Introduction

Respect for autonomy is usually considered one of the most important reasons in favour of legalising euthanasia. However, in recent debate this has been challenged by arguments claiming that autonomy is primarily a value that needs to be promoted or protected. This may imply, or so it has been argued, that euthanasia and physician-assisted suicide (PAS) are impermissible since they destroy the autonomy of the patient (Materstvedt 2003). Accordingly, it has been maintained that considerations of autonomy cannot be used as a reason in favour of euthanasia or PAS and that such arguments are self-contradictory (Doerflinger 1989; Gordijn and Janssens 2000). The idea of autonomy as a value has also been taken as a reason not to offer palliative, continuous deep sedation in end-of-life care (Eckerdal 2004). The question of whether respect for autonomy may support a right to euthanasia has been labelled as one of the central topics in palliative care ethics (Hurst and Mauron 2006).

The aim of the present paper is to investigate the rationale behind the claim that respect for autonomy is incompatible with euthanasia and assisted suicide. The argument deserves attention for two reasons. First, it has become influential in palliative care guidelines and policy documents, and it has been put forth by leading theorists in the field. Second, it is theoretically interesting since it supposes quite a different normative relevance of autonomy than the standard bioethical conception of autonomy as a ground for a right to self-determination (including the right to decide when to end one’s life).

The outline of our paper is as follows: starting with some general remarks about the concept of autonomy as it is used in bioethical debate, we will go on to present the above-mentioned arguments as they have been propounded in current discussions. After that, we will present two different interpretations of the idea that autonomy is valuable: autonomy conceived of as a personal prudential value—something that benefits us in that it makes our lives better—or as an unconditional value in a Kantian sense. In doing this, we will argue that autonomy as a value, however interpreted, cannot provide a reasonable ground for the position that euthanasia or assisted suicide is always wrong.

The concept of autonomy in bioethics

Autonomy is a central concept in bioethics. The core of the concept as it is understood in this context concerns personal self-governance: being in control of oneself and choosing one’s own way (Dworkin 1988). Most theories about personal autonomy acknowledge that it presupposes a capacity for intentional action on the basis of one’s own rational deliberations as well as freedom from controlling instances. Sometimes a distinction is made between freedom and autonomy, with freedom specifically referring to the absence of external hindrances (Tännsjö 1998). Freedom in this sense corresponds to Isaiah Berlins’ notion of negative liberty (Berlin 2002). The term freedom is also used in the sense of positive liberty, which is a broader notion concerning the actual presence of self-governance. Both aspects of liberty are highly relevant to autonomy as the concept is understood in contemporary bioethics (Beauchamp and Childress 2001).

Another distinct question is what normative relevance autonomy has, i.e. in what way autonomy is important. Its relevance can be understood in terms of a right to be respected, (Beauchamp and Childress 2001) and in terms of a valuable end to be promoted (Juth 2005). In Kantian ethics, autonomy is a fundamental moral value and the basis of our moral obligations towards ourselves as well as towards other persons, and thus gives rise to duties and not primarily to rights (O’Neill 2002).

The central question for this paper is whether there is a reasonable interpretation of the normative relevance of autonomy that can back up the claim that euthanasia or PAS is always wrong, or that considerations of autonomy cannot be a reason for complying with a patient’s request for PAS or euthanasia.

Autonomy-based arguments against PAS and euthanasia

The idea that autonomy is incompatible with euthanasia emanates from the presumption that autonomy is valuable, and therefore needs to be protected. This reasoning has been put forth in several different contexts. Gordijn and Janssens (2000) maintain that respect for autonomy in palliative care implies negative as well as positive obligations for healthcare personnel. They argue that healthcare personnel may be obliged to enhance or restore patients’ autonomy, but they have no obligation to act if their actions lead to loss of autonomy. Since death means that the necessary conditions for autonomy are lost, respect for autonomy can therefore not be a reason in favour of complying with a patient’s request for euthanasia (Gordijn and Janssens 2000). The European Association of Palliative Care (EAPC) similarly argues that respect for autonomy in palliative care “primarily is about strengthening and restoring autonomy […] not to destroy it” (Materstvedt et al. 2003). According to Materstvedt, the chair of the EAPC ethics task force on palliative care and euthanasia, the value of autonomy means that euthanasia and PAS are unjustifiable since they “damage autonomy in the most fundamental way” (Materstvedt 2003). Doerflinger develops a similar argument in the paper “Assisted Suicide: Pro-Choice or Anti-Life” (1989). In this paper, Doerflinger argues that there cannot be an autonomy-based right to PAS (which is the subject he discusses), since death means total loss of future freedom. He concludes that options in healthcare must be limited in order to safeguard “freedom itself”.

