New research out of the Yale School of Management is the first to find that humans have a poor intuition of how incentives motivate performance.

In the study, researchers found that subjects performed better in a letter-sorting task when wrong answers meant losing a reward already granted than when a positive answer meant gaining a corresponding reward. In contrast, subjects in the study thought positive incentives would be more motivating than the negative one. The finding has implications for helping both business managers and individuals understand incentives and motivation.

“There is so little work in how people’s predictions of what motivates them may be [the] opposite of what actually motivates them. It wasn’t clear what to expect,” said Yale School of Management professor and study co-author Ravi Dhar in an email to the News. “I would say it was exciting to find the difference.”

In the positively framed condition of the study, subjects received money for successfully unscrambling letters. In the negatively framed condition, subjects started with the maximum possible amount of money from the positive condition and lost money for each unsolved anagram. The study design ensured that a given level of performance was equally compensated in both conditions.

As the researchers expected, subjects spent more time working on the anagrams in the negative condition, signaling that the prospect of losing money they already had encouraged subjects to work harder. Researchers then interviewed a different set of undergraduate students about whether positive or negative framing would be more effective. Contrary to the experimental results, these subjects thought positive framing would be more effective.

“I was a little surprised that people thought they would be more motivated by positive framing,” said Yale School of Management professor Nathan Novemsky. “I hope this will stimulate more research in the area.”

The study also found that the effect disappeared in the oldest group of the study, with positive and negative conditions being roughly equally reinforcing. Dhar said this finding shows the importance of considering how age differences may drive findings in psychological research.

The study was published on Sept. 23 in the Journal of Experimental Psychology: Applied.