Do Rewards Kill Motivation?
In 1969, Edward Deci paid students a dollar per puzzle solution, then stopped the money and watched through a one-way window as their interest died. His field experiment paid newspaper headline writers fifty cents per headline — until the fund ran out. Across 128 studies, expected tangible rewards undermined later interest while positive feedback strengthened it. The rebuttal is here too: the damage happens under restricted, avoidable conditions. The difference between a reward that feeds motivation and one that taxes it is not the prize. It is the bargain.
By MyAudioBooks.ai ·
This free listen has rotated out. The story remains available to read.
The experimenter left the room and told the student he would be gone only a few minutes: "I shall be gone only a few minutes, you may do whatever you like while I am gone." On the table sat a puzzle called Soma, seven plastic pieces that could be assembled into millions of configurations. There were magazines to read. There was a window to stare out of. And behind a one-way window, in a laboratory at the University of Rochester in nineteen sixty-nine, Edward Deci stood with a timer, watching what the student chose to do when nobody was asking.
Some students had been paid a dollar for every configuration they completed in the previous session. Others had never been paid a cent. The money, the paid students had been told, was now exhausted. Our verdict is that this eight-minute observation is the seed of one of the most useful and most abused findings in psychology: rewards can kill motivation, but only certain rewards, delivered in a certain spirit. The pay stops, the play stops — if the pay was ever running the play.
At My Audio Books dot A I, you can create your own audiobooks from prompts, turn your documents into audio, all with one subscription, and store your items in your own personal library.
The Dollar and the Puzzle
The elegance of Deci's nineteen seventy-one design is that it measures motivation where nobody can perform it for an audience. Across three sessions, students worked with the Soma puzzle, a task chosen because college students found it genuinely enjoyable — in the paper's own ratings, the sessions averaged between seven point two five and eight on a nine-point enjoyment scale, with no significant differences between the groups. Everyone started on the same footing: this was a thing people did for fun.
In the second session, the experimental group was offered a dollar for each configuration reproduced within a thirteen-minute limit. The control group received the same configurations with no pay. Then, in the third session, the payments stopped. The paid students were told there had only been enough money for one session, which is a beautiful detail, because it means the experiment reproduced the exact arc of every bonus scheme ever devised: the work begins for its own sake, the money arrives, the money leaves.
The measurement was the eight minutes after each session, when the student believed the experimenter had stepped out to the teletype. The criterion for working was strict — manipulating and looking at the pieces, not merely holding one while daydreaming. And the configurations on the table were impossible to complete, so no student could confound the result by finishing a puzzle and stopping satisfied. The only thing left to measure was whether the person still wanted to play. The paper's summary of the outcome across its experiments: "when money was used as an external reward, intrinsic motivation tended to decrease."
Sit with the perversity of that. The activity did not change. The puzzle was the same seven pieces of plastic. The student's ability had not changed, and neither had the number of interesting configurations. The only new ingredient was a dollar, and the dollar had somehow made the puzzle less worth doing. If you have ever watched a beloved hobby curdle the moment it became a side hustle, you already know the feeling. Deci gave it a laboratory and a timer.
The Question Behaviorism Actually Answered
To see why this was such a subversive result, you have to remember who owned the motivational machinery of the era. The nineteen ninety-nine meta-analysis opens its own backstory with a generous admission: by nineteen seventy-one, hundreds of studies in the operant tradition of B. F. Skinner had established that extrinsic rewards can control behavior. Given right after a behavior, rewards reliably made that behavior more likely, and the effect persisted for as long as the reward contingency stayed in place. That is not a small finding. It is the entire technology of reinforcement, and it works.
