The overjustification effect: why paying for play can spoil the fun, and when it does not

The overjustification effect is the drop in later interest in an activity you already enjoy after you do it for an expected tangible reward. You paint for fun, then you paint for pay, then painting feels like work even after the pay stops. Deci showed the pattern with puzzles in 1971. Lepper, Greene and Nisbett showed it with drawing in preschool children in 1973. The effect holds under clear conditions, and it disappears under others.

• Intrinsic motivation means you do the activity because the activity itself pleases you. Extrinsic motivation means you do it for a separate outcome such as money, grades or praise. Our guide to intrinsic and extrinsic motivation defines both types.
• Deci paid students 1 dollar per puzzle in one session only. In the next unpaid session they spent less free-choice time on puzzles than before and less than unpaid controls.
• Lepper and colleagues promised preschool children a certificate with a gold seal for drawing. One to two weeks later the promised group spent less free-play time drawing than children who got the same certificate by surprise or got nothing.
• The hinge is expectation. The same reward harms when promised beforehand and does little harm when it comes as a surprise after the act.
• Tangible rewards tied to doing the task undermine most. Verbal praise that tells you what you did well usually does not undermine and can help.
• Rewards for dull tasks with low initial interest can help. The risk sits with tasks you already like.

What the overjustification effect is

The effect describes a shift in the perceived reason for acting. Before the reward, the reason is internal: I draw because I like drawing. After an expected reward, the reason looks external: I draw because I get a certificate. When the reward stops, the external reason is gone and the internal reason feels weaker than before. Economists call the same shift crowding out: the outside motive pushes the inside motive aside.

Three parts must all hold for the label to fit. The activity starts with real intrinsic interest. The reward is expected in advance, tangible, and tied to doing the task. Later interest is measured after the reward stops, with free choice and other options present. If any part is missing, the result has another name. Paying you to do what you dislike is simple incentive. Praising good work is feedback. A surprise treat after study is a gift.

The measure matters. Researchers do not ask only whether you liked the reward phase. They compare free-choice time before the reward with free-choice time after it ends, against a control group that never got the reward. Deci left puzzles and magazines in the room and timed puzzle use while the experimenter was away. Lepper observed classrooms from behind a one-way mirror and timed drawing during free play. Both designs test what remains when the pay stops.

The two studies that named it

Deci ran the first clear test in 1971 at Carnegie Mellon University. The paper appeared in the Journal of Personality and Social Psychology (Deci, 1971). The task was Soma, a spatial puzzle from Parker Brothers that most students found absorbing.

The design used three sessions on separate days. Each session had a break in the middle with an eight-minute free-choice period. The room held more puzzles and some magazines. The experimenter left, and observers timed puzzle use. In session one, nobody heard about pay. In session two, the experimental group got 1 dollar for each puzzle solved and the control group got nothing. In session three, both groups worked unpaid, with the paid group told that funds ran out. The paid group spent more free-choice time on puzzles in session two than in session one, by about 68 seconds on average. In session three they fell below their own baseline and below the controls. A second Deci experiment with verbal praise instead of money found no such drop. Praise that approved performance kept interest steady. A field test with headline writers at a student newspaper pointed the same way. The full 1971 text is freely hosted by the theory group (Deci, 1971, PDF).

Lepper, Greene and Nisbett ran the classroom version in 1973 with preschool children aged three to five. The paper appeared in the same journal (Lepper et al., 1973). Teachers first named children who already liked drawing during normal baseline observation. The team assigned them at random to three groups. The expected group agreed to draw in order to get a certificate with a gold seal and ribbon. The unexpected group drew the same way and got the same certificate but heard about it only after finishing. The no-reward group drew with no mention of any award. The full text is freely hosted by MIT (Lepper et al., 1973, PDF).

The team brought drawing back to the classrooms one to two weeks later and watched in secret. Children in the expected group spent a smaller share of free-play time with the markers than either other group. The overall test was significant at F equal to 3.25 with p below .05. The planned contrast of expected against the other two was significant at F equal to 6.19 with p below .025. Within the expected group, interest fell from baseline to follow-up with t equal to 2.61 and p below .02. The other groups showed small non-significant rises. Counts of pictures drawn in the session did not differ at 2.61, 2.44 and 2.33. Quality did differ. Blind judges rated expected-group pictures at 2.18 on a five-point scale. The other groups scored 2.85 for unexpected and 2.69 for no reward. The gaps were significant at t equal to 3.01 with p below .01 and t equal to 2.08 with p below .05. The promised reward lowered both later choice and same-day care.

