Information and the Partial Reinforcement Acquisition Effect

Abstract

Three groups of rats were trained on a delayed reinforcement schedule in an L-shaped runway. The running response, completed just before the turn of the runway, initiated the delay and exposed the rats to one of two delay stimuli. For two of the groups partial reward (50%) followed the delay. For one of these groups the delay stimuli and trial outcomes were correlated, making the delay stimuli informative with respect to the trial outcome some seconds before the outcome occurred. For the other group the delay stimuli and outcomes were uncorrelated, so information was not delivered until the occurrence of the trial outcome. The third group was reinforced on every trial (CRF), so no information followed responding at all. The results were that the partial reinforcement animals trained with correlated stimuli ran most quickly, followed by the partial reinforcement animals trained with uncorrelated stimuli, and then by the animals that were reinforced on every trial. The results were explained by the hypothesis that stimuli antedating the goal are increased in reinforcing strength when they transmit information, with the increase being proportionally greater the further the stimuli are from the goal.