BOOK-LENGTH RESEARCH · DEVELOPMENTAL READING EDITION
Chapter I — The Need That Outlives the Duty
When a person who has helped for years must leave, what—not just whom—does anyone still owe?
← Complete contents · Chapter 1 of 8 · Updated November 2026
In this chapter
- 1.1. A person leaves; the need does not
- 1.2. Two continuities—and several kinds of discontinuity
- 1.3. Why boundaries are not enough
- 1.4. Five different events concealed by the word 'transfer'
- 1.5. The invalid conservation inference
- 1.6. The strongest rival explanations
- 1.7. Care before consent, and the distribution of burdens
- 1.8. A serious loss without a determinate wrongdoer
- 1.9. Counterexamples as constraints, not illustrations
- 1.10. What would count as a contribution?
- 1.11. Method and plan of argument
- 1.12. Conclusion: The past does not disappear, but it does not govern by default
1.1. A person leaves; the need does not
On the morning Mara tells Leila she cannot continue, the arrangements are still working. Clara, Leila’s mother, is eighty-two and needs regular assistance with meals, medications, mobility, and the ordinary administrative burdens of living with declining health. Mara has provided most of this assistance for nearly five years. She is not Clara’s daughter, employee, or court-appointed guardian. She became involved when Leila faced a temporary work crisis, agreed to cover several difficult weeks, and continued because she could help and because each temporary extension seemed preferable to leaving a vulnerable person without care. Over time her assistance became the architecture of the household. Leila accepted a position with longer hours. Clara came to depend on Mara’s familiarity with her routines. The names of alternative providers faded from practical consideration. No single conversation converted Mara’s initial willingness into a permanent promise; nevertheless, what had begun as extraordinary help gradually became the assumed condition of everyone else’s ordinary life.
Mara has now developed a serious medical condition. Her clinician advises against the lifting and interrupted sleep that caregiving requires. She tells Leila that she will stop providing direct care in three weeks. She offers to document medication routines, introduce an available replacement, and answer reasonable questions during the handoff. Leila replies that three weeks is not enough. The replacement is expensive; Clara dislikes unfamiliar caregivers; and Leila cannot immediately abandon her employment. Leila does not deny Mara’s illness. She claims instead that Mara has made herself too important to withdraw on terms of her own choosing. The arrangement has continued long enough to create obligations that did not exist when Mara first offered help.
Leila's objection has force. Five years of help shaped what the family believed it could safely plan. Extended care is not always morally equivalent to a single generous act; reliance may grow as an arrangement continues. But Leila moves too quickly from one truth to another. Clara still needs care. Does that mean Mara must provide it—or remain until Leila considers the handoff satisfactory? Clara's need is not in dispute. Mara's obligation is. Its extent depends on what Mara promised, what expectations she reasonably encouraged, what duties exist apart from a promise, what her health permits, and what Leila, institutions, or others owe.
The imagined case is intentionally less dramatic than the familiar opposition between the selfish deserter and the innocent dependent. Mara need not have behaved wrongly, and Clara may nevertheless be exposed to serious hardship. Leila may be understandably frightened without possessing authority to compel Mara to continue. An adequate ethical account cannot evade the difficulty by declaring one person’s needs illegitimate. Clara’s need is legitimate; Mara’s health-related limit is legitimate; Leila’s practical difficulty may be legitimate as well. The problem is that the legitimacy of all three does not, by itself, establish who owes which form of assistance.
This chapter begins from that gap. The puzzle is not whether people may leave relationships or whether the vulnerable matter. Both propositions deserve a place in any credible moral theory. The puzzle is whether the continuing existence of one person’s need licenses an inference to the continuing existence, transfer, or expansion of another person’s directed obligation. In many everyday judgments, this inference occurs almost without being noticed. The person who has always supplied the benefit is treated as its presumptive guarantor. The person who depends on its continuation is treated as entitled to select the conditions under which the provider can cease. A proposed replacement is treated as though accepting the practical work necessarily means accepting the entire moral position of the predecessor. None of these transitions is automatic.