A follow-up with school-age children used a gold-star good-player award and found the same hinge. Children told in advance about the award later spent less time on the task. Children surprised with the same award did not lose interest. Expectation, not the object itself, did the work.

Why it happens

Two theories explain the shift, and they agree on the practical point while they word the cause in different ways.

Self-perception theory says you learn what moves you by watching your own acts the way an outsider would. Psychologist Daryl Bem proposed it in 1972. When a clear outside reason stands next to your act, you credit the outside reason. You drew and you got a certificate, so drawing must have been for the certificate. The inference feels natural because inner motives are hard to see and outer rewards are easy to see. When the reward ends, the credited reason ends with it.

Cognitive evaluation theory says the same events change where you feel the cause sits. Edward Deci and Richard Ryan built it as part of self-determination theory. Each person tracks a perceived locus of causality, the felt address of the cause. Inner address means I choose this. Outer address means I am steered by pay, deadlines or watchers. A reward read as control moves the address outward and weakens autonomy, one of three basic needs alongside competence and relatedness. A reward read as useful news about skill supports competence and leaves autonomy intact. Money for showing up reads as control. Specific praise for accurate work reads as news. That is why one undermines and the other sustains.

Both theories predict the same boundary. Salient expected pay for a liked act moves the story from I want this to I do this for that. Remove the that and little story remains.

When it holds and when it does not

The effect is conditional. A 1999 meta-analysis of 128 studies by Deci, Koestner and Ryan found clear undermining for expected tangible rewards. It found a helpful effect for informational verbal rewards (Deci et al., 1999). A 2016 quantitative review of 65 single-case data sets found no mean undermining overall. It found larger undermining where baseline response was already high relative to rewarded response (Levy et al., 2016). The two reviews differ in samples and methods, but they point students to the same table. Use it to predict the result before you set a reward.

Students often mix up reward words. Pay for output is not the same as removal of an unpleasant state, and rules that set how often a reward arrives change the reading. Our guide to reinforcement schedules sorts those terms so the table above stays clear.

Four common mistakes follow from ignoring the table. The first is to ban all rewards while revising. Rewards still move dull setup work such as sorting cards, copying terms or clearing a backlog, where interest starts near zero. The second is to pay yourself per page for a topic you love. That is the exact cell where undermining peaks. The third is to reward mere showing up. Tokens for attendance teach attendance. Tokens tied to accurate recall teach recall. The fourth is to read a short-term lift as proof of safety. Interest during the paid phase usually rises. The test is the unpaid week after.

What it means for studying

Treat liked topics and dull chores with different rules. For a topic you already open for fun, keep it reward-free. Let the act stay its own reason. For dull setup with no pull, use a small defined reward to start, then drop it once the material itself carries you. Tie any reward to quality, not to mere completion. A rule such as no treat until you type five definitions from memory without notes beats a rule such as one sweet per page read. The first rewards accurate recall. The second rewards turning pages.

Keep choice visible. Self-chosen order, self-chosen lists and self-paced drills protect autonomy where imposed quotas harm it. Psychology World fits that slot in a small way. You save the motivation terms you meet into a named list, such as Exam 1 or Memory. Then you drill that list with flashcards or by typing answers until the terms hold. The drill is effortful, but nobody pays you per card, so the reason stays inside the act. A daily word notification brings you back without setting a price on each return.

Make praise informational. After a drill, note what was accurate and what needs one more pass. General praise such as good job carries little news. Specific notes such as three of five definitions exact, two confused with nearby terms carry news about competence and support the next pass. If you study with children, use the same split. Praise the careful drawing, not the sitting still. Keep promised prizes for tasks they avoid, not for play they already choose.

As of 2026, the evidence supports this short version for exams. Expected tangible pay for liked study lowers later free choice. Surprise treats, specific praise and quality-tied rewards do not show the same drop. Effects in labs run over minutes to days and in classrooms over one to two weeks. Claims about whole semesters go beyond what these studies test.

How it differs from nearby ideas

Bribery and overjustification point in opposite directions. A bribe pays you to do what you would not do alone. Overjustification pays you to do what you would do alone, then weakens the alone part. Crowding out is the economist label for the second case: outside money crowds inside motive aside.

Cognitive dissonance looks similar in one number and opposite in meaning, so students mix the two. In Festinger and Carlsmith's dollar study, students paid 1 dollar to call a dull task fun later rated it as more fun than students paid 20 dollars. Small pay gave no good outer reason, so belief moved inward to justify the act. In overjustification, expected pay gives a strong outer reason for a liked act, so motive moves outward. Later interest then falls. Small pay internalizes a dull act. Expected pay externalizes a liked act. The dollar amounts match by accident. The mechanisms run in reverse.