TO STUDY THE EFFECTS OF OVERTRAINING ON REVERSAL AND NONREVERSAL SHIFTS OF CUES, 96 NORMAL (MEAN AGE = 117 MONTHS, MEAN IQ = 109.8) AND 96 RETARDED CHILDREN (MEAN AGE = 119 MONTHS, MEAN IQ = 70.1) WERE TESTED ON A SIMULTANEOUS 2-CHOICE DISCRIMINATION TASK. SUBJECTS WERE TRAINED ON TWO LEVELS (CRITERION AND OVERTRAINING). FOLLOWING TRAINING ON THE ORIGINAL TASKS WITH SHAPE, HEIGHT, AND BRIGHTNESS AS DISCRIMINANDA, THE SUBJECTS WERE TRANSFERRED TO A REVERSAL SHIFT (RS), NONREVERSAL SHIFT-IRRELEVANT DIMENSION (NRS-ID), OR NONREVERSAL SHIFT-NEW DIMENSION (NRA-ND) TASKS. SMALL TANGIBLE REWARDS WERE OFFERED. SIXTEEN SUBJECTS WERE TRAINED IN EACH OF THE 12 EXPERIMENTAL CONDITIONS. RESULTS IN THE CRITICAL TEST CONDITIONS INDICATED (1) A SIGNIFICANT SHIFT EFFECT (P IS LESS THAN .001), INDICATING THE RELATIVELY GREATER DIFFICULTY OF RS AND NRS-ID PROBLEMS, (2) A SIGNIFICANT TRAINING EFFECT (P IS LESS THAN .001) FROM OVERTRAINING FOR BOTH NORMAL AND RETARDED SUBJECTS IN ALL TRANSFER TASKS EXCEPT FOR NORMAL SUBJECTS ON NRS-ID PROBLEMS, AND (3) NO RELIABLE LEARNER OR LEARNER SHIFT INTERACTION EFFECT (P FOR BOTH IS LESS THAN .10). THE STUDY THEREFORE SUGGESTED THAT THERE MAY NOT BE A FUNDAMENTAL DISCONTINUITY BETWEEN THE DISCRIMINATION PERFORMANCE OF NORMAL AND RETARDED SUBJECTS. THE NEED FOR FURTHER RESEARCH INTO THE REASONS FOR THE FACILITATING EFFECTS OF OVERTRAINING WAS INDICATED. A SECOND STUDY INTRODUCED VERBALIZATION IN AN ATTEMPT TO INVESTIGATE ITS EFFECT UPON TRANSFER PERFORMANCE ON RS, NRS-ND, AND NRS-ID PROBLEMS OF 48 RETARDED SUBJECTS. RESULTS INDICATED THAT, WHILE VERBALIZATION SIGNIFICANTLY FACILITATED LEARNING THE ORIGINAL TASK (P IS LESS THAN .001), IT HAD A LIMITED FACILITATING EFFECT UPON TRANSFER TASKS. THREE TABLES PRESENT DATA, AND A BILBIOGRAPHY LISTS 11 ITEMS. (DF)