Duties to Oneself and Their Alleged Incoherence

ABSTRACT

Duties to oneself are allegedly incoherent: if we had duties to ourselves, we would be able to opt out of them. I argue that there is a constraint on one’s ability to release oneself from duties to oneself. The release must be autonomous in order to be normatively transformative. First, I show that the view that combines the division of the self with the second-personal characterization of morality is problematic. Second, I advance a fundamental solution to the problem of the incoherence of duties to oneself, one that does not rely on any division of the self, temporal or otherwise. I build upon the prevalent idea that, in releasing others from duties, we exercise the power of consent. The transformative force of consent partly derives from our autonomy. Invoking a plausible characterization of autonomy, I argue that release from duties requires the right kind of mental state.

KEYWORDS:

• duties to oneself
• release from duties
• second-personal
• diachronic duties to oneself
• self-consent

1. The Challenge

A familiar objection to the possibility of moral duties owed to oneself runs as follows. Duties are by their nature such that, if one has a duty, one cannot release oneself from it (P₁). If there were duties to oneself, these would be duties from which one could release oneself at one’s discretion (P₂). Hence, the very notion of a duty to oneself is internally incoherent [Singer 1959: 202–5]. Henceforth, I refer to this objection as ‘the Challenge’ (for discussion, see Kading [1960], Wick [1960, 1961], Mothersill [1961], Knight [1961], Singer [1963], and Cholbi [2015]; for earlier formulations, see Hobbes [1651: 184] and Kant [1797: 6: 417]).

Each of (P₁) and (P₂) is taken to be uncontroversial. (P₁) follows analytically from the very concept of a duty, since it is in the nature of a duty that it is binding. If it were possible for one who has a duty to release oneself at one’s discretion from this duty, such a duty would not be binding in any meaningful sense. As Marcus Singer [1959: 202] puts it, ‘[i]t is essential to the nature of a duty that no one can release oneself from it by not wishing to perform it or by deciding not to perform it, or, indeed, in any other way whatsoever’.

As for (P₂), it follows from the contention that it is possible to be released from a duty by its object—namely, the person to whom φ-ing is owed. It follows that if A has a duty to A to φ, then, even though A qua subject of that duty cannot release herself from it on pain of contradiction, A qua object of that duty can do so. Since A qua subject and A qua object of said duty are identical, A can release herself from the duty to herself.

The literature discusses a number of counterarguments (or instance, Hills [2003], Rosati [2011], Cholbi [2015], Oakley [2017], Muñoz [2020]). For example, arguably neither (P₁) nor (P₂) is universal in scope. Alison Hills [2003] invokes the intuition that some rights are inalienable and argues, contra (P₂), that duties correlative to inalienable rights are non-releasable, whether they are owed to others or to oneself. Tim Oakley [2017] argues, contra (P₁), that no matter who the object of a duty is, any agent qua subject of that duty could release herself from it. Given our pro tanto moral right to self-direction, by settling on a course of action, we can change what we ought on balance to do, thereby releasing ourselves from what would otherwise have been our all-things-considered duty.

The Challenge also begs the question. Connie Rosati [2011] notes that whether we can release ourselves from duties to ourselves by giving up rights depends on what grounds those rights. She suggests that our rights are grounded in the agent-neutral value of persons, which no person can consistently disregard. Hence, generally, we cannot release ourselves from duties that have their source in the normative status of persons by giving up our rights [ibid.: 128–31, 134]. Alternatively, Michael Cholbi [2015] claims that the Challenge question-beggingly presupposes that one cannot be accountable to oneself.

Another response stands out among the others. Paul Schofield [2015] dismisses the Challenge on the ground that moral duties arise as a result of a legitimate (actual or hypothetical) issuing of claims or demands. One could have a duty to oneself to φ if a hypothetical justified claim is issued from one perspective occupied at time t to another perspective at time t′. The resulting duties are immune to the Challenge because an agent could waive a valid claim only from the perspective from which the claim had been issued. Yet, in order to be able to exercise that power, the person has to stand in the actual perspective. Since it is impossible to waive a claim hypothetically, Schofield argues, diachronic duties to oneself are invulnerable to the charge of incoherence [ibid.: 521].

