The Overjustification Effect: When Rewards Backfire

You pay people to do a thing. They do more of it. That’s the whole theory of incentives — and for tasks nobody wants to do, it holds. The trap is what happens when the task was already interesting.

Here’s the honest version. When someone already finds an activity rewarding, paying them for it can reduce how much they do it once the pay stops. This is the overjustification effect, and it’s one of the better-replicated findings in motivation research: across 128 experiments, tangible rewards offered up front reliably undermined intrinsic motivation. It isn’t universal — the effect is strongest for tasks people already enjoy, and it barely touches boring work. But if you lead people who are intrinsically driven, the reward you add can quietly become the reason they stop.

What is the overjustification effect?

The name is the mechanism. When you already do something for its own sake, you attribute your effort to internal reasons: it’s interesting, it’s yours, you care. Add a salient external reward and you now have two justifications — and the external one is louder. Your mind quietly re-files the behavior: I’m doing this for the money. Remove the money, and the original internal reason has been crowded out. You’ve been over-justified.

The critical variable is the starting point. Overjustification only has something to undermine when intrinsic interest is already there. For a task you’d never touch unpaid, a reward has nothing to displace.

What’s the actual evidence that rewards backfire?

The landmark study is deliberately simple. In Lepper, Greene & Nisbett (1973), researchers at Stanford’s Bing Nursery School selected 51 preschoolers who already loved drawing with felt-tip markers during free play. They split them into three groups. One was promised a “Good Player Award” for drawing (expected reward). One drew and got the same award as a surprise (unexpected reward). One drew with no award at all (control).

The award itself changed nothing in the moment — every child drew. The effect showed up one to two weeks later, when the markers reappeared in the classroom with no reward on offer. The children who had been promised a reward now spent about half as much free-choice time drawing as the children who were never offered one. The surprise-reward and no-reward groups didn’t differ. Only the expected reward did the damage — the one that let a child think “I draw to get the certificate.” Take the certificate away, and the reason to draw went with it.

Is this a one-off finding or a real pattern?

It’s a pattern, and it survived the most rigorous test anyone has run on it. In 1999, Deci, Koestner & Ryan published a meta-analysis in Psychological Bulletin pooling 128 experiments on extrinsic rewards and intrinsic motivation — the definitive quantitative review of this literature.

The results were consistent. Rewards contingent on doing the task, completing it, or performing well all significantly reduced free-choice intrinsic motivation, with effect sizes of d = −0.40, −0.36, and −0.28 respectively. Tangible rewards that were expected in advance were the reliable offenders. And the harm was larger for children than for college students — the more the reward looks like the point, the more it displaces the original interest.

The same analysis found the flip side, which is the part managers should tattoo somewhere. Positive verbal feedback — praise — enhanced intrinsic motivation (free-choice behavior d = +0.33; self-reported interest d = +0.31). Unexpected rewards, and rewards not tied to the behavior, didn’t do the damage either. It isn’t that recognition is dangerous. It’s that a specific structure — an expected, tangible, contingent payment for something already interesting — is.

When don’t rewards backfire? (The boundary most write-ups skip)

Overjustification is real but narrow, and honest use means naming its limits.

A 2017 quantitative review by Levy and colleagues in the Journal of Applied Behavior Analysis looked for the effect in a very different setting: 65 data sets of people with intellectual and developmental disabilities, comparing behavior before reinforcement to behavior after it was withdrawn. The overall effect size did not differ from zero. Behavior after rewards was as likely to be higher as lower. Where the target behavior did dip, it tended to be behavior that had already been occurring at high rates beforehand — exactly the “already interested” precondition the theory predicts.

That’s the boundary. Overjustification needs pre-existing intrinsic interest to undermine. Pay someone to do tedious, low-interest work and withdrawing the pay won’t sink them below where they started — there was no intrinsic motivation to crowd out. The effect targets your most engaged people doing your most engaging work, which is precisely where it’s easy to miss.

How should a leader actually use this?

Incentives aren’t the enemy. Misapplied incentives on already-motivated people are. Practical translation:

• Don’t put a price on work people already love. Bonusing your most intrinsically driven person for the project they’d do anyway is the exact Lepper setup. You risk teaching them the certificate was the point.
• Prefer feedback over transaction for interesting work. Specific, genuine praise raised intrinsic motivation in the meta-analysis. Recognition informs; a per-unit payment re-frames.
• Use tangible rewards where interest is low. For dull-but-necessary work, there’s little intrinsic motivation to undermine — incentives largely just help.
• Make rewards unexpected when you can. Surprise rewards after the fact didn’t undermine anything in Lepper’s data, because they never reframed why the person was working.
• Watch the withdrawal, not the payout. The damage is invisible while the reward flows. It shows up when it stops — like the markers reappearing two weeks later.

This is also why sustainable behavior change rarely runs on willpower and payouts. It runs on identity — becoming someone for whom the behavior is simply what you do, no external justification required. That’s the same reason identity-based habits outlast incentive schemes, why chasing a dopamine “reset” misreads how the reward system works, and why self-compassion beats self-punishment after a lapse. Hypnosis-based change works on the same layer — rehearsing the internal reason a behavior is yours, rather than bolting an external one on top. For the fuller evidence map, see the pillar guide on AI hypnotherapy and behavioral change.

The takeaway

Rewards don’t kill motivation across the board — that’s the lazy version. What the research actually shows is sharper and more useful: expected, tangible rewards for work someone already finds intrinsically interesting can quietly replace the internal reason with an external one, and the cost only surfaces when the reward stops. Praise your best people. Pay them well and fairly. Just don’t turn the work they love into a transaction — and if you must incentivize, aim it at the work nobody wants to do anyway.