The virtual reality (VR) technology market has been rapidly expanding with a projected global growth from $166.7 billion in 2022 to $2273.4 billion in 2029 [33]. Virtual reality has emerged as a significant asset across a diverse range of domains, including research [13, 43], healthcare [2, 28, 48], training [24, 32], education [17, 25], and entertainment [22, 23] as it offers a richer, more immersive, and interactive experience. However, a notable challenge in virtual environments (VEs) is user guidance. Where should the user move next? Where should the user look next? Is there an important alert to attend to? What if it is behind the user? Effective guidance mechanisms are essential for user experience in applications ranging from medical training to interactive educational experiences such as exhibitions. Without effective guidance, users may experience frustration, reduced engagement, or insufficient navigation, which limits the utility of VEs in critical and demanding scenarios.

This research focuses on informal educational environments, particularly virtual exhibition spaces such as museums, which enable people to visit them without travel. The design of these experiences poses an interesting human–computer interaction (HCI) challenge to balance users’ freedom with the agenda of the exhibit designer. Users must be able to explore on their own and engage with exhibits that interest them while at the same time, the exhibit designers may have a goal; for example, “I want patrons to learn how important water is to their daily lives and several ways to save water” or “I want patrons to appreciate the dramatic change between Romantic artists and Realist artists and Realism’s influence on social class structures.” Previous research on guidance systems in real museums has demonstrated that most users value their freedom and resist following a set path [16, 36]. An exhibit will often evoke emotion and even a strong sense of presence or flow [40] and as such, it is important for the user interface to be as simple and unobtrusive as possible so that it does not interrupt that experience [10]. This goal overlaps with some of Weiser’s goals of ubiquitous computing [64], which include providing invisible, seamless, or natural interfaces for systems that surround us. It also overlaps with the concept of digital nudging [63]—designing a system that offers users a choice while strongly favoring that they make certain choices over others. The virtual exhibition designer may want to give patrons freedom while nudging them toward certain goals. While visitors in a real exhibition will likely augment their experience with at most their phone and headphones, virtual exhibitions offer many more options for these seamless guidance interfaces. Thus, this research seeks to answer two research questions: What are the potentially suitable types of seamless guidance cues (visual, auditory, and haptic) for virtual exhibitions? How do different types of guidance cues impact users’ ability to locate visual targets in a virtual exhibition?

Multiple types of guidance exist in VEs, which include the following: (1) direction of what the user should do next [34, 66], (2) directing the user to a particular locomotion path [14, 60, 62], and (3) directing the user to attend to specific visual targets [7, 49]. This research focuses on the last category. Unlike more typical visual guidance in the form of alerts (e.g., in a nuclear power plant control room, an airline cockpit, or a hospital surgery), where compliance with the alert guidance is highly consequential, guidance in virtual exhibitions must remain optional and almost invisible. As noted by previous researchers, providing guidance to specific visual targets without constraining the user can pose a challenge in balancing the salience of the guidance cue with the presentation of a freely browsable virtual environment [49].

Previous guidance systems for multiple visual search tasks span a range of sensory modalities, typically visual, audio, and haptic. Although significant research has been conducted to explore the potential of these cues across modalities, each study has its limitations. For example, Felix et al. primarily altered the field of view to explore visual guidance but neglected the rear space [7]. Similarly, although some research has investigated the use of mirrors for rearview guidance, the applications remain limited. Ha et al. employed a mirror to extend the field of view for assembly tasks [20], focusing on a very specific and close space in the front view. Boonsuk et al. used a mirror to detect ten identical red barrels in the rear space [6], where uniform targets with the distinct color help make them easier to locate. Additionally, many studies focused narrowly on one sort of guidance cue. For example, Rebai et al. explored visual guidance cues with a focus on rear space but omitted auditory or haptic comparisons [53]. Some studies focused solely on active cues and overlooked the potential of passive aids [11, 38, 54, 68]. Moreover, several lacked a control condition where no cues were provided, making it difficult to evaluate the inherent challenges of the task or the environment itself [54, 68].

To address these gaps, the present study compares a control condition, a passive visual aid, active audio alerts, and active haptic alerts to evaluate their effectiveness in guiding users to a collection of specific visual targets. This study uses a scavenger hunt task to simulate real-world search scenarios, which offers a unique approach that balances user autonomy with the effectiveness of cue salience.

