To test the stimulus-aftereffect hypothesis and the informational hypothesis, human subjects received partial reinforcement in which a number of N-R transitions and the values of H2 and H3 were varied under the gambling task. The continuous reinforcement group was added for a control. After eleven acquisition trials, the PRE was demonstrated, but resistance to extinction was not an increasing function of both the number of N-R transitions and informational measures. However, resistance can be regarded as a function of N-length. The extinction data suggest that the stimulus-aftereffect interpretation is more persuasive than the discriptive theory of information, considering the mechanisms of the few-trial PRE.