Combining Virtual Reality Visual Novels and Social Stories to Support Social and Emotional Development in Children with Autism Spectrum Disorder

Abstract

Autism Spectrum Disorder (ASD) is a neurodevelopmental condition affecting social interaction, communication, and emotional regulation, often requiring specialized educational approaches. Traditional teaching methods may not meet the needs of children with ASD, while visual and interactive formats tend to be more effective. This study examines the use of a virtual reality (VR) visual novel incorporating social stories to support social–emotional learning in children with Level 1 ASD. The intervention used an interactive, choice-based VR environment simulating everyday social situations. Four primary school students participated in a one-week case study, during which they engaged with the VR prototype. Teachers observed the students and provided feedback through questionnaires and interviews. Results showed high engagement with observable social communication behaviors and emotional responses in one participant but no consistent short-term changes across the group. All teachers expressed positive views, noting the tool’s visual clarity, ease of use, and alignment with ASD learners’ needs. Usability challenges were minimal but included pacing delays and difficulties managing impulsive behavior. Although limited by sample size and short duration, the findings indicate that VR visual novels are a promising and acceptable approach for practicing social–emotional skills in inclusive classrooms. Further research is needed to evaluate long-term effects and enhance adaptability.

1. Introduction

According to Ousley and Cermak [1], Autism Spectrum Disorder (ASD) is a neurodevelopmental condition marked by significant impairments in social communication, interaction, and language skills, alongside the presence of restricted and repetitive behaviors, interests, and activities.

The updated fifth edition of the Diagnostic and Statistical Manual of Mental Disorders [2] brought about notable revisions to the diagnostic criteria for autism. The previously used term Pervasive Developmental Disorders (PDD) has been replaced with Autism Spectrum Disorder [3]. Significant modifications include the removal of specific diagnostic categories such as Autism, Asperger Syndrome, and PDD-NOS (Pervasive Developmental Disorder-Not Otherwise Specified). Symptoms are now classified into three levels of severity: Level 3, which indicates a need for very substantial support; Level 2, requiring substantial support; and Level 1, where support is necessary. This study investigates a virtual reality-based intervention targeting social and emotional development in children diagnosed with Level 1 ASD. The focus on Level 1 autism is intentional, as the project’s design and objectives cater specifically to individuals who exhibit higher independence and require minimal support. Level 1 autism, characterized by relatively mild communication and social interaction difficulties, aligns well with the project’s aims, facilitating effective implementation and assessment of the intervention. In contrast, children with Level 2 or 3 autism necessitate more intensive and personalized support, which exceeds the project’s scope and design.

Social skills are the ability to interact well with others, including cooperation, sharing, and following rules. They help people work in groups, solve problems, and connect with others. Empathy, or understanding others’ feelings, is key. But without ethics, social skills can be misused. Teaching these skills with a moral base early on supports strong relationships later in life [4].

Communication skills are the ability to send and receive messages clearly through speech, writing, and body language. They include clear speaking, active listening, and using non-verbal cues like gestures and expressions. Focusing too much on technique can make communication feel cold. Balancing clarity with empathy helps create real connection. Strong communication skills support better relationships and success at work [5].

Emotional skills are the ability to understand and manage your own emotions and others’. They help with stress, relationships, and solving conflicts. While helpful, too much control can cause emotional fatigue. Balancing regulation with honest expressions is key. Emotional intelligence supports both personal and career success [6].

For the purposes of this study, the term “social and emotional skills” refers to two interrelated domains. The first involves social communication abilities such as verbal expression, turn-taking, and appropriate response in social interactions. The second pertains to emotional understanding, including the ability to recognize, interpret, and respond to others’ emotional states. These constructs were examined using a combination of teacher observations, a structured interview protocol, and a custom post-session questionnaire assessing observable behaviors and emotional responses during and after the intervention.

The core outcome of this study is the observation of changes in social communication behaviors and emotional responses following exposure to a VR-based visual novel. These outcomes were assessed through educator observations and structured feedback. The theoretical basis for the intervention draws on experiential learning theory [7], which emphasizes learning through active engagement and reflection, and social learning theory [8], which supports the idea that modeling and role-playing within interactive narratives can enhance social–emotional learning in children with ASD.

1.1. Social Stories in Education

One early intervention used in therapy for children with ASD is the application of social stories. These are concise narratives created to help children understand and respond to social situations by illustrating appropriate behaviors. Social stories can be written or visual and aim to clearly explain everyday social concepts and interactions. They are typically delivered by therapists either verbally or with basic 2D visual support. However, the limited engagement offered by 2D visuals may reduce the effectiveness of the message, affecting how well the child absorbs and applies the intended social lesson [9].

1.2. Visual Novels in Education

Visual novels, a genre of interactive digital storytelling originating in Japan, combine narrative storytelling with dialog and visual elements. They typically feature decision points that lead players through branching storylines with multiple outcomes. These decision paths offer a form of replayability and interactive engagement [10]. Visual novels are increasingly being explored for their educational potential, not just as games but as tools for teaching.

Although many educational visual novels have not been thoroughly evaluated for their instructional value, some studies have reported positive outcomes. For instance, educational visual novels have been linked to enhanced self-efficacy and the experience of flow [11], as well as greater learning, interest, and knowledge retention compared to traditional text-based materials [12]. Such narrative-based games can also help develop language proficiency, reading comprehension, creativity, and critical thinking skills [13]. These findings suggest that the interactive narrative format of visual novels holds promise for education, especially when learners can explore scenarios and consequences in a safe virtual environment.

1.3. Immersive Technologies in Education

Virtual reality has emerged as a transformative technology in education, standing out from other ICT tools due to its immersive qualities. VR involves a suite of technologies that generate highly interactive, three-dimensional environments that replicate real or imagined scenarios. These environments allow users to engage through multiple senses and interact with virtual elements in real-time [14]. For maximum educational benefit, VR must be clearly defined and properly integrated into instructional frameworks.

With technological advancements driving more digitalized and gamified learning experiences, education is increasingly adopting strategies from video games to create immersive and interactive learning environments. In the context of special education, VR’s controlled and multisensory environment is particularly appealing for children with ASD. It offers a safe space to practice social interactions without the unpredictable and stressful elements of real-life situations. However, it has not yet been proven whether the use of VR technology combining social stories with visual novel elements can be effective for children with Level 1 ASD.

