"One dramatic incident last month — a session that went badly — is now driving every design choice you make. You're calibrating on the vivid memory, not the base rate. You know this. You're still doing it. What would you do?"
The signal we'd send: Name the incident explicitly, in your own notes, as a single data point — not a pattern. Kahneman's research on availability bias shows that vivid, recent events dominate our probability estimates far beyond their statistical weight. The corrective isn't forgetting the bad session; it's deliberately gathering the base rate: review the last twelve sessions, count the actual bad ones, and calibrate design choices on that number. One outlier doesn't revise the method. It asks you to look at the full voyage log.
What We Can't Stop Paying For
In 1985, Hal Arkes and Catherine Blumer published the landmark experiment that named and confirmed what anyone who has watched a failing project continue to receive funding already suspected. Their subjects, having paid for a ski trip they could no longer enjoy due to an injury, consistently chose to attend the worse trip over a better, free one — simply because they had "already paid." The sunk cost had become a reason to act irrationally in the present.
Arkes, H.R. & Blumer, C. (1985). The psychology of sunk cost. Organizational Behavior and Human Decision Processes, 35(1), 124–140. The foundational experimental study demonstrating sunk cost effects: subjects paid more to attend worse outcomes rather than "waste" prior investment. Replicated across ski trips, business investments, and policy decisions.The most vivid real-world demonstration of sunk cost thinking at scale is the Concorde programme. By the mid-1970s, British and French government analysts had concluded that the supersonic passenger jet would never be commercially viable. The technical achievement was genuine; the economic case was not. But both governments had already committed hundreds of millions of pounds. Stopping felt like admitting the money was wasted. So they kept going — for decades — subsidizing an operation that lost money with every flight until 2003.
Cairns, R.D. & Galbraith, J.W. (1990). Artificial competitiveness and the Concorde. The RAND Journal of Economics, 21(2), 299–312. Documents the decision-making structure behind Concorde continuation post-1975: both governments acknowledged non-viability but continuation was driven by prior commitment, not projected return. A canonical sunk cost case study in policy economics.This is the sunk cost fallacy in its clearest form: resources already spent cannot be recovered by continuing. The decision in front of you should be evaluated on its own merits — what it costs from this point forward, against what it gains from this point forward. The past is not a variable. But human cognition treats it like one, and keeps paying for the ghost of what it has already spent.
The Forty-Five Minute Problem
A trainer is 45 minutes into an activity that clearly isn't working. The group has lost energy. The learning isn't landing. Participation has dropped from three-quarters to two people carrying the rest. The practitioner can feel the room cooling.
But they spent an hour preparing this activity. They adapted it from a source they trust. They explained it to their co-facilitator yesterday with genuine enthusiasm. And they are 45 minutes in. Stopping now feels like abandoning the investment — the prep time, the explanation, the 45 minutes already spent building toward a payoff that hasn't arrived.
They keep going. Sometimes the room recovers. More often, it doesn't — and the practitioner spends the debrief quietly reconstructing why the second half fell apart, never quite naming that the answer was visible at the 45-minute mark. The decision not to cut was a sunk cost decision. The cost of making that decision is the learning that didn't happen in the second half.
The practitioner isn't weak or unobservant. Sunk cost effects intensify in proportion to the investment, the visibility of the investment, and the social cost of acknowledging the change. All three are present in a live training room. Arkes and Blumer documented exactly this: the larger the prior investment, the stronger the pull to continue despite evidence.
What the Words Know
The practitioner who pivots at 45 minutes is not wasting the preparation. The preparation happened; it has its own value; it produced the first 45 minutes of a session. The practitioner who continues for another 45 minutes of declining returns is not honoring the preparation — they're paying tribute to a ghost.
The Evidence Chain
Kahneman and Tversky's Prospect Theory, developed in 1979, provides the mechanism underneath sunk cost fallacy. Losses feel roughly twice as painful as equivalent gains feel pleasurable. This asymmetry — loss aversion — means that abandoning an investment feels like a loss that the brain weighs far more heavily than the equivalent gain from the better decision. The practitioner who pivots is not just making a practical choice; they're absorbing a psychological loss that their brain has pre-weighted against them.
Kahneman, D. & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica, 47(2), 263–291. Established loss aversion as a systematic bias: losses are weighted approximately 2× more heavily than equivalent gains. The foundational mechanism explaining why abandoning prior investment feels disproportionately costly.Barry Staw's research on escalation of commitment (1976, 1981) added a crucial organizational dimension: decision-makers who are personally responsible for an initial decision are significantly more likely to continue investing in it despite negative feedback than those who inherit the decision. Ownership of the original choice intensifies the sunk cost pull. For practitioners, this means the activities they designed themselves, the methods they advocated publicly, and the approaches they've defended to their clients are the most dangerous candidates for sunk cost continuation.
