Since 2016, VR technology has rapidly matured as a subset of XR, exhibiting accelerated and diversified applications across industries (Chen et al., 2020; Velev and Zlateva, 2017). However, the success of VR technology relies not only on advancements in hardware and software but also on user perception and interaction experiences. Research has shown that presence and comfort influence XR experiences (Mondellini et al., 2022). Therefore, user experience in VR can be assessed through indicators such as presence, workload, usability, flow, and latency.

Existing studies offer valuable insights into topics such as interface design, cognitive load, time perception, and emotional impact. For instance, Bi et al. (2024) examined how scene-switching and time visualization methods (e.g., electronic clock vs. bomb countdown) affect creativity during brainstorming activities, highlighting how time distortion in VR environments can influence user experience and task fluidity. Hartfill et al. (2024) found that cognitive load decreases when users perceive more control over their virtual avatars, contributing to time compression. Meanwhile, research by Che et al. (2025), van Weelden et al. (2024), and De Witte et al. (2024) suggests that higher levels of immersion do not always enhance user experience, as excessive immersion can lead to increased cognitive load or discomfort. Bartyzel et al. (2025) and Kim et al. (2025) analyzed user performance, workload, and subjective experiences in VR training, offering valuable insights for future VR training system designs.

Given the expanding scope of this field, a comprehensive understanding of user experience is crucial (Rhiu et al., 2020). We adopted the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) protocol (see http://prisma-statement.org) to systematically review current research on VR user experience. Figure 1 describes our screening process. Our objective was to analyze studies examining the impact of VR design on user experience (UX), cognitive load, emotion, and time perception. As Suh and Prophet (2018) recommended, limiting the search to a single database ensures reproducibility, rigor, and transparency while mitigating inconsistencies caused by variations in search functions and algorithms.

We conducted a keyword search in the Scopus bibliographic database using the following search terms: (“VR” OR “virtual reality”), with (“UX” OR “user experience” OR “usability”), (“interface” OR “user interface”), (“emotion*“), (“cognitive load” OR “cognit*” OR “workload”), (“time perception”), and (“design”). The initial search was performed on 19 November 2024. As 2016 is widely recognized as the year in which VR started (Chen et al., 2020; Velev and Zlateva, 2017), only literature published in English from 2016 was included. To ensure relevance, there was a focus on the subject areas of computer science and social science. Note that this review excluded surveys, editorials, conference papers, and documents without abstracts or full texts. Initially, 342 documents met the search criteria. After non-compliant documents were removed, 61 underwent a full-text review, resulting in 15 relevant UX studies for VR design.

Table 1 summarizes selected literature discussing user experiences in VR. These studies span various countries and industries, including entertainment, healthcare, education, gaming, and social applications. Topics include research on visual stimulation intensity in VR environments (Asish et al., 2022; Batistatou et al., 2022; Chiossi et al., 2022; Latini et al., 2024), multimodal interfaces (Dzardanova et al., 2024; Jacucci, 2017; Yuan et al., 2023), auditory stimulation (Bosman et al., 2024; Picard et al., 2023), physiological signals (Qu et al., 2022; Raees and Ullah, 2020), dissimilar avatars (Cheymol et al., 2023), and the number of users in VR environments (Birt and Vasilevski, 2021). Despite significant progress, many research gaps remain in interface design. For example, researchers have investigated waiting experiences in mobile apps (Chen and Li, 2022; Cheng et al., 2023; Pibernik et al., 2023) and customer service robots (Wintersberger et al., 2020); however, waiting experiences in VR devices have yet to be examined (Heidrich et al., 2020). As previously mentioned, existing HCI models may be insufficient to fully explain the complexities of VR environments. UX research can contribute to refining our understanding by empirically investigating ways to enhance user satisfaction, reduce cognitive load and adverse effects, inform VR system development and design, establish standardized evaluation methods, and explore multimodal sensory interactions (Mondellini et al., 2022). Therefore, UX research should be considered essential in the development of any VR application.