This study explores how a VR application could support children with Level 1 ASD in enhancing their social and emotional abilities. Specifically, we investigate whether a VR visual-novel-style social story can effectively improve social interaction and communication skills among children with ASD in a mainstream educational setting. To answer this, a case study was conducted using a VR visual novel designed specifically for primary school children with ASD, situated within a mainstream educational environment. The VR story provided relatable social scenarios and offered users the chance to make decisions, mimicking real-life social interactions. This controlled, virtual setting offered a safe space to practice responses, understand social cues, and build social confidence. Assessments conducted before and after the intervention, along with feedback from teachers and parents, indicated noticeable improvements in participants’ social behaviors, including better eye contact, turn-taking, and increased empathy.

2. Related Work

The integration of immersive technologies, particularly VR, into educational interventions for children with ASD has gained increasing attention in recent years. Recent studies suggest that VR environments may offer benefits in supporting the development of social, communication, and emotional skills in children with ASD by enabling structured, repeatable interactions in safe contexts [15,16]. However, most interventions to date either focus on skill rehearsal through gamified tasks or provide open-ended virtual spaces without embedded narratives. There remains a lack of empirical research examining how VR-based visual novels—integrating structured social stories with interactive storytelling—might support socio-emotional learning, particularly in school-based applications targeting children with Level 1 ASD.

A central theme across several studies is the use of controlled, quiet environments to reduce distractions and enhance participants’ engagement with virtual settings. For instance, Frolli et al. [17], Zhao et al. [18], and Moon [19] all emphasize the importance of using dedicated spaces to ensure that participants can fully immerse themselves in virtual environments. These controlled environments are particularly beneficial in ASD interventions, where focus and attention are critical.

Many of the existing VR-based interventions for ASD focus on individualized learning experiences where a child interacts one-on-one with virtual characters or objects. Studies by Cheng et al. [16], Beach and Wendt [20], and Ip et al. [21] found that self-paced, personalized VR experiences can be effective. One emerging area is the introduction of collaborative learning models, where participants engage in social decision-making with guiding characters or peers inside the VR environment. Inspired by early efforts like those of Parsons [22], multi-user interactions in VR remain underexplored in ASD contexts.

Another common aspect of the literature is the use of experiential, scenario-based learning. VR studies such as those by Ip et al. [21,23] and Bekele et al. [24] demonstrate the efficacy of scenario-based learning in raising social interaction and emotional regulation, key areas of difficulty for children with ASD. This decision-making approach is consistent with the findings of studies such as George et al. [25], which highlighted the importance of active participation in promoting social–emotional growth.

Intervention duration is a variable that differs widely across studies. Some researchers have implemented long-term VR interventions over several months [17,25], while others, like Cheng et al. [16] and Didehbani et al. [26], had shorter durations of 5 to 6 weeks. As seen in Alvarado et al. [27] and Manju et al. [28], who noted that extended practice may yield more significant outcomes. Balancing intervention depth with implementation constraints remains a challenge for future applications.

An area of consensus across studies is the focus on social and emotional skills development. Many interventions have concentrated on improving communication, social interaction, and emotional recognition, as these are core deficits in ASD. Virtual environments are particularly effective in this regard because they offer a controlled space where children can practice these skills without the risk of real-world consequences. Studies like those by Soltiyeva et al. [29], Frolli et al. [17], and Vahabzadeh et al. [30] have shown that VR scenarios can improve emotional regulation, social attribution, and reasoning skills.

Additionally, VR has been shown to be particularly effective for teaching specific social behaviors, such as initiating conversations, responding to social cues, and understanding emotional expressions. This aligns with the findings from studies such as those by Moon [19], Beach and Wendt [20], and Ip et al. [21], where participants demonstrated increased engagement and improved communication skills. In particular, Moon [19] observed that children with ASD showed greater confidence in initiating conversations with virtual characters.

However, challenges remain in ensuring that VR interventions are universally effective. Some studies, such as those by Soltiyeva et al. [29], noted that children with ASD may focus more on static objects than dynamic ones, which can influence their engagement with social tasks.

Despite these advancements, a gap remains in evaluating whether VR-based social stories—delivered through interactive visual novels—can effectively enhance both social and emotional skills in children with ASD. Most existing interventions do not integrate narrative, decision-making, and reflection within VR formats. There is also limited evidence from classroom-based implementations that include teacher feedback.

From this context, the following research questions and hypothesis are proposed:

RQ1: What social communication behaviors are observed in children with Level 1 ASD during and after participation in a VR-based social story presented as a visual novel?

RQ2: How do children with Level 1 ASD respond emotionally during the VR experience, as observed and reported by educators?

RQ3: What insights do educators offer regarding the acceptance, user experience, and practical feasibility of implementing VR visual novels in inclusive classroom settings?

H1: Children with Level 1 ASD will demonstrate observable social communication behaviors during and after participation in a VR-based visual novel intervention.

H2: Children with Level 1 ASD will exhibit emotional responses during the VR experience that are identifiable by educators.

H3: Educators will perceive the VR-based visual novel as acceptable, user-friendly, and feasible for use in inclusive classroom settings.

3. Materials and Methods

The proposed VR-based visual novel intervention reflects the trends and challenges identified in previous studies on VR and ASD. While it shares many of the same objectives—such as improving social, emotional, and communication skills—it distinguishes itself by incorporating collaborative decision-making, a flexible, narrative-driven structure, and a focus on real-time social reflection. This combination aims to enhance the learning experience and provide children with a more comprehensive and immersive tool for developing essential life skills. Further research should explore the impact of extended interventions and personalized adaptations to ensure that VR can be maximally effective in supporting children with ASD.

3.1. Research Design

This study was designed as an exploratory case study in a real school setting, using a mixed-methods approach to evaluate the VR intervention. The primary objective of this study was to examine how children with Level 1 ASD demonstrate observable social communication behaviors and emotional responses during and after participating in a VR-based visual novel.

Virtual reality was used to deliver immersive, individualized learning experiences that minimize the stress and distractions typically encountered in real-life social situations. The intervention was structured as a single one-hour, one-on-one session conducted in a quiet lab space. Children interacted with two narrative-based, decision-driven VR scenarios modeled after visual novels. These sessions were preceded by a brief introduction to communication concepts.

