Skip to main content
Jonathan Johnson-SwagelEnterprise applied AI · Public work Search

Search the site

Share Jonathan’s profile

View introduction text

Jonathan Johnson-Swagel is a Chicago-based applied AI platform leader. His experience includes shared platform architecture at AWS and global transformation at Uber.

RESEARCH · DEVELOPMENTAL READING EDITION

Chapter II — What It Means to Owe

The difference between need, duty, claim, permission, and the authority to change an obligation.

← Complete contents · Chapter 2 of 8 · Updated November 2026

2.1. The grammar of an unfinished obligation

Imagine that a clinician tells a patient she will retire at the end of the year. The patient requires regular medication management, has spent several years establishing an effective treatment regimen, and has good reason to fear that discontinuity will produce serious risk. The clinician's departure is legitimate: she is not required to practice indefinitely. The patient's need is also legitimate: discontinuity may damage health, and a retirement announcement does not render that concern any less urgent. Now suppose the clinician offers to send records to a receiving practice but the practice has not agreed to take the patient. Has the clinician fulfilled her obligations? The question cannot be answered by deciding whether retirement is permissible, whether the patient deserves care, or whether someone else could competently perform the work. Each proposition might be true while the status of a particular duty remains unsettled.

The difficulty begins with the word “owe.” We can owe a payment, an explanation, care, honesty, gratitude, an apology, or loyalty. We may even speak of owing someone another chance or our love. Yet these are not interchangeable claims. What may the other person demand? What may we refuse? What counts as fulfillment, and who can release us? Someone may deserve serious consideration without being entitled to the response they want. A claimant may hold a firm right without controlling every permissible way of fulfilling it. And someone who has benefited profoundly from another's generosity does not thereby acquire a continuing right of access to that person's life.

Before evaluating how responsibilities change, we need to distinguish moral reasons, duties owed to particular people, claim-rights, liberties, and powers to change or release a duty. We must also distinguish carrying out an act from satisfying the obligation that required it, and identify who may properly hold the agent accountable. This vocabulary is not original. Hohfeld's analysis of rights, work on promising by Hart, Raz, Owens, and Shiffrin, Darwall's account of second-personal accountability, and theories of special obligations and care provide its foundations (Wenar, “Rights”; “Promises,” Stanford Encyclopedia of Philosophy; Darwall 2006; Chadha-Sridhar 2026). The question is what, if anything, their strongest combined account leaves unexplained.

Our concern is not lexical tidiness. When an institution announces that a service has been “transferred,” the term may conceal whether a successor assumed an independently grounded duty or whether an original agent retained responsibility for ensuring performance. When family members insist that a caregiving obligation “continues,” they may be saying only that the need remains, or instead asserting that the same person owes a particular act. When someone declares an obligation “finished,” the declaration may describe factual performance, a valid waiver, exhaustion of a defined term, or a unilateral refusal to recognize another's claim. The verbal similarity of these events creates opportunities for moral error. Their distinctions must be established before their legal and institutional consequences can be judged.

2.2. Reasons, requirements, and relational claims

Suppose a stranger has dropped a bag of groceries beside a crowded sidewalk. I have a reason to help. If assistance would take a few moments and impose negligible cost, that reason may be powerful; if the stranger faces an acute danger that I alone can avert, I may be morally required to intervene. But those claims do not yet settle whether the stranger holds a directed right against me to carry every parcel home or whether an identified third party has standing to enforce the entire duty. A reason to act and a claim possessed by another person need not have identical logical structures.

A moral reason and a duty are not always sharply separate. Some philosophers understand duties as especially weighty reasons; others emphasize the special accountability they create. Nor is every duty owed to one named person. A general duty of beneficence may require serious concern for others while leaving discretion about whom to help and how. A promise to deliver medicine to a particular patient at a particular time fixes both beneficiary and action more closely. This is the distinction often discussed through perfect and imperfect duties. It does not mean one class matters and the other is optional: imperfect duties can be binding even when they leave latitude in how to act (Kant and Hume on Morality, Stanford Encyclopedia of Philosophy).

Three questions help separate these issues. First, does a fact matter morally—a patient's vulnerability, a relative's sacrifice, a former partner's reliance, or an institution's ability to prevent harm? Second, does that fact place a requirement on anyone? Third, is the requirement owed to an identifiable person who has standing to demand its fulfillment? An urgent need may give everyone reason for concern without imposing the same duty on everyone. A general duty need not give one beneficiary control over how it is fulfilled. Even when someone holds a genuine claim to a specified act, that does not entitle them to demand an entirely different one. These questions are related, but answering one does not settle the others.

