Development and Validation of a Tool for VBOI (Virtual Body Ownership Illusion) Level Assessment

Abstract

Virtual Body Ownership Illusion (Virtual BOI) refers to the perceptual, cognitive, and behavioral changes that occur due to the illusion that a virtual body is one’s own actual body. Recent research has focused on inducing Virtual Body Ownership Illusion (Virtual BOI) using various physical conditions of VR environments such as haptic feedback and 360-degree immersion, among others. The level of Virtual BOI has been recognized as an important factor in VR-based clinical therapy programs where patient immersion is crucial. However, a common issue is the lack of standardized evaluation tools for Virtual BOI, with most experiments relying on ad hoc tools based on experimental conditions or lacking consideration for the physical design elements of VR. This measurement tool was designed to consider the characteristics of recent VR devices, such as haptics and hand tracking, in the design of experiments and questionnaires. The tool is composed of sub-attributes related to VR technology, including Embodiment, Presence, Visuo-tactile, Visuo-proprioceptive, and Visuo-Motor. Based on a review of the existing literature, we hypothesized that the Virtual BOI scores would vary depending on manipulation methods, viewpoints, and haptic conditions. An experiment was conducted with 39 participants, who performed the same task under four different conditions using a virtual hand. Virtual BOI scores were assessed using the evaluation tool developed for this study. The questionnaire underwent CFA, and three items with factor loadings below 0.5 were removed, resulting in a total of 14 items. Each subscale demonstrated high reliability, with Cronbach’s alpha values greater than 0.60. When developing experiments, clinical programs, or VR content related to Virtual BOI, the evaluation tool presented in this study can be used to assess the level of Virtual BOI. Additionally, by considering technological elements such as haptics and hand tracking, VR environments can be designed to enhance the level of Virtual BOI.

1. Introduction

1.1. Background

HMD VR technology has made significant advancements, offering a wide field of view, high resolution, and feedback for auditory and tactile sensations, among others. These advancements induce a high level of immersion for users, making VR applications such as gaming, therapy, and collaborative programs increasingly versatile and effective [1]. In this way, HMD VR can provide experiences similar to reality, enabling its utilization in fields such as psychological therapy, rehabilitation, and behavioral neuroscience [2]. In HMD VR, psychological therapy and behavioral neuroscience treatment programs can reduce levels of anxiety triggered by specific stimuli or induce a sense of ownership over virtual bodies of different forms or sizes, thereby altering the perception of the actual body [3]. In essence, patient immersion in the treatment program arises from the experience of owning a body in the VR environment. In various fields of VR environments, making a virtual body feel like one’s own body, akin to the actual self, is commonly defined as the Body Ownership Illusion (BOI) [4].

One of the prominent experiments related to BOI is the Rubber Hand Illusion (RHI) paradigm. This experiment involves simultaneously touching a participant’s real hand and a rubber hand using a small brush, demonstrating that it can induce an illusion as if the experience were actually occurring from the rubber hand [5]. However, the Rubber Hand Illusion (RHI) paradigm imposes physical constraints due to the presence of a rubber hand, limiting its applicability for expanded research. However, BOI using virtual reality technology has been adopted in research because it allows for the creation of scenarios with various conditions and modifications (such as different situations or physical constraints) [4].

There have been various studies in different conditions and fields aiming to develop therapeutic programs using Virtual BOI. Scarpina et al. conducted a study on 15 obese patients and 15 normal-weight participants to investigate changes in the ownership perception of a virtual body with an abdominal region and the perception of the actual body [6]. Falconer et al. investigated depression treatment in VR by inducing Virtual BOI among depressed patients. They conducted an 8 min scenario in a VR environment and subsequently demonstrated improvements in depression severity, self-criticism, and self-compassion, highlighting the potential for depression therapy in VR [7]. Pamment and Aspell induced a Full Body Illusion in chronic pain patients, where it seemed as though their nerves were separated from their own bodies [8]. They confirmed a reduction in pain intensity. Weber et al. conducted an experiment with post-stroke upper limb paralysis patients using mirrors to present virtual movements, aiming to propose the potential of a VR-based therapy program for chronic stroke patients [9]. Alphonso et al. researched the potential for reducing pain in patients through the observation of virtual bodies [10].