SOR theory is a psychological model widely applied in consumer behavior research. Its core concept posits that a stimulus influences an individual, leading to a response (Mehrabian and Russell, 1974). According to Jacoby (2002), the stimulus (S) represents the environment an individual encounters at a specific moment, while the organism (O) encompasses various internal states such as attitudes, beliefs, values, motivations, personality, knowledge, experiences, emotions, tendencies, and cognition. Vieira (2013) further explains that the response (R) reflects an individual’s willingness or intention to enter or exit a particular environment. Later, Bitner (1992) incorporated cognition and physiology to expand the SOR theory to the service domain. Past studies have used SOR theory to elucidate behavioral dynamics in XR technologies (Mala Kalaiarasan et al., 2024) and to examine the impact of VR on service perception, emotions, and behavior. Examples include virtual presentations of tourist hotels (Surovaya et al., 2020), virtual tourism experiences (Kim et al., 2020; Latifi et al., 2024; Zhu et al., 2023), and museum exhibition experiences (Chang et al., 2018). Thus, SOR theory is an appropriate framework for the exploration of users’ perceptions and responses to VR stimuli. The current study is the first to apply this framework to VR game-loading interfaces.

Figure 2 shows the research framework of the current study.

This study used SOR theory to explore the psychological and behavioral dynamics of waiting experiences in VR games. Similar to previous studies that used specific designs as environmental stimuli (Chang et al., 2018; Kim et al., 2020), this study considers scene interactivity, waiting length, visual stimulation intensity, and background environments. Following Godefroit-Winkel et al. (2022) and Su et al. (2020), this study viewed individual perception as an internal processing outcome, reflecting users’ subjective time experience and perceived emotions in response to various stimuli in VR environments. Additionally, this study includes cognitive load as a behavioral response, as measured by physiological signals collected using the VR headset. Suh and Prophet (2018) also considered cognitive overload a reaction in their systematic review of the applications of SOR theory to immersive technologies.

Emotions are crucial in user research due to their influence on behavior, perception, and cognition to varying degrees (Somarathna et al., 2023), as well as on distortion of users’ time perception (Cui et al., 2023). Gable et al. (2022) found that the motivational direction of emotions affects time perception. For example, emotions associated with approach motivation, such as anger, can accelerate time perception due to strong goal orientation. Conversely, emotions linked to avoidance motivation can extend time perception, showing how emotions regulate time perception through their motivational direction. Emotional responses have been shown to enhance presence in VR (Baños et al., 2004; Riva et al., 2007) and significantly impact the enjoyment and quality of the VR experience (Wienrich et al., 2018).

Past VR research has focused on how sensory environments influence emotional arousal. For example, Batistatou et al. (2022) found that participants experienced more pleasant emotions in green environments. van der Ham et al. (2019) noted that appropriate sound design could enhance users’ positive emotions or reinforce specific emotional states, aiding deeper connections with VR content and narratives. Liao et al. (2020) confirmed the impact of multisensory (visual and auditory) stimuli on emotions. Jacucci (2017) studied how multimodal synthesis (such as haptic and facial expression stimuli) can recognize and influence participants’ emotions. Previous studies have explored the effective use of VR as an emotion induction mechanism (Bosman et al., 2024; Dey et al., 2022; Somarathna et al., 2023); however, few studies have focused on emotional changes during VR waiting times. Waiting in VR is often more unpleasant for gamers because they are awkwardly confined by the headset (Zwiezen, 2020). As such, users cannot divert their attention to escape the loading screen, potentially triggering negative emotional experiences (Heidrich et al., 2020).

