Time perception research in real world

Time perception refers to how people experience and understand time. It is subjective and can be affected by various factors in the physical world [6]. These factors are typically categorized as either external (e.g., environmental conditions) or internal (e.g., individuals’ physiological and psychological states).

External factors refer to various elements arising from the surrounding environment that influence time perception, such as the environment’s high heat acclimation, which can impair time awareness, and contribute to perceiving time longer [7]. Similarly, medicine that increases dopamine levels can prolong subjective time perception by affecting brains, for instance, high-concentration attention leads the brain to deal with extra information [8]. Environmental stimuli, such as music tempo and sound intensity also influence time perception. Specifically, slower tempos and louder sounds often lead to a distorted sense of time, making it feel longer [9, 10]. Visual stimulation, such as flicker rates, can also alter time perception and increase time estimates [11].

Internal factors, however, are linked to an individual’s inherent physiological and cognitive state on time perception. These factors can affect subjective time perception. For instance, conditions such as depression can slow time perception due to changes in brain function and neuroscience. People with mental illnesses generally experience time delay [12]. Emotions, such as fear and anxiety, can also impact time perception. Increased anxiety tends to make time feel longer because the heightened arousal from negative stimuli intensifies the perception of time passing slowly [13, 14]. Sex differences also played a role. Women typically perceive time intervals as being longer than men [15].

The internal clock model is widely accepted in psychological time research [16]. This model suggests that an internal clock system exists inside the human brain, including a pacemaker emitting pulses captured and accumulated by a switch, and the accumulated count can be used to judge and represent time intervals, that model could help individuals perceive time [17]. The model has been instrumental in understanding how various factors influence time perception, particularly in scenarios where time is distorted, such as during states of boredom when time seems to pass more slowly owing to reduced brain activity [2].

Time perception research in VR

VR has enormous potential to offer immersive experiences that significantly affect time perception. Previous research has shown that while engaged in VR activities such as games or immersive videos, users often lose track of time, experiencing what is commonly called “time flying.” This effect is likely owing to the high level of engagement in VR, which distracts users from external time cues and body rhythms [18]. Schneider et al. attempted to use VR to relieve breast cancer pain during chemotherapy. They found that the patients’ duration judgment during VR-immersive chemotherapy underestimated the actual time by 28 percent, demonstrating how VR can distract attention from discomfort [19]. Lugrin et al. highlighted that comparing waiting times in the real world and virtual environment(VE) in the same situations, participants perceived time as being shorter than in real waiting room conditions because people focus more on the VR content, and the higher arousal, distracted their attention [20].

Recently, avatars have become the main target in VR time perception research. Unruh et al. highlighted that the presence of avatars and embodiments in VR can affect time perception. Participants reported the perceived time to be faster when they had self-controlled avatars [21]. Unruh et al. also investigated the effect of avatars with different levels of embodied representation, such as hands, arms, and the full body, which are controlled by the participants, on their time perception. They found that when participants controlled the avatar only with hands without any other part of the body, it contributed to the perception of time passing more slowly than in other conditions [22]. In a study by Landau et al., participants who embodied avatars with a child’s appearance perceived the duration of time in the VE as longer than those with other avatar appearances [23].

In addition to avatars, various virtual objects, and elements, such as audio-visual stimuli and object motion, influence time perception in VR. Picard et al. examined the effects of audio, visual, and combined stimuli when participants performed object-sorting tasks in a VE. Their results demonstrated that a combination of visual and audio stimuli significantly decreased the judgment of time duration. Furthermore, synchronized stimuli, such as virtual and audio elements, can lead to time compression or dilation, depending on the tempo [24]. Yamamoto and Miura explored the effect of the coherence of visual object motion on time perception. They found that participants who observed four line segments forming a coherent motion experienced altered time perception, which was influenced by perceived motion speed [25].

Time perception research on exercise intensity

The relationship between exercise and time perception has been the focus of numerous studies, aimed at understanding the underlying mechanisms. For instance, Lambourne’s study on cycling found that participants perceived time to be approximately 15% during exercise than during rest. This can be explained by the internal clock model: cycling alters the internal clock pulse rate, leading to an overestimation of time [4]. Petrizzo et al. demonstrated that running can cause individuals to overestimate the time spent on visual stimulation tasks, suggesting that exercise acts as a distraction that influences time evaluation [26].