In addition to experiments related to treatment programs, there have been studies inducing BOI under various conditions such as visuo-tactile, visuo-proprioceptive, visuo-motor, and different avatar forms. Rubo and Gamer demonstrated in a VR environment that adaptation to virtual bodies and the regulation of movement behavior can occur based on the congruence of visuo-tactile cues [11]. Lugrin et al. investigated the impact of the anthropomorphism level and realism of avatars on Virtual BOI [12]. Kondo et al. compared the effects of visuo-motor synchronization between avatars utilizing only body parts such as hands and feet and avatars representing the whole body [13]. Preston conducted research on BOI concerning visuo-proprioception, investigating the effects of increasing the distance between the real hand and a fake hand [14]. He found that the strength of BOI could decrease when the fake hand was positioned far away from the real hand and body. Indeed, key technologies in VR such as haptic feedback, 360-degree environments, and body tracking can induce Virtual BOI in users, and this is being investigated under various conditions.

Therefore, this paper conducts various statistical validations to develop a universal evaluation tool that can be applied in diverse experimental environments, moving away from arbitrarily constructed evaluation tools for experimentation. Furthermore, alongside Embodiment, which corresponds to the subjective perception of Virtual BOI, Presence is considered as a subscale for measuring the level of user immersion in the VR environment. To enable the association with the physical design of the VR environment, including haptics, embodiment, and interaction methods, we introduce the subscales of Visuo-tactile, Visuo-motor, and Visuo-proprioceptive. This facilitates the future design of environments where VR designers can induce Virtual BOI through this evaluation tool.

This paper aims to develop an evaluation tool capable of measuring the level of Virtual Body Ownership Illusion (BOI) by incorporating subjective factors of users, synchronization, and physical design elements of VR environments such as haptics, which most existing surveys fail to adequately capture. Given the absence of a standardized evaluation tool for Virtual BOI and the inadequate representation of physical aspects of VR environments in many questionnaires, we seek to construct items that combine subjective elements with synchronization and haptics, thereby facilitating the measurement of Virtual BOI.

1.2. Literature Review

We conducted a literature review on survey studies related to measuring Virtual Body Ownership Illusion. Currently, in most studies, subjective levels of Virtual Body Ownership Illusion are evaluated by using items derived from the Rubber Hand Illusion (RHI) or by modifying them to fit the experimental conditions when constructing the questionnaire.

Table 1. Sub-scales of Virtual BOI questionnaires.

References	Sub-Scales	Objectives	Procedure/Results
[15]	VRBody/MyBody, Mirror, Features, TwoBodies	Generating a questionnaire to verify whether inducing BOI through a represented altered body in a VR environment is possible.	Evaluating the level of BOI through subscales by using representations of a child’s body and a scaled-down adult’s body.
[16]	Acceptance, Control, Change	Developing a standardized measurement tool for Virtual Body Ownership Illusion.	- Verifying Virtual BOI through scenarios where different forms of bodies (Woodie, Robot, Human, Custom) are observed via a mirror. - Items with inadequate factor loadings and consistency were discarded, and PCA (Principal Component Analysis) was conducted to derive three subscales.
[17]	Ownership, Agency, Change	Developing a standardized measurement tool for Virtual Embodiment.	- BOI was measured in three experiments: synchronization based on participants’ actions and movements, representation of avatars resembling the user’s appearance, and detection of participants’ gaze and facial expressions. - CFA was conducted on the data obtained from the experiments, revealing three subscales.
[18]	Ownership, Agency, Tactile sensations, Location, Appearance, Response	Developing a standardized measurement tool for Avatar Embodiment.	Reviewed questionnaires from over 30 studies with high relevance to Avatar Embodiment to derive six subscales.
[19]	Appearance, Response, Ownership, Multi-sensory	Developing a standardized measurement tool for Avatar Embodiment.	- Data were collected from over 400 questionnaires across nine experiments. - PCA was conducted, resulting in 16 questions and four subscales.

