The Paradox of Duties to Oneself

ABSTRACT

Philosophers have long argued that duties to oneself are paradoxical, as they seem to entail an incoherent power to release oneself from obligations. I argue that self-release is possible, both as a matter of deontic logic and of metaethics.

KEYWORDS:

• duties to oneself
• moral obligation
• reasons
• rights
• self-obligation
• promises to oneself

Disclosure Statement

No potential conflict of interest was reported by the author.

Notes

1 Precursors include Kant and Hobbes (see section 3).

2 Wick [1960: 161–2, 1961] introduced this response, with Mothersill [1961: 207] noting that, for him, ‘duties to oneself … must be of a completely different order from duties of the familiar sort’. See also Knight [1961] on non-legal duties and Eisenberg [1968] on social contractual duties. Paton [1990: 225–6] argues for ‘non-contractual’ self-obligations, entailing neither rights nor release; she also concedes that we can’t make binding self-promises. Hills [2005: 138] agrees about self-promises, rejecting juridical duties to oneself. These authors’ inspiration is Kant [MS 6:383, LE 29:117, 632]. (I quote from the Cambridge editions and use pagination from the Akademie Edition; ‘MS’ is for The Metaphysics of Morals; ‘LE’, Lectures on Ethics.)

3 Kant argues that non-juridical duties require us to act from the right motive, but we can’t be coerced into having nobler motives: when coercion works, it is the motive! (See [MS 6:219].)

4 On whether Kantian duties to oneself require ‘two selves’, see Kant [MS 6:417], Denis [1997: 334–5], Potter [2002: 375n8], Timmermann [2006: 509, 512], Reath [2006: 356], Kerstein [2008: 206], and Schofield [2015: 514n24].

5 Kading [1960: 155] gives two more cases where promisees lack powers of release: one where they have also promised us not to release us; and another where they transfer right of release to a third party. Kading’s [ibid.: 356] fourth case is quite different; it is meant to show that you might be unable to release me from my duties merely regarding you.

6 Singer’s [1963: 134] main objection is that, contingent inability aside, the idea of self-release is still ‘self-contradictory’. I don’t know what to make of this point (but I will return to it in section 3). He has three other replies to Kading: (1) we wouldn’t say that the promisee post-mortem has a right; (2) someone else might have the power to release the promisor after the promisee dies; and (3) the promisee has the power to release pre-mortem [Singer 1963: 133–5]. The first point isn’t persuasive (why can’t the dead have rights?), and the others are irrelevant. For discussion, see Hill [1991].

7 Schofield [forthcoming] extends the account to synchronic duties.

8 Note that Schofield [2015: 13] doesn’t rest his argument on any metaphysics of future ‘selves’ as momentary agents. On the contrary, he assumes that people endure over time.

9 On release from self-promises, see Hill [1991], Habib [2009], Rosati [2011], and Fruh [2014].

10 Kant, too, says that a ‘contradiction’ would follow if ‘the one imposing obligation (auctor obligationis) could always release the one put under obligation (subiectum obligationis) from the obligation (terminus obligationis)’ [MS 6:417].

11 Singer has no other objections to self-promises, besides a suggestion that talk of self-promising could be interpreted as a metaphor for ‘settled determination’ to act [1959: 203].

12 Here’s a third way to hear Singer’s objection: we can’t release ourselves from obligations because they are categorical—they bind us, no matter what we want (cf. Wick [1960: 161] and Hills [2003: 133]). (By contrast, hypothetical norms, like ‘You have to hurry, if you want to be early’, bind us only if we want a certain end.) But no one thinks that we can erase obligations just by wanting out of them. Duties to oneself could be categorical—that is, not based in desires—while still being sensitive to the choice to release oneself. Analogously, for Hobbes, the laws aren’t grounded in what the lawmaker wants, but they do depend on what he chooses to repeal.

13 The arrow is the ‘accessibility relation’. In S4 violations, the relation is non-transitive, as it is in our case. (To fully depict the case, we would add arrows from w₃ to w₂ and w₁; from w₂ to w₁; and from each world to itself.)

14 We would have to add the same arrows as before for a full description (see note 13).

15 Hills [2003: 131] presupposes that duties are reasons, in asking ‘When should reasons for action be classed as “duties to the self”?’ (See also Schofield [2015: 10].) For Reath [2006: 241], a duty is a reason owed to whoever ‘is the source of reasons for one to act.’ Fruh [2014: 65] claims that we ‘expect obligations to be weighty things that well-reasoning, conscientious, non-akratic agents will have to make room for in their deliberations’—that is, reasons.

16 Of course, whether the duty is in play also depends on what you do, since it’s up to you whether to release me. But if you are perfectly happy to grant a release, the presence of the duty will depend on my choice of whether to get released, and the duty becomes unstable for me, so it won’t itself give me a reason to pick you up as promised. (I might still have other reasons to pick you up.)

17 Nagel writes: ‘to have a reason to promote an end, one must expect that there will be a reason for it whether one undertakes to promote the end or not’ [1970: 53].

18 The same goes for waiving rights against oneself, if A’s duty to B = B’s right against A.

19 Steiner [2013: 240n21] writes: ‘I cannot have rights against myself; I cannot be both plaintiff and defendant in a legal suit nor, presumably, in its moral counterpart.’ By contrast, Gilbert (2018: 177–8) thinks that I may enforce any of my commitments (e.g. promises) by making demands of myself. My view is that self-enforcement is usually silly because it is otiose. Why not just do the act?

20 My thanks to two referees at this Journal for insightful comments that enormously improved the paper, especially its last two sections. (Without them, I would not have included the Stability principle, and the argument would have been much weaker.) I am also grateful to the editors for their patience and excellent advice. For written comments, I thank Tamar Schapiro, Kieran Setiya, and Caspar Hare. For helpful discussion, my thanks go to David Balcarras, Jacqueline Broad, Thomas Byrne, Matthias Haase, Ginger Schultheis, Brian Hedden, Nathaniel Baron-Schmitt, Paul Schofield, Quinn White, Julia Driver, Conni Rosati, Oli Rawle, and Jack Spencer.