1.2. Two continuities—and several kinds of discontinuity
The central fact in Mara’s situation is that practical and normative continuities can come apart. Clara’s needs extend across the date on which Mara is entitled to stop providing direct care. The practical arrangement may end while the reasons to protect Clara remain. Conversely, Mara may still owe a particular form of notice or disclosure even when she no longer owes physical caregiving. There is no contradiction in these judgments. They concern different objects.
Three things can continue, and they need not continue together. Clara's need may persist. The work of caring for her may pass from Mara to a professional. But Mara's responsibility might remain, change within agreed limits, end, or give way to a separate duty owed by someone else. Uninterrupted care does not tell us which moral change occurred. Mara acted from affection and informal commitments; a professional may act under a paid agreement specifying competence, oversight, and conditions of termination. The acts look similar. The sources and limits of responsibility may be entirely different.
The converse is possible too. If the replacement fails to arrive, Clara’s care may suffer even though Mara has satisfied every obligation she legitimately held. Describing this outcome as morally regrettable does not establish Mara’s wrongdoing. Moral failure in the situation—if there is one—might consist in Leila’s failure to arrange alternatives, a public system’s failure to provide accessible support, a professional provider’s breach of a newly assumed obligation, or some combination. Sometimes there may be grave loss without anyone having committed an actionable moral wrong. Such cases resist the comforting assumption that every serious need has a determinate private guarantor.
The distinction is compatible with the claim that obligations are temporally extended. Promises can bind after the motivations for making them disappear. Parental responsibilities are not created and dissolved each morning according to preference. Professional duties may constrain the manner of retirement or transfer. What fails is a more ambitious inference: that the persistence of a practical dependency, coupled with the historical fact of one person’s assistance, establishes the persistence of whatever particular obligation would most effectively maintain the earlier arrangement. Duties require grounds, and the grounds of an earlier duty may or may not justify its extension to the present case.
The moral stakes are not exhausted by the provider’s freedom. A theory that takes justified withdrawal seriously but leaves a vulnerable person unsupported has not thereby explained how unmet need should be addressed. It has shown only that one proposed assignment of responsibility is inadequately grounded. Some needs give rise to general duties of beneficence; some activate specific roles; some are properly matters of institutional justice. These different grounds must not be collapsed into one undifferentiated expectation of care. Ong and Savulescu’s account of relational responsibility insists upon the wider social setting in which caregiving burdens are distributed, including the responsibilities of collectives to address conditions that impose unreasonable burdens on individuals (Ong and Savulescu 2025/2026). That insight prevents the exit question from swallowing the more comprehensive question of justice.
The present argument accordingly declines two rival simplifications. The first treats ongoing dependence as an automatic title to the previous provider’s future labor. The second treats the provider’s permissible withdrawal as evidence that no further moral question remains. The first threatens to make generosity a source of indefinite liability; the second threatens to treat the absence of one directed duty as if it exhausted society’s reasons to respond. Between them lies a problem of moral specification: exactly which duty exists, who bears it, to whom is it owed, what authorizes changes in its performance, and what determines whether it has ended?
1.3. Why boundaries are not enough
The language of boundaries has considerable practical merit. It tells people who feel submerged in the demands of others that an encounter with another person’s need does not automatically transfer ownership of that need to them. It protects a domain of legitimate refusal, especially where a person has learned to experience the disappointment of others as evidence of their own moral failure. Yet the vocabulary is insufficient for the present problem. To say that Mara may set a boundary supplies only one part of the judgment. It does not establish whether Mara has previously promised a transition period, whether she must disclose information that only she possesses, whether Leila has a right to demand a particular handoff, or whether someone else is now obligated to assume care.
A boundary usually identifies a limit on what may be required of the person asserting it. The problem here concerns the structure of the remaining moral relations. Who holds the claim? What is its object? Can the claim be waived by its beneficiary? Is its performance delegable? Is it an obligation to provide a result or to perform a particular act? Does the putative successor assume the original duty or independently incur another one? Those questions cannot be resolved by asserting a general right to self-direction. Nor can they be resolved by invoking care in the abstract. The moral significance of dependency must be translated into more specific reasons, duties, powers, and restrictions.
