Temporal phase discrimination was measured as a function of spatial separation of the stimulus components. In contrast to many previous studies, phase discrimination thresholds were measured directly, rather than inferred from the ability to discriminate synchronous from antiphase stimuli, or from segregation or shape tasks. For abutting bars, relative phase thresholds were closely proportional to temporal frequency. The proportionality corresponded to a threshold temporal offset of 2.5 to 9.5 msec, across subjects. Introduction of a small gap (0.125 deg or greater) led to a dramatic (3 to 7-fold) increase in thresholds for temporal phase discrimination, and thresholds were no longer proportional to temporal frequency. Insertion of a third bar filling the gap resulted in a recovery of the low thresholds, provided that its modulation was consistent with apparent motion across the three bars. Below 8 Hz, phase discrimination thresholds across three bars were equivalent to thresholds for two abutting bars. Above 8 Hz, phase discrimination thresholds for the 3 bar combination were lower than thresholds for two adjacent bars, implying that phase information was integrated across all three bars.
Phase discrimination thresholds do not appear to reflect the properties of a single mechanism. Especially at high temporal frequencies, low thresholds for phase discrimination are closely tied to the presence of apparent motion. Temporal phase discrimination is markedly impaired by a small separation of stimulus components. Moreover, the inability to detect phase differences across gaps corresponds to the loss of phase-dependence of vernier acuity thresholds across gaps.