Temporal and nontemporal control in a midsession reversal task with variable intertrial intervals

To assess the degree of temporal control in the midsession reversal task, pigeons learned a simultaneous discrimination with two stimuli, S1 and S2. Choices of S1 were reinforced on the first 40 trials and choices of S2 on the last 40. Variable intertrial intervals (ITIs) separated the trials. The pigeons learned to reverse preference from S1 to S2 near trial 40. To see if pigeons had learned a temporal discrimination, we either doubled or halved the mean of the ITIs during test sessions. If choices were based on temporal cues, preference should reverse at the same time in the session but on half as many trials into the session when the mean ITI doubled and twice as many trials later when the mean ITI halved. Variable ITIs should reduce generalization decrement and thereby reveal temporal control more clearly. The reversal trial changed in the directions predicted by timing, but the magnitude of the change was smaller than predicted. Most individual choice patterns were consistent with temporal control, but a few were consistent with control by trial number or by the events of the previous trial. Variable ITIs seem to reduce the weight of temporal cues relative to the weight of nontemporal cues.