represents the subscales of the questionnaire used to assess Virtual BOI levels. Banakou, Groten, and Slater divided the subscales into VR Body/My Body, Mirror, Features, and TwoBodies to investigate the impact of BOI induced by externally transformed bodies on cognition and behavior, using the questionnaire utilized in the RHI paradigm [15]. In a VR environment, 30 adults experienced embodying themselves as 4-year-old children and as adults with the same height as the children. Through the survey, the experiment confirmed whether BOI decreased due to visuo-motor incongruence and examined its effects on cognition and behavior. Roth et al. conducted a scale development study to measure Virtual Body Ownership Illusion using data from the fake mirror scenario experiment [16]. They conducted Principal Component Analysis (PCA) on this and extracted three factors: ‘Acceptance’, ‘Control’, and ‘Change’. Roth and Latoschik constructed the Virtual Embodiment Questionnaire and conducted Confirmatory Factor Analysis (CFA) based on data from three experiments with a total of 196 participants [17]. As a result, they identified three factors: Ownership, Agency, and Change. Gonzalez-Franco and Peck defined the feeling of the user’s body being replaced by their avatar in VR environments as Embodiment [18]. To measure this Embodiment, they proposed a standardized Embodiment questionnaire consisting of Ownership, Agency, Tactile sensations, Location, Appearance, and Response, along with a review of past similar questionnaires. Peck and Gonzalez-Franco collected over 400 survey data from nine experiments with the previously developed questionnaire [19]. They removed questions that were too specific or redundant, as well as those unrelated to other questions. The final questionnaire consisted of items grouped into four scales: Appearance, Response, Ownership, and Multi-sensory. Comparing the improved questionnaire with the original one, they found that participants aged 30 and above had significantly lower Embodiment scores compared to participants aged under 30.

However, constructing questionnaires arbitrarily to assess the level of Virtual Body Ownership Illusion may lead to inconsistency and reduced reliability across experiments. Since many researchers often directly quote or modify the Rubber Hand Illusion (RHI) questionnaire, such an approach may not be effective in VR environments characterized by features such as Immersion, Presence, Visuo-motor, and Visuo-tactile.

Furthermore, to statistically validate the evaluation tool, we aim to conduct assessments by designing variations in levels of various conditions that may influence the level of Virtual Body Ownership Illusion (BOI), such as haptic feedback (Haptic/Non-haptic), perspectives (1PP/3PP), and interaction method (Controller/Hand Tracking). We will measure participants’ Virtual Body Ownership Illusion (BOI) using the evaluation tool and conduct factor analysis to assess the reliability of the evaluation tool. Furthermore, based on previous research on Virtual Body Ownership Illusion (BOI), we will formulate hypotheses regarding the outcomes that can be observed under different conditions. We aim to validate whether the experimental results obtained using the evaluation tool proposed in this study align with these hypotheses. Based on these statistical validation results, we aim to propose an evaluation tool that can be utilized for various experiments and content development aimed at inducing Virtual Body Ownership Illusion (BOI).

2. Sub-Scales for Questionnaire

To create a questionnaire for measuring Virtual Body Ownership Illusion (BOI), we established five subscales and restructured existing Virtual BOI and Rubber Hand Illusion (RHI) items according to the following definitions. The definitions for each subscale are as follows.

2.1. Embodiment

Embodiment is defined in the context of the role of the actor’s body and the embodiment of cognitive processes, encompassing the role of the body in cognition [20]. Mottelson et al. defined Embodiment as experiencing an object as if it were one’s own body [4]. Furthermore, Embodiment can arise through alignment with Visuo-motor and Visuo-tactile inputs, and in VR environments, it can be triggered [21].

2.2. Presence

Tham et al. described Presence as an indicator of the overall mediated experience in VR, encompassing the quality of social interaction, the realism of the environment (graphics, sound, etc.), the degree of immersion generated by the interface, the user’s ability to perform significant tasks, and the social impact of events occurring within the environment [22].

2.3. Visuo-Tactile

Experiencing the body is primarily mediated by visuo-tactile experiences, which can be defined as information about contact events between the body and the surrounding environment [21]. In VR environments, when the virtual body appears to be touched visually, participants experience tactile sensations through vibration systems attached to their actual body or hands [23].

2.4. Visuo-Motor

The induction of Virtual Body Ownership through visuo-motor processes manifests as participants visually confirming the artificial body’s movements while simultaneously engaging in active or passive movements themselves [21]. In essence, the attributes associated with visuo-motor processes, which involve visually monitoring and perceiving one’s own movements and actions, make significant contributions to Body Ownership Illusion (BOI) [24].