An analogous argument has also been put forth in the discussion concerning continuous deep sedation (CDS). CDS means that the consciousness of the patient is reduced without any intention to restore it, in order to avoid intolerable suffering (Morita et al. 2002). Although practised in many European countries (Cherny and Radbruch 2009), the procedure is controversial.Footnote 1 Although there are important differences between the practices of CDS, PAS, and euthanasia, similar arguments have been used both for and against these practices. It has, for instance, been argued that since autonomy is valuable and sedation leads to loss of autonomy, it is wrong to sedate competent patients during their last phase in life (Eckerdal 2004). We will return to this discussion in the subsequent sections.

This reasoning implies that physicians have obligations to preserve and promote patients’ autonomy and that these, in some situations, have priority over patients’ wishes about how to end their lives. Thus, autonomy is seen as a value to protect and promote rather than a right for individuals to decide about their care.

The arguments presented above can be questioned on several grounds. In this paper, we will neither discuss whether autonomy is valuable in itself, nor how the alleged value of autonomy is to be balanced against other potentially valuable things, such as wellbeing or absence of suffering. In the following we will assume that autonomy is valuable in itself, in order to focus on the question of whether death through euthanasia or PAS necessarily means an unacceptable loss of autonomy for the person in question.

In order to investigate this we need to achieve a better understanding of the ways in which autonomy can be considered valuable. We will examine what we believe are the two most plausible interpretations of autonomy as a value in healthcare. As we shall see, neither interpretation gives a compelling reason for holding every instance of euthanasia to be wrong. On the contrary, we will argue that autonomy, also conceived of as a value, may serve as a well-founded reason in favour of these practices. This does not mean that requests for euthanasia or PAS always should be respected, nor do we believe this to be a sufficient reason for legislative or regulatory changes in Europe or elsewhere. The aim of the paper is merely to show that autonomy-based arguments against euthanasia, PAS, or CDS cannot establish that these practices always and necessarily are wrong.

Autonomy as a prudential value

The first notion of the value of autonomy that seems relevant for the way the argument has been presented is the idea of autonomy as a personal prudential value—something that benefits us and makes our lives better. Such a notion of the value of autonomy can be found in J. S. Mill’s classical writings on liberty (Mill 1982).Footnote 2 In recent years the idea has been elaborated by Lindley (1986) as well as in contemporary bioethics (Beauchamp and Childress 2001; Juth 2005). The value of autonomy here resembles the value that usually is ascribed to wellbeing or preference satisfaction in consequentialist ethical theories. If the value of autonomy is seen in this way, the question of whether or not to respect a patient’s wish for euthanasia or PAS boils down to concerns for the patient’s overall prudential interests (welfare in a very broad sense). If we look at the argument against PAS and euthanasia from this perspective, it must lend credence to the view that it is in the patient’s own best interests to remain autonomous, rather than to die or lose consciousness.

There are two ways of understanding such a position. One possible interpretation is that capacity for autonomy is what is valuable (Husted 1997), whereas another view would rather stress the value of exercising one’s autonomy (Lindley 1986). Let us start with the first assumption, that merely capacity and not exercise constitutes the value of autonomy. If capacity is valuable without relating it to actual exercise, the complete infringement of a person’s actual abilities to act autonomously is unproblematic from the point of view of autonomy as long as the mere capacity for autonomous acting is maintained. But if it is in a person’s interest to be autonomous, then it is reasonable to assume that it must also be of some interest for that person to actually exercise their capacity. It seems strange to hold the capacity for autonomy as highly valuable and not ascribe any value to the exercise of it. Such a position may be coherent but is, in our view, not very plausible.

However, if we would adhere to the idea that the exercise of autonomy (also) is valuable (which of course presupposes capacity), there does not seem to be a simple clear-cut answer on how to respect autonomy when persons decide to end their lives. Rather the value of respecting a certain autonomous wish must be weighed against the value of future autonomy.Footnote 3 In order to take a closer look at how to balance the different aspects of the value of autonomy, we will turn to the argument presented by Doerflinger (1989).

Doerflinger writes that patients who are capable of choosing to end their lives are also capable of making many other important decisions. The problem with suicide is that it removes the possibility of making all these other choices.Footnote 4 Doerflinger argues by way of analogy by referring to the well-known discussion of whether a right to freedom entails a right to give up one’s freedom and become a slave. Doerflinger’s answer is no, the freedom of the individuals must be protected against their own decisions. He then compares the situation of a slave with that of a corpse and argues that, in terms of freedom, the slave is better off, since he at least is free to choose which attitude to take towards his predicament whereas the corpse cannot think at all. Therefore, or so it can be argued, if one holds that there is no autonomy-related right to voluntary slavery, one must come to the same conclusion regarding assisted suicide.