I maintain that the Challenge consists not in the fact that we can release ourselves from purported duties to ourselves, but rather in the presupposition that we can do so by simply opting out of them: that is, that no basis or justification of sorts is required for doing so (cf. Rosati [2011] and Schaab [2021]).¹ My objective is twofold. First, I show that Schofield’s characterization of duties to oneself as irreducibly second-personal and diachronic is problematic in its own right. Consequently, it can neither provide a solution to the Challenge, nor lend support to further ideas of synchronic second-personal duties to oneself (section 2). Second, I follow through on the prevalent idea that, in releasing others from duties, we exercise a normative power [Rosati 2011; Muñoz 2020; Schaab 2021]. I argue that there is a constraint on one’s ability to release oneself from duties to oneself: the release must be autonomous if it is to be normatively transformative. The power to release is akin or equivalent to the normative power of consent, and its transformative force derives partly from the value of personal autonomy. A person whose decision to release an agent from a duty is non-autonomous undermines that very feature in virtue of which the power to release is normatively transformative (section 3). Thus, a fundamental solution to the Challenge emerges that does not appeal to any division of the self, temporal or otherwise.² Afterwards, I defend my solution from two possible lines of criticism.

2. The Second-Person Standpoint and Duties to Oneself

Schofield [2015] argues that duties to oneself are immune to the Challenge from internal incoherence. On his understanding, these duties arise as a result of a hypothetical second-personal issuing of claims between perspectives separated in time.³ One can have a duty to oneself to φ if a hypothetical justified claim is issued from one perspective occupied at a later time t to another perspective at an earlier time t′. When a duty is generated in this way, the agent cannot release herself from it, because one can waive a valid claim only from the perspective from which it was issued. However, in order to be able to exercise the power of release, an agent has to stand in the actual perspective. As a result, one can never release oneself from diachronic duties to oneself [ibid.: 521].

Schofield [ibid.: 511] arrives at his account by extrapolating from Darwall’s [2006] view of morality as irreducibly second-personal. He asserts that, in issuing a valid claim, it is not the metaphysical distinctness of the relating parties that matters, but rather the fact that these parties occupy different perspectives, and that the interests composing these perspectives conflict [Schofield 2015: 518–19]. This is allowed for by Darwall’s view [2006: 9]: ‘It is the perspective one assumes in addressing practical thought or speech to, or acknowledging address from, another … and, in so doing making or acknowledging a claim or demand on the will.’

However, the diachronic account of duties to oneself is problematic. First, it is vulnerable to its very own version of the non-identity problem. Second, it delivers counterintuitive implications whenever the choice must be made between doing what will result in a greater aggregate benefit for a cluster of perspectives, and doing what will produce a greater benefit per perspective.

A moral doctrine faces the non-identity problem when it implies that the victim of a wrongdoing lacks grounds to complain because the very act by which she was purportedly made worse off also caused her to exist and thereby benefited her. So long as the victim’s life is worth living, the benefit of existence outweighs the loss in well-being caused by the same act, thus rendering incoherent the victim’s complaint that she has been morally wronged [Parfit 1984: ch. 16]. According to Schofield, one can have a duty to oneself to φ if a hypothetical justified claim is issued from one perspective (P) occupied at time t to another perspective (P’) occupied at time t′. The wrongful action from perspective P causes perspective P to exist. Can we make sense of the complaint issued from perspective P at time t?

Imagine 15-year-old Bob who decides not to exercise, leading to his being mildly overweight at the age of 60. Although from the perspective of mildly overweight 60-year-old Bob, life is worth living, his weight causes minor health issues, and prevents him from going hiking with his friends. Counterfactually, if 15-year-old Bob were to choose to exercise, this would lead to him later being a fit 60-year-old who is able to enjoy beautiful sunsets with his mountaineering friends. From the perspective of mildly overweight 60-year-old Bob, it was morally wrong of 15-year-old Bob to decide not to exercise. Yet this decision caused the perspective of mildly overweight 60-year-old Bob to exist, at the expense of the perspective of the very fit 60-year-old Bob.

On the original formulation of the non-identity problem, one is wronged by being made worse off [Kumar 2003: 99–118]. However, in the second-person framework, being made worse off is not a necessary or sufficient condition for being wronged. Thus, even though 60-year-old Bob might have been made worse off by his younger self, it does not necessarily follow that he has been wronged by it.

Schofield does not tell us which valid moral principle guides an intra-personal second-personal relation. It is natural to suppose that, by adopting Darwall’s moral framework, Schofield also shares Darwall’s contractualist inclinations. Indeed, Schofield points in the direction of a contractualist moral principle. Inspired by Rawls’s observation that utilitarianism does not take seriously the separateness of persons, Schofield [2015: 524] suggests that we should take seriously the distinction between perspectives:

Some interests and ends … attach to temporal perspectives, and a contribution to the good of the person overall will not necessarily constitute a contribution to the good enjoyed from each of those perspectives. Indeed, advancing the good of a person overall might require significant suffering from some perspectives.

This suggests that the relevant moral principle must be justifiable to each perspective. If so, then Schofield’s account is vulnerable to the non-identity problem, after all. For how can we make sense of the complaint that 15-year-old Bob wronged himself from the perspective of mildly overweight 60-year-old Bob, if this particular perspective did not even exist when the offending act occurred?

