Relative importance of perceptual and mnemonic variance in human temporal bisection

We investigated the relative contributions of perception and reference memory to behavioural variability in a temporal discrimination with human subjects. We used two temporal bisection tasks. In both tasks each trial consisted of a sequential presentation of three intervals, two standards, and a probe, and subjects were asked to judge the similarity of each probe against the two standards. In a "single bisection", the standards' duration was constant across trials. In a "roving bisection", the two standards were trial unique. We compared our results with the predictions from a model related to Scalar Expectancy Theory, with the added assumption that the decision process minimizes the expected number of errors given the information available. The model shows that if errors in reference memory were dominant the psychometric function should be identical for single and roving tasks, and if perceptual errors were dominant the psychometric function should be steeper for the single than for the roving bisection. As we found that psychometric functions were steeper for the single than for the roving tasks, we concluded that perceptual errors are dominant.