Particularly in large VR games, waiting times often exceed users’ comfortable threshold of 10 s. In these games, it can take more than a minute for a player to enter. This prolonged waiting creates cognitive friction, where users face unexpected outcomes from seemingly intuitive interfaces or functions. The mismatch between expected and actual results leads to frustration (Ericson, 2022), often triggering negative emotions during these long waiting experiences. Cheng et al. (2024) found that interactions with features like doodles and emojis in VR can enhance positive emotions. This suggests that different interaction features can evoke varying emotional states. Based on the aforementioned literature, it seems different stimuli trigger different emotional values. Thus, this study proposes the following hypotheses:

H4a: In interactive interfaces, different visual stimulation intensities during waiting will lead to differences in positive emotions.

Cognitive load refers to the limited capacity of human cognitive resources, which can be seen as working memory. Executing tasks occupies working memory. A well-designed system allows users to focus on the task and spend less mental effort on irrelevant aspects (Sweller, 1988). Kleygrewe et al. (2024) pointed out that users’ VR experience and gaming frequency influence their cognitive resource consumption in VR. Compared to typical HCI scenarios, VR covers the user’s visual field, creating unique effects on human vision. Compared to the experience of using a 2D graphic website, VR exposes users to more sensory stimuli, requiring them to allocate more attention and cognitive resources to the game (Fisher et al., 2018) This affects the size of the cognitive load experienced by users.

There are still research gaps regarding cognitive load in the HCI field. As more systems compete for users’ attention, a better understanding of cognitive workload becomes critical in HCI research (Kosch et al., 2023). As the processing of visual stimuli requires a certain amount of cognitive resources, different types and complexities of visual messages lead to varying levels of cognitive load (Valtchanov and Ellard, 2015). Cognitive load has often been used as a metric to explore the level of impact of visual stimulation on users. For example, engaging visual details in VR environments may cause additional cognitive load (Cruz et al., 2023). When cognitive load is increased, users often experience symptoms similar to visual fatigue (Gowrisankaran et al., 2012). When green elements are present in a VR environment, users experience lower cognitive load and more efficient information searching, improving overall performance (Latini et al., 2024). Chiossi et al. (2022) also found that reducing visual complexity can improve cognitive load. Additionally, previous research has shown that users facing higher cognitive load during waiting perceive time as passing more quickly (Chen and Li, 2022; Lallemand and Gronier, 2012). The literature thus confirms that visual resources such as visual stimulation intensity and spatial vision utilize working memory. Interactive environments may divert users’ attention, affecting cognitive load (Shelton et al., 2021). Therefore, improving visual presentation efficiency and reducing cognitive resource usage are crucial from a design perspective. This study thus uses cognitive load as an evaluation index to explore the impact of visual stimulation on users. Based on the above theories, this study proposes the following hypotheses:

H5a: In interactive loading interfaces, different visual stimulation intensities will lead to differences in cognitive load.

Methods for measuring cognitive load are generally classified into physiological indicators (Gupta et al., 2019; Lee, 2014; Souchet et al., 2022) and self-report assessments (Shelton et al., 2021). Physiological measurements capture cognitive load through eye-tracking, heart rate variability (HRV), electroencephalography (EEG), respiration, and electrodermal activity (EDA), providing objective insights into cognitive states (Paas et al., 2003; Whelan, 2007).

Self-report assessments primarily rely on questionnaires, the most well-known being the NASA Task Load Index (NASA-TLX) developed by Hart and Staveland (1988). Originally designed to evaluate pilots’ mental and physical workload, this questionnaire includes six dimensions: mental demand, physical demand, temporal demand, performance, effort, and frustration (Hart and Staveland, 1988). In VR cognitive load research, NASA-TLX has been widely used for subjective workload measurement (Bi et al., 2024; Che et al., 2025; Chiossi et al., 2025; De Witte et al., 2024; Hartfill et al., 2024; Kim et al., 2025; van Weelden et al., 2024; Vorwerg-Gall et al., 2023). Derived versions include the Raw Task Load Index (RTLX) (Hart, 2006), which has been broadly validated (Lovasz-Bukvova et al., 2021; Rodríguez-Fernández et al., 2024), and the Simulation Task Load Index (SIM-TLX), which incorporates task complexity and situational stress to better align with VR cognitive load measurement (Harris et al., 2020; Urbano et al., 2024).