Exercise intensity is a vital factor affecting time perception during exercise. Hanson and Lee found that increased exercise intensity leads to slower perceived passage of time when people running at higher exertion levels like Rating of Perceived Exertion (RPE) of 17, and the estimated time is significantly lower than at moderate levels, such as RPE 13 [27]. Edwards and McCormick also found that exercise intensity distorted the participants’ time perception. The study found that performing Wingate cycles and rowing ergometer boats at slight and medium intensities did not affect time perception while exercising at maximum intensity caused participants to perceive a shorter duration [28]. Additionally, Hanson and Buckworth found that women chose to run at a relatively higher intensity when they could be selected by themselves, compared to men. Simultaneously, a higher intensity leads to a relatively slow feeling of time passage [29]. Moore and Olson indicated that time perception differed before and after high-intensity exercise. They tested people before and after exhaustive cycling and found that high-intensity exercise usually induced time distortion owing to fatigue. Under the fatigue state of people, duration judgment increases, and the relative error significantly increases [30].

In general, heightened sensory awareness and physical discomfort during intense exercise contribute to a longer perceived duration. This phenomenon is often linked to the release of catecholamines, which trigger hyperarousal—a state where the brain processes more information than usual, making time seem to pass more slowly than it does [31, 32].

However, while exercise intensity has been proven from people’s internal perspective will affect their time perception. But, the stimuli of exercise intensity from outside will also work on people’s time perception still not known, like the surrounding avatars. Thus, this is still important to investigate whether the intensity of the surrounding avatars affects people’s time perception.

Time perception research on motion speed

The motion speed also significantly influenced time perception. For example, running compared with the static mode, running led people to percieve that time was going longer [33]. By contrast, Rietzler et al. found that controlling an avatar with a slowed-down motion in a VE results in a shorter estimated duration [34].

In addition to self-motion, external motion stimuli also affect time perception as well. Kaneko et al. demonstrated that participants in a virtual car perceive a longer time when experiencing a faster motion speed [35]. Similarly, Landeck et al. immersed participants in virtual tunnels with numerous different sections and found that a stronger sensation of self-motion and fast speed increased subjective time [36]. Additionally, Landeck’s further research showed that external object motion within VE, such as pendulum movements, caused time to be perceived as passing more quickly than simpler motions like linear or rotary because the complexity of the pendulum motion captured more of the participants’ attention [37]. Dynamic images and speed cues can influence people’s time perception, often causing them to overestimate the duration [38]. Another study compared a virtual object’s motion speed to a human movement speed, such as point-light displays (PLDs), with only a few points of light representing the major joints or key parts of a moving body. After watching themselves and other people’s dancing movements, which are recorded as PLDs at different speeds. The result showed that the PLDs’ moving speed also affects people’s time perception, with a faster motion speed of human movement frames leading to longer duration judgment [39].

In summary, both the internal and external motion speeds affect time perception. Images and PLDs with human motion information and speed properties are effective. In VR environments, such as VR gyms, the influence of motion speed is further compounded by social and interpersonal factors, where avatar behavior can introduce additional stimuli that affect users’ time perception [40]. Furthermore, human information, such as dynamic facial expressions and high-arousal emotional displays, induces a longer perceived duration [41]. Thus, displaying only PLDs, despite being understood as a human performance action, loses considerable information compared to avatars [42].

Hence, it is crucial to extend prior research on self-perceived motion speed and visual stimuli such as PLDs by investigating the impact of the surrounding avatar motion speed on individuals’ perception of time.

Contributions

Previous research on avatars and time perception in VEs has focused on user-controlled avatars. However, the effect of surrounding avatars in VR gyms remains under-explored. The surrounding avatars, which can act as non-player characters (NPCs) or other users’ avatars, play a significant role in the social spaces of VR gyms. They differ from other virtual objects because of their greater emotional and social connection points, which more closely resemble real human users, with movements that mimic human behavior [43]. Given their potential impact on time perception, it is essential to investigate how these surrounding avatars affect participants in VR gyms. This study addresses this gap by examining two key attributes of the surrounding avatars: motion speed and exercise intensity. We aim to provide insights that could guide the design of more engaging and effective exercise experiences in future VR gymnasiums.

