Pigeons were rewarded on a fixed-interval (FI) schedule. On occasional unrewarded tests, they usually showed a break-run-break pattern of responding. Across trials, the start of the run correlated positively with the end, suggesting variance across trials in (a) clock speed, (b) the delay to start the clock, or (c) a single criterion time used on a trial. If variance in criterion time exists, then a task in which the start and the end of the run are based on independent criterion times should produce a reduced start-end correlation. In 2 such tasks, the start-end correlation dropped to near zero, although other correlations were comparable to those found in the standard FI task. This provides evidence for variance across trials in a single criterion time used on a trial.