Nor should the alternative to boundary-setting be a romantic ideal in which every obligation is negotiated until all parties are satisfied. A person leaving an abusive relationship may justifiably refuse further interpersonal bargaining. A caregiver facing genuine danger may need to stop immediately. A professional may have obligations to patients that are governed by standards substantially different from the norms of friendship. Attention to these differences is not an inconvenience to be eliminated by a universal maxim; it is what an adequate theory must preserve.
Betzler’s ethics of divorce provides an important counterweight to accounts that celebrate the right to exit while neglecting the person whose life has been organized around the relationship. Her discussion of explanation, participation, transitional goods, and relational freedom makes clear that departure may be permissible without making its manner ethically indifferent (Betzler 2024). Yet Lopez-Cantero’s subsequent critique reveals a serious danger in extending deliberative duties too far: requiring reasons or participation may cease to be action-guiding when love ends without a further explanation, and it may subject the person leaving to intrusive demands (Lopez-Cantero 2026). Their disagreement cannot be settled by observing that the departing person has boundaries, or that the person left behind deserves care. The rival considerations need a theory of the forms and limits of obligation.
The larger question is therefore not how to become independent of others. It is how to specify interpersonal responsibilities so that legitimate obligations retain their force without allowing the historically successful exercise of care to become evidence of unlimited authority over the giver. This is a philosophical question about reasons and normative relations, not a clinical theory of why individuals struggle to say no. It has practical consequences, but its proper method is first to determine which inferences about responsibility are valid.
1.4. Five different events concealed by the word 'transfer'
A commonplace instruction—'make sure someone else takes over'—can describe several ethically distinct events. First, a person may delegate an activity while retaining responsibility for ensuring that it is properly performed. A clinic might assign an administrative task to an employee while retaining institutional duties toward a patient. The task changes hands; the accountability may not.
Second, a duty-holder may be released from an obligation by an agent with the appropriate authority. The person owed performance may waive a promise when waiver is valid and when no independent obligations prevent release. The original obligation then ends. Its termination need not create an identical duty in any successor.
Third, a new participant may undertake a distinct obligation. A caregiver who accepts employment acquires duties through the new professional and contractual relationship. Those duties may supply equivalent practical benefits while differing in scope from the informal responsibilities of the previous helper.
Fourth, the original duty may be modified while remaining in place. A promise to provide direct service may, in certain circumstances, become an obligation to furnish specified support through an agreed alternative arrangement. But the legitimacy of that modification must be established; it cannot be inferred solely from the convenience of the new arrangement.
Fifth, the original duty may permissibly terminate while the beneficiary’s need remains unaddressed. This is a morally difficult possibility, but denying its existence by definition would erase the distinction between the existence of a need and the authority to impose a particular directed obligation. If the need ought to be met, the responsibility must be grounded elsewhere rather than silently assigned back to the person who previously helped.
These distinctions are neither unknown nor original to the present inquiry. Shiffrin’s analysis of hidden delegations demonstrates how apparent transfers of contractual rights can also involve normatively significant transfers of duties, and why the identity of the duty-bearer matters (Shiffrin 2023). Hohfeldian analysis provides a grammar for separating claims, liberties, powers, and immunities; one person's claim that another perform an act does not itself establish an unlimited power to alter every duty related to the relationship (Wenar 2025). Sakharova’s critique of a one-sided account of contractual correlativity further challenges the idea that a claimant simply governs the promisor’s normative position (Sakharova 2024).
The burden of a dissertation is not to announce this architecture as undiscovered. It is to show what a grounds-sensitive account of changing responsibilities can explain beyond those already powerful resources. To do that, we first need a more exact description of the inference that creates the problem.
1.5. The invalid conservation inference
The temptation to infer duty from need is best understood as a failed conservation inference. Where a practice has met an important need over time, observers often speak as if the obligation to maintain that practice must remain somewhere in the system. If one person may stop, another must take over; if no one has taken over, the original provider must still owe continuation. This reasoning may express an admirable refusal to abandon the vulnerable. But it can also move from an evaluative proposition—that the need matters and its neglect is undesirable—to a directed normative proposition identifying a determinate duty-holder. The former does not entail the latter.