For the problem of moral transition, the most important consequence is negative. The persistence of a reason for action does not itself prove the persistence of the previously assigned obligation. If a patient continues to need treatment, the urgency of that need provides powerful reasons to organize care; it does not by itself explain which duties remain with the retiring clinician, which arise for the receiving clinician, and which belong to the organization that operates the medical practice. A theory that jumps from need to attribution has moved too quickly. Yet a theory that concludes from the absence of a particular directed duty that no one has reason to act makes the opposite mistake. The moral problem requires both careful attribution and attention to what remains important when attribution fails.

2.3. Hohfeld: four incidents, four different questions

Wesley Newcomb Hohfeld's contribution was to distinguish normative relations frequently collected under the undifferentiated term “right.” The familiar scheme separates claim-rights, privileges or liberties, powers, and immunities, each paired with a correlative position held by another agent. A claim corresponds to a duty; a liberty to the other's no-claim; a power to a liability to normative alteration; and an immunity to the other's disability or lack of power. These are analytic distinctions, originally articulated in legal theory, which can illuminate moral relations without determining their grounds (Wenar, “Rights”; “Legal Rights,” Stanford Encyclopedia of Philosophy).

Let A be an agent and B another person. If B holds a claim against A that A return a borrowed document, A has the corresponding duty. If A has a liberty to decline a non-obligatory favor, B lacks a claim that A provide it. If B has a valid power to release A from a promise, B can alter A's normative position by an appropriate act. If A has an immunity against B's attempt to require some new performance, B lacks the relevant power to impose that obligation. An actual moral relationship can contain several of these incidents at once. The presence of one does not make the others automatic. A person might be entitled to payment while lacking authority to require romantic attention; someone might be free to end an engagement yet bound to return property entrusted during it.

The distinction between liberties and powers is particularly important for relational endings. A liberty to stop doing something is not identical to the power to extinguish every obligation associated with the relationship. Mara, the caregiver described in Chapter I, might be permitted to cease physically demanding work while lacking unilateral authority to declare a separate promise of advance notice discharged. Conversely, someone who receives the care may possess a valid claim to specified assistance without acquiring the power to extend its duration indefinitely. If the patient in the opening case has a claim to responsible notice, the claim's content must be specified before we can say whether it has been satisfied. It cannot be amplified without justification into a right to prevent retirement.

Likewise, the distinction between claims and powers matters when the recipient of a service accepts a substitute. Suppose A promises B a sum of money, and C offers to pay B instead. If B has authority to accept C's payment in full satisfaction of the obligation, A's duty may be discharged once the appropriate act occurs. But this differs from C independently promising B the same amount while A's obligation remains. It also differs from an unauthorized third party paying money that B may receive without agreeing to release A. Which transaction occurred depends on the terms of the original duty, the normative powers of the parties, and the acts undertaken; the identical receipt of funds does not answer these questions.

Hohfeld's vocabulary does not solve the moral problem of who ought to hold each position. It tells us what it would mean for a person to have a power of release; it does not prove that the person possesses it. The distinction therefore supplies discipline rather than an answer. An ethical theory must justify which powers, immunities, and claims belong to whom and why. The desire to render all personal relations into Hohfeldian tables should also be resisted. Some responsibilities may be imperfect, plural, uncertain, or insufficiently settled to justify such determinate assignments. The appropriate use of the analysis is diagnostic: to prevent a conclusion about one normative relation from being silently substituted for a claim about another.

2.4. The claimant, the beneficiary, and the authority to demand

A further distinction is needed among the person who benefits from an act, the person to whom a duty is owed, and the person who can authoritatively demand or enforce performance. These positions often coincide, but not always. A child may benefit from an act that a parent is required to undertake by reason of the child's welfare, while a court or other institution may have an authorized role in overseeing its performance. A contractual beneficiary may gain from a promised payment without possessing the powers granted to the original contracting party. And someone may have a justified complaint about the failure of a public institution even if the relevant right is not individually waivable by that person.

Consider a public agency responsible for providing emergency accommodation. The people who need shelter have interests that ground the agency's obligations. A particular applicant may have a claim under a legal or moral account of fair provision. Yet it would be mistaken to infer from her need that she can unilaterally determine the agency's entire resource allocation or release it from duties owed to everyone else. The standing to receive, demand, adjudicate, release, or redesign an obligation may be distributed across different agents and institutional procedures. None is exhausted by the bare existence of a need.

The distinction also matters between intimate parties. Suppose a parent financially supports an adult child, and another family member urges the parent to continue because discontinuation would generate hardship. The family member may have good reasons to criticize withdrawal; that does not give the family member the same authority as the adult child to accept or reject a proposed arrangement. Equally, the adult child's wish for continued support does not by itself settle the moral status of the parent's resources or the claims of other dependents. The facts that someone is affected and that someone has standing to demand a particular response are connected but not equivalent.

