Perceived duration of an event sometimes deviates from its physical duration. This so-called ‘time distortion’ (Herbst et al., 2013; Kliegl and Huckauf, 2014) has been a useful tool for understanding time-processing mechanisms. The phenomenon of duration adaptation means that after repetitive exposure to a stimulus of relatively short duration, a subsequent stimulus of long duration is perceived as being even longer, and after repetitive exposure to a stimulus of relatively long duration, a subsequent stimulus of short duration is perceived as being even shorter (Walker et al., 1981; Heron et al., 2012). Generally, adaptation is an ecologically useful mechanism, which enables us to flexibly and rapidly change our sensitivity to external stimuli according to the environment around us. Adaptation methods have been widely applied in studies of sensory perception to investigate the mechanisms underlying dynamic and environment-dependent adjustments to our perceptual processes (Webster, 2011). Adaptation-induced time distortion is analogous to sensory adaptation, and is a powerful tool for investigating time-processing mechanisms.

Indeed, many studies have investigated the time-processing mechanisms associated with this duration adaptation paradigm. Heron et al. (2012) proposed the existence of duration-selective channels that respond selectively to a narrow range of stimulus durations centered on a preferred duration. It was suggested that these channels cause duration adaptation. Walker et al. (1981) revealed that duration adaptation does not transfer across modalities, and Heron et al. (2013) reported that duration adaptation precedes multisensory integration. These results implicate modality-dependent clock systems, and suggest that duration adaptation occurs at relatively early stages in sensory processing. Conversely, it has been shown that visual duration adaptation is contingent neither on the orientation of the stimulus (Li et al., 2015a) nor on the hemifield of stimulus presentation (Li et al., 2015b). These results alternatively suggest that duration adaptation occurs at a relatively later stage in sensory processing. Previous studies have reported that duration adaptation occurs only for durations less than 1 s (i.e., sub-second). As aforementioned, most studies suggest that duration adaptation is related to sensory processing, and assume that duration adaptation is a phenomenon limited to the sub-second range, which is crucial for sensory perception. However, duration adaptation for durations of greater than 1 s (i.e., supra-second) has not been explicitly tested in previous studies, and therefore it remains unclear whether duration adaptation occurs for the supra-second time range.

Several studies have suggested that different time-processing systems are used depending on the stimulus duration. For example, it is widely accepted that there are two distinct time-processing systems divided at a boundary of about 1 s (Buhusi and Meck, 2005; Gooch et al., 2011). The involvement of different neural networks for sub- and supra-second processing was reported by studies that used different methods such as fMRI (Lewis and Miall, 2003a,b; Wiener et al., 2010), neuroanatomy (Hayashi et al., 2014), transcranial magnetic stimulation (TMS; Koch et al., 2007), and neuropsychology (Gooch et al., 2011). A pharmacological study showed that drugs that impair the processing of sub-second durations do not impair the processing of supra-second durations, and conversely, drugs that impair the processing of supra-second durations do not impair the processing of sub-second durations (Rammsayer, 1999). Furthermore, unique genotypes are associated with sub-second versus supra-second duration processing (Wiener et al., 2011). Psychophysical analyses have revealed that the individual Weber fraction for sub-second durations does not correlate with that of supra-second durations (Hayashi et al., 2014). Additionally, increases in cognitive load selectively affect performance in a temporal task of supra-second duration (Rammsayer and Lima, 1991). These differences between the sub-second and supra-second processing systems suggest that duration adaptation in sub-second durations may not apply to duration adaptation in supra-second durations.