For this study, participants were tasked with finding seven specific paintings within a virtual environment of the remodeled Louvre Museum. Their performance was measured by (1) how often they referred to the task list of seven to assess how confident they felt with the received cue to finish tasks while navigating and (2) the total time taken to complete the task to analyze the extent to which cues helped participants quickly find target paintings among others. The primary hypothesis of this study is that adding visual, auditory, and haptic cues can help users better locate visual targets in a virtual space compared to solely relying on the front view. In addition, to gain insight into the relationship between the three guidance systems and a quality experience in the VE, other metrics were evaluated, including the level of workload, sense of presence, and cybersickness of users. Moreover, users’ satisfaction with the guidance cues’ efficiency and effectiveness was evaluated through qualitative open-ended questions.

This study contributes to the field of HCI and VR research, particularly in the design of guidance interfaces for immersive environments. Specifically, the study identifies three potentially suitable types of seamless guidance in VR environments: passive visual aids, active audio alerts, and active haptic alerts. Furthermore, it demonstrates that haptic and audio cues are effective for designing guidance interfaces in immersive VR environments through a task-based experiment. The findings are particularly relevant to virtual museums, educational VR applications, and interactive simulations, offering practical guidance for improving user experience in informal learning environments.

Attention orientation is important for virtual tasks as it impacts the ability to process information, navigate environments, and perform tasks efficiently [4]. It involves focusing on specific tasks or objects while disregarding irrelevant information [14], which makes it challenging to measure accurately [58]. Traditional methods, such as self-reports and observations, often capture partial data due to the dynamic nature of VR environments [61]. Previous VR research has explored a variety of methods to guide attention orientation [5, 11, 13, 35, 41, 52]. Typically, attention is measured by the total time users require to complete tasks or the number of right and wrong responses [3, 5, 12, 52, 68]. However, as highlighted in Carlos Lievano et al.’s comprehensive literature review, these methods need to account for factors such as individual differences among users, experimental design and settings, and the characteristics of extended reality (XR) technology in each study [58].

Sensory cues are a key aspect in shaping attention orientation in VEs [37]. These external stimuli, such as visual, auditory, haptic, olfaction, and gustation, can be used to guide users’ attention toward specific points of interest. However, they play varying degrees of importance and are particularly critical to navigation and completing tasks due to the challenge of perceiving distance, motion, and direction in VR environments [13, 32]. Among sensory cues, visual, auditory, and haptics appear to have been studied the most [4, 11, 32, 37, 39, 43, 62].

Vision is the primary sensory stimulus humans rely on for interaction with the world [31]. Researchers have investigated the importance of vision in virtual navigation and performance through different perspectives. For instance, a study on the effect of different field of view angles (102∘, 52∘, and 32∘) on a visual search task for training suggested that a higher field of view led to better training effectiveness and virtual performance [45]. A study evaluating gaze-aware visual cues on driving within a VR game showed that the cues improved driver attention and response to unexpected pedestrian crossings [8]. Another study compared the impact of visual cues on finding targets within versus outside the visual field. The results showed that cues significantly improved performance when targets that needed to be located occurred outside the visual field [53]. In addition, some studies introduced the concept of a rearview mirror in VR. Ha et al. presented a virtual mirror to uncover hidden spatial information, enhancing precision in 3D manipulation. The mirror was automatically controlled and positioned where it could highlight the spatial relationship between the manipulated object and the other objects close to it. The results showed that the mirror was significantly helpful for users in manipulating 3D objects [20]. Another study compared user interfaces with a 90∘ rearview mirror to a 180∘ rearview mirror in a scavenger-hunt-style target location task. The results showed that although participants primarily used the front views to acquire targets, they did sometimes use the rearview mirrors [6]. These studies suggested that besides the default front view, a rearview mirror that extends a user’s view of the world could be beneficial in a visual search task. Based on this research, the present study chose to include a rearview mirror as one of its guidance systems.

Audio can also play a significant role in attention orientation, particularly in VEs. Researchers have explored its impact in different scenarios to highlight its effectiveness in directing attention [11, 38, 57]. A study comparing participants’ attention in both static and dynamic VEs with and without 3D sound cues found that audio cues enhanced participants’ attention when scanning for targets [38]. Suzuki et al. introduced a novel method to study sound localization in VEs. The “active listening” method leverages dynamic spatial information generated by participants’ movements to improve spatial hearing analysis [57]. The orientation of attention through audio localization has been further investigated using different audio modalities, such as mute, mono, and ambisonics [11]. Research found that salient object sounds had the greatest effect in directing one’s attention toward the source [11]. Based on these studies on audio cues, the present study chose to include a simple audio proximity alert as one of its guidance systems.