Software and hardware

A fitness bike (ALINCO, AFB6119) with a magnetic load system was used by the participants for cycling. The maximum load capacity of this fitness bike was 120 kg, and various sports modes could be selected. The entire experimental system ran smoothly on a desktop PC(Intel(R)Core^™ i7-12700, 16 GB RAM, NVIDIA GeForce RTX 3060). We chose META QUEST 2 as our Head mount display device to provide participants with a truly immersive virtual experience. This HMD has a 90° field of view, a resolution of 1832 × 1920 pixels per eye, and a 90Hz refresh rate. To creating the experimental environment, we have chosen Blender (v. 3.6.5.0) as the main tool to create a the 3D model and Unity3D (v. 2021.3.9f1) as the main development tool. Using 3D modeling, we constructed a VR gymnasium environment consisting of many different training devices, such as treadmills, barbells, and chest push frames. Regarding the design of the entire virtual space, we opted for a low-poly aesthetic to craft 3D models constituting the VR gym. While this design choice lent a distinctive style to our environment, it could potentially trade off the degree of realism. However, based on our subjective experience, this stylistic decision did not markedly affect the overall virtual immersive. Unity3D was used to create an entire environment and arrange each position of the training device and VR application (Fig 1). A specific experimental room was laid out to allow participants to easily slip into the VR gymnasium. A virtual bicycle station was placed within the experimental area, permitting participants to engage in experimental tasks while immersing themselves in the VR gymnasium scenario.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. The VR gym design 3D model of training machine.

These were used as the main scenario during the whole experiment.

https://doi.org/10.1371/journal.pone.0311860.g001

For the surrounding avatars and the avatar representing participants, both of them used low-poly style models from “Low Poly Animated People” [46], which can fully improve visual coherence and presence in VR gym environments. In addition, the outside appearance of the surrounding avatar assumed that of an athlete, which was more suitable for VR gym (Fig 2A and 2B). The default surrounding avatar was positioned centrally in the VR gym to ensure maximum visibility. Two barbells with different numbers of weight pieces be used to do squat by the surrounding avatar: one with three big plates and another smaller plate on each side of the barbell pole, another one only has barbell pole-like (Fig 2A-1 and 2A-2). The surrounding avatar animation, a barbell squat, was obtained from the Mixamo website [47].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. The Avatar used in the experiment: A) Surrounding avatar A-1) Surrounding avatar under heavy-workload condition A-2) Surrounding avatar under light-workload condition B) Participant’s avatar B-1) Participant’s avatar with cycling gesture B-2)User’s viewpoint.

https://doi.org/10.1371/journal.pone.0311860.g002

To represent the participants within the virtual gym, we crafted an avatar situated on a stationary bike (Fig 2B-1). The participants perceived the VR gym through a first-person perspective to observe the environment, allowing them to change their sight view by simply moving their heads (Fig 2B-2). Considering the nature of the experiment, whether it was a sitting or cycling task, the participants remained postured on a stationary bike until each experimental section finished. Consequently, the participant’s avatar was fixed to the virtual station bike in the VEs (Fig 2B-1).

Participants

We recruited 28 participants for the experiment through an online communication group. The recruitment period ran from September 14 to October 5, 2023, and the experiment was conducted from September 23 to October 10. The average patient age was 26 years, with a standard deviation of 2.4 years. Among the participants, there were 9 females and 19 were male. All the participants were unaware of the purpose of the study. The participants had normal or corrected-to-normal vision. All participants provided written informed consent. Participants were free to withdraw at any time during the experiment. This experiment was approved by the Research Ethics Committee of the Japan Advanced Institute of Science and Technology and was conducted in strict compliance with its contents.

Experimental conditions

In real-world gyms, people typically do squats or other resistance exercises following a mode like one training set with one rest set, which is called “Rest between sets (RBS)” [48]. For squats, people typically spent 20, 30, and 40 s to complete each training set.

