The sunk cost fallacy is the tendency to continue investing in a losing proposition because of what has already been spent rather than what is expected going forward. Rational decision-making considers only future costs and benefits — past expenditures are gone regardless of your next action. But recognizing sunk costs is harder than it sounds: identity attachment ("I've spent 5 years on this degree"), social pressure ("we've committed to this strategy"), and loss aversion (abandoning feels like admitting failure) all push toward irrational continuation. Rational quitting is a skill: explicitly separating past investment from expected future value, setting pre-commitment criteria for when to quit, and reframing quitting as redirecting resources toward higher-expected-value opportunities.
For an ongoing project or commitment you are unsure about, perform the "fresh start" test: if you were not already involved, would you start this project today given what you now know? If no, the only reason to continue is sunk cost bias. Practice setting kill criteria in advance: "I will quit if X does not happen by Y date."
From expected value decision-making, you know that rational choices should be guided by the probability-weighted outcomes of each option going forward. The sunk cost fallacy is what happens when past expenditures -- money, time, effort, emotional investment -- distort this forward-looking calculation, causing people to continue investing in losing propositions because of what has already been spent rather than what is expected to come.
The logic against sunk costs is simple and decisive. Suppose you have spent $40,000 and two years building a product that shows no market demand. If you continue, you will spend another $20,000 and six months. If you stop, you lose the $40,000 already spent. But here is the key: you lose the $40,000 either way. It is gone regardless of your next decision. The only relevant comparison is between the expected future value of continuing (spending $20,000 more for an uncertain payoff) and the expected future value of stopping (saving that $20,000 and six months to redirect toward a better opportunity). Past investment is the same in both futures, so it provides zero information for distinguishing between them. Any weight given to "we've already invested too much to quit" is irrational -- it conflates a backward-looking accounting fact with a forward-looking decision.
Yet sunk cost bias is remarkably persistent, even among people who understand the principle, because multiple psychological forces reinforce it. Identity attachment makes abandonment feel like admitting you were wrong ("I've spent five years on this career path -- quitting means those years were wasted"). Social pressure punishes quitting ("we committed to this strategy publicly -- we can't reverse course now"). Loss aversion makes the certain loss of the sunk cost feel worse than the uncertain gain from switching, even when expected values favor switching. And escalation of commitment creates a self-reinforcing cycle: each additional investment deepens the attachment that makes quitting feel costlier. These forces operate simultaneously, and they explain why even savvy decision-makers fall prey to the fallacy despite knowing better.
The practical countermeasures are concrete. The fresh start test asks: "If I were not already involved, would I start this project today given what I now know?" If the honest answer is no, the only reason to continue is sunk cost bias. Kill criteria set in advance -- "I will quit if we don't have 100 users by June" -- separate the quit decision from the emotional context of accumulated investment, because they are made before the psychological forces of continuation have built up. And reframing helps: quitting a failing project is not "admitting failure" -- it is redirecting scarce resources toward higher-expected-value opportunities. Rational quitting is a skill, and like any skill, it requires practice and deliberate effort to overcome the powerful intuitions that push toward irrational continuation.
Topics in reflective domains aren't scored by quiz answers. Read, reflect, and mark when you've thought it through.