The purpose of this study was to examine the Self Evaluative Motivation (SEM) Model proposed by Sakurai (1984a), using four kinds of extrinsic rewards: verbal, token, verbal-token, and only feedback from outcomes. The control group received no rewards nor feedback. According to the SEM model, the experimental and the control groups should select the following tasks after getting each reward mentioned above. Four experimental groups should more difficult tasks than the control group because they have higher feeling of competence. The token, the verbal -I- token, and the control groups should select more dissimilar tasks than the verbal and the feedback groups because the former groups have little feeling of self-determination. After exercises, subjects were assigned lucky-puzzle tasks and given each extrinsic rewards or feedback. Then they were asked which task they wanted to try among four kinds of tasks. Most results supported the present predictions.