But look at the dependent variable, as the researchers say — look at what was being measured. The operant finding is about what you do while the contingency is operative. When the rewards were terminated, the behavior eventually returned to its pre-reward baseline. So the behaviorists had answered the question how do I get you to do this now, and their answer was correct. Deci was asking a different question: what happens to your relationship with the activity once the apparatus is removed? One question is about performance under a contingency. The other is about interest after it. The whole undermining literature lives in the gap between those two questions, and most of the confusion in the popular versions comes from answering the first and assuming you have answered the second.
This is also why the experiment had to be built like a trap with the door left open. If Deci had simply measured output during the paid session, he would have confirmed the behaviorists: pay people and you get more configurations. The interesting measurement was the eight minutes when the session was over and the student thought nobody was watching. That is where motivation lives when it is not being rented.
The Headline Writers
A laboratory puzzle is one thing. So Deci took the same design into a job, and the field study is the part of the paper that should be required reading for every manager who has ever proposed a bonus. The setting was a biweekly college newspaper. Eight students wrote headlines on two four-person staffs, one for the Tuesday edition and one for the Friday. For five months, the students had no idea they were in an experiment at all.
The Tuesday staff became the paid group. Their payment was fifty cents per headline, introduced with a cover story about a budget fund that had to be used up by the end of the semester, and a warning not to tell the other staffs in case it caused animosity. After three weeks, they were told the money was exhausted and the payments ended. The Friday staff worked the same task with no payments at all. And through the whole thing, a man they believed to be a fellow staff member sat in the room as the supervisor, surreptitiously recording how long each headline took.
That is a genuinely extraordinary piece of fieldwork. The measure of motivation was the speed of the work itself, with absences tracked as a measure of attitude, and a follow-up period five weeks into the next semester to see whether any effect survived. The paper's conclusion joins the lab's: money as an external reward tended to decrease intrinsic motivation, and the framework had to explain why. Think about what the payment had quietly changed. Headline writing was once something these students did for the paper, for the craft, for each other. The fifty cents reclassified it. Now it was a transaction — and when the transaction ended, the reason to do the work had to be rebuilt from scratch.
The mechanism Deci reached for came from his colleague Richard deCharms: when an external reward is introduced for an activity you already loved, your sense of personal causation shifts outward. You were the origin of the behavior; now you are, in deCharms' memorable phrase, "a pawn" to the source of the reward. Psychologists call this the locus of control — whether you experience the cause of your own behavior as living inside you or outside you. The reward does not merely pay you. It reassigns authorship of the act. And an activity you no longer author is an activity you no longer own.
One Hundred Twenty-Eight Studies
By the late nineteen nineties, this was no longer a handful of experiments; it was a literature with a war inside it. So in nineteen ninety-nine, Deci, Richard Koestner, and Richard Ryan published a meta-analysis of one hundred twenty-eight studies in Psychological Bulletin, and its pattern is far more interesting than the bumper sticker it inspired.
The effects split by the structure of the reward. Rewards contingent on engaging in the task, on completing it, or on performing it to a standard all significantly undermined later voluntary engagement, with effect sizes of minus zero point four zero, minus zero point three six, and minus zero point two eight respectively. Engagement-contingent and completion-contingent rewards also significantly undermined self-reported interest. The pattern held for tangible rewards and for expected rewards overall. In other words, the classic arrangement — you can expect a concrete payoff if you do this thing — reliably damaged the thing it was meant to encourage.
And then the finding that should be printed next to every performance-review template: "Positive feedback enhanced both free-choice behavior (d = 0.33) and self-reported interest (d = 0.31)." Positive feedback moved motivation up, in the same analysis that found tangible expected rewards pushing it down. The 1971 paper had seen the same asymmetry at small scale: "when verbal reinforcement and positive feedback were used, intrinsic motivation tended to increase." Praise and payment are not two versions of one thing called reward. In the data, they point in opposite directions.
There are two more results in that abstract that complicate the slogan usefully. Tangible rewards tended to be more detrimental for children than for college students — the people we most often bribe with sticker charts are the people for whom the bribe is most corrosive. And verbal rewards tended to be less enhancing for children than for college students, which means you cannot simply swap candy for compliments and assume the problem is solved. The developmental lesson is uncomfortable: the younger the person, the more carefully you should handle the whole apparatus of evaluation.
