TL;DR: A promise gives another person a reason to organize their choices around your future action. Changed circumstances can justify asking to revise it, but your request is not itself permission to withdraw. Consider what was undertaken, who relied on it, whether genuine release is possible, and what obligations remain if performance becomes impossible or wrong. The argument below treats renegotiation as a moral conversation about another person's options, not a technique for making inconvenient commitments disappear.
Begin with a promise that was sincere
Leila promises to translate a short catalog for a community exhibition by Friday. She is volunteering, knows the expected length, and believes she can complete it. The organizer, Tomas, stops looking for another translator and schedules the printing. On Tuesday, Leila receives an attractive paid assignment that would occupy the time she had reserved. She would now prefer to withdraw.
The situation is deliberately ordinary. There is no secret intention to deceive when the promise is made, no contract clause to interpret, and no claim that a court should enforce the arrangement. The case and its later variations are original thought experiments about moral responsibility. They let us isolate a question that everyday language often blurs: does a change in the promiser's preference change what the promisee is entitled to expect?
If Leila had merely said that she probably had time, Tomas might have needed to keep looking. Her promise did something stronger. It invited him to treat the translation as settled enough to make further decisions. That invitation cannot be reversed solely by revising her private estimate of what would now be best for her. Something about the relationship has changed, even if no money moved between them.
This does not settle the case by declaring every promise inviolable under every circumstance. It establishes who must enter the reasoning. Leila's new opportunity matters, but Tomas's altered options matter too. Asking whether she would still choose to promise today is therefore the wrong test. The relevant fact is that she already made a commitment on which someone else was invited to rely.
Two historical arguments help locate the issue without deciding all its modern variations. In Book III, Part II, Section V of A Treatise of Human Nature, Hume examines promising as a human convention that enables cooperation across time, rather than a mysterious act of will that creates obligation by itself. In Groundwork of the Metaphysics of Morals, Kant uses an intentionally false promise to examine whether its maxim could be universalized. His example concerns deceit at the outset; Leila's original sincerity makes her later problem different.
The argument developed here is narrower than a reconstruction of either philosopher's complete account. It asks how a sincere commitment can be reconsidered while still taking seriously the person to whom it was made. Neither historical text should be turned into a ready-made rule that every contemporary scheduling conflict has one obvious answer.
A promise changes the other person's position, not just their prediction
A prediction gives someone information about what may happen. A promise adds an undertaking: the promiser takes responsibility for a future action in a way the other person can address. “I expect to finish Friday” and “I will deliver it Friday; you can plan on that” may describe the same forecast, yet they invite different kinds of reliance and different responses if the forecast changes.
The words alone are not always decisive. A casual “sure” can accept a serious request when both people understand the context. A ceremonial “I promise” can be part of a joke everyone recognizes. The practical question is whether one person reasonably understood the other to undertake a commitment, and whether that understanding was mutually available. This is an interpretive problem, not a search for a magic phrase.
For Leila and Tomas, the context is clear. Tomas asked whether the translation was covered so he could arrange printing. Leila accepted the task and date. She did not merely report her current optimism. She participated in closing an option that Tomas would otherwise have needed to keep open. That closure is part of what must be considered when she asks to reverse the decision.
Reliance matters, but it is not the whole source of obligation
Suppose Tomas has not yet booked the printer. Leila might argue that no harm has occurred, so the promise can be withdrawn without asking. That conclusion is too quick. A person may have a claim to an undertaking before they visibly reorganize their life around it. Otherwise promises would become binding only after the promisee rushed to incur a cost.
Reliance changes the urgency and difficulty of release, but it should not be treated as a contest in which the first person to spend money wins moral authority. Tomas can reasonably expect Leila to ask even if he has not yet acted. If releasing her is easy, he may have a strong reason to do so. The ease of release does not mean she has already obtained it.
The reverse case is also revealing. A person can reasonably rely on behavior that was never promised. Tomas may assume that a longtime volunteer will help again, even though Leila has not agreed this year. His disappointment could deserve consideration without creating the same obligation as an accepted task. Harm, expectation, and promise are related concepts; none should automatically replace the others.