We conducted a within-participants experiment using three independent variables. The first was motion speed with three levels: I) slow motion speed, II) medium motion speed, and III) fast motion speed. The frequency was used to define three levels: for slow motion speed, the surrounding avatar performed 19 squats per minute; for medium motion speed, it was performed 28 times per minute; and for fast motion speed, it was performed 38 times per minute. The motion speed of the surrounding avatar was based on “The Ruffier Squat Test”, which is a quick fitness test that checks heart health by measuring heart rate before and after doing squats. it needs people to squat 30 times in 45 seconds. However, since this is a high-intense test, we reduced the frequency to 38 times per minute as the fast-motion speed condition in our study. Furthermore, a 30% reduction was applied to create the other two conditions [49]. The second independent variable was the exercise intensity of the surrounding avatar, with two different exercise intensities: I) low and II) high. For the low exercise intensity, the surrounding avatar lifted a barbell pole without any extra weight plates; for the heavy intensity, the surrounding avatar lifted a barbell pole with three big and two small weight plates on each side of the barbell. The last independent variable was the state condition of the participant. The participants observed and estimated the duration while either sitting on a stationary bike or cycling on a stationary bike.

In designing the duration of each trial, we took into account previous research by Landeck et al [36]. The duration of each trial was set to 20, 30, and 40 seconds. Each default time that we set for each trial was presented with equal frequency and in a random sequence. So for each experiment session, participants observed the surrounding avatar doing exercise with RBS, and the task was to estimate time for each exercise set, with the rest time allocated for answering questions. This task design makes the duration judgment of each workout set more relevant to actual gymnasium training styles. Additionally, previous research found that judgment long term is more affected by emotions [50]. Relatively, short-term duration judgment can reflect people’s time perception more accurately. Meanwhile, Droit-volet et al. found that the feeling of time passage had a positive correlation between several seconds and several minutes of judgment [50, 51]. Also, when time exceeds 1 scend duration judgment and feeling of time passage also correlate.

Procedure

The entire procedure of the experiment is shown in Fig 3. Participants first signed a consent form after the experiment explanation after they completed a demographic questionnaire to provide basic information. Next, they then wore disposable socks and indoor sneakers, adjusted the bike seat height to match their height, and put on the HMD (Fig 4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. The procedure for the VR experiment.

https://doi.org/10.1371/journal.pone.0311860.g003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. The scene of the experiment.

https://doi.org/10.1371/journal.pone.0311860.g004

First, they had a practice session to get used to the task, which included 6 trials and answering questions 6 times. Following the practice session, participants took a 5-minute rest in a separate room. The main experiment consisted of four sessions: two cycling sessions and sitting sessions. To reduce the potential bias caused by cycling fatigue, there were no two consecutive cycling sessions. For the sitting session which lasted 18 minutes, participants sat on the bike and were first shown a hint scene: “You are in the observation stage, please observe and familiarize yourself with the environment,” to help them immerse into the VR gym. After that, participants entered and observed the environment for 1 minute and familiarized themselves with the VR gym. Then, they began the first trial. After each trial, they answered two questions. After answering, they proceeded to the next trial. This process continued until all 19 trials were completed, one of the trials was without avatar condition as the baseline in the experiment. The cycling sessions, lasting around 19 minutes, differed only in the hint scene: “You are in the warm-up stage, please start cycling. The experiment will start after 2 minutes warm up.” The remaining trials followed the same format as the sitting sessions. The stationary bike was set to a 30-watt mode to ensure all participants consumed the same amount of energy regardless of their pedaling frequency.

After each session, the experimenter helped participants remove the HMD. There was a 10-minute rest period between sessions, during rest time, participants wore eye masks and headphones to listen to light music, promoting deep rest and reducing fatigue and outside interference. Throughout the experiment, participants completed an equal number of trials involving combinations of three motion speeds, two exercise intensities, three observation times, and one blank scene without any surrounding avatar, counterbalanced in randomized sequences, totaling 76 trials. After the experiment, a brief interview was conducted, which focused on the effect of the surrounding avatar’s motion speed and exercise intensity on their perception of time.

