# The Sunk-Cost Effect: When Past Spending Traps a New Decision

> Understand when past spending can distort a new choice, what sunk-cost studies show, and how to compare future options without assuming you should quit.

- Canonical URL: https://www.brainmaxx.ink/blog/sunk-cost-effect-when-past-spending-traps-decisions/
- Publisher: Brainmaxx.ink (https://www.brainmaxx.ink/)
- Author: Maya Chen (https://www.brainmaxx.ink/authors/maya-chen/)
- Published: 2026-10-23 (October 23, 2026)
- Updated: 2026-10-23 (October 23, 2026)
- Category: Cognition
- Topics: sunk costs, decision making, cognitive biases, planning
- Reading time: 12 min read

![The Sunk-Cost Effect: When Past Spending Traps a New Decision technical infographic](https://www.brainmaxx.ink/uploads/sunk-cost-effect-when-past-spending-traps-decisions-evidence-infographic.webp "flags:infographics")

## The short answer

Before putting more into a project, separate what is already unrecoverable from what your next choice can still change. The time or money already spent may feel like a reason to continue. The useful question is what continuing would now cost and provide compared with the available alternatives.

That does not mean unfinished work should always be abandoned. A project near completion may offer substantial value for little additional effort. Leaving can also create real costs. The distinction is between using those future consequences to decide and treating past expenditure itself as something the next investment will recover.

## Definitions and scope

A sunk cost is a past investment that cannot now be recovered through the decision under consideration. If the same nonrefundable fee stays paid whether you continue or stop, that fee does not change the difference between the options from this point onward.

The sunk-cost effect concerns past investment influencing a subsequent choice even when it should not supply that advantage. Arkes and Blumer investigated such choices in their 1985 paper through several tasks. Their studies do not diagnose every person who persists. [Arkes and Blumer, 1985](https://cognition.aau.at/bg/BA/Arkes%20%26%20Blumer%201985.pdf).

Be precise about “unrecoverable.” A possible refund is still relevant if one option preserves it. An item you own may have resale value or another use. A cancellation charge may be a future consequence. Those belong in the current comparison; calling the original purchase a sunk cost does not make every associated consequence disappear.

Past experience can also contain information. If earlier attempts reveal that a task is harder than expected, that evidence can help estimate what remains. The distinction is between learning from the experience and saying that having spent a lot, by itself, makes another commitment worthwhile.

## How it works

### Using something already paid for differs from funding more work

Two everyday decisions often receive the same label. One is whether to use something you already bought. Another is whether to put further resources into an ongoing project.

Roth, Robbert and Straus's meta-analysis distinguishes **utilization decisions**, such as using a prepaid service, from **progress decisions**, such as funding the next stage. Its results support sunk-cost effects while also showing variation among tasks and decision contexts. The review covers earlier monetary-cost research; it is not a universal estimate of how often any one person will make a mistake. [Roth and colleagues, 2015 issue, published online in 2014](https://link.springer.com/article/10.1007/s40685-014-0014-8).

The practical difference matters. Having a paid admission ticket does not force you to attend. Completing a partially built project may, however, require a new purchase, more time or another person's work. Spell out the next decision rather than combining the original payment, the remaining commitment and the final outcome into one emotional total.

Neither situation has a rule that “stop” is always the correct answer. You may still enjoy the event, and the remaining project may still be worthwhile. A useful comparison asks what each option offers now, including its burdens and alternatives.

## What the evidence supports

### A classic choice task makes the conflict visible

In Arkes and Blumer's ski-trip questionnaire, college students imagined having bought two nonrefundable tickets for the same weekend. They expected to enjoy the cheaper trip more. Thirty-three chose the more expensive trip and twenty-eight chose the cheaper one. These were hypothetical choices by a particular student sample, not observed holidays or a population estimate of irrationality.

Neither ticket could be sold or returned. Choosing one did not recover its price. Real choices may involve additional uncertainty, obligations or benefits, so the task does not establish a general rule to abandon expensive commitments. [Original experiment, pages 126–127](https://cognition.aau.at/bg/BA/Arkes%20%26%20Blumer%201985.pdf).

### Time investments need their own evidence

It is tempting to assume that money, effort and time always influence choice in the same way. Navarro and Fantino's 2009 paper reports a set of questionnaire and behavioural experiments suggesting sunk-time effects, including a role for personal responsibility. The abstract supports the existence of effects in those tasks; it does not provide a personal probability or show that every long-running activity traps its participants. This article uses the abstract rather than a detailed methods review. [Navarro and Fantino, abstract](https://pubmed.ncbi.nlm.nih.gov/20037669/).

A 2023 registered report by Petrov and colleagues revisited selected studies from Soman's earlier work with an online U.S. sample. It found a money-versus-time difference in one task, effects for both domains in another, and no support for its conceptual replication of the proposed accounting intervention. The retained sample of 821 people completed the combined survey; these are not three independent samples of 821. Its findings argue for task-specific interpretation, rather than a rule that time investments never matter. [Petrov and colleagues, 2023](https://storage.googleapis.com/jnl-up-j-irsp-files/journals/1/articles/883/6565da6fe01ed.pdf).