Appealing as this analogy may seem at first, it rests on a quite problematic assumption. It seems to assume that we can compare the freedom of a corpse with that of a living person. This is absurd; corpses are neither free nor unfree, the predicate ‘free’ cannot be accurately applied to a corpse. When comparing the freedom of a slave with a free person it seems obvious that a life of slavery is likely to involve less freedom than a life of freedom. But this does not prove that a person would live a less free life if they were to have a request for assisted suicide respected at the end of their life than if such a wish on their part were not to be respected. Thus, the analogy loses force. If we assume (as the argument presupposes) that we can measure autonomy, not only the sheer quantity of autonomous decisions seems relevant but also their relative importance to the person. It seems unreasonable to claim that persons are better off just because they make, or are able to make, many choices, if these choices do not matter to them, or if they are made possible at the cost of disrespecting more important decisions.

Although a person in slavery may be free in some respects, their freedom could be infringed in so many substantial ways that a life of slavery would be highly unlikely to benefit their overall autonomy. In fact, if it is possible to ascribe a negative value to every infringement of a person’s autonomy, it is possible that a slave will end up with a negative balance. In order to make the analogous argument that it is better for a patient to remain autonomous than to be assisted in dying, one must make a case for the negative value of disrespecting the patient’s choice being outweighed by the positive value of future autonomy. This claim may hold in some cases, as when persons with relatively long and healthy lives ahead of them want to end their lives, as in the case of a temporarily depressed and suicidal teenager. But, for instance, in situations where patients are suffering from end-stage terminal illnesses, the claim seems much less compelling.

Imagine a case where after careful consideration a patient decides that they want to end their life in order to escape suffering at the very final stage of an incurable illness. Would disrespecting their wish to die be more beneficial to their autonomy than actually helping them to end their life the way they want to? We believe that, from the point of view of autonomy as a prudential value, it is unreasonable to claim that any amount of future autonomous action, no matter how small, is of greater value to a person than having a central wish respected towards the end of life. If the value of autonomy is to be conceived of also in terms of actually exercising one’s autonomy, having control over how to end one’s life is a plausible candidate for valuable exercise of one’s autonomy.

The Kantian argument; autonomy as an unconditional value

As argued above, the idea of autonomy as a prudential value may rule out certain acts of assisted suicide or euthanasia. It does not, however, provide us with a reason for concluding that all acts of euthanasia, PAS, or CDS are wrong. But there is another way of interpreting the autonomy argument against PAS and euthanasia, namely one which rests on Kantian ethics and the idea of autonomy as an absolute, unconditional value.Footnote 5 This is a completely different notion of the value of autonomy from the idea of a prudential value outlined above. According to this view, autonomy is not good because it is good for us, and not something that should be promoted because more is better. As Velleman (1999) argues, there is an important difference between what is good for a person and the value of the person himself. This latter value is a fundamental value which all prudential values depend on. Thus, on this view autonomy is intrinsically linked to our dignity as human beings and the absolute value of personhood.

Velleman, like Doerflinger, takes the case of slavery as an example. Velleman argues that people have no right to sell themselves into slavery, not because slavery harms them (or is prudentially bad for them), but because slavery violates personhood. And the same reasoning could, according to Velleman, be applied to suicide. Suicide is wrong, not because it is prudentially bad for us but because it is a violation of the value of personhood. Many contemporary Kantians similarly argue that PAS and euthanasia are impermissible for this very reason. However, in the following we will argue that considerations of autonomy also gives Kantian ethics a possible reason to accept at least some instances of suicide as well as euthanasia.

Kantian ethics is built on the notion of a categorical imperative, which comes in three different formulations. The first formulation of the categorical imperative says: “Act only in accordance with the maxim through which you can at the same time will that it become a universal law” (Kant 2002: 37). According to Kantianism, this rules out practices that would be inconsistent in terms of universal practice (for instance, if everyone were to lie, there would be no trust, and with no trust there can be no lies—therefore the practice of lying is inconsistent with the categorical imperative). Kant argued that suicide cannot be universalised and is therefore wrong. However, it is questionable whether it is inconsistent in such a strict sense so that the maxim cannot be universalised.Footnote 6 Let us assume that a person in terrible suffering and with little time left to live takes his life. There does not seem to be any logical inconsistency with a world just like ours where people in such predicaments choose to end their lives (Nell 1975: 79f).

