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 III — How Relationships Produce Reasons

How promises, reliance, caregiving, and imposed roles can create—or fail to create—obligations.

← Complete contents · Chapter 3 of 8 · Updated November 2026

3.1. The commitment that was never spoken

At the beginning, nobody thought an agreement was necessary. When her neighbor Rosa returned from the hospital, Elise left a prepared meal at her door, brought the recycling bins inside, and stopped by to check whether prescriptions had been collected. Within weeks these gestures became routine. Rosa knew which evenings to expect food, and Elise knew the appointments that were difficult for Rosa to reach. As Rosa regained some strength, Elise began helping with correspondence and household bills. She did not ask to become a caregiver. She did not promise to continue indefinitely. Yet as months passed, Rosa abandoned the search for a paid service; her adult children assumed that Elise had the matter in hand; and Elise, seeing the arrangements that now depended on her, increasingly hesitated to explain that she had never agreed to become their foundation.

There are two familiar but incomplete ways to describe Elise's position. One says that the obligations are hers because Rosa has come to rely upon her. The other says that they cannot be hers because she never promised anything. The first draws a morally important connection between another person's dependence and the continuity of one's actions, but risks making all effective generosity into an involuntary commitment. The second protects the moral difference between a freely undertaken responsibility and a benefit supplied without any undertaking, but treats the absence of a promise as capable of exhausting the question of what can be owed. Neither description adequately captures the fact that the parties' conduct has a history, that the history has altered the distribution of risk, and that reasons for assisting can emerge from more than one source.

Chapter II distinguished needs, duties, claims, and powers. Those distinctions tell us what it means to say that something is owed; they do not tell us why it is owed. A theory can classify a claim precisely and still fail to justify it. It can name the supposed duty-holder without explaining why that person should bear the duty. The question now is one of moral grounds: what makes a particular act binding on this person, toward this recipient, at this time?

We are not searching for unwritten contracts hidden inside every relationship. Promises, assurances, repeated conduct, dependence, accepted roles, joint activity, vulnerability, and social institutions can all affect what people owe one another. But they do so differently. Some create definite duties to particular people; some give reasons to help without specifying a single obliged provider; others locate responsibility in an institution rather than the person visibly supplying care. Several grounds may operate together, and one may limit another. A caregiver's promise can matter without vindicating a social system that presumes her endless availability. Someone may lack a special claim against a former helper and still have an urgent claim to public assistance. The question is which facts justify which obligations.

3.2. The difference between creating a duty and recognizing one

A distinction between the creation and recognition of duties must govern the analysis. A promise can, under appropriate conditions, change the normative landscape: something that was once optional may become owed because the parties undertook a commitment. But a person's recognition that another is hungry does not necessarily create hunger's moral significance, nor does their refusal to acknowledge a duty prove its absence. Moral grounds may precede a person's recognition, or may depend upon acts through which the parties deliberately alter what they owe.

The distinction becomes crucial when one party can dictate what an arrangement supposedly means. A public institution may have depended for years on unpaid help and later insist that the volunteer had accepted responsibilities that were never agreed. Conversely, parents cannot disclaim their minor child's basic welfare simply because they did not volunteer for parenthood. The first case requires evidence of a valid undertaking, a justified expectation, or another ground of duty. The second involves obligations that do not depend entirely on consent. Calling both “relationship-generated obligations” without explaining that difference would obscure the issue.

We should distinguish why a duty is binding from what brings it into force toward a particular person. The first is its ground; the second, its trigger. Philosophers disagree about how to formulate that relationship, but Chadha-Sridhar provides a relevant example. Her 2026 account places the moral ground of care duties in a natural duty, while particular relationships activate it toward identifiable people. It therefore rejects both the view that care always requires a voluntary undertaking and the view that relational facts alone supply all the duty's force (Chadha-Sridhar 2026). This distinction is central to her work, not an innovation of this dissertation.

Even when care is owed, we must still ask what is owed. A duty to secure adequate care need not require the same person to perform every task indefinitely. Parents may meet some obligations through competent alternative provision. Neighbors may owe warning before ending an arrangement without owing another year of direct help. A local government may have to address a serious service gap without assigning it to any one employee. The ground of the duty, the circumstances that activate it, and its particular content must each be established. A useful theory should explain those differences rather than treat them as exceptions.

The difference among ground, trigger, and specification will later allow us to test claims that duties “migrate” through relationships. If the original provider's obligation arose from a promise while the successor's duty arose from an accepted professional role, practical continuity may conceal normative discontinuity. But the observation itself is not an original doctrine. It tells us what must be investigated before any claim about succession can be defended.

3.3. Promises as deliberate changes in what is owed

Promises occupy a distinctive place because, at least in paradigm cases, an agent intentionally undertakes a responsibility directed toward another. The act allows individuals to coordinate activities whose value depends upon mutual assurance. A promise to meet a friend at a hospital, to care for a dependent during a scheduled absence, or to finance a course of study can make particular plans rational in ways that vague goodwill cannot. The promisee need not remain constantly anxious that the promisor will reevaluate the entire arrangement whenever circumstances become less convenient. Commitments provide a protected interval within which another person may plan.