Schofield might reply that the ends and interests attached to each perspective are not sensitive to the psychological identity of those perspectives. These are ends and interests that someone is expected to have from any perspective understood as a type. They are independent of the identity of any perspective to which they are attached, and are to be respected no matter what one’s actual perspective is. Even if the perspective of mildly overweight 60-year-old Bob did not exist at the time of the offending act, 15-year-old Bob could already know what he owes to his future self.

This renders the account akin to a spare wheel on a vehicle—one that does not do any work [Hooker 2002: 53–76]. If what we owe to ourselves is not sensitive to the identity of the perspectives from which the claims are issued, but is fixed in virtue of, say, our status as persons, then it is not the fact of a claim that explains what moral duties we have, but the moral principle of respect that we owe to ourselves as persons. Second-personal address does not explain what makes an action wrong, but presupposes its wrongness.

Another problem concerns a choice between doing what will result in a greater aggregate benefit for a cluster of perspectives, and doing what will produce a greater benefit per perspective.⁴ Schofield claims that we should take seriously the separateness of perspectives and the distribution of good among them within a person’s life. He also claims that if his account is correct, then it indicates the need to re-evaluate the basic assumption that any person is free to balance her gains against her losses [2015: 526]. In determining which perspective we should give weight to, we cannot consider the goods or losses of many perspectives added together. Instead, we are to compare pairwise the reasons for objecting to a choice of an action that hold from one perspective, with the reasons for objecting to an alternative choice of action that hold from another perspective.

Imagine that at a particular time you should undergo surgery causing 100 units of pain. If you don’t, you will suffer chronic pain for the rest of your life, the total units of which would amount to 600. Arguably, it is permissible and perhaps even rational, for you to submit to the surgery now in order to avoid 600 units of future pain. Imagine further that your life is divided into 10 perspectives, and that, from the point of view of each, there will be only 60 units of pain to suffer. Your current perspective, taken individually, would object to suffering 100 units of pain. At the same time, if you do not undergo the surgery, each and every one of your 10 future perspectives would object to suffering 60 units of pain. We arrive at the counterintuitive result that it would at most be supererogatory for you to have the operation. This perplexing situation would not arise if you considered the costs and benefits of surgery from the point of view of your life overall. However, on Schofield’s view, prudential considerations are constrained by duties to oneself [ibid.: 525–6n47].

In sum, the diachronic account of duties to oneself is problematic in its own right and does not provide a tenable solution to the Challenge.⁵

3. Release and Opting Out

The crux of the Challenge consists not in the claim that we can release ourselves from duties to ourselves, but in the presupposition that we can do so by simply opting out of them. The rationale behind this contention derives from the presumed analogy between duties to others and duties to oneself. If genuine duties to oneself exist, they must be similar in relevant respects to duties to others. Since the objects of duties to others have the power and discretion to release their subjects from those duties, the same must hold in the case of duties to oneself.

I will argue that the explanation for why we can release others from their duties to us suggests that the release must be autonomous in order to be normatively transformative. If this is correct, then there is ipso facto a non-arbitrary line between releasing oneself from a duty to oneself and violating it. Consequently, if we have releasable duties to ourselves, it is possible to fail to release ourselves from them.

First. whenever we release others from their duties towards us, we immediately bring about a normative change without necessarily altering the balance of existing considerations.⁶ The desired change typically transpires from a speech act, expressly intended to bring about this very change. This is how theorists conceive of a normative power (see, for example, Raz [1999: 464–5] and Owens [2012: 4–5]; cf. Enoch [2014: 302–3]).⁷

Second, the normative power of release is equivalent or akin to the normative power of consent. Both permit doing what it is otherwise wrong to do. The normative change resulting from the use of either power does not depend on the uptake by the released party or by the consentee. Intuitively, the conditions for valid consent are the same as those for valid release. For example, one would not be ‘off the hook’ if one were to be released from one’s obligation by a drunk or confused person. Much like consent, ‘release’ is an umbrella term that refers to a range of actions (speech acts included), the recognized intrinsic result of which is the making-permissible of what was otherwise impermissible (for others who implicitly assimilate these powers, see, for example, Tadros [2011], Owens [2011], and Dougherty [2015]).

Third, the ground of the normative power of release/consent is in the content-independent value of individual autonomy. This means, on the one hand, that normatively transformative release/consent does not require that some human interest be served by it. On the other hand, it requires the object of a duty to have a specific, autonomy-conferring, mental state.

Consider the former claim first. On the view that it denies, the normative work is done by a combination of the agent’s will and some human interest (both necessary and jointly sufficient) that is served by the release/consent. It would then follow that, without any clear interest of mine being served by my releasing a person from her duty toward me, I fail to turn an impermissible act into a permissible one. However, I can release another person from, say, her promise to me, even if no interest of mine is served by it.