At My Audio Books dot A I, you can listen to this story and thousands of others that explore the hidden science and mechanics behind the headlines.
The Magic Markers
The cleanest demonstration of the whole argument involves no money at all, and it took place at the Bing Nursery School at Stanford, in work published by Mark Lepper, David Greene, and Richard Nisbett in nineteen seventy-three. The researchers first watched the classrooms and identified preschool children who already loved drawing — children who chose the activity on their own during ordinary free play. Only then did the experiment begin, because you cannot study the undermining of intrinsic interest in children who never had any.
Each child came individually to a room with a child-sized table, a set of magic markers, and a sheaf of paper. For one group, the researchers produced a Good Player Award — a colored card with a gold star and a red ribbon, spaces for the child's name and school — and the child agreed in advance to do the drawing in order to win it. For the second group, the same award existed but came as a complete surprise after the drawing was done. The third group neither expected nor received anything. The reward was identical in the first two groups. The only difference was whether it was a bargain beforehand or a gift afterward.
The result supported the prediction precisely: the children who had drawn in order to obtain the award showed less subsequent intrinsic interest in drawing than the children in either of the other two conditions. Not less than the rewarded children — less than both other groups. The surprise award did not carry the same cost. And the authors underline how little is required to produce the effect: explicitly contracting to do an activity for a reward can undermine interest "even when the reward is insubstantial or merely symbolic." A gold star on a three-by-five card is enough to turn play into a job.
Hold this study next to the meta-analysis's finding that tangible rewards were more detrimental for children, and the managerial version of the argument writes itself. The bargain is the problem, not the prize. A child can receive a certificate and flourish, and a child can sign up for the same certificate and learn that drawing is something you do to get certificates. The object did not change. The authorship did.
The Rebuttal
If this were a different kind of book, we would stop here with a clean morality tale about the evils of money. But the same literature contains a serious counterattack, and intellectual honesty requires putting it at full strength. In nineteen ninety-six, Robert Eisenberger and Judy Cameron published a review in the American Psychologist with a title that is itself a challenge: Detrimental Effects of Reward, Reality or Myth? Their conclusion is a direct rebuke to the undermining consensus: "detrimental effects of reward occur under highly restricted, easily avoidable conditions."
Read that as a scientist rather than a partisan and it becomes more useful, not less. Eisenberger and Cameron argue that the anti-reward literature had been overgeneralized into textbooks and popular media, and that when you sort the experiments properly, the reliable damage is much narrower than advertised. In their analysis, "Only expected tangible rewards are seen to have a reliable decremental effect on task duration." Unexpected rewards do not show the reliable decrement. And when they pushed one level further, into whether the reward depended on quality, completion, or mere performance, the reliable decrement on performance appeared only with performance-independent tangible reward — payment for simply doing the thing, regardless of how well.
Notice what has happened. The rebuttal does not rescue the year-end bonus for work you already love. It narrows the indictment to nearly the same defendant the original work accused. Both sides are now pointing at the expected, tangible, just-for-doing-it reward — the bribe you knew was coming, for an activity you might otherwise have enjoyed. The behaviorists add a genuinely constructive point of their own: rewards can also be used to build generalized creativity, if you design them around conditioning principles rather than slogans. Our read is that the fight is real but the practical overlap is larger than either camp's fans admit. Nobody serious is defending the controlling bribe. The argument is about how much else gets caught in the blast radius.
The Rewards That Don't Backfire
So the useful question is not whether rewards kill motivation but which rewards do, and the literature offers a working taxonomy. Expected and tangible is the danger zone: the promised cash, the guaranteed prize, the if-you-do-this-you-get-that arrangement that turns an activity into a means. The expectedness matters because it lets the reward move into the decision itself. You are no longer doing the puzzle; you are doing the dollar. And when the dollar leaves, it takes the reason with it.
