Why We Keep Throwing Good Resources After Bad: The Sunk Cost Fallacy Examined.

In 2016, a research team at the University of Oxford did something unusual with brain-damaged patients. They gave them investment decisions (the same kind of decisions that trip up hedge fund managers, military strategists, and anyone who has ever stayed in a bad relationship too long) and discovered something that inverts four decades of assumptions about sunk cost psychology. The patients with damage to their ventromedial prefrontal cortex did not just make different decisions. They made better ones.^[33] The part of the brain that drives persistence, the part that whispers you've come this far, was gone. And with it went the fallacy.

That finding, published in Nature Human Behaviour, is the sharpest evidence yet that the sunk cost fallacy is not a failure of logic. It is not a gap in education. It is a feature of neural architecture: a system that treats prior investment as a reason to continue, regardless of whether continuing makes sense. Removing the system does not impair decision-making. It improves it.^[33]

The sunk cost effect is one of the most replicated phenomena in behavioural science. Roth and Robbert's meta-analysis of 98 independent effect sizes confirmed a moderate, reliable distortion: d = 0.496 across four decades of research.^[18] The effect shows up in laboratory vignettes, corporate boardrooms, and (as a parallel line of work in rodents has demonstrated) even in species that have never heard of a balance sheet.^[21]

The mistake most people make when they first encounter the sunk cost fallacy is to assume it is about stupidity. It is not. Haita-Falah's laboratory analysis demonstrated that raw fluid intelligence (the kind measured by IQ tests) does not protect against the bias.^[19] What matters is not how smart you are but how your brain encodes investment. The sunk cost effect runs on the same loss aversion circuitry that Kahneman and Tversky identified in their foundational work on prospect theory: losses feel approximately twice as painful as equivalent gains, and abandoning a sunk cost registers neurally as a loss.^[3][12]

This is why awareness campaigns fail. Telling someone about the sunk cost fallacy is like telling someone afraid of heights that gravity is constant. The information is correct. It changes nothing about the felt experience. The bias persists because the neural machinery that produces it is older than language, older than culture, and, if the cross-species evidence holds, older than the mammalian cortex itself.

Thaler's mental accounting framework explains the bookkeeping layer: people open a mental account when they make an investment and experience acute discomfort at the prospect of closing that account in the red.^[4][5][8] The account stays open. The investment stays active. The loss stays unfelt until exit forces the write-down.

The sunk cost fallacy begins in the oldest part of the decision-making system. When you invest effort, time, or money into a course of action, the brain does not simply record the expenditure. It recalibrates value. In animal models, Eshel and Malenka demonstrated that dopamine release in the striatum (the brain's central reward-processing hub) scales with the effort required to obtain a reward: the harder you worked, the more the reward pathway insists the outcome must be worth it.^[31] This animal-model evidence suggests a neurochemical basis for why prior investment inflates perceived value, though the extent to which this mechanism operates identically in human sunk cost decisions has not yet been directly established.

That matters because the distortion starts before conscious reasoning enters the picture. The ventral striatum and dorsolateral striatum are not deliberation centres. They are valuation centres. By the time a person sits down to evaluate whether to continue a failing project, the reward circuitry has already inflated the perceived value of the investment. Not because the project is good, but because the effort was large.^[31][37]

Li and colleagues' 2013 fMRI study, the first neuroimaging study of the sunk cost effect, found that higher sunk costs activated lateral frontal and parietal cortices associated with risk-taking, while lower incremental costs activated the striatum and medial prefrontal cortex associated with reward processing.^[13] No overlapping brain areas responded to both. The sunk cost effect, at the neural level, is not a single computation. It is a shift in which network runs the decision.

The second node in the circuit is the ventromedial prefrontal cortex (vmPFC). This region performs two functions that appear contradictory but are actually complementary. Haller and Schwabe's 2014 imaging study found that previous investments reduced vmPFC contribution to current decision-making.^[14] The more you had already spent, the less the rational value-comparison system participated. The vmPFC was being suppressed; its prospective-value function was going quiet.

Holton and colleagues' 2024 study revealed the other face of the same coin. The vmPFC also drives goal commitment through selective attention: as participants approached goal completion, vmPFC activity increasingly prioritised goal-related stimuli over alternatives, a process called selective attentional capture.^[33] These are not contradictory findings. They describe two functions of the same region: the vmPFC's role in calculating future value is suppressed by sunk costs (Haller), while its role in locking attention onto the current goal is amplified (Holton). The net effect is a brain that stops comparing options and starts defending the choice it has already made.

The causal proof came from Holton's lesion patients. With the vmPFC damaged, participants showed significantly reduced persistence bias, and their overall decision performance improved.^[33] The region that drives commitment also drives the fallacy. Remove it, and people make better choices.