Why does a promise bind? Competing answers emphasize the social practice of promising, the assurance knowingly given, the promisee's reliance, or the authority a promisor gives another person over future conduct. These explanations differ. A reliable prediction need not rest on an obligation, and a valid promise may bind before anyone has acted in reliance on it. The Stanford Encyclopedia of Philosophy surveys conventional, expectation-based, assurance, reliance, and trust accounts (“Promises,” 2022). A serious comparison must preserve their disagreements instead of treating them all as a single theory of commitment.

Scanlon's principle of fidelity is especially important because it incorporates intentional assurance and the promisee's knowledge of the assurance. His account is not that any expectation automatically binds the person expected to perform. It identifies morally significant conditions under which one person intentionally leads another to expect performance, knowing that the other wants this assurance (Scanlon 1998, chapter 7; “Promises,” §3.1). This offers a clear contrast to Elise's case. Rosa might predict that meals will arrive because they have often arrived; whether Elise intentionally assured Rosa that the pattern would continue is a separate matter. A theory that moves from predictability to promise erases the moral significance of communicative intent.

The same distinction makes promises resistant to mere preference change. If Elise had expressly committed to six months of meals while Rosa recovered, then Elise's wish to reclaim her evenings would not ordinarily extinguish the undertaking. The promise matters precisely because it survives fluctuations that would otherwise destabilize reliance. But the promise's terms would still require interpretation. Was Elise undertaking personal preparation of each meal or a commitment to ensure that meals were available? What if the work became medically unsafe? What if Rosa voluntarily released her? The questions demonstrate why a promise's persistence does not establish unlimited authority to redefine its content.

Raz, Owens, and Shiffrin each supply important perspectives on the moral significance of normative authority, promises, and intimate assurance. Their accounts should be reconstructed separately in a later chapter rather than invoked as a composite endorsement of the present project. For our purposes, the provisional lesson is narrower: a legitimate undertaking can create obligations that are neither reducible to beneficial consequences nor indefinitely extensible beyond the authorized object. Where the parties disagree about whether an act of promising occurred, the dispute concerns communicative history and the valid conditions of undertaking, not the desirability of continued help alone.

3.4. Reliance: prediction, encouragement, and moral responsibility

Reliance introduces a different problem. Someone may alter their conduct because they reasonably expect another to act. The expected performance might be desirable, even essential, without having been promised. If the expectation was deliberately induced, a person may have strong reasons not to defeat it without warning or justification. Yet dependence on another's behavior does not automatically give the dependent person a right to require its continuation. The moral significance of reliance therefore depends on how it was formed, what the relied-upon person knew, the costs of disappointed expectation, and the opportunities the parties had to clarify their roles.

Consider three variations. First, Rosa assumes Elise will bring meals for a year despite Elise's clear statement that the help is temporary. Second, Elise repeatedly assures Rosa that she need not arrange alternatives, while privately doubting she can continue. Third, Elise makes no promise but discovers that Rosa has abandoned other care because she trusts the established routine; despite foreseeing serious harm, Elise says nothing. The same unmet need may give rise to different duties. Rosa's unilateral assumption in the first case is a weak basis for a claim. The second involves an assurance that may have induced harmful reliance. The third requires closer examination of foreseeability, due care, and fair dealing. Neither “no promise, no duty” nor “reliance always binds” can settle all three.

Scanlon's work on expectations and due care helps explain duties that arise without an explicit promise. Someone may be wrong to encourage a false expectation when another person is likely to act on it at serious cost (Scanlon 1998, chapter 7). The absence of a spoken promise therefore does not settle every duty to communicate. Nor is induced reliance a new philosophical discovery. The harder questions are when reliance gives someone a claim to continued action, when it warrants a warning instead, and when other agents or institutions should bear the risk.

Reasonableness cannot be treated as an objective fact independent of social circumstances. A beneficiary with limited access to alternatives may reasonably rely upon an informal arrangement that a wealthier person would treat as easily replaceable. The provider may know that a fragile arrangement has become the only available means of maintaining a basic good. Yet the resulting moral pressure cannot automatically become unlimited liability. If dependence itself created all the relevant authority, a person could acquire control over another's future simply by becoming sufficiently reliant. We need an account that registers the seriousness of vulnerability while refusing to reward strategically cultivated dependency.

The strongest formulation distinguishes the reason to protect someone against the harm of an expectation one has unjustifiably induced from the supposed duty to preserve every expectation one can foresee. Sometimes reasonable notice is the right response. Sometimes an express commitment requires actual performance. Sometimes the circumstances justify withdrawal while other persons or institutions must address the unmet need. These are different conclusions requiring different grounds. Our project will fail if it reaches them by measuring the magnitude of harm alone.

3.5. Joint commitment: when agency is genuinely shared

A relationship may produce obligations that cannot be reconstructed as a series of isolated assurances. Two people can undertake a project together, understand themselves as jointly engaged, and acquire distinctive standing to expect one another's cooperation. The significance of this phenomenon is especially clear in ordinary examples: two people walking together do not normally regard it as permissible for one simply to disappear without a word, even where nobody uttered a formal promise to remain. Margaret Gilbert's work analyzes such cases in terms of joint commitment, arguing that shared activity generates directed obligations and rights belonging to its parties (Gilbert 2006, 2013).