Measures

In this experiment, we designed the following two questions to measure how long the participants felt in each trial and their perception of time passage, as used by Lugrin et al., Unruh et al., and Landeck et al. [20, 22, 36, 37]. Both of these questions were displayed in Chinese and were only explanation during the experiment. All participants fully understood the meaning of the two questions by reading the experiment explanation document.

• Q1: Intuitively (no more thinking), how long do you think you existed in the VR gym? Please use seconds as the unit of estimated time and say it directly and orally.
• Q2: From your physical feeling, how fast is the time passage?

Q2 is a 7-point Likert scale adopted from the Inventory on Subjective Time Self and Space (STSS) [52]. Q1 was used to measure participants’ subjective time perception, “individual’s objective estimation of the duration of a specific period”. Q2 was used to measure “participant’s subjective perception of the rate at time passage”. Based on the specific situation in this experiment, we have revised the question slightly. For Q1, we added a description of how to answer the question orally and the details of duration judgment from existing in the VR gymnasium. For Q2, “from your physical feelings” was added; participants needed answers from their physiological responses rather than relying on their brain’s thinking to get more accurate and avoid errors on the feeling of time passage.

Duration judgment

The results of the duration judgment are presented in Table 1 and Fig 5. Although the assumption of normality was violated for the dependent variable (duration judgment), we converted the data logarithmically [53]. After conversion, most data were normal. However, several conditions were still violated; therefore, we also proceeded with ANOVA because it is quite robust in terms of normality [54, 55].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Main results of the dependent variable duration judgment: Scores on motion speed, exercise intensity, and participants’ state, with error bars.

https://doi.org/10.1371/journal.pone.0311860.g005

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. The result of duration judgment.

https://doi.org/10.1371/journal.pone.0311860.t001

For the comparison between the surrounding avatar’s different motion speeds, there was a significant effect on the duration judgment (F(1.33, 35.98) = 29.98, p < .001, ). After that, there also was an interaction significant effects between motion speed and participants’ state (F(3.51, 84.92) = 8.61, p = .005, ). Furthermore, there was a significant difference in duration (F (1.72, 46.35) = 1362.87, p < .001, ) on duration judgment.

For the comparison between the surrounding avatar’s exercise intensities, the results revealed no significant difference in the duration judgment between exercise intensities (F(1, 27) = 0.03, p = .865, ). For the comparison of participants’ states, the result presented that there was no significant difference (F(1, 27) = 0.52, p = .479, ).

The Bonferroni test was used to perform post hoc tests following the significant main effect of motion speed on duration judgment. Bonferroni-adjusted post hoc comparisons revealed that all pairwise differences between the surrounding avatar motion speeds for three durations were significant. The analysis revealed distinct differences among the various motion speed conditions. the fast motion speed condition (M = 3.02, SE = 0.04) compared with the medium motion speed condition (M = 2.94, SE = 0.04) (fast motion speed >medium motion speed (t(27) = 5.112, p < .001)), the medium motion speed condition (M = 2.94, SE = 0.04) compared with the slow motion speed condition (M = 2.84, SE = 0.05) (medium speed >slow speed (t(27) = 4.558, p < .001)).

To further understand the interaction effects on the participants’ states and the surrounding avatar’s different motion speeds, we also performed a post hoc comparison with Bonferroni adjustment. The results showed that at the slow motion speed, the cycling state (M = 2.88, SD = 0.45) was not significantly different(t(27) = -1.993, p = .0565) from the sitting state (M = 2.80, SD = 0.42). Similarly, under the medium motion speed, the sitting condition (M = 2.93, SD = 0.38) compared with the cycling condition (M = 2.95, SD = 0.43) was not significantly different (t(27) = -0.632, p = .532). In addition, under fast motion speed, the sitting condition (M = 3.03, SD = 0.371) was significantly higher (t(27) = 0.971, p = 0.340) than the cycling state (M = 3.00, SD = 0.42).

Feeling of time passage

The results of the feeling of time passage are presented in Table 2 and Fig 6. For the dependent variable of the feeling of time passage, the result of the assumption showed that normality was violated. Although the normality assumption was violated, we continued with ANOVA because of its robustness [54, 55].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Main results of the dependent variable feeling of time passage: Scores on motion speed, exercise intensity, and participants’ state, with error bars.

https://doi.org/10.1371/journal.pone.0311860.g006

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. The result of the feeling of time passage.

https://doi.org/10.1371/journal.pone.0311860.t002

The results showed that the feeling of time passage was significantly affected by two independent variables related to the surrounding avatar, the motion speed (F(1, 27) = 116.99, p = .019, ), and the exercise intensity (F(1, 27) = 6.25, p = .019, .