These examples expose a hazard in the apparent simplicity of the language of relational accountability. To whom is the agent answerable? For which act? Who can speak on behalf of the claimant, and under what conditions? Whose agreement effects discharge? An answer must sometimes account for people who are unable to participate fully in deliberation, for multiple affected parties whose claims conflict, and for institutions whose responsibilities cannot be collapsed into the will of a single beneficiary. The question is not whether such complexity renders moral requirements impossible, but whether a proposed theory has the conceptual resources to identify whose authority is being exercised.

2.5. Directed duties and the difference between being owed and being needed

The term “directed duty” marks a requirement owed to a determinate person or set of persons, rather than simply a requirement to perform an act whose benefits may be widely distributed. Its philosophical importance lies in the special relation between the duty-bearer and the claimant: if A owes B a promised performance, the wrong of nonperformance is not exhausted by a diminution of aggregate welfare. B may have standing to demand an explanation, protest the breach, seek appropriate repair, or grant a valid release. But the scope of that standing remains a matter of theory, not a consequence of the word “directed.”

Stephen Darwall's second-personal account provides a particularly exacting challenge. In The Second-Person Standpoint, Darwall argues that moral obligation is intimately connected with authority to address claims and hold others accountable (Darwall 2006). On this approach, to understand a requirement as an obligation is, in part, to understand the standing of persons to make demands within a community of moral equals. This perspective prevents a simplistic reduction of obligation to efficient welfare provision: the recipient of a violated promise is not just a location in which a bad outcome has occurred. She may be someone to whom another person has failed to answer.

Yet the second-personal perspective also sharpens our question. If a relationship changes, do the same persons retain the same authority to make demands about every aspect of it? If B possesses standing to hold A accountable for a promise of financial support, does that provide standing to require A's continuing intimate presence? If B may properly demand an explanation for a particular breach, can B demand indefinite access to A's inner reasons for ending a relationship? The answer cannot be yes without qualification, because the reasons that ground one authority do not necessarily ground another. It also cannot simply be no across the board, because some forms of answerability genuinely persist across relational transformation.

The point can be illustrated through a case in which two wrongs produce comparable outcomes but differ in whom they wrong. Suppose an employer pays a worker late, and a family member provides enough money to prevent financial hardship. The worker's material need may be met, but the employer may still have failed in an obligation owed to the worker. The family member's generosity can mitigate harm without by itself satisfying the employer's moral or legal duty. The same reasoning applies to a professional who breaches confidentiality while ensuring that the patient receives excellent treatment: clinical benefit does not necessarily answer a distinct claim to privacy. These examples show why outcome-equivalence is an unreliable proxy for the discharge of directed obligations.

But the concept has limits. Not every failure to supply an important benefit wrongs a particular individual in the same way as breach of promise or violation of privacy. The fact that a person can reasonably complain that society has failed to make adequate care available need not imply that every available helper has wronged her individually. A full account must distinguish interpersonal claim, general social duty, institutional obligation, and the prudential interest of the person in need. Moral concern is abundant; authority to impose a determinate duty is not inferred from concern alone.

2.6. Promises: the power to bind and the limits of control

Promises are a natural starting point because they make the temporal character of obligation unmistakable. A person who promises to do something tomorrow commonly intends to change what will be owed tomorrow. If the promise did not persist through ordinary shifts of convenience, it could not provide the assurance on which coordinated action depends. Yet the normative significance of promising is not straightforwardly reducible to the promisee's expected benefits. The same outcome can occur by accident without being a promise fulfilled; an action of equal value can fail to keep a specific commitment; a promisee may release a promisor even where performance would otherwise remain possible. These phenomena motivate theories that give the act of promising a distinctive role in altering interpersonal normative relations.

Philosophers explain promises in different ways. Raz treats a promise as an exercise of normative power: through an undertaking, people can change what they and others may expect or require, especially in shared activities. Owens emphasizes the promisee's authority interest—having some legitimate say over what the promisor must do, not just predicting their behavior. Shiffrin stresses meaningful commitments made between persons who owe each other equal respect. These are distinct arguments, and the full differences matter. For present purposes, they agree on one point: a promise is not simply evidence that another person's performance would be useful (Raz 2014; Owens 2012; Shiffrin 2008; “Promises,” Stanford Encyclopedia of Philosophy).

What follows for the person whose character changes after a promise? Suppose a person promises a friend six months of help with a demanding project. Midway through, the promisor discovers that the work no longer expresses their values. If the original promise was valid and the burden remains within its expected scope, dissatisfaction alone may not eliminate the duty. Indeed, a practice of committing oneself would be greatly weakened if the agent could dissolve every obligation by redescribing subsequent preferences. A promise can be valuable partly because it remains operative when the immediate motivational conditions that produced it no longer exist.