The study followed a single-group design with pre- and post-intervention evaluations. Over the course of the following week, classroom teachers observed student behavior during typical school activities. They later completed structured questionnaires and participated in interviews, providing both behavioral and experiential feedback on the tool’s feasibility and educational value.

This exploratory case study follows precedent in early-stage ASD (Level 1) intervention research, where small samples allow in-depth observation under realistic school conditions [16,22]. The focus is on feasibility, not generalizability. Additionally, the limited sample size reflects the study’s objective to identify children with ASD within a mainstream educational setting rather than a special education school. This decision was based on the higher likelihood of encountering Level 1 cases in typical school environments. However, the prevalence of students with ASD in such settings is relatively low, and the specific school involved had a limited number of eligible participants available at the time of the study. This was the convenience sample we had access to, as it was not possible to recruit more participants.

The limited duration of the intervention was due to school-imposed time constraints and the fixed timeline for data collection. Despite these limitations, the study offers important exploratory insights that may inform the design of longer, more scalable interventions in future research.

3.2. Participants

This study involved four students diagnosed with Autism Spectrum Disorder Level 1, who were participating in an inclusive educational program at a Greek primary school. For confidentiality, we refer to the children using pseudonyms. The above information was obtained through conversations with the personal special education teacher of the year for each child.

George (11 years old, fifth grade) is sociable, well-liked, and enjoys interacting with peers and teachers, although he struggles with extended conversations. Mary (10 years old, fourth grade) is active and sociable but becomes frustrated when she feels different from her peers. Giannis (12 years old, sixth grade) is polite and cheerful. He has limited verbal skills, rarely initiates conversations, but follows social prompts well when guided. Spyros (8 years old, second grade) is polite and sociable but reacts emotionally to disruptions in routine. All students attended mainstream classes and received one-on-one support from special education teachers in a quiet computer lab to reduce distractions. Each child worked individually with a teacher in 45 min sessions, followed by a week of behavioral observation. The small sample size and the use of an existing inclusive classroom context align with the exploratory, case-study nature of the research.

The study was approved by the school administration and implemented as an enhancement to the existing ICT-based curriculum. Although the VR intervention was not part of the standard educational program, it was delivered within the school setting under teacher supervision. Written informed consent was obtained from the parents of all participating students and from the teachers involved, in line with ethical standards for educational research.

3.3. VR Intervention Design

We developed a VR prototype application that delivers a social story in the format of an interactive visual novel. The VR environment, created with the Spatial.io platform and Unity 3D ver. 2021.3.21f1, consisted of two linked virtual scenes themed around the daily life of a child named “Peter.” The first scene, entitled “Peter’s World”, represented a home setting specifically, Peter’s bedroom (Figure 1), and the second scene represented a school setting entitled “Peter’s School” (Figure 2). These scenes were populated with familiar objects such as toys, furniture, a classroom environment, etc., to make the experience relatable and to simulate real-world contexts. The application was designed to run on a standard PC, representing a non-immersive VR setup. In this context, non-immersive refers to the absence of head-mounted displays or sensory isolation, offering instead a screen-based interactive experience (no VR headset was used, for ease of integration in the school setting). The environment followed a desktop-based VR format, including interactive 3D scenes, dialog, and narrative decision-making features consistent with non-immersive VR used in educational contexts. The dialog and text within the game were in Greek, the native language of the participants and their teachers, to ensure accessibility and comprehension.

The educational content of the intervention was centered on teaching appropriate social greetings and fostering empathy. The storyline followed a day in the life of Peter, a virtual character who models both appropriate and inappropriate social behaviors. For example, if someone greets Peter by saying “hello” and Peter responds by saying “hello” back, this is considered an appropriate social behavior. At the start of the VR experience, in Peter’s virtual bedroom, a friendly guide character welcomes the player (the child) and provides instructions for how to proceed. A prominent visual overview (the “Big Picture”) is presented to outline the session’s goals and steps, helping the child understand what will happen next. The child can interact with various items in the room. For example, clicking on a computer in the virtual bedroom plays a short video demonstrating how to greet someone politely, emphasizing saying “hello” and making eye contact. This acts as a video-based social story, showing a scenario of Peter greeting a friend. Next, the player clicks on a bookcase, which opens a written version of the same social story about greetings. Presenting the lesson in both video and written formats reinforces understanding and caters to different learning preferences. After reviewing these materials, the player clicks on a door, which transitions the VR scene from Peter’s home to his school.

In the “Peter’s School” scene, the player is engaged in a virtual school environment with classrooms and hallways. The character Peter appears as a guide within this scene; he introduces himself to the player and explains the next interactive task. The player is guided along a path marked by visual indicators on the floor, leading to the next interactive activity (Figure 2).

Upon reaching a specific spot, indicated by a highlighted area or circle in the virtual schoolyard, the player is presented with a situation and a set of response options (Figure 3). For instance, the scenario might involve Peter encountering a classmate and the player must choose how Peter should respond (options could include a proper greeting, an improper greeting, or ignoring the classmate). These choices are presented as text or icons on the screen, and the child must select an option. If the player selects a socially inappropriate or incorrect option, the game provides immediate gentle feedback: the screen might indicate that a “heart”, which serves as a life point, is lost and an explanation is given that this choice was not the best. The feedback is visual (e.g., a heart icon disappearing) and narrative, but non-punitive, encouraging the child to reflect and try again. The player can then attempt the scenario again or proceed to the next part of the story if they choose correctly. Through this branched storytelling, the child experiences the consequences of different social choices in a safe environment.

However, it is important to acknowledge that observed engagement may have been influenced by the novelty effect. The introduction of a new digital medium, particularly VR, can generate heightened interest independent of content quality. While most students responded positively during the session, some of this enthusiasm may reflect short-term excitement rather than lasting impact. Future studies should therefore examine sustained use over time to differentiate between initial novelty and genuine educational effectiveness.

