Judging multi-minute intervals retrospectively

Abstract

A total of 50 participants were asked to perform five different cognitive tasks lasting 120, 210, 300, 390 and 480 s, respectively. After completing the series of tasks, they were asked to estimate retrospectively the duration of each one. Psychophysical analyses linking psychological time to physical time revealed that the value of the power law exponent was about .47, but was .79 when the estimate of the total duration of the session was taken into account—a value lower than unity, indicating that shorter durations have been overestimated, and longer durations underestimated. The Weber fraction, or the ratio of variability to time, ranged from .59 (at 120 s) to .21 (at 480 s). Overall, the study shows that it is possible to make certain changes in the traditional retrospective timing method and thus adapt it for further investigations of the mechanisms involved in memory for the duration of past events.

This research was presented at the 45th Annual Meeting of the Psychonomic Society held in Toronto in 2005 and was supported by a research grant from the Natural Sciences and Engineering Council of Canada and from the Social Sciences and Humanities Council of Canada to S.G., and by an NSERC summer scholarship to M.P. We would like to thank Laurent Gagné-Roy for his help in collecting data.

Notes

¹Ordering the length of the tasks before estimating duration was introduced after Participant 7 gave equal estimates for all tasks.

²In the literature on prospective timing, variability estimates are usually the product of a relatively long procedure (multiple trials). A traditional psychophysical procedure is to draw a psychometric function and extract an estimate of 1 standard deviation, which is operationally adopted as the measure of variability and is used to analyse the validity of Weber's law for time (Killeen & Weiss, 1987). We found no report in the literature linking the variability estimate adopted in the present study and the traditional one. Moreover, we recognize that, although the minimum and maximum values reported have to be based primarily on a temporal-based process, the confidence interval reported is likely to also reflect partially the output of some decisional process.