Performance of goldfish (Carassius auratus) in patterned sequences of rewarded and nonrewarded trials.

The performance of goldfish was studied in a series of experiments with patterned sequences of reward (R) and nonreward (N) for response to a given stimulus (S3). Trials with two other, readily discriminable stimuli, response to one always rewarded and response to the second never rewarded, were interpolated in such a way that the outcome of any S3 trial could not be predicted from the events of the immediately preceding trial. The main purpose was to look for control of response to S3 based on memory of N like that found in previous work with pigeons. In Experiments 1 and 2, blocks of R and N trials with S3 were programmed, R trials in the first half of each session and N in the last half, or N trials in the first half and R in the last half. In Experiment 3, successive acquisition and extinction of response to S3 was studied, with the N-to-R and R-to-N transitions either within sessions or between sessions. In Experiment 4, the effects of partial and consistent reinforcement on extinction of response to S3 were compared. The results obtained are similar in important respects to those for pigeons but different in their failure to demonstrate control of performance by memory of N.