To make the distinction precise, let N(B,t) denote that person B possesses a significant unmet need at time t. Let H(A,B,φ,t₀) denote that A previously performed service φ for B. Let D(A,B,φ,t) denote a directed moral duty owed by A to B to perform φ at t. The following inference is invalid absent further premises:
N(B,t₁) ∧ H(A,B,φ,t₀) ⟹ D(A,B,φ,t₁).
Suppose an acquaintance brings meals to a patient for a week after hospitalization and makes clear that the arrangement is temporary. The patient still cannot cook when the week ends. The need persists, but the week's generosity does not by itself create a personal duty to keep bringing meals. That conclusion does not leave the patient morally invisible. A bystander may owe emergency aid where the cost is slight; the discharging institution may owe appropriate care planning; relatives may have duties arising from their roles. The question is not whether anyone should help. It is why this particular helper would owe this particular assistance.
The opposite inference is equally defective. If A is permitted to cease φ, it does not follow that A has no remaining duties toward B. Permission to end one performance does not logically cancel a prior promise of notice, a requirement to return entrusted property, an independently grounded duty to avoid causing foreseeable injury, or obligations arising from an earlier wrong. A clinician who is entitled to retire is not therefore entitled to disregard every professional responsibility concerning the safe disposition of records. A person who may end an intimate partnership may retain a contractual or independently justified duty of financial support. The permission concerns an action under particular circumstances; it does not erase the normative history of the relationship.
A third invalid inference concerns functional substitutes. Let S be a proposed successor who begins performing ψ in a manner that supplies benefits comparable to φ. The fact that S now provides help does not demonstrate that A's original duty was validly transferred to S or that A has been discharged. A person may accomplish a task on another's behalf without acquiring the standing, responsibility, or authority that made the original bearer accountable. A task can be performed adequately while the arrangement governing who may direct it, contest it, terminate it, or answer for failure remains defective. Conversely, a legitimate successor may incur duties through a new undertaking without inheriting the former provider's entire moral history. Practical equivalence and normative equivalence must be argued separately.
The word 'conservation' here names a recurring pattern of reasoning, not an empirical psychological effect that this dissertation has established. It is a conceptual diagnosis of a possible inference in ordinary and institutional judgment. The central research question is whether careful normative analysis of that inference yields a genuine explanatory contribution beyond familiar accounts of directed duties, permissions, normative powers, and remedial responsibility. A chapter that simply proves the elementary invalidity of an inference would not meet the standard of an original dissertation. The difficult work begins where the inference's conclusion is sometimes true—where the original provider really does owe continued support—and where the grounds that make it true must be identified.
1.6. The strongest rival explanations
Rights theory already supplies a disciplined language for distinguishing these questions. In Hohfeldian terms, a claim-right corresponds to another agent's duty, while a liberty indicates the absence of a contrary duty; a power concerns the ability to change normative relations, and an immunity corresponds to another's lack of such a power. These are different normative incidents, although actual legal and moral arrangements may combine them (Wenar 2025). A beneficiary might hold a claim against a provider without holding an unrestricted power to require any form of substitute performance. A provider might enjoy liberty to cease one activity while remaining subject to a distinct duty. Before drawing philosophical conclusions about transfer, one must specify the relevant incidents and the agents to whom they attach.
These distinctions describe the structure of a claim, not the reason it is justified. They help explain what it would mean for Clara to have a claim against Mara, but do not show that such a claim exists. Nor do they decide whether Leila could modify it or whether the community owes Clara another form of support. A theory of rights must do more than classify obligations: it must explain why they bind particular people. Philosophers of rights already recognize this distinction between identifying a claim and justifying it. This project uses that distinction; it does not claim to have discovered it.