2.5. Visuo-Proprioceptive

Virtual Body Ownership Illusion (Virtual BOI) manifests at different levels based on the visual spatial configuration and the Visuo-proprioceptive information of the actual body (concerning body position, posture, balance, and movement) [21]. Even in the absence of stimuli such as Visuo-tactile or Visuo-motor, the relative positioning and alignment of the actual body and the virtual body significantly influence the induction of Virtual BOI.

3. Method

3.1. Experiment Design

Prior to the experiment, participants were given a brief explanation of how to use the VR equipment and a brief overview of the experiment for 5 min. In this experiment, as the primary focus was on inducing Virtual Body Ownership Illusion (Virtual BOI) in VR, participants were instructed to focus on their interaction between their hands and the VR environment, rather than excessively engaging in gameplay. To exclude preconceptions about different conditions in the VR environment, specific descriptions regarding the conditions were not provided. All participants wore the same type of VR device for the experiment. The first task of this experiment involved observing the hands for one minute under static conditions. Following that, participants were tasked with writing their own name in the VR environment and picking up objects with different weights for 2–3 min. Afterward, participants engaged in 10 games of equal difficulty within the VR environment, allowing them to experience various actions such as grabbing objects, painting, and solving puzzles.

Based on the existing literature, we hypothesized that the presence of haptic feedback, perspective, and controllers could influence Virtual Body Ownership Illusion (Virtual BOI) scores. Therefore, the following conditions were established (

Table 2. Virtual BOI experiment condition description.

	Condition 1	Condition 2	Condition 3	Condition 4
Haptic feedback	Yes	No	Yes	No
Interaction method	Controller	Hand Tracking	Controller	Controller
Perspectives	First-Person Perspective	First-Person Perspective	Third-Person Perspective	First-Person Perspective

). The evaluation, consisting of 5 sub-scales, namely, Embodiment, Presence, Visuo-tactile, Visuo-proprioceptive, and Visuo-motor, along with 17 items, was presented using a 7-point Likert scale. Participants experienced all four conditions, and they evaluated the experience after each condition concluded. Each condition lasted an average of 15 min, with the entire experiment taking about an hour on average. Figure 1 shows a diagram of the experiment.

3.2. Experiment Apparatus

The experiment utilized the Meta Quest Pro (Reality Labs, Menlo Park, CA, USA) as the experimental device, and for performing tasks to induce Virtual Body Ownership Illusion (Virtual BOI), the VR game called “Hand Physics Lab” (v.1.2.0) by Holonautic (Luzern, Switzerland) was employed. The game allows players to perform various tasks such as grabbing and painting using virtual hands. The game has control over the perspective, hand tracking, controllers, and haptic feedback levels. Additionally, after completing each task, scores are provided based on performance. Figure 2 shows a snapshot of the game.

3.3. Hypotheses

The objective of this experiment was to verify whether the evaluation tool could demonstrate variations in participants’ levels of Virtual Body Ownership Illusion (Virtual BOI) based on conditions such as perspective (1PP/3PP), interaction method (Hand Tracking/Controller), and haptic feedback (Haptic/Non-Haptic).

Fröhner et al. demonstrated that haptic feedback significantly enhances Embodiment scores for virtual hands [25]. Rubo and Gamer showed that participants in their experiment adapted more easily to the shape of the virtual body assigned to them when Visuo-tactile correspondence was consistent [11]. Therefore, in this study, we aim to investigate whether Visuo-tactile and Embodiment scores align with the findings regarding the presence or absence of haptic feedback as revealed in previous research.

Previous research has revealed that stronger Virtual Body Ownership Illusion (Virtual BOI) occurs in the first-person perspective condition compared to the third-person perspective condition [26]. Mottelson et al. found that perspective (i.e., first-person or third-person perspective) consistently influences Body Ownership Illusion (BOI) [4], and Pozeg et al. differentiated between first-person perspective and a 90° rotated perspective (third-person perspective), with the first-person perspective significantly enhancing the integration of vision and touch [27].

Therefore, the following hypotheses were formulated:

• Virtual Body Ownership Illusion (Virtual BOI) scores will vary significantly depending on the conditions (H1).
• The Virtual Body Ownership Illusion (Virtual BOI) scores in Condition 4 (Non-Haptic) are expected to be lower than those in Control 1 (Haptic) (H2).

