The study

The quotations presented in this article are from participants who took part in a study that involved interviewing females aged between fifteen and twenty-six years who had a diagnosis of anorexia nervosa and were known to the psychiatric services. The interviews were semi-structured and covered a range of areas about participants' experiences of, and attitudes towards, anorexia nervosa. The focus of the study was on decision-making around anorexia nervosa and its treatment (Tan, Reference Tan2006).

For the main study, twenty-nine participants were recruited from four different centres across the south of England. Standard international diagnostic criteria for anorexia nervosa were used (American Psychiatric Association, 1996). Sixteen participants were under eighteen years old. At the time of interview the Body Mass Index of all but six participants was below 20 with a mode of 17 and the lowest value of 12. Further study details are given in Hope and colleagues (2011).

The number at the end of each quotation (e.g. P12) is the participant number. P1 − P11 were in a pilot study (Tan et al., Reference Tan, Hope, Stewart and Fitzpatrick2006a; Tan and Hope, Reference Tan, Hope, Widdershoven, MacMillan, Hope and Van Der Scheer2008) and no quotations from these participants are included here.

The burdensome world of anorexia nervosa

If there is one word that captures participants' experiences of living with anorexia nervosa that word is burdensome. Many participants, it is true, found in anorexia nervosa some advantages and positive aspects (Serpell, Treasure, Teasdale and Sullivan, Reference Serpell, Treasure, Teasdale and Sullivan1999), but, except in its early stages, the overwhelming experience was as a burden and a struggle.

From the accounts of the people we spoke to, anorexia nervosa is not a life choice in the sense of a way of living that has been chosen through reflection on the balance of positive and negative aspects or on deeply held life goals.

Anorexia nervosa is not burdensome in the ways that, we imagine, a hunger striker might find things difficult. The hunger striker typically has a political commitment that is part of that person's significant life goals. As part of that commitment he decides to go on hunger strike. It is not easy to starve oneself and it must be very difficult to maintain the refusal of food in the light of the strong physiological desire for food. But the political commitment gives the direction and the strength to maintain the self-starvation. The burden for the hunger striker, we imagine, is the burden of doing something physiologically difficult and dangerous for the sake of one's ideals.

For the person with anorexia nervosa, by contrast, the self-starvation can be only too easy. It is the eating to put on weight that is difficult. The burden is in part the conflict between wanting, on the whole, to avoid the serious physical dangers from self-starvation, and the anxiety associated with eating and gaining weight. But for many the burden is a great deal more than this. For many it would not be an exaggeration to say that they are in the grip of anorexia nervosa and this loss of control is itself a further burden.

Many of the study participants described a sense of being trapped.

The struggle to get free is exhausting, an exhaustion no doubt exacerbated by the physical weakness of starvation. It can be so exhausting that, like a scallop in the starfish's embrace, there comes a point when it is easier to give up resisting: to give in to those forces that drive the person to continuing weight loss.

The preoccupation with shape and weight leads to another set of burdens that arises even if the person gives in to the anorexia nervosa. The mind is filled with thinking about how to lose more weight, or is ruminating about feeling fat. Life is also impoverished because of the ways in which anorexia nervosa interferes with social activities.

Life can be further impoverished by the physical problems associated with low weight.

Finally, the pain caused to family and friends can be a burden to some.

In a nutshell, the burdens of anorexia nervosa can be:

• Terror of weight gain, and even of food itself;

• Conflict arising from knowledge of the dangers if the terror is not faced;

• Exhaustion due to the conflict – the constant struggle against the terror;

• Lack of control leading to a feeling of being trapped;

• Obsession with shape and weight which, combined with the lack of control, leads to

• Impoverishment of life – both intellectually and socially;

• Physical weakness and other physical consequences as a direct result of starvation; and

• Hurting others – those who love them and who face the pain of their self-harm.

Agency

If living with anorexia nervosa is such a burden, why not stop the behaviour and simply eat to put on weight?

One of the questions we asked all participants was what they thought others should know about anorexia. The most common answer was that others should know just how difficult it is to put on weight. It is so difficult that it often feels impossible.

Participants were well aware that people who do not have anorexia nervosa find it difficult to appreciate how there can be any difficulty in eating to put on weight.

