1. The Variety of Duties and the Variety of “Mere Reasons”

Duties are a varied bunch, and the challenge is to find an account that can identify the underlying thread. In this Section I mention seven aspects of duties, and then ask which are unique to duties and which pertain also to nonduty “mere reasons”:

1. I. Most duties are pro tanto (i.e., defeasible): duties that in appropriate contexts can justifiably be infringed in order to ensure that more important things are done. Examples include many promissory duties and duties to do one’s best in one’s role. But in any context, there will also be some things one is, all things considered, duty-bound to do or not to do. These normally include duties not to kill, torture, assault, or show disrespect.Footnote ³

2. II. Some duties are of great importance, such as—again—duties not to kill, assault, torture, and further duties correlating with basic human rights.Footnote ⁴ Other duties are unimportant, such as trivial promissory duties, duties not to fuss and grumble, or duties created by time limits for parking spaces, deadlines for returning library books, or park bye-laws forbidding ball games.

3. III. We can distinguish duties by the form of normativity within which they fall. Some are moral, some prudential, some legal, and it is not clear whether or when these classes overlap. Are there distinctively aesthetic or epistemic duties?

4. IV. Duties can be borne by persons, by corporate entities (businesses, states, and clubs), and perhaps by some nonhuman animals (e.g., dogs and horses are given commands that arguably constitute duties). But it seems mistaken to see duties as borne by will-less entities like plants, rocks, or cars, or by abstract entities like the color blue. Further, there are many nonhuman animals with a will, but without the necessary capacities to bear duties: butterflies, deer, haddock, and spiders.Footnote ⁵

5. V. Some duties are permissibly enforceable, such as duties not to kill, assault, torture, contractual duties, and duties of emergency rescue, while others are not, such as duties of gratitude, of romantic fidelity, or duties not to fuss and grumble.Footnote ⁶ (In later sections, I will argue that there is a technical sense in which all duties are permissibly enforceable, and that this is a distinctive type of enforceability which “mere reasons” lack. But this is not our everyday notion of enforceability. In our everyday sense, duties not to kill are enforceable, but duties of gratitude are not).

6. VI. Some duties are “directed” or “owed to” others, such that their violation is not just wrong but wrongs a particular party.Footnote ⁷ Examples cut across the distinctions above: “directed” duties include duties not to kill, duties of gratitude, promissory duties, contractual duties, and duties of romantic fidelity. Other duties are arguably undirected—though this is controversial. Typical examples include duties not to damage the unliving natural world (e.g., rock formations), and duties of general beneficence.Footnote ⁸ Furthermore, among the duties that are “directed,” some correlate with rights (e.g., duties not to kill and promissory duties) while others do not: gratitude seems to be a directed duty, owed to a particular party, but it seems mistaken to say that the relevant party has a right to be thanked.Footnote ⁹

7. VII. Some duties are created, such as legal duties or the duties constitutive of games, carparking systems, or library bye-laws; others are not, such as fundamental moral duties. Among created duties, we can distinguish between those that are morally justified and those that are not. We can also distinguish between those created duties generated or annulled by the direct reflexive exercise of duty-creating or duty-waiving powers (such as promissory duties, or legal duties created by formal legislative acts, or the waiving of duties through consent or forgiveness), those created duties generated or annulled by human actions that are not aimed at duty-creation or waiver (such as the unplanned emergence of conventions or habits, or the unplanned conception of a child), and those created duties whose existence or annulment are the result of events in the nonhuman natural world triggering prior conditions on general duties (such as an asteroid impact generating duties of rescue).Footnote ¹⁰

Can the seven distinctions above be drawn similarly in relation to “mere reasons?” My view is that distinctions (I)–(IV) apply as readily to nonduty “mere reasons” as to duties, while it seems that distinctions (V)–(VII) pertain only to duties. Let me explain. First: many nonduty reasons are pro tanto, but sometimes reasons can stack up such that ɸ, for which one has a nonduty reason, is what all things considered one should do.Footnote ¹¹ Second, some nonduty reasons are more important than others, and some are very trivial.Footnote ¹² Third, we can distinguish moral from prudential from aesthetic from epistemic nonduty reasons, and can distinguish nonduty reasons within created realms of normativity, such as law, etiquette, and games.Footnote ¹³ Fourth, like duties, “mere reasons” can be borne by humans, by corporate entities, and potentially by some nonhuman animals, but not by rocks, plants, or colors.Footnote ¹⁴