The VR visual novel thus combines multiple pedagogical strategies: video modeling (through the greeting demonstration video), reading and observation (through the written social story), and learning by doing (through interactive choice-making in the school scenario). By integrating these elements, the intervention provides repetition and reinforcement of the target skill (appropriate greetings) in various forms. It also engages the student in active learning, as they must make decisions and see the outcomes. The overall experience is designed to last around 15–20 min, fitting within a typical session with a special education teacher. The experiential, game-like structure of the intervention aims to maintain the children’s interest and provide a low-pressure setting in which they can practice essential social behaviors like greeting others, recognizing social cues, and understanding how their choices affect those around them.

Although basic emotional expressions were embedded into the character animations (e.g., smiling or neutral faces), they were limited in range and subtle in execution. Feedback from teachers and observed outcomes suggest that greater emotional fidelity may be needed for effective learning transfer. Future versions of the application should include varied emotional expressions (e.g., anger, sadness, excitement) presented through more distinct facial cues and potentially through symbolic representations (e.g., animated GIFs, expressive icons, or color changes). These enhancements would aim to better support emotional recognition for children with ASD, who often rely on exaggerated or visually explicit cues to interpret affective information.

3.4. Procedure

Prior to the VR sessions, researchers consulted with the special education teachers to identify suitable participants and to understand each child’s individual needs. These discussions ensured that the chosen participants met the criteria necessary for the study’s objectives, and that the educational environment was suitable for conducting the intervention. Based on consultations with the teachers, the decision was made to avoid the use of VR headsets, as they expressed concerns about possible discomfort or distraction for the students.

Following the selection process, participants were introduced to fundamental concepts related to communication. This introductory phase aimed to establish a baseline understanding of social interaction and communication skills, preparing the children for the upcoming intervention.

Each child then participated in a one-hour one-on-one session with their special education teacher in a quiet computer lab setting. At the beginning of the session, the child donned headphones and was seated at a computer where the VR visual novel was loaded. The teacher provided basic instructions on how to interact with the VR environment, e.g., using the mouse to click on objects and choices and then guided the child into the experience as needed. The child engaged with the “Peter’s World” and “Peter’s School” VR scenarios as described above. The teacher was present to offer encouragement or minimal assistance, but the children were encouraged to make choices independently within the game.

After completing the VR intervention, which took approximately one hour including introduction and breaks, each child returned to their regular class routine. Over the following week, the teachers closely observed each child during normal classroom and playground activities, paying particular attention to any changes in social interaction, communication attempts, or emotional responses compared to prior behavior. This observational period was intended to capture any immediate, real-world effects of the VR experience on the child’s social behavior.

At the end of the one-week observation period, teachers completed a structured questionnaire for each participating child. In addition, brief semi-structured interviews were conducted with each of the four special education teachers. In these conversations, teachers were asked to elaborate on their experiences using the VR tool, their perceptions of student engagement and learning, and any practical issues or suggestions for future implementation. These interviews were audio-recorded and later summarized for key themes.

3.5. Research Instruments

To evaluate the outcomes of the intervention and gather educator feedback, we used two primary instruments: a custom teacher questionnaire and follow-up teacher interviews.

The decision to utilize questionnaires and conduct conversations with the teachers was based on the fact that each participating teacher is the designated personal special education teacher for their respective students. Given their close, ongoing relationship with the children, they are uniquely positioned to provide meaningful insights into each child’s social and emotional development. This approach is particularly important, as children with autism often experience difficulties with verbal expression and communication, which can limit the reliability of self-reported data.

The questionnaire was developed by the research team and administered via Google Forms for convenience and consistency. It was divided into three sections, combining both closed-ended and open-ended questions to collect quantitative ratings and qualitative insights.

The first section is about usability and engagement, how easily children navigate the VR environment, their level of enjoyment, and any observed emotional responses. Questions included Likert-scale items on navigation difficulty.

The second section focuses on the core research objectives (RQ1 and RQ2), asking whether the teacher observed any social communication behaviors or emotional responses following the VR intervention. For example, teachers indicated whether the student showed new or improved social behaviors in the days following the session (such as initiating greetings more often, using phrases from the game in real life, or demonstrating greater empathy/recognition of others’ emotions). These questions typically used before/after comparison (no change, slight improvement, moderate improvement, etc., or not applicable).

The final section explores teachers’ perspectives toward the use of VR visual novels in autism education addressing RQ3. This part was grounded in the Technology Acceptance Model [31]. Teachers reflected on perceived usefulness and ease of use, and indicated whether they would recommend similar approaches. An open-ended question invited suggestions for improvement and noted any challenges faced.

The questionnaire design included both closed-ended questions, for measurable and standardized results [32,33], and open-ended ones to capture deeper insights and personal observations [34,35].

To complement the questionnaire, in-depth interviews with the teachers were conducted after they submitted their questionnaires. In these semi-structured interviews (approximately 20–30 min each), teachers were prompted to discuss their overall impression of the VR intervention, elaborate on any questionnaire responses if needed, and share practical considerations (e.g., how feasible it would be to use this tool regularly in the classroom, or what support/training would be required). The interview responses provided contextual depth and helped clarify why certain outcomes were observed.

This combination of instruments allowed for a comprehensive evaluation of the VR intervention’s effectiveness (through observed student outcomes) and its acceptability and practicality from the educators’ perspective. The mixed-methods data were then analyzed to answer each of the research questions.

4. Results

After the VR intervention and the one-week follow-up period, data were collected through teacher observations, questionnaires, and interviews. The results are presented below according to the research questions, focusing on key findings related to social communication skills (RQ1), emotional understanding (RQ2), and teacher perspectives (RQ3). All four participating teachers completed both the questionnaire and the interviews, resulting in a complete dataset for analysis. The main finding is that noticeable positive outcomes were not evident within the first week following the intervention.

4.1. Observed Social and Communication Behaviors (RQ1)

When it came to engagement, three out of four teachers agreed that the students were highly involved and interested while using the visual novel. Only one teacher gave a neutral response. Similarly, emotional responses were mostly positive. According to three teachers, the children appeared excited or happy during the sessions. One teacher did not observe strong emotional reactions, reporting a neutral stance. These responses suggest that the VR experience was engaging and emotionally stimulating for most of the participants.

