“The use of rewards—what might be called positive coercion—does not work in the long run any better than threat and punishment, or negative coercion. In the reward, the child senses the parent’s desire to control no less than in the punishment. The issue is the child’s sense of being forced, not the manner in which the force is applied. This was well illustrated in a classic study using magic markers.2 A number of children were screened to select some who showed a natural interest and inclination for playing with magic markers. Those who did were then divided into three different groups. For one group, there was no reward involved and no indication what to do with the markers. Another group was given a small reward to use the markers, and the third was promised a substantial reward. When retested sometime later, the group that had been most rewarded showed the least interest in playing with the magic markers, while the children who had been left uninstructed showed by far the greatest motivation to use them. Simple behaviorist principles would suggest it ought to have been the other way around, another illustration that behavioral approaches have no more than short-term efficacy. At work here, of course, was residual counterwill in response to positive coercion. In a similar experiment, the psychologist Edward Deci observed the behaviors of two groups of college students vis-à-vis a puzzle game they had originally all been equally intrigued by. One group was to receive a monetary reward each time a puzzle was solved; the other was given no external incentive. Once the payments stopped, the paid group proved far more likely to abandon the game than their unpaid counterparts. “Rewards may increase the likelihood of behaviors,” Dr. Deci remarks, “but only so long as the rewards keep coming... Stop the pay, stop the play.” We”