Effects of inconsistent reinforcement on reversal and nonreversal shifts.

Abstract

3 EXPERIMENTS WERE PERFORMED TO INVESTIGATE WHY INCONSISTENT REINFORCEMENT DURING DISCRIMINATION LEARNING HINDERS REVERSAL. IN EACH OF 2 EXPERIMENTS, RATS TRAINED ON A 75:25 BRIGHTNESS DISCRIMINATION REACHED A CHOICE CRITERION OF EXTINCTION MARGINALLY MORE RAPIDLY THAN SS TRAINED ON A 100:0 PROBLEM, BUT SUBSEQUENTLY REVERSED MORE SLOWLY BECAUSE THEY FORMED STRONGER POSITION HABITS. THIS IMPLIES THAT INCONSISTENTLY REINFORCED SS REVERSE SLOWLY NOT BECAUSE THEY EXTINGUISH SLOWLY, BUT BECAUSE THEY FAIL TO LEARN ADEQUATELY TO ATTEND TO THE CUE RELEVANT BOTH DURING ORIGINAL TRAINING AND REVERSAL. EXP. III SUPPORTED THIS IDEA BY SHOWING THAT INCONSISTENT REINFORCEMENT RESULTS IN MORE RAPID LEARNING OF A NONREVERSAL SHIFT PROBLEM IN WHICH THE ORIGINALLY RELEVANT CUE BECOMES IRRELEVANT. (PsycINFO Database Record (c) 2006 APA, all rights reserved)