The appeal of Gilbert's position lies in its ability to capture a moral feature neglected by purely individualistic accounts: the parties have done something together, and the shared activity can alter what each is entitled to expect. It is insufficient to describe one person's conduct simply as useful to the other. The participants may have jointly constituted a pattern of action, and the norms of continuation and release depend on that shared undertaking. The view does not deny the possibility of ending the activity; it makes the terms of unilateral ending a substantive question.

This creates a serious rival to a dissertation that too quickly privileges the right to withdraw. Suppose Elise and Rosa have explicitly adopted a joint plan for six months of post-hospital recovery. Elise's contribution is not simply one charitable act among many but part of a mutually understood project. If she withdraws without discussion in week three, she may be violating a joint commitment even if she never described her role as an ordinary promise. The theory of joint commitment explains why the relational form of the undertaking matters.

Yet not every practice that becomes stable is jointly authorized. Rosa and Elise might have coordinated repeatedly without agreeing on who possessed authority to determine the arrangement's duration. Elise's presence in the household may be interpreted by Rosa as a shared long-term project, while Elise still regards it as a sequence of freely offered favors. The important question is whether the constitutive conditions of joint commitment are satisfied, not whether an external observer can discern a functioning routine. A joint commitment cannot be inferred from social convenience alone.

The problem becomes sharper under unequal power. A person can feel unable to resist an ostensible joint project because departure would bring punishment or exclusion. The existence of shared words or collective action does not by itself settle the legitimacy of the ensuing obligation. The theory must distinguish genuine joint undertaking from coerced coordination and ensure that explaining the normative force of shared agency does not launder another person's dependence into authority over the less powerful participant. Gilbert's position demands serious examination; it cannot be dismissed as another version of contractual voluntarism nor adopted without the qualifications required by cases of coercion.

3.6. Vulnerability and the Goodin challenge

Suppose we accept that Rosa's need does not automatically place a permanent obligation on Elise. A further question still presses: why would the absence of a promise be decisive where Rosa has become especially vulnerable to Elise's actions? A person responsible for a dependent's daily medication may possess knowledge that no one else can immediately reproduce. A caregiver who abruptly disappears might expose another to avoidable danger even without intending the harmful outcome. These cases are difficult because their moral pressure originates partly in the relationship's effects rather than in an agreement. The vulnerability is not a subjective demand invented by the claimant; it is an alteration in the ways another person's interests can be harmed.

Robert Goodin's Protecting the Vulnerable is an indispensable predecessor. He argues that special responsibilities must be understood through the vulnerabilities people have to one another's actions and choices, not only through explicit undertakings. His analysis connects promises, families, friendships, professional relations, and public responsibilities with a common concern for persons whose interests depend upon what others do or omit (Goodin 1985). This significantly weakens any theory in which special duties are exhausted by consent. It also supplies a plausible explanation for why a person can owe care to a stranger encountered in particular circumstances, even when no ongoing relationship or promise has been established.

Yet vulnerability is morally extensive. A worker may be vulnerable to an employer's staffing decision; a child to a parent's choice; an immigrant to an administrative rule; a patient to a clinician; and a poor household to the operation of public benefits. If every vulnerability generated an unrestricted duty on the actor whose choices matter, the distribution of obligations would be unmanageable and often unjust. Goodin's challenge is therefore twofold: explain why vulnerability creates particular responsibilities in some circumstances, and explain how those responsibilities should be allocated among persons and institutions. Treating vulnerability as sufficient without specifying the relevant relation threatens to make the agent with the greatest practical influence the default guarantor of another's interests.

The degree of vulnerability and the degree of control also vary independently. Elise may have great influence over Rosa's immediate safety because Rosa has few alternatives, but Elise's own health and economic stability may be fragile. A public agency may control the availability of paid home care yet never appear in Rosa's daily experience. A family member may possess ample resources and relational obligations but have delegated the work to Elise without ever formally discussing its duration. The visible caregiver is therefore not necessarily the sole or principal actor whose morally relevant choices explain the dependency.

This suggests a distinction between being a source of exposure and being the rightful bearer of a protective duty. The former is often a matter of causal and practical relation; the latter requires normative assessment. The distinction is not new; it is one of the difficulties Goodin himself addresses in allocating responsibility. But it will be essential when the dissertation examines claims that a person must continue providing a benefit simply because discontinuation exposes another to substantial risk. The strength of that exposure may deepen moral concern without determining the specific content or permissible duration of the duty.

3.7. Care duties, the concept of care, and the possibility of bad care

A difficult challenge to promise-centered theories comes from duties of care that nobody chose. In “Why Do We Owe Duties to Care?” (2026), Chadha-Sridhar defends a hybrid account: a general natural duty grounds the obligation, while particular relationships determine when it becomes owed to someone specific. This differs both from the claim that every duty must be volunteered for and from the claim that any socially assigned role is automatically legitimate. Elise may have a duty to help Rosa in an emergency even without a promise. Yet a family or institution cannot make Elise permanently responsible simply by assigning her the work. The distinction between the duty's moral ground, the relationship that activates it, and the justice of assigning it is already developed in this scholarship.