By contrast, I doubt that any consideration that amounts to just a nonduty “mere reason” is permissibly enforceable, at least in its guise as a “mere reason.” A legislature might decide to make some nonduty “mere reasons” enforceable through law, but in doing so, it transforms the relevant “mere reasons” into duties. As “mere reasons,” the relevant considerations are not enforceable. Further, moral considerations that look like enforceable reasons independently of legislation—for example, reasons not to assault, or to respect one’s contracts, or to care for one’s children—all seem to look already like moral duties. I will not spend long on this; if it is correct, then a distinction between the enforceable and the unenforceable can be drawn only within the realm of duties, as all “mere reasons” are unenforceable while they have the status of “mere reasons.”

Similarly, it seems to me that no “mere reason” is directed or owed to someone—or at least not in the quasi-juridical way that duties can be. This is a controversial claim, and I do not intend to rule out the possibility of “mere reasons” that are directed in some alternative sense.Footnote ¹⁵ Of course, there can be “mere reasons” constituted by particular people’s wellbeing or autonomy: your thirst is a reason for me to bring you a cup of tea; your desire to visit the library is a reason for me to give you a lift; these really are “mere reasons” rather than duties for me. (Further, such “mere reasons” based on your good might themselves bear a tight relationships to directed duties owed to you: e.g., duties owed by me to you to give the mere reasons appropriate weight in my reasoning, and duties of apology owed to you if I fail to do so.) But I believe that nonduty “mere reasons” grounded in the good of particular people are not themselves directed reasons in the special way that duties can be directed. It is well known that a duty can pertain to a particular person, or be morally grounded in their good, or involve action entwined with them and their good, while nonetheless being directed or owed to someone else. Raz notes that the British state’s legal duty to make child benefit payments is owed to parents, such that they are wronged if the payments are withheld, but the child’s well-being is the ground of the duty, and the child is the core party to whom the duty pertains.Footnote ¹⁶ I do not believe that we can make sense in the case of nonduty “mere reasons” of any contrast like that between (a) the party to whom a duty is owed, that is, the party who is wronged by its violation, and (b) parties whose good grounds the duty or to whom the duty pertains or with whom the duty is somehow bound up. In the relevantly quasi-juridical sense captured by (a), there seem to be no distinctively directed “mere reasons.” This is reflected in the absence of any linguistic marker of such directedness for “mere reasons”: duties are “owed to” those to whom they are directed, while by contrast it is unidiomatic to regard reasons as “owed to” anyone (except in the sense that one sometimes has a duty to explain one’s reasons to another). On the same basis, I think there are no “mere reasons” that correlate with rights—where this would mean reasons that do not amount to duties, but whose nonfulfillment in itself necessarily constitutes an infringement of a right.

Our analysis of duties needs to explain the asymmetry above: why there are no enforceable or directed “mere reasons” while some duties are enforceable and some are directed. There are theorists who make the contrast sharper by claiming that all duties are enforceable and directed, while no “mere reasons” are.Footnote ¹⁷ I have some sympathy with these views, though I do not endorse them. A key question—ultimately addressed in Section 6 below—is why some duties are enforceable and directed, while no “mere reasons” are.

2. Created and Uncreated Duties and “Mere Reasons”

What about the final contrast (VII)? Can there be created nonduty “mere reasons?” There is a tough question here. The easy part is this: of course both nonhuman and human events can create “mere reasons” through triggering conditions on prior general reasons: I have a general reason to respond to others’ needs when I can, and a volcano, a phone call, or a hungry stomach can trigger more specific reasons to provide funds to support a rescue, to listen to someone and to feed them, “mere reasons” that often do not amount to duties.