Promises matter because other people make plans around them. If someone could withdraw every assurance whenever it became inconvenient, cooperation would be precarious. But the existence of a promise does not give its beneficiary unlimited authority over the promisor. Owens emphasizes the promisee's interest in having authority over the commitment; Sakharova challenges accounts that make that authority too nearly sovereign. Their disagreement concerns what control a promise legitimately creates, how participants stand in relation to one another, and where that control ends—not whether promises have moral force (Owens 2012; Sakharova 2024).
Shiffrin's analysis of contractual assignment is especially threatening to the novelty of this dissertation, and therefore especially valuable. She argues that a transfer presented as an assignment of rights may covertly delegate duties and that a beneficiary's legitimate concern with the identity of the person undertaking an obligation can make the transfer morally significant. Her analysis shows why the transfer of a financial claim should not be assumed to involve only the reassignment of a commodity-like entitlement. The creditor's obligations toward the debtor may travel in ways that ordinary assignment practices fail to acknowledge (Shiffrin 2023). It would be untenable to claim that recognizing the difference between moving a task and relocating a duty constitutes a novel contribution after this work.
A charitable reconstruction of Shiffrin therefore establishes a stringent standard. Suppose we can describe a caregiving transition as a hidden delegation and identify when the identity of the substitute matters. Unless our analysis produces a further, defensible implication, it will have illustrated Shiffrin rather than surpassed her. The same standard applies to rights theory: to insist that permission to leave and a duty to provide notice are distinct is to use an existing conceptual apparatus properly. The dissertation earns its place only if it exposes a remaining uncertainty within the strongest combined account of the relevant rights, authority interests, and special obligations.
At this stage, one candidate remains open. Informal relationships are often marked by the absence of a clear promisee, release authority, or accepted substitute, even though significant reliance exists. Their moral life may not be adequately mapped by describing a single directed duty and its formal delegation. Yet this observation cannot serve as a novelty claim until compared with contemporary care ethics and theories of remedial responsibility, which were developed partly to account for responsibilities that arise outside voluntary contracts.
1.7. Care before consent, and the distribution of burdens
The caregiver case looks different when viewed from the standpoint of a theory of care rather than the standpoint of promissory authority. Mara is not the sole protagonist. Clara's vulnerability is not an artifact of someone else's expectations; it is an enduring feature of her present condition. Nor can the moral importance of care depend entirely on whether a caregiver previously volunteered. People enter caring relationships through birth, disability, illness, social arrangements, accidents, and institutions in which the choice to assume a particular responsibility is constrained or absent.
Chadha-Sridhar's account of the grounds of duties to care is an important recent challenge to any view on which only voluntary undertaking can generate interpersonal responsibilities. She argues for a hybrid non-voluntarist position on which natural duties provide the normative basis and particular relationships activate those duties toward determinate persons (Chadha-Sridhar 2026). A dissertation that treats the absence of an express promise as decisive against care obligations would overlook that argument. It would also be poorly positioned to explain the responsibilities of parents, guardians, or public institutions whose duties are not adequately reconstructed as exercises of ordinary contractual consent.
Yet identifying a duty of care is not the same as establishing that an identified person owes every form of care the beneficiary needs. There are questions of intensity, duration, opportunity cost, institutional background, and the legitimacy of a preexisting distribution. Ong and Savulescu argue that special responsibilities must be understood within broader networks, and that collective institutions have duties to mitigate unreasonable burdens (Ong and Savulescu 2025/2026). The significance of this view for Mara's case is not that it automatically frees her. It is that the moral field includes more actors than the recipient of care and the person who has historically provided it. An arrangement's success in concealing institutional absence does not establish that the individual filling the gap bears indefinite responsibility for doing so.
Miller's account of remedial responsibility identifies several reasons someone might be required to address a problem: they caused it, contributed to it, benefited from it, have unusual capacity to help, or belong to a community that must respond (Miller 2001). Different reasons may point toward different people. Mara might have a duty arising from a clear promise to remain until replacement care was secured. Leila's reliance might matter if Mara knowingly encouraged it. A public service might bear primary responsibility if it withdrew without planning. And a nearby stranger might owe urgent rescue regardless of any earlier relationship. The continuing need does not select its duty-bearer automatically.