To explore the effects of the surrounding avatar’s different exercise intensities and speeds on the feeling of time passage, we conducted a post hoc pairwise comparison with Bonferroni adjustment, and the results showed that faster motion speed led to a higher feeling of time passage (fast motion speed>medium motion speed (t(1, 27) = -8.432, p < .001); medium motion speed>slow motion speed (t(1, 27) = -12.817, p < .001)).

For the participants’ states, there was no significant difference in the feeling of the time passage (F(1, 27) = 1.82, p = 0.177, ). There was a significant interaction between motion speed and participants’ state (F(1.78, 47.93) = 10.26, p < .001, ).). Other interactions did not show any significant differences. Subsequently, we performed a post hoc Bonferroni correction for the interaction effect of the participant’s state and motion speed. The results showed that under the slow-motion speed condition, the sitting state (M = 3.06, SD = 0.99) was significantly lower (t(27) = -2.305, p = 0.029) than the cycling state (M = 3.30, SD = 1.11). However, under the medium motion speed condition, the sitting state (M = 4.38, SD = 0.80) was significantly lower (t(27) = 0.656, p = 0.518) than the cycling state (M = 4.30, SD = 1.02). Under the fast-motion speed condition, the sitting state (M = 5.48, SD = 1.05) was significantly lower (t(27) = 1.683, p = 0.104) than the cycling state (M = 5.28, SD = 1.20).

Summary of experiment results

Our results strongly indicate that the attributes of the surrounding avatar significantly influenced the participants’ time perception. For the hypothesis that the participants judge the duration longer in fast motion speed and high exercise intensity conditions compared to slow motion speed and low exercise intensity conditions, responses to Q1 (“How long do you think you existed in the VR gym?”) showed significant differences between conditions with varying motion speeds, and the participants judged the duration to be longer in faster motion speed conditions. Thus, this result supports H1-1. However, the results did not show significant differences in the estimated duration between the two exercise intensity conditions; therefore, H2-1 was rejected.

For the feeling of the passage of time, we hypothesized significant differences between the motion speed and exercise intensity conditions. Responses to “From your physical feeling, how fast is the time passage?” demonstrate significant differences between the different motion speed conditions. Participants felt time passed was faster in the fast-motion speed condition than in the slow-motion speed condition. Therefore, H1-2 were supported. Meanwhile, for the exercise intensity conditions, high exercise intensity conditions led to a faster feeling of the time passage. Therefore, H2-2 is supported.

For H3, we hypothesized that participant’s cycling states would lead to higher duration judgments and faster feeling of time passage. However, our results did not align with our expectations, as the findings indicated no difference in time perception between the cycling and sitting states. One possible reason that in this study the surrounding avatar attracts almost all the attention of the participants is that participants can not spare their attention to cycling or not cycling, so the participant’s state, cycling or sitting, can not affect people’s time perception.

Difference between duration judgment and feeling of time passage

In this section, we discuss the effects of surrounding avatars on time perception. Our results showed that motion speed and exercise intensity played important roles in the participants’ time perception. However, the two attributes had different effects: while the motion speed affected both the duration judgment and the feeling of the passage of time, the exercise intensity only affected the latter. This difference may be due to the different mechanisms between duration judgment and the feeling of passage of time. Droit–Volet et al. examined the relationship between duration judgment and the passage of time with concurrent measurements and duration judgment and found that the feeling of time passage should be considered distinct. Moreover, duration judgment and the feeling of the passage of time lack a direct connection [56]. Another study has also highlighted that duration judgment is affected by the internal clock, and when the pacemaker of the internal clock increases, the emitting pulses speed up, and the duration judgment will be longer. However, the passage of time is affected by emotional valence, arousal, attention to time, and time expectation [57]. As evidence, patients with depression experience a slowing down of time passage, although they do not exhibit any deficit in duration judgment compared to healthy individuals [58].

