before treating a patient or enrolling a participant in a trial. So her critique goes to the heart of contemporary medical ethics: if O’Neill is right, medical ethicists have systematically misunderstood two of the most fundamental concepts they deal with—respect for autonomy and informed consent.

This paper has four sections. Section 1 distinguishes between three different ways of talking about respect for autonomy, and looks in more detail at the one that has come to be central to bioethical writing on informed consent—namely, the idea that we should respect autonomous choices. Section 2 argues, following O’Neill, that it is implausible to think that the purpose of informed consent requirements is to respect autonomous choices. Section 3 argues that O’Neill’s proposed reworking of autonomy and informed consent is inadequate. O’Neill’s approach requires us to adopt a more modest view of the purpose of informed consent procedures. In her view, the purpose of informed consent procedures is simply to avoid deception and coercion, and the ethical justification for informed consent derives from a different ethical principle, which she calls principled autonomy. I argue that contrary to what O’Neill claims, we cannot derive the wrongness of coercion from principled autonomy, and so its credentials as a justification for informed consent procedures is weak. Section 4 argues that we do better to rethink autonomy and informed consent in terms of respecting persons as ends in themselves, and a characteristically liberal commitment to allowing individuals to make certain categories of decisions for themselves.

THREE CONCEPTIONS OF AUTONOMY

There are at least three different things that “autonomy” is used to refer to when medical ethicists claim that we should respect a patient or a research participant’s autonomy:

1. Autonomy sometimes refers to the capacity to make autonomous choices, with the underlying claim being that people who are capable of making autonomous choices are worthy of respect whereas people who lack this capacity are not.

2. Autonomy sometimes refers to autonomous choices, with the underlying claim being that autonomous choices are worthy of a respect that non-autonomous choices are not.

3. Autonomy sometimes refers to a sphere of decisional privacy, with the underlying claim being that we should respect autonomy by allowing persons to make certain sorts of choices for themselves without coercing or otherwise interfering with them.

While all three senses of respect for autonomy are found in the literature (and sometimes side by side in the same article), it is the second that has come to predominate, following influential expositions by Faden and Beauchamp³ and by Beauchamp and Childress.⁴ O’Neill’s argument is directed against this second conception, which she refers to as “individual autonomy”, and as we shall see shortly, makes a powerful case against taking it to be a fundamental value in medical ethics.

In general, respect is an attitude one adopts towards someone or something, which is characterised by the judgement that the respected thing or person places legitimate limits on what one may do with it, or to it. Something that is worthy of respect can be either intrinsically or extrinsically worthy of respect. When something is extrinsically worthy of respect, it is worthy of respect because of something else. For example, the pen that Lincoln used to sign the Emancipation Proclamation could plausibly count as extrinsically worthy of respect.⁵ (We might think it wrong to destroy it, for instance, but the reason it would be wrong to do so is the role the pen played in history, not anything related to the particular structure or craftsmanship of the pen itself.) Something that is intrinsically worthy of respect, on the other hand, is worthy of respect just in virtue of what it is. Many people think that human beings are worthy of respect just in virtue of being human.

The different accounts of respect for autonomy suggest different kinds of respect claims. Sense 1 suggests that autonomy is a feature of persons that grounds a duty to treat them with respect: we should respect persons (at least in part) in virtue of their capacity for autonomy. In this view, autonomy is intrinsically worthy of respect: a being that is autonomous is worthy of respect just in virtue of being autonomous. This line of thinking has its roots in Kant’s⁶ conception of persons as ends in themselves: “rational beings are called persons inasmuch as their nature already marks them out as ends in themselves, i.e. as something which is not to be used merely as means and hence there is imposed a limit on the arbitrary use of such beings, which are thus objects of respect”.

Sense 3 implies that autonomy as decisional privacy is only extrinsically worthy of respect: we should respect people’s decisional privacy not for its own sake, but because doing so will have beneficial consequences. This line of thinking has its roots in Mill.⁷

It is less clear what kind of respect claim underlies sense 2. Clearly, sense 2 implies that autonomy is a feature of certain choices and not others, and suggests that the reason to respect those choices is that they are autonomous. But what is the nature of the respect that autonomous choices are due: should we treat autonomous choices as worthy of respect in the way that Lincoln’s fountain pen is, or in the way that a person is?

