The Effect of Partially Delayed Reinforcement and Trial-Distribution on the Extinction of an Instrumental Response

Abstract

In 1951, Crum, Brown, and Bitterman reported an investigation which has not received sufficient attenion.l Rats were trained to traverse a runway, one group being rewarded immediately on every trial, while a second group was rewarded immediately on half the trials and after a delay of 30 sec. on the remaining half (partial delay of reinforcement). In subsequent tests, the latter group was found to extinguish less rapidly than the former; that is, partial delay of reinforcement increased resistance to extinction. This outcome appeared to contradict the HullSheffield interpretation of partial reinforcement in terms of the carryover the aftereffects of reinforcement from trial to trial.2 Since both groups of rats had been rewarded on every trial, it was argued, the difference in rate of extinction must have been due to events taking place within the training trials.3 The primary purpose of the present experiment was to replicate the finding of Crum, Brown, and Bitterman, which has the same far-reaching implications for behavior-theory as the paradox of partial reinforcement itself. A second purpose was to study the effect of distribution of practice, a variable which has led to inconsistent results in work on partial reinforcement.4