Keeping these distinctions prevents two mistakes. The promiser cannot erase an undertaking by saying the other person should never rely on anyone. The promisee cannot create an undertaking merely by announcing that they have relied heavily. The relationship needs an account of what was actually communicated, as well as what each person subsequently did.
The scope of a promise includes its ordinary conditions
Leila promised a short catalog translation, not an unlimited amount of exhibition work. If Tomas adds several new essays, requests daily revisions, and changes the audience, he may be proposing a different task. Leila can accept, reject, or renegotiate that expansion without thereby repudiating the original promise. Faithfulness does not require treating every later request as if it had been included from the beginning.
There will be borderline cases. Correcting an error may belong to completing a translation; repeatedly rewriting a newly expanded catalog may not. The moral conversation should identify the disputed change instead of accusing one person of inflexibility and the other of unreliability. A promise has enough stability to support cooperation only if its content can be discussed without being either frozen into absurdity or expanded without limit.
A further boundary concerns what Leila was entitled to undertake. Suppose she promised that another volunteer would finish the illustrations, despite never obtaining that person's agreement. Her confident statement cannot by itself impose a duty on the illustrator. It may instead give Leila a responsibility to correct the assurance she made without authority. The recipient's reasonable expectation and the third person's freedom both matter.
This is why a promise about someone else's action differs from a promise to seek their help. “I will ask the illustrator today” concerns an action Leila controls. “The illustrations will be ready Friday” may imply an assurance over another person's labor. If the stronger assurance was given carelessly, the right response is not to pressure the illustrator into making it true at any cost. It is to acknowledge the gap between what was promised and what Leila could legitimately arrange.
The same distinction limits the scope of renegotiation. Tomas and Leila can revise their own arrangement, but they cannot agree that an uninvolved volunteer must absorb the consequences. A mutually convenient solution for the two central parties can still be unfair to someone outside their conversation. Resolving a promise therefore requires attention to who bears the revised burden, as well as who has agreed to the revision.
Change the circumstances and observe which reason changes
The paid assignment case is not the only kind of revision. Several variations show why “circumstances changed” cannot be a complete justification. It describes a difference between two moments, but says nothing yet about the moral relevance of that difference. A better offer, unexpected impossibility, and a newly discovered wrong do not all do the same work.
Variation one: a more attractive opportunity appears
In the original case, Leila still can perform. She would simply prefer the paid assignment. That preference supplies a reason to ask Tomas for release, especially if another competent translator is readily available. It does not supply an automatic right to abandon the catalog. If promises dissolved whenever a better option appeared, their practical function would shrink toward a forecast of continuing convenience.
This does not require Leila to pretend that the new opportunity is unimportant. She can explain its significance honestly. What she should avoid is converting desire into necessity: saying she “cannot” do the translation when the accurate statement is that she now wants to do something else. The difference affects Tomas's understanding of the request and his opportunity to respond.
Tomas may release her generously. He may propose a partial handover. He may explain that the printing schedule leaves no workable replacement. His response should be considered rather than anticipated as an obstacle to be managed away. Asking is meaningful only if the other person's answer can affect what happens.
Variation two: a genuine conflict makes performance unreasonable
Suppose Leila unexpectedly becomes responsible for an urgent family matter that occupies the available time. The moral reason for changing the commitment is now stronger than a more attractive assignment. The competing responsibility may justify not completing the translation even if Tomas cannot readily release her. That conclusion concerns the weight of conflicting obligations, not proof that the original promise never existed.
Notice what remains. Leila should communicate as early as she reasonably can, identify the work state, and avoid leaving Tomas to discover the failure at the deadline. If she can transfer a partial draft or explain an unresolved issue without neglecting the urgent responsibility, that may reduce the damage. If she cannot, the moral assessment should recognize the actual constraints rather than invent unlimited time for conscientious behavior.
An excusing circumstance can change blame without restoring lost options. Tomas may reasonably feel disappointed and still recognize that Leila acted as she should. Those positions are compatible. Moral evaluation should not force the promisee to deny the cost in order to show compassion, or force the promiser to claim that justified nonperformance caused no harm.