Chadha-Sridhar's “The Legal Concept of Care” (2026) complicates the picture further. Tort, medical, and family law regulate different aspects of care, but the paper argues that their uses are instances of a shared concept rather than unrelated meanings. It distinguishes caring actions, attitudes, and dispositions. Tort law emphasizes dispositions and actions; medical law emphasizes acts of care; family law may involve both actions and attitudes (Chadha-Sridhar 2026, “The Legal Concept of Care,” Introduction and abstract). The point is not that the paper provides a theory of moral succession. Rather, continuing to perform the same caregiving tasks need not mean that the underlying attitudes, roles, or obligations continue unchanged. Each requires separate examination.

The distinction matters to the claimant's position. A family member may legitimately receive a particular standard of physical care without possessing authority to compel the inward affection of its provider. A professional may be obliged to undertake careful acts while no obligation exists to sustain the intimacy associated with friendship. A former romantic partner may be entitled to truthful communication while lacking a right to the personal attitude that once accompanied it. These judgments cannot be derived solely from the grammatical classification of the relevant duty as one of care; they require interpretation of its object and authority. The concept of care is useful partly because it has several distinguishable aspects, not because all instances fall under one uniform standard of performance.

In “Bad Care,” Chadha-Sridhar offers a further warning: care can be morally objectionable because of how its caring aim is pursued, not only because it is badly executed or underfunded. Paternalistic care may seek to meet someone's needs while disregarding their legitimate choices. The intention to help can then be implicated in the wrong itself. The paper does not claim that care is usually harmful. It argues that being caring gives an act a default, rather than guaranteed, positive moral status (Chadha-Sridhar 2026, “Bad Care,” §§1–3). This differs from saying that good care might take place within an unjust labor arrangement: even the manner of trying to meet someone's needs can be morally defective.

Suppose Elise finds a replacement who reliably supplies Rosa's meals and medication but disregards Rosa's informed preferences, despite her capacity to decide. The tasks have been completed, yet the care may still be inappropriate. The failure is not just discourtesy: treating Rosa's need as permission to override her choices can compromise the care itself. But the opposite rule—that a recipient may veto every intervention—is also too broad. Some duties of emergency rescue and safeguarding operate under legitimate limits on choice. We must therefore assess the quality of the care, the duty to provide it, and the authority to determine its form without treating those questions as interchangeable.

The care literature challenges two shortcuts. We cannot assume that an arrangement should continue simply because it supplies care: the provider may have been assigned an unjust burden, and the recipient may be subjected to paternalism. Nor can we assume that a replacement is adequate merely because the same tasks are performed. What the duty protects, how care is delivered, whose choices count, and whether the burden is fairly allocated can come apart. An argument that the original duty has been discharged must address those dimensions. Chadha-Sridhar and other care ethicists already supply important resources for doing so; the insight should not be presented as a new discovery about succession.

Consider two arrangements that last equally long. In the first, a caregiver accepts a legitimate time-limited role, respects the recipient's informed preferences, and provides competent help until the agreed end. The need continues, but the original role may properly end while other duties remain. In the second, the caregiver supplies the same tasks but uses the recipient's vulnerability to control unrelated choices and silence objections. Greater dependency does not make the second arrangement morally superior or grant the caregiver wider authority. The fact that material needs were met cannot by itself establish that the relationship was legitimate.

None of this means that every preference determines what care must be given. Emergencies, nonvoluntary duties, and scarce alternatives can constrain choice. The point is to distinguish a duty to care, an activity called caring, and the moral quality of the care delivered. Genuine duties can arise within unjust systems that assign their burdens unfairly. Assistance offered with benevolent intentions can still wrong a recipient when it disregards legitimate choices. Existing care ethics already separates these considerations. To add anything philosophically distinctive, this dissertation must identify a further consequence for how duties change when the relationship changes.

3.8. Kittay: care for the one who cares

A theory of responsibility can make a subtle error by focusing on the person who needs direct assistance while treating the caregiver as an unlimited means of meeting that need. Eva Feder Kittay's Love's Labor offers a foundational corrective. Her analysis of dependency emphasizes that care work itself creates or intensifies the caregiver's vulnerability to social and economic disadvantage. Someone who gives sustained care can become unable to maintain employment, obtain rest, build savings, or preserve the conditions of their own flourishing. Kittay's account of doulia proposes that a just society must support the person who sustains the dependent rather than demand that the caregiver absorb the full burden alone (Kittay 1999).

The significance of this argument extends beyond the intuitive claim that caregivers should be thanked. A person whose work makes the life of another possible can become socially dependent precisely because the work is treated as a private obligation. The care recipient and the carer need not occupy symmetrical positions of dependence, but the relation is not ethically exhausted by one person's need and the other's capacity. The caregiver is also a person whose conditions of life must be secured, and the larger community may owe forms of support that render care sustainable. The idea of dependency work therefore questions the sharp division between autonomous agents who give and dependent beneficiaries who receive.