One might find this conclusion too quick. Perhaps the normative power of release/consent serves our interest in controlling what others do to us [Owens 2012: 166ff.; Tadros 2016: 213ff.]. Whenever someone else’s actions coincide with what we want, it is pro tanto good for us. Nevertheless, control interest is but one of many human interests and, as such, it competes for our attention. Hence, we should be asking whether release serves our overall interest. Again, the answer is ‘no’, because I have the power to release others from their duties to me, whether or not the release benefits me overall [Schaber 2018].

This brings us to the following point. If release from a duty can be normatively transformative despite its going contrary to our interest, then it must be our volition that effects the change. If so, then it is natural to suppose that the explanation for volition’s normative authority lies in the distinctive structure of our will and the significance that we attach to it. Specifically, it lies in the fact that we are autonomous agents capable of reflecting upon and endorsing or rejecting our first-order motivating states—the feature that is widely regarded as an object of value.

In valuing autonomy, it is not the mere capacity that we value. After all, one can possess it without exercising it. Rather, we value its exercise. I suggest, first, that the normative power of release/consent enables us to exercise our capacity for autonomy. Second, this power is normatively transformative in virtue of the fact that it serves that purpose.

In view of this, I propose that there must be a constraint on one’s ability to release others and, by analogy, ourselves from duties owed to us: the release(/consent) must be autonomous in order to be normatively forceful. If a person’s decision to release an agent from a duty is non-autonomous, then she undermines that very feature in virtue of which our power to release/consent is normatively transformative.

On a widely shared view, autonomy captures something significant about the relationship between one’s psychological states and actions. An agent who is competent⁸ and who has a sufficient range of options is autonomous with regard to her decisions and actions only if she holds an attitude of reflective endorsement toward them.⁹ If the consideration of autonomy curbs the successful use of our power of release/consent, then we can further specify the constraint as follows: release/consent requires an autonomy-conferring attitude of reflective endorsement.

This conclusion is significant for the Challenge because it shows that there is a principled basis behind the use of our normative power of release. One might fail to render permissible what it is otherwise morally wrong to do, and hence it is incorrect to think that one can simply opt out of duties to oneself. Furthermore, unlike previous solutions to the Challenge, this conclusion does not imply unpalatable views about duties to oneself: it neither stymies theorizing about such duties by implying that they are rare concomitants of inalienable rights, nor suggests that, unlike duties to others, duties to oneself are such that one can never be released from them.¹⁰

For the purposes of illustration, assume that we have the prima facie duty not to cause (or risk causing) ourselves gratuitous harm. Consider the following scenario. Bob is a medical student who wishes to be taken seriously by his senior colleagues. Occasionally, however, Bob’s insecurity gets the better of him and he seeks his colleagues’ approval by means that conflict with his considered judgment. In a fit of bravado, Bob injects himself with an experimental drug, which both causes him severe and lasting migraines and creates a risk of serious side effects. Bob knew about the risks and understood that the medical community couldn’t cheer for such recklessness and unprofessionalism. Hence, he acted non-autonomously and perhaps even regrets what he did. Compare Bob to Saloni—the leading scientist in the project—who injects herself with the same drug but for different reasons. Saloni believes herself to be on the verge of a scientific breakthrough. She has just one test left to make but cannot bear to wait to find a volunteer to test the drug. In full awareness of the likelihood of harm, she decides to test the drug on herself.

The difference between Bob’s and Saloni’s actions is that the former failed to comply with his duty not to cause himself gratuitous harm while the latter released herself from it. Moreover, it is independently plausible to think of Saloni’s action as a release from her duty to herself. In full awareness of the potential harms, she chooses to act on her higher-order desires and commitments. If asked to explain her action, Saloni will probably appeal to the importance and timeliness of the potential scientific discovery, as well as to its potential significance for her career and sense of self-esteem. Bob’s action seems more plausibly characterized as a failure of sorts. If confronted, he would cringe at his own action, admitting that he was overcome by his own insecurities about his in-group status and that he achieved the opposite of what he wished to achieve. In a sense, Bob failed to act on his will.

Two things are important to emphasize. First, Bob’s failure to successfully render gratuitous self-harm morally permissible due to a failure of his local autonomy does not imply that he generally lacks the normative power to do so. A failure to give valid consent does not imply that the agent lacks the power to release/consent. Second, and more importantly, we need not deny our power of release in order to vindicate duties to ourselves. The Challenge falsely presupposes that acting against a duty to oneself just is to release oneself from it, because there is no non-arbitrary line between releasing oneself from a duty to oneself and violating it. However, unlike violating one’s duty to oneself, releasing oneself from it must be autonomous. Consequently, the core assumption behind the Challenge is mistaken.