These distinctions forbid a convenient moral arithmetic in which capacity, benefit, causation, promise, and history are added together until the largest total indicates the responsible person. The importance of these considerations does not make them commensurable or interchangeable. A promise supplies a different kind of reason from a general duty to aid. A duty of an institution to provide equitable access has a different normative structure from an informal obligation of friendship. Even where the resulting acts are identical, the reasons that justify requiring them need not be.
The dissertation must therefore address not only the quantity of burden but the authority under which burden is assigned. A person's historical competence is relevant to whether they can help, but not sufficient to establish that they are the legitimate party to whom the costs of everyone else's planning should be charged. At the same time, others' freedom not to be appropriated as a resource cannot be a universal solvent for relationships involving real promises and serious dependence. The moral task is not to minimize obligation but to identify its grounds and legitimate limits.
1.8. A serious loss without a determinate wrongdoer
One of the obstacles to clear thinking about relational change is the impulse to locate every serious loss in a single person's misconduct. The impulse has understandable sources. To identify a responsible wrongdoer promises an actionable answer: demand repair, require renewed performance, or allocate the cost to the person who broke the arrangement. Yet loss and wrongdoing are not coextensive. In Mara's case, a justified cessation of unsafe physical work can coexist with Clara's continuing need and Leila's real hardship. The facts may support sorrow, anxiety, and a powerful public reason to provide care without proving that Mara wrongs either person by ceasing direct service.
The philosophical literature on moral dilemmas and moral residue helps explain why such circumstances remain ethically troubling. Agents can appropriately regret losses that accompany actions they were justified in performing, and regret is not identical to culpability. It would beg the question to treat every painful consequence as evidence that someone has committed an all-things-considered wrong (McConnell and others, discussed in the Stanford Encyclopedia of Philosophy, “Moral Dilemmas”). The difference between regret and guilt matters here because a theory of transition that equates unavoidable loss with misconduct will repeatedly overassign duties to whichever participant happens to remain visible.
A no-wrongdoer case does not show that no one has responsibilities. Emergency rescue duties can be owed regardless of fault. Institutions may owe support because persons have independent claims to basic goods. Families may have obligations grounded in their relationships rather than in prior injuries. The task is to prevent the inference from “someone ought to respond” to “the person who has been responding must continue.” Even the claim that some assistance ought to occur may be indeterminate in its personal allocation. Moral reasons can be strong while the question of who is entitled or required to act remains contested.
Calling an arrangement voluntary does not settle whether it was fair. A daughter may provide years of unpaid care because disability services were unavailable. She may be legally free to stop, yet unable to do so without severe consequences for those she loves. Her obligations cannot be assessed simply by asking whether she signed up. Sex, class, disability, immigration status, and institutional design shape who becomes responsible and who pays the price of leaving. Treating these conditions as irrelevant can disguise an unfair distribution of labor as personal commitment. We must examine whether any undertaking was genuinely voluntary, what duties exist without consent, and whether institutional failure has been normalized through one person's sacrifice.
These concerns reveal why a narrow thesis about continuity is inadequate. It is not enough to state that obligations can end while needs remain. We must also explain when obligations never legitimately belonged to the apparent provider, when they were genuinely undertaken, and when public duties exist whether or not any private party succeeds in withdrawing. That inquiry cannot be completed by introspection about the provider's boundaries. It concerns the moral distribution of vulnerability across a social order.
1.9. Counterexamples as constraints, not illustrations
Consider first a parent who wants to cease caring for a young child. The child’s needs remain, and the parent’s role is not reducible to voluntary helpfulness. It would be absurd to infer from the general value of autonomy that the parent may terminate obligations at will. The theory must preserve duties grounded in parenthood, guardianship, and the child's protected interests. An authorized adoption or other appropriate arrangement may eventually relocate aspects of care, but the mere availability of a competent substitute does not automatically release the parent from all relevant responsibility. A theory that cannot distinguish this case from the generous neighbor is insufficiently sensitive to the grounds of obligation.