Participants themselves were often surprised by how difficult it is to eat to put on weight. The idea that it is not possible might seem like a convenient excuse to avoid responsibility for the behaviour. Most participants thought carefully about this issue. Many were torn between accepting that they must be able to eat to put on weight, and feeling that, at least at some stages, they were just not able to do so.

One way of accounting for the behaviour of people with anorexia nervosa would be in terms of rational choices. On this view, if the individual overall believes that the advantages outweigh the disadvantages she will continue with the anorexic behaviour. Once she wants to put on weight, however, she will do so. According to this perspective, people with anorexia nervosa may have unusual and self-destructive preferences but they do not have any problem with the ability to act in accordance with these preferences. In short, if a person with anorexia nervosa does not eat to put on weight it is because she does not want to; it is not because she wants to but is unable to do so.

This analysis does not seem consistent with the experiences of the participants in this study. These participants had all, to some extent, accepted treatment, although often with much ambivalence. But even when they were clear that they wanted to eat to put on weight they still experienced great difficulty in doing so.

What can the barriers be to eating a reasonable amount of food? For those who could articulate these barriers the most commonly cited was anxiety, often described as terror.

Indeed, for this participant, even seeing a chocolate wrapper could provoke the anxiety.

Many participants described how the sense of lack of control creeps up on them. They begin the anorexic behaviour perhaps unaware of the dangers and by the time they start to be concerned and wish to stop, they find that the behaviour has taken hold.

From participants' accounts it seems clear that even when the person wishes to eat to put on weight it can be very difficult to do so. Some participants described how they had decided to accept treatment, knowing that this would involve eating more than minimal amounts of food, but when the food was set down in front of them they felt unable to eat it. Since most participants did not want to come to serious harm as a result of self-starvation, the difficulty in weight gain does not seem to be due to indifference to consequences. It is a matter of common sense that some actions are more difficult to perform for some people than for others. It is easy, for example, for a fit young person to walk 100 yards; impossible for a person whose legs are paralysed; and difficult but not necessarily impossible for someone with arthritis. How difficult it is for the person with arthritis depends to a significant extent on how bad is the arthritis. The more painful it is, the harder it will be to walk. We might even consider that the pain is so bad that the person really cannot walk 100 yards. But even when this claim seems reasonable there might be extreme circumstances in which the person could walk the 100 yards: to avoid a raging fire for example.

We suggest that an analogous situation occurs in anorexia nervosa. The anxiety associated with eating to put on weight, and even just with eating, is such that it can be very difficult to do so. On the view we are proposing, voluntariness should be considered a graded concept: actions are more, or less, difficult to perform.Footnote ² For people with anorexia nervosa, eating to put on weight may be very difficult even when the person is, at least at a conscious level, unambiguously motivated to do so.

The ability to act can be affected by the setting in subtle ways. Some study participants described how they felt unable to make the decision to accept treatment, and to eat more than minimal amounts of food. If, however, the decision was taken out of their hands, and they were forcefully told that they had to eat the food placed in front of them, they were able to do so. For some this seemed to relate to a question of responsibility: if they felt that they themselves would be responsible they could not take the decision, nor carry out the act. If, however, they felt others were responsible they were ‘released’ from their inhibition and able to act. In short, they could only act if they did not feel responsible for the act.

The difficulties with voluntary action clearly impact on the ability of people with anorexia nervosa to pursue life goals. They can also affect those goals because overcoming anorexia nervosa will normally involve facing and overcoming the fears and anxieties associated with eating to put on weight. These anxieties can therefore significantly limit opportunities in a way perhaps analogous to how pain from arthritis can limit opportunities. Respecting autonomy requires more from the physician, therefore, than simply respecting competent decisions. It requires enabling the patient to develop her autonomy: to become independent from controlling influences, and to maximise her capacity for intentional action and reflection (see Beauchamp and Childress, Reference Beauchamp and Childress2009, p. 100). Treating a disorder such as anorexia nervosa is not only about curing a disease but also about fostering self-development (Tan, Hope and Stewart, Reference Tan, Hope and Stewart2003a). In other words, promoting autonomy may be a goal of treatment. This dual role for autonomy may lead to conceptual and practical problems because respecting capacitous choice may conflict with promoting autonomy.