Some scholars have raised concerns about the applicability of self-report questionnaires. Buchner et al. (2025) argued that contextual constraints limit self-report measures, whereas physiological measurements enable non-intrusive cognitive state assessment (Halbig and Latoschik, 2021). Foy and Chapman (2018) similarly highlighted that self-reports suffer from subjectivity, lack of real-time synchronization, and an inability to simultaneously capture subjective ratings during simulation-based training. Additionally, Kosch et al. (2023) suggested that the widespread use of NASA-TLX in HCI applications may be more a result of historical precedent and convenience rather than because it is the optimal measurement tool. They argued that the questionnaire is prone to individual biases and lacks tailored design and empirical validation for HCI environments. In contrast, physiological signal measurements provide a more comprehensive approach to capturing real-time feedback (Zagermann et al., 2016). Furthermore, physiological indicators of cognitive load represent genuine physiological responses to real-world stimuli, aligning with the SOR theory framework adopted in this study (Suh and Prophet, 2018).

This study thus used the HP Reverb G2 Omnicept, a VR headset with physiological signal measurement capabilities. Cognitive load was measured on a scale from 0 to one through a machine-learning model built on data from 738 subjects, as detailed in HP’s white paper (Company, 2021). In reports from the HP research team, their classification accuracy reaches 79.08% (Siegel et al., 2021). In recent years, many VR studies have utilized this headset for cognitive load measurement (Ahmadi et al., 2023; Lataifeh et al., 2024; Reddy et al., 2022). Reddy et al. (2022) validated the feasibility of using non-invasive sensors, such as the HP Reverb G2 Omnicept, for estimating cognitive load in VR. Their study also identified pupil diameter variation and fixation count as reliable indicators for assessing task complexity and cognitive load.

The pilot study involved semi-structured interviews with five game development experts to design the experimental system. Subsequently, researchers recruited 20 participants to explore the differences in users’ time estimates and emotions between interactive and non-interactive loading interfaces under varying waiting times.

For both types of interfaces, researchers set the waiting times at 15 s, 30 s, and 60 s. The experimental data for this phase are shown in Table 2. As the waiting time increased, players’ perceptions of waiting time also increased. However, time perceptions for the interactive interface were lower than those for the non-interactive interface. Therefore, H1a was supported. Additionally, different loading interfaces significantly affected time perceptions at 15 and 60 s (p < 0.05). Therefore, H2a was supported. That is, the longer the waiting time, the more significant the difference in time perceptions for interactive interfaces compared to non-interactive interfaces.

The author then explored the impact of different loading times (15 s, 30 s, and 60 s) and interface interactivity (non-interactive vs. interactive) on the subjects’ emotions, the results of which are shown in Table 3. For 15 s of waiting time, the difference in interface type triggered positive emotions. The statistical analysis shows that the Z value was −2.153 with a p-value of 0.0313, while negative emotions showed a Z value of −3.209 with a p-value of 0.0013. For 30 s of waiting time, the Z value for positive emotions was −3.416 with a p-value of 0.0006 and that of negative emotions was −3.258 with a p-value of 0.0011. Finally, for 60 s of waiting time, the Z-value for positive emotions was −3.727 with a p-value of 0.0002 and that of negative emotions was −3.658 with a p-value of 0.003.

These results support H1b, which states that the interactive interface will engender more positive emotions and less negative emotions than the non-interactive interface. In addition, these results also support H2b, which states that the longer the waiting time, the greater the difference in emotion for the interactive interface compared to the non-interactive interface.

The main objective of the pilot study was to investigate the relationship between time perceptions and emotions for different waiting times and interactive vs. non-interactive loading interfaces. The results of the study revealed the following: (1) an interactive loading interface can shorten perceptions of waiting time; (2) an interactive loading interface is highly effective at shortening perceptions of waiting time for long waiting periods (60 s), and the interactive interface generally elicited shorter perceptions of waiting time than did the non-interactive interface; and (3) an increase in waiting time leads to an increase in perceptions of waiting time, and the longer the waiting time, the more positive emotions decrease and the more negative emotions increase. In addition, using an interactive interface is more likely to maintain positive emotions and reduce negative emotions than using a non-interactive interface.