Consider, next, a person withdrawing from an abusive partner who threatens self-harm unless personal contact continues. The threat may involve urgent needs for protection. It does not therefore establish a right to control the withdrawing person's body, intimacy, or continuing exposure to danger. Assistance through emergency services or other safe channels may sometimes be appropriate, but it would be ethically perverse to treat the threatened need as an unrestricted source of coercive authority. A theory that imposes a categorical duty of negotiated transition would fail precisely where the relevant risks are most severe.
Consider a professional who can no longer perform a service but holds information indispensable to a safe transition. Depending on the professional role and applicable rules, providing records or a competent handoff may be obligatory even when personally continuing the service is not. A helpful substitute does not erase questions of confidentiality, appropriate authorization, and accountability. Yet it would be a mistake to generalize the requirements of a licensed, regulated profession directly to friendship, family life, or informal mutual aid. The former may involve explicit public standards that do not exist in the latter.
Finally consider an organization whose production system depends upon a volunteer's exceptional skill. Because the volunteer has repeatedly prevented failures, executives have delayed hiring staff, building redundancy, or creating documentation. The volunteer announces withdrawal. Management insists that the volunteer owes a long transition because the organization will otherwise suffer significant losses. The losses are foreseeable; the question is who created and who legitimately bears the risk. If the organization chose to make the volunteer indispensable without negotiating a corresponding commitment, the very dependence it now cites may be evidence of its own planning decisions rather than the volunteer's enlarged duty. But if the volunteer previously accepted a clear handoff obligation and induced justifiable reliance, that undertaking changes the case. Neither “the volunteer is free” nor “the organization depends on them” supplies a complete judgment.
These cases are designed to constrain the theory rather than decorate it. A proposal that assigns transition duties wherever harm can be mitigated risks coercion. A proposal that locates all obligation in explicit promises risks abandoning persons whose vulnerability activates non-voluntary duties. A proposal that prioritizes continuity risks preserving unjust arrangements. A proposal that assigns costs to the most capable agent risks converting competence into an unbounded public resource. A proposal that treats every outcome-equivalent replacement as normatively equivalent risks neglecting the authority and accountability through which service is legitimately provided.
The same conceptual hazards arise in artificial-intelligence services, but the analogy must be handled cautiously. When a provider retires a personalized system, users may lose access to useful histories, practices, and forms of assistance. The provider may have privacy, portability, contractual, or fair-notice obligations. A new system that performs similar tasks is not automatically a lawful or morally adequate successor to the original arrangement. Yet no claim about the “duties” of the AI system itself is needed to make this point. The relevant questions concern human and institutional responsibility and the consequences of technology-mediated reliance.
1.10. What would count as a contribution?
We should now state the research burden without euphemism. Nothing proved in this chapter establishes a novel philosophical principle. Hohfeldian analysis already distinguishes kinds of normative standing. Owens investigates why obligations and powers over them matter to persons. Shiffrin shows that changes of apparent rights ownership can conceal moral transfers of duty. Sakharova challenges an overly unilateral view of claim-holder control. Betzler and Lopez-Cantero contest the scope of residual duties arising when intimate relations end. Chadha-Sridhar supplies a non-voluntarist grounding of care duties, Ong and Savulescu situate such duties in wider social environments, and Miller develops considerations relevant to the allocation of remedial responsibility. A serious doctoral project must assume these theories are stronger and more comprehensive than a passing summary can establish.
There are at least three ways the dissertation might eventually contribute. It might demonstrate a genuine inconsistency among otherwise plausible accounts and defend a principled resolution. It might identify a type of case systematically misclassified by the strongest combination of existing theories and explain why the error matters. Or it might establish a previously unrecognized normative consequence concerning the authority to assert that a duty has been transferred, satisfied, or dissolved. The last possibility is especially challenging because earlier work on moral repair, contractual waiver, and delegation already explores the allocation of such authority. The thesis must therefore be earned in direct comparison with primary texts and cases in which theories are asked to produce differing judgments.