Staw, B.M. (1981). The escalation of commitment to a course of action. Academy of Management Review, 6(4), 577–587. Demonstrated that personal responsibility for an initial decision significantly amplifies escalation of commitment to failing courses of action — directly applicable to practitioners who design and advocate for their own methods.Thaler and Sunstein, in Nudge, identify commitment devices as one of the primary tools for overcoming sunk cost effects: pre-committing to exit criteria before an investment begins removes the psychological weight of the in-the-moment loss aversion. The session design equivalent is the practitioner who, before delivering an activity, names the conditions under which they will cut it early — and commits to that standard in advance.
Thaler, R.H. & Sunstein, C.R. (2008). Nudge: Improving Decisions About Health, Wealth, and Happiness. Yale University Press. Chapter 3 on commitment devices: pre-commitment to exit criteria reduces the in-the-moment cost of abandoning an investment by framing departure as plan-adherence rather than loss.In Kegan and Lahey's immunity to change framework, the sunk cost fallacy maps onto a competing commitment that is almost always some version of: I need to not be wrong. I need to not be foolish. I need the prior investment to have been worth making. The deep assumption protecting that competing commitment is usually: if I abandon something I invested in, it means the original decision was a failure — and that reflects on my competence, judgment, or character.
This is why sunk cost effects intensify when the investment was publicly visible, when the decision-maker was personally responsible, and when the audience includes people whose respect matters. The immunity isn't protecting the investment — it's protecting the self-concept of the person who made it.
Constructive stewardship offers a reframe: sunk costs are data, not verdicts. The ship is sunk. The navigator is not. The wisdom of a prior investment is not measured by whether you continue it — it's measured by what you carry forward. Meliorism is not about never being wrong; it's about the quality of learning that comes from having been wrong, and the quality of the next decision that learning makes possible.
The practitioner who cuts an activity at 45 minutes, names what they observed, pivots to something better, and uses the debrief to draw meaning from both halves of the session — that practitioner has not wasted the preparation. They've converted it into judgment. That's the salvage operation worth running.
The Room as Exit Criteria
The sunk cost fallacy is not a personal weakness — it's a cognitive pressure that intensifies in direct proportion to social visibility. A room full of people who watched you set up an activity is exactly the condition that maximizes the pull to continue. Design your container to make pivoting feel like expertise, not retreat.
The physical room can't predict which activities will need cutting — but it can make mid-session pivots feel intentional rather than desperate. A visible "parking lot" surface (wall paper, whiteboard, or flip chart) does double duty: it collects material that gets set aside when you pivot, which communicates to the room that the work from the cancelled activity isn't being discarded — it's being held. Put the parking lot where the whole group can see it. When you pivot, write the activity title on it. This gesture says: we were here; that work existed; we're moving because of what we learned, not because it was wrong to try. The room's architecture should make "we're changing course" look deliberate.
Before you run any high-stakes activity, state your exit criteria aloud to the group: "We'll run this for 30 minutes. At the 20-minute mark I'm going to check in — if the energy is moving, we'll continue; if it's not, we'll extract what we have and go a different direction." This is Thaler and Sunstein's commitment device applied live: you've pre-committed to a decision rule in front of witnesses, which means executing that rule at 20 minutes is plan-adherence, not retreat. The social cost of abandoning sunk investment flips — now, not checking in at 20 minutes would be the deviation. When the check-in arrives, make it literal: ask the group "Is this landing?" Their answer is data. You're modeling that data beats investment.
When you pivot, name the mechanism. Not apologetically — precisely: "I designed this activity and I've been running it for twenty minutes. Here's what I'm observing: the energy dropped at the fifteen-minute mark and hasn't recovered. The preparation I put in doesn't change what I'm seeing. So we're going to extract what we have and move." Then move. This does something few practitioners ever do in the room: it shows that expert judgment is responsive to real-time data, not loyal to prior investment. Your clients are watching you do exactly the calculation they need to do in their own leadership — sunk costs are not arguments for continuation. Seeing a practitioner model that, live, is worth more than an hour of theory about it.
Grit as Cover Story
"Don't quit. Honor your commitments. Persistence is the mark of character. Grit is the difference between people who succeed and people who don't. Finishing what you started is what separates professionals from amateurs. The breakthrough is always just ahead — that's exactly when people give up."
