Reversal learning in conditional discriminations is not controlled by reinforcer density

Three experiments have examined whether a whole-partial reversal effect is due to shift in reinforcement density across phases, between whole and part ial reversal in both matching (or nonmatching) -to-sample discriminations u sing 12 different stimulus sets (Experiments 1 and 2) and three concurrent discriminations (Experiment 3). In Experiments 1 and 2, rats were trained o n nonmatching (or matching) -to-sample discriminations and then either give n reversal training on 12 stimulus sets (W), on 9 out of them (P-9), on 6 o ut of them (P-6), or 3 out of them (P-3). Group W reversed faster than the other three partial groups. Group P-3 reversed faster than Group P-6, which in turn reversed faster than Group P-9. In Experiment 3, rats were concurr ently trained on three discrimination tasks and then either given reversal on a total of three discrimination tasks (W), on two out of them (P-2), or on one out of them (P-l). Group W reversed faster than Group P-l, which in turn reversed faster than Group P-2 after overtraining. These findings prov ide evidence that rats form functional classes, and that the whole-partial reversal effect is not due to shift in reinforcement density.