Despite this engagement, most teachers did not observe clear improvements in students’ social or communication behaviors during the post-session observation period. Only Spyros’s teacher noted observable progress, such as increased turn-taking, clearer verbal expression, and more confidence in initiating conversations. These outcomes are consistent with prior studies reporting limited real-world impact from short-term VR interventions [16,17,18,19,22,23,24,25,26,36].

Feedback on verbal communication was similarly divided. While three teachers saw no improvement, Spyros showed moderate improvement. These results reflect patterns found in earlier research, where some children showed benefits in verbal and non-verbal communication, while others did not [16,17,18,19,22,23,24,26]. This variation may stem from individual learning differences, including cognitive styles, familiarity with digital tools, or the ability to generalize virtual scenarios to real life. Thus, the intervention might require personalization and longer exposure to produce consistent outcomes across students.

These findings indicate that although the VR environment captured attention and elicited positive emotions, its effect on real-life social and communication behaviors was limited during the short-term intervention. This is also reflected in the descriptive questionnaire results (Appendix A,

Table A1. Frequency and Mean of Teacher Responses (N = 4).

Item	Mean	SD	Very Easy/Observable/Definitely	Easy/Moderate/Probably	Neutral/Slight	No Improvement
1. How easy was it for children to navigate the virtual space?	3.0	0.82	1	1	2	0
2. Did the children find the virtual novel engaging?	3.25	0.50	0	3	1	0
3. Did the children seem happy or excited while playing the game?	3.25	0.50	0	3	1	0
4. Did the game help children improve social skills?	2.00	1.15	1	0	0	3
5. Did the game help children improve communication skills?	2.00	1.15	1	0	0	3
6. Verbal expression improvement after the game?	2.25	0.96	0	1	0	3
7. Improvement in understanding others’ feelings?	2.25	0.96	0	1	2	1
8. More appropriate emotional responses?	2.25	0.96	0	1	0	3
9. Plan to continue using VR visual novels?	3.75	0.50	1	3	0	0

Scale adjusted contextually per item: 1 = No Improvement/Very Difficult/Definitely Not, 4 = Observable Improvement/Very Easy/Definitely.

), where mean ratings for improvement in social and communication skills ranged between 2.00 and 2.25. Responses were given on a 4-point Likert scale, where 1 = Not at all, 2 = A little, 3 = Moderately, and 4 = Very much.

Table 1. Questionnaire responses on improvements in social and communication skills.

Special Needs Teacher	Did the Game Help the Children Improve Their Social Skills (e.g., Initiating Conversations, Interacting with Others) in Real-Life Situations?	Did the Game Help the Children Improve Their Communication Skills (e.g., Using Specific Words or Phrases from the Game in Real-Life Conversations)?	Have You Noticed Any Improvement in How the Children Express Themselves Verbally After Playing the Game?
Spyro’s Teacher	Observable improvement	Observable improvement	Moderate improvement
George’s Teacher	No improvement	No improvement	No improvement
Mary’s Teacher	No improvement	No improvement	No improvement
Gianni’s Teacher	No improvement	No improvement	No improvement

presents a summary of teacher responses regarding perceived behavioral changes.

4.2. Emotional Responses During VR Use (RQ2)

As demonstrated by

Table 2. Questionnaire responses on improvements in emotional skills.

Special Needs Teacher	Did the Game Help the Children Understand Others’ Feelings Better in Real-Life Situations (e.g., Recognizing Emotions, Empathy)?	Do You Plan to Continue Using Virtual Reality Visual Novels or Similar Technologies in the Future for Teaching Children with Autism?
Spyro’s Teacher	Moderate improvement	Moderate improvement
George’s Teacher	No improvement	No improvement
Mary’s Teacher	Slight improvement	No improvement
Gianni’s Teacher	Slight improvement	No improvement

, feedback related to emotional skills showed only modest improvements and emphasized the difficulty of achieving deeper emotional learning in a brief intervention. Teachers observed varied impacts on emotional understanding. Two of the four teachers noted slight progress in their students’ ability to recognize or label others’ emotions after the VR experience. For example, these students were marginally more attentive to characters’ facial expressions or feelings in the stories, suggesting the game’s scenarios may have begun to increase their awareness of emotional cues. One teacher (Spyros’) reported a more moderate improvement in emotional understanding—this corresponded to the same student who showed gains in social communication, hinting that this particular child (among the four) derived more benefits from the VR intervention. Nevertheless, all teachers agreed that the children’s ability to respond appropriately to others’ emotions did not change over the short duration of the study. In practical terms, teachers did not observe the students demonstrating more empathy or adjusting their behavior in response to a peer’s feelings following the VR sessions. Only in one case was there a minor positive change in how a child reacted to others’ emotions (a single teacher described some improvement in this regard). Thus, while the VR social story may have helped a few children begin to identify emotions (e.g., recognizing when a character was happy or upset), it did not appreciably improve the children’s real-world emotional responses or emotion regulation skills in the immediate term.

These results for RQ2 indicate that translating emotional insight from the VR game into everyday empathy remained a challenge. The game’s design did include interactive choices and consequences aimed at emotional learning—for instance, players had to choose how the character Peter should respond in social situations. An immediate visual feedback mechanism was provided: if a less appropriate choice was selected, the player would lose a “heart” life, signaling the choice was not ideal. However, despite this feedback-driven learning approach, teachers did not witness a clear carryover of improved emotional understanding or empathy outside the VR context during the short evaluation period. As presented in Table A1, mean ratings related to emotional understanding and response remained modest (M = 2.25 on a 4-point Likert scale), indicating that most teachers perceived only minimal change in this domain.

4.3. Educators’ Perspectives and Technology Acceptance (RQ3)

Table 3. Teacher feedback on future use and suggested improvements.

Special Needs Teacher	Do You Plan to Continue Using Virtual Reality Visual Novels or Similar Technologies in the Future for Teaching Children with Autism?	Do You Have Any Suggestions for Improving the Virtual Reality Experience for Children with Autism?
Spyro’s Teacher	Definitely	The virtual reality tool was very informative and fun for the student. So I wouldn’t change anything.
George’s Teacher	Probably	No Change
Mary’s Teacher	Probably	No Change
Gianni’s Teacher	Probably	If possible, the options could be accompanied by corresponding icons or GIFs, in order to maximize the autonomy of students who do not understand the meaning of words and primarily respond to visual stimuli.

clearly highlights that a key positive finding was that three out of four teachers were willing to use similar VR tools in the future. Spyros’s teacher was especially enthusiastic, noting how appealing the visual content was for autistic learners. One teacher explained that children with autism respond well to images and minimal text, which aligns with research supporting visual strategies in ASD education.

