This was also the topic of Dan Pink's TED talk, http://www.ted.com/talks/dan_pink_on_motivation.html http://www.ted.com/talks/dan_pink_on_motivation.html, which

This was also the topic of Dan Pink's TED talk, http://www.ted.com/talks/dan_pink_on_motivation.html http://www.ted.com/talks/dan_pink_on_motivation.html, which came across the RSS feed about 9 hours ago, http://news.ycombinator.com/item?id=782171 http://news.ycombinator.com/item?id=782171. In that presentation, the speaker talked about the candle experiment and some of its variations. He mentions that for purely mechanical skills, rewards indeed have been shown to repeatedly increase productivity. But for non-mechanical tasks, where the task is not immediately solvable or requires calculation, the promise of a reward to the top performer often diminishes the time in which the task is completed. He didn't mention any deeper analysis though. For instance, one can wonder about the difference in the attitude between the top performers in the reward and non-reward groups. Does a participant get pre-occupied with the fact that they have not solved the problem first, and thereby reduce the amount of attention they are giving to the problem at hand? Do they have the same amount of concentration throughout the task, or after what they perceive to be a "deadline" do they stop caring. A major difference between Pink's talk and this article is that this posits that the previous questions don't matter. Instead, it hints at the fact that once any reward is offered, the work done suffers because it diminishes one's intrinsic motivation. I wonder if this can be tested against the hypothesis that the reward becomes an object measuring stick to the quality of work; while another person's respect or praise might not be as easily objectified and measured motivating one to do more work.