  –

  The Embodiment scores in Condition 4 (Non-Haptic) are expected to be lower than those in Control 1 (Haptic) (H2a).

  –

  The Visuo-tactile scores in Condition 4 (Non-Haptic) are expected to be lower than those in Control 1 (Haptic) (H2b).

• The Virtual Body Ownership Illusion (Virtual BOI) scores in Condition 3 (3PP) are expected to be lower than those in Control 1 (1PP) (H3).

  –

  The Embodiment scores in Condition 3 (3PP) are expected to be lower than those in Control 1 (1PP) (H3a).

  –

  The Visuo-proprioceptive scores in Condition 3 (3PP) are expected to be lower than those in Control 1 (1PP) (H3b).

3.4. Subjects

From 15 to 31 October 2023, an experiment was conducted with a total of 39 participants, with an average age of 22.2 years. There were 21 female participants and 18 male participants. After the experiment, participants received approximately USD 15 as compensation for their participation.

4. Results

4.1. Confirmatory Factor Analysis

Confirmatory Factor Analysis (CFA) was conducted using R (Lavaan package) to assess the consistency of the sub-scales of the questionnaire developed.

Table 3. Confirmatory Factor Analysis results.

Subscales	Questions	Factor Loading	Cronbach’s Alpha
Embodiment	Q1. Did it feel like the virtual body was your own body?	0.894	0.822
	Q2. Did it feel like your (real) body was moving towards the virtual body or like the virtual body was moving towards your (real) body?	0.742
	Q3. How much did it feel like there was more than one body besides your actual body?	−0.253
	Q4. How much did the virtual body resemble your (real) body?	0.729
	Q5. How natural was the interaction with the virtual environment?	0.873
Presence	Q6. How closely did the experience in the virtual environment match real-life experiences?	0.857	0.818
	Q7. Could you predict what would happen next based on your actions?	0.62
	Q8. How much delay did you experience between your actions and the expected outcomes?	0.322
Visuo-proprioceptive	Q9. Did it feel like your body was where the virtual body appeared to be?	0.911	0.681
	Q10. How well could you observe the virtual body from various perspectives?	0.570
	Q11. Was the sensation felt in the virtual body perceived to occur at the location where it was seen to be touched?	0.792
Visuo-tactile	Q12. Was the sensation felt as if it originated from touching the virtual body?	0.852	0.742
	Q13. How strong was the sensation felt simultaneously in both locations (virtual body/real body)?	0.851
Visuo-motor	Q14. Did it feel like your actual body was being affected by something?	0.467	0.829
	Q15. How actively could you move or manipulate the virtual environment using the controller (or hand tracking)?	0.904
	Q16. Could you control the virtual body as if it were your own body?	0.847
	Q17. Did it feel like the movements of the virtual body were influenced by the movements of your own body?	0.663

presents the results of this analysis. After the first CFA, items 3, 8, and 14 were excluded due to factor loadings below 0.5. After the exclusion, in the second analysis, the test using non-parametric statistics indicated statistically significant results (χ² = 127,841, df = 67, p < 0.001). Thus, each item and scale can be considered appropriate as components for evaluating VBOI.

The reliability values assessed by Cronbach’s alpha for each of the sub-scales, Embodiment (α = 0.822), Presence (α = 0.818), Visuo-Proprioceptive (α = 0.681), Visuo-tactile (α = 0.742), and Visuo-motor (α = 0.829), were all >0.6, indicating high reliability. Thus, we can conclude that the survey items we constructed are appropriately structured for each of the sub-scales.

4.2. ANOVA

4.2.1. Virtual BOI Scores

To verify whether the evaluation tool presented in this study indicates different levels of Virtual Body Ownership Illusion (Virtual BOI) according to conditions, ANOVA was conducted. The results of ANOVA revealed significant differences in Virtual BOI scores across conditions, thus supporting H1 (F = 59, p = <0.001).

Table 4. ANOVA for BOI scores.

	Sum of Squares	Df	Mean Square	F	p
Between Groups	22,678	3	7559	59	<0.001
Within Groups	19,474	152	128
Total	42,152	155

summarizes the results of ANOVA.

