It was considered that the reward would bring about a better performance, and indirectly facilitate learning; however, all given the reward either pre or post the performance would make no significant difference between them. And it was found that the pre-reward group seemed better than the post-reward group in comparing with the non-reward group in speed and accurateness. This proved that the former group was stronger than the latter in incentive motivation. The difference between the reward and non-reward all depended on the kind of perfoamance. If the material for performance was a kind of association and contiguity, then there would be no differences among them. This was found in the test of word paired number association. However if the material was not a kind of association like accurateness of motor performance, then significant difference was found among the pre-reward and post-reward and the non-reward groups. This proved that accurateness was, greatly affected by emotion and incentive motivation. As far as the speed was concerned, it seemed there was a kind of learning between contiguity and reinforcement. Furthermore, it was found association and speed would be improved by practice significantly; however, accurateness was not so, the errors of non-reward group were increasing (p＜O.O1). Therefore in the theory of Spence’ excitatory potential [E=H(D+K)] under the assigned job, the effect of reward had to be considered with the kind of learning that was performed.