Teachers highlighted the engaging visual nature of the VR novel as a major advantage for students with ASD. The rich graphical environment and minimal text—for example, the colorful 3D scenes and cartoon characters shown in Figure 1—were seen as well-suited to the needs of autistic learners. They noted that many children with ASD are visual learners and respond better to images or cues than to lengthy verbal explanations. The VR story’s format, which combined simple language with interactive visuals, helped maintain the students’ focus and motivation. Teachers also valued the controlled, safe environment of the VR application, commenting that it allowed students to practice social situations (like greeting a friend or sharing) without the stress of real-life consequences. They felt this engaged yet structured setting could encourage children to work on social–emotional skills more willingly.

Alongside their positive feedback, teachers identified a few practical challenges and suggestions to improve the VR tool. One commonly mentioned issue was the pacing of the game. In its current form, the transition between scenes or the appearance of new content was somewhat slow. Two teachers noted that this slow image loading disrupted the flow of the session; students would have to wait for the next scene or choice to appear, which occasionally caused them to lose interest or become impatient. In one notable case, a student began selecting answers impulsively due to the pacing and possibly the child’s own impulsiveness. This student consistently chose the first available response option before hearing all the narrative options or awaiting the full scenario. Even when the teacher prompted the child to listen to all choices, the student often continued to click quickly through the options, seemingly misunderstanding the task or simply eager to progress. As a result, this student frequently encountered the game’s corrective feedback (losing a heart life for an inappropriate choice), but the immediate penalty did not deter the impulsive behavior. This behavior led to some frustration and diminished the educational value for that child, as he was not fully processing the social decision-making aspect. Additionally, that same student at times disengaged from the activity, requiring the teacher’s intervention to refocus his attention on the VR scenario. Such observations underline the need for more adaptable pacing or guidance within the game for learners who struggle with self-regulation.

When asked for improvements, most teachers were satisfied with the tool as-is: three out of four did not suggest any changes, describing the experience as enjoyable and informative for their students. One teacher wrote that the VR story was an “innovative and fun way” to work on social skills, reinforcing that they found the content pedagogically valuable. However, the fourth teacher provided a constructive suggestion to enhance accessibility: they recommended adding small icons or animated GIFs alongside the text of the answer choices. The idea was that having a visual symbol for each choice (for example, a happy or sad face icon next to options about emotional responses) could help students who struggle with reading or comprehension to better understand the meaning of each choice. This would make the decision points in the game more immediately clear to non-readers or to those who need visual support to process language. Implementing such multi-modal cues could improve the experience for children with lower literacy or greater cognitive challenges.

In summary, educators’ perspectives on the VR visual novel were very positive regarding its usefulness and fit for ASD education. They embraced the tool’s engaging qualities and saw potential for continued use, while also recognizing areas for improvement like technical smoothness, support for impulsive learners, and additional visual aids. Appendix A (Table A1) shows that teacher willingness to reuse the VR tool was rated highly (M = 3.75 on a 4-point Likert scale), suggesting a generally positive perception of its classroom applicability.

5. Discussion

This study investigated the effectiveness of a VR visual novel in improving social and communication skills in four primary school students with Level 1 ASD. All participants exhibited polite and cheerful behavior, with no major emotional outbursts. The VR sessions took place in a quiet computer room, supervised individually by their special needs teachers, aligning with other studies that utilized calm, controlled settings for ASD interventions [16,17,18,19,22,23,24,25,26,29,36]. Unlike the peer-pairing method in Parsons [22], this study opted for individualized sessions.

Three students engaged readily with the game, while Mary was initially resistant due to the perception that it was only for “autistic” children. Her reaction emphasizes the importance of designing inclusive tools that avoid stigmatization and promote universal appeal. Spyros and George responded enthusiastically to the game’s content, supporting findings from Soltiyeva et al. [29]. In contrast, Giannis needed clear guidance to use the interface, reflecting participation difficulties seen in similar studies. For instance, in [29], students required extra time and assistance during the “Greeting task” with the virtual farmer, highlighting variability in user autonomy.

Teachers’ feedback on navigation varied. Spyros found it very easy, George’s teacher rated it as easy, while Mary and Giannis gave neutral responses. This suggests that the game interface worked well for some students but not all, pointing to the need for adaptable and intuitive design to accommodate diverse learners. All teachers found the game engaging, though one expressed neutrality. Nonetheless, most observed signs of excitement and sustained attention, indicating that the visual and interactive format successfully captured student interest—a common benefit noted in ASD-related VR studies.

5.1. Social Communication Outcomes

The limited observable changes in social communication behaviors align with Hypothesis 1 (H1), which proposed that children would demonstrate such behaviors during and after participating in the VR intervention.

In our short, one-week trial, only one out of four children demonstrated any notable gains in social interaction. Such limited progress is consistent with the literature suggesting that short-term VR interventions often yield minimal changes in real-life social functioning for children with autism. For example, Frolli et al. [11] found that while VR experiences can be highly engaging, a brief exposure did not immediately translate into improved social initiation or reciprocity in children on the spectrum. Similarly, studies by Ip et al. and Bekele et al. reported mixed or negligible social skill gains after short VR programs (5–6 weeks or less). Based on Alvarado et al.’s [27] and Manju et al.’s [28] studies, it is also important to mention that both studies did not yield the positive results they had anticipated. Both studies pointed out that one potential reason for this outcome could be the limited duration of the sessions. Manju et al. and Alvarado et al. suggested that with more time and additional repetitions, the results might have been more favorable. One likely explanation is that social communication behaviors require time and repeated practice to generalize from a virtual environment to everyday settings. In our study, the children enjoyed and participated actively in the VR social story, but most did not start greeting classmates more often or conversing more fluidly in the span of a week. This outcome underscores a point noted by prior researchers: longer and more sustained interventions are often necessary for children with ASD to internalize and apply new social skills outside of therapy sessions. Additionally, the variability in responses among the four students reflects individual differences that are well-documented in ASD intervention studies. Each child with ASD has a unique profile of needs, cognitive abilities, and comfort with technology. The one student (“Spyros”) who showed moderate improvements in communication may have had a predisposition to benefit from the game—for instance, stronger language ability or a higher motivation to interact—whereas the others might require different types of support. This heterogeneity of response is echoed in earlier VR research: some children show clear benefits in social engagement, while others show little change. Regarding verbal communication, while three teachers did not observe any development in students’ verbal expression, the teacher of Spyros noted moderate improvement. This aligns with prior research indicating potential benefits of VR in enhancing both verbal and non-verbal communication, though such outcomes tend to vary across individuals [16,17,18,19,21,22,23,24,26,36]. It suggests that personalization and sufficient duration are key for VR interventions targeting social skills. An implication for practice is that VR-based social training should be integrated as a long-term, individualized supplement to real-world practice, rather than a one-off activity. With iterative use over time and perhaps adaptive content to match each child’s social understanding level, VR experiences might produce more tangible improvements in face-to-face social communication.

