Rational Autonomy and Decision-Making Capacity (2024)

In his judgement on a famous case concerning an adult refusal of treatment, Lord Donaldson of Lymington made the following observation:

An adult patient who … suffers from no mental incapacity has an absolute right to choose whether to consent to medical treatment … This right of choice is not limited to decisions which others might regard as sensible. It exists notwithstanding that the reasons for making the choice are rational, irrational, unknown or even non-existent.¹

This oft-cited judgement champions patient autonomy over medical paternalism; patients with decision-making capacity are afforded the right to make ‘unwise’ decisions with regards to their own health.

There are two notable features of this judgement. First, it implicitly acknowledges that the ‘right of choice’ is only afforded to those who do not suffer from mental incapacity. If mental capacity is closely related to considerations of autonomy (a claim I shall further defend below), then this suggests that the extent to which an individual is able to make an autonomous treatment decision has a considerable bearing on whether that decision should be respected. Second, this judgement also seems contrary to one of the central claims that I have advanced in this book, namely, that autonomous decision-making requires deciding on the basis of what one believes one has reason to do. The Donaldson judgement implicitly objects to this sort of account on the basis of an anti-paternalist concern: the worry that rationalist conceptions of autonomy will allow physicians to ignore a patient’s wishes if they run contrary to medical opinion, on the basis that the decision is not rational and therefore not autonomous. According to this objection, rationalist autonomy pays mere lip service to the idea of individual self-government, and in fact simply amounts to indirect paternalism.

Part of my aim in this chapter is to respond to this anti-paternalist objection. I shall argue that we should reject the claim that a rationalist conception of autonomy (and a fortiori decision-making capacity) must have this sort of substantive connotation. In the next section, I shall begin by providing a brief general overview of capacity and competence, and outlining two prominent accounts. I shall then introduce and critique a prominent view of the relationship between decision-making capacity and risk. In the second half of the chapter, I shall delineate and respond to two different versions of the anti-paternalist objection to a rationalist conception of decision-making capacity.

1. Competence, Capacity, and Competing Values in Their Assessment

The terms ‘competence’ and ‘capacity’ are sometimes used interchangeably in the bioethical literature. This is perhaps understandable, since in common usage, both concepts are used to broadly denote the ability to perform a certain task. For instance, outside of the bioethical context, we might describe someone as a competent driver if they are able to drive a car well; similarly, we might say that an Olympic sprinter has the capacity to run 100 m in under ten seconds.

However, it is important to acknowledge that the two terms can be used to mean somewhat different things in bioethics. For instance, in England and Wales, the concept of capacity is typically treated as a legal concept that is defined by the specific criteria set out in the Mental Capacity Act 2005. In contrast, ‘competence’ is understood to refer to a clinical concept, which may take into account a broader set of considerations than the legal conception of capacity.² Somewhat confusingly though, in the US context, the meanings of these terms are reversed. For instance, Beauchamp and Childress observe:

Several commentators distinguish judgments of capacity from judgments of competence on the grounds that health professionals assess capacity and incapacity, whereas courts determine incompetence.³

This disagreement on the use of terminology is unfortunate. In the absence of any other convincing justification for adopting one use over the other, I shall follow the use employed in the English and Welsh context, which treats decision-making capacity rather than competence as the operative concept in the legal domain (of which I shall say more below).

Whilst the distinction between competence and decision-making capacity thus has some importance, it is important not to overstate this difference. First, as I shall explain, there is often a considerable degree of overlap between the two. Second, whilst only the courts have the authority to determine whether patients have decision-making capacity, they will typically defer to professional judgements. As such, the practical implications of the clinician’s assessment of competence will typically be similar to a legal determination of decision-making capacity.⁴ That said, the two are separate, and judges do not always follow the professional assessment.⁵ Whilst acknowledging this distinction between competence and decision-making capacity, I shall phrase the remainder of my discussion in terms of the latter (which I shall abbreviate to ‘DMC’) alone. However, my claims should also be understood to extend to competence unless explicitly stated otherwise.

DMC can be understood as a range property in that it is a binary property that does not itself admit of degrees, even though we assign it on the basis of an individual’s possession of certain abilities that do admit of degree.⁶ As such, in considering whether a patient has DMC we are asking whether they have the requisite degree of the relevant abilities that are necessary to perform a certain task. The task in question in most discussions of DMC in the medical context is that of providing valid informed consent to (or refusal of) medical treatment.⁷

In understanding the relevant task for DMC in this manner, I am adopting a straightforward understanding of the relationship between DMC and local autonomy.⁸ The reason that we limit the right to consent to or refuse medical treatment to those who have DMC, is that these individuals alone have the capacity to make an autonomous decision about this matter. This is important in cases of treatment refusal because only if a patient has DMC to refuse treatment will there be an autonomy-based justification for omitting to provide a treatment that could serve to outweigh considerations of beneficence that speak in favour of providing it. In the case of consent to treatment, DMC is significant in so far as we should only recognize the normative authority of waivers of rights that would otherwise preclude the permissibility of treatment, if the decision to waive the right (by so consenting) was autonomous. We may note that a significant virtue of this approach to understanding the relationship between autonomy and DMC is that it allows for a straightforward explanation of what the conditions of DMC are, and why DMC matters morally.

Of course, in light of the differences between the institutional and non-institutional senses of informed consent, we should also acknowledge that what it means to have DMC to give informed consent in the institutional sense may differ from what it means to have DMC to give informed consent in the sense of an autonomous authorization. In any case, though, one of the primary aims of an account of DMC is to outline the causally necessary conditions for an individual to make a decision to consent in accordance with the constitutive conditions of valid consent. If one holds that the conditions of autonomous decision-making map on to the requirements of informed consent, then there will be a close connection between the concepts of DMC, informed consent, and autonomy.

The fact that DMC is treated as a range property implies that in order to assess whether a patient has it, we must set a threshold level for the abilities associated with providing valid consent, such that an individual qualifies as having DMC to make a particular decision once they have passed that threshold for all the necessary abilities. Naturally, this raises the question of where we should set these thresholds. It is possible to ask this question in an abstract idealized sense, where the only relevant consideration is whether a particular ability is necessary for providing either an autonomous authorization or institutionally valid consent, depending on the sense of consent that we mean to invoke. However, this question is most typically asked in non-ideal contexts. An important implication of this is that non-ideal theories of DMC have a further aim of resolving a conflict between competing moral values that arise in light of the epistemic obstacles that arise in non-ideal contexts. To understand why this so, it is important to be clear about the moral values in question.

One of the most significant reasons underlying the claim that patients should be allowed to make their own treatment decision is that we attribute significant value to personal autonomy: following a long liberal tradition, it is widely held that individuals with DMC should be free to decide to act in ways that are not conducive to their well-being.⁹ Indeed, we typically afford such individuals what has been termed ‘legal capacity’ in the medical context, in that we afford these individuals certain rights (and responsibilities), including the right to refuse beneficial medical treatment.¹⁰ However, this right is often understood to be conditional on DMC. As I wrote above, one justification for limiting this right to patients with DMC is that only in such cases will there be an autonomy-based justification of omitting to provide a treatment that could serve to outweigh considerations of beneficence. Moreover, we might claim that the power to waive certain claims is also conditional on DMC, in so far as we should only recognize the normative authority of autonomous decisions to waive one’s rights.

Although this link between DMC and legal capacity is widely endorsed, it is by no means universal; it has recently been challenged by the Convention on the Rights of Persons with Disabilities (CRPD).¹¹ Whether we should accept the link between legal capacity and DMC, or agree with the CRPD that legal capacity should be afforded on other bases is beyond the scope of my discussion here. I shall confine myself to the narrower question of the relationship between autonomy and DMC. However, my discussion concerning the role of autonomy and well-being in this chapter lends some indirect support to maintaining a relationship between DMC and legal capacity.

