Of all the places regression to the mean hides, the cruelest is inside the feedback we give each other. Because when reward and punishment land right after extreme results — and they almost always do — the world hands us a lesson that is not just wrong but backwards. It teaches us that harshness works and encouragement fails. Generations of coaches, teachers, managers, and drill instructors have absorbed that lesson, and it was never anything but arithmetic.
The definitive telling comes from the psychologist Daniel Kahneman, and it’s worth knowing by heart.
Before you read — take a guess
A flight instructor notices: every time he praises a cadet for a superb landing, the next landing is worse; every time he screams at a cadet for a terrible landing, the next is better. He concludes punishment works and praise backfires. What did he actually witness?
Kahneman’s flight instructors
Kahneman was teaching Israeli air force flight instructors about the value of positive reinforcement when one of the most experienced instructors pushed back. In his experience, he said, praise was counterproductive: “On many occasions I have praised flight cadets for clean execution of some aerobatic maneuver. The next time they try the same maneuver they usually do worse. On the other hand, I have often screamed at cadets for badly executed maneuvers, and in general they do better the next time. So please don’t tell me that reward works and punishment does not.”
Kahneman realized the instructor had stumbled into a statistical trap. Performance on a hard maneuver is skill plus luck. An exceptionally good maneuver sits at a lucky peak — so the next attempt is almost certain to be worse, praise or no praise. An exceptionally bad one sits at an unlucky trough — so the next is almost certain to be better, scream or no scream. The instructor only ever praised after the peaks and yelled after the troughs, so he saw praise “followed by” decline and punishment “followed by” improvement. He had constructed, out of pure regression, a perfectly convincing and perfectly false theory of motivation.
Kahneman's punchline
“Because we tend to reward others when they do well and punish them when they do badly, and because there is regression to the mean, it is part of the human condition that we are statistically punished for rewarding others and rewarded for punishing them.” The feedback lands after an extreme; the extreme regresses; we credit the feedback.
Why the lesson always comes out backwards
This isn’t a fluke of aviation. It’s structural, and it will happen to you unless you watch for it. Two human tendencies collide:
- We react most to extremes. Nobody calls a meeting over an average quarter. We praise the standout, punish the disaster, intervene at the peak and the trough — the very observations most inflated by luck.
- Extremes regress on their own. The peak drifts down; the trough drifts up.
Put them together and the pattern is inescapable. Line it up:
| What triggered your reaction | Your reaction | What regression does next | The false lesson you “learn” |
|---|---|---|---|
| An exceptionally good result (lucky peak) | Praise / reward | Result drifts down toward the mean | ”Praise makes people complacent — it backfires” |
| An exceptionally bad result (unlucky trough) | Criticism / punishment | Result drifts up toward the mean | ”Criticism works — be tough” |
Both conclusions are the opposite of what controlled studies actually find (which is that positive reinforcement generally works well). The feedback didn’t cause the reversion — it just had front-row seats.
A manager reviews the data: employees she praised after a banner month usually declined afterward, while employees she reprimanded after a dismal month usually improved. Before concluding 'praise hurts, reprimands help,' what is she missing?
The same illusion, everywhere: superstition and quackery
Once you see the shape, you find it running the whole machinery of superstition and folk cures. Any ritual performed at an extreme gets credited with the reversion:
- Rain dances. You dance during the drought — an extreme dry spell. Droughts end (weather regresses to normal). The dance “worked.”
- Lucky charms and rituals. The team fires the coach during a slump (an extreme low); the slump ends on its own; the firing gets the credit. The player breaks out the “lucky socks” after a cold streak that was going to end anyway.
- Quack cures. People buy the miracle tonic when they feel worst; self-limiting illnesses regress toward better; the tonic collects a glowing testimonial. Regression is a perpetual-motion machine for generating “proof” that inert remedies work — which is exactly why medicine trusts controlled trials over anecdotes (lesson 03).
- Managerial over-reaction. Gut a strategy after one bad quarter, bench a player after one cold week, launch a reorg at the trough — then watch the inevitable bounce and conclude the drastic move was a masterstroke. Over-reacting to noise, then crediting the reaction for the regression.
Because the “evidence” arrives reliably, on schedule, and feels causal. Every time you punish a trough you’re rewarded with improvement; every time you praise a peak you’re punished with decline. A superstition that gets confirmed almost every single time is nearly impossible to shake by experience alone — you’d need to know the math to doubt what your own eyes keep showing you. That’s why regression to the mean has to be taught: raw experience actively argues for the wrong conclusion.
The pitfall: don’t overcorrect into “encouragement is useless”
One caution, so you don’t misuse the model. Regression does not prove that praise, criticism, coaching, or medicine do nothing. It proves something narrower and sharper: the naïve before/after “evidence” you get from reacting to extremes can’t tell you what your feedback did, because reversion is tangled up in it. The real effect of praise vs. punishment is a genuine empirical question — and when it’s studied properly, with controls, positive reinforcement generally comes out ahead. So the lesson isn’t “stop encouraging people.” It’s “stop learning what works from uncontrolled feedback applied to extremes.” Trust designed comparisons, not the seductive after-the-extreme story.
The trap to name out loud
The illusion of learning from feedback: because we intervene at extremes and extremes regress, real life systematically rewards toughness and punishes kindness in appearance only. Anyone who “learned from experience” that harshness gets results, and praise spoils people, likely learned regression to the mean and mistook it for psychology.
Which conclusion does regression to the mean actually license about the flight instructor's observation?
When to use it
Any time you catch yourself concluding “X works” from what happened after an extreme outcome — a punishment that seemed to fix a slump, a ritual that seemed to end a drought, a remedy that seemed to cure a flare-up, a crackdown that seemed to turn a team around — pause. Ask whether the outcome you reacted to was extreme, and whether the improvement would have arrived on its own. Especially resist the emotionally satisfying lesson that toughness works: it is the single conclusion regression most reliably fakes.
Recap
In Kahneman's flight-instructor story, why did praise appear to hurt and screaming appear to help?
Check your answer to continue.
Where this goes next
You’ve now seen regression fake the most convincing lesson of all — that toughness works — right inside our everyday feedback. Lesson 05 is the defensive capstone. We’ll separate regression from two things people constantly confuse it with — a real trend and the gambler’s fallacy (they are not the same, and mixing them up causes its own errors) — and assemble the toolkit that keeps you from being fooled: expecting reversion, demanding control groups, and leaning on larger samples.