The tough question is whether nonduty “mere reasons” can be created through intentional reflexive willing: willing “that some consideration be a reason, where that willing is that in virtue of which the consideration is a reason” (Chang, Reference Chang2020, 292).Footnote ¹⁸ Such a reflexive process is how promises, legislation, and commands create moral, legal, and military duties, and how consent and forgiveness work to annul duties. It is debatable how many people’s wills must be involved for the relevant change to be effected: must my promise be accepted by its addressee if it is to generate a new duty—and is the same true when consent waives a duty?Footnote ¹⁹ It is also debatable how deliberate and knowing the relevant agents must be to qualify as engaged in appropriately reflexive willing. However these questions are answered, in each case a single-person/multiperson, transparently-deliberate/less-intentional communicated will-to-change-a-duty effects the relevant change simply in virtue of the act of communicated will. While I trip you, wounding your ankle, I might tell you that I am doing this in order to place myself under new duties of medical care for you, but my will-plus-telling is not here what effects the change in duties: the change is effected by my creation of your new need. By contrast, if I promise to provide you with new medical care and you accept this promise, our valid communicated acts of will are enough on their own to create the new duty.Footnote ²⁰

My question now is this: is a similar process of reflexive willing able to create (or annul) nonduty “mere reasons?” Some think so.Footnote ²¹ But on inspection, it is unclear, and we need a good theory of the distinction between duties and “mere reasons” to help clear things up. Candidate processes for reflexively creating new nonduty “mere reasons” include inviting, requesting, assuring, and perhaps creating new games; one might also think of offering, suggesting, or encouraging; on the “annulling” side, one might think that sometimes consent waives a “mere reason” rather than a duty. But a close look at these processes is indecisive: the cases can, I suggest, plausibly be analyzed as reflexively creating or annulling only duties and never “mere reasons.”

For example, when I invite you to a party, I exercise a power of permission, allowing you to attend when otherwise you would have been duty-bound not to; the invitation also typically expresses my desire that you attend. But it seems mistaken to think of my invitation as itself a reason to attend: when wondering whether to go, it would be appropriate for you to consider whether I want you to be there, whether you want to attend, what goods—friendship, neighborliness, love, and fun—are bound up with attendance, and what the opportunity costs are. But it would be inappropriate for you to consider the fact that I invited you as an additional reason to attend, beyond those mentioned. The invitation makes it appropriate to consider the reasons listed, which would otherwise not be considerations for your attention (given your general duty not to attend parties uninvited). But the invitation is not itself an extra reason. By contrast, if you had promised to attend, then the fact that you promised would itself be a ground for a duty to attend. Of course, in accepting the invitation, you are typically promising attendance, so your acceptance creates a duty to attend. But prior to or independently of acceptance, the invitation creates no new reason to attend. And acceptance creates a new duty, not a new “mere reason.”

The same point seems to apply to other processes that look at first glance to involve the reflexive creation of nonduty reasons: when I issue a request, I offer the person subject to the request the opportunity to place themselves under a duty to satisfy my request by accepting it.Footnote ²² Perhaps before they have accepted my request, we might think it has generated a new “mere reason”—either to accept or reject the request. But is this simply a nonduty reason? Or does the relevant person have a duty here, namely, a duty of respect for my request and a duty to decide whether to accept it?

The discussion above is indecisive, but it is enough for now to present a sharpened challenge. At the end of Section 1, we concluded that we need an account of duties that explains why “mere reasons” cannot be enforceable or directed, while duties can. We should add that our account ought in addition to explain why, although some duties can be created simply through a reflexive communicated will-to-create the relevant duties, it is more doubtful that “mere reasons” can be so created (or perhaps annulled) simply through a reflexive act of will.

3. Sentimentalist and Rationalist Accounts of Duties

In the literature, there are two prominent accounts of the nature of duties: first, “sentimentalist” or “accountability” approaches, and secondly, “rationalist” or “constraint” approaches.Footnote ²³ I will argue that both approaches capture important truths but require supplementation to address shortcomings and to meet the challenges outlined in Sections 1–2 above.