Returning to the role of the values of autonomy and well-being in the context of DMC, we may note that an ideal system of assessing DMC would be one that identified all and only those people who are able to provide valid consent as having DMC. However, there are a number of obstacles to employing such an ideal system of assessing DMC. One reason for this is that there is considerable disagreement about the nature of autonomy; this much should be clear from the first half of the book. As such, there is bound to be controversy in determining the abilities that our assessment of DMC should be aiming to identify. The second and perhaps more significant barrier though is that many of the abilities that we might plausibly agree are necessary for providing valid consent are mental abilities that do not admit of straightforward external assessment. We are thus likely to make mistakes in our assessments of these abilities with the relatively crude tools at our disposal.

In the light of these epistemic barriers and the errors that they make likely, our decision about where to set the relevant threshold for DMC in the non-ideal context must unavoidably make a moral judgement about how to balance the value of patient autonomy, against the moral reasons grounded in the duties of beneficence and nonmaleficence. The benefit of setting a low threshold for DMC is that our assessment is likely to be more sensitive, in the sense that it will more reliably identify true positive cases; that is, on a low threshold approach, individuals who are able to make autonomous decisions will typically qualify as having DMC.

However, the cost of employing a low threshold is that our assessment is thereby unlikely to be particularly specific. Employing a low threshold increases the chance that our test of DMC will lead to false positive assessments; there is a higher chance that individuals will qualify as having the DMC to make a treatment decision, when they are not in fact capable of making the decision autonomously. The cost of such false positive assessments is that allowing such patients to make decisions that put them at risk of harm cannot be justified by an appeal to the value of their autonomy; we may be understood as harming them by allowing them to make their own decisions. Low threshold approaches to the assessment of DMC thus place greater emphasis on defending patients’ decision-making authority, at the expense of protecting non-autonomous patients from harm.

In contrast, the benefit of setting a high threshold for DMC is that our assessments will be much more specific, in the sense that they will more reliably pick out true negative cases; that is, on a high threshold approach, individuals who are not able to make autonomous decisions will typically not qualify as having DMC. The cost of employing a high threshold is that our assessment is thereby unlikely to be particularly sensitive, in the sense that employing a high threshold increases the chance that our test of DMC will lead to false negative assessments; on such an approach, there is a greater chance that individuals will qualify as lacking DMC, when they are in fact capable of making a decision in an autonomous fashion. The moral cost of such false negatives is that prohibiting a person from making an autonomous decision about their treatment when they are capable of doing so runs contrary to the liberal tradition that affords greater weight to the duty to respect autonomy than to the duty of beneficence.¹² High threshold approaches to the assessment of DMC thus place greater emphasis on protecting non-autonomous patients from harm, at the expense of jeopardizing the decision-making authority of some patients.

With this discussion in mind, we may observe that in outlining criteria of DMC in non-idealized contexts, we have to answer three questions. First, is the criterion plausibly a necessary condition of autonomous decision-making or providing consent in its institutional sense? Of course, this question is also relevant when we are thinking about DMC in ideal contexts. Second, do we have the methods to reliably assess the abilities in question? Third, what are the implications of our criteria for the balance that we are aiming to strike between the competing moral reasons we face in making assessments of DMC in non-ideal contexts? Over the course of this chapter, I shall outline different permutations of the anti-paternalist objection to rationalist theories that place different emphases on these questions.

2. Two Cognitivist Accounts of DMC

Grisso and Appelbaum developed a particularly influential account of DMC which is largely echoed in the Mental Capacity Act in England and Wales. According to Grisso and Appelbaum, DMC requires the ability to:

Communicate a choice
Understand relevant information
Appreciate the situation and its consequences and
Manipulate information rationally¹³

These conditions bear a striking resemblance to those that are adopted in the Mental Capacity Act 2005 (henceforth MCA). A necessary (but not sufficient condition) for an individual’s lacking DMC on the MCA is that she lacks any of the following abilities, as outlined in section 3(1) of the Act:

(a): The ability to understand the information relevant to the decision.
(b): The ability to retain the information for long enough to be able to make a decision.
(c): The ability to use or weigh that information as part of the process of making the decision.
(d): The ability to communicate their decision.¹⁴

Grisso and Appelbaum’s theory and the approach evidenced in the MCA are similar in that they both emphasize cognitive capacities. However, it is worth highlighting three striking differences. First, the MCA does not adopt a criterion relating to the ability to appreciate information; appreciation goes beyond mere understanding of material information in requiring that individuals are cognizant of the fact that material information applies to them and their situation.¹⁵ Second, although the MCA adverts to the need to weigh and use information, it makes no reference to the need to do so rationally, unlike the counterpart criterion in Grisso and Appelbaum’s approach. Third, the MCA supplements Grisso and Appelbaum’s functional approach with a further diagnostic criterion. According to the MCA, the fact that a patient lacks one of the above abilities is not sufficient for establishing that she lacks DMC. For that to be the case, the patient’s lacking the ability in question must also be attributable to ‘an impairment of, or a disturbance in the functioning of, the mind or brain’.¹⁶ The MCA thus incorporates both a functional and diagnostic test of DMC.¹⁷

I shall discuss whether these differences are philosophically warranted over the course of this chapter. Here though, we may note that the cognitive approach endorsed by both accounts is broadly compatible with the procedural analysis of autonomy that I have offered so far in this book. Indeed, section 1(4) of the MCA explicitly states that: ‘A person is not to be treated as unable to make a decision merely because he makes an unwise decision’. In this regard, it echoes both Lord Donaldson’s remarks quoted at the beginning of this chapter, as well as Buchanan and Brock’s claim that standards of DMC should focus ‘… not on the content of the patient’s decision but on the process of the reasoning that leads up to that decision’.¹⁸

The abilities outlined above are also all plausible candidates for abilities that are causally necessary for autonomous decision-making on my account. As I argued in the previous chapter, patients must be able to understand certain information about their decision in order to be autonomous with respect to it. Furthermore, being able to retain information is central to one’s ability to make a decision on the basis of that information, and a criterion referring to the practical element of communication is also congruous with the practical orientation of my account of autonomy. Finally, my account also lends theoretical support to the criteria of appreciation and weighing and using information rationally; I shall make the case for this claim in section 3 where I shall also critically engage more broadly with these cognitivist approaches. Here though, I shall conclude my discussion of these two prominent approaches by highlighting two further general and widely accepted features of DMC.

The first is that DMC is typically understood to be contextually dependent.¹⁹ Different local decisions will require different degrees of aptitude in the particular abilities relevant to DMC. Accordingly, although a patient may lack DMC to make certain sorts of decisions, this does not entail that they lack DMC to make any decisions for themselves. For example, whilst an agent may be able to understand material information pertaining to a decision about relatively simple treatments, such as whether she ought to have surgery on a broken bone, she may not be able to understand material information pertaining to more complex treatment options which could lead to various possible outcomes, such as in the treatment of cancer. Similarly, the fact that an individual is globally autonomous does not entail that they will make a locally autonomous decision.

The threshold level of DMC required to make a certain decision will depend upon the complexity of the information that is material to the patient. This is particularly relevant when we consider DMC in children. The majority of the provisions in the MCA apply to children who are 16 and over;²⁰ once a person has reached this age, they are thus presumed to have DMC unless proven otherwise.

It would be a mistake to make the overly general claim that children simply lack the capacity for decisional autonomy, even though the law denies children the authority to make certain decisions for themselves. It is clearly the case that some children below 16 can hold and exercise the sorts of abilities discussed above with respect to at least some decisions. Indeed this is recognized in the legal concept of Gillick competence, according to which a child ‘… has a right to make their own decisions when he reaches a sufficient understanding and intelligence to be capable of making up his own mind on the matter requiring decision’.²¹ Accordingly, in England and Wales, children who are under 16 but who are ‘Gillick competent’ can provide valid consent to some medical procedures; however, their refusal to consent to treatment that is deemed to be in their best interests may be overridden by someone with parental responsibility, by virtue of the Family Law Reform Act.²² Notably, the latter is also true of children under 18.

