Reward magnitude and learning in a single-presentation discrimination.

Abstract

The purpose of this experiment was to study the effect of amount of reward on performance in a single-presentation discrimination. Rats were reinforced for running down a gray alley to a black goalbox (S) but were not reinforced in a white goalbox (S~). Half the ,5s received a large reward, and half received a small reward. Generalizing from single-runway studies (Crespi, 1943), one would expect response speed in S+ to be faster with the larger reward. However, there is no evidence on whether magnitude differences in S+ influence terminal performance in S~, and one interest of this study was to answer this question. Spence (1956) has proposed the hypothesis that magnitude of reward affects a motivational factor (incentive or rg) that is learned and must be elicited by discriminative stimuli to facilitate performance. According to this hypothesis, differences in amount of reward in S should not influence terminal performance in the nonreinforced stimulus, S~. This prediction contrasts with a comparable prediction for variations in another motivational variable, drive (i.e., hours of deprivation), which is independent of a conditioning histoiy; increases in hours of deprivation would be expected to increase terminal performance in S, but two experiments reported by Spence (1958) give conflicting evidence on this point. A second question of interest was whether performance in S+ would improve if S~ trials were eliminated after extensive discrimination training. It is known, at least for classical defense conditioning (Gynther, 1957), that performance of 5s given simple conditioning is superior to that of 5s given discrimination training. The present study, using a. locomotor response, compared these two conditions within the same 5s. Following discrimination training, 5s were shifted to all S trials, and