Extinction of food-reinforced instrumental behavior in Japanese quail (Coturnix japonica).

Japanese quail (Coturnix japonica) were reinforced with food for traversing a runway for either 18 or 36 trials, administered at a rate of 1 trial per day. Then, all animals received 18 extinction trials. The latency to run from the start box to the goal box was the dependent variable. Extinction was significantly slower in animals that had received 50% partial reinforcement during acquisition, whether relative to a group matched in terms of acquisition trials (36 trials, twice the number of reinforced trials) or relative to a group matched in terms of reinforcements (18 trials). The latter group was also matched in terms of the temporal distribution of acquisition trials with the partial reinforcement group, being trained only on days when the partial group was scheduled to receive a reinforced trial. Thus, there was evidence of a spaced-trial partial reinforcement extinction effect. A comparison of groups receiving large versus small reward magnitudes yielded no evidence of the spaced-trial magnitude of reinforcement extinction effect, even though the large-reward group consumed approximately 3 times more food than the small-reward group. Moreover, a comparison of groups that received 36 versus 18 acquisition trials produced no evidence of the spaced-trial overtraining extinction effect, even though acquisition latencies were significantly lower for the group that received 36 acquisition trials. These results are discussed in relation to comparative research on learning phenomena involving incentive downshift manipulations.