This coheres neatly with a rationalist approach, since it seems plausible to claim that children are perfectly capable of recognizing and appreciating certain kinds of reasons. To give a simple example, consider the reasons grounded by an individual’s hedonic likings and dislikings. Even very young children can recognize that they have stronger reasons to choose, for example, an ice-cream flavour that they have enjoyed previously (say chocolate) over one that they have disliked previously (say coffee). Ceteris paribus, it seems plausible to say that we ought to allow even a young child to make a choice between alternative ice-cream flavours, because doing so simply requires weighing two of the same kind of reasons (concerning the child’s hedonic liking of a certain taste), reasons to which the child is well-equipped to respond. The child can thus make an autonomous choice in this circ*mscribed choice domain.

However, there are some reasons that children are not well-equipped to recognize and appreciate—even if it seems plausible to allow a five-year-old a degree of autonomy about which flavour ice-cream to have, we would be reluctant to allow her to make her own decisions about how often ice-cream should feature in her diet. The reason for this is that whilst a child of this age is able to recognize her reasons to eat ice-cream (namely, that she enjoys the taste), she is less able to appreciate and weigh facts that give her reasons to refrain from eating ice-cream, namely, that frequent consumption of ice-cream would be bad for her health. However, as a child’s general cognitive capacities develop, so too will their ability to recognize and weigh more complex kinds of reason-giving facts.

A second notable general feature of capacity is that it has often been the case that the gravity of a decision has been understood to influence the relevant threshold of DMC that is necessary for making that decision: The more serious the consequences of a decision, the higher the threshold for DMC.²³ For instance, in the case of Re T, the Court of Appeal stated:

What matters is that the doctors should consider whether at that time he had a capacity which was commensurate with the gravity of the decision. The more serious the decision, the greater the capacity required.²⁴

Call this the ‘sliding-scale view’ of DMC.²⁵

It should be noted that the sliding-scale view is often understood to operate only within certain thresholds. If an individual exhibits the abilities that contribute to DMC to a particularly high degree, then she should be understood to have the DMC to make her own medical decisions, no matter how detrimental the consequences of her decision are for her individual well-being. This reflects the liberal view that considerations of autonomy should trump those of beneficence.

More generally though, the sliding-scale view has somewhat puzzling implications. For instance, Culver and Gert note that in some situations, a choice to refuse treatment may have very serious consequences, whilst consenting to treatment may not. They note that the sliding-scale view thus has the somewhat puzzling implication that a patient might have sufficient DMC to choose option B (because it concerns a low-risk medical intervention and thus implies a relatively lower threshold for DMC), but lack capacity to choose option A (because the gravity of refusing consent implies a higher standard of DMC on the sliding-scale view). However, in making this choice between A and B, she has to understand and weigh the same information about her alternative options.²⁶

Partly on the basis of this implication, Culver and Gert claim that the sliding-scale view conflates the distinct concepts of DMC and rationality, conceived in a substantive sense; patients only qualify as having DMC if they make decisions that are rational in the view of the medical profession.²⁷ One reason that this is problematic is that it unhelpfully conflates two separate concepts. Whilst this is Culver and Gert’s particular worry,²⁸ it might be argued that this conflation is particularly concerning for those who endorse the anti-paternalist objection, because of the paternalistic connotations of conflating DMC with rationality (in this substantive sense). In the next section, I shall outline some ways in which the sliding-scale view might be defended, and consider whether these defences might be used to deflect this charge of indirect paternalism. I shall argue that the prominent justifications offered for standard interpretations of the view fail in this regard, but that we should not wholly disregard a revised version of the sliding-scale view. In section 4, I shall argue that, contrary to Culver and Gert, the concepts of DMC and rationality do in fact overlap in some sense, but not in the manner that undergirds this particular criticism of the sliding-scale view.

3. Sliding-Scale, Risk, and Value

It might be argued that the sliding-scale view can be justified in both ideal and non-ideal contexts by considerations of autonomy alone. One such justification might claim that there is a positive linear relationship between the degree of risk that a decision concerns, and the degree of the requisite abilities it takes to make that decision autonomously.²⁹ Call this ‘the linear justification’ of the sliding-scale view.

The linear justification may seem appealing, because we can think of some examples that fit this picture of the relationship between risk and autonomous decision-making. The reason for this is that in some cases, decisions with more serious consequences can involve more complex information, and the weighing of a greater number of considerations and options. To illustrate, trivial decisions, such as deciding what to eat for lunch, typically do not require understanding the sort of complex information that might be involved in decision-making regarding a range of different cancer treatments.

Yet, this justification of the sliding-scale view is overly simplistic. There is not always a straightforward linear relationship of the sort it appeals to between appropriate requirements of DMC and the risks the decision concerns. In some cases, decisions can plausibly have grave consequences without necessarily requiring the individual to understand highly complex information, or to compute a large number of options. Choosing to refuse a blood transfusion when one is bleeding profusely has grave consequences, but it is not particularly difficult to understand why that might be the case. In stark contrast, one can also think of extremely low-risk decisions that might require understanding much more complex information and weighing of options; for example, those involved in playing a strategic board game.

4. Rationalist DMC in the Ideal Context, and the Anti-Paternalist Objection

In the previous section, I considered the extent to which the widely accepted sliding-scale view of DMC has paternalistic connotations. I now want to consider whether similar charges could be raised against the implications that my rationalist conception of autonomy has for our understanding of DMC. In this section, I consider the implications that the theory has in the idealized context, where the only question we have to consider is whether the rationality criterion sets out a plausible necessary condition of DMC. I shall consider its implications in the non-ideal context in the following section. To begin this discussion, I shall explain the implications that my theory has for the discrepancies between the two accounts of DMC that I outlined above.

The first discrepancy was that Grisso and Appelbaum’s approach incorporates a criterion of ‘appreciation’ that is absent in the MCA test. Such a criterion is a highly plausible addition to an adequate set of criteria for DMC by the lights of the theory that I have outlined in this book.³⁸ Appreciation can clearly be a necessary ability for autonomous decision-making, since the failure to appreciate information in this way can clearly be inimical to an individual’s holding decisionally necessary beliefs. For instance, if a patient fails to believe that they are seriously ill and will die without medical intervention when that is in fact the case, then they lack a belief that is crucial for making an autonomous decision in that context; in such an epistemic situation, they will be unable to perceive an important set of reasons, namely those in favour of undergoing medical treatment.³⁹

In accordance with my earlier discussions in this book, we may also note that in addition to the ability to appreciate information in this sense, the ability to meet minimal standards of theoretical rationality will also be necessary for individuals to avoid some of the false beliefs that are inimical to their decision-making, and to use and weigh information in the manner that connotes autonomous decision-making.

The criterion of appreciation also represents a way in which the patient’s evaluations feature in DMC criteria, since appreciation involves ‘assigning values to information’.⁴⁰ This feature is also relevant to the second discrepancy between the two accounts, namely that Grisso and Appelbaum make explicit reference to the need to manipulate information rationally in one’s deliberative process, whilst the MCA does not. Naturally, my view lends support to the former approach, and provides a theoretical basis for adding further content to this requirement.

Grisso and Appelbaum are predominantly concerned with the theoretical rationality that decision-making capacity requires, suggesting that this criterion requires that patients are able to ‘… reach conclusions that are consistent with their starting premises’.⁴¹ Whilst I agree that this is an important part of DMC, I claim that we ought to understand this ability (and the ability to ‘use and weigh’ information to which the MCA refers) in a manner that reflects the rationality condition defended in the previous chapter. To have the ability to ‘weigh and use’ information in one’s decision-making process is to have the ability to make a decision in accordance with what one values, that is, with one’s personally authorized preferences. To weigh information in the manner that autonomy requires is to consider the bearing that material information has on ends that agents value (and their pursuit thereof), and to consider the strength of the relevant competing reasons. To do this, a patient must be able to recognize that they have self-interested reasons to want certain things, and they must be able to use the information provided to them to decide what course of action to pursue, in the light of both descriptive facts and their own values. In short, DMC should incorporate considerations of practical, as well as theoretical rationality.