Haptic cues, particularly vibrotactile feedback in VR, represent another influential cue in orienting attention. Vibration belts with directional cues could help prevent collisions with obstacles in an immersive virtual set [68]. Hong et al. explored the potential of vibrotactile simulations in redirecting visual spatial attention during a visual task. The findings suggested that when haptic cues are valid, they significantly improve response time, indicating the cues’ impact on attention enhancement [59]. Another study demonstrated that short, repetitive vibrations could reduce cognitive load if users’ gaze were guided toward nearby interactive objects to correct inaccuracies [54]. Localization of haptic cues has also been examined under two conditions with high and low validity. The results indicated that high validity of haptic cues oriented one’s attention more naturally and intuitively than low validity [68]. Based on this research, the present study chose to include haptic vibrotactile proximity alert as one of its guidance systems.

A variety of studies have also focused on the combination of such cues [12, 28, 43, 47]. For instance, one study conducted experiments involving multiple simultaneous visual search tasks where visual, audio, and vibrotactile cues were applied. The findings indicated that on-head vibrotactile cues effectively guided users’ attention toward targets without causing any distraction in other simultaneous tasks when compared to audio and visual cues [12]. Another study focused on multisensory integration suggested that multisensory stimuli could be significantly helpful in target detection only if the environment and task required a high perceptual load [37]. Metrics such as task-completion time, identification accuracy, and range of head movement were monitored to assess these effects. The variability in findings suggests that cues may perform differently depending on scenarios and combinations of other cues. To keep the present study simpler and focus on the impact of each cue, the authors did not combine cues. However, the choices of metrics and dependent variables used in previous studies influenced the variables measured in the present study.

Several studies have explored the role of sensory cues, particularly in virtual museums. A mixed-method study of three Indonesian museums examined how virtual multisensory spaces are perceived, focusing on sensory systems as the key factor influencing respondents’ experiences. The findings revealed that visual and auditory senses dominated virtual spatial experiences while chemical senses were weaker [15]. Similarly, Guo et al. identified the prominence of visual and auditory cues over haptic and taste cues in creating a holistic digital museum experience. Additionally, it demonstrated that emotional state and sense of presence mediate the impact of multisensory cues [19]. Another study introduced a multisensory virtual museum experience by integrating tactile feedback via 3D-printed artifacts and sensors, enhancing the sense of presence. Experimental results from 32 participants demonstrated that this multisensory approach significantly improved realism, immersion, and user preference compared to traditional audiovisual methods [27]. Although these studies highlighted the potential of sensory cues in creating an immersive virtual museum experience, they neglect to investigate the impacts of individual sensory cues on achieving a seamless experience. Moreover, they focus primarily on immersion and presence, overlooking the critical aspects of how sensory integration affects visitors’ sense of freedom and engagement with the virtual environment, which are essential for a comprehensive experience.

The present study seeks to compare passive visual aids like the rearview mirror with active attentional guidance in the form of audio and haptic alerts in the context of a scavenger-hunt-style multiple visual search task (e.g., find seven objects in any sequence). Although extensive human factor research exists in the design of effective alerts (e.g., for medical conditions in a surgical room [39], airline cockpits [44], and nuclear power plant control rooms [69]), these alerts usually serve as alarms. In a scavenger-hunt-style task, where the goal is to allow the user to freely explore while more subtly alerting them to objects of interest, the fields of ubiquitous computing and ambient information systems are perhaps more relevant. These fields focus on methods of alerting users to new information in the background without distracting focus from a primary task [42]. Based on the four dimensions of ambient systems [42], the authors chose simple alert systems for this initial study. The alerts had low information capacity (“a target is near”), medium notification level (a single sound or vibration), low representational fidelity (the alert did not resemble the target), and low aesthetic emphasis (the alerts were not integrated with the artistic or architectural design of the VE).

This study uses a mixed-method approach aimed at uncovering the effectiveness of different cues in orienting attention and task completion in a virtual recreation of the Louvre Museum. The investigation combined quantitative assessments of task performance with qualitative analyses of participants’ survey and questionnaire responses following the VR task, providing a comprehensive understanding by capturing both measurable outcomes and subjective experiences.

As illustrated in Figure 1, participants were first introduced to the task and the target paintings, followed by a practice session in a simple virtual room to familiarize themselves with the controls. They were then randomly assigned to one of four conditions, each featuring a distinct type of cue. Upon completing the VR experiment, participants filled out a questionnaire to provide feedback on their experience. The following sections present a detailed description of the task design, study design, study variables, and the data analysis approach.