Through observation of the participants, this study found that after waiting for a certain period, the participants started to pay attention to external factors such as the environment and objects in the loading interface. Some respondents thought that the difference between interactive and non-interactive interfaces was insignificant in terms of time predictions. Post-experiment interviews suggested that this might be related to users’ familiarity with the long waiting times of VR games. This study also found that the level of players’ engagement in the previous game may have affected their waiting experience in the next game.

The formal study incorporated additional visual stimulation elements to further explore their effects on time perceptions, emotions, and cognitive load.

In this stage, the authors analyzed the influence of stimuli of different intensities on the interactive and non-interactive loading interfaces by averaging the trends of the dependent variables. The results presented in Table 4 show that the lowest cognitive load was found for environmental details (None) x visual stimulation (None) in the non-interactive interface (0.521), and the lowest cognitive load was found for environmental details (None) x visual stimulation (Yes) and environmental details (Yes) x visual stimulation (None) in the interactive interface (0.54).

The effects of different visual stimulation intensities on time perceptions (F (3,31) = 0.447, p = 0.721) and cognitive load (F (3,31) = 0.503, p = 0.683) were not significantly different for the interactive interface. Therefore, H3a and H5a were not supported. In the case of the non-interactive interface, visual stimulation intensity was not significantly different for time perceptions (F (3,31) = 0.902, p = 0.451); thus, H3b was not supported. However, cognitive load reached a significant difference (F (3,31) = 2.892, p = 0.039), thereby supporting H5b.

The emotion questionnaire consisted of 14 questions on positive and negative emotions. The questionnaire used a 7-point Likert scale, and reliability was confirmed with a Cronbach’s α coefficient of 0.758. The results show that different visual stimulation factors affected positive and negative emotions for both the interactive and non-interactive interfaces. The means of positive and negative emotions were respectively 2.85 and 2.25 in the absence of visual stimulation without environmental details; however, when both visual stimulation and environmental details were present, positive emotions decreased slightly to 3.08, and negative emotions increased slightly to 2.17, which implies that visual stimulation intensity may affect participants’ emotions. Detailed data are shown in Table 5. In the case of the interactive interface, visual stimulation intensity did not reach a significant difference for positive emotions (F (3,31) = 1.15, p = 0.328), which means H4a was not supported. Negative emotions (F (3,31) = 3.109, p = 0.039) reached a significant difference, providing support for H4c. For the non-interactive interface, visual stimulation intensity was not significant for both positive (F (3,31) = 1.425, p = 0.24) and negative (F (3,31) = 1.655, p = 0.182) emotions; thus, H4b and H4d were not supported.

Based on SOR and time perception theories, this study explores how sensory stimulation in VR game loading interfaces affects individuals’ internal states (time perceptions and emotions) and responses (cognitive load). Previous studies on loading interfaces often used simulations. They thus lacked comprehensive research on waiting scenarios in real VR games, typically focusing on 360-degree image interactions or optimizing algorithms to enhance viewing smoothness (Bendre et al., 2024; Sun et al., 2018; Zeynali et al., 2024). This study fills the research gap on loading interfaces in VR environments.

As Suh and Prophet (2018) and Kourouthanassis et al. (2015) have suggested, sensory stimulation is crucial for enhancing user experiences in immersive environments. Visual stimulation in immersive technologies evokes cognitive and emotional states, leading to behavioral changes. This study considers the effects of visual stimulation and interactivity, combining the characteristics of VR and interactive waiting experiences in the gaming industry. This study expands the research field of loading interfaces by exploring the impact of interactivity and visual stimulation during waiting times.