Table 5. Mean values for experiment conditions on VBOI score.

Dependent Variables	Experiment Conditions	Means	SD
Virtual Body Ownership Illusion Scores	Condition 1	5.6	0.66
	Condition 2	5.2	0.72
	Condition 3	3.3	1.04
	Condition 4	5.0	0.90

displays the mean values of Virtual BOI scores for each condition. Condition 1 (M = 5.6, SD = 0.66), with haptic feedback provided by the controller and a first-person perspective, and Condition 2 (M = 5.2, SD = 0.72), with hand tracking and a first-person perspective, exhibited higher scores compared to other conditions. Conversely, Condition 3 (M = 3.3, SD = 1.04), with haptic feedback provided by the controller and a third-person perspective, showed lower scores.

To investigate the statistical differences in Virtual BOI scores between conditions, Tukey’s test was conducted, and

Table 6. The result of Tukey’s HSD test between experiment conditions on BOI scores.

Independent Variable (I)	Independent Variable (J)	Average Difference (I–J)	p
Condition 1	Condition 2	0.430	0.112
	Condition 3	2.308	<0.001
	Condition 4	0.625	0.007
Condition 2	Condition 3	1.877	<0.001
	Condition 4	0.194	0.738
Condition 3	Condition 4	−1.683	<0.001

represents the results of Tukey’s test.

The results of Tukey’s test showed a statistically significant difference between Condition 1 and Condition 4, where the only difference was the presence of haptic feedback. H2 was supported. Thus, it was observed that the level of BOI varies depending on the presence of haptic feedback. Condition 3, which is in a third-person perspective, differs significantly from Condition 1, Condition 2, and Condition 4, which are all in a first-person perspective. This difference is statistically significant, supporting H3. Thus, in a third-person perspective, participants’ BOI was not well induced. This indicates that BOI scores are lower when experiencing the body from a third-person perspective compared to a first-person perspective.

4.2.2. Haptic Feedback

Statistical validation was conducted to verify whether not only Virtual BOI scores but also the sub-scales of Visuo-tactile and Embodiment in this questionnaire align with the findings regarding the presence or absence of haptic feedback as revealed in previous research.

The ANOVA results confirmed that there are differences in Embodiment scores based on the presence or absence of haptic feedback (F = 53.64, p = <0.001).

Table 7. Mean values for haptic feedbacks on Embodiment score, along with ANOVA results.

Dependent Variable	Conditions	Mean	SD	F	p
Embodiment Score	Condition 1 (Haptic)	5.7	0.74	53.64	<0.001
	Condition 4 (Non-Haptic)	4.7	1.12

presents a summary of the ANOVA results. Condition 4, which had no haptic feedback, had lower Embodiment scores compared to Condition 1, which had haptic feedback, and this difference was statistically significant, supporting H2a.

Statistically significant differences were also observed in Visuo-tactile scores based on the presence or absence of haptic feedback (

Table 8. Mean values for haptic feedbacks on Visuo-tactile score, along with ANOVA results.

Dependent Variable	Conditions	Mean	SD	F	p
Visuo-tactile Score	Condition 1 (Haptic)	5.0	1.11	10.73	<0.001
	Condition 4 (Non-Haptic)	4.1	1.58

). The Visuo-tactile scores were lower in Condition 4 compared to Control 1. Therefore, H2b was supported.

Based on these results, it was confirmed that the sub-scales of Visuo-tactile and Embodiment, as established in the questionnaire, indeed reflect differences in scores based on the presence or absence of haptic feedback. This indicates the ability to measure attributes related to Virtual BOI that vary according to physical conditions.

4.2.3. Perspectives

Condition 3, in a third-person perspective, exhibited statistically significant differences in Virtual BOI scores compared to Condition 1, Condition 2, and Condition 4, all of which were in a first-person perspective (Table 6).

The ANOVA results confirmed significant differences in Embodiment scores based on perspective (

Table 9. Mean values for perspectives on Embodiment score, along with ANOVA results.

Dependent Variable	Conditions	Mean	SD	F	p
Embodiment Score	Condition 1 (1PP)	5.7	0.74	53.64	<0.001
	Condition 3 (3PP)	2.9	1.18

). Significant differences were found in Embodiment scores based on perspective, with the third-person perspective scoring lower in Embodiment compared to the first-person perspective. H3a was supported.