Sentimentalist-accountability approaches define duties in terms of the attitudes or actions their violation justifies. According to one version of this approach, a duty is defined as a normative consideration whose violation would (absent excuses) justify blame, guilt, resentment, punishment, apology, and potential forgiveness: attitudes not justified by failure to act on “mere reasons.”Footnote ²⁴ According to a related approach, a duty is a normative consideration whose existence justifies actions of demanding or claiming—either by the moral community, or by the specific person to whom the duty is owed—that are not justified by a “mere reason.”Footnote ²⁵

There are technical problems for sentimentalist-accountability approaches.Footnote ²⁶ But the core idea seems to address our key question about duties and nonduty “mere reasons”: when duties are violated or infringed, distinctive accountability attitudes and demands are plausibly triggered, attitudes and demands inappropriate to the violation or infringement of “mere” nonduty reasons.

My unoriginal worry is that there is no noncircular way of distinguishing the attitudes and demands definitive of duties from the attitudes and demands definitive of nonduty “mere reasons.”Footnote ²⁷ My neighbor has a reason—but no duty—to paint their gray exterior walls so that they are less ugly for the street. If my neighbor ignores their neighborly aesthetic reasons without justification, then I would be justified in making an adverse judgment of my neighbor’s will, could ask my neighbor for an explanation, and I could justifiably feel heated, difficult emotions that are phenomenologically, motivationally, and functionally identical to those involved in resentment. Because my response is justified merely by the ignoring of a nonduty reason rather than the violation of a duty, it might be said that it does not qualify as resentment proper, but phenomenologically, motivationally, and functionally, it could be identical; we could call it resentment*. The distinction between the resentment* appropriate to the ignoring of nonduty “mere reasons” and the genuine resentment appropriately triggered by duty violations (or infringements) is a distinction that, I believe, can only be defined circularly: the genuine, nonasterisked reactive attitudes at the center of the sentimentalist-accountability approach are simply those reactive attitudes justifiably triggered by duty violations (or infringements) as opposed to those triggered by overlooking “mere reasons.”

Some theorists—notably Darwall—are happy to accept this conclusion, taking duty, blame, accountability, and demandability as forming an inter-definable circle (Darwall, Reference Darwall2009; Darwall, Reference Darwall and Shafer-Landau2012). It is a problem for those who seek a noncircular account that distinguishes duties from “mere reasons” on the basis of the reactive attitudes or actions that duties distinctively justify.Footnote ²⁸ I believe a noncircular account is achievable and illuminating—but only if we supplement the sentimentalist approach with something different.

The standard further approach is a “rationalist-constraint” view that defines duties by the distinctive role they should play in the agent’s reasoning. For some, duties are “constraints” or “trumps” whose action-demanding force requires that they sometimes be respected even at the expense of the greater good (Nozick, Reference Nozick1974, 28–30; Dworkin, Reference Dworkin and Waldron1984; Kagan, Reference Kagan1998, 72). For others, duties “silence” certain considerations that would otherwise speak against the duty-required action (McDowell, Reference McDowell1998, 91–93), or duties “exclude” certain countervailing reasons from consideration, or from being acted on (Raz, Reference Raz1975, Ch. 2; Scanlon, Reference Scanlon1998, 156–157).Footnote ²⁹ Each position enables the theorist to claim that what distinguishes the reactive attitudes or demands appropriate to duty violation or infringement is that they are accountability responses appropriate to special reasons of the relevantly constraining kind—to trumps, silencers, or excluders. “Mere reasons” by contrast, are plausibly less complex, possessing only weight or importance.

I mention two shortcomings of the rationalist-constraint approach, before examining the challenges of Sections 1–2. First, the proposals from rationalist-constraint theorists seem heterogeneous: trumping and constraining differ from excluding which (on at least some readings) differs from silencing. It would be satisfying to explain the unity within this heterogeneity, especially as different duties fit some of the approaches better than others: your moral duty not to kill one innocent person seems to outweigh the countervailing fact that in so doing you could save three other innocent people from similar killing. But does it silence that fact or merely override or trump it? I am uncertain whether saving three innocents from a similar fate is a genuine reason to kill the one—a reason that is outweighed, and on which I should not act, but a reason nonetheless. By contrast, your duty not to kill one innocent person more plausibly silences the fact that in doing so you would gain an inheritance. The inheritance, while perhaps a genuinely “greater good” to you from the killing, is no reason at all to kill. When we consider unimportant morally justified duties, matters seem different: prima facie it is doubtful that my duty not to park on a double yellow line silences anything. It excludes from my consideration reasons such as the inconvenience of having to walk further to the shops, without rendering these countervailing reasons nonexistent or silent. Furthermore, some reasons seem to be exclusionary without qualifying as duties: Raz mentions that a financially expert friend’s advice might be an exclusionary reason for me to invest in certain ways, but it is no duty (Raz, Reference Raz1975, 77). What unifies the kinds of trumping, silencing, or excluding that are definitive of duties and not of “mere reasons?”