In our hypothetical, Elise's repeated assistance may have obscured the absence of formal services, reduced Rosa's family's perception of the burden, and simultaneously narrowed Elise's options. She may have declined travel, lost opportunities for paid work, or absorbed emotional strain. Those facts do not eliminate Rosa's needs. They reveal that a moral evaluation focused exclusively on whether the need is met can ignore the conditions under which someone is being made to meet it. A just arrangement cannot be determined solely by measuring beneficiary outcomes.

Kittay therefore poses a formidable constraint on a theory of legitimate relational transition. If a public institution's reliance on unpaid care has increased the provider's vulnerability, then requiring the same provider to ensure indefinite continuity could entrench the injustice. But permitting a caregiver to withdraw without any provision for the recipient would fail to address the interdependence Kittay places at the center of equality. The answer must be sought in the structure of support and reciprocal responsibility, not in the moral heroism or moral innocence of an isolated individual.

A common misreading of care ethics depicts it as endorsing self-sacrifice whenever anyone needs something. Kittay's work demonstrates why that reading is wrong: the dependency worker's needs are themselves subjects of justice. This insight also bears on disability. A philosophy that presents a dependent person's life as an excessive burden risks diminishing the person it purports to protect. The injustice lies in the institutional refusal to distribute support and protect all participants, not in the dependent person's existence or need for assistance. Any argument about relational withdrawal must make that distinction explicit.

3.9. Tronto and the politics of who takes responsibility

Joan Tronto distinguishes four elements of care: noticing a need, taking responsibility for responding, performing the work, and attending to the recipient's response. Her corresponding ethical concerns include attentiveness, responsibility, competence, and responsiveness (Tronto 1993). These stages need not belong to one person. Someone may recognize a problem, another may organize assistance, and someone else may provide it. The distinctions matter when an arrangement changes because the departure of the worker does not automatically settle who remains responsible for planning or responding.

This point is especially important for our dissertation because its apparent conceptual discovery may already be contained in Tronto's account. The fact that Elise notices Rosa's needs is not equivalent to Elise accepting permanent responsibility for organizing care. The fact that she performs the work does not necessarily establish that she alone was authorized or required to take charge of long-term planning. Tronto's distinctions separate recognition, responsibility, performance, and reception long before our proposed language of normative discontinuity. The dissertation must acknowledge that precedent rather than market it under new terminology.

In Caring Democracy, Tronto places care within democratic politics rather than treating it as a private domestic concern. When society depends on largely invisible caregivers without providing adequate public support, the distribution of care becomes a matter of justice and institutional design (Tronto 2013). A fair transition therefore cannot be assessed only by asking whether an individual has been generous enough. We must also ask whether the institutions allocating need, labor, and decision-making have met their responsibilities. A departing caregiver should not automatically inherit the entire burden of repairing a support system that was never adequately sustained.

It would nevertheless be a mistake to claim that society's responsibility absolves every individual. Public provision can be delayed, inadequate, or inaccessible. The person who encounters a vulnerable stranger may have an immediate duty to prevent severe harm where assistance is reasonably available. A family member may have undertaken a particular duty the state does not displace. A caregiver who possesses vital information may owe communication where withholding it would create avoidable danger. Tronto's political analysis enlarges the field of responsibility; it does not supply a general exemption from interpersonal obligation. The challenge is to specify how the layers operate together.

Who provides care is only part of the question. Someone must identify the need, decide how to respond, control resources, and assess whether the response is adequate. Those roles may be divided among the recipient, relatives, professionals, and public institutions. When one caregiver leaves, the labor may shift while decision-making authority stays in place. Planning can change hands without direct work changing hands. And the recipient's legitimate judgment remains relevant even when a worker departs. Care ethics has already shown why these distinctions matter. The present inquiry must build upon that work, not present it as a new discovery.

3.10. The difference between dependence and its legitimate allocation

Dependency can arise from the ordinary human condition or from another person's discretionary choices. A child depends on adults because of developmental vulnerability. An adult may become dependent through illness, injury, or age. A worker may become dependent on a firm because specialized training and reputational recognition are difficult to carry elsewhere. A student may become dependent on one mentor because the institution has not provided alternative access to resources. These situations differ in the sources of vulnerability, the agents who control its conditions, and the reasons the dependency is or is not an acceptable feature of the relationship.

A morally adequate theory must distinguish dependence as a relation from dependence as an assignment of duty. Suppose a disabled adult needs assistance to live in the community. That need is real regardless of whether a daughter lives nearby, whether a spouse is capable of providing care, or whether the state provides adequate support. A relative's presence may make direct aid feasible and may interact with special duties, but it does not alone prove that public responsibilities have been transferred to the relative. The dependency relation describes facts about a person and available conditions of assistance; the assignment of a duty is a normative judgment about what another agent owes.