5.2. VR Impact on Emotional Skills

The outcomes related to emotional understanding and empathy were modest, offering only partial support for Hypothesis 2 (H2), which proposed that children would exhibit identifiable emotional responses during the VR experience, and were somewhat incongruent with studies that found broader emotional gains with VR tools [16,17,18,21,22,23,24,25,26,36]. Our results showed only slight improvements in emotional recognition for half of the participants and essentially no change in their empathic responses. In contrast, several previous studies have documented broader gains in emotional skills after children with ASD engaged in VR or similar interactive training [12,14,26]. The discrepancy between those findings and the present study’s limited impact likely stems from differences in intervention length and content depth. Our VR application was used over a very short term, and while it included key emotional scenarios (such as recognizing a friend’s sadness or happiness and choosing how to react), the children may not have had enough repetition to master these concepts. Despite this, the pattern we observed—where one student out of four showed a moderate improvement in emotion-related skills—is not unusual. It mirrors the mixed outcomes noted in certain VR studies that reported only some participants benefiting emotionally.

Similarly, when evaluating how students responded to others’ emotions, only Spyros showed improvement. This again matches limited but positive results from Didehbani et al. [26], Cheng et al. [16], Parsons [22], Lorenzo et al. [36], and Moon [19]. The mixed outcomes suggest that while VR can help some children improve emotional skills, others may not benefit without additional scaffolding.

There are several possible reasons why emotional learning did not generalize widely in our case. One consideration is the fidelity of emotional cues in the VR game. While the visual novel format did present characters with basic facial expressions and scenarios involving feelings, these cues might not have been sufficiently explicit or varied for all learners. Children with ASD often require exaggerated or clearly labeled emotional signals to reliably identify feelings. It has been suggested that virtual interventions for ASD should incorporate enhanced emotional content—such as more pronounced facial expressions, body language, and contextual hints—to effectively teach emotion recognition and appropriate responses. In line with this, an improvement for future iterations of our VR tool would be to refine the emotional storytelling: for example, including a wider range of emotion scenarios (joy, frustration, fear, etc.), using animated characters that visibly show those emotions, and narrating the characters’ feelings. Moreover, empathy-based scenarios could be added to encourage perspective-taking—situations where the player must infer how another character feels and choose a caring response. Such additions might deepen emotional engagement and learning, helping more children not only identify emotions but also practice responding to them in a supportive virtual context.

The findings for RQ2 suggest that a short, single-exposure VR story was insufficient to substantially improve emotional skills for most participants, though it hints at potential benefits given the right conditions. Future research and design improvements should emphasize repeated practice and richer emotional content to harness VR’s capabilities for teaching empathy and emotional understanding to ASD learners.

5.3. Teachers’ Attitudes and Willingness to Use VR

One of the most encouraging findings related to Hypothesis 3 (H3), which proposed that educators would find the VR visual novels acceptable, easy to use, and feasible for classroom implementation. Despite the only modest student outcomes observed over one week, the special education teachers saw clear value in the tool, which has important implications for real-world adoption.

The teachers indicated a willingness to continue using VR-based social stories, reflecting a recognition of the perceived usefulness and ease of use of this approach. In technology integration research, such teacher endorsement is crucial: if teachers believe a tool is beneficial and not overly difficult to implement, they are far more likely to incorporate it into their curriculum. Our results suggest that the VR visual novel met these criteria. Teachers found the system accessible, as evidenced by their feedback that navigation was generally easy for students (teachers rated the interface from neutral to very easy for the students to navigate). In our context, educators clearly perceived the VR game as a useful supplement to their teaching—it captured student interest and provided a novel way to practice social scenarios—and they were comfortable managing it within the class setting. These findings echo broader trends in inclusive education where teachers are increasingly open to leveraging technology, especially interactive and visual tools, to support diverse learners.

This study builds directly on our prior systematic review [12], which highlighted the potential of VR visual novels incorporating social stories for supporting social–emotional learning in ASD contexts, but also noted a lack of empirical classroom-based applications. In response to these gaps, the present study contributes an in situ evaluation of such an intervention, emphasizing teacher feedback, emotional design, and usability challenges. It represents a shift from theoretical synthesis to applied research, offering practical insights grounded in real-world educational use. The teachers’ feedback also provides valuable insights into educational practice and future intervention design. First, the fact that teachers uniformly expressed interest in using the VR tool again implies that integrating VR into special education is feasible when proper support is in place. In this study, the VR sessions were conducted in a structured environment with a special educator present, which likely contributed to the smooth implementation. It highlights that such tools are best used not in isolation but as part of a guided learning activity, where educators can scaffold the experience (for example, pausing to discuss a scenario or helping a student who gets stuck). Teachers in our study saw the VR visual novel as complementary to their teaching rather than a replacement. One teacher noted that repeated exposure to the game could yield better outcomes over time—an insight that concurs with the need for sustained intervention. This suggests that teachers are willing to incorporate the VR sessions regularly (e.g., weekly over a term) if they see potential benefits for student growth. From an educational standpoint, using VR social stories repeatedly across a semester could become a new form of social skills “exercise” integrated into the routine, much like role-playing activities or group discussions are used traditionally.

