Extinction as a function of partial reinforcement and distribution of practice.

Abstract

72 albino rats were trained to run down an alley to food. Half received reinforcements on all training trials, and half randomly on 50% of the trials. Half of each group were trained with a 15-sec. interval between trials and half with a 15-min. interval. In each of the 4 training groups, the responses of half of the subjects were extinguished with 15-sec. interval and half with the 15-min. interval. After massed training, resistance to extinction was significantly greater for 50% reinforcement groups than for 100% reinforcement groups; but after spaced training, no difference was found between the 2. These results verify a prediction that the increased resistance to extinction generally found with partial-reinforcement technic results from training in performing the conditioned response in the presence of cues for extinction, namely, the changed stimulus pattern resulting on a given trial when reinforcement is omitted on the preceding trial. The hypothesis of derivation follows from a more general hypothesis which explains part of the decrement characteristic of all extinction as due to generalization to a changed stimulus pattern resulting when reinforcement is omitted.