The sunk cost fallacy is not a character flaw dressed up as loyalty — it's a decision error that benefits the people who sold you the investment. When a training methodology isn't working, the methodologists benefit from your continuing. When an institutional program isn't producing outcomes, the institution benefits from you staying enrolled rather than asking for accountability. When a leadership approach is failing, the consultants who sold it benefit from escalated commitment rather than honest evaluation. "Don't quit" is not moral wisdom — it's a frame that distributes the cost of a bad investment onto you and away from whoever designed the investment. Persistence is virtuous when the path is sound. It's expensive when the path is wrong and you're carrying the tab alone.
Ask yourself: who benefits when practitioners keep running methods that aren't working? Who benefits when organizations continue programs past the point of evidence? Who benefits from the cultural story that abandoning a failing investment equals weak character? The answer is almost never the learners in the room.
Your clients are making sunk cost decisions right now — about jobs that stopped working, relationships that have calcified, beliefs about themselves that pre-date the evidence they now have. They need three specific capacities to work with this clearly:
The sunk cost audit: Can they name what has already been spent, separately from what the next decision actually costs? Most people can't separate these columns. Practice helps: "Here's what I've already spent. It's gone regardless. Now: what does the next six months cost, and what does it gain? What would I decide if the prior investment had never existed?" Until those columns are separated, the ghost of prior spending haunts every next choice.
The interests question: "Whose interests does it serve for me to continue?" is not cynicism — it's precision. Teach clients to ask it about every institutional message they're receiving about persistence. Their employer, their training program, their ideology — all of these have interests in their continued enrollment. That doesn't mean all of them are wrong. It means the question is worth asking, specifically and without apology: who benefits from my continuing, and is that person or institution carrying any of the cost of being wrong?
The data reframe: The decision to stop is not an indictment of the decision to start. A sunk investment is evidence that you tried something based on the best available information at the time. If new information contradicts the original read, updating your decision is not failure — it's what the original decision was always meant to produce. Practitioners who help their clients see stopping as a form of scientific competence — not moral weakness — are doing something the "grit" culture rarely offers: an exit that preserves dignity and produces learning.
Your Assignment This Week
The navigator's practice is not to be indifferent to investment — it's to be precise about what information the investment contains. The preparation was data: it told you this group, this room, this moment needed something specific. When that something stops working, the data has updated. Continuing to honor the sunk investment is not loyalty to good judgment. It's loyalty to the judgment you had before you had this new data.
BIBLIOGRAPHY
- Arkes, H.R. & Blumer, C. (1985). The psychology of sunk cost. Organizational Behavior and Human Decision Processes, 35(1), 124–140. — The founding experimental study. Multiple experiments demonstrating sunk cost effects across contexts including professional decisions.
- Kahneman, D. & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica, 47(2), 263–291. — Established loss aversion; provides the psychological mechanism underneath sunk cost fallacy.
- Staw, B.M. (1976). Knee-deep in the Big Muddy: A study of escalating commitment to a chosen course of action. Organizational Behavior and Human Performance, 16(1), 27–44. — First study on escalation of commitment; personal responsibility as an amplifier of sunk cost effects.
- Staw, B.M. (1981). The escalation of commitment to a course of action. Academy of Management Review, 6(4), 577–587. — Review and theoretical framework for escalation; directly applicable to practitioner decision-making under social visibility.
- Thaler, R.H. & Sunstein, C.R. (2008). Nudge: Improving Decisions About Health, Wealth, and Happiness. Yale University Press. — Commitment devices as sunk cost antidote; pre-committing to exit criteria as practical application.
- Cairns, R.D. & Galbraith, J.W. (1990). Artificial competitiveness and the Concorde. The RAND Journal of Economics, 21(2), 299–312. — The Concorde case fully documented; policy sunk cost as historical record.
- Kahneman, D. (2011). Thinking, Fast and Slow. Farrar, Straus and Giroux. — Chapters 31–32 integrate sunk cost with loss aversion in the accessible synthesis; applied professional examples.
- Kegan, R. & Lahey, L.L. (2009). Immunity to Change. Harvard Business Review Press. — Competing commitment framework applied to sunk cost: the competing commitment to not appear wrong as the psychological driver of escalation.
You're a year and real money into a program the evidence now says isn't working. Stopping feels like admitting the whole year was wasted. What would you do?
The year is data, not a verdict. What you learned in twelve months of a program that didn't deliver is the clearest diagnostic you'll ever have — you now know specifically what doesn't work for this context, this population, this mechanism. That knowledge doesn't disappear when the program does. The sunk cost isn't the year: the real cost would be spending another year protecting the story that the first year was a good investment.