What Sunk Cost Is Doing to Your Decision
Holding on to something past its usefulness is not a failure of will. It is a predictable output of how people value what they have already spent — and the effects have been measured precisely enough that you can work with them.
What follows alternates: a finding, then what it means for the specific thing you are holding on to.
Finding: people honour costs they cannot recover
The classic demonstration involved theatre season tickets. Buyers were randomly given full price, a small discount, or a large discount. The tickets were identical and the discount was random, so there was no reason for it to affect anything.
Full-price buyers attended significantly more performances over the following months.
The amount already paid — money gone, unrecoverable, identical entertainment on offer — changed how much people used the thing.
What that means for what you are holding
Ask what you would choose if you were arriving at this situation today, with none of the history and none of the investment.
If the answer differs from what you are currently doing, the difference is being produced by costs you cannot get back. That is not a reason to continue; it is the specific bias the experiment isolated.
The question has to be asked in that form. Asking "should I keep going" invites you to justify the past. Asking "would I start this today" does not.
Finding: losses weigh roughly twice as much as gains
Across many studies, people require a gain of about twice a given loss before a gamble becomes attractive. Losing fifty feels roughly as significant as winning a hundred.
The ratio varies by person and context, but the asymmetry is robust and it is large.
What that means for the arithmetic
Whatever you would give up by letting go is being weighted at roughly double, automatically, before you have thought about it. What you would gain is being weighted at face value.
So a decision that feels close to even is probably not even. If letting go feels like a marginal call, the asymmetry suggests it is already the better option and is being dragged back to parity by the weighting.
This is one of the few places where knowing the size of a bias is directly actionable — you can adjust for a factor of two.
Finding: escalation grows with responsibility
Commitment to a failing course of action increases when the person deciding is the one who started it. People who inherit a struggling project are markedly more willing to end it than the people who launched it.
Self-justification is the usual explanation: ending it concedes that starting it was a mistake, and the concession is what is being avoided rather than the loss itself.
What that means about who should decide
If you chose the thing, you are the worst-placed person to judge whether it should continue — not through weakness, but through a documented and specific effect.
The workaround is to borrow someone else's position. Describe the situation to a person who has no history with it and watch which way they lean. Their judgement is not better than yours in general; it is better on this particular question because they carry no responsibility for the original decision.
Finding: the bias needs a self to protect
An intriguing detail: sunk-cost behaviour is inconsistently observed in other animals and in very young children. The effect appears to strengthen with the development of a sense of self that has a reputation to maintain.
The evidence here is more contested than the earlier findings and worth holding loosely. But the direction is suggestive.
What that means about the discomfort
If the bias runs on self-image, then what makes letting go hard is not the loss of the thing. It is the admission attached to it.
Which is testable. Ask whether you would find it easier if nobody knew you had ever been involved. If yes, you are protecting a reputation rather than a resource, and those call for different responses. Reputations recover faster than people expect, and they recover faster from a clean ending than from a long visible decline.
The twelve minutes
Make a cup and take one thing through the four questions, writing the answers rather than thinking them.
- Would I start this today? Arriving fresh, no history.
- What am I weighting at double? Name the specific loss, then halve it deliberately and see whether the decision moves.
- What would someone with no history say? Ask an actual person if one is available; the imagined version is weaker but still useful.
- Would this be easier if nobody knew? If yes, the thing being protected is not the thing you thought.
Twelve minutes is enough for one thing. It is not enough for a category — "my career", "this relationship" — and attempting one of those in a single sitting produces a lot of feeling and no decision.
Finally, one honest limit. None of this says letting go is correct. Sunk cost is a bias in the direction of continuing, so correcting for it makes ending more likely — but persistence through difficulty is also how most worthwhile things get finished, and the same four questions asked in a bad week will point at quitting things that should not be quit. The questions identify whether the past is doing the deciding. They do not tell you what the answer should be.
One counterweight to all of the above: some things are worth keeping precisely because you invested in them. A block of ceremonial cacao paste in a cupboard is not a sunk cost — it keeps for months and it will still be good.