To further clarify the importance of values to DMC criteria, consider the following case:

Apathetic Andrea: Andrea suffers from clinical depression. Her physician explains to her that there are a number of treatment options available (including various anti-depressants, and forms of psychiatric counselling), and provides her with extensive information about each option and their possible outcomes. Andrea understands this information, retains it, and can compare how medically effective each option is against the other. However, Andrea is pathologically apathetic, and does not care at all what happens to her; she is convinced that everyone despises her, despite clear evidence to the contrary. Nothing can persuade her that her life is in any way worthwhile. Although she considers the information about each of her treatment options, she believes that this information is simply irrelevant. She simply does not care.⁴²

In evidencing the ability to compare the relative effectiveness of each intervention, Andrea can plausibly be described as having the ability to ‘weigh information’. However, if Andrea were to make a treatment choice in this scenario, it seems problematic to claim that her decision was autonomous, despite the fact that she meets the MCA criteria. The reason for this, I suggest, is that Andrea is unable to engage in rational deliberation about what to do, because she is unable to regard herself as having self-interested reasons to pursue her own well-being.⁴³ We might say that she is, in some sense, ‘value-impaired’.⁴⁴

Furthermore, Andrea’s apathy is grounded by a theoretically irrational belief in her own lack of worth; she has what we might describe as an evaluative delusion.⁴⁵ To repeat claims I made in Chapter 2, this is not a claim about the truth or falsity of the content of the belief. Rather, there are grounds for claiming that Andrea holds this evaluative belief in a theoretically irrational sense, in so far as she holds it unshakeably in a manner that is immune to evidence. It is a form of an evaluative delusion, which, as I argued in previous chapters, can serve to undermine decisional autonomy.⁴⁶

Grisso and Appelbaum’s approach might partly capture Andrea’s lack of DMC by adverting to the necessity of appreciation to DMC, and the importance of being able to assign value to information in the process of appreciation. However, as I have explored previously in the book, agents who are not value-impaired in this way can still fail to make autonomous decisions, because of the role their values are playing in their decision-making process. In some cases, an individual’s decisions may reflect the force of a motivating desire that the individual does not rationally endorse. In others, we may be concerned that the values that ground the agent’s decision are not authentic to her, and are thus not a suitable ground for her autonomy.⁴⁷ The problem in such cases is not that agents fail to assign value to information; it is that the values they assign do not really reflect what they want.

Such cases represent arguably the most challenging cases for any procedural theory of autonomy, as well as raising questions about how we should delimit the scope of the clinical category of evaluative delusions. This deserves its own discussion, which I shall postpone until the following chapter. Here though, I want to begin considering the anti-paternalist objection to incorporating considerations of theoretical and practical rationality into criteria of DMC even in ideal contexts.

The general thrust of this objection is that incorporating such considerations would render standards of DMC too demanding, and would lead to physicians being able to overrule patient choice. The anti-paternalist objection is arguably the most pressing objection facing the rationalist account to autonomy I have defended in this book. I have noted that it is implicitly incorporated within Lord Donaldson’s judgement quoted at the beginning of the chapter. However, the objection as I have just phrased it captures two distinct but related concerns. Here, I shall consider the objection in its purest form, as an objection to rationalist criteria as elitist even in ideal contexts. In the next section, I shall consider an epistemic version of the objection that can be raised against rationalist criteria in non-ideal contexts, according to which such criteria make it more likely that physicians will be able to overrule patient choice because of limits to our ability to accurately identify rational decision-making.

The elitist version of the anti-paternalist objection has been raised explicitly by a number of philosophers. For instance, in defending the standard account, Faden and Beauchamp write:

If conscious, reflective identification with one’s motivation were made a necessary condition of autonomous action, a great many intentional, understood, uncontrolled actions that are autonomous in our theory would be rendered non-autonomous.⁴⁸

Nelson et al. go further, arguing that to claim that authenticity of any stripe is a condition of voluntariness is ‘… both conceptually unsatisfactory and morally dangerous’.⁴⁹

The first thing to acknowledge about this objection is that it is understood to pertain to all of the procedural theories of reflective autonomy that I surveyed in the first two chapters. This observation alone might seem to render the objection implausible. To see why, reconsider Frankfurt’s view of autonomy; on Frankfurt’s view, autonomy requires that one identify with one’s first-order motivating desire with a second-order volition. Crucially for Frankfurt, human beings can be distinguished from other creatures by virtue of the fact that they alone are able to form second-order desires.⁵⁰ Accordingly, far from being elitist, Frankfurt might claim that the standards set in his theory of reflective autonomy are simply the standards for how we assess personhood.

However, it might be claimed that rationalist theories of autonomy of the sort that I have defended are particularly vulnerable to this objection.⁵¹ For instance, John Christman writes that:

… the property of autonomy must not collapse into the property of ‘reasonable person’, where the idea of being self-governing is indistinguishable from the idea of being, simply, smart.⁵²

There are several things to say in response to this objection.⁵³ First, phrased in this way, the objection appears to assimilate rationality and ‘smartness’; yet, one need not be ‘smart’ in order to be rational. On the theory that I have developed here, agents need only be able to pursue the outcome of their desire on the basis of their belief that the outcome is something that they have reason to pursue. It is not at all clear why this should be intellectually demanding; to suppose otherwise is to conflate the separate concepts of rationality and intelligence.⁵⁴

One might instead interpret Christman’s concern here to be that a rationalist approach to autonomy entails that it will be reserved only for those who think through their choices. I struggle to see why we should find it problematic to claim that an individual will only be able to make a locally autonomous choice if they think through it in basic ways. Indeed, this is just why we reserve the right to make one’s own medical decisions for those who qualify as having decision-making capacity. In my view, the most plausible way of cashing out this concern is that we have reasons to be sceptical of a theory of autonomy that entails that individuals will only qualify as being globally autonomous if they think through every single one of their choices and make them in a maximally autonomous way. Yet a rationalist account does not require this—as long as one does not understand global autonomy as the aggregative accumulation of various locally autonomous choices (a suggestion I rejected in the introduction to the book), then it is quite compatible with an individual’s making a considerable number of choices over their lives that are not locally autonomous.

That said, as I explored in Chapters 1 and 2, on a rationalist account, autonomy does require rational reflection in a way that other accounts do not, even if such reflection can be unconscious or dispositionally produced. I have suggested in previous chapters that accounts that do not appeal to actual reflection of this sort fail to accommodate paradigm cases of individuals who lack autonomy (i.e. the standard account), and that accounts appealing to hypothetical reflection (such as Christman’s account) also face other challenges concerning their operationalization in bioethical contexts. Ultimately though, even if the rationalist has to bite the bullet and accept that autonomy is more challenging on his account than it is on others, this does not entail that it is beyond the capacity of most human beings. On the contrary, like Frankfurt, advocates of rationalist theories can suggest that having the abilities associated with rational decision-making is just part of what it is to be a person. Indeed, in the very first sentence of On What Matters, Parfit claims that humans are ‘… the type of animal that can both understand and respond to reasons’.⁵⁵ Similarly, in his defence against a similar objection, Joseph Raz points out that ‘[t]o want to be rational is to want to be a person’.⁵⁶

Rationality conditions of decisional autonomy do not entail that autonomy-based protections will only be afforded to those who think through their choice with intellectual precision and accuracy, nor does it unduly preclude individuals from having decision-making authority. However, it might be argued that the rationalist account is elitist in a different sense, in that it places too much emphasis on cognitive elements of decision-making capacity and fails to acknowledge the importance of affective attitudes and emotional experience to DMC.⁵⁷ However, the rationalist account can be understood to incorporate affective elements of DMC in so far as many of our affective attitudes and emotional experiences can give rise to values, and ground certain sorts of reasons. Consider for example the experience of love; although the experience of love is not itself typically the output of rational deliberation (we ‘fall in’ love, rather than rationally deliberate ourselves into it), it can clearly give rise to other evaluative judgements that we come to reflectively endorse, and reasons to act in certain ways towards others. Moreover, as Nomy Arpaly has persuasively argued, emotions might plausibly be a source of reasons which may not be accessible at the time of deliberation, but which may nonetheless ground rational behaviour on a broadly coherentist approach.⁵⁸

