Overjustification effect
Paying people for something they already enjoyed can make them enjoy it less.
When an external reward is introduced for an activity a person already found interesting, their later willingness to do it without the reward tends to fall. The proposed mechanism is that the reward supplies an explanation for the behaviour, displacing the interest that was there before. The effect is real and bounded, not a general law that rewards backfire.
How it shows up in software
The risk shows up when a product attaches points, badges or payments to something users were doing for their own reasons: writing reviews, answering questions, contributing code, sharing work. The reward can raise volume while lowering quality and hollowing out the motive, and removing it later leaves you worse off than before you added it.
Using it well
- Leave intrinsically motivated behaviour alone. If contributions are already happening, reduce friction instead of adding incentives.
- Use informational feedback, which tells people their work was good or useful, rather than tokens that stand in for value.
- Make rewards unexpected and retrospective when you use them, since the undermining effect is tied to rewards anticipated in advance.
- Reward for a quality threshold rather than for participation, so the reward points at the thing you actually want.
Where it turns manipulative
- Introducing payment into a volunteer community, then withdrawing it when the budget ends, can destroy contribution that existed for years beforehand.
- Gamifying creative or caring work with points tends to strip the meaning people found in it, which is documented in workplace settings.
- Citing the effect selectively, to justify not paying contributors for work you profit from, uses psychology as a cost-saving argument.
Where you have seen it
Wikipedia
Contribution runs on no points or payment, with recognition through peer review and editorial standing.
Stack Overflow
Reputation is tied to votes on answer quality and unlocks moderation abilities rather than converting to money.
What the research says
- Lepper, Greene and Nisbett, 1973Mixed evidence
Preschoolers who expected a certificate for drawing spent less free time drawing afterwards than those who got no reward or an unexpected one.
The expected-reward distinction is the key result and is often dropped when the finding is retold.
- Deci, Koestner and Ryan, 1999Mixed evidence
Meta-analysis of one hundred and twenty-eight experiments found expected tangible rewards reliably reduced free-choice persistence on interesting tasks, while verbal praise did not.
The boundaries matter more than the headline. It applies to tasks that were already interesting, to rewards expected in advance, and to tangible rewards. Outside those conditions it largely disappears.
- Cameron and Pierce, 1994Contested
A competing meta-analysis concluded the undermining effect was narrow and largely limited to one reward type, which triggered a long methodological dispute.
The dispute was never cleanly resolved. Anyone telling you rewards always destroy motivation, or never do, is skipping this exchange.
Grades are a judgement about the evidence, not about how useful the idea is. Plenty of contested effects are still worth knowing, as long as you do not cite them as settled.