Perceptual time distortion and its dependence on physical and perceived properties of visual stimuli(知覚時間歪みの視覚刺激の物理特徴と知覚特徴への依存性)

ラブレンテヴァ ソフィア

 Chapter 1: Introduction
 People can perceive time on different scales, from milliseconds and seconds to hours and days. Time perception on each scale serves a distinct purpose and relies on different mechanisms. Being able to discriminate subsecond time intervals is essential for many behaviors ranging from language and music to sports and driving, and largely relies on the processing of visual, auditory and other perceptual information. Early models of sub-second duration perception suggested the existence of a single dedicated timing mechanism in the brain, resembling an artificial clock (Creelman, 1962; Treisman, 1963). However, later studies discovered several phenomena of time perception that could not be explained by a single dedicated mechanism. Therefore, later models were extended to include modality dependent timing mechanisms, for example a visual timing mechanism specializing in the processing of the time of visual stimuli (Johnston et al., 2006). The visual system processes a wealth of information about visual stimuli, such as their luminance, color, motion and shape. To understand how the local visual timing mechanism works, it is necessary to clarify which stages of visual processing contribute to duration encoding. For example, it has been shown that temporal frequency adaptation at early stages of visual processing can affect the perceived duration of a subsequently presented stimulus (Johnston et al., 2006). Moreover, the perceived durations of visual stimuli are affected by the speeds of their motion, which are encoded at later stages of visual processing in the dorsal pathway of the visual system (Kaneko & Murakami, 2009).
 Shape, which is processed by the ventral pathway of the visual system, is another important aspect of visual stimuli that has not yet been extensively explored in the context of visual timing. It is not clear whether the information about the shapes of stimuli can affect their perceived durations. Unlike temporal frequency and speed, shapes are usually relatively stable over time and do not contain inherent temporal information. Therefore, shape processing may not affect perceived duration. Alternatively, some models of time perception suggest that any brain network can be time sensitive (Paton & Buonomano, 2018), in which case networks that process the shapes of visual stimuli might contribute to duration perception.
 The aim of my thesis is to investigate whether and how shape processing of visual stimuli affects their perceived duration. Since low-level properties of the stimuli can affect their perceived duration, it is important to exclude low-level contributions by using stimuli that are matched in their low-level attributes but differ in terms of shape processing. This can be achieved by using a phenomenon called spatial crowding. Crowding is a phenomenon in which a target stimulus that is easy to discriminate when it is presented alone becomes difficult to discriminate when it is surrounded by other stimuli called flankers (Whitney & Levi, 2011). Strength of crowding can be manipulated by varying the distance and/or similarity between the target and flankers (Bouma, 1970; Kooi et al., 1994). My investigations capitalized on the above fact to create identical target stimuli whose shapes were easy to discriminate in some conditions and difficult in others. To investigate whether and how differences in shape processing influence duration perception, I addressed the following questions:
 (1) Does the strength of the temporal illusion named the temporal oddball effect depend on shape discriminability, or does the illusion occur only due to low-level information even when shape processing is disrupted (Chapter 2)?
 (2) Does a difference in shape discriminability directly influence perceived duration (Chapter 3)?

 Chapter 2: Temporal oddball effect in stimuli that are affected by crowding
 The study shown in this chapter investigated whether and how shape processing affects the temporal oddball effect: when two stimuli of the same duration are presented in succession, the second stimulus is perceived as having a longer duration when it is a novel (oddball) stimulus than when it is a repeat of the first stimulus (Tse et al., 2004).

 In a typical experiment investigating the oddball effect, stimuli differ in both low-level properties and shape. In such conditions it is not clear which difference contributes to the oddball effect. I tested whether being able to process differences in stimulus shape was necessary for the oddball effect to occur. Each stimulus consisted of a target letter surrounded by two flanker letters. Stimulus discriminability was manipulated by inducing five different levels of crowding strength by using five different distances between the target and flankers. Two stimuli were presented sequentially in each trial, and the target letter in the second stimulus was different from (novel stimulus) or the same as (repeat stimulus) the target letter in the first stimulus. The strength of the oddball effect was measured as the probability of perceiving the novel stimulus as longer than the repeat stimulus. If shape discrimination is necessary for the oddball effect, then the oddball effect should be weaker when the difference in stimulus shape is difficult to discriminate. In this case it can be concluded that shape processing can affect duration perception through the temporal oddball effect.