Differences in Visuo-proprioceptive scores based on the presence or absence of haptic feedback were also observed (

Table 10. Mean values for perspectives on Visuo-proprioceptive score, along with ANOVA results.

Dependent Variable	Conditions	Mean	SD	F	p
Visuo-proprioceptive Score	Condition 1 (1PP)	5.5	0.79	45.07	<0.001
	Condition 3 (3PP)	3.3	1.03

). Differences in Visuo-proprioceptive scores based on perspective support H3b. This indicates that the Visuo-proprioceptive sub-scale, representing the user’s experience related to the physical location information of the object, indeed exhibits different scores based on perspective.

5. Conclusions

This study presented a VR-BOI evaluation tool applicable in VR environments with various conditions involving the presence of virtual bodies. This evaluation tool consists of the following sub-attributes related to Virtual BOI: Embodiment, Presence, Visuo-tactile, Visuo-proprioceptive, and Visuo-Motor. Based on sub-attributes that can be perceived in one’s body, and conditions allowing the evaluation of Virtual BOI based on these changes, we constructed the evaluation tool. We derived conditions that could induce different levels of Virtual BOI through a literature review. This includes perspectives (1PP/3PP), haptic feedback (Haptic/Non-Haptic), and interaction methods (Controller/Hand Tracking). The evaluation tool allowed us to assess the induced Virtual BOI levels under each condition, and we conducted statistical validation processes, including Confirmatory Factor Analysis (CFA) and Analysis of Variance (ANOVA), to verify the reliability of the evaluation tool.

We aimed to validate the consistency of the evaluation tool’s sub-scales through Confirmatory Factor Analysis (CFA). Through this, we removed three items with factor loadings < 0.5 and derived Cronbach’s alpha scores for each sub-scale based on the remaining items. All five sub-scales, Embodiment, Presence, Visuo-proprioceptive, Visuo-tactile, and Visuo-motor, demonstrated Cronbach’s alpha scores > 0.6, indicating the high reliability of the evaluation tool proposed in this study.

In this study, to validate the accuracy of the evaluation tool, hypotheses derived from previous research were formulated, and the Virtual BOI scores obtained from experiments were examined to determine whether they aligned with these hypotheses. The results confirmed that all hypotheses were supported. First, we examined whether Virtual BOI scores varied across the four conditions presented in the experiment. ANOVA results revealed statistically significant differences in Virtual BOI scores among the conditions. Furthermore, among the conditions, Condition 1 with a first-person perspective using a controller with haptic feedback and Condition 2 with a first-person perspective using hand tracking both showed Virtual BOI scores of 5 or higher. On the other hand, Condition 3 with a third-person perspective exhibited the lowest Virtual BOI scores, in the range of 3 points. Tukey’s test results indicated significant differences between the first-person perspective conditions and the third-person perspective condition, demonstrating that perspective plays a significant role in inducing Virtual BOI.

We also aimed to investigate whether the scores of sub-scales varied according to the conditions. Therefore, we compared the Embodiment scores and Visuo-tactile scores of Condition 1 and Condition 4, which differed only in the presence of haptic feedback, as well as the Embodiment scores and Visuo-proprioceptive scores of Condition 1 and Condition 3, which differed only in the perspective condition. ANOVA results showed that the scores of the sub-scales also differed among the conditions. Specifically, Condition 1 exhibited higher scores than Condition 3 and Condition 4, with statistically significant differences. These results indicate that the sub-scales proposed in this study sufficiently reflect the physical environmental conditions provided in the VR environment.

The evaluation tool proposed in this study can be utilized as an assessment tool for various fields of VR environments, including games and clinical therapy programs, related to Virtual BOI. Virtual BOI serves as a measure, especially in clinical therapy programs, to indicate how immersed patients are in VR therapy. By utilizing the evaluation tool presented in this study to assess the extent of induced Virtual BOI in VR environments, diverse forms of VR environments can be evaluated while considering the physical design elements associated with real VR environments.

In future research, additional analyses could be conducted to assign weights to each sub-scale. Moreover, expanding the range of the body displayed in VR to include not only hands but also the entire body or limbs could enable experiments and the statistical validation of the evaluation tool across various body conditions.