3.1. Autonomy and Responsibility

The suspicion may creep in that the advocated view implies that we are never morally responsible for violating our duties to ourselves. If the view suggests that to autonomously act contrary to a duty to oneself just is to release oneself from it, then it follows that a violation of such a duty is always non-autonomous, and therefore is not something for which one is responsible. The resulting view would then win the battle at the cost of losing the war.

There are two aspects of this worry—the presupposition that one cannot violate a duty to oneself autonomously, and the contention that an agent is never morally responsible for her non-autonomous morally wrongful actions, either because she is not a responsible agent in general or, if she is, because she is excused for violating them. Let me address these issues in turn, starting with the former.

If a release from a duty (or consent) must be autonomous in order to be normatively transformative, then autonomy is necessary for release/consent. This says nothing about whether autonomy suffices for it. While I maintain that acting non-autonomously contrary to a duty to oneself amounts to a violation of that duty, from this it does not follow that, when an agent autonomously acts contrary to her duty to herself, necessarily she thereby releases herself from it.¹¹

Speaking of duties to oneself, whether autonomy suffices to release oneself from them depends on the correct first-order account of such duties, and on how the value of autonomy interacts with the considerations that ground them. To offer an account of duties to oneself is a task for another paper. However, the idea that autonomy is not just necessary but also sufficient for release/consent is controversial.

Many theorists take autonomy to be a significant source of our moral status and the ground of our duty to treat others and ourselves with respect. If so, then it is an especially serious moral wrong to destroy or significantly undermine one’s autonomy. Even if it is psychologically possible for a person to be autonomous in her decision to start taking heroin, or to set herself on fire, it might not therefore be permissible for her to do so. Plausibly, permissible self-harm must be inflicted for a good reason (see, for instance, Tadros [2011: 31–47]). For example, it might be morally permissible for me to save a battalion of soldiers even if I would suffer a serious concussion and would be left with diminished cognitive capacities. It is doubtful, however, that it would be morally permissible for me to suffer similar harm in order to save somebody else’s finger. Autonomy is not the only source of moral duties. Well-being is another morally salient consideration that grounds moral duties. Granted, the consideration of an individual’s autonomy typically outweighs the consideration of her well-being: it is widely thought to be morally permissible for others to harm me with my consent. Yet it might still be incorrect to conclude that autonomy thereby suffices for releasing from duties grounded in the moral salience of well-being. Some forms of self-harm can invalidate consent. For example, it might never be morally permissible for someone to kill me, even if I purport to consent to it [Bergelson 2008; Tadros 2011].

The claim that autonomy suffices for release/consent goes against widely shared intuitions. Showing that one cannot autonomously violate a duty to oneself requires showing these intuitions to be mistaken. If the above considerations are compelling, then this burden of proof is not mine.

Consider now the contention that an agent is never morally responsible for her non-autonomous morally wrongful actions, either because she is not a responsible agent in the first place or because she is excused for violating them. One can fail to act autonomously in various ways. At one end of the spectrum, we have maximally locally non-autonomous acts, such as occasional akratic, negligent, or confused actions. Moving towards the other end, we find progressively broader failures of autonomy. These might include actions manifesting general confusion about one’s ends and values, compulsive actions, and actions caused by brainwashing, coercion, hypnosis, or systematic manipulation. Finally, globally non-autonomous agents lack the basic competencies necessary for autonomy, such as the capacities to critically reflect on one’s motives (including one’s organizing values and commitments) in order to form intentions, and to effectively realise them.

A globally non-autonomous agent is not a proper subject of moral requirements because the capacities necessary for autonomy are also necessary for being a responsible agent—that is, for being a sensible target of aretaic appraisal and reactive attitudes. Moreover, the competencies necessary for autonomous action are arguably not sufficient for moral responsibility. Susan Wolf [1987] argued that being a morally responsible agent requires more than the ability to act on motives that one endorses on the basis of one’s values. One must also have moral competence—that is, the capacity to recognize and respond to moral considerations. Someone like Jojo, the son of a dictator, who has a mind perverted by the circumstances of his upbringing, is not a morally responsible agent because he lacks moral competence. Jojo might endorse his father’s ghastly value system, and hence it is not impossible for him to act autonomously. Still, some have the intuition that Jojo cannot be regarded as a morally responsible agent because he lacks moral competence.¹²

On most occasions, however, a person who non-autonomously acts contrary to her duty to herself will be autonomous enough to qualify as a responsible agent in a general sense. Common cases involve failures of local autonomy due to weakness of will, negligence, or confusion about one’s motives. Furthermore, it would be implausible to think that a failure of local autonomy always provides an excuse for particular violations of duties. Consider first the interpersonal context. We do not excuse a weak-willed adulterer for cheating on their spouse.¹³ If a person negligently forgets an important promise made to her friend, we do not take her to be excused for doing so. The same holds for intra-personal moral wrongs. Linda, a diabetic craving a cake, might tell herself ‘What harm is there in my eating a piece or two?’ Linda is aware of the risks. Being both weak-willed and negligent of her health, she acts frivolously. When she later feels guilty, her guilt seems appropriate.