The argument this paper presents points to two serious difficulties for those who think that autonomous choices are intrinsically worthy of respect. First, even if it is true to claim that autonomous choices are intrinsically worthy of respect, this claim can have little relevance to bioethics, since only few choices will count as autonomous in the relevant way, and, second, given that informed consent procedures protect both autonomous and non-autonomous choices indiscriminately, the claim that it is respect for autonomous choices which leads us to seek informed consent is highly dubious. The final section defends an account of respect for individuals’ healthcare choices that makes such choices extrinsically worthy of respect.

PROBLEMS WITH INTRINSIC RESPECT FOR AUTONOMOUS CHOICES AS A JUSTIFICATION FOR INFORMED CONSENT

The claim that any and all choices that individuals make are intrinsically worthy of respect is manifestly implausible, given that some choices are “self-centred, pig-headed, impulsive, random, ignorant, out of control and regrettable or unacceptable for these and many other reasons” (O’Neill,² p 28). So it is clear that we cannot let just any old choice count as autonomous if we are going to claim that all autonomous choices are intrinsically worthy of respect.

Any account that wants to claim that autonomous choices are intrinsically worthy of respect must meet two constraints. First, its criterion for what makes a choice autonomous must correctly latch onto a feature that does make a choice worthy of respect just in virtue of possessing it: if it did not do so, then it would remain unexplained why we should treat autonomous, but not non-autonomous choices, as worthy of respect. It is plausible to think that any such criterion would have to be fairly demanding; and it would certainly have to go beyond the factors of being adequately informed and acting voluntarily, given that there are morally repugnant acts and choices which share these features, and which we would presumably not want to count as worthy of respect in themselves.

Second, it must place autonomous choices within the ability of every normal adult. This requirement is not rendered necessary by the concept of autonomous choice: it is perfectly coherent to have a concept of autonomy (like Nietzsche’s conception⁸) under which few people, and few actions, turn out to be autonomous. But such a position would clearly be incompatible with the antipaternalistic assumptions of modern bioethics.

However, no account of autonomous choices could meet both constraints simultaneously: if an account makes autonomy sufficiently demanding that actions which meet it are worthy of respect for their own sake, then the account will be too demanding to allow the vast majority of the choices that patients make about their healthcare in a hospital setting to count as autonomous. However, if the account is sufficiently lax as to allow the ordinary choices of patients in a hospital setting to meet it, then we will no longer have reason to think of such choices as worthy of respect in their own right.

In practice, defenders of the intrinsic respect-worthiness of autonomous choice attempt to mask this problem by shifting their claims according to the context. When they talk in terms of why autonomous choices should be respected, they give a fairly demanding account which refers to authenticity, second-order desires, or self-mastery, and so on. But when they come to apply their principle of respect for autonomy in practice, they tend to be much more permissive, and assume that respect for autonomy is appropriately operationalised through the seeking of informed consent. However, as O’Neill argues, it is highly implausible to think that informed consent requirements could provide the appropriate response to the intrinsic respect-worthiness of autonomous choices: “[b]y insisting on the importance of informed consent we make it possible for individuals to choose autonomously, however that is to be construed. But we in no way guarantee or require that they do so” (O’Neill² p 37). Informed consent does nothing to ensure that autonomous choices are respected in a way that non-autonomous choices are not: as O’Neill² puts it, “Requirements for informed consent are relevant to specifically autonomous choice only because they are relevant to choice of all sorts” (p 38). Hence, it seems at best somewhat misleading to claim that in seeking informed consent we are acting out of the intrinsic respect-worthiness of autonomous choices.

O’Neill’s argument points to two fundamental problems: first, even if it is true to claim that autonomous choices are intrinsically worthy of respect, this claim has little relevance to bioethics, since only few choices will count as autonomous in the appropriate way, and, second, given that informed consent procedures protect both autonomous and non-autonomous choices indiscriminately, at the very least it requires further explanation why seeking informed consent is the best way to respect autonomous choices. For, if our true goal were to respect autonomous choices, it would seem to be better to adopt a policy which allowed us to discriminate between those choices that are autonomous and those that are not. So, O’Neill’s argument leaves us with a problem for bioethicists’ standard views of both respect for autonomy and of the justification of informed consent procedures.