The rationalist account does speak against unreflective emotional states grounding autonomous decision-making and action, but this seems quite plausible; a person who acts in a fit of rage and later decides that this did not reflect their evaluative judgements is not appropriately described as having acted autonomously. This suggests that we need to have a nuanced understanding of the role of emotions in autonomous agency; it is neither the case that they alone can ground autonomous decision-making, nor that their mere influence impedes it. The issue turns on whether our emotional states are connected in the right way to our evaluative judgements.⁵⁹

Similar remarks apply to relational influences on decisional autonomy. Cognitive tests of DMC have been criticized on the basis that they overlook relational (as well as emotional) influences on autonomous decision-making.⁶⁰ However, the approach to DMC that I am outlining here is quite compatible with relational influences. First, as I discussed in Chapter 5, many of the abilities that are necessary for decisional autonomy are socially mediated; accordingly, in claiming that decisional autonomy requires practical and theoretical rationality, I am implicitly accepting that the relational and social conditions that are necessary for individual rationality will also be necessary for DMC. We may also observe that relationships can be central to the content of our values. Nonetheless, whilst acknowledging these important points, it is best to maintain some conceptual distance between relational influences and DMC. The reason for this is that there are many interpersonal effects on the voluntariness of decision-making that do not adversely affect DMC, even if they undermine decisional autonomy in other ways. For example, I suggested that deceived agents and coerced agents may lack autonomy with respect to their decisions, but it can still make sense to describe them as retaining the abilities that are causally necessary for (counterfactually) making that decision autonomously. Indeed, as I explained in Chapter 5, it is only by virtue of the fact that the victim of coercion retains their rational capacities that coercion is able to dominate the victim’s will.

Perhaps part of the explanation for why rationalist theories of autonomy are deemed elitist is that critics assume that these theories are substantively rational rather than procedurally rational.⁶¹ This is one plausible way of reading Lord Donaldson’s judgement; he seems to understand rational decisions as being co-extensive with decisions that others regard as rational, or with those that accord with impersonal reasons, ranked in a certain objective way.⁶² However, on the theory that I have defended, agents may act on the basis of their beliefs about facts that provide them with either personal or impersonal self-interested reasons. This is crucial, since there is scope for considerable intersubjective variability in what agents have personal self-interested reason to do; moreover, rational agents may differ with regards to the weight that they place on different impersonal reasons. The upshot of this is that whilst we may agree with Lord Donaldson that patients should be free to make decisions that are irrational from the impersonal perspective, we should reject his claim that patients should be free to act in accordance with decisions that have no rational basis.

In a slightly different vein, Nelson et al. partly ground their criticism of authenticity-based accounts of voluntariness on the basis that we often make voluntary choices that are inauthentic. They write:

Anomalous actions sometimes arise from choices that are out of character as a result of surrounding events that are unprecedented in the actor’s experience, such as serious disease.⁶³

To illustrate further, they appeal to the following example:

A patient might request a highly invasive treatment at the end of life against his previous judgment about his best interests because he has come to a conclusion that surprises him.⁶⁴

Whether or not this is a compelling objection to the accounts of authenticity that the authors have in mind, this example is not particularly problematic for the account that I have developed here. The reason for this is that the objection assumes that authenticity must require a far greater degree of stability than is necessary. It is true that autonomy requires a degree of stability in our overall evaluative nexus; we will be unable to adequately pursue the long-term plans that undergird our global autonomy if we frequently abandon the values that provide their basis. However, this is quite compatible with the claim that our local autonomous choices can run contrary to some of the evaluative judgements that we have long held dear. Such departures from a pre-existing value can be authentic if the change is intelligible to the agent, by virtue of its coherence in the overall nexus of her other acceptances and preferences; in short, her character. However, such choices may undermine autonomy if they are not a response to the agent’s own judgements about what is good for her, but produced by other irrational drives (such as fear) that can disconnect the agent’s motivation from her evaluative judgements. The crucial question is thus not whether or not the agent can autonomously choose contrary to a previously held evaluative judgement, but rather why the agent in question has chosen contrary to that judgement.

In contrast to the terms in which the objection above has been stated, we have little reason to believe that the patient in the above case, who changes her judgement about what is in her best interests, is acting contrary to her values generally, even if she is now deciding contrary to a particularly long-standing evaluative judgement. In fact, we have good reasons to suppose that facing a serious medical condition will tend to prompt individuals to reconsider the values that undergird the reasons for their practical choices. Not only that, such agents will be carrying out this reflection at the same time as acquiring epistemic access to vital reason-giving facts about the precise nature of the situation that they now find themselves in. For instance, they may now be acutely aware of the fact that illness can drain a person of their reserves of determination.⁶⁵

In contrast, the evaluative judgements that they had previously made about such situations were made without such awareness. In Nelson et al.’s case, the patient’s understanding of the comparative strength of their reasons to avoid severe pain on the one hand, and to avoid death on the other, will naturally be sharpened and altered by being placed in a situation in which they are confronted with the reality of having to choose to act on the basis of one of these reasons. As such, the fact that the patient in question is now making a request that is in conflict with her previous evaluative judgement does not entail that it qualifies as non-autonomous on the rationalist view I have defended; it can instead be a rationally intelligible adaption to one’s radically different circ*mstances.⁶⁶

So, in phrasing their spin on the anti-paternalist objection, Faden and Beauchamp are quite right to claim that ‘many intentional, understood, uncontrolled actions that are autonomous in our theory would be rendered non-autonomous’ on a rationalist theory. But the reason for this is that many intentional, understood, and uncontrolled actions are not autonomous. The main remaining worry undergirding the elitist objection in the context of biomedical ethics is that incorporating the rationality condition I have suggested into a conception of DMC will serve to increase the number of patients who will lack DMC. I have in mind here patients such as those who suffer from conditions that render them unable to make treatment decisions in accordance with what they believe they have reason to do in light of their own evaluative judgements. Whilst such patients would lack DMC on the approach that I advocate, I do not take this to be a flaw of the theory. On the contrary, it is a flaw of the standard view that it finds such patients competent to make their treatment decisions, and regards their choices as autonomous. Whilst these patients are able to express a ‘choice’, it is one that is unconnected to what they themselves believe they have reason to do in light of their own values.

In fact, the standard theory itself comes very close to advocating a similar viewpoint in its stipulation that psychiatric disorders can represent internal forms of controlling influence that undermine autonomy. However, as I argued in previous chapters, in the absence of something like an account of authenticity, the standard account lacks a unified explanation of what it is that makes these disorders controlling in the sense that undermines decisional autonomy. As I shall discuss in the next chapter, my account allows for a far more nuanced understanding of the ways in which certain psychiatric disorders can, but need not, undermine autonomous decision-making.

Strikingly, despite these philosophical objections to rationalist criteria of DMC, there is evidence to suggest that prominent accounts of DMC (including those invoked by the courts) seem to incorporate either a rationality constraint, or something similar in order to acknowledge ways in which volitional deficiencies can undermine DMC.⁶⁷ However, as I shall begin to explain in considering a different permutation of the anti-paternalist objection grounded by epistemic considerations, some versions of this view appear to place undue emphasis on unreliable proxies for procedurally rational decision-making.

5. Rationalist DMC in Non-Ideal Contexts and the Epistemic Anti-Paternalist Objection

Dispensing with the elitist conception of the anti-paternalist objection may suffice for justifying the adoption of a rationalist approach to DMC in ideal contexts. However, in accordance with my analysis above, in the non-ideal context we must consider two further questions about the application of a rationalist conception of DMC. First, do we have reliable methods to assess the supplementary abilities that I have considered so far? With regards to appreciation and theoretical rationality, it seems that the answer to this question is ‘yes’. It is true that assessing these abilities requires going beyond the mere assessment of the individual’s ability to understand information. However, these abilities plausibly admit of empirical assessment using similar methods to those that we use to assess understanding. Indeed, clinical tests for competence such as the Macarthur Competence Assessment Tool already use semi-structured interviews to assess appreciation.⁶⁸

Whilst clinical assessment tools might plausibly assess appreciation and theoretical rationality, it is less clear that they will be able help physicians ascertain whether their patient is making their treatment decision in accordance with the requirements of practical rationality that I have outlined. It is one thing to establish that a patient can meet requirements of theoretical rationality in their deliberations in the manner that the MCA test and Grisso and Appelbaum’s approach seems to imply. It is quite another to claim that they are weighing information rationally in accordance with their evaluative judgements, and making their decision in accordance with that weighting. In turn, this raises a further question about incorporating a rationality condition into our understanding of DMC in non-idealized contexts. Given that we are likely to make errors in our assessment of this ability, what implications might this criterion have for the balance that we are aiming to strike between the competing moral reasons at stake in setting thresholds of DMC? Is it justifiable to heighten the epistemic obstacles we face in making assessments of DMC by adding requirements of practical rationality?