My argument does not imply that an agent is never morally responsible for violating her duties to oneself. In most cases of a non-autonomous violation of a duty to oneself, an agent will be autonomous enough to qualify as a responsible agent. Admittedly, if an agent acts contrary to her duty to herself due to extreme duress, manipulation, or hypnosis, she will be excused for it. This, however, is not true of all, or even the most mundane, instances of non-autonomous wrongdoing.

3.2 Autonomy, Consent, and the Demandingness Objection

One might object that the above criteria for release/consent are too demanding. If release/consent requires autonomy, and if autonomy requires an attitude of reflective endorsement toward the choices that we make, then one cannot consent unless one has this autonomy-conferring attitude toward the state of affairs to which one consents.¹⁴ Generally, given that in high-stakes contexts consent can make the difference between permissible actions and grave wrongs, a more demanding account is preferable to a less demanding one. Nevertheless, it might still be too much to demand an attitude of reflective endorsement as the necessary condition of consent.¹⁵

Consider a person asking for yet another cosmetic surgery procedure with the self-proclaimed motive to look more attractive, while her real motive has nothing to do with how she looks but rather with how she wants to feel about herself (a goal for which surgery is not the right means). Finally, imagine that, upon introspection, she would be able to conclude that, whatever the gain in aesthetic value, it does not outweigh the risk. The standard model of autonomy implies that such a patient would fail to consent because (a) she is unaware of her true motive, and so her action and her motive are not suitably related, and (b) she ranks her desire to stay safe higher than her self-proclaimed desire for a more appealing face. Yet it seems counterintuitive that, if the surgeon informed the patient of the risks, and asked the right questions, she still acts culpably by proceeding to operate. The surgery, the objection goes, must be consensual.¹⁶

This objection rests on a failure to appreciate the epistemological aspect of consent [Guerrero forthcoming]. Consent plausibly requires the right kind of mental state.¹⁷ This presents the problem of knowing another person’s mind, and points to the gap between the objective permissibility of an action and the culpability of an agent who acts as if the other consents.

On the one hand, it is possible that A has the consenting mental state, but that B is nevertheless culpable for acting as if A consents, because B lacks a justifiable belief that A has the consenting mental state with regard to the object of consent. By acting as if A consents, B takes an objectionable moral risk, provided that it is permissible for B not to act at all.¹⁸ On the other hand, it is possible that A does not have the consenting mental state, yet B is not culpable for acting as if A consents, because B justifiably believes that A has the consenting mental state with regard to the object of consent. In our cosmetic surgery example, probing questions might help the surgeon to notice cues suggesting that the patient is confused about her motives. The surgeon would then act culpably by proceeding to operate, as her belief that the patient has the consenting mental state is unjustified.

Two issues are worth mentioning. First, one cannot avoid gauging the consenter’s mental state by insisting that consent also requires a performative act. Instead of signifying the consenting mental state, the performative act might be the product of hypnosis or deception. This does not mean that the performative act is unnecessary for consent. For example, in his response to Alexander [2014], Richard Healey [2015] argues that control over spheres of life protected by the value of autonomy is not the only function of consent. Consent also regulates normative relations between individuals on the basis of mutual recognition and respect. For example, by actively seeking another person’s consent before attempting to have sex with them, we demonstrate that we recognize their authority over their body. Similarly, it is important for each of us to be seen as someone who recognizes other people’s control over their bodies. To play this role, consent must be explicit.

Second, given that consent might make the difference between permissible action and a grave moral wrong in high-stakes contexts, a more demanding view is preferable to a less demanding one. The epistemology of consent supports the idea that moral context matters. Guerrero [2007, forthcoming] convincingly argues that whether B is non-culpable for acting as if A consents depends, in part, on the moral context [2007: 69]:

The more morally significant the actions that a belief in p (or absence of a belief in p) will support or license, the more stringent the epistemic demands that must be met before one can act as if one is justified in believing that p.