PRINCIPLED AUTONOMY

O’Neill’s proposed solution is twofold. First, to urge a less grandiose conception of the purpose of informed consent procedures: in her view, the purpose of informed consent is to ensure that no one is coerced or deceived, and it is not to ensure that autonomous choices are respected. Second, to introduce another conception of autonomy, which she calls calls “principled autonomy”, and to argue that this conception of autonomy should be foundational for bioethics.

O’Neill’s conception of principled autonomy, although new to bioethics, has a long pedigree elsewhere: for it is the conception of autonomy that Kant articulated in his universal law formulation of the categorical imperative. As O’Neill² expresses it, principled autonomy is “a matter of acting on certain sorts of principles, and especially on principles of obligation ... principled autonomy is expressed in action whose principle could be adopted by all others” (pp 84–5). O’Neill argues that this conception of autonomy allows us to see that the wrongs that informed consent aims to protect against—coercion and deception—are wrongs independent of an appeal to respect for autonomous choices. I shall suggest that O’Neill’s conception of principled autonomy fails to evade the classic difficulties that affect the universal law formulation of the categorical imperative, and that pace O’Neill principled autonomy in fact fails to rule out coercion, and so its credentials as a justification for the requirement to seek informed consent are weak.

O’Neill² argues that we can derive a requirement not to coerce, and a requirement not to deceive from the idea of principled autonomy: “an agent who adopts a principle of coercion must also will some effective means of coercion (violence, intimidation, whatever else might work). So an agent who (hypothetically) wills a principle of coercion as a universal law must also (hypothetically) will that everybody use some effective means of coercion. However, since there will be at least some coercive action in any world where all are committed to a principle of coercion, at least some persons would then be unable to adopt a principle of coercion because their capacities for action would be destroyed or undermined by others’ coercive action ... Coercion is necessarily a minority pastime, and universal coercion cannot be willed without internal contradiction” (pp 86–7).

There is an ambiguity in this argument. Is the thing that makes coercion morally bad supposed to be the fact that coercion destroys other persons’ capacities for action, or the fact that coercion cannot be universalised? If the thing that is morally bad about coercion is that it destroys other persons’ capacities for action, then it looks like principled autonomy is not playing any real role here, and the real work is being done by the claim that it is wrong to undermine others’ capacities for action. (In this case we might ask whether the value in the background is in fact an idea of respect for persons.)

But, if the moral problem with coercion is supposed to be the fact that it cannot be universalised, then the argument faces the following two problems. First, it seems wrong to claim that the mere fact that something cannot be willed universally without contradiction shows that doing that thing is wrong. For example, you cannot universally will the maxim of leaving work an hour early to beat the traffic without contradiction: for if everyone acted on this maxim, the rush hour would merely start an hour early, and you would still be stuck in the traffic. However, leaving work an hour early to beat the traffic is not immoral. So even if universal coercion could not be willed without contradiction, this would not be sufficient to explain why it is immoral.

Second, it is untrue to claim either that coercion is necessarily a minority pastime or that universal coercion cannot be willed without contradiction. O’Neill’s argument seems to presuppose that the relationship of coercion is transitive: that if A can coerce B, and B can coerce C, then A will be able to coerce C and hence C will be unable to adopt a maxim of coercion, as he will be unable to coerce A or B. But the relationship of coercion is not transitive, and so there is no contradiction in a society in which everyone is able to coerce someone else. For example, suppose A is the CEO of a company, B the line manager and C the underling. While the CEO might be able to coerce the line manager, and the line manager may coerce the underling, the underling may still be able to coerce the CEO, because, say of some secret he knows about the CEO.

So even assuming that O’Neill is right, and that it is appropriate to have a less grandiose conception of informed consent that does not appeal to respect for autonomous choices, it is wrong to think that principled autonomy can justify the requirement for informed consent.