Secondly, some genuine duties seem too trivial to have a trumping, silencing, or exclusionary character. Consider a very unimportant conditional promise: I promise “to give you my spare blank sheet of paper unless doing so carries any direct cost, opportunity cost, or inconvenience for me or others.” This is a genuine promise that creates a genuine duty. But it demands only a trivial action, and the slightest inconvenience is allowed to work as a countervailing reason against fulfillment. The promise has still created a genuine duty, but what reasons does this duty exclude, silence, or outweigh for me as the promisor?

Where have we got to? I have said that there is truth in the sentimentalist-accountability view that duties distinctively trigger reactive attitudes and demandability, but that we need to supplement this approach to avoid circularity. A rationalist-constraint supplement adds that duties are distinctively trumping, silencing or exclusionary, but we need more to explain the unity underlying the different ways that duties constrain, to explain why some “exclusionariness” (as in reasons to follow my financially expert friend’s advice) is not constitutive of duties, and to explain how very trivial duties constrain (where even important nonduty “mere reasons” do not). Such a further account should also answer Section 1’s questions by explaining why constraints of the relevant sort are potentially enforceable and directed, while nonduty “mere reasons” are not, and Section 2’s question by explaining why relevant constraints are creatable through reflexive willing, while it is less clear whether “mere reasons” are.

The rationalist-constraint approach as sketched so far does not answer these questions: it is unclear why having a trumping or exclusionary character should be necessary for a reason to be enforceable, directed, or reflexively creatable. To get this clear, we need more. In particular, we should look not just at how my duties constrain me, but also at how my duties constrain others’ actions in relation to me. In the literature, the sentimentalist-accountability approach is relational, focusing on how duty violation by me justifies attitudes and actions in others: I suggest that for a full account of duties we need a similarly relational rationalist-constraint approach.Footnote ³⁰

4. A New Millian Account: Duties as Will-Silencing Reasons

My proposal is that a duty to ɸ is a reason for the agent to ɸ that silences the significance for others that would normally attach to that agent’s will-not-to-ɸ.Footnote ³¹

This is a relational rationalist-constraint analysis of duties. It makes the distinction between duties to ɸ and nonduty “mere reasons” to ɸ depend on precisely how the agent’s decision about ɸ-ing is others’ business. It says that A’s duty to ɸ is a reason for A to ɸ whose nature silences for (some) others the weight they should normally attach to respecting A’s will as such. In particular, the idea is that when A has a “mere reason” to ɸ, A’s decision not to ɸ deserves some respect from others simply as an expression of A’s will, giving others pro tanto reason to back off and allow A to ɸ. By contrast if A has a duty to ɸ then this is a reason for A to ɸ whose nature silences what would otherwise be others’ reason to respect A’s will-not-to-ɸ and back off. This means that when A has a duty to ɸ, the agent’s will-not-to-ɸ can permissibly be overridden by (some) others who can thereby “exact” or “enforce” A’s ɸ-ing, if there are no countervailing reasons against doing so.