Promoting autonomy involves trying to motivate and help the person to face her fears and overcome them in order to achieve her ultimate goals. Although on this analysis it is justified, and indeed mandated, for health professionals to use their skills and powers of persuasion to try and enable the person to find the courage to overcome the anorexia nervosa, it does not follow that if the person refuses treatment such refusal is incompetent. This issue requires further consideration.

Consider a person with anorexia nervosa who unambiguously wishes to be free from it and wishes to put on weight. She is very fearful, however, of weight gain and experiences great anxiety when she attempts it. Ideally she would like to experience no anxiety and to gain weight. Given the anxiety that she does, as a matter of fact, feel, she chooses to avoid the anxiety and not to gain the weight. Is such a refusal a competent decision? It might be said that if there is reduced voluntariness to eat to put on weight then there is reduced competence to decide to eat to put on weight. But this, we believe, is wrong. The situation seems analogous to that of the person with arthritic pain deciding whether or not to walk the 100 yards for a particular reward. If the patient with anorexia nervosa understands the benefits from treatment it is up to her whether or not to face her fears and experience the anxiety for the sake of those benefits. We should no more force her into hospital and into treatment than we should force the person with arthritis to walk the 100 yards.

The difficulties with agency impair autonomy through reducing opportunities and restricting life goals, but such reduced agency does not provide grounds for considering that a decision to refuse treatment is incompetent. Thus a choice might not be fully autonomous while still being competent. In making this argument we have assumed that the person has a clear and unambiguous desire to be rid of the anorexia nervosa, and that the reason for refusing treatment is that despite this desire the person prefers to avoid the unpleasantness of treatment. Given that anorexia nervosa is such a burden, in so many ways, it seems reasonable to expect the person with anorexia nervosa to be clear that her fundamental goal is to be rid of the condition and to put on weight. The only difficulty, it would seem, is facing the anxiety that eating to put on weight causes.

Preferences or beliefs?

From the accounts of the participants in our study, however, such a clear and stable goal is not the usual experience, at least not until the person is firmly on the road to recovery. The lack of ability to hold on to a stable goal to overcome the anorexia nervosa is no doubt a central reason why motivation to undergo therapy can be so fragile. One participant, when asked what is the hardest thing about having anorexia nervosa, replied:

What is the cause of this continual fluctuation, this constant battle? Part of the answer, we believe, is that the goal of achieving a healthy weight is itself fragile because the person's beliefs about her shape and weight, and about the dangers of self-starvation, fluctuate. These beliefs fluctuate because there is a mismatch between the objective evidence and the evidence from her affective responses.

It is often said that people with anorexia nervosa misperceive themselves as fat when they are thin (Bruch, Reference Bruch1962; Garner, Garfinkel, Stancer and Moldofsky, Reference Garner, Garfinkel, Stancer and Moldofsky1976). But of course even if they do experience a misperception there is contradictory evidence as well – evidence that the person is not overweight. This evidence includes the testimony of family, friends and professionals, as well as objective measures such as weight and waist size. If the false evaluation that she is overweight originates from a false visual perception of her body (or its mirror reflection) but she is otherwise rational, what does she make of all this contradictory evidence?

One participant brought this issue up:

Participants' accounts suggest that they feel negatively about themselves when they see their body (either in reflection or directly), and this feeling leads to the judgment that they are too fat. Indeed, the judgment that ‘I am too fat’ in this context is ambiguous. It might mean ‘I am significantly overweight by population norms and by standards of health’; or it might mean ‘I realise that I am underweight by these objective standards but nevertheless I am too fat for me’.

The first of these meanings is a straightforward objective statement that if said by an underweight person is false. The second meaning, however, is a subjective judgment (caused by an emotional response) and is not rendered false by evidence about her weight or waist size. One participant was questioned about how she responded to evidence that she was in fact very low in weight. Her answer suggests that on reflection her belief that she was overweight was a value judgment.

From the participants' accounts the ‘perception’ that the body is fat when it is significantly underweight does not seem to be a straightforward (mis)perception analogous to those distorting mirrors found in fairgrounds. Rather than false perceptions of body size causing the desire to lose weight it may be that the belief that one is fat and the apparent misperceptions are part of the same set of affective experiences that cause the person with anorexia nervosa to be terrified of weight gain in the first place.

