The Sunk Cost Fallacy: Deciding What Is Worth Continuing
You have spent weeks reading a difficult book that no longer serves your purpose. Stopping feels wasteful because of the time already invested. Yet the remaining chapters would require still more time, and you cannot recover the earlier hours by continuing.
The sunk cost fallacy describes decisions in which an unrecoverable past investment receives inappropriate weight in choosing what to do next. The key word is unrecoverable. Past information can matter, but a cost that cannot be changed should not be treated as a benefit of further investment.
This distinction helps with ordinary choices about projects, learning routines, and commitments. It does not mean that perseverance is foolish or that quitting is always the correct response.
Separate the past cost from the future decision
Suppose you bought a ticket to an event and cannot obtain a refund. On the day, another activity would be more worthwhile. The ticket price is already gone under either choice.
The relevant comparison concerns the remaining options: attending, staying home, or doing something else. Travel time, enjoyment, obligations, and future consequences may matter. The nonrefundable payment does not become recoverable merely because you attend.
Arkes and Blumer's 1985 paper examined how prior investment can influence continued commitment. The research provides a basis for the concept, but it does not decide every real-world case without examining its actual costs and benefits.
A useful question is: what changes from this point forward if I choose one option rather than another?
Persistence can still be rational
A project may be close to completion, and the remaining work may offer substantial value. Continuing can be sensible even after a large investment. The reason is the expected value of what remains, not the mere existence of what was spent.
Past work may also create assets that are still useful: knowledge, equipment, relationships, or a partially completed product. These are not simply sunk costs if they affect future options.
For example, completing a course may be worthwhile because only one relevant module remains and the resulting skill is useful. That differs from continuing solely because abandoning the course would make the earlier effort feel embarrassing.
The distinction prevents a common overcorrection: labeling every commitment to a long project as irrational escalation.
A worked example: revising a content project
Imagine spending a month on an article series. New evidence shows that the original central claim is poorly supported. Continuing unchanged would preserve the appearance of consistency, but it would produce misleading material.
Several options remain. You could abandon the series, rewrite it around the stronger evidence, or salvage accurate sections for another purpose. The earlier work does not disappear from history, but it need not dictate the next action.
Compare the remaining effort and likely value of each option. A revision may be worthwhile if it produces useful work with a manageable investment. Abandonment may be reasonable if the project no longer serves its goal.
The lesson is not “ignore the past.” It is “use the past as information and resources, rather than as a demand for further spending.”
Why stopping can feel like accepting a loss
Continuing may preserve hope that the original decision will eventually look successful. Stopping makes the disappointing outcome more explicit. This emotional difference can influence how the alternatives are framed.
You may also worry about explaining the decision to others. A project can become tied to identity: “I am someone who finishes what I start.” That value may support useful persistence, but it can also make revision feel like a personal failure.
These are understandable concerns, yet they should be named separately from the project's future value. A decision can acknowledge disappointment without using additional effort to avoid acknowledging it.
The article on decentering offers a related distinction: a thought such as “stopping means I failed” can be noticed without automatically treating it as the final decision rule.
Use a forward-looking comparison
Describe the options as they exist now. For each, estimate the remaining effort, plausible benefit, uncertainty, and consequences for other priorities. You do not need precise numbers when they are unavailable, but you should make the assumptions visible.
Then ask whether you would choose the same future commitment if you encountered the opportunity today with the existing resources already available. This question can expose the influence of an unrecoverable cost.
It is a reasoning prompt, not a universal formula. Some obligations are ethical or contractual, and some consequences are difficult to quantify. The comparison should include those factors rather than pretending every choice is purely financial.
Distinguish sunk costs from switching costs
Leaving a project can create new costs. You may need to hand work over, learn another system, or explain a change. These are future consequences and therefore relevant.
Similarly, a refundable payment is not fully sunk. A reusable tool may retain value. A relationship may be affected by withdrawal. A careful analysis identifies which costs can still change and which cannot.
This distinction matters because “ignore sunk costs” is sometimes used to dismiss legitimate transition expenses. The principle is narrower: do not count an unrecoverable past expense as if continuing could undo it.
Write the categories separately when the decision is confusing. Past irrecoverable effort, remaining effort, recoverable value, and transition costs answer different questions.
Set review points before commitment grows
Before beginning a substantial project, identify what would count as progress and when the plan should be reviewed. A review point makes reconsideration part of the process rather than a sign that something has gone wrong.
For a learning routine, progress might mean being able to explain and apply a concept after a delay. If repeated sessions produce no improvement, inspect the method and prerequisites rather than adding hours automatically.
The overview of neuroplasticity emphasizes observable learning outcomes. A routine should not be continued solely because it carries an appealing brain-related explanation or because much time has already been spent.
Review criteria should remain flexible enough to accommodate new evidence. They are aids to judgment, not a mechanism for avoiding judgment.
Discuss the decision without rewriting history
A useful explanation can acknowledge that the original choice was reasonable given earlier information and that the current choice differs because conditions changed. Revision does not require pretending the original decision was either perfect or foolish.
Likewise, stopping a project does not establish that every hour was wasted. Skills or information may remain valuable even if the original outcome was not achieved.
Describe what was learned and what will happen next. This makes the decision easier to evaluate than a defensive argument about whether the past investment must be justified at any cost.
Common questions
Is finishing a bad book always a sunk cost fallacy?
Not necessarily. You may value completion, need the material, or expect later sections to be useful. The issue is whether the remaining reading is justified by current goals, not whether someone else would enjoy it.
Does the principle mean emotions should be ignored?
No. Feelings can reveal values and concerns. They should be considered alongside the actual options, while distinguishing disappointment about the past from benefits expected in the future.
What if I cannot estimate future value?
Make the uncertainty explicit and consider a limited next step that provides information. Avoid treating uncertainty as a reason to continue indefinitely without review.
What is the simplest useful question?
Ask: “What would justify the next unit of effort?” If the only answer is that much effort has already been spent, reconsider the comparison. If there is a clear future benefit, evaluate that benefit on its merits.
Sources
- Arkes, H. R., & Blumer, C. (1985). The psychology of sunk cost. https://www.sciencedirect.com/science/article/pii/0749597885900494