128 HOODED RATS WERE TRAINED IN A BLACK-WHITE DISCRIMINATION TASK IN A Y MAZE ACCORDING TO A 2 * 2 * 2 * 2 FACTORIAL DESIGN. LARGE OR SMALL REWARD, CORRECTION OR NONCORRECTION PROCEDURE, DISTINCTIVE OR NONDISTINCTIVE TRIALS WERE COMBINED IN THE 8 POSSIBLE COMBINATIONS DURING ACQUISITION. FOR 1/2 THE SS IN EACH GROUP REVERSAL LEARNING BEGAN UPON REACHING CRITERION, FOR THE OTHER 1/2 200 ADDITIONAL TRIALS WERE GIVEN BEFORE REVERSAL. OVERLEARNING FACILITATED REVERSAL LEARNING WHEN REWARD WAS LARGE, BUT NOT WHEN REWARD WAS SMALL, AND THE OVERLEARNING REVERSAL EFFECT (ORE) WAS MORE LIKELY TO OCCUR UNDER CORRECTION THAN UNDER NONCORRECTION PROCEDURES. DISTINCTIVENESS AS MANIPULATED IN THIS EXPERIMENT, DID NOT HAVE A LARGE DIFFERENTIAL EFFECT ON THE ORE. OVERLEARNED SS IN BOTH CONDITIONS PERFORMED BETTER DURING REVERSAL THAN DID MASTERY SS. THE RESULTS FOR REWARD AND TRAINING PROCEDURE WERE DISCUSSED WITH REFERENCE TO FRUSTRATION THEORY. (19 REF.) (PsycINFO Database Record (c) 2006 APA, all rights reserved)