This study applied SOR theory to validate the feasibility of designing interactive loading interfaces for VR games and the effectiveness of incorporating psychological stimuli related to waiting into game-loading interfaces. The results confirmed that such effects directly result from the stimuli on the individual’s internal and external states. By adding interactive elements and environmental details during waiting times, this research significantly improved players’ waiting experiences and mitigated the adverse effects caused by prolonged waiting periods. Moreover, physiological data measurements indirectly showed that the responses triggered during VR game waits might align with similar paradigms found in past research.

Compared to the non-interactive loading interface, the interactive loading interface resulted in shorter time perceptions, increased positive emotions, and decreased negative emotions. As waiting times increased, the interactivity of the interface increased the variance in time perceptions and emotional responses. In the interactive interface, players’ attention was focused on the interactive elements; therefore, after a certain threshold, the effect of visual stimuli on time perceptions and cognitive load decreased. For the non-interactive interface, increases in visual stimulation were associated with increases in cognitive load.

Visual stimulation elicited both positive and negative emotions. Players’ interactions also seemed to amplify the effects of visual stimuli on negative emotions. Interactions may also satisfy players’ intrinsic motivation and trigger expectations for further interactions with other visual stimuli; thus, attention should be paid to players’ expectations when designing interfaces with visual stimulation.

First, this study focused on the direct impact of stimuli on internal and external variables within SOR theory instead of treating individual internal responses as mediating variables. Future research should consider the effects of time perceptions and emotions on cognitive load. Additionally, to enable the participants to master the game quickly, the game levels in this study were designed to be less challenging. This may have affected players’ intrinsic motivation. Further research is necessary to examine the consistency of the findings across different game genres. In terms of experimental design, we used repeated testing. However, the long experimental time may have triggered the anchoring and practice effects of repeated experimental testing (Liu et al., 2024). Experimental times and randomization should be carefully considered in future studies.

Additionally, this study did not measure a baseline for the dependent variables before the experiment, which may affect the interpretation of results. Participants’ initial states (e.g., emotion and cognitive load) could influence their perception of time, as psychological factors such as anxiety or focus level may impact time perception independently of the VR environment manipulation. Future studies should consider incorporating baseline measurements—such as static waiting tasks or time estimation in a quiet environment—before the experiment to ensure that subsequent changes in measurement can be attributed to the VR interface manipulation rather than individual differences.

Furthermore, future research could adopt a more comprehensive approach in terms of cognitive load measurement. While some scholars question the reliability of subjective measures (Buchner et al., 2025; Kosch et al., 2023), many studies suggest combining subjective and objective assessments to provide a more holistic analysis (Ayres et al., 2021). For instance, existing research widely employs NASA-TLX alongside physiological measurements, including electrodermal activity (EDA) (Armougum et al., 2019; Chiossi et al., 2022; Kim et al., 2025), heart rate variability (HRV) (Che et al., 2025; Reddy et al., 2022), electroencephalography (EEG) (Chiossi et al., 2025; van Weelden et al., 2024), and electrocardiography (ECG) (Weiß and Pfeiffer, 2024). Future research should integrate baseline measurements with multimodal cognitive load assessment to enhance data reliability and interpretability. Furthermore, the selected measurement methods may have failed to capture subtle differences in time perceptions due to visual stimulation; future research could include more refined measurement methods or more sensitive experimental designs to detect the effects of visual stimulation on time perceptions.

Because individual player preferences may have affected players’ responses to the stimuli presented on the loading interface, we suggest that the population of subjects be further categorized to increase the accuracy of the findings. Numerous VR studies have focused on the interaction effects of multimodal sensory experiences on user experience (Dzardanova et al., 2024; Jacucci, 2017; Yuan et al., 2023). Future experiments could consider incorporating auditory (Bosman et al., 2024; Picard et al., 2023) and tactile interactions as stimuli. Finally, since the stimuli in waiting scenarios are influenced by individual player experiences, we suggest that future studies could further segment the participant groups to derive more accurate conclusions.