A system that functions across the sub- and supra-second durations, termed the interval-independent system, has also been proposed (Rammsayer and Troche, 2014). The traditional pacemaker-accumulator model assumes that there exists a single timing mechanism (Treisman, 1963). Lewis and Miall (2009) showed that precision (the coefficient of variation of the perceived duration) decreases constantly across the sub- and supra-second durations without any “break point”. Such continuous perceptual characteristics implicate an interval-independent timing mechanism. Another study reported that central tendency also occurs across the sub-second and supra-second durations (Lejeune and Wearden, 2009). Central tendency describes a phenomenon in which relatively short durations are likely to be overestimated and relatively long durations are likely to be underestimated in tests of multiple durations. The observation of central tendency across the sub- and supra-second durations suggests that multiple durations were processed in an interval-independent system. If systems that function across the sub- and supra-second durations are responsible for duration adaptation, then it follows that duration adaptation should also be observed for supra-second durations.

In the present study, we investigated how duration adaptation affects the perceived duration of the sub- and the supra- second stimuli, and whether these effects remain within the sub- or the supra-second systems, or extend across the systems. We propose three possible patterns for duration adaptation. First, duration adaptation may occur only in the sub-second durations and not in the supra-second durations. This scenario would suggest that duration adaptation involves only the sub-second system. Second, duration adaptation may occur in the sub-second durations and in the supra-second durations, but may not occur across both the sub- and supra-second durations. If this is the case, the supra-second duration would not be perceived as being longer after adapting to the sub-second duration, and the sub-second duration would not be perceived as being shorter after adapting to the supra-second duration. This scenario would indicate that duration adaptation affects the two distinct systems independently. Third, duration adaptation occurs in the sub-second durations, in the supra-second durations, and across the sub- and supra-second durations. This scenario would indicate that duration adaptation involves systems extending across the sub- and supra-second systems.

In Experiment 1, we used sub-second durations for the adapting and test stimuli, aiming to replicate previous studies. We also used supra-second durations to examine whether duration adaptation occurs within supra-second durations. In Experiment 2, we used sub- and supra-second durations to examine whether duration adaptation occurs across the sub- and supra-second durations.

The aim of this experiment was to examine whether duration adaptation occurs in sub-second durations and in supra- second durations.

The aim of this experiment was to examine whether duration adaptation occurs across the sub- and supra- second durations.

Our results indicate that duration adaptation occurred not only for sub-second durations but also for supra-second durations, and occurred across the sub- and supra-second durations.

Previous studies have reported duration adaptation only for sub-second durations (Walker et al., 1981; Heron et al., 2012). Their results imply that duration adaptation only involves the sub-second system. However, our results from Experiment 1 disproved that possibility. Our results from Experiment 2 also denied the possibility that duration adaptation affects the two distinct systems independently. Our results suggest that duration adaptation involves both the sub- and supra-second systems.

Rammsayer and Troche (2014) proposed a model of interval timing using a statistical approach and suggested two possibilities for time-processing systems. Based on their theory, we propose two possible alternatives to explain how duration adaptation involves time processing systems. One explanation is that duration adaptation involves an interval-independent system, not two distinct systems. This is the simplest explanation that is in accord with our results. This explanation is based on the possibility proposed by Rammsayer and Troche (2014) that temporal information is processed by two distinct systems at the initial level, while both these systems are controlled by a common, interval-independent superordinate processing system at the next level. Our alternative explanation is that duration adaptation involves both the sub- and supra-second systems, which are not distinct but are associated with each other. This explanation is based on the possibility proposed by Rammsayer and Troche (2014) that the sub- and supra-second systems are functionally related. That is, durations near 1 s are not processed by either the sub- or supra-second systems, but are processed by both the sub- and supra-second systems, and the boundary between the sub- and the supra-second system is not fixed at 1 s, but is more continuous. It should be noted that these two explanations are not mutually exclusive. In either case, the idea of common or continuous mechanisms between sub- and supra-second systems are also supported by fMRI studies indicating that various brain regions including the inferior frontal gyrus, the somatosensory area, and the supplementary motor area process both sub- and supra-second durations (Lewis and Miall, 2003a; Morillon et al., 2009; Wiener et al., 2010).