Even if one would reject the argument that purposively shortening one’s life is not universalisable according to the first version of the categorical imperative, there could nevertheless be reasons why it would be acceptable according to Kantian ethics. According to the second formulation of the categorical imperative, we should always act in such a way that we treat humanity, whether in ourselves or in others, as an end, and never merely as a means to an end (Kant 2002: 46f). In order to respect our own dignity, there are constraints on what actions are permissible for us. Suicide could then be seen as being wrong, since it means that we use ourselves (our autonomy) as a means and not as an end. We cannot justifiably trade an absolute value (autonomy or dignity) against interest-relative goods such as pain relief, since this would mean our prioritising what is valuable to us over the value we carry in ourselves (Velleman 1999).

Velleman nevertheless argues that there is a Kantian argument in favour of at least some instances of PAS and euthanasia. He argues that the Kantian objection to suicide applies to cases where suicide is committed in order to obtain benefits or escape harm, and that this leaves open the possibility of suicide being justified for other reasons, namely when it is committed out of concern and respect for our dignity. Velleman writes that it sometimes is justifiable to destroy objects of value if they would otherwise deteriorate in ways that would offend that very value. Similarly, patients who are about to lose autonomy and dignity due to suffering and distress caused by terminal and incurable illness may be justified in ending their lives in order to die with dignity.Footnote 7 If we accept this line of reasoning, the Kantian argument can be used as a reason against PAS or euthanasia that is motivated by concern for suffering but not by concern for loss of autonomy and dignity. This would mean that an absolute ban on assisted suicide and euthanasia cannot be maintained on this reading of the autonomy-argument against PAS and euthanasia.

Kantian autonomy and the noumenal self

The Kantian argument may be interpreted in yet another way as a basis for rejecting PAS or euthanasia. Materstvedt and Bosshard (2009) develop such an argument when discussing the relationship between CDS on the one hand and PAS and euthanasia on the other. They argue that permanently removing a person’s consciousness, as in CDS, means a loss of valuable autonomy but also that this does not necessarily make CDS wrong. If loss of autonomy comes after autonomous consent has been given, the value of autonomy is properly respected. However, this does not apply to PAS or euthanasia. Materstvedt and Bosshard argue that even though there is no important difference from the point of view of the patient between being permanently unconscious and dead, there is still an ethically relevant difference which they account for by the Kantian notion of personhood. According to Kant, the world is divided into an empirical realm and a noumenal realm. The empirical realm concerns the world as it appears to us; it is a contingent world governed by the laws of nature. The noumenal world, on the other hand, is outside of time and is governed by reason, not by causation. Human beings have a part in the noumenal world, which makes it possible for us to act autonomously. Materstvedt and Bosshard write that CDS only affects the empirical person (the body), and although it makes autonomous action impossible, it does not take away the basic capacity for autonomy and thus leaves the patient’s noumenal self intact.

Interesting as this idea may be, it entails two main problems. First, the metaphysical assumption of two separate realms and two separate selves is metaphysically extravagant and quite difficult to make congruent with modern science. We believe it to be quite problematic to base normative assumptions on this kind of speculative metaphysics. Second, and more importantly, respect for autonomy here seems to imply that the important autonomy to respect is the autonomy of our noumenal self, not that of our empirical self. As Berlin (2002) points out, this does not necessarily imply respect for our actual wishes and desires. If our noumenal self is causally independent of our empirical self, how can it be threatened by anything we do to the empirical person? However, autonomy as it is usually understood in bioethics concerns our actual, not ideal, wishes and desires. It does not concern something located in another realm. It seems highly counterintuitive to claim that a person is free in a morally relevant sense if they would be deprived of all actual, empirical freedom. The question for anyone relying on such a notion of personhood is what respect for autonomy actually means, if it means anything at all.

Conclusions

Respect for autonomy is typically considered a central reason for allowing physician-assisted suicide and euthanasia. However, according to certain interpretations of the normative relevance of autonomy, these practices are unacceptable. According to one interpretation, autonomy is a prudential value to be promoted and protected. According to another, Kantian, interpretation, autonomy is an unconditional value, which leaves us with a duty to remain autonomous. Both these interpretations of autonomy give ground for claiming that PAS or euthanasia may be wrong in some cases, but not all but not all. If we suppose that autonomy is a prudential value, respecting current autonomous decisions must be weighed against the value of future autonomy. In a Kantian approach, reasons against PAS or euthanasia motivated by concern for suffering can be complemented by pro-arguments based on respect for autonomy and dignity. We conclude that neither of the two interpretations of the value of autonomy that we have discussed can be used to show that all cases of assisted suicide or euthanasia are wrong.