Timescale- and Sensory Modality-Dependency of the Central Tendency of Time Perception

Abstract

When individuals are asked to reproduce intervals of stimuli that are intermixedly presented at various times, longer intervals are often underestimated and shorter intervals overestimated. This phenomenon may be attributed to the central tendency of time perception, and suggests that our brain optimally encodes a stimulus interval based on current stimulus input and prior knowledge of the distribution of stimulus intervals. Two distinct systems are thought to be recruited in the perception of sub- and supra-second intervals. Sub-second timing is subject to local sensory processing, whereas supra-second timing depends on more centralized mechanisms. To clarify the factors that influence time perception, the present study investigated how both sensory modality and timescale affect the central tendency. In Experiment 1, participants were asked to reproduce sub- or supra-second intervals, defined by visual or auditory stimuli. In the sub-second range, the magnitude of the central tendency was significantly larger for visual intervals compared to auditory intervals, while visual and auditory intervals exhibited a correlated and comparable central tendency in the supra-second range. In Experiment 2, the ability to discriminate sub-second intervals in the reproduction task was controlled across modalities by using an interval discrimination task. Even when the ability to discriminate intervals was controlled, visual intervals exhibited a larger central tendency than auditory intervals in the sub-second range. In addition, the magnitude of the central tendency for visual and auditory sub-second intervals was significantly correlated. These results suggest that a common modality-independent mechanism is responsible for the supra-second central tendency, and that both the modality-dependent and modality-independent components of the timing system contribute to the central tendency in the sub-second range.

Fig 1. Schematic of the experimental procedure.

Two brief flashes or tones were sequentially presented, and participants made button presses to reproduce the interval. Correct or incorrect feedback was presented immediately after participants’ responses.

Fig 2. Group means of the reproduced intervals.

The left panel indicates the results for sub-second timing. The right panel indicates the results for supra-second timing. Blue lines represent results for visual stimuli, and red lines represent results for auditory stimuli. Dotted lines correspond to accurate reproduction of the stimulus intervals. Error bars represent the standard error of the mean.

Fig 3. Quantification of the central tendency.

The slopes of the linear regressions are presented for all four conditions. A and V represent auditory and visual stimuli, respectively. Sub and sup represent sub-second and supra-second intervals, respectively. Error bars represent the standard error of the mean. *** p < .001, n.s. = not significant.

Fig 4. Within-individual correlations of the central tendency across different sensory modalities (a) and timescales (b).

Regression index represents the slope of the linear regression of the reproduced intervals to the stimulus intervals. Each small circle represents an individual’s data. V and A stand for the visual and auditory conditions, respectively. Sub and supra represent the sub-second and supra-second intervals, respectively. *** p < .001.

Fig 5. Psychometric functions of visual (a) and auditory (b) intervals for a typical subject.

The probabilities at which the subject judged that the comparison (350–650 ms) was longer than the standard (500 ms) were plotted, and fitted to the cumulative normal distribution function. The discrimination sensitivity was defined as the Weber fraction, the ratio of the just noticeable difference (JND) to the standard interval.

Fig 6. Within-individual correlations of the central tendency across visual and auditory modalities.

Regression index represents the slope of the linear regression of the reproduced intervals to the stimulus intervals. Small open circles represent each individual’s data. Bold circles and bars represent the mean and the standard error of the mean, respectively. * p < .05.