A promise can endure without giving its beneficiary unlimited control. Performance may become unsafe, the commitment may have been coerced, its original terms may contain limits, or fulfilling it may wrong someone else. In such cases the duty may change or cease to require its original performance. Promising to help with a project does not promise permanent access, affection, or agreement with the beneficiary's beliefs. A sound account must explain what power the promise created, when it binds, and who may release the promisor. An undertaking is not advance consent to every later demand made in the name of the same relationship.

This is a difficulty for accounts of relationships framed as comprehensive exchanges. An intimate promise can be made within a relationship whose continuation neither party can guarantee. Some commitments are directed toward particular conduct regardless of affection; others presuppose the continuing voluntary participation of both persons. To infer that a promisee owns all the future consequences of the promise would convert a bounded power to undertake an obligation into an unlimited power to alienate oneself. But to infer from the inalienability of some personal participation that no promise can survive changed emotional commitment would render other-directed assurances unreliable. The theory requires content-sensitive distinctions.

2.7. Consent, waiver, and normative powers over discharge

An obligation can change not because the duty-bearer acts, but because an appropriately situated agent exercises a power. A promisee may be entitled to release a promisor; a creditor may accept a settlement; a patient may authorize a permissible form of information sharing. These are normative acts: they do not simply alter the physical circumstances surrounding a duty, but can change the duty's legal or moral status. Yet the fact that an individual can receive a benefit does not show that individual possesses every power to waive obligations associated with providing it.

Consider the patient who tells a clinician that there is no need to send a summary to a new provider. That statement may validly waive a particular preference or authorization. It may not waive all professional, safeguarding, or record-retention duties that arise from standards designed to protect other interests. A promisee can often release another from an ordinary private commitment, but cannot necessarily authorize conduct that wrongs a third party or defeats an independently binding public duty. An individual's power of release is therefore itself limited by the grounds and objects of the obligation.

Nor should a verbal declaration of satisfaction be mistaken for a normative release. A claimant might express gratitude while still expecting fulfillment, or accept substitute assistance under economic pressure without intending to discharge the original debtor. Whether a waiver is valid may depend upon knowledge, voluntariness, the nature of the right, and institutional constraints. An account of obligation cannot read release directly from someone's temporary relief, especially where the party seeking discharge has superior power or controls the information on which the claimant must rely.

The converse possibility matters too. A claimant who is entitled to waive a duty may decline to do so despite the availability of a practically equivalent substitute. This can be normatively significant where the identity of the performer is part of what was promised. But a valid claim to a specific performance is not a general power to demand any action the claimant prefers. An obligation's object and a claimant's power to modify it need separate justification; neither can be deduced from the mere statement that the claimant is owed something.

Our inquiry into relational transition will therefore require a distinction among three events: a factual act that supplies the benefit originally sought; an authorized exercise of normative power that changes or discharges a duty; and a moral judgment that performance, whether original or substituted, meets the applicable standard. These events often occur together. They can also come apart. The claim that an obligation has ended is incomplete until we know which event is being described and what facts make it effective.

2.8. Special obligations without voluntary undertaking

The attraction of promising as a model is also its limitation. Much of human moral life is structured by responsibilities that do not arise from a freely negotiated undertaking. A child has claims upon caregivers before being capable of consenting to a contract. People with serious disabilities may be dependent upon institutions whose obligation to provide adequate support does not derive from individual bargains. An urgent danger may generate duties of rescue among strangers. Some duties arise from one's role, one's relationship to another, one's responsibility for creating an important situation, or the demands of justice rather than from the voluntary assumption of a private commitment.

The literature on special obligations asks what can justify distinctive duties to family members, friends, compatriots, patients, and other participants in enduring relations. The diversity of cases should warn against a universal explanation. Filial duties have been explained through gratitude, debt, friendship, and goods distinctive to the parent–child relationship; parental duties have been explained through causation, voluntarily assumed roles, the child's interests, and the political organization of guardianship; professional duties can be founded in both accepted roles and public standards of practice (“Special Obligations,” Stanford Encyclopedia of Philosophy). These are contested theories, not components whose conjunction is automatically true.

An especially relevant distinction, noted in discussions of parenthood, concerns an obligation to ensure that a child receives adequate care as opposed to an obligation personally to perform each act of care. A parent may be required to secure appropriate care without being required to administer every meal, provide every lesson, or remain the exclusive provider. Yet that does not mean finding any willing substitute necessarily discharges the original responsibility; the adequacy, authority, and stability of the arrangement matter. The conceptual distinction between securing a good and personally delivering it anticipates much of this dissertation's concern with practical and normative continuity (“Special Obligations”; “Parenthood and Procreation,” Stanford Encyclopedia of Philosophy).

