You Started Paying People for What They Loved. They Stopped Loving It.
In 1973, a team of researchers at Stanford gave preschool children who already loved drawing a choice: draw and get a certificate with a gold seal and a ribbon, or just draw. The kids who received the reward drew enthusiastically during the session — but in the weeks that followed, with no reward on offer, they drew significantly less than children who had never received one.
The intrinsic motivation that brought them to the drawing table had been replaced by something weaker. They no longer drew because drawing was worth doing. They drew for the reward — and without it, drawing wasn't worth the effort.
The researchers — Mark Lepper, David Greene, and Richard Nisbett — called this the overjustification effect. It's one of the most replicated findings in motivational psychology. And most organizations building performance incentive systems have never heard of it.
What Happens When You Add "Incentive" to Intrinsic Work
The mechanism isn't complicated. When you start paying people for something they already do because they find it meaningful, you give them an external explanation for their behavior.
Before the reward, the behavior was self-explanatory: I do this because I want to, because it matters to me, because I'm the kind of person who does this. After the reward, there's now an alternative explanation: I do this to get paid. The self-referential motivation — the identity story — competes with the transactional one.
Psychologists call this the self-perception account: people infer their own motivations by observing their behavior and its context. If the context prominently features an external reward, the reward becomes part of the causal story people tell themselves about why they do what they do. Over time, the intrinsic explanation weakens. When the reward disappears, so does the behavior.
Edward Deci's early studies in the 1970s established this through repeated experiments with college students doing puzzle-solving tasks — tasks they genuinely enjoyed. When he introduced payment, intrinsic interest dropped. When he removed payment, the drop persisted. The rewards had restructured how people understood their own motivation. This research became foundational to Self-Determination Theory, the framework Deci developed with Richard Ryan that remains the most empirically supported theory of human motivation.
The Workplace Version
The overjustification effect doesn't show up in psychology papers. It shows up in quarterly performance reviews.
A software engineer who writes unusually good documentation — thorough, clear, actually useful to readers — isn't doing it because of an OKR. She's doing it because she believes documentation matters, because she's embarrassed by codebases she can't navigate, because craft is part of her identity. If her manager decides to "encourage" this behavior by making it a tracked metric and tying part of her review score to documentation quality, something shifts. Documentation was something she owned. Now it's something she's being evaluated on.
The behavior might improve in the short term — she writes even more. But the motivation behind it has changed. If the metric disappears next cycle, so does the documentation. Or worse, the documentation continues but it's now optimized for what the metric measures rather than what actually helps readers.
This is the pattern. The behavior that was intrinsically motivated, that exceeded what any rational incentive would have produced, gets captured by an incentive system that makes it legible to management — and in doing so, makes it transactional. The surplus motivation, the discretionary effort that came from genuine care, evaporates.
The Contingent/Non-Contingent Distinction
The research offers a useful nuance: not all external rewards damage intrinsic motivation equally.
Contingent rewards — rewards that depend on completing a specific activity or reaching a specific performance level — are the damaging ones. "You'll get X if you do Y" creates a conditional relationship that most directly triggers the overjustification mechanism.
Non-contingent rewards — unexpected, given after the fact as acknowledgment rather than as incentive — have less damaging effects on intrinsic motivation and can sometimes reinforce it, particularly when they come in the form of informational feedback ("that was excellent work because of X") rather than controlling feedback ("good job, keep it up").
The distinction matters practically. The annual performance bonus — contingent on metrics, announced in advance — is the most problematic structure for preserving intrinsic motivation in high-quality creative or intellectual work. The spontaneous recognition of excellent work, specific about what made it excellent, carries far lower risk. The former tells people what their work is worth. The latter tells them why it was good.
What This Means for Creativity, Research, and Care Work
The effect is strongest for activities that are intrinsically interesting, that people do voluntarily without external prompting, and that involve the kind of exploratory, open-ended engagement that doesn't lend itself to simple output metrics.
Writing, design, research, engineering on frontier problems, teaching — these are exactly the domains where intrinsic motivation is most productive and most vulnerable to incentive corruption. The person who reads across adjacent fields because they're genuinely curious produces better ideas than the person who does it to hit a learning-and-development target. The difference in output isn't marginal.
This is one reason the autonomy loss that comes with structured AI tool adoption hits creative workers so hard: it's not just that the tool changes the work, it's that the tool plus the metrics that follow it can restructure what the work means to the person doing it. When work becomes a set of tasks to be optimized rather than a domain to be explored, motivation follows that shift in framing.
There's a related phenomenon in learning: the moment someone becomes very good at something and starts being recognized externally for that skill, they often lose the exploratory relationship with the domain that made them good at it in the first place. The gap between accumulating experience and actually getting better is partly explained by this — people optimize for the thing that's being rewarded rather than for genuine skill development.
What to Do With This
Pay people fairly for their work. That's non-negotiable, and "pay them less so they stay intrinsically motivated" is not what this research supports. Market-rate compensation doesn't systematically damage intrinsic motivation — the relationship between pay and identity is more complicated than a simple external/internal split. People can want fair pay and also genuinely love their work.
What damages intrinsic motivation is building elaborate performance incentive structures around activities that people already care about. Turning craft into metrics. Making motivation visible, quantified, and rewarded in ways that give people an external explanation for behavior they previously owned internally.
The most productive approach for high-quality intellectual work is to hire people who genuinely care about the domain, pay them fairly enough that compensation isn't a constant distraction, and then stay out of the way of the motivation that got them to care in the first place. The incentive systems that improve output the most are often the ones that remove obstacles — unclear ownership, meeting bloat, approval bottlenecks — rather than add rewards.
The 1973 drawing study still lands because it captures something most people have felt: the moment when something you did for its own sake became something you did for someone else's recognition of it. That shift is not subtle. And it's very hard to reverse.
Pay people fairly for their time. Leave their curiosity alone.
Photo by Mykhailo Petrenko