Many people who suffer from anorexia nervosa do not usually think that they are overweight with respect to others, so much as feel that they are ‘fat’, or ‘huge’ in an emotionally laden way (usually with an associated feeling of self-disgust). The goal of such people is usually to be ‘thin’ rather than some particular centile or objective weight; or, more accurately, ‘thin enough for me’ – an elusive unattainable goal which always floats just out of reach even as weight is lost and risk to health and life increases.

One way of unpacking this is that the person does perceive herself correctly, and at some level does believe that she is thin, but the affective responses lead her to discount the perception and associated belief, and not give them the appropriate epistemic weight. Indeed for some people ‘feeling fat’ may be more related to a feeling state than it is to an actual weight; it can even become a kind of shorthand for a particular emotional state.

A similar point emerged with regard to the physical risks of self-starvation, including death. One participant described her beliefs thus:

At another point in the interview she said:

This participant appears to be making two distinct statements. First, a statement about her beliefs: she did not believe that she was in danger of death. Second, a statement about her preference to die rather than to put on weight. These two ideas are, however, closely connected. Her apparent preference to die rather than put on weight may have been affected by difficulties she had in really believing that she is going to die.

Why do so many people with anorexia appear not to believe the risks of serious harm? One participant explained it thus:

And another:

It is not clear that these participants had a value that puts the risk of death as of less importance than shape and weight. It is rather that the risk of death has a low salience. And again this is likely to be related to the emotions. Things are salient to us in part because of their affective connections. This issue of salience may be particularly significant in the context of making decisions as opposed to justifying decisions. The lack of salience of the physical risks seems in part due to the focus on shape and weight.

We propose the following hypothesis for what is going on when people with anorexia nervosa refuse treatment even when at significant risks from self-starvation. The starting point is that people with anorexia nervosa experience high levels of anxiety around the idea of gaining weight or of being overweight. When the person sees her body image or her body directly she feels these high levels of anxiety. She does not directly misperceive her body as overweight but she experiences her affective responses as telling her she is fat.

Our beliefs are affected by our affective states (see, for example, Loewenstein, Weber, Hsee and Welch, Reference Loewenstein, Weber, Hsee and Welch2001), and generally beliefs and emotions work together in a constructive way. In anorexia nervosa the emotions tell the person one thing: that she is fat and that she is at more danger from being fat than thin. The objective evidence and the testimony of others tell her the opposite. There is a kind of internal tug-of-war comparable with that described by the neuroscientist Elkhonon Goldberg (Reference Goldberg2009, p. 121), and attributed by him to a dissociation between cognitive and emotional contributions associated with different areas of the brain. The emotional responses of a person with anorexia nervosa to her body – including the sense of feeling fat − are so powerful that together with the lack of salience of the contradictory evidence she is led to believe that she is fat. Her refusal of treatment may be based on these feelings and associated beliefs together with the lack of salience of the dangers of self-starvation.

On this account a false belief is guiding her refusal of treatment even though overall it is not clear that she holds this belief. On the traditional analysis one might therefore consider her as lacking competence. But when physicians question her, emphasising the objective evidence that she is actually severely underweight and at risk of harm, she cannot rationally deny this. But she is still unable to bring herself to believe what she is being told or to accept treatment because of her affective experiences. So she must justify such refusal when called on to do so, for example as part of the assessment of her competence. The only logical argument is if she puts forward the view: I would prefer to risk death than put on weight. And this looks like a preference. On the classic analysis she is competent because the conclusion to refuse treatment follows, given her preferences, even if she correctly believed that she is underweight and at risk of death. But this preference is not one that she clearly held before the discussion with her physician. It was constructed to create a rational argument for refusing treatment that can satisfy her physician.

The power of the emotions in directing the person's decisions calls into serious doubt the validity of the person's refusal of treatment even though when pushed to justify the decision the person apparently does so on the basis not of false beliefs but of self-destructive preferences.