The care literature offers a more direct challenge to models that give voluntariness too much weight. Chadha-Sridhar argues that duties to care can be grounded in natural duties while particular relational ties trigger their application. Ong and Savulescu insist that relationships should be understood within the broader structures that shape vulnerability and distribute caring burdens. These positions do not imply that every person with a relevant connection must supply unlimited assistance. Indeed, limiting unreasonable burdens is central to the latter account. But they require us to ask what the relationship contributes to the grounds of an obligation even when no promise was made (Chadha-Sridhar 2026; Ong and Savulescu 2026).

The morally attractive claim that nobody owes a favor unless they consented to provide it is therefore too crude. It is plausible with respect to many ordinary, optional services, but not as a general theory of care, parenthood, rescue, or institutional justice. Equally crude is the claim that the existence of a socially recognized relationship automatically identifies the complete content of its duties. A daughter might have reasons to support an elderly parent without owing co-residence, unpaid full-time nursing, or an indefinite relinquishment of professional life. A friend might be required to avoid cruel abandonment without being required to answer every call. The content of a special obligation cannot simply be read from the label of the relationship.

This difference bears directly on withdrawal. A person may legitimately cease a particular form of assistance without ceasing to be a parent, friend, citizen, or other relevant relation. In some cases, duties survive by changing their form; in others, the role itself ends, and its characteristic duties may end with it. But the status of the role, the nature of the duty, and the conditions of alteration require separate explanation. Our future inquiry should not presume that every change in form is a transfer, or that every reduction in performance is a failure of moral fidelity.

2.9. The object of a duty: performance, protection, and provision

A duty cannot be adequately identified without specifying its object. Compare a promise to deliver medication at six o'clock with an obligation to ensure that appropriate medication is available. The first is directed toward a relatively definite performance. The second may concern an outcome or standard for which several acts and arrangements could count as adequate. The difference affects whether delegation is permissible, whether personal identity matters, and whether alternative performance discharges the obligation. It also affects the seriousness of a claimed breach: a person can fail to carry out a particular promised act while nevertheless ensuring the protected interest is fully secured, or perform the promised act while failing in another duty to protect the person's safety.

We should be cautious about sorting all duties into a neat binary between acts and outcomes. Professional and fiduciary responsibilities often involve standards of reasonable diligence; their discharge may depend upon competent processes rather than successful outcomes alone. An undertaking to exercise careful judgment differs from a guarantee that judgment will produce a favored result. Equally, relationships generate duties of restraint—do not disclose a confidence, do not exploit another's vulnerability, do not misrepresent the nature of a service—that cannot be replaced by simply adding more beneficial activity. The object of a duty may therefore be action, restraint, protection, process, result, or some structured combination.

This complexity becomes acute when a relationship ends. A person who owes a defined payment may sometimes discharge the duty by arranging authorized payment through another person. But a person who owes a truthful admission of wrongdoing cannot necessarily substitute an anonymous charitable donation. An institution that owes protection against discrimination cannot automatically discharge that duty through a public statement of good intentions while continuing discriminatory practices. And a caregiver whose duty is to ensure adequate arrangements may be responsible for a handoff even though the recipient can obtain care from elsewhere, depending on the role and scope of the original undertaking. Practical benefits do not, by themselves, determine whether the specified object has been supplied.

The distinction also guards against making normative equivalence too narrow. Some duties are genuinely capable of substitution. A promise to provide transport may be adequately fulfilled through a safe paid ride if the parties' agreement and surrounding circumstances permit it. Insisting that the promisor drive personally when the original concern was timely transportation could fetishize the identity of the performer rather than honor the intended commitment. By contrast, another promise may have been made precisely because the promisee trusted that particular person's presence. Whether substitution is legitimate is partly a question of the duty's correct interpretation, not only a question about moral authority after the fact.

An exact account of relational transition must therefore avoid both presumed fungibility and presumed nonfungibility. The content of an obligation can determine whether another person may perform it; the authority of the claimant and the identity of the original bearer can determine what counts as valid modification; and independent obligations may govern the means by which the alternative is arranged. The evaluation has multiple dimensions. We should not compress them into the single question whether the recipient was left no worse off.

2.10. The continuing person and the changing duty

A person's moral identity is not a fixed set of desires, capacities, beliefs, and attachments. People develop new commitments, acquire disabilities, recover abilities, revise convictions, and cease to occupy roles through which earlier obligations arose. Yet a theory that permits duties to vanish whenever the duty-bearer changes threatens the possibility of stable interpersonal trust. A theory that requires every earlier commitment to retain identical force regardless of material change threatens to freeze persons within their past.

The difficulty can be framed without assuming a metaphysical theory of personal identity. The central question is not whether Mara is numerically the same individual after illness, or whether a person who no longer loves is literally a different person. In ordinary cases, personal continuity is sufficiently clear to attribute history and responsibility. What changes is the relevant factual and normative setting: capacity, risk, institutional position, the object of the promise, the claimant's standing, and the availability of acceptable alternatives. The moral issue is not solved by replacing an account of obligation with an account of authenticity.