The picture is complicated by the fact that there are very often independent countervailing reasons not to make people do their duty. My view is prefigured by Mill, though in its detail his view differs from mine because he fails to clarify that it is only the agent’s duty-contrary will that is no impediment to “exaction”:

This quotation from Mill does not quite get things right, because a person bound by a duty is sometimes justifiably entitled to complain about its exaction simply because the exaction has some harmful effect on that person themselves or on wider values, rather than on others or on the exacting party. In Section 5, we will see cases where values of political equality and romantic love prevent a duty’s exaction. We need to qualify Mill’s view so that it holds only that the agent bound by a duty necessarily has no entitlement to complain about being made to fulfil it if the cost to the agent and others is only the overriding of the agent’s will. Footnote ³²

I think this is an attractive rationalist-constraint relational basis for the distinction between duties and nonduty “mere reasons.” When I am faced with a nonduty “mere reason,” my decision to ignore this reason is, qua my will, itself a reason for others not to intervene and to allow me to act, even if my decision is foolish, dangerous, silly, or morally questionable. Consider examples like my neighbor’s decision not to paint her gray exterior wall to beautify the street, or my decision not to bring you a cup of tea when you are clearly in need of one. Simply as a choice, as an expression of autonomous will, such a decision deserves respect from others. Or at least this is true insofar as the decision conforms to or contravenes a nonduty “mere reason.” By contrast, when I am faced with a duty, my contrary decision or will does not deserve such respect from at least some of those who would otherwise have had to respect it: willing contrary to duty is no reason in itself for some others not to intervene to ensure that I do my duty. There are often further reasons for nonintervention by others, but my duty-contrary will is no such reason, as it would be were I facing only a “mere reason.”

This thesis that duties are will-silencing reasons addresses the questions left hanging for rationalist-constraint accounts at the end of Section 3. It provides a unifying feature, in that all duties, whether functioning from the duty-bearer’s perspective as variously “trumping outweighers,” as “silencers” or simply as “exclusionary,” necessarily stop the bearer’s duty-contrary will from functioning as a reason for (some) others’ respect. Across the cases mentioned in Section 3—my duty not to kill you to save three, my duty not to kill you to gain an inheritance, and my duty not to park on a double-yellow line—my contrary will to kill or to park constitutes no reason in itself for others’ respect. Of course there are often further reasons for others not to make me keep off the double-yellow line (or indeed for others not to make me refrain from killing), but my mere will to park on it is no good reason in itself for anyone to back off and allow me so to park—as it would be if I had only a “mere reason” and not a duty here.

Furthermore, the thesis that duties are will-silencing reasons also explains why some “exclusionary” reasons are not duties: my decision to ignore my financially expert friend’s investment advice, despite the exclusionary reason not to do so constituted by the friend’s expertise, deserves respect from others as my will—even if there are further will-independent reasons (e.g., having to do with my family’s wellbeing) for others not to respect my decision. My advice-contrary will would not deserve such respect qua will if my exclusionary reason were a duty.

Finally, I think that even the most trivial duties satisfy the “Will-Silencing” condition, such as my weak conditional promise to give you a sheet of paper unless so doing carries any costs. Were I to decide not to give you the paper despite it being utterly convenient to do so, that decision in itself gives you no reason—no reason of respect for my decision as a decision—not to “simply make” me hand it over, if such a thing were possible with no causal or constitutive side-effects other than the overriding of my will. That is the nature of duties. Of course, there will normally be additional reasons not to intervene; Section 5 below clarifies this.Footnote ³³

Have we now reached the point where I address my core questions by showing how this relational rationalist-constraint view explains why duties are potentially enforceable, directed, and creatable through reflexive willing, while nonduty “mere reasons” are not? Not quite. We will get there, but first we need a section on apparent counterexamples, for I am sure readers have doubts.Footnote ³⁴

5. Unenforceable Duties and Liberalism

One concern is that the proposal—“Duties as Will-Silencing Reasons”—cannot accommodate rights to do wrong; another is that it cannot accommodate overridden pro tanto duties. Showing how it handles such cases generates a third concern: that it makes duties surprisingly “small” in their practical implications.

Rights to do wrong involve the agent being permitted to act contrary to duty. Following Jørgensen, Setiya mentions as examples the agent’s rights to be insufficiently beneficent and the right to vote for a fascist political party; he adds the right to break a promise; Barry and McTernan mention the unenforceability of duties of romantic fidelity and of gratitude.Footnote ³⁵ In each case, the agent is subject to a duty—to give to charity, to vote for the just party, to keep their promise, and to be romantically faithful and grateful—and yet others seem rightly bound to “back off,” to allow the agent to act contrary to her duty. Does this not then show that the agent’s duty-contrary will in such cases deserves respect, falsifying “Duties as Will-Silencing Reasons?”