Based on these ideas, in the next two sections, we separately discuss the effects of the attributes of the surrounding avatars on the two dependent variables.

Relationship between participant’s state and motion speed

Our findings revealed a significant interaction effect between the participants’ states (sitting or cycling) and the motion speed of the surrounding avatar. This interaction affects duration judgment and the subjective feeling of the passage of time; thus, we supposed that it could be shared by the same underlying factor.

Specifically, the avatar’s motion speed and the participants’ state jointly influenced the feeling of time passage and duration judgment. This could be due to the different motion speeds of the surrounding avatars, which attracted the participant’s attention to the VR gymnasium. For instance, as explained in the previous section, factors related to temporal clues affect people’s judgments of duration and their feelings about the passage of time. Therefore, in the condition of fast motion speed, the surrounding avatar’s squatting speed attracts the participant’s attention because there are many temporal clues released by the avatar’s fast motion speed. Thus, whether the participants stated cycling or sitting alone did not affect their time perception. The converse holds in the slower motion condition; the slower motion of the avatar provides less temporal stimulation, allowing participants to be affected by the surrounding avatar and cycling. During cycling, the brain processes more information, including temporal aspects, such as cycling frequency and body rhythm, which may contribute to a heightened perception of time. Previous studies have shown that physical activity can influence the judgment of duration. In contrast, when sitting, the lack of additional stimuli beyond the avatar may result in less pronounced effects on time perception [4, 26].

Compare with actual duration

Our results indicated that all experimental conditions on duration judgment were underestimated, even without the influence of surrounding avatars, participants consistently underestimated the duration in the reference conditions. This finding aligns with previous research on time perception in VR, which can distort an individual’s perception of time, independent of other factors [18]. In VR, people often feel that time passes more quickly because they lose access to natural time cues, such as bodily rhythms, and environmental signals, such as sunlight or other factors. This phenomenon has been observed in similar contexts; for example, Lugrin et al. found that people waiting in VR perceived time as shorter than waiting in the real world under identical conditions [20]. This also explains why nearly all the conditions for duration judgment were underestimated.

Contribution to VR design

Our findings suggest that the surrounding avatars in a VR gymnasium may strongly affect people’s time perception. These findings may contribute to the VR design in two ways.

Limitation and future works

In this study, we gained detailed insights into time perception in VR gyms. We believe that our research results represent a first step toward helping realize the effectiveness of the surrounding avatar while working in VR gyms. However, this requires a long time. This section highlights some of the limitations of this study.

First, the surrounding avatar played the role of a zeitgeber in the VR gyms, affecting participants’ time perception. In this study, squats were selected as the exercise for the surrounding avatar. Compared with squats, which have a regular cycle, other exercises such as running on a treadmill [62], do not have a regular cycle. Thus, it is necessary to investigate whether the attributes of the surrounding avatars in these cases have similar effects in the future.

Second, the effects of motion speed and exercise intensity on time perception may not be linear. We selected a common range of motion, speed, and exercise intensity as the conditions. However, it is necessary to further investigate whether these effects persist under extreme settings. It is possible that when the surrounding avatar’s motion speed and exercise intensity are significantly higher than usual, the effects may diminish or disappear because the participant may not consider the surrounding avatar zeitgebers in such an unusual situation. Therefore, in future work, we will replicate the experiment in a wider setting to explore these effects.

Third, the surrounding avatar was always in a fixed position throughout the trials. This is somewhat artificial compared to a real VR gym situation. A real VR gym might contain many interactions between users and surrounding avatars. Such social behavior might be crucial in time perception because it induces participants to have a better immersive feeling. Additionally, previous studies have shown that time perception is generally associated with the environment and situations [63]. Hence, further research in more interactive settings is required.

Fourth, for the experimental design of each trial, we followed Landeck et al. ’s design in the 20s, 30s, and 40s [36], and used this trial design to simulate the way how the people work on anaerobic exercise in a real gym. However, it was relatively short to measure people’s time perception about the content of exercise. The purpose was to obtain precise time perception and simulate more real situations in a VR gym. However, further experimentation and long-time judgment are required to test the same conditions in a more realistic time condition.