The classic model of competence, as Louis Charland (Reference Charland1998; Reference Charland2002) has argued, is based on a false psychology. Specifically it fails to understand the enormous importance of affective states in our cognitive processes. As such it asks only ‘Is this a false belief or is this an unusual preference?’, and a great deal depends on the answer. But where decisions are affected by strong emotions, as is the case in anorexia nervosa, the distinction between a false belief and a (self-destructive) preference is not robust.

Ambivalence

We are hypothesising that there is a marked dissonance between the beliefs and goals that result from the person's affective experiences, that is, their feelings and emotions as experienced over long periods of time, and those that follow from more objective evidence. People with anorexia nervosa can understand the evidence that they are unhealthily underweight and at physical danger, and they can see the ways in which anorexia nervosa impoverishes their life. At the same time, the profound emotions tell them that eating to put on weight is dangerous and that they will be safer if they lose still more weight.

It is no wonder that many participants spoke of significant ambivalence with two conflicting parts of their mind, each with different preferences. These conflicting preferences lead to ambivalence about motivation.

There is a close and interactive relationship between affective and cognitive states: between how participants feel and how they think. Furthermore, participants describe a shifting between ‘mindsets’. A mindset in this sense is similar to Teasdale's (Reference Teasdale, Clark and Fairburn1997) use of ‘mind in place’ and refers to a somewhat coherent grouping of beliefs, goals and affective states. In one mindset the goal is to gain weight, and it is possible to eat more without feeling overly anxious. But, as described in the quotation above, the person quickly reverts to the other mindset: the anorexic mindset. She panics, she feels she is not able to eat to put on weight and perhaps she is no longer sure that to be healthy she needs to gain weight.

Buchanan and Brock (Reference Buchanan and Brock1989) suggest that if a person frequently changes her mind then she might lack competence. The descriptions of the two mindsets suggest that people with anorexia nervosa experience not only frequent changes in preferences and beliefs but may be profoundly ambivalent at any one time.

Such instability and internal ambivalence, what we have elsewhere called substantial inner conflict (Hope et al., Reference Hope, Tan, Stewart and Fitzpatrick2011), provides grounds for considering that the decisions of at least some people with anorexia nervosa are not competent decisions. Craigie (Reference Craigie2009) has developed the argument that a person may lack competence when she bases a decision on a consideration that she later comes to recognise that she overvalued. Craigie explicitly applies her analysis to people with anorexia nervosa. The problem in making a competence assessment is exacerbated by the possibility that the ambivalence is not articulated.

A similar point is made in the quotation at the start of this article. In the face of such ambivalence, and instability of view, is it possible to identify one of the two ‘mindsets’ as the one that should be respected, the one that is in some sense the ‘authentic’ self?

Authenticity

In another article (Hope et al., Reference Hope, Tan, Stewart and Fitzpatrick2011) we have examined the struggles with authenticity of the study participants. We described how, for almost all participants, the relationships between anorexia nervosa and a sense of the authentic self were a significant issue.

For many, anorexia nervosa was seen as separate from the real self, although this was experienced in various ways. The idea of a power struggle between these was a common theme.

Many participants described how the relative strengths, or sizes, of the two selves was weight dependent: the lower the weight, the stronger the anorexic part of the self. The two parts of the self had different and often conflicting desires. A few described a hierarchical relationship involving ‘second order desires’ (Frankfurt, Reference Frankfurt1971; Dworkin, Reference Dworkin1988): the desire not to have one set of desires or at any rate to overcome its behavioural consequences. Most, however, described their experiences as of ambivalence and conflict without there being a hierarchical relationship between conflicting desires.

In contrast with those who described, or experienced, separate selves, many participants described how they viewed the anorexia nervosa as integral to a single self. For this group, treatment could be problematic (Tan et al., Reference Tan, Hope and Stewart2003a).

We discussed in a previous article (Hope et al., Reference Hope, Tan, Stewart and Fitzpatrick2011) the question of whether it is justified to override the treatment refusal of a person with anorexia nervosa on the grounds that the refusal may be inauthentic. We argued against such a view. We thought that the current theoretical and empirical understandings of authenticity are not sufficiently well developed to provide a secure basis for such an argument. Furthermore, at the time when decisions about treatment refusal need to be made, it is unlikely that the person herself would refuse treatment and at the same time say that her refusal is inauthentic.