Moral stakes can affect the scope of action undertaken on the basis of the belief that p. Consider the belief that there is nobody inside some particular abandoned house. In order to arrive at this belief, I knocked on the door, peeked inside, asked people who live nearby, and even returned and did the same a week later. These joint measures justify my belief that there is nobody inside the abandoned house. I can legitimately appeal to this belief in my effort to calculate how many people live in town, but I cannot appeal to it in order to justify the demolition of the house.¹⁹ Similarly, in the cosmetic surgery case, having asked the patient the right questions, the surgeon may come to believe that the patient has the consenting mental state. Yet, given the risks involved and, ex hypothesi, the lack of obvious need for the procedure, this might not be enough to justify proceeding to operate, provided that it is permissible for the surgeon not to operate. Perhaps a consultation with a psychologist is incumbent upon the patient, and the surgeon who fails to request this is culpable either of unexcused consent violation or of taking an objectionable moral risk.²⁰

In sum, if consent requires a mental state, and if in high-stakes contexts consent makes the difference between permissible and wrong action, it does not seem excessively demanding for consent to require an attitude of reflective endorsement. The difficulty of knowing another mind accounts for the discrepancy perceived between the objective moral status of an action and the culpability of an agent who acts as if there is (or isn’t) consent.

This epistemological aspect of release/consent reveals the important difference between releasing others from their duties to us and releasing ourselves from our duties to ourselves. Typically, we enjoy privileged access to our own mental states. This explains why we often feel regret and guilt after we fail to act in accordance with our better judgments, values, and commitments. Self-release involves a decision based on the understanding that there are significant morally relevant considerations that balance out the justification behind a duty that currently applies. Regret and guilt are, arguably, signs that the agent knew, even if only semi-consciously, that there were no such considerations in place when she acted against that duty.²¹

4. Conclusion

In this paper, I have defended the idea of duties to oneself by rejecting allegations of its internal incoherence. It is alleged that the very idea of duties to oneself is self-contradictory: if one had a duty to oneself, one could simply opt out of it. This supposedly follows, by analogy, from the fact that we have the power to release others from their duties to us. As the problem arises partly because the subject and the object of a duty owed to oneself are one and the same person, some have sought to solve it by relying on temporally distinct perspectives within a person’s life. I have argued that this approach suffers fatal flaws.

A more fundamental solution to the objection from internal incoherence must involve the analysis of the normative power of release. This power is equivalent to the normative power of consent, and its transformative force is at least partly grounded in the value of autonomy. There is a constraint on one’s ability to release oneself from duties to oneself: the release must be autonomous in order to be normatively forceful. A person whose decision to release herself from a duty is non-autonomous fails to effect the intended normative change. Consequently, if we had releasable duties to ourselves, it would nevertheless be possible to fail to release ourselves from them, since not all attempts to exercise our power of release are licensed by that very power.²²

Disclosure Statement

No potential conflict of interest was reported by the author.

Additional information

Funding

I benefitted from the Swiss Government Excellence Scholarship and the Lund Gothenburg Responsibility Project (LGRP) funded by the Swedish Research Council.

Notes

1 Schofield [2021] claims that, on its strongest reading, the Challenge tells us that duties to oneself are not binding. However, from the fact that an agent can release herself from a duty to herself, it does not follow that this duty is not binding in the first place (see, e.g., Muñoz [2020]). Schofield presupposes that, since our power to release is located in our will, it is therefore unconstrained. In this paper, I show this presupposition to be unwarranted.

2 Compare Rosati [2011] on self-promise. Unlike Rosati, who relies on the notion of rights against oneself, I analyse our power to release others from their duties to us. This avoids conceding the existence of rights against oneself. One could either claim that the system of correlation between rights and duties, originally adopted by Hohfeld [1917] by analogy with legal relations, should apply solely to the inter-personal domain of morality, or one could accept the system of correlation, but argue that rights against oneself are epiphenomenal.

3 Schofield allows for the possibility that not all moral reasons are second-personal [2015: 514n23].

4 Muñoz [2020] has recently made a similar observation.

5 Schofield [2019] and Schaab [2021] propose that an agent can engage in synchronic intra-personal second-personal moral address from the perspectives of her conflicting practical identities. I don’t have the space to discuss this idea in detail. I think, though, that such proposals overstate the phenomenon of inner conflict. Typically, agents strive to harmonize their interests by assigning varying importance to their projects based on their settled conceptions of the good. More importantly, similarly to the diachronic account, such views face the spare wheel objection. Their proponents owe us an account of what makes a claim issued from a perspective of a certain practical identity legitimate in a way that does not render the fact of that address superfluous. Schaab’s [2019, 2021] idea that we can acquire duties to ourselves by making commitments, and that it is the commitments that make claims issued from certain perspectives legitimate, is appealing. However, the resulting view then seems to account only for commitment-based duties to oneself. If I have a duty not to cause myself unnecessary harm, then, unless it can be shown to be incompatible with some commitment that I made to myself, the Challenge retains traction.