The danger is especially acute where a supposed role is an effect of unequal institutional design. In some systems, public agencies tacitly assume that women will provide informal care and then classify their labor as a natural family resource. When support is withdrawn because such a caregiver is available, institutional allocation helps manufacture the very private dependency later invoked as justification for further private duty. Ong and Savulescu examine the wider social ecology of care and the danger of treating involuntarily assumed responsibilities as limitless; their framework includes duties on collectives with the capacity and resources to reduce unreasonable burdens (Ong and Savulescu 2026).

The resulting pattern cannot be addressed by declaring that dependency is bad. Dependence can be a valuable consequence of trust, education, intimacy, coordinated specialization, and shared life. Parents and children depend on one another; collaborators develop joint skills; friends rely on companionship; patients reasonably trust skilled professionals. The injustice emerges when the conditions of dependence are arranged, maintained, or interpreted so that one person's vulnerability becomes an unanswerable claim against another's agency, or the caregiver's labor becomes a pretext for the withdrawal of otherwise justified institutional support.

There is an important danger in treating all departures as acts of liberation. A powerful actor can sometimes withdraw from a relationship that they themselves structured to make the other party precarious. A firm may acquire essential knowledge from contractors and then terminate the arrangement without securing access to what was promised; a parent may exploit a child's dependence and later invoke personal freedom to avoid responsibility. In these cases, the history of the relationship may supply robust grounds for continuing duties. The correct conclusion is not that past dependency has no moral force. It is that the grounds of particular duties must be reconstructed rather than presumed from the practical survival of need.

What changes the normative result is not the bare existence of dependence but the interplay of morally relevant facts: whether an obligation was genuinely undertaken, whether another's vulnerability was knowingly induced, whether the relation carries an independently grounded role duty, whether the assignment rests upon a just social practice, whether withdrawal would breach a protected interest, and whether other agents or institutions hold primary responsibilities. No single consideration can be treated as a universal master key. The task of a constructive theory will be to show how these facts can guide judgments without becoming an indeterminate instruction to “consider all circumstances.”

3.11. The difficult case of reasonable but unauthorized expectation

We can now revisit Elise's assistance without assuming that any single framework settles it. Rosa's continuing need supplies reasons for care; Elise's conduct may have cultivated reliance; informal practices may have assigned a caregiving role; and wider institutions may have allowed other parties to retreat from responsibility. The strength and kind of Elise's obligation depend on which of these factors has genuinely normative force. But an additional complication remains: a recipient may form an expectation that is both psychologically understandable and normatively unauthorized.

Imagine Rosa correctly understands that Elise will probably help tomorrow, because Elise nearly always does. Rosa makes a rational decision not to hire an expensive replacement that day. Elise knows about the decision but has never represented herself as permanently available. Is the resulting reliance reasonable? In one sense, yes: Elise's history is strong evidence of what she is likely to do. In another sense, Rosa's expectation may extend beyond anything Elise has undertaken or may fairly be asked to continue. Epistemic rationality about future behavior and moral entitlement to that behavior are not the same. One can predict correctly that someone will be generous without acquiring a right to generosity.

Now suppose Elise repeatedly assured Rosa that she would remain available, discouraged her from arranging alternatives, and knowingly allowed her to incur substantial costs. The moral history has changed, even if Elise never uttered a formal promise. Intentional assurance and induced reliance may create obligations. Yet the duty need not be indefinite personal care. Its content might instead involve reasonable notice, reimbursement of avoidable costs, cooperation in an authorized handoff, or fulfillment of a limited commitment. What matters is how Elise generated the expectation and what response that history justifies—not dependency considered in isolation.

A third variation introduces a manipulative beneficiary. Rosa deliberately rejects affordable alternative services, insists that she will be helpless without Elise, and threatens to denounce Elise as cruel if she withdraws. Rosa's need remains real. But the means through which an expanded demand is asserted may be illegitimate. The moral authority of an apparent claim is not established by maximizing the beneficiary's dependence. A view that rewarded strategic escalation of vulnerability would undermine autonomy and create perverse incentives for both parties. Nevertheless, the presence of manipulation does not automatically remove general duties to prevent an imminent emergency. The right response may be to distinguish safe, proportionate assistance from the demanded form of continuing personal access.

These cases show why a theory of the grounds of obligation must separate reliability as evidence, reliance as a state of affairs, reliance that another person knowingly induces, and reliance protected by a legitimate undertaking. There may be overlapping duties grounded in several of these considerations, but they cannot be treated as interchangeable. The objection is familiar to Scanlonian approaches and theories of promise; it is not an original discovery. Our prospective contribution must come from the ability to specify difficult transitions without erasing either relational history or the limits of morally legitimate demand.

3.12. Four matched cases: separating the grounds

A useful method is to hold the severity of the beneficiary's need constant while altering one fact about the relationship. Suppose Rosa's practical condition, available resources, and risk of interrupted care remain identical across four cases. The question is whether a directed duty belongs to Elise, and if so on what basis.

In Case A, Elise makes a six-month promise of direct care, Rosa accepts it, and the burden falls within what both reasonably understood. The intentional undertaking supplies a strong ground. Breaking the commitment solely because the work has become inconvenient will generally require a further justification. The relevant challenge concerns terms, permissible modification, and competing reasons. Promissory theory does most of the explanatory work.