Variation three: the requested performance becomes wrongful
Now suppose the organizer reveals that the translation is to conceal material information from the exhibition's audience. Leila's promise should not be treated as a machine that converts wrongful conduct into a duty merely because she agreed before learning the facts. The nature of the requested act matters. A commitment can fail to justify performance when its fulfillment would wrong others.
This variation differs from private disappointment with the exhibition's artistic choices. Disagreeing with a preference is not the same as discovering deception. Leila owes a careful account of what changed and why she believes it alters the moral status of the task. Otherwise “my values changed” can become an unexamined phrase covering every inconvenient commitment.
Tomas's willingness to release her would not be the only relevant authority here. If performance would wrong third parties, his insistence cannot by itself make the action acceptable. This marks a limit to a simple permission model. The promisee has an important position in the relationship, but the relationship does not exhaust the moral world.
Variation four: the original agreement was misunderstood
Suppose Leila thought the catalog contained ten pages and Tomas thought she had accepted forty. If neither recognized the mismatch, there may be no single shared undertaking with the scope either person now asserts. The problem is partly interpretation and partly repairing a coordination failure. Repeating “a promise is a promise” adds little until the content of the alleged promise is established.
The response should consider what each person had reason to understand. If the page count was clearly stated and Leila overlooked it, her mistake does not automatically transfer all the cost to Tomas. If Tomas changed the document after she accepted, the new burden cannot simply be folded into her original consent. Responsibility follows the actual communication, not whichever interpretation is more convenient now.
Release is a decision the promisee must be able to make
A request to change a promise can itself be respectful. It recognizes that the other person's position matters enough to consult. The moral problem arises when asking becomes a performance after the promiser has already decided to withdraw. “Would you mind?” then functions as a demand for reassurance rather than a real question.
A more honest request separates the reason, the proposal, and the status of the existing commitment. Leila might explain the new assignment, ask whether a replacement or revised deadline would work, and make clear that she has not yet accepted incompatible work. Tomas can then assess an actual option rather than retroactively approve a fait accompli.
This approach is not always possible. In the urgent family variation, Leila may already know that the translation cannot be completed. Then she should state the situation directly instead of asking for permission she cannot meaningfully wait for. The conversation concerns notification, mitigation, and what remains owed. Calling it a request for release can obscure the actual decision structure.
Silence, politeness, and pressure are different from willing release
Suppose Tomas says, “Do whatever you need,” while clearly trying to avoid a public confrontation. Leila should not always treat those words as a complete settlement. The context matters. Did Tomas understand the options? Was there an opportunity to discuss the printing cost? Was Leila asking privately, or in front of people whose approval made refusal difficult?
This does not mean a promisee must pass an elaborate authenticity test before their answer counts. It means the promiser should not create pressure and then cite the resulting politeness as evidence of free agreement. A brief clarification can help: “I want to make sure I am not leaving you with an impossible deadline. Would the handover actually work?”
Power differences make the issue sharper. A manager can ask an employee to release a personal commitment while making refusal professionally costly. A dependent family member may feel unable to object. In such cases, the appearance of agreement may deserve less weight than the stronger party would like. The person seeking release has reason to make refusal practically possible, not merely verbally available.
A substitute can change the plan without being an equivalent performance
Leila finds another translator and assumes the problem is solved. Perhaps it is. But the replacement may need context, offer different quality, or require Tomas to spend time on coordination. If the original undertaking depended on Leila's particular knowledge, a substitute changes more than the name attached to the task. Tomas should be able to assess the proposed handover.
The same applies to partial delivery. A half-completed translation can be useful if its state and unresolved issues are clear. It can be costly if another person must first discover what is trustworthy. The moral value of a handover lies in the options it preserves for the recipient, not in how much effort the departing promiser can display.
Nor can Tomas release every claim belonging to everyone else. He may forgive Leila while a printer has already reserved work or another volunteer has rearranged their schedule. The details determine whether Leila owes those people anything directly, but their interests should not vanish because the two central parties reach a friendly understanding. Release can settle one relationship without settling the whole situation.
Three objections to taking promises this seriously
Does this let the past imprison the present?