These papers also differ in date, participants, wording and design. A review can include earlier experiments, so four cited papers do not amount to four independent replications of a single result. None tests the complete practical routine offered below as a guaranteed way to improve everyday decisions.

## Misconceptions and uncertainty

### Why past spending can feel persuasive

Several accounts may contribute: avoiding the feeling of waste, defending an earlier decision, viewing stopping as accepting a loss, or believing more investment will justify the first commitment. The original paper discusses competing explanations, and the later review notes that escalation can involve drivers beyond sunk costs.

Use those accounts as possibilities, not explanations assigned to someone from a sentence they say. “We've already spent months on it” may express frustration, supply useful context or omit a sound reason for continuing. Ask about the next costs and benefits before concluding that the person is biased.

Likewise, being willing to continue is not a measure of low intelligence or weak character. A decision can be difficult because the information is incomplete, the alternatives are poor or the outcome matters. A label adds little unless it helps clarify the actual choice.

## Myth or Real?

![Myth or Real: past costs can influence choices in studied tasks. The ski example was a scenario, not observed travel. Compare real next options.](https://www.brainmaxx.ink/uploads/sunk-cost-effect-when-past-spending-traps-decisions-myth-or-real.webp)

### Claim

**Social paraphrase:** Past spending can influence a later choice even when either option leaves the original payments unrecoverable.

### Verdict

**REAL**

### Facts

In a hypothetical ticket task, some students chose the costlier trip despite expecting more enjoyment from the cheaper one.

The scenario is not observed travel, a personal probability or a rule that continuing is always wrong.

REAL applies to a conditional pattern in studied choices. The ski-ticket question asked students to imagine two overlapping nonrefundable bookings; it did not follow actual holiday decisions. The response pattern fits the sunk-cost concern, but one answer does not identify a unique motive, intelligence level or stable personal trait. The later registered report used several task variants within one combined online sample, rather than independent cohorts for every scenario. Money and time findings varied by task, and its conceptual accounting intervention was unsupported in that implementation. Neither report tests this complete everyday project organiser. The review and the abstract-only time study remain separate sources with their original access limits; unresolved reporting inconsistencies are not invented corrections. A future comparison still includes refunds, resale or alternative uses, real exit costs, commitments and effects on others. Useful information learned from earlier work is not erased. Continuing may be reasonable when prospective consequences support it. The optional project comparison can clarify a reason; it promises no improved outcome or personal odds. Respect preferences, skip needless repetitive analysis and stop distressing checking. Consequential financial, legal, employment, healthcare or safety matters need relevant qualified advice, facts and process. Persistent distress or daily-life interference warrants appropriate professional support.

### Takeaway

For one ordinary optional comparison, assess the real next options, changeable consequences and obligations.

### Sources

Claim discussion: [Public discussion of the Arkes and Blumer ski-ticket example](https://www.reddit.com/r/BehavioralEconomics/comments/1x1tg07/the_sunk_cost_fallacy_in_1985_33_of_61_students/).

Evidence: [Arkes and Blumer (1985): The Psychology of Sunk Cost](https://cognition.aau.at/bg/BA/Arkes%20%26%20Blumer%201985.pdf); [Petrov and colleagues (2023): Sunk Cost Effects for Time Versus Money](https://storage.googleapis.com/jnl-up-j-irsp-files/journals/1/articles/883/6565da6fe01ed.pdf).

Checked 2026-10-20. Scope: Only the conditional existence of past-cost influence in studied decisions. Public theatre numbers, coin-flip findings, commercial quiz and comments are outside this check. No universal quit rule, personal diagnosis or tested organiser benefit.

![The Sunk-Cost Effect: When Past Spending Traps a New Decision concept-model](https://www.brainmaxx.ink/uploads/sunk-cost-effect-when-past-spending-traps-decisions-concept-model.webp)

![The Sunk-Cost Effect: When Past Spending Traps a New Decision practical-support](https://www.brainmaxx.ink/uploads/sunk-cost-effect-when-past-spending-traps-decisions-practical-support.webp)

## Practical meaning and sources

The following is an editorial decision aid, not a tested intervention or a treatment. Use it when an ordinary, reversible planning conversation feels dominated by what has already been spent. You can skip it when the answer is already clear or the exercise would become another round of repetitive checking.

### 1. Name the next commitment

Replace “Should we finish?” with the specific action now available. That might be approving another work session, ordering a part or reviewing one uncertain requirement. State what it would require and what it is expected to produce.

An all-or-nothing question can hide intermediate choices. You may be able to reduce scope, pause for a quote or stop one part while keeping useful work. Do not assume those alternatives exist; find out whether they are actually available.

### 2. Separate past expenditure from changeable consequences

Write two short lists. One contains resources already spent that cannot be recovered. The other contains costs, benefits and obligations that differ among the options from now on.

The first list acknowledges what happened. It does not have to disappear from the conversation. The second list does the comparison work. Include any real exit charge, recoverable payment, salvageable work, remaining effort, useful outcome and effect on other people that the choice can change.

Avoid giving an uncertain benefit an exact number simply to make the sheet look complete. An unknown can remain a question to investigate. A rough estimate should be labelled as such and tested against a plausible alternative estimate.

### 3. Compare the next step with a real alternative

Ask what else could be done with the time, attention or money still available. “Do nothing” may be one option, but it is not the only possible alternative. Compare realistic choices with the same objective and time horizon.

A useful prompt is: “Given what we know today, what would make this next step worth doing?” This retains information learned from earlier work. It does not pretend you can erase the actual starting position or make an existing obligation vanish.

If the remaining work now offers a valuable result for a manageable commitment, continuing can be reasonable. If the main justification is recovering a payment that cannot actually be recovered, ask for a reason tied to a future consequence.

### 4. Decide what information is worth obtaining

Sometimes a small check can resolve the uncertainty: an actual quote, a clearer requirement or a short discussion with someone who owns the next dependency. Identify what decision the information would change before investing in another investigation.

Information gathering also consumes resources. Avoid treating endless analysis as automatically safer. Set a review point appropriate to the ordinary task, then either decide or state exactly what remains unknown. This guide supplies no universal time limit or formula for the value of information.

### 5. Record the reason and a review condition

Write the chosen next step and the prospective reason for it. If continuing, identify an observable condition that would prompt another look: a revised quote, a missed requirement or an outcome that changes what remains worthwhile.

A review condition is useful only if it can be checked and still fits the situation. It is not a promise that you can predict every outcome. Changing course after meaningful new information can be consistent with a sound earlier decision.

Consider this **invented project example**. You have paid a nonrefundable workshop fee, but completing an optional display would now require another evening and a materials purchase. Put the fee in the unrecoverable-past list. Put the remaining evening, materials, any actual cancellation obligation and the value you expect from displaying the work in the changeable-consequences list. Compare those with another realistic use of the evening. The fee stays paid either way. You might reasonably continue if making the display still matters enough to you, or stop if the remaining commitment no longer fits. This example supplies no measured benefit, ideal price or instruction to ignore an obligation; it shows which question each item belongs to.

### Reasons to continue can survive the check

| Reason offered | Question to clarify it |
| --- | --- |
| “Most of the work is already done” | What remains, and what useful result would completing it now provide? |
| “We will lose the payment if we stop” | Can the next choice actually recover it, or is it already unrecoverable? |
| “Stopping affects other people” | What concrete consequences or commitments differ between the available options? |
| “Another attempt could work” | What evidence supports that expectation, and what would the attempt cost? |

The aim is to make the justification assessable. Completion value, coordination and contractual consequences can be real. They should not be discarded merely because someone mentions earlier spending in the same conversation.

### Limits and when to get help

Do not use a sunk-cost label to make another person's preferences irrelevant or pressure them to quit. People can value completion, enjoyment, relationships or learning differently. Ask what matters to the decision and distinguish a preference from an unsupported prediction.

This ordinary project guide does not resolve investment trades, contracts, employment disputes, healthcare choices or personal safety decisions. Those may require relevant qualified advice, facts and obligations outside a simple comparison. It is not a diagnosis and provides no test for a mental health condition. If decision distress or repeated checking persistently interferes with daily life, seek appropriate professional support rather than treating another worksheet as a diagnosis or cure.

For the everyday case, success is a clearer reason for the next step. It may support continuing, changing scope, waiting for specific information or stopping. The past investment has happened; the part you can still shape is what follows.

## Sources

### Evidence and access limits

The article uses the actual Arkes–Blumer PDF, the actual Roth–Robbert–Straus review PDF and the actual Petrov registered-report PDF, plus Navarro–Fantino's abstract. The review's abstract and results contain inconsistent numerical counts, so no pooled count or effect-size estimate is used here. The registered report's abstract, results table and discussion support the task-specific summary; a contradictory direction statement in its conclusion is excluded. These access and reporting limits do not become invented corrections. No original-data reanalysis, personal experiment, clinical benefit or validated everyday decision procedure is claimed.

## Frequently asked questions

### Does identifying a sunk cost mean I should always stop?

That does not mean unfinished work should always be abandoned.

### Can earlier experience still provide useful information?

Past experience can also contain information. If earlier attempts reveal that a task is harder than expected, that evidence can help estimate what remains.

### Is this a test for a mental health condition?

It is not a diagnosis and provides no test for a mental health condition.

---

Cite as: "The Sunk-Cost Effect: When Past Spending Traps a New Decision", Brainmaxx.ink, updated 2026-10-23, https://www.brainmaxx.ink/blog/sunk-cost-effect-when-past-spending-traps-decisions/