One particular concern we might have in this regard is that medical professionals might exploit our epistemic limitations about the practical rationality of others to unjustifiably revoke patients’ decision-making authority, in order to prioritize considerations of beneficence. The thought here is that in view of our epistemic limitations in this regard, adopting a rationalist criterion of DMC would most likely lead physicians to make judgements about the rationality of a patient’s decision based on the content of the patient’s decision, or their disease status.⁶⁹ Even if we agree that DMC should not be defined by appeal to such substantive considerations, perhaps the epistemic barriers we face in assessing rational DMC may leave us with little choice but to adopt a substantive approach to assessing capacity, inevitably increasing false negative assessments. This is in tension with both the proceduralist spirit of the MCA and the account of autonomy that I have defended, and it is precisely what Lord Donaldson was seeking to defend against in his judgement outlined at the outset of this chapter.

Indeed, there is some evidence to suggest that this sort of problem is already arising with respect to the manner in which the MCA is interpreted. Although the MCA does not make explicit reference to requirements of rationality or authenticity in its ‘use and weigh’ criterion, the manner in which the law has been interpreted in the context of refusals of treatment from patients suffering from anorexia nervosa suggests that the criterion has been understood to preclude individuals from qualifying as having DMC if their decisions are grounded by apparently ‘compulsive motivations’. For instance, the MCA code of practice suggests that patients suffering from anorexia nervosa may lack DMC, not because of any deficiency in their ability to understand material information, but rather because ‘their compulsion not to eat may be too strong for them to ignore’.⁷⁰ Furthermore, Jillian Craigie and Alisa Davies have highlighted a number of legal judgements that suggest that courts in England and Wales tend to view the desires that are symptomatic of anorexia nervosa as amounting to compulsions that are incompatible with DMC.⁷¹

The claim that compulsions undermine decisional autonomy is broadly compatible with my procedural account of autonomy, although I shall say more about this in the next chapter. In practice though, assessments of what constitutes a ‘compulsion’ in these contexts may be grounded by non-procedural considerations. As Camillia Kong has argued, the assessment of a compulsion can treat ‘compulsion’ as a thick concept;⁷² it may incorporate substantive considerations either directly or indirectly through an appeal to the patient’s diagnostic status. The concern here is that this interpretation of the MCA test coupled with our epistemic limitations threatens to unjustifiably collapse the ostensibly procedural test of capacity into a diagnostic status-based test, whereby anorexic patients are simply assumed to lack capacity because they are assumed to be subjects of compulsion in their decision-making.⁷³

I am sympathetic to Kong’s concern about this in the specific context of anorexia nervosa, and I shall consider the issue in more detail in a case discussion in the next chapter. To conclude this chapter though, I want to consider this epistemic form of the anti-paternalist objection in a more abstract sense outside of this specific context. Do our epistemic limitations give us decisive reasons not to incorporate considerations of rationality into non-ideal assessments of DMC?

If Kong’s critical analysis of the current interpretation of DMC is correct, then the flaw in this interpretation seems to lie in the fact that the courts are relying on individuals’ disease-status, and perhaps even the content of the patients’ decision, as an exhaustive and reliable proxy for procedurally rational decision-making. Before considering whether this mistake must be inevitable, it is important to note that the use of proxies to enable one to overcome epistemic barriers to accurate assessments of DMC is not problematic per se. It is quite coherent to claim that the content of an individual’s decision can provide evidential support for an assessment of DMC, whilst denying that it can provide a sufficient ground for a judgement that an individual lacks DMC. To use Colin Gavagahn’s memorable phrase the content of a decision can serve as a ‘warning flag rather than a stop sign’ in assessments of DMC.⁷⁴ Indeed, although the above discussion suggests that the MCA is not always interpreted correctly on the following point, the Act nonetheless implicitly endorses the view that substantive considerations can play a non-exhaustive role in assessments of DMC. Recall that the MCA stipulates that ‘A person is not to be treated as unable to make a decision merely because he makes an unwise decision’. As Herring notes, the use of the word ‘merely’ here suggests that the fact that a decision is unwise can factor in one’s assessment of DMC; it just cannot be the only factor.⁷⁵

The account of autonomy that I have defended can help to elucidate why this approach to evidential proxies can be justified. When a patient has made their treatment decision autonomously, they should be able to justify that decision by appeal to what they understand to be the reason-implying facts about their treatment options, and its coherence with their other evaluative judgements. In many cases, the reasoning behind a decision will be quite transparent to third parties. In some cases though, the rationale for the content of a patient’s decision may be opaque to others. If, in such cases, the content of a particular decision is contrary to what the patient has impersonal reason to do, or if it appears incongruous with other elements of the patient’s character system, that gives us reason to investigate the patient’s deliberative process in a deeper fashion. Crucially though, in such cases, the content of the decision should not serve as the end-point of an assessment of the patient’s DMC. Rather, the fact that the rationale for the decision is opaque should act as a springboard for investigating the individual’s reasons for making that decision, and also how it relates to her core preferences and acceptance.⁷⁶

Proxies, however, are problematic if they are understood to be wholly sufficient for assessments of procedural DMC, or if they are in fact unreliable ‘warning flags’ for that which we are seeking to identify, in this case, procedurally rational decision-making. However, incorporating a rationalist condition into one’s account of DMC does not entail that we must rely on unreliable proxies or indeed proxies alone.

This point has an important bearing on the broader epistemic concern that grounds this permutation of the anti-paternalist objection, namely the concern that incorporating judgements of rationality into assessments of DMC will unavoidably lead to more false negative assessments. This concern is only warranted if we accept that we lack evidential methods for accurately assessing the ability that we are attempting to capture here. Whilst one may plausibly object that we do not currently employ such methods, there is some scope for optimism about the possibility that we might accurately assess rational decision-making. First, there is currently a great deal of interest in using neuroscientific approaches to assessing the neural underpinnings of clear deficits in rational decision-making, and there have been calls to use such evidence in assessments of DMC.⁷⁷ However, this research is at an early stage, and may not be appropriate for many patients.

Yet there are other alternative methods we might adopt in tackling this epistemic barrier to assessments of practical rationality that have far greater clinical feasibility. For instance, Natalie Banner and George Szmukler’s ‘Radical Interpretation’⁷⁸ approach advocates that in assessing DMC, clinicians should focus not on the content of a belief or decision, but rather upon the relationships between that belief and decision to other elements of their ‘mental economy’:

The epistemic standards of ‘coherence’ and ‘correspondence’ thus provide a framework within which decisions and behaviour, whether unusual or not, can be interpreted and understood. It is only in virtue of the implicit background structure of interconnected beliefs, actions, and so forth, that individual beliefs (or values) can be picked out as normatively inappropriate, and therefore potentially indicative of an impairment that could undermine capacity: a note of discord in an otherwise fairly coherent and harmonious symphony of intentional behaviour.⁷⁹

As such, on the radical interpretation approach, and more generally on the view of autonomy that I have defended, a clinician’s substantive assessment of a particular belief or decision should be understood to motivate a broader kind of enquiry into the agent’s character system, rather than wholly constituting an assessment of DMC; substantive assessment here does not lend support to a substantive approach to autonomy or unwarranted paternalism.⁸⁰ The above discussion also suggests that, in situations in which there is disagreement between the physician and their patient about the best treatment option, it is not only appropriate for the physician to ask their patient to explain the reasons underlying their decision, but in fact necessary for establishing that the decision was made in the right way.

