THE ERROR FACTORS WHICH OPERATE DURING THE DISCRIMINATION LEARNING OF NORMAL CHILDREN WERE STUDIED. THREE EXPERIMENTS WERE CONDUCTED TO COMPARE RESPONSE-SHIFT AND RESPONSE-PRESERVERATION TENDENCIES OF DIFFERENT AGE GROUPS. THE FIRST EXPERIMENT INCLUDED 4-, 6-, 8-, AND 10-YEAR-OLDS WHO WERE GIVEN REWARD AND NONREWARD PROBLEMS. THE SECOND EXPERIMENT INCLUDED 4- AND 3-YEAR-OLDS WHO WERE GIVEN EITHER 75 PERCENT REWARD AND 25 PERCENT NONREWARD PROBLEMS, OR THE REVERSE. THE THIRD EXPERIMENT INCLUDED 4-, 6-, 8-, AND 10-YEAR-OLDS WHO WERE GIVEN REWARD PROBLEMS ONLY. A LEARNING SET PROCEDURE OF TWO-TRIAL PROBLEMS AND MULTIDIMENSIONAL STIMULI WAS EMPLOYED IN THREE TREATMENTS. BOTH RESPONSE-SHIFT AND RESPONSE-PERSEVERATION TENDENCIES APPEAR TO BE IMPORTANT SOURCES OF ERROR IN CHILDREN'S DISCRIMINATION LEARNING. (RS)