In Case B, Elise never promises continued assistance but knowingly creates a materially misleading impression that Rosa need not secure another caregiver; Rosa acts upon that impression. A duty concerning warning, correction, or mitigation may arise without the same duty of exact performance found in Case A. The difference must be defended through principles governing assurance, due care, and induced reliance. It cannot be inferred from equal severity of need.

In Case C, Elise provides occasional help while expressly stating that the arrangement is temporary, and Rosa independently chooses to rely on its continuation. The moral argument for exact performance is weaker. A duty to respond to an emergency might still exist if Elise is uniquely placed to prevent serious harm at modest cost. But that would arise from a different ground, not from an automatically renewed promise that no one made.

In Case D, Elise is legally appointed or otherwise legitimately charged with an ongoing caregiving role, and the relevant duty is not exhausted by the presence or absence of a voluntary promise. The scope of her responsibility depends upon the role's justification, applicable standards, and permissible conditions of replacement or withdrawal. Role-based and natural-duty accounts now become central.

The cases refute a simple severity rule: the same degree of unmet need does not establish identical directed obligations against the same person. They also refute a pure voluntarism rule: Case D can generate duties without ordinary contractual consent. But these results are already available from established work on promising, vulnerability, special obligations, and care. The cases are necessary to clarify the argument; they are not yet evidence that the dissertation has made a novel contribution.

Hold the words of an undertaking constant and change the social conditions. In one society, a caregiver freely agrees to one year of help with access to paid leave, professional services, and genuine alternatives. In another, the same agreement is made under threats of destitution, exclusion, or severe gendered punishment. Treating the words alone as decisive would miss the difference in voluntariness. Chadha-Sridhar's theory likewise denies that gravely unjust systems can create care duties solely by assigning them to particular people. Any separate emergency or independently grounded obligation must still be considered; rejecting the legitimacy of an imposed role does not make the recipient's need disappear.

3.13. A minimal model of normative grounding

A limited notation helps identify the relevant facts without treating them as a formula for deciding every case. Let N(B,t) denote B's important need at time t, and V(A,B,t) denote A's influence over B's vulnerability. Let U(A,B,φ,t₀) denote A's sufficiently clear voluntary undertaking to perform φ for B, A_s(A,B,φ,t₀) an intentional assurance, and R(B,A,φ,t₀) B's reliance. Let J(A,B,t) denote a legitimate joint commitment, K(A,B,t) an independently justified care relationship or role, and Q(A,B,t) a morally legitimate social assignment of care. Finally, D(A,B,φ,t) denotes A's directed duty to perform φ for B. Naming these variables does not establish that any single one always produces a duty.

The formal possibilities reveal why no single fact is enough. The proposal N ∧ V ⇒ D—that need together with vulnerability or influence produces a duty—ignores cost, available alternatives, and the principles governing the relationship. The simpler U ⇒ D captures ordinary undertakings but misses duties we never volunteered for, as well as promises that were invalid or conditional. The rule R ⇒ D treats reliance as sufficient without asking who invited it or whether the expectation was justified. And K ⇒ D risks turning an assigned role into a duty even when the assignment is unjust or its demands exceed the role's legitimate scope. These symbols mark candidate implications, not established moral laws.

The notation identifies questions, not automatic implications. A promise-based explanation must show that U was valid and specify its terms. An account based on reliance must explain why the reliance was protected; vulnerability alone requires a principle that links exposure to a particular person's duty. Shared activity is not sufficient evidence of J, a legitimate joint commitment. And a care-based account must distinguish the reason the duty exists from the relationship that activates it. Chadha-Sridhar's natural-duty approach illustrates that last difference while resisting the idea that an oppressive role assignment automatically creates a moral obligation (Chadha-Sridhar 2026).

Moral judgment cannot be reduced to a score calculated from promises, need, vulnerability, and benefit. Duties may overlap, leave room for choice, or depend on resources an institution has not supplied. Declaring that “every factor matters” explains too little; declaring one factor always decisive would be equally unsound. A stronger approach identifies specific invalid inferences, asks who may validly change a duty, and tests whether its judgments differ from those available under the strongest existing theories.

Time adds a further difficulty. A promise made at t₀ may create an obligation that continues at t₁, even while the promisor's capacity or circumstances change. A valid waiver, an altered role, or a new institutional arrangement may also change what is owed. Sometimes the apparent continuation of “the same duty” actually conceals several distinct duties. A volunteer's promise, an agency's public responsibility, and a professional's new undertaking may jointly maintain care without transferring one obligation between them. Shiffrin and theories of normative powers already anticipate important parts of this analysis; any claim of novelty must go beyond them.

3.14. Three objections that cannot be evaded

One objection is that this approach could make promises unreliable. If every change of circumstances prompts a fresh debate about what is owed, nobody can plan confidently around another person's word. The objection is powerful where the promise was clear, voluntary, and essential to reasonable reliance. A valid undertaking ordinarily remains binding even when the person who made it wishes otherwise. Examining a duty's grounds is not permission to renegotiate every inconvenient commitment. It establishes what the commitment actually required and the limits it legitimately contained.