Contingency is the second dimension. Payment for engaging, for completing, for performing — all three showed undermining effects in the meta-analysis, which means you do not escape the problem simply by making the reward contingent on quality. A performance-contingent prize is still a lever, and a lever still reassigns authorship. This is where the popular advice to just pay for performance runs into the evidence: performance-contingent rewards were among the significant underminers, not an exemption from them.
But the same analysis refuses to be a blunt instrument, and this is the part to read carefully if you design incentives for a living. The authors explain that every reward carries two messages at once: a controlling message and a competence message. Engagement-contingent rewards are almost pure control. Performance-contingent rewards are different, because when a person does well enough to earn a level of reward that signifies excellent performance, the reward also delivers substantial positive competence information — and the authors note that this can significantly affirm competence and offset some of the negative effects. So the same bonus can read as a leash or as a verdict on your mastery, and which reading dominates depends on how it is framed and what it actually says about your work.
The safer categories are nearly the photographic negative. Positive feedback and verbal praise enhanced motivation in both the original experiments and the meta-analysis. Unexpected tangible rewards did not show the reliable decrement in the rebuttal's analysis. The common thread is that these rewards arrive as information rather than as a leash. They tell you that you are competent, that your work landed, that someone noticed. They do not tell you what to do next. We think that is the cleanest way to hold the distinction: a reward that speaks about your ability tends to feed motivation; a reward that negotiates for your behavior tends to tax it.
And there is a boundary condition the slogan never mentions: none of this concerns tasks you never wanted to do. The entire undermining literature is about intrinsic motivation — interest that already exists. You cannot undermine what is not there. For the tedious, the necessary, and the unloved, external structure is not corrupting a pure motive; it is supplying the only motive in the room. The mistake is not using rewards. It is using them indiscriminately, on the activities that were carrying themselves.
How to Pay People
For a manager, the translation is uncomfortable but actionable. If your team loves the work, do not hang a meter on the parts they love. The moment you attach an expected per-unit payment to the craft itself, you have changed what the craft is for, and the Deci design tells you what happens when the meter is switched off. Use compensation to make the unloved parts possible — the administration, the compliance, the graveyard shift — and use evaluation to deliver information about mastery rather than to ration approval by the unit.
For a parent or a teacher, the child result is the one to sit with. Tangible rewards were more detrimental for children, and verbal rewards were less enhancing for them, which is the literature's way of saying that a young person's motivational system is easier to disrupt and harder to repair with slogans. So praise the strategy and the effort when you give feedback, and be very slow to put a standing price on reading, drawing, or practicing. The point of the activity was supposed to be the activity. Once it becomes the sticker, you will have to keep supplying stickers.
For yourself, the protocol is to notice the moment the ledger appears. A hobby that starts invoicing you is a hobby at risk. If you monetize something you love, do it with your eyes open: keep a part of the practice that is never for sale, never measured, never delivered. Protect a version of the puzzle that no one pays you to solve, because the unpaid version is the one that will still be there when the market loses interest.
And when you must reward, make it a message and not a bargain. Specific, unexpected, about the work: that is the pattern the evidence keeps pointing to. Before you introduce any incentive, run the experiment mentally to its third session. Ask what happens to the behavior on the day the reward ends, because that day always comes. If the honest answer is that the behavior ends with it, then you were never building motivation. You were renting compliance, and the rent will go up. The pay stops, the play stops — but only when the pay was ever the point of the play. Keep the reason inside the person, and the rewards can stay where they belong: on the sidelines, applauding.
At My Audio Books dot A I, you can create fiction, non-fiction, and turn your documents into audio, all stored in one place with a single subscription — plus get instant access to thousands of audiobooks and deep-dive investigations. Learn more today at My Audio Books dot A I.