The purpose of this study was to compare the effect of three sensory cues on attentional orientation in VR. Of the 71 participants, three chose to stop the task before completion due to cybersickness. Their performance data were excluded from our analysis, resulting in n = 17 for each of the four groups.

In line with the previous literature [12, 32, 45, 55], the findings of this study indicated that the guidance cues impacted users’ attention orientation. The audio and haptic cues (active cues) significantly reduced the frequency of referring to the task list, which supported our predictions. However, the rearview mirror condition (passive cue) had unexpectedly high task list use and time to complete compared to the active alert systems. A reason for this could be that the interface design or mirror placement made users less certain about the location of the painting targets or the mirror partially blocked their view, and they had to double-check their task list more regularly than with the active alert conditions. In terms of time for task completion, the rearview mirror required significantly greater time than even the control condition as well as the other two conditions. It may have added extra cognitive or visual demands, slowing down participants’ performance. On the other hand, the audio and haptic conditions did not significantly differ from the control condition, indicating that these sensory cues did not drastically impact the task-completion time while adding some value in terms of guidance. This result suggests that the audio and haptic interfaces are closer to the desired seamless interaction that offers the user additional guidance without breaking the sense of presence. This result also contrasts with previous studies [59], suggesting that user differences, task characteristics, and environment may play a crucial role [1]. Further investigation is needed to explore these influences in more detail.

The survey responses resulted in two core themes: engagement and effectiveness. These two themes were crucial for assessing the effectiveness of cues because they showed that guidance cues impacted the participants’ experience interacting with the VE and completing the task. Almost all users in the four conditions enjoyed the task they were assigned to, as they all mentioned in the survey, unaware of the existence of other conditions. They were engaged sufficiently to finish the task successfully regardless of the number of times they referred to the task list or the speed at which they finished the task. Survey responses indicated that haptics and audio were considered highly effective, especially for locating the more challenging paintings, which were those with intricate details and multiple objects, as opposed to simpler paintings featuring a single figure with few elements. Moreover, the paintings that remained elusive after longer exploration were primarily found by relying on the receiving haptic or audio cues; five participants mentioned that they were waiting for the cue to assist them to locate the last painting. Notably, this type of experience was noted in only a small number of participants in the rearview mirror condition. It is also notable that some users explained that they forgot to check the mirror and just relied on their front view. This result might stem from the habit that people use a rearview mirror primarily for safety reasons in vehicles rather than as a tool to locate targets.

The three questionnaires did not result in statistically significant differences between different conditions but offered important insights into the investigated cues on the designed task in VR. The results of NASA-TLX indicated that none of the cues resulted in a significantly increased workload even in the rearview mirror condition, where participants referred to the task list the most and spent a longer time to complete the task. Some variables had means that seemed notably different, but the high standard deviations removed the possibility of statistically significant differences. However, this pattern suggests that with a greater sample size, the mean differences might become significant; these areas are thus worthy of further research. Examples include the NASA-TLX Physical Demand and Frustration (with mirror having larger means than other conditions) and Temporal Demand (with control having a larger mean than other conditions). The mean presence score was lowest with the mirror condition and highest with the haptic condition (though not significantly different). The higher presence with the haptic condition is consistent with previous findings [18]. The audio condition also performed well, highlighting its potential to enhance immersion [26]. However, when confronted with a different scenario featuring a more challenging task or graphically different VE, these cues might cause a different sense of presence, necessitating further investigation.

Systems for attention orientation are crucial in VR tasks that require the guidance of users, and a balance of offering seamless, invisible guidance while still offering users freedom to explore is critical in virtual exhibition environments. The findings of this study suggested that the active alert systems (audio and haptic cues) reduced the users’ need to refer to the task list; they confidently counted on receiving help throughout the entire process with minimal interruptions. On the other hand, participants of the passive guidance system (rearview mirror condition) most frequently checked the task list to complete the task. Furthermore, the total time spent in the virtual Louvre Museum to locate all targets and complete the task was significantly longer in the rearview mirror condition compared to the other three conditions, possibly due to the mirror’s limited effectiveness in guiding users and partially obstructing the front view.

The survey results highlighted that the haptic and audio cues were the most favorable cues in finding elusive targets after extended exploration, with some participants even waiting for cues to guide them to the target’s location. These cues seemed to successfully achieve the desired balance of seamlessness while not inhibiting user exploration of the environment. Since this study focused on a virtual museum setting, caution is needed when generalizing the results to other scenarios, as the novelty of different VEs and task requirements may influence outcomes. However, the findings of this research emphasize the potential of well-designed guidance cues and offer VR designers guidelines for designing VEs that require a balance of user freedom and guidance.