Accuracy of delayed matching to sample was studied in pigeons at different combinations of length of intertrial interval and length of delay. When intertrial interval and delay were varied between sessions in Experiments 1 and 2, accuracy increased monotonically with intertrial interval and decreased monotonically with delay. Evidence was found for constancy of performance at equivalent ratios of intertrial interval to delay, and percentage of correct choices was linearly related to the log of this ratio. In Experiments 3 and 4, intertrial interval was manipulated as a within-sessions variable. In contrast to the effect of this variable when manipulated between sessions, accuracy improved only from the shortest interval to the next shortest interval and remained constant at all longer intervals. In Experiment 4, it was found that performance improved as a direct function of the mean intertrial interval for sessions and that this relation was not affected by the degree of intertrial interval variability within sessions. These findings resemble the effects of temporal variables on autoshaping, and the possibility is discussed that some common processes are involved in delayed matching and autoshaping.
