Sunk cost fallacy: examples, relationships, and a better next decision

What does sunk cost fallacy mean?

A sunk cost is a cost that cannot be recovered through the choices now available. The fallacy occurs when that irrecoverable amount becomes a reason to continue even though the forward-looking comparison does not support continuing. The important question is not whether the earlier spending was large. It is whether the next action can change it, and what that action changes instead.

In the sunk cost fallacy experiment, distinguish past spending from the future payoffs. Watch what happens to the recommended comparison when only the past cost changes. The model isolates a decision mechanism. Real commitments involve uncertainty, other people, and consequences beyond the simple amounts on the board, so use the exercise to structure questions rather than automate an answer.

Not every consequence of a past decision is sunk

A nonrefundable ticket is different from a refundable ticket. A previous payment is different from an avoidable cancellation charge. Training already completed may be sunk as an expense while still creating skills that affect what you can do next. Classifying the original spending correctly does not require ignoring every continuing benefit or obligation related to it.

Write separate lines for irrecoverable costs, recoverable resources, future benefits, future costs, and switching consequences. This prevents an overcorrection where someone uses the phrase sunk cost to erase legitimate contractual duties or transition needs. The opportunity cost experiment adds another line: what the same time or money could do elsewhere if you stop committing it here.

A worked example: finish the project or change direction?

Consider an invented project with eighty tokens already spent, forty needed to finish, and an expected future benefit of thirty. An alternative uses the same remaining time and costs twenty for an expected benefit of thirty-five. These amounts are stipulated teaching values, not a business forecast. They let us compare the future choices without hiding the past spending inside either option.

1. Record eighty as irrecoverable under both available choices. Continuing does not make it return, and changing direction does not create a second loss of eighty.
2. Compare the future net amounts: finishing offers thirty minus forty, or −10; the alternative offers thirty-five minus twenty, or +15, under the stated assumptions.
3. Check uncertainty, commitments, and switching effects before deciding. If those assumptions remain credible, the future comparison favours changing direction; the eighty explains the history rather than overturning the comparison.

How the idea applies to relationships without reducing them to money

People search for sunk cost fallacy in relationships because shared history can feel like an obligation to preserve any current arrangement. But years together do not fit a simple profit formula. Care, trust, responsibilities, mutual willingness to change, and future wellbeing matter. A useful question is whether the relationship's next chapter has a credible path that both people want, rather than whether the past deserves respect.

Acknowledge the past without requiring it to determine the future. Ask what can improve, what evidence supports that possibility, what boundaries or support are needed, and what each person is willing to do. These questions are not a rule to stay or leave. If safety or coercion is involved, a cost-benefit game is the wrong decision tool; appropriate human support and safety take precedence.

Why persistence can feel like repairing the original decision

Arkes and Blumer's research on sunk costs examines how earlier investments influence subsequent choices. One practical concern is that continuing can feel like proving the first decision was sensible. Yet a new expense does not alter the earlier information or guarantee recovery. A decision made under uncertainty can have been reasonable at the time and still deserve revision now.

In the disposition effect game, a purchase-price label can influence what someone sells even when the forward payoffs stay unchanged. In the war of attrition game, being last to quit can win the prize while producing an overall loss. These are different concepts, but both make it easier to see why emotionally satisfying persistence and a good prospective outcome can diverge.

A five-minute routine before spending more

Begin with the next decision, not a judgment of your entire history. Write what continuing actually requires, what stopping frees up, and which alternative is available. Specify a review date and observable progress rather than making an indefinite promise to give the effort one more chance. A bounded trial can produce information; repeated unbounded extensions can simply postpone evaluation.

OpenStax's discussion of opportunity cost provides a useful companion: scarce resources have alternative uses. That does not require monetising everything important. It asks which other goal cannot receive attention while this one continues. A clear alternative makes the decision more concrete than a general feeling that you should not waste what you already spent.

• What past cost is genuinely unrecoverable?
• What benefit or obligation still exists regardless of the sunk expense?
• What does the next step cost from today?
• What is the best available alternative?
• What evidence and review date would justify continuing?

Leaving is not automatically rational, and staying is not automatically a fallacy

Continuing can be sensible when future benefits exceed future costs, when transition consequences are significant, or when another trial is likely to produce valuable information. Stopping can be costly too. The fallacy is not persistence itself; it is using an irrecoverable expense to replace the forward-looking comparison. The same outward action can come from careful reasoning or from a misleading reference point.

Nor should every disappointment trigger a change. Long projects include difficult periods, and many commitments deserve sustained work. The task is to distinguish a credible reason to continue from a refusal to re-evaluate. Ask someone who is not defending the original choice to review the assumptions. Give them enough information to challenge the forecast rather than merely ask them to approve your preferred conclusion.

Practise on a low-stakes decision first

Choose an unfinished book, unused subscription, or optional hobby purchase. List what is recoverable and what is not, then compare your next hour with its alternative. A small example makes the distinction easier to practise before applying it to a high-stakes project. It also reveals how often the sentence I already paid for it substitutes for an actual description of future value.

The outcome might be continuing. That is fine if the reason is what the next step offers. Learning this concept is successful when the justification becomes clearer, not when every commitment is abandoned. Respecting effort and revising a decision can coexist. Neither requires pretending the earlier experience had no value.