Choose-long errors at delays shorter than the training delay persist when enhanced between-trial summation of duration memories are precluded in pigeons.

Pigeons were trained to match 2- and 10-s durations of houselight to red and green comparisons. Following acquisition with a 0-s baseline delay, they were tested with delays of 0, 10 and 20s. There was a strong tendency to choose the short-associated comparison stimulus at both the 10- and 20-s delays (i.e., a choose-short effect) and no bias at the 0-s delay. This test was repeated after baseline training with a constant 10-s delay. As in the first delay test, a choose-short effect was obtained at the 20-s delay. In contrast to the first test, no bias was obtained at the 10-s delay and a strong tendency to choose the long-associated comparison (i.e., a choose-long effect) was obtained at the 0-s delay. Importantly, the choose-long effect was obtained under conditions which insured that the temporal spacing between a 0-s delay trial and the preceding trial was equal to that in training. These results are inconsistent with the temporal summation account of the choose-long effect and are most readily interpreted within a perspective emphasizing the subjective shortening of temporal memories over time.