6 In some cases, such as when allowing a stranger to enter my home, I remove the only existing reason not to φ. In other cases, release does not change the overall balance of reasons for or against φ-ing. Instead, this generates permission to disregard conclusive reasons to the contrary [Raz 1999: 85–106].

7 Schaab [2021] holds that the bare fact that, in releasing from a duty, we exercise a normative power debunks the contention that, in acting contrary to a duty to oneself, we thereby release ourselves from it. He seems to agree, however, that this formal distinction between a release from, and a violation of, a duty makes for a poor case for duties to oneself.

8 An agent is competent if she is able to (a) critically reflect on her basic motivating elements as well as her organizing values and commitments, (b) form intentions to act on their basis, and (c) effectively realize her intentions in the absence of obstacles [Christman 2009: 155].

9 This is meant to capture the following conditions of autonomy. (1) The right kind of relation must obtain between the agent’s actions and her first- and higher-order motivating states: (a) motives must explain actions qua reasons for action [Davidson 2001: 12–25], and (b) actions must be motivated by desires that correspond to the agent’s higher-order motivating states [Frankfurt 1988: 11–25]. (2) An agent’s higher-order motivating forces must be caused or formed without coercion, manipulation, or any other activity eroding the agent’s authorship over her own life [Christman 1991]. (3) Neutrality with regard to the subject-matter and the evaluative status of the agent’s preferences: an action can be autonomous even if motivated by value judgments with which others disagree. (4) Autonomy requires that we be disposed to actualize the endorsement of our mental states and plans by subjecting them to critical reflection [Arneson 1994: 49]. (5) The presence of a sufficient range of varied and meaningful options [Raz 1986: 372]. Specific accounts of autonomy may not accept all five of these conditions. This is not the place to adjudicate between various conceptions of autonomy. For a concise statement of this model of autonomy, see Moles [2007: 56–61].

10 Schofield [2021] maintains that duties to oneself must be shown to be non-releasable. This is infelicitous, since it does not accord with the complexities of life. Suppose that I made a commitment to myself to pursue a career as an artist only to realize later that, while this is still my dream job, the external pressure is more than I can bear. I could then justifiably release myself from that commitment. Similarly, I can release myself from the (supposed) duty of self-care when some cause that I hold to be important requires me to go on a hunger strike.

11 See Tadros [2011] for an argument that validity of consent partly depends on its content. While I have been speaking of whether autonomy suffices for consent, others might be speaking of whether consent suffices for the permissibility of actions. This disagreement is verbal.

12 See Talbert [2012] for the position that it would not be unfair to blame an agent with impaired moral competence.

13 Admittedly, it is easy to interpret my position here as implying that a weak-willed adulterer does not consent to sex. This seems to recommend a less demanding criterion of consent. In the remainder of this paper, I argue against this suggestion. If the argument there is on target, then this suggests, I believe, that we need a more nuanced theory of consent—one that would allow for degrees of consent. Additionally, I am inclined to think that, while the parallel between consent and self-consent is useful and appropriate for the analysis of the Challenge, there are limits to how far it plausibly goes. So, I do not think that there is a bullet here for me to bite.

14 Others have suggested that consent requires the mental state of acquiescing [Westen 2016: 25–63], intending [Hurd 1996], or waiving one’s right [Alexander 2014]. My view comes closest to that of Guerrero [forthcoming], who argues that consent requires the mental state of affirmative endorsement.

15 My discussion concerns moral permissibility and should not be taken to imply anything about legal permissibility.

16 On the relation between autonomy and consent, see, e.g., Manson and O’Neill [2007: 21] and Beauchamp and Childress [2009: 100–1].

17 Some hold that a signification of agreement is both necessary and sufficient for consent [Malm 1996; Archard 1998: ch. 1; Wertheimer 2000], but this is now a minority view. For its difficulties, see Alexander [2014].

18 There are other complications. For example, B might be excused from taking the risk because the urgency of the medical context requires her to act quickly.

19 The example comes from Guerrero [forthcoming].

20 Some argue that consent cannot require autonomy, because psychological or verbal coercion does not necessarily invalidate consent. For example, if the threatened harm is not too severe, and the coercer is within his or her rights to make the conditional announcement (‘Either we have sex or I will break up with you!’), then the victim can still consent [Conly 2004]. I find this position counterintuitive. If someone acquiesces to sex because of a threat or coercion, the resulting sexual act is not consensual.

21 This is not to deny that the feeling of guilt or regret can be irrational.

22 I am grateful to Andrés Moles, János Kis, Emma Bullock, Brad Hooker, Paul Schofield, David Enoch, David Mark Kovacs, Máté Veres, the reviewers from Ethics and from this journal, and audiences in Zurich, Budapest, Reading, Copenhagen, and Gothenburg, Linköping.