The puzzle was called Soma: a set of odd three-dimensional pieces that you fit together to match figures on a sheet, some easy, some hard. Edward Deci, a psychologist still working on his doctorate, wanted a task people find interesting for its own sake. This was it.
He brought students in for three sessions, at least a day apart. In the middle of the first session, the experimenter left the room for eight minutes, saying more material was needed. There were magazines on the table, and more puzzles. The students could do whatever they liked. They did not know they were being watched, or that the seconds they spent on the puzzles were being counted.
In the second session, one group was offered $1 for every puzzle they solved. In the third, those same students were told the budget had run out. Nobody would be paid this time.
Then the experimenter stepped out again, for another eight minutes. The students who had been paid spent less of their eight minutes on the puzzles than they had in the very first session. Those who had never been paid spent a little more.
What the dollar bought
The money hadn’t made the puzzle any less clever. In Deci’s reading, it had changed the answer to a quieter question: why am I doing this? In the first session, the answer was “because it’s interesting.” After a session of being paid per puzzle, the answer had become “because they pay me.” When the pay stopped, so did the reason.
Deci’s own experiments showed it wasn’t every reward. In a third study, students who were told “You did very well” kept their interest. In later work, people paid just for showing up, rather than per puzzle, didn’t lose interest either. What seemed to do the damage was a reward that felt like it was steering the person, a dollar handed over for each piece of behavior the experimenter wanted.
Picture a hobby that someone starts paying you for by the piece. While the money comes in, you keep going. When it stops, the hobby can feel like unpaid work.
What people need
Deci and his colleague Richard Ryan spent the next decades building those findings into a theory. They named it in a 1985 book, Intrinsic Motivation and Self-Determination in Human Behavior. By 2000 they had boiled it down to three psychological needs. Competence is the sense of being effective and growing in what you do. Autonomy is the sense that what you do comes from you. Relatedness is the sense of being connected to people who care about you in return.
Autonomy here doesn’t mean working alone. Ryan and Deci define it as a feeling of volition that can accompany any act, whether you do it by yourself or for a group.
Pay can make people do a task. Choice, skill and belonging make them want to.
In their account, money isn’t the enemy. What undermines interest is a reward experienced as control, one that turns the dollar into the only answer to “why.”
The dull task
Most work is not a Soma puzzle. Say a team lead has to get a junior analyst to recheck two thousand old invoices before an audit. Nobody will find it interesting, the analyst included.
She could set a Monday deadline and tell him she’ll review every row. He will probably do it. She could also tell him what the auditors are looking for and why a missed invoice costs the company, admit the job is tedious, and let him decide how to split it up and in what order.
SDT doesn’t promise that the second version makes rechecking invoices fun. Its claim is narrower. In a 1994 laboratory study by Deci and colleagues, people given a meaningful reason for an uninteresting task, along with support for their autonomy and for feeling related, took the task on as their own more fully. Under controlling conditions there was less of that, and what there was tended to be the kind people keep up to avoid guilt or anxiety, taken in without really being accepted.
If the theory holds, the difference shows up at the next audit, when nobody is checking every row.
Small rooms and big claims
The 1971 study was small. A 2019 review written by Ryan and two co-authors says the first experiment had only 24 participants and its result was only marginally significant. The design had other weaknesses too.
The finding didn’t rest there. A 1999 meta-analysis by Deci, Richard Koestner and Ryan pooled 128 studies, almost all laboratory experiments with tasks people found interesting. When people expected a tangible reward for doing, finishing or doing well at the task, they later spent less free time on it. Rewards that came as a surprise, or weren’t tied to the task, showed no such effect, and positive feedback raised interest. For dull tasks, the authors found no significant effect of rewards either way. It was a fight as much as a summary. A 1996 meta-analysis had called the undermining effect a “myth,” and the 1999 paper, by Ryan’s account, documented the miscounted and miscategorized entries behind that verdict.
Putting the theory to work is harder than measuring it. Deci, Ryan and James Connell trained managers in the service division of a large office-machine company, in a project funded by Xerox, to support their technicians’ autonomy. The managers’ approach clearly changed. The effect on the technicians was, in the authors’ words, “less conclusive.”
Economists found a cousin of the puzzle result in Israel. Ten day-care centers in Haifa took part in a study, and six of them began fining parents who arrived more than ten minutes late. Late pickups rose significantly. (More on that pattern under Motivation Crowding Out.)
When the fine was dropped, late pickups did not return to where they had been.