Role of magnitude of reinforcement in spaced-trial instrumental learning in turtles (Geoclemys reevesii)

The runway performance of turtles (Geoclemys reevesii) was studied as a function of magnitude of reinforcement under spaced training conditions (one trial per day). A large reward magnitude (24 pellets) produced faster acquisition than a small magnitude (2 pellets). After a shift from the large to the small magnitude, latencies of the shifted animals continued to be significantly lower than the latencies of the non-shifted, small-magnitude controls, and similar to those of the non-shifted, large-magnitude controls. There was no evidence of the successive negative contrast effect. Extinction latencies were also significantly lower after training with the large magnitude than after training with the small magnitude; that is, there was no evidence of the magnitude of reinforcement extinction effect. The results are discussed in relation to the comparative analysis of the so-called paradoxical effects of reward, a family of learning phenomena that has not yet been found in a variety of experiments with fish, amphibians, and reptiles trained under widely spaced conditions.