Some obligations have content deliberately insulated from such changes. A repayment duty does not ordinarily depend upon the debtor's continuing enthusiasm. Others are expressly conditional: an agreement to help so long as one remains capable may terminate when capability fails. Still others demand adjustment under principles not wholly determined by the original undertaking. A duty of care may persist as a requirement to arrange adequate alternatives once direct performance becomes unsafe. But it may also legitimately end without any personal successor duty, leaving unmet need to be addressed through different social and institutional obligations. Which outcome obtains is not read from the abstract desirability of responsibility's persistence.

Betzler's account of divorce and Lopez-Cantero's critique illustrate why changes in emotional commitment generate distinct questions of moral authority. Ending a partnership may be necessary for the value of freely sustained romantic life, while the history of the relationship can generate residual interests in explanation, deliberation, and practical transition. Yet requiring explanation or participation may become ethically questionable where it burdens the departing person in ways not justified by the reasons for the breakup (Betzler 2024; Lopez-Cantero 2026). Their exchange demonstrates why we cannot settle all cases by treating either the past or the present self as normatively sovereign.

The useful analytical distinction is therefore between the historical grounds of a duty and the present conditions under which it is to be performed. A duty may have been generated by a past promise while its present content is affected by impossibility, release, countervailing obligations, or legitimate modification. It may also have been grounded in a role that no longer exists. To say the duty changes is not yet to explain how. The future chapters will examine the grounds and powers capable of producing such changes, resisting the temptation to turn the obvious fact of human transformation into a theory of moral release.

2.11. Delegation, substitution, and the duty that does not move

Consider two arrangements that produce the same immediate result. In the first, A promises to accompany B to an essential appointment, then arranges for C to drive B there. In the second, C independently promises to take B, and B releases A from the earlier commitment. Both arrangements may leave B with transportation. Only in the second is the formal structure of release apparent. In the first, A may still be accountable if C does not arrive, depending on what A promised and what the parties legitimately understood about substitution. The practical observation that “C took over” does not reveal the normative relationship.

The problem is sharpened when the identity of the duty-bearer is morally significant. A family friend who undertakes to care for someone entrusted to them is not necessarily permitted to transfer the entire arrangement to a stranger without consulting the person whose trust was part of the undertaking. Similarly, an organization cannot assume that a contractual assignment carries every moral duty attached to the relationship simply because another party now receives payments. Shiffrin's study of consumer debt exposes a structurally related error: transactions described as transfers of claims may in effect delegate obligations owed to the debtor, including responsibilities whose identity-sensitive performance matters to the debtor's legitimate interests (Shiffrin 2023). The significance of the case is not that duties cannot ever be delegated. It is that the transfer must be identified and justified rather than concealed under the apparent transferability of a financial asset.

We should distinguish transfer of the act, assumption of a new duty, authorized release of the original agent, and continued accountability for a delegated act. An agreement that C will perform A's task may create an obligation on C without extinguishing A's original responsibility. A's duty may be one to ensure that B receives a particular benefit, in which case competent delegation is one permitted means of performance. Alternatively, A may have promised personally to appear, in which case C's performance may not satisfy the promise without a valid alteration of its terms. Or A's original undertaking may have expired; C's new commitment then creates responsibility without receiving it from A at all. These possibilities are analytically distinguishable even when the outward result is the same.

The relevance of the duty-bearer's identity is not an invitation to an absolute doctrine of personal non-substitutability. Some promises concern outcomes whose value is independent of who provides them. To insist that the original promisor must personally transport B when a safe, agreed alternative is available could assign moral significance to an irrelevant feature of the act. The appropriate distinction depends on the object of the duty, the expectations legitimately created, the authority of the relevant claimant, and the conditions of any replacement. A theory of relational transition must explain when identity is constitutive of what is owed and when it is incidental.

We should therefore reject the proposition that obligations behave as conserved objects moving automatically from one person to another. In a genuine novation, waiver, or legal assignment, a particular practice may authorize some modification of responsibility; in informal social life, different grounds may yield functionally similar transitions. The relevant normative events must be separately established. This is a methodological caution rather than a newly demonstrated doctrine: contract theory and the analysis of normative powers have long distinguished various forms of release, transfer, and delegation. Any originality claim beyond that familiar architecture must be defended through a difference-making case.

2.12. Breach, residue, regret, and repair

An obligation may remain morally significant after its original performance becomes impossible or is no longer required. But the remaining significance might take several forms. It may be a directed duty of compensation for wrongful breach, an obligation to acknowledge damage, a reason to apologize despite justified nonperformance, a duty to return property, or an appropriate attitude of regret that does not itself license any further demand. Conflating these possibilities generates the illusion that any unfinished moral business must be an outstanding performance duty.