If every past commitment dominates later judgment, promising would become a trap. People change, learn, and encounter responsibilities they could not foresee. An account of promises that denies those facts would make sincerity dangerously expensive. The objection is strongest when the original undertaking becomes much more burdensome than either party reasonably contemplated.
The reply is not to make promises weightless. It is to distinguish a reason for reconsideration from unilateral permission to disregard another person. A changed self still inhabits relationships shaped by earlier choices. That continuity does not require identical preferences forever. It requires addressing the effects of change on people who were invited to plan around a commitment.
A promise can therefore create a strong reason without creating an absolute reason. The strength depends partly on its content, the competing obligations, and the available alternatives. This leaves room for judgment, but judgment is unavoidable. The possibility of hard cases does not show that an ordinary better offer should routinely defeat an earlier undertaking.
Would explicit conditions make every promise safer?
It is tempting to solve the problem by attaching a long list of exceptions. Some conditions are useful. A person can promise a draft subject to receiving the source material by a stated date. That clarifies a dependency both parties can understand. But a promise qualified by “unless I later prefer something else” may provide little of the assurance the recipient sought.
The purpose of a condition should be to locate a real uncertainty, not to preserve all the promiser's freedom while borrowing the appearance of commitment. Otherwise the promisee receives the psychological benefit of reassuring words without the practical benefit of a dependable undertaking. Precision helps only when it makes the allocation of uncertainty more honest.
There is also a cost to excessive specification. Ordinary relationships cannot list every possible contingency. Trust includes the expectation that people will interpret unforeseen cases reasonably. The answer to uncertainty is a combination of clear central terms and a willingness to discuss genuine changes, rather than pretending that a complete exception catalogue can replace moral attention.
Why not pay compensation and move on?
Compensation can repair some losses. If Tomas must pay another translator, covering a reasonable additional cost may matter. But it does not follow that every promise contains an unstated option to buy withdrawal at a price chosen by the promiser. Some losses concern timing, trust, or opportunities that cannot readily be purchased again.
Even where money is useful, the question is whose options it restores. A payment that leaves Tomas with no translator before printing does not solve the relevant problem. Conversely, a timely replacement that genuinely meets the need may matter more than a dramatic apology. Repair should respond to the structure of the loss rather than the promiser's preferred gesture.
Treating compensation as sufficient by definition also privileges the person able to pay. They could routinely impose disruptions on others and regard the relationship as settled afterward. A more defensible view treats compensation as one possible remaining obligation, alongside explanation and practical assistance, rather than a universal permission to break commitments whenever the price seems affordable.
An apology should leave the other person's judgment intact
Suppose Leila ultimately fails to deliver. An apology should identify the undertaking, acknowledge what happened, and avoid demanding a favorable moral verdict in exchange for showing remorse. “I know you relied on the Friday delivery, and my late notice removed your chance to arrange another translator” recognizes a specific loss. “I hope you understand I am a good person” asks Tomas to manage Leila's self-image.
Explanation still matters. The urgent family case differs from quietly accepting the better offer. But an explanation should illuminate the decision rather than erase the recipient's experience. Tomas can understand why Leila acted and still need to discuss the consequences. He need not pretend that everything was easy in order for reconciliation to begin.
Repair may also require changing how future commitments are made. Leila might offer a narrower scope, avoid accepting overlapping work, or state an uncertain dependency earlier. These changes should respond to the failure that occurred. A broad vow never to disappoint anyone again is another promise whose content cannot reasonably be sustained.
The moral discipline is modest but demanding: do not treat your changed preference as the other person's consent. Ask when asking is possible, explain when it is not, and preserve what options you reasonably can. A promise is reconsidered well when the person who trusted it remains a participant in the decision rather than an inconvenience to be talked past.
Sources
- David Hume: A Treatise of Human Nature, Book III, Part II, Section V—Of the Obligation of Promises, published in the Treatise's 1739–1740 sequence.
- Immanuel Kant: Groundwork of the Metaphysics of Morals, 1785, Jonathan Bennett's modernized English version; the false-promise examples occur in Chapters 1 and 2. The contemporary cases and argument about renegotiation above are original analysis, not attributed positions of Hume or Kant.