This chapter's more limited contribution is an organizing diagnosis. The practical continuance of a need does not, without further normative grounds, identify the bearer of a directed duty; the practical continuance of assistance does not, without further grounds, prove the continuance of a prior obligation; and the justified ending of an activity does not, without further grounds, establish the absence of all remaining responsibilities. Each claim identifies an inference requiring justification. Together they define a research problem; they do not yet solve it.
A plausible account of changing obligations must identify their sources, requirements, and legitimate limits. It should explain both why some duties survive a person's departure and why others do not, without treating either continuity or release as the default. The test is comparative: can this account resolve a difficult case more convincingly than the strongest established theories? A synthesis that cannot do so may still be valuable, but its contribution would be clarification rather than a new philosophical doctrine.
1.11. Method and plan of argument
The inquiry is primarily normative and conceptual. It does not begin by asking survey participants whom they blame, although such evidence might later illuminate ordinary patterns of moral attribution. People's intuitive judgments would not, by themselves, justify a duty. Neither does the inquiry attempt to derive moral rules from the frequency with which informal assistance becomes expected. Descriptive prevalence cannot settle normative authority.
The method proceeds in four stages. It first reconstructs the strongest available accounts of duties, claims, powers, promises, care, and remedial responsibility. It then isolates the facts that might change an obligation: a promise, induced reliance, vulnerability, prior wrongdoing, an accepted role, or a publicly assigned responsibility. Comparing cases in which one fact changes at a time helps identify which facts matter. Finally, it tests proposed principles against difficult cases involving children, coercion, unjustly imposed care, professional accountability, and institutional planning failures. The goal is an explanation that survives serious counterexamples, not appealing answers built into the examples themselves.
The argument proceeds in stages. Chapter II asks what it means to owe someone a duty, distinguishing claims, liberties, powers, immunities, and general reasons to help. Chapter III examines how promises, reliance, relationships, and care create obligations. Chapter IV asks what survives a justified ending, including duties to repair or assist and losses that involve no wrongdoing. Chapter V distinguishes someone else's performance of a task from a valid change in responsibility. Chapter VI tests an account of those changes against established theories. Chapter VII brings the account to cases in family life, professional practice, institutions, and AI services. Chapter VIII states what can be concluded—and what remains unproved.
This order is also an intellectual safeguard. We cannot decide in advance that the caregiving case exposes a new moral principle and then select theories too weak to explain it. We must first let established accounts of rights, promises, care, and social responsibility address the case on their own terms. The later chapters will ask whether any problem remains after that comparison. The result might be a new implication, a sharper application, or a rigorous synthesis; the argument must determine which.
1.12. Conclusion: The past does not disappear, but it does not govern by default
When Mara stops caring for Clara, something valuable may be lost. The loss deserves attention irrespective of whether Mara acted wrongly. Clara's need does not vanish when Mara's body reaches its limit. Leila's life may have been built around a history that now imposes real costs. Yet none of this establishes, without more, that Mara must preserve the arrangement, that Leila may dictate the terms of her withdrawal, or that a proposed replacement automatically inherits the original obligations. Their moral relationship must be described with more care than the binary language of abandonment and freedom permits.
The example also exposes a difference that the next chapter must make precise. Clara's need is real, and Mara's history with her is morally significant. Yet a reason to care about Clara is not necessarily a claim against Mara, and a claim against Mara would not automatically give Leila authority to dictate every means of performance. A successor who provides comparable help may occupy a different moral and professional position. Each of those relations must be examined before an ending can be called abandonment or a handoff can be called complete.
The inquiry begins with questions, not a rule that resolves them in advance. What made an obligation binding? To whom was it owed? Who had authority to alter or release it, and within what limits? Our histories matter because they can create responsibilities that survive change. But history does not, by itself, give another person authority over the rest of our lives. The philosophical task is to explain when the past makes a justified claim on the future and when that claim exceeds its proper reach.
Continue reading
← All chapters · Chapter II: What It Means to Owe →
This is a developmental philosophical manuscript. It has not undergone peer review or doctoral examination. References and the research-status statement are available on the book's main page.