Based on previous studies (Ivry, 1996; Becker and Rasmussen, 2007), Heron et al. (2012) suggested that there are duration channels that respond selectively to a narrow range of stimulus durations centered on their preferred duration, and that changes in the channels’ responses cause the aftereffect. This model is called the channel-based model (Heron et al., 2012). If this model explains duration adaptation, the largest aftereffect would be observed when the difference between the adapting and the test duration is optimal, and the aftereffect magnitude would become smaller when the difference becomes larger or smaller than the optimal value. This type of aftereffect is called ‘repulsion-type aftereffect’ (Schwartz et al., 2007), and was not observed in our experiments. However, it should be noted that our results do not refute the channel-based model. In the present study, the difference between the adapting and the test durations is smaller than in Heron’s study. The ratio of test duration to minimum adapting duration and the ratio of maximum adapting duration to test duration were eight in Heron’s study, while the ratios were 2 or 1.4 in Experiment 1, and 2.5 in Experiment 2 of our study. This difference could explain why the ‘repulsion-type aftereffect’ was not observed in the present study.

The parameters used in our study were selected based on the assumption that the boundary between the sub- and supra-second systems is around 1 s (Buhusi and Meck, 2005; Gooch et al., 2011). Some researchers have reported the boundary at around 500 ms (Rammsayer and Lima, 1991; Rammsayer, 1999), while others have reported the boundary at around 2 s (Morillon et al., 2009). Even if the boundary is not around 1 s, our conclusion that duration adaptation occurs across the sub- and supra-second systems is still justified. If the boundary is 500 ms, we can say that, from the sub-sub-supra condition in Experiment 2, an aftereffect was observed for the supra-second duration after adapting to sub-second durations. Conversely, we cannot say that an aftereffect was observed for the sub-second duration after adapting to supra-second durations. However, previous studies have reported that such an aftereffect occurs. For example, Heron et al. (2012) showed that duration adaptation occurs for 320 ms after adapting to 640 ms. Therefore, we can speculate that an after effect would be observed for the sub-second duration after adapting to supra-second durations. Alternatively, if the boundary is 2 s, we can say that an aftereffect was observed for the sub-second duration after adapting to supra-second durations, from the sub-supra-supra condition in Experiment 2. We can also say that, from the supra-second condition in Experiment 1, the aftereffect was also observed for the supra-second duration after adapting to sub-second durations.

Previous studies have suggested that duration adaptation is associated with relatively early stages in sensory processing because duration adaptation precedes multisensory integration (Heron et al., 2013) and does not transfer across modalities (Walker et al., 1981). Additionally, sub-second duration tuning cells are found in primary sensory areas such as the primary visual cortex in the cat (Duysens et al., 1996) and in subcortical structures such as the inferior colliculus in the bat (Casseday et al., 1994; Faure et al., 2003). Here, we propose that duration adaptation is also associated with stages that involve both sub- and supra-second durations. The present study indicates that duration adaptation spans a millisecond-to-second time scale. Duration-tuned cells for the supra-second range are found in the prefrontal cortex in primates (Yumoto et al., 2011). If duration-selective cells act as duration channels and cause adaptation, it follows that duration adaptation should occur in the supra-second range. A recent fMRI study showed that the inferior parietal lobule exhibits duration-selective neural activity (Hayashi et al., 2015). These studies imply that the neural induction of duration adaptation is not restricted to subcortical structures or primary sensory areas that play key roles in primary sensory processing, but is rather distributed among the various brain regions that comprise timing networks. These multiple brain areas may work cooperatively as a network and consist of a single interval-independent mechanism. The present study provides psychophysical evidence for a general duration-selective mechanism in the human brain.

Conceived the experiments: SS, YM, YH, YY. Designed the experiments: SS, YM. Performed the experiments: SS. Analyzed the data: SS, YH. Wrote the paper: SS, YM, YH, YY.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.