The second objection is that a focus on the illegitimacy of imposed roles may obscure the beneficiary's reality. A disabled person requiring daily care does not become less dependent because the existing arrangement is unjust. If a daughter withdraws from exploitative caregiving expectations, the need for safe provision remains. The moral solution cannot consist in telling the dependent person that nobody owes anything. A theory must recognize the duty's possible relocation to institutions or other actors while distinguishing what can be required of the individual who is refusing injustice. Kittay, Tronto, and Ong and Savulescu supply substantial resources for this analysis. Any purported new theory that ignores them should be rejected.

The third objection is that the thesis is redundant. An informed critic could assemble Scanlon's assurance and due-care principles, Goodin's vulnerability analysis, Gilbert's joint commitment, Chadha-Sridhar's hybrid non-voluntarism, and Tronto's care framework to explain every case constructed so far. That critic would have a substantial argument. Repetition of the same verdict in a different vocabulary is not philosophical novelty. The dissertation must either identify a case where the combined accounts fail to specify the legitimacy of a transition, expose an inconsistency among their competing explanations, or restrict itself to a genuinely new institutional or practical problem. Intellectual discipline requires the possibility that no broad foundational innovation survives.

An additional methodological objection concerns hindsight. Once a relationship has failed, it is tempting to narrate its history as if one person always intended the ending and another always intended the dependency. That approach retrospectively converts unstructured assistance into deliberate undertaking or interprets ordinary care as strategic capture. A careful philosophical analysis must describe what participants knew and could reasonably foresee at the relevant times. Later consequences may supply evidence about earlier conduct, but they should not be allowed to rewrite its normative meaning without argument.

3.15. The explanatory burden and its relation to prior work

The author's previous scholarship already develops several of the problems this chapter has raised. Reflexive Capture examines how formative relationships can intensify the dependencies through which their own power operates. When the Gift Becomes Capacity addresses the conversion of informal assistance into organizational planning dependence before a continuing productive role is authorized. After the Gift defends the receiver's standing and the finitude of the giver, and Differential Modal Foreclosure explains how formally available alternatives can lose the conditions that make them credible. These works provide substantive intellectual ancestry, but they do not establish this dissertation's originality simply by being the author's own.

The danger is an argument that migrates from one level of explanation to another without noticing. A causal mechanism may show how an organization developed dependence on a volunteer, but not whether the volunteer owes continued work. A philosophical account may establish that the volunteer's future cannot be owned, but not whether a limited promise or a duty of warning remains. An institutional critique may establish that public services are inadequate, but not whether a particular local caregiver has an emergency obligation. Moving properly among these levels requires distinguishing facts, grounds, specifications, and permissions.

The comparisons in this chapter undermine two simple rules. Repeated assistance or serious vulnerability cannot, by itself, make one historical provider responsible indefinitely. Yet the absence of an explicit promise does not rule out duties arising from care, emergency, role, or justifiably encouraged reliance. The stronger conclusion is constructive: we must identify the particular facts that make a duty binding and distinguish those facts from the social conditions that may have assigned its burden unfairly. Existing scholarship supplies much of this explanation.

This distinction now creates a new question. Once an obligation has a legitimate ground, what happens when the relationship changes? A promise may survive withdrawal, a role may end under defined conditions, and another person's duty may arise without receiving anything from the first provider. Chapter IV examines endings; Chapter V examines substitution and release. Chapters VI and VII then test whether those distinctions generate a judgment the strongest available theories cannot already explain.

3.16. Conclusion: no one fact makes a relationship a debt

Elise and Rosa's history may have created obligations that neither imagined at the start. Some may arise from assurances, others from vulnerability or shared activity, and still others from institutions that should have provided dependable support. Those sources of responsibility are not interchangeable. Need does not guarantee unlimited access to the former helper; a promise may remain binding after enthusiasm fades; a role may create duties without consent while an unjust allocation may lack authority. Several legitimate duties may also coexist, owed to different people and carried by different agents. The moral task is to identify each, not force them into a single story about dependence.

The differences become especially visible across kinds of relationship. A child's claim to care, a friend's expectation of loyalty, a patient's entitlement to professional diligence, and a stranger's claim to emergency rescue may all demand serious attention. But they arise under different principles and permit different forms of fulfillment. The public agency that depends on an unpaid caregiver introduces still another question: whether the institution has improperly shifted its own burden onto a private person. These cases cannot be settled by ranking need alone.

Before deciding that one person's relationship with another makes them responsible for that person's future, we must identify the source, content, and limits of the alleged duty. We must also ask who had authority to create or assign it. Rosa's need for care cannot be dismissed because no private person has agreed to guarantee it. But Elise's earlier generosity cannot make her that guarantor simply because everyone else has organized their plans around her availability. Both truths must remain visible.

Rosa's dependence, then, is neither an illusion nor a blank check drawn on Elise's future. What Elise owes depends partly on what she undertook and encouraged, partly on duties that need no promise, and partly on the responsibility others have failed to assume. Chapter IV carries that analysis into the moment when a relationship ends. A legitimate departure may leave promises, losses, and duties of care behind—but which ones, and why?

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.