Examining Virtual Reality Interventions for Social Skills in Children with Autism Spectrum Disorder: A Systematic Review

Autism Spectrum Disorder (ASD) is a neurodevelopmental disorder that is characterized by limitations in social communication and interaction, self-repetitive behaviors, and the presence of limited interests (American Psychiatric Association [APA], 2022). The prevalence of ASD, which typically emerges in the first years of life, is increasing at an alarming rate due to multiple factors, including the broadening of diagnostic criteria, heightened public awareness, and more frequent diagnoses among women and adults (Russell et al., 2022). A report published by the U.S. Centers for Disease Control and Prevention (CDC) indicates that one in 36 children is affected by ASD (Maenner et al., 2023). This represents an increase from the previous CDC report, published in 2021, which concluded that one in 44 children is affected by ASD (Maenner et al., 2021).

One of the distinguishing characteristics of ASD is persistent deficits in social communication and social interaction across multiple contexts (APA, 2022). Difficulties in initiating and maintaining social communication, interaction, and reciprocal engagement are among the core features of children with ASD. Therefore, targeting the domain of social skill development is critically important for the success of children with autism (Silveira-Zaldivar et al., 2021).

In their 1988 study, Morgan and Jenson defined social skills as comprising both verbal and nonverbal behaviors that facilitate mutually beneficial social interactions. Goodwin (1999) defined social skills as abilities that enable individuals to function independently in society and enhance their quality of life. As can be gleaned from the definitions provided, social skills serve to enhance and enrich the quality of individuals' lives. This is achieved by enabling them to interact effectively with others in their daily lives, develop friendships, participate in leisure activities, and become integrated into a group (Turhan & Vuran, 2015). Consequently, these experiences are of paramount importance for individuals with ASD (Vuran, 2007).

Experts have historically utilized various methods to teach social skills to children with ASD, including video modeling, behavioral skills training, social stories, peer- and sibling-mediated interventions, antecedent interventions, manualized curricula, pivotal response training, milieu teaching, naturalistic interventions, reinforcement, script fading, discrete trial training, and self-management (Radley et al., 2020). However, they have faced challenges due to the significant time and effort required to create diverse instructional scenