A recent expansion of hypothesis theory has resulted in a new interpretation of the partial-reinforcement extinction effect (PREE). According to this interpretation, the classical lOO%-reinforcement and 50%-reinforcement groups receive different solution-experiences on Problem 1. This, in turn, produces different expectations about the solution on Problem 2. Three experiments, using a nonreversal shift paradigm, were performed to test this view. The results were that, as hypothesized, a nonreversal shift was learned faster than a shift from a random-reinforcement condition.