Suppose A promises to attend B's important ceremony but encounters an unforeseen emergency in which leaving would endanger another person. A misses the event. It may be true that the emergency justifies nonattendance. It may also be appropriate for A to express regret and offer an explanation. Yet those judgments do not settle whether B holds a directed claim to financial compensation or whether A has wronged B all things considered. Theories of moral dilemmas and moral residue dispute precisely what regret can reveal: some hold that justified action can leave morally significant regret without establishing that the agent committed a wrong; others give unresolved requirements a more robust normative status (“Moral Dilemmas,” Stanford Encyclopedia of Philosophy).

We should distinguish the persistence of an unsatisfied interest from the persistence of a duty, and both from the moral relevance of what has happened. Someone who suffered a disappointment may deserve acknowledgment without being entitled to choose the form of redress. Someone wrongfully harmed may have a directed claim to repair even where the wrongdoer has substantially improved their character. But an injured person cannot always demand the recreation of a relationship that no longer legitimately exists. The appropriate remedy depends on the nature of the wrong and its surviving effects, not simply the magnitude of the recipient's continuing dissatisfaction.

This distinction has a direct overlap with the author's separate manuscript The Moral Remainder, which argues that institutional transformation does not automatically constitute evidence that specific obligations have been discharged. This dissertation should not duplicate that argument under a different name. Its more limited task here is to distinguish duties that arise from wrongs from duties that may remain after permissible changes, and to specify what kind of normative event a “remainder” is supposed to name. An obligation may be discharged without an unwanted consequence disappearing; a valuable reform may occur while a duty of compensation remains; a moral loss may persist without a wrongdoer or identifiable holder of a claim. The distinctions are indispensable for what follows.

2.13. Institutional duties and the plurality of bearers

The temptation to convert practical continuity into moral continuity is strongest where institutions are involved. Institutions preserve procedures, records, financial flows, and services across changes of personnel. A patient may continue to receive medical treatment after the clinician retires; a student may continue to attend lessons after a teacher leaves; a beneficiary may continue to receive payments after a creditor transfers a claim. Continuity can be morally valuable because it protects people against the costs of disruption. But a continuous service should not be mistaken for evidence that all underlying obligations were appropriately reassigned or fulfilled.

An institution can itself be the bearer of duties that are not reducible to the personal duties of the staff currently performing tasks. A school can owe adequate educational provision even while individual teachers change. A hospital can possess responsibilities to organize continuity of care that are distinct from the personal professional obligations of any particular clinician. An employer can bear obligations toward workers that do not vanish when a manager resigns. The assignment of tasks inside the institution is not necessarily a transfer of the institution's underlying duty to those outside it, nor does successful operation always prove that its obligations have been discharged in a legitimate manner.

These claims require limits. Organizations do not possess all the agency, vulnerabilities, or accountability structures of individuals, and different moral theories disagree about the nature of collective responsibility. It would be mistaken to personify an institution and then infer that its duties operate exactly like a friend's promises. Institutional obligations are often mediated by public rules, roles, distributive principles, and administrative authority. A theoretical comparison must identify these sources rather than hide the difference under a general language of relation.

Miller's framework for distributing remedial responsibility is pertinent because the bearer of an obligation to remedy a problem need not be whoever is physically nearest or historically most helpful. Causal contribution, moral responsibility, benefit, capacity, and relationships of community can give different grounds for attribution (Miller 2001). Ong and Savulescu, likewise, argue that care responsibilities should not be confined to the apparent caregiver–recipient pair; a broader social ecology may impose duties on institutions to protect people against unreasonable caregiving burdens (Ong and Savulescu 2026). Any account that allocates obligations solely by the identity of the current provider would neglect these established insights.

The resulting problem is not simply which agent is cheapest or most effective. If a public institution has a duty to supply an accessible service, private generosity that temporarily meets the need need not discharge the institution's obligation. If an employee consistently fills a staffing gap, the availability of that labor may conceal the employer's responsibility to maintain sustainable coverage. If an AI system gives users useful assistance, its practical success does not by itself establish which human or corporate actor is accountable for privacy, security, reliability, or termination. The transition of benefits across agents is one fact; the legitimacy of the underlying arrangement is another.

2.14. A minimal formal map

To prevent ambiguity in later chapters, the analysis will use provisional notation. Let R(A,B,φ,t) mean that A has a moral reason concerning performance φ for B at time t. Let D(A,B,φ,t) mean that A owes B a directed duty to perform φ at t. Let L(A,φ,t) denote A's liberty to perform or omit φ, relative to a specified counterpart and duty; let P(X,ΔD,t) denote X's valid normative power to alter a specified duty; and let I(A,X,ΔD,t) denote A's immunity against X's attempted alteration of that duty. Let N(B,t) mark a significant unmet need of B, and H(A,B,φ,t₀) record A's historical provision of φ for B. These symbols are schematic, not a claim that every moral relation can be exhaustively formalized.