No. In each case, all others have powerful reasons (duties indeed) not to prevent the agent from acting contrary to her duty, but these reasons to “back off” do not spring from or constitute respect for the agent’s will qua will; instead, they have independent grounds. For example, the fascist voter’s will to vote fascistically—to vote contrary to their duties of justice—does not in itself qua will constitute any reason for others’ respect. It is rather that values of political equality and democratic fairness give others reason to respect the fascist voter’s duty-contrary will. If it were possible to “simply make” the fascist voter vote for a more just party—that is, to make them vote nonfascistically without any side-effects on other values such as political equality or democratic fairness or other interests—then this would be permitted.Footnote ³⁶ The trouble is that such “simple making” is not an option, for any enforcement of a nonfascist vote, even through magical overriding of the voter’s will, would entail illegitimate unequal political recognition for the relevant voter. Drawing on remarks by Setiya (Reference Setiya2022, 1128), we can make the same point about compelling a wealthy person to engage in charitable beneficence: insofar as there are good reasons against compelling such charity, they do not spring from respect for the uncharitable will of the wealthy person, but rather from wider considerations about the social value of respect for legal property rights—considerations that rule out the possibility of side-effect-free compulsion of such charity.

In some cases, the impossibility is deeper: what the agent has a duty to do simply cannot be done other than through an autonomous will. Magically making a lover behave faithfully by overriding their will would not get us true romantic love with autonomous commitment to the loved one—instead, it would deliver a simulacrum. The same is true of gratitude. So, making the agent fulfil duties of romantic fidelity or gratitude when they choose not to is deeply impossible: even magically overriding the agent’s will fails to deliver what the duty requires in these cases, not because it generates unavoidable costs on other values (as in the earlier cases above), but because the required action can only be performed through an autonomous will.Footnote ³⁷

These reflections support “Duties as Will-Silencing Reasons”: in each apparent problem case, the agent’s duty-contrary will is not itself a reason against making them fulfil their duty; instead, other reasons tell against this. Even in the romantic fidelity case, it is not the agent’s contrary will qua will that gives others reason not to make the agent behave faithfully—it is, rather, the nature of romantic love that entails that it is impossible to “make” anyone engage in it. In each of the cases listed, the agent really does have a right to do wrong, but one springing from factors other than simple respect for her will qua “wrong will.”Footnote ³⁸

A similar response should be made about overridden pro tanto duties, and we can be brief here. When I infringe my duty to meet you as promised, in order to rescue someone in dire need instead, it is not my decision not to fulfil my promise that gives you and others reason to “back off” and let me perform the rescue: it is, rather, the value of the rescue that gives you and others such reason. Even overridden pro tanto duties stop the agent’s will as such from deserving respect: it is the duty-overriding factors that ask others to respect the agent’s will in such cases, not the bare fact of the agent’s willing.

A doubtful reader might worry that this reveals the analysis of duties to turn on a “small,” overly specific feature: whether the agent’s will in itself qua will generates reasons for some others to respect it or not. Making duties “small” is, of course, a price one should expect to pay in developing an account that can encompass trivial promissory and institutional duties. Nonetheless, it is worth highlighting that the will-focused kind of Millian exactability that I have isolated is hardly unimportant. It is at the heart of liberalism and hence of contemporary morality and law to think that an agent’s will normally deserves some respect simply qua will, independently of an assessment of the quality of what is willed. If duties mark the point where we depart from this liberal assumption—the point at which the agent’s reasons to ɸ are of such a character that they make her choice not to ɸ deserve no respect qua choice from at least some other(s)—then duties mark a highly significant feature in normative topology, even if the examples in this section show that it is not always a feature that makes much difference.

In the next section, I return to our initial puzzles and explain the close link between this “will-silencing” feature of duties and enforceability, directedness, and reflexive creatability-at-will.

6. Duties’ Enforceability, Directedness, and Creatability