The concern that the manner in which the MCA is interpreted may currently lead to substantive assessments of capacity must be taken seriously. However, my suggestion is that this interpretation is a result of over-generalizations and misconceptions about both the nature of particular disorders and plausible demands of rationality, rather than a problem with incorporating considerations of rationality into assessments of DMC per se. It is possible to assess the extent to which a patient is making their decision in accordance with the kind of rationality condition I have outlined, without relying on the patient’s disease status or substantive considerations alone as crude proxies.

Yet, it might finally be objected that engaging in radical interpretation is highly burdensome for health care teams, who simply do not have the time to engage in this sort of detailed discussion with every patient. However, my claim that radical interpretation is the best way to accurately assess ‘rational DMC’ and to thus facilitate our ability to afford decision-making authority to patients appropriately, is quite compatible with there being stronger moral reasons that outweigh those in favour of its use. Such reasons might include considerations pertaining to the just allocation of scarce resources in health care, including the medical team’s time and energy. However, we cannot have our cake and eat it too. If we believe these other moral reasons are stronger, and therefore advocate an approach to patient decision-making that does not incorporate deep consideration of the patient’s reasons and values, this only means that we must acknowledge that we are trading off the value of giving decision-making authority to the people who actually deserve it (and protecting those who deserve protection from harm) against other moral values. It does not mean that the radical interpretation approach does not facilitate our ability to make true positive and true negative assessments of DMC.

Of course, that is not to say that radical interpretation is a flawless evidential mechanism in this regard. Even if it is highly accurate, there is still the possibility that some individuals might unjustifiably be denied decision-making authority on the basis that they have incorrectly been assessed as lacking practical rationality in their decision-making. But, this cost has to be weighed against the costs of two features of the status quo. The first is that the low threshold approach to the DMC in the MCA means that it is likely that a number of individuals currently qualify as having DMC when they are not able to make autonomous decisions about treatment. Second, due to the vague wording of the ‘use and weigh’ criterion, there is scope for widely varying interpretations for the general applications of this criterion.⁸¹ In addition to the concerns about how this may open the door to substantive considerations determining assessments of capacity, considerations of justice speak against leaving the interpretation of the ‘weigh and use’ criterion to the discretion and intuitions of different courts. The values of the individual either should or should not matter for all patients whose DMC is under consideration.

Ultimately, the question must boil down to how important we think rationality is to autonomy, and whether it is sufficiently important to include it amongst our criteria of DMC, given the costs of raising the threshold of DMC in a non-ideal context. My own view is that practical rationality warrants inclusion because of its centrality to autonomous decision-making. Practical rationality as I outlined it in the first chapters of this book is not just one ability among several that are relevant to DMC; it is central to the value of autonomous decision-making in so far as it allows us to direct our lives in accordance with our own values. It thus grounds the moral significance of all the other abilities that we typically accept are necessary to autonomous decision-making. Understanding information, retaining it, and ‘weighing’ it only matters for autonomy if we assume that agents have the ability to link that information (and how much weight it is given) to their own values; similarly we should only be concerned about the decision a patient communicates if it is a communication of a decision grounded by their values.

In the next chapter, I shall bring this theoretical discussion of DMC to bear on some practical cases that will serve to further elucidate features of my account. In doing so, I shall consider further the concern that assessments of DMC in the context of anorexia nervosa are in danger of collapsing the proceduralist test of the MCA into a status-based test that indirectly incorporates a substantive conception of autonomy.