For present purposes, four non-entailments define the methodological task. First, N(B,t₁) and H(A,B,φ,t₀) do not alone entail D(A,B,φ,t₁). Second, L(A,omit φ,t₁) does not entail the absence of all duties owed by A concerning B: another duty with a different object may persist. Third, the fact that C performs φ does not entail that D(A,B,φ,t₁) has been discharged; that conclusion depends upon the duty's content and applicable rules of fulfillment or release. Fourth, a claimant's possession of D(A,B,φ,t₁) does not automatically imply P(B,replace φ with ψ,t₁), an unrestricted power to require a different performance.

There are important qualifications. The liberty notation is incomplete unless the relevant counterparty and act are specified: a person may have liberty relative to one actor while being subject to another duty arising from a different source. Likewise, the meaning of directed duty is disputed across moral theories; it should not be treated as definitionally equivalent to every use of “ought.” Our notation is intended to expose equivocations, not to manufacture rigor through symbols. A formalism that silently assumes its contested premises is weaker than an unformalized argument that makes them explicit.

These non-entailments also do not prove the absence of obligations. The first inference can become valid when supplemented by a binding undertaking or other sufficient grounds. The third may be valid under a practice that authorizes delegation and discharge. The fourth may become true where the promisor genuinely conferred a particular power of modification. The point is to specify the missing premise. A successful constructive theory must supply such premises, explain their comparative significance, and avoid generating duties by treating the resulting benefit as a surrogate for moral authorization.

2.15. Three rival objections

One objection is that this method makes relationships sound like contracts. People who care for friends seldom think in terms of claim-rights or powers, and a framework that treats every disagreement as a dispute between rights-holders can miss the meaning of shared life. The objection matters. We need this vocabulary not to govern affection but to clarify conflicts about what one person may demand of another, especially where vulnerability or unequal power raises the cost of error. Its value must be tested in those conflicts. If the distinctions obscure rather than illuminate what matters, the framework should yield to a better description.

The second objection charges that the theory is too protective of the departing agent. If every claim must be independently justified and every successor arrangement authorized, a person could withdraw from a dependent relationship while insisting that no perfectly specified duty requires further assistance. The danger is real, particularly where the original relationship was structured by unchosen care duties or severe power imbalances. The answer cannot be the bare assertion that the departure is permitted. General duties, role-based obligations, equitable burden-sharing, and institutional responsibilities remain open. The failure of one inference does not license refusal to examine other grounds. The proposed method guards against unwarranted personal appropriation only if it also recognizes genuine duties where voluntary promises are absent.

A final objection concerns originality. The distinctions developed here are already central to Hohfeld, Darwall, Raz, Owens, Shiffrin, care ethics, and theories of special obligation. This chapter does not offer a new theory of rights or duties. Its contribution is to establish a common vocabulary in which disagreements about withdrawal and succession can be framed precisely. Whether the resulting analysis yields anything beyond a synthesis of established views must be decided through the later comparisons, not asserted here.

That comparison needs a clear standard. A distinction is useful when it removes ambiguity from a case; it is philosophically original only if it also supports a justified implication the best rival accounts do not already capture. The constructive chapters must therefore present those rivals fairly, identify where their answers diverge, and explain why one answer is better grounded. An attractive example or a new name for the same obligation will not meet that standard.

2.16. Conclusion: Obligation as an answerable relation

A need is not automatically an obligation assigned to whoever can meet it. To say that someone owes another person an act is to make a claim about why that person is bound, what they must do, and who may hold them accountable. The relation may also give someone power to demand, waive, replace, or enforce performance, but such powers require justification of their own. The patient who still needs care does not necessarily have an indefinite claim against the retiring clinician. The clinician's release from one duty does not cancel every other responsibility. And a successor can provide the same benefit without inheriting the original obligation.

These distinctions matter because disputes about leaving can contain several questions at once. A person may have a reason to help without owing the requested act to a particular claimant. They may owe a specified act without granting the claimant broad control over their life. They may also be released from one commitment while remaining subject to duties created by law, role, wrongdoing, or the needs of others. The task is not to choose between care and freedom in the abstract, but to identify which moral relationship is actually at issue.

With those distinctions established, the next question is how a duty comes into existence. Chapter III asks why a promise, an encouraged expectation, an accepted role, or a vulnerable person's circumstances sometimes creates a claim against someone specific—and why the same facts sometimes do not. Only then can we decide what survives when the arrangement that gave the duty its form changes.

Continue reading

← Previous chapter   ·   Next chapter →   ·   Full contents

This is a developmental scholarly manuscript, not peer-reviewed or defended. The bibliography and research-status statement are accessible from the main contents page.