 The target letter discriminability was indeed affected by the crowding strength. However, the oddball effect occurred in all stimulus configurations, irrespective of whether the stimulus shape was easy or difficult to discriminate, and the strength of the oddball effect did not interact with crowding strength. Therefore, the temporal oddball effect does not require shape processing and can be based on the similarity of low-level stimuli features.
 These findings provide some insight into the mechanism of the temporal oddball effect. It is unclear what visual processes cause the temporal oddball effect. Since I have revealed that shape processing is not necessary for the temporal oddball effect, other intermediate visual processes must be considered as candidates for the cause of the temporal oddball effect. One such process could be ensemble coding, which is consistent with the fact that stimuli rendered indiscriminable during crowding contribute to ensemble coding (Manassi & Whitney, 2018).

 Chapter 3: Duration perception of stimuli affected by crowding
 Introspective reports from the first experiment suggested an intriguing possibility that crowding itself might affect duration perception in the following way: stimuli that are difficult to discriminate because of crowding are perceived as having shorter durations than identical stimuli that are not affected by crowding and thus are easy to discriminate. In this chapter I explore this possibility.
 I examined whether the perceived durations of physically identical stimuli could differ depending on how they were affected by crowding and how easily they were discriminated. Each stimulus consisted of a target letter surrounded by ten flankers, some of which shared their color with the target while the others did not. There were two stimulus types, defined by which flankers shared their color with the target: the nearest to the target (crowded stimulus) or the furthest from it (non-crowded stimulus). Both crowded and non-crowded stimuli were presented in each trial in random order. The participants indicated which stimulus they perceived as having a longer duration.
 I found that stimuli were perceived as having longer durations when they were unaffected by crowding and easy to discriminate. These results were replicated with two different types of shapes, which means that the phenomenon is not stimulus specific but depends on general shape recognition. Changes in perceived duration could not be explained by perceptual shifts in stimulus onsets and/or offsets. Therefore, I concluded that differences in shape processing can directly affect duration encoding. I also found some evidence that easy-to-discriminate stimuli are processed with better temporal resolution. Longer perceived durations of these stimuli might be a byproduct of such improved temporal resolution.
 Taken together with previous studies, these findings imply that information processed at different stages throughout the visual system, including the information about stimulus shapes, contributes to time perception.

 Chapter 4: General discussion and conclusions
 I have demonstrated that shape processing in the visual system can affect duration perception in some situations, such as when judging the duration of a stimulus affected by crowding, but that the temporal oddball effect does not depend on shape processing. These findings imply that perceived duration of highly similar stimuli can be affected by visual processing at multiple levels of the visual system, thereby supporting distributed mechanisms of time perception.
 If distributed models of duration processing are correct, then to develop a full theory of how duration is encoded such models must answer the following two questions. First, it must reveal the types of information that contribute to duration encoding and how they do so. Second, it must elucidate how information from different sources and across different levels of processing is integrated to form a single internal representation of an interval duration.
 In this thesis I showed that low-level visual information about physical features of a stimulus and high-level perceptual information about the stimulus shape can both affect how stimulus duration is perceived. These findings support the notion that any visual information can contribute to duration encoding, rather than being limited to only some special kinds of information, such as stimulus magnitude or salience.
 I also found that perceived stimulus duration contracts due to crowding more when the stimulus is attended to, which highlights the role that attention plays in duration perception. On the other hand, I demonstrated that a difference in physical features of stimuli, that cannot be consciously discriminated and are thus unavailable for attentional control, can cause the temporal oddball effect. Thus, the findings of this thesis suggest that information about duration of the stimulus is integrated based on both bottom-up and top-down processes.
 In conclusion, my thesis supports distributed models of time perception and expands these models by proposing visual shape as relevant information for duration perception.