HOW WE SHOULD THINK ABOUT RESPECT FOR AUTONOMY AND INFORMED CONSENT

O’Neill’s argument shows at the very least that there is a grave need for greater clarity in what we mean by respect for autonomy, and why we think it is important. Without this, our thinking will be confused and contradictory. So how should we proceed? Assuming that each of the conceptions of respect for autonomy that we started from (respect for persons in virtue of their capacity for autonomy, respect for autonomous choices and respect for decisional privacy) have some legitimate role to play in our ethical thinking, the challenge is to find a way of taking each into account in the appropriate way. I will sketch one such account, rooted in the work of the political philosopher John Rawls, which allows us to reconcile the new insights that O’Neill has brought to the debate about respect for autonomy and informed consent with some of bioethicists’ more traditional concerns.

I suggest that we take respect for persons in virtue of their capacity for autonomy to be the most fundamental sense of respect for autonomy, and to rethink our conceptions of respecting autonomous choices and of informed consent in terms of this value. Rawls⁹ argues that human beings are worthy of respect in virtue of what he calls the two moral powers—namely, the capacity for a conception of the good and the capacity for a sense of justice. All beings who have the two moral powers are moral equals, and must be treated with equal respect.⁹

We should make sense of respect for autonomous choices and of informed consent within this broader framework. Those who have the two moral powers are in Rawls’¹⁰ words, “self-authenticating sources of valid claims” (p 23). That is, just in virtue of having the two moral powers, a person has a right to make claims on others and to have their views taken seriously: so a requirement to respect the choices of persons follows from the more basic respect. But this requirement is circumscribed by the more fundamental requirement to respect each person as an equal in virtue of their two moral powers. Hence, there is no requirement to respect choices or evaluative perspectives that are incompatible with equal respect for all, and indeed, absent special circumstances, there is a requirement to actively combat such choices and values.

It is important to notice that respecting autonomous choices entails different duties in different normative contexts: in Elizabeth Anderson’s words, “to respect a customer is to respect her privacy by not probing more deeply into her reasons for wanting a commodity than is required to satisfy her want. The seller does not question her tastes. But to respect a fellow citizen is to take her reasons for advocating a position seriously. It is to consult her judgment about political matters, to respond to it in a public forum, and to accept it if one finds her judgment superior to others”.¹¹ It follows that before we can think about the duties that follow from a requirement to respect autonomous choices, we need to work out what norms would appropriately structure choices of this type.

Informed consent procedures will be an appropriate way of respecting autonomous choices only in certain circumstances. Informed consent procedures give the person from whom consent is sought a right of veto over something being done to them, which they are allowed to exercise for arbitrary reasons. One important justification for thinking that many kinds of choices should be left up to the choice of the individual in this way is the sheer intractable variety of views of the most worthwhile life for a human being. As Rawls¹² argues, there are strong reasons to think that, even among conscientious and reasonable persons given unlimited time to discuss, we would still not find a consensus on how a human being should live. Moreover, any attempt to enforce a state-sanctioned conception of good life will surely have unacceptable results, as was evidenced in the Reformation and its aftermath (Rawls,¹⁰ p 34). The liberal solution to these difficulties has always been to treat a large swathe of choices about how the individual wishes to live his or her life as private, thus allowing each person freedom to pursue their own conception of the good without forcing them to justify themselves to others, where such choices do not impinge unfairly on the interests of others.

Most medical decisions that patients make fall squarely within this territory staked out by the liberal principle of allowing each person to pursue their own conception of the good, but not all do. In particular, public health cases can raise a problem, where parents refuse newborn screening on behalf of their children (thus imposing an increased risk of harm on their child)¹³ or where someone refuses treatment for a serious infectious disease (thus putting other people in danger), and in such cases there is a case (though often a defeasible one) for not allowing the person the right to refuse, which informed consent implies.

Informed consent procedures, on this account, are justified by this liberal idea of treating decisions about conceptions of the good as private. Their justification is not that the decisions that people make as a result of informed consent procedures will be autonomous in such a way as to render these decisions intrinsically worthy of respect, but rather that it would be wrong and counterproductive in nature to attempt to enforce a particular conception of the good, and the best way to avoid doing so is to allow each person the privacy to make decisions in line with their own values. If this rough sketch is correct, we have the beginnings of a theory that will allow us to defend the role of respect for autonomy and informed consent in medical ethics, while also allowing us to draw any such limits to the use of informed consent procedures that we think, on balance, to be morally required by our commitments to public health measures.