The Israeli Air Force flight instructors experiment is a teaching episode in which Daniel Kahneman exposed a powerful causal illusion. Experienced instructors believed that harsh punishment improved flight cadets’ performance whereas praise made it worse. By replacing a disputed lesson about training with a simple coin-throwing demonstration, Kahneman showed that the apparent effects of feedback could arise from regression to the mean: unusually good or bad performances tend to be followed by less extreme ones even when nothing has caused the change.

Setting and Dispute

The episode began when Kahneman was teaching flight instructors in the Israeli Air Force about effective training. He presented what he regarded as an important principle of skill training: rewards for improved performance work better than punishment for mistakes. Research involving pigeons, rats, humans, and other animals supported the proposition, but the instructors’ experience seemed to contradict it.

One of the most seasoned instructors responded after Kahneman’s enthusiastic presentation. He acknowledged that rewarding improvement might work for birds but rejected its application to flight cadets. He described praising cadets after clean aerobatic maneuvers, only to see them perform worse on the next attempt. By contrast, when he screamed into a cadet’s earphone after a bad maneuver, the cadet generally did better the next time. From the instructor’s perspective, punishment appeared effective and reward appeared counterproductive.

The instructor’s observations were accurate, but his explanation was not. Praise was given after performances far above average, while punishment followed performances that were unusually poor. Because extreme performances partly reflected random fluctuations in skill or circumstance, the next performance was likely to be closer to the average regardless of the feedback. The instructors had attached a causal interpretation to the ordinary movement of a variable that was not perfectly stable.

The Demonstration of Regression to the Mean

Rather than answer with algebra, Kahneman marked a target on the floor. He asked each officer to turn away from it and throw two coins at the target in immediate succession without looking. The distances from the target were measured, and the results of both throws were written on a blackboard. The first attempts were then ordered from best to worst.

The pattern was unmistakable. Most, though not all, officers who performed best on the first throw did worse on the second, while those who performed poorly at first generally improved. The sequence reproduced the instructors’ experience with aerobatic maneuvers, yet no praise or punishment had intervened between the throws. Poor performance was followed by improvement and good performance by deterioration because the first result contained chance variation, not because the second result had been produced by feedback.

For Kahneman, the demonstration produced one of the most satisfying moments of insight of his career. The instructor had been right about the sequence of events and completely wrong about its cause. A cadet praised after an unusually good maneuver was likely to have been unusually lucky on that attempt and therefore likely to perform less exceptionally on the next one. A cadet punished after an unusually bad maneuver was likely to improve naturally as the random element receded. The same pattern could therefore make ineffective punishment look successful and effective praise look harmful.

The Feedback Trap

The instructors were caught in what Kahneman called an unfortunate contingency. Since they punished cadets when performance was poor, they were commonly followed by an improvement; since they praised cadets when performance was excellent, they were commonly followed by a decline. The feedback system thus rewarded the instructors for being nasty and statistically punished them for being nice, even if their interventions had no effect.

The episode illustrates a broader problem in human judgment. People naturally look for causes immediately before changes in performance, and feedback is especially tempting as an explanation because it is intentional, visible, and emotionally charged. But when the initial observation is an extreme score, the correct prediction is often a movement toward the average. Without a comparison group or a recognition of random fluctuation, natural regression can be misread as evidence that an intervention worked.

Significance for Training and Judgment

The experiment does not establish that praise and punishment are psychologically identical in every training situation. Its force lies in showing that the instructors’ informal evidence could not support their conclusion. Their personal experience was real, but it was structurally incapable of distinguishing the effect of feedback from the statistical tendency of extreme performances to become less extreme.

The lesson belongs to Kahneman’s larger account of the mind’s susceptibility to causal stories. A sequence containing an action and a subsequent change feels explanatory even when the change was already likely to occur. The coin throws made that hidden probability visible: once performance was ranked by the first attempt, the second attempt revealed the expected return toward the center without any intervention at all.

The event also gives regression to the mean a memorable human consequence. Instructors who tried to help after success could feel that their kindness had been followed by failure, while those who reacted angrily to failure could feel vindicated by improvement. The resulting pattern could shape future behavior, reinforcing harshness through misleading feedback. What appeared to be a lesson about how to train cadets was therefore also a lesson about how easily life’s feedback can reward the wrong interpretation of events.