Teachers’ qualitative comments about what worked and what did not in the VR design can directly inform design improvements. They praised the visual-rich content and simplicity of text, reinforcing that visual supports are key in ASD. This aligns with established practices like visual schedules and social stories on paper, now elevated through VR’s technology and interactive activities, offering higher user engagement. Therefore, developers should continue to prioritize clear, supportive visuals and perhaps add even more visual scaffolds to cater to children who have limited reading skills or who process information visually. Conversely, the issues of impulsive responding and momentary disengagement point to a need for features that can manage attention and pacing. An implication is that VR educational games for ASD might incorporate built-in prompts or restrictions to prevent rushing through content—for instance, forcing a short pause before selecting an answer or having the narrator voice all options before input is accepted. Such a feature could gently guide impulsive learners like the student in our study to slow down and consider their choices, thereby improving the educational value of the interaction. Additionally, maintaining a child’s engagement might require the system to detect inaction or boredom and respond (perhaps with a visual cue or a change in activity) to re-capture attention. These design considerations could enhance the effectiveness of VR interventions.

5.4. Limitations and Future Work

This study has several limitations that constrain the generalizability of its findings and inform directions for future research. First, the study’s small sample size (N = 4) limits the broader applicability of its findings. This design is typical for pilot-stage studies aiming to test usability and acceptance before scaling to larger trials. Future work will incorporate larger samples and repeated sessions to assess generalizability and longitudinal effects. Likewise, according to the studies by Alvarado et al. [27] and Manju et al. [28], neither study achieved the expected positive results. Both studies suggested that the limited length of the sessions might have been a contributing factor to this outcome. Additionally, the sample limited to children diagnosed with Level 1 ASD restricts applicability to the broader spectrum. Children with greater support needs may respond differently to VR-based interventions. Future studies should include a more diverse participant pool across age groups and ASD severity levels to better understand who benefits most and what adaptations may be necessary.

Another constraint was the short observation window—teachers had only one week to assess the tool’s impact. This brief timeframe likely was not sufficient to observe meaningful changes in behavior or skill development. The one-week implementation period may have been too brief to observe more substantial or lasting changes. Most effective interventions cited in the literature span several months, providing more opportunities for adaptation and growth. Future work should extend the intervention over a longer period—ideally through multi-week curricula with repeated sessions—to better assess sustained engagement and learning outcomes.

Another limitation concerns the measurement approach. The custom post-intervention questionnaire used in this study was designed specifically to reflect the VR scenarios and expected social–emotional behaviors, based on teacher input. While it offered valuable context-specific insights, it lacks the psychometric validation of standardized instruments. Future research should consider combining scenario-specific items with validated, theory-based tools to enhance both contextual relevance and measurement reliability.

Technical and design limitations impacted usability. Teachers reported slow scene transitions and lack of features to support impulsive users. One student clicked through choices rapidly, reducing the activity’s learning value. Future iterations should improve technical performance and add interface features that regulate pacing and support self-regulation, such as locking choices until all options are read or providing adaptive prompts.

In light of these limitations, we propose several recommendations for future research and development. Researchers should aim to include larger, more diverse participant groups with a range of autism spectrum levels.

Based on our findings and aligned with related research, we recommend implementing VR-based social story interventions over a minimum of four weeks, with two sessions per week, to allow for repetition and gradual skill internalization. Each session should include opportunities for guided reflection, especially in weeks 2 and 3, where prompting students to explain their decisions can support emotional awareness and perspective-taking. To support diverse learner needs, the interface should allow for adjustable pacing, visual supports (e.g., icons or GIFs), and options for emotional cue intensity. A larger-scale pilot could involve 25–30 students across multiple schools, using a structured design with weekly sessions and pre/post evaluations. Such a scheme would allow for analysis of scalability, differentiated outcomes, and longer-term behavioral change, offering evidence to inform classroom integration and policy-level adoption.

Additionally, giving teachers the ability to adjust the VR experience—such as modifying visual content, pacing, or support strategies—could better accommodate individual student needs. Such flexibility would help create a more personalized and effective learning experience. Future studies should examine the clarity and range of character expressions, testing prototypes with varying levels of facial expressiveness. Including visual examples (e.g., screenshots) could support evaluation and design refinement. Such improvements may enhance emotional recognition for children with ASD by increasing the emotional fidelity of the VR experience. Lastly, investing in improved hardware would reduce technical problems like lagging visuals, which can interrupt focus and reduce the educational value of the tool. Reliable, high-performing equipment is key to making VR educational experiences successful and sustainable.

6. Conclusions

This study examined the use of a VR-based visual novel integrating social stories to support social–emotional learning in children with ASD. Findings related to RQ1 and RQ2 indicated that observable social communication and emotional responses were limited during the short-term implementation. While most students did not show improvements in social interactions or emotional responses, one participant demonstrated modest gains in both areas, suggesting potential benefits for some learners even in brief interventions.

In contrast, teacher feedback (RQ3) was highly positive. All participating educators expressed a willingness to reuse the tool, citing its engaging design, accessibility, and suitability for the needs of students with ASD. They highlighted the VR environment’s visual strengths and the value of a controlled setting for practicing social behaviors without real-world stressors. This level of acceptance suggests strong practical feasibility for classroom integration.

Although the results confirm that a single week of VR use is insufficient to produce consistent behavioral improvements, the combination of strong teacher endorsement and observed student engagement points to the educational potential of this approach. As with traditional interventions, time and repetition appear necessary for skills to generalize. The use of VR visual novels should be viewed as a supplementary method that requires thoughtful planning and longer implementation to be effective.

Future research should explore extended and more intensive use of VR social stories, perhaps combining them with other therapeutic techniques, to determine if improvements in social communication and emotional empathy can be achieved with time. More frequent sessions and a larger-scale sample may also yield stronger outcomes, enhancing the robustness and generalizability of findings. Additionally, developing more adaptive and personalized VR tools will be important. This includes refining the content to be more inclusive (addressing a wider range of abilities and scenarios), adding adaptive features that can tailor the experience to each child’s needs (for example, adjusting difficulty, providing additional support or feedback based on performance), and ensuring technical reliability and smooth user experience. By pursuing these directions, researchers and practitioners can work.