1: Re T (adult: refusal of medical treatment), my emphasis. For a philosophical approach to capacity that endorses this sentiment, see Draper, ‘Anorexia Nervosa and Respecting a Refusal of Life-Prolonging Therapy’, 125–6.
2: Tan and Elphick, ‘Competency and Use of the Mental Health Act—a Matrix to Aid Decision-Making’.
3: Beauchamp and Childress, Principles of Biomedical Ethics, 114.
4: Grisso and Appelbaum, Assessing Competence to Consent to Treatment, 11.
5: Re SB; for discussion, see Herring, Medical Law and Ethics, 164.
6: McMahan, The Ethics of Killing, 250. For an exploration of why the possession of rationality conceived as a range property of agents is a sufficient basis of moral equality, see Carter, ‘Respect and the Basis of Equality’.
7: Buchanan, ‘Mental Capacity, Legal Competence and Consent to Treatment’. It is also possible to talk about capacity in the more generalized sense of what capacities an individual requires in order to be a globally autonomous agent, as opposed to an individual who has the capacity to make an autonomous local decision. See Dworkin, ‘Autonomy and the Demented Self’, 10. Dworkin notes that global autonomy might require further diachronic evaluative capacities, and that it might also be understood to ground the general right to autonomy.
8: I briefly consider and reject two alternative approaches to understanding the relationship between autonomy and DMC in the next chapter.
9: Kleinig, ‘The Nature of Consent’.
10: Re C (Adult: Refusal of Treatment); Craigie and Davies, ‘Problems of Control’; Blumenthal, ‘The Default Legal Person’.
11: Convention on the Rights of Persons with Disabilities (CRPD); for discussion, see Bartlett, ‘The United Nations Convention on the Rights of Persons with Disabilities and Mental Health Law’.
12: For a discussion of how this philosophical view is reflected in the Common Law regarding adult refusals of treatment, see Gavaghan, ‘In Word, or Sigh, or Tear’, 35. See also Clarke, ‘The Neuroscience of Decision Making and Our Standards for Assessing Competence to Consent’.
13: Appelbaum and Grisso, ‘Assessing Patients’ Capacities to Consent to Treatment’; Grisso and Appelbaum, Assessing Competence to Consent to Treatment.
14: Mental Capacity Act 2005, 2(1), 3(1).
15: Appelbaum and Grisso, ‘Assessing Patients’ Capacities to Consent to Treatment’.
16: Mental Capacity Act 2005, 2(1).
17: Herring, Medical Law and Ethics, 157.
18: Buchanan and Brock, Deciding for Others, 50.
19: See also Brock, Life and Death; Herring, Medical Law and Ethics, 157.
20: However, some provisions, such as those pertaining to advance directives, making a lasting power of attorney, and deprivation of liberty safeguards (amongst others), only apply to those who have reached the age of 18.
21: Gillick v. West Norfolk and Wisbech Area Health Authority and Department of Health and Social Security, HL 17 Oct. 1985.
22: Family Law Reform Act. For a useful overview of the law in this area, see Hope, Savulescu, and Hendrick, Medical Ethics and Law, ch. 10.
23: Buchanan, ‘Mental Capacity, Legal Competence and Consent to Treatment’.
24: Re T (Adult: Refusal of Medical Treatment).
25: For support of the sliding-scale view, see Buchanan and Brock, Deciding for Others; Buchanan, ‘Mental Capacity, Legal Competence and Consent to Treatment’; Eastman and Hope, ‘The Ethics of Enforced Medical Treatment’; Drane, ‘The Many Faces of Competency’.
26: Culver and Gert, ‘The Inadequacy of Incompetence’, 636.
27: Ibid., 632.
28: Culver and Gert do not themselves subscribe to this view, as they believe that it is permissible to overrule irrational decisions. Their complaint though is that the concepts of capacity and rationality should not be conflated.
29: Buchanan and Brock, Deciding for Others, 52–5.
30: Beauchamp and Childress also object to what I call the linear justification view. See Beauchamp and Childress, Principles of Biomedical Ethics, 76.
31: As Craigie points out, although this full-blown sliding-scale view was previously endorsed in the Common Law in England and Wales, since the enactment of the MCA, it has only been adopted in this attenuated sense. Craigie and Davies, ‘Problems of Control’, 13.
32: Buchanan, ‘Mental Capacity, Legal Competence and Consent to Treatment’.
33: Eastman and Hope, ‘The Ethics of Enforced Medical Treatment’.
34: For detailed discussion of issues facing the balancing account, see Wicclair, ‘Patient Decision-Making Capacity and Risk’.
35: Buchanan, ‘Mental Capacity, Legal Competence and Consent to Treatment’, 417.
36: Buchanan recognizes the implication that raising the threshold of capacity ‘also increases the number of instances in which people are incorrectly assessed as not legally competent’. Ibid., 417. He suggests that advocates of the epistemic justification must simply assume that the severity of the harm of false positives does not increase with the severity of the harm at stake. My argument below is that this assumption is unjustified.
37: Beauchamp and Childress, Principles of Biomedical Ethics, 76–7.
38: Beauchamp and Childress take their criterion of understanding to incorporate appreciation. Ibid., 88–93 and fn. 32.
39: Note that a patient can perceive these reasons, even if she only rejects the negative connotations of her illness. For instance, a patient who accepts the descriptive claim that she is not ‘healthy’ in a biostatistical sense, but who also denies that she has strong reasons to be healthy in this sense, may still be able to understand that she has very weak reasons to undergo medical treatment. For a detailed discussion of insight into mental disorder and implications for capacity, see Holroyd, ‘Clarifying Capacity’.
40: Appelbaum and Grisso, ‘Assessing Patients’ Capacities to Consent to Treatment’.
41: Appelbaum and Grisso, 1636.
42: I base this case on Appelbaum et al.’s observation that depressed patients may have decreased motivation to protect their interests, perhaps associated with feelings of hopelessness that may alter the nature of patients’ treatment decisions. Appelbaum et al., ‘Competence of Depressed Patients for Consent to Research’, 1380. See also Rudnick, ‘Depression and Competence to Refuse Psychiatric Treatment’.
43: For discussion of similar cases and the MCA see Rudnick, ‘Depression and Competence to Refuse Psychiatric Treatment’.
44: Brock, ‘Patient Competence and Surrogate Decision-Making’, 130. In describing the depressed person as value-impaired, he notes that ‘[t]here may be no failure in their understanding or reasoning about this outcome’; he argues that ‘mental illness that distorts what they value from what it would otherwise be can result in incompetence to decide about treatment’.
45: Fulford, ‘Evaluative Delusions’.
46: It may also be understood as a particularly damaging form of evaluative delusion, in so far as it encapsulates a paradoxical identification with what one loathes rather than what one values. See Radoilska, ‘Depression, Decisional Capacity, and Personal Autonomy’.
47: This issue with cognitivist tests of capacity has sometimes been parsed as a failure to incorporate considerations of volitional control. Craigie and Davies, ‘Problems of Control’, 2. I shall consider this framing in my discussion of anorexia nervosa in the next chapter.
48: Faden and Beauchamp, A History and Theory of Informed Consent, 264.
49: Nelson et al., ‘The Concept of Voluntary Consent’.
50: Frankfurt, ‘Freedom of the Will and the Concept of a Person’.
51: Christman, ‘Autonomy and Personal History’, 14; Hyun, ‘Authentic Values and Individual Autonomy’; Hill, Autonomy and Self-Respect, 49. Notably, Ploug and Holm, ‘Doctors, Patients, and Nudging in the Clinical Context—Four Views on Nudging and Informed Consent’ suggest that Christman’s criticism can also be weighed against the form of rationality implied by the standard account.
52: Christman, ‘Autonomy and Personal History’, 14.
53: A closely related objection in this context is that rationalist conditions rule out the autonomy of individuals who prefer a life of non-reflection or spontaneity. Hyun, ‘Authentic Values and Individual Autonomy’, 199; Double, ‘Two Types of Autonomy Accounts’, 73; Blumenthal-Barby and Naik, ‘In Defense of Nudge–Autonomy Compatibility’. However, this objection misconstrues the nature of procedurally rationalist theories by raising what is essentially a substantive complaint. The rationalist theory can quite easily accommodate the thought that an individual can autonomously live a life of non-reflection or spontaneity, as long as they do so because they believe that way of life is valuable—this is a procedural rather than substantive matter. For a similar reply to this objection, see Ploug and Holm, ‘Informed Consent, Libertarian Paternalism, and Nudging’.
54: See Baron, Rationality and Intelligence for an account of how the two differ.
55: Parfit, On What Matters, 31.
56: Raz, Engaging Reason, 18.
57: Charland, ‘Anorexia and the MacCAT-T Test for Mental Competence’; Vollmann, ‘“But I Don’t Feel It”’; Christman, The Politics of Persons, 144; Mackenzie, ‘Three Dimensions of Autonomy’, 31; Kong, ‘Beyond the Balancing Scales’, 234–5.
58: Arpaly, ‘On Acting Rationally against One’s Best Judgment’. Marilyn Friedman also recognizes that features of emotion and character can constitute reasons. See Friedman, Autonomy, Gender, Politics, 9.
59: For a nuanced discussion of how emotions can help us perceive practical reasons, see Tappolet, ‘Emotions, Reasons, and Autonomy’. In cases in which an agent does not reflectively endorse their emotional states, then altering these emotional states may serve to enhance their autonomy. For example, see Douglas et al., ‘Coercion, Incarceration, and Chemical Castration’.
60: See Camillia Kong’s recent defence of a relational approach to mental capacity in Kong, Mental Capacity in Relationship.
61: Culver and Gert seem to understand rationality in this sense in disputing the role of DMC. Culver and Gert, ‘The Inadequacy of Incompetence’.
62: Similarly, when Draper calls for a distinction between incompetence and irrationality, she provides examples of decisions that appear irrational, but does not offer an account of irrationality per se. Yet, she endorses an account of competence that requires the ability to ‘weigh information in a balance to arrive at a choice’. Draper, ‘Anorexia Nervosa and Respecting a Refusal of Life-Prolonging Therapy’, 126. As such, it appears that she accepts the claim that competence is incompatible with some sorts of practical irrationality.
63: Nelson et al., ‘The Concept of Voluntary Consent’. See also Beauchamp and Childress, Principles in Biomedical Ethics, 103.
64: Nelson et al., ‘The Concept of Voluntary Consent’. The authors draw this example from Jaworska, ‘Caring, Minimal Autonomy, and the Limits of Liberalism’, 82.
65: Gavaghan, ‘In Word, or Sigh, or Tear’, 249.
66: See also Meynen, ‘Depression, Possibilities, and Competence’; Gavaghan, ‘In Word, or Sigh, or Tear’, 246–9.
67: Craigie, ‘Competence, Practical Rationality and What a Patient Values’; Craigie and Davies, ‘Problems of Control’; Brock, ‘Patient Competence and Surrogate Decision-Making’; Buchanan and Brock, Deciding for Others.
68: Grisso, Appelbaum, and Hill-Fotouhi, ‘The MacCAT-T’. For criticisms, see Baergen, ‘Assessing the Competence Assessment Tool’; Kim, ‘When Does Decisional Impairment Become Decisional Incompetence?’; Banner and Szmukler, ‘“Radical Interpretation” and the Assessment of Decision-Making Capacity’.
69: Banner and Szmukler, ‘“Radical Interpretation” and the Assessment of Decision-Making Capacity’.
70: Mental Capacity Act Code of Practice, 4.22.
71: Craigie and Davies, ‘Problems of Control’.
72: Kong, ‘Beyond the Balancing Scales’.
73: Ibid.
74: Gavaghan, ‘In Word, or Sigh, or Tear’, 252.
75: Herring, Medical Law and Ethics, 165.
76: Gavaghan also defends further probing into the internal consistency of the patient’s reasoning process in such contexts. See Gavaghan, ‘In Word, or Sigh, or Tear’, 247.
77: Clarke, ‘The Neuroscience of Decision Making and Our Standards for Assessing Competence to Consent’; Peterson, ‘Should Neuroscience Inform Judgements of Decision-Making Capacity?’
78: Banner and Szmukler, ‘“Radical Interpretation” and the Assessment of Decision-Making Capacity’.
79: Ibid., 385.
80: Ibid., 389–92.
81: Bartlett, Blackstone’s Guide to the Mental Capacity Act 2005, 51.