A summary of the experiences of living with anorexia nervosa

From the accounts of the study participants, living with anorexia nervosa is a burden. It can take over one's life and curtail freedom. One response is to try and fight free from the grip of anorexia nervosa, but the ensuing struggle is exhausting and unpleasant. Eating to put on weight can be terrifying because of the extreme anxiety it engenders, and giving in to the anorexia can lead to a life that is impoverished and one-dimensional, and health is severely compromised. The study participants did not want to put themselves at risk of death. The natural response, it might seem, would be to have an unambiguous and sustained wish to overcome the anorexia nervosa, to seek treatment and to do one's best to ensure that the treatment is successful. But this is not the case for most people much of the time. Why is that? In a nutshell, the answer is because of the affective components that affect motivation and belief.

Treatment involves two distinct levels of co-operation: first, agreeing to put on weight and to abandon plans to lose more weight; and second, actually eating more in order to put on weight (Tan, Reference Tan2006 ). Agreeing to put on weight is difficult in the face of the strong affective experiences that ‘I am (still) too fat for me’. Furthermore the feelings of security that accompany the thought (and actual experience) of weight loss provide added motivation to lose weight, and give the message that weight loss is safe. Actually eating to put on weight is also very difficult because of the extreme anxiety and feelings of self-disgust that accompany the intake of food, particularly intake that is perceived as excessive. This anxiety is not only unpleasant in itself, but also gives the message that eating, and weight gain, are dangerous. Unfortunately, diets designed for restoration are not only considerably larger than what most patients would have been used to taking, but also the amounts of food would (gradually) exceed what the person's friends and families would be eating to maintain weight. These emotional difficulties in eating are exacerbated by the physical difficulty in eating when weight is low, for example, due to the sensations of bloating caused by slowing of gastric emptying.

Despite the anorexia nervosa, however, participants could appreciate the evidence that they are significantly underweight and should put on weight. This conflict between two sets of preferences and beliefs leads to conflicting mindsets: an ‘anorexic’ and a ‘non-anorexic’ mindset. The experience of anorexia nervosa is of conflict between these two mindsets, even when such conflict is not articulated. This experience of two mindsets with conflicting preferences and beliefs leads people with anorexia nervosa to become concerned with the issue of the authenticity of the self. Is one mindset (e.g. the non-anorexic one) the person's authentic self? Or is the self some combination of the two?

The result of these experiences is that autonomy may be compromised in the four ways outlined near the start of this article and discussed in the main sections above.

Enforcing treatment

We have argued that the difficulties with voluntariness and agency, whilst impairing autonomy, are not grounds for considering that a decision to refuse treatment is non-autonomous. Although concerns with authenticity are significant for people with anorexia nervosa, and a relevant focus in therapy, the concept does not help in deciding when it is right to override treatment refusal (Tan, Stewart, Fitzpatrick and Hope, Reference Tan, Stewart, Fitzpatrick and Hope2010).

In contrast, we have argued that treatment refusal may be based on affective experiences leading to false beliefs (albeit beliefs that are held ambiguously) and should for that reason be considered as an incompetent decision. It is not clear, however, that such treatment refusal would be assessed as incompetent, or as lacking decision-making capacity, using current legal and clinical assessment methods. There has been controversy over how such methods should be understood and applied in specific situations (David, Hotopf, Moran, Szmukler and Richardson, Reference David, Hotopf, Moran, Szmukler and Richardson2010). In order to have decision-making capacity (or to be competent) it is generally regarded that the person must be able to make a choice. The Mental Capacity Act 2005 in England and Wales states that a person must be able to use or weigh the information relevant to the decision as part of the process of making the decision. Our argument is that there are subtle but important ways in which such an ability can be impaired in the setting of anorexia nervosa: ways that might be missed or ignored in making a capacity assessment. A crucial point is that accounts of capacity (competence), even such a detailed and sophisticated account as that of Grisso and Applebaum (Reference Grisso and Applebaum1998), make a distinction between a decision based on a false belief (incompetent) and one based on unusual or self-destructive preferences (competent). We have argued that, in the setting of anorexia nervosa, this can be a distinction without a difference. A decision may be based on a false belief but justified in terms of self-destructive preferences. Moreover, excessive anxiety may compromise the person's awareness of these issues. Competence, and autonomy, may therefore be significantly compromised even when competence assessment using standard criteria fails to pick this up (Tan, Hope and Stewart, Reference Tan, Hope and Stewart2003b; Tan et al., Reference Tan, Hope, Stewart and Fitzpatrick2006a; Tan, Stewart and Hope, Reference Tan, Stewart and Hope2009).

On some analyses, competence is compromised if decisions, preferences and goals fluctuate significantly over time (Buchanan and Brock, Reference Buchanan and Brock1989; Craigie, Reference Craigie2009). But assessment of competence is generally approached as though it can be effectively carried out at one point in time. This again may be problematic in the setting of anorexia nervosa, where inner conflict and variation in preferences is a common experience. Competence assessment is further complicated by the fact, stressed by some of the study participants, that people with anorexia nervosa may fail to articulate the inner conflict they are experiencing. Some patients, in our clinical experience, who have refused, for example, to eat the meal provided for them, have subsequently complained that we were not sufficiently forceful with them, and that we were too respectful of their refusal.

Whether the current methods of assessment identify these ways in which competence may be compromised will depend on how competence is understood, how the methods are applied, and how much of the inner conflicts are articulated. We conclude that with the current understanding of competence and the current assessment tools and approaches it is not appropriate for law and policy to require that people with anorexia nervosa can be treated despite refusal only if they lack competence. We believe that from the accounts of people with anorexia nervosa reflecting on their experiences there is a significant group of people who could be assessed as competent but whose refusal of treatment for the reasons given may not be autonomous. Furthermore, what a person with anorexia nervosa says she wants may depend on the context and the ways in which she is asked. For example, a person faced with a meal that she is being encouraged to eat may express the view that she would prefer to die than eat to put on weight. In a context where the dilemmas that she faces are being discussed, the same person may say that of course she would not prefer to die, but that eating to put on weight is very difficult.

Although it may be right for society to allow people with illness to risk death or serious harm by respecting treatment refusal, it is only right if such refusal is autonomous. We believe that there is sufficient doubt around the autonomy of treatment refusal in the case of anorexia nervosa for it to be wrong necessarily to risk such harm on the grounds that treatment refusal must be respected. In order to further develop policy around enforcing treatment on patients with mental disorder, the accounts from people who have experienced the disorder are important.

Autonomy and clinical management

In respecting patient autonomy a clinician's role might be described as helping release the patient from the prison of anorexia nervosa. In doing this it will be important to understand the following points.

1. (1) The patient may be experiencing a greater level of inner conflict and ambivalence towards the anorexia nervosa than she is expressing and may vary in her beliefs and preferences from one day to the next. Thus, motivation to recover may be highly unstable, which can be problematic for family and clinicians. Patients can be relieved when clinicians explicitly raise the question of inner conflict, and helping patients voice this conflict and articulate the different mindsets can enhance collaboration and assist the process of recovery.

2. (2) Eating to put on weight is usually very difficult. This difficulty with agency, the near impossibility of eating to put on weight, can impact negatively on a motivation to get better. Some patients feel more able to eat if the decision is taken out of their hands. Indeed, some can only eat if they do not feel responsible for the act. The clinician needs to work jointly with the patient to find ways of sensitively providing assistance with eating. Being aware of the high anxiety, indeed torment, facing many patients when they eat can lead to more empathic and productive strategies to assist this process, including strategies to manage anxiety.

3. (3) The powerful affective components of anorexia nervosa impact not only on agency but also on beliefs. Even when she appears to accept the objective evidence regarding her shape, weight and the dangers of self-starvation, the decisions of a person with anorexia nervosa (for example to refuse treatment) may be driven by false beliefs about these issues. Thus competency can be compromised in subtle ways and self-stated preferences may fluctuate. In making sense of the self-stated preferences of a person with anorexia nervosa, it is important to be aware of the impact of strong affect, not only on the ability to weigh up the relevant issues in coming to a decision, but also on the formation of the key beliefs regarding eating, shape, weight and the consequences of starvation.

4. (4) As a result of substantial inner conflict, the patient may be struggling with issues of authenticity and identity. A crucial aspect of therapy is enabling the patient to build up an identity with which she feels comfortable.