Sunk Cost Fallacy

A simple explanation

The sunk cost fallacy is the persistence in a course of action — a project, a relationship, a degree, an investment, a campaign — because of what has already been spent on it, when a clear-eyed evaluation of what remains ahead would not justify continuing. Money, time, effort, attention, identity: any resource that has already been committed can become the reason for committing more, and the more that has been committed, the harder it becomes to stop.

The bias is not commitment itself. Commitment is load-bearing, and persistence past the point of short-term discouragement is often what produces real outcomes. The bias is the moment commitment-protection takes over from forward-looking evaluation — when the decision is no longer is this worth doing from here but can I afford to have already wasted what I have.

An everyday example

You bought a non-refundable ticket to an event a month ago. On the day, you are unwell, the weather has turned, and the event has lost most of its appeal. You go anyway. The ticket money is gone whether you go or not; the cost of going — the discomfort, the lost rest, the strain — is paid only if you go. The decision the evening asks of you is a forward-looking one. The decision you make is a backward-looking one. The Reward System protects the original commitment by routing you out the door.

The same mechanism runs on much larger scales. The graduate programme half-completed despite the loss of interest. The business kept open past the point of viability because of years of effort. The relationship continued past the point of mutual care because of the years already shared. In each case, the question being answered is not the question the situation is asking.

Why do I keep going on something I know isn't working?

Because admitting that prior investment will not produce its expected return is, somatically, painful — and the Reward System routes by short-term somatic cost. To stop is to convert prior expenditure into acknowledged waste, which the body feels as a small loss in real time. To continue is to defer the conversion. The System, asked which option lowers immediate autonomic load, supplies the deferral with conviction.

The deferral has its own costs. Each additional unit of resource committed in the deferring direction increases the prior investment that future stopping would have to convert into waste. The bias gets denser as it runs, which is why escalation of commitment is so often the form sunk-cost reasoning takes in practice.

The behavioral loop

A loop that hides because each step looks like determination:

1. Original investment — resources are committed to a course of action: money, time, effort, identity, public commitment.
2. Early feedback — signals begin to suggest the course is not delivering on its expected return.
3. Forward-looking question — the situation begins to ask from here, with what remains, is this worth continuing?
4. Backward routing — the Reward System routes the question backward: given what we have already put in, can we stop?
5. Continuation — the course is continued, often with additional resources, to protect the prior commitment from being recategorised as waste.
6. Identity entanglement — your self-concept becomes increasingly bound to the course; stopping now would require revising not only the decision but who you have been.
7. Escalation — further investments are made under the same logic; the prior commitment grows, raising the cost of stopping at each step.
8. Sealed pattern — the course is sustained well past the point where forward-looking evaluation would have stopped it, and the cost is paid in the resources that could have built whatever else.

Emotional drivers

Four feelings, often layered:

• A felt loyalty to the original decision and the self who made it.
• A pre-emptive shame at the prospect of acknowledging waste.
• A fear of the identity revision that stopping would require.
• A quiet relief, with each continuation, at having postponed the reckoning.

What your nervous system does

The Reward System treats commitment-protection as a category of reward, because honoured commitment in ancestral environments produced relational trust, group standing, and predictability of self-image. The autonomic system files continuation with a small ease and files stopping with a small surge — the surge of converting prior effort into acknowledged loss. The System routes by the gradient.

Across years, the gradient deepens. The autonomic system has built its predictions around the continuing course; the model the body would have to rebuild if you stopped grows with each step. The cost of stopping rises, not because the situation has changed but because the body has refined its prediction that this is who you are.

The DojoWell interpretation

The sunk cost fallacy is one of the clearest examples of a Reward System deposit paid in commitment-protection and reclaimed in forward-looking value. The System's original request — honour what we have committed to — is a deep one and a load-bearing one. The substitute, never asked for explicitly, is the protection of the original commitment past the point where the situation it was made under still applies. The substitution does not feel like substitution. It feels like character.

The density signature is false_progress because the bias does not register as a cost in real time. Each continuation feels like principle, persistence, commitment. The system logs continuous effort. The residue accumulates somewhere else: in the resources poured into courses that no longer earn their forward value, in the opportunities not taken because the resources were committed elsewhere, and in the slow erosion of self-trust as the gap widens between what you can see clearly and what you can act on.

The work is not to abandon commitment as a virtue. Commitment that survives short-term discouragement is part of how real outcomes are built. The work is forward-looking re-evaluation — at regular cadence, with the prior investment explicitly set aside — so that continuation is a chosen verdict each time rather than the protective default the System supplies.

How do I evaluate what's left to do without being pulled by what I've already done?

You install a small set of practices that re-pose the question the System rerouted. The System will not re-pose it on its own.

Three moves:

1. Ask the from-here question. If you were arriving at this course today, with no prior investment, would you commit the resources still required to continue?
2. Separate the identity question from the decision question. Stopping a course is not the same as revising who you have been; the conflation is the System's work.
3. Pre-schedule a reckoning. Set a date in advance — six months, a year — at which forward-looking re-evaluation will run regardless of how it feels in the moment. Pre-commitment beats in-the-moment courage.

Practical steps

1. For one current commitment, list what continuing further will cost from today forward. Resist the small pull to include what has already been spent. The exclusion is the discipline.
2. For the same commitment, ask whether the original case for it still holds under current conditions. If it does not, name what changed.
3. Identify three small experiments that would test continuation cheaply. Cheap tests break the escalation pattern.
4. Reframe stopping as resource recovery. The resources you do not commit to a failing course are available for whatever else; the framing helps the System relax its protection.
5. Distrust language of betrayal and waste. When stopping is described as betrayal of the original commitment or as wasting what came before, the bias is running the rhetoric.

Reflection questions

• Which of your current commitments would you not begin today, knowing what you now know?
• Where has loyalty to a prior decision crowded out the question of what is still worth doing?
• What identity revision are you avoiding by continuing a course that no longer earns its forward value?
• Which resources would become available if you stopped one course you have been continuing on momentum?

Frequently Asked Questions