Abstract
Some stimulus properties (e.g. contrast) are represented by neural response level, while others (e.g. spatial frequency and orientation) are represented by mechanisms that are narrowly tuned for a subset of the available range. Contrary to these properties, the representation of time is less well studied, less well understood and may be highly non-linear in the presence of eye movements. We use an equivalent noise paradigm to analyze how the perception of time changes with the duration of a signal and in the presence of a saccade. The mean duration of a signal composed of 200 random dots was manipulated by changing the duration of each dot according to a Gaussian distribution whose standard deviation was systematically manipulated. Observers were required to estimate the mean duration of two such signals that were presented sequentially and to indicate which lasted longer in a 2AFC task with feedback. Weber fractions for mean duration discrimination were extremely high (50–75%) and increased with signal duration. The reduction in time sensitivity at longer intervals was caused by a steady increase in internal noise over time. Estimates of sampling efficiency indicated that remarkably few elements were employed to perform this task at all durations. Sensitivity to time was impaired by the presence of a saccade in the middle of the trial (enforced with an eye tracker), which was attributed almost entirely to a large increase in temporal noise. These results suggest that duration is represented by low-resolution time channels, analogous to a family of noisy ‘clocks‘, and that saccades increase temporal uncertainty, which allows for the possibility of reversals in apparent time without invoking non-linear or relativistic models of time perception.