Sunk cost fallacy: why you keep paying for a choice you would not make again

A sunk cost is money, time, or effort you already spent and cannot get back. The sunk cost fallacy is the tendency to continue with a choice because of that past payment, even when quitting serves you better. Rational choice looks only at future costs and benefits. You commit the fallacy when past costs decide for you.

That feeling is familiar. You sit through a boring film because you paid for the ticket. You finish a course you dislike because you paid the fee. You stay in a project that fails because you spent months on it. Each time the same question helps: would you still choose this if you had not paid yet. If the answer is no, the past payment is deciding, not the future value.

For the wider family of errors this belongs to, see the overview of ten biases.

What counts as a sunk cost

A cost is sunk when three conditions hold. It is already incurred. It cannot be recovered. No future action changes it.

Money qualifies when the ticket is non-refundable, the fee is paid, or the equipment is bought and cannot be resold for its price. Time qualifies when the hours are spent and no choice brings them back. Effort qualifies when the work is done and only its result remains.

A prospective cost is different. It is still avoidable. Think of a ticket you have not bought. Think of a weekend you have not spent. Think of the next project phase you have not funded. These are prospective. Rational decisions weigh prospective costs against future benefits. They ignore sunk costs. Economists call this the bygones principle: bygones are bygones.

The error is common because the brain treats the two kinds apart with effort. Quitting feels like waste. Continuing feels like rescue. Both feelings point at the past. Neither measures the future.

The two studies that named it

Richard Thaler first described the pattern in 1980. He observed that paying for the right to use a good increases the rate at which people use it. People who paid more for a service used it more, not because they liked it more, but because they had paid.

Hal Arkes and Catherine Blumer gave it the classic test in 1985. They defined the effect in one sentence. It is a greater tendency to continue an endeavor once money, effort, or time has been invested (Dijkstra and Hong 2019, quoting Arkes and Blumer). The paper itself is (Arkes and Blumer 1985).

Their ski-trip questionnaire is still taught. Participants imagined two ski trips booked for the same weekend by mistake. One was a 100 dollar trip to Michigan. The other was a 50 dollar trip to Wisconsin. They were told they would enjoy the Wisconsin trip more and could not return either ticket. Most participants, 54 percent, said they would take the Michigan trip. The larger past payment won over the better future experience (The Decision Lab 2020).

Their theater study tested real behavior. People bought season tickets at full price, with a small discount, or with a larger discount. The full-price group attended a mean of 4.11 shows. The small-discount group attended 3.32. The large-discount group attended 3.29. The more people had paid, the more they forced themselves to attend (The Decision Lab 2020).

Later studies extended the pattern beyond questionnaires to time investments and course choices (The Decision Lab 2020).

Why it happens

No single cause explains every case. Research points to five mechanisms that often combine. The table below sorts them.

Loss aversion and framing come from behavioral economics. Overoptimism was shown in early betting studies. Personal responsibility was shown when participants managed a prior failing investment versus inheriting one from a predecessor. Desire not to appear wasteful appears in ticket and event studies (Asana).

A newer line of work adds affect. In four studies, Dijkstra and Hong measured how bad each sunk-cost scenario made people feel, then recorded their choice. More negative feeling predicted more sunk-cost choice. The effect ran through integral affect, the feeling elicited by the scenario itself, not background mood. When participants wrote a justification for their decision, both the negative feeling and the fallacy dropped. Under high cognitive load, the link between feeling and choice grew stronger (Dijkstra and Hong 2019).

Related biases clarify the picture. Cognitive dissonance pushes you to justify a prior payment by valuing the choice more. When evidence favors persistence, you notice it faster and weigh it more, which protects the original decision.

Sunk cost, escalation, and the Concorde case

Students mix three terms. They overlap but they are not the same.

A sunk cost is the unrecoverable past payment. The sunk cost fallacy is the bias: letting that payment drive the next choice. Escalation of commitment is the behavioral pattern: putting more resources into a failing course over time. The Concorde fallacy is the famous case. Britain and France kept funding the supersonic airliner after its economics had turned poor (Wikipedia).

Use the terms in this order. Name the sunk cost first. Call the error the fallacy. Call the continued funding escalation. Call Concorde the example. Examiners reward that order.

Four everyday cases

Each case below follows the same test. Strip out the past payment. Compare only future options.

1. 1.The boring film. You paid for pay TV and after five minutes the film is bad. A free channel shows a film you prefer. The rational choice is the free film. The payment does not return either way. Staying to get your money worth buys boredom with time.
2. 2.The wrong course. You paid 150 dollars for a course with a 75 percent pass rate. A free course with an 85 percent pass rate appears. Coleman found that over-budget payers stuck with the paid course more than under-budget payers, even with the worse odds (The Decision Lab 2020). The fee is gone. The pass rate is the future.
3. 3.The tired relationship. You invested years, shared friends, and daily routines. The relationship now costs sleep and calm and returns little. Hattis, a family therapist, asks clients to price the continuation: lost sleep, extra spending, weekly repair work. Naming those prospective costs makes the past visible as past (Greater Good Berkeley 2025). This example only illustrates the mechanism. It does not replace advice for your own situation.
4. 4.The failing project. Your team spent six months on a tool users reject. A simpler alternative exists. Continuing to prove the first choice right spends the next six months too. A review date with preset kill criteria protects the next budget. Age matters here with care. In one study, adults aged 58 to 91 were less prone to the fallacy than adults aged 18 to 27. They also stayed more consistent across invest and no-invest versions. Treat it as one finding, not a rule about age.

A harder moral form exists. In a University of Waterloo study, participants imagined a harder choice. They had already run deadly tests on 900 monkeys. Then they were asked about running tests on 100 more to finish a cure, after learning a cheaper cure existed elsewhere. Those with the past cost were more willing to continue (Greater Good Berkeley 2025). The study used hypothetical scenarios, so read it as evidence about judgment, not about real clinics.

The quit question and three fixes

Ask one question before you continue. Would I still choose this if I had not invested yet. Veronika Tait, who studies the effect, recommends exactly this blank-slate test (Greater Good Berkeley 2025). If the answer is no, stop or switch. If the answer is yes, continue for future reasons and name them.

Three fixes support that test.

1. 1.Write the future only. List next costs and next benefits for each option. Exclude paid fees, spent hours, and prior effort from both sides. Choose the better future.
2. 2.Justify in one sentence. Write why the next step is worth its next cost. Dijkstra and Hong found that a short justification reduced both negative feeling and sunk-cost choice. A vague defense fails this test. A concrete gain passes it.
3. 3.Set the exit before you enter. Fix a review date, a cost cap, and a kill criterion at the start. When the date comes, the rule decides, not the past payment. Teams use goals and decision matrices for this for the same reason.

Keep grit for a different case. Grit persists because future returns justify persistence. The fallacy persists because past payments demand it. The quit question tells the two apart.

Save this term with its neighbors in Psychology World. Put sunk cost, escalation of commitment, loss aversion, and framing on one bookmark list and drill the list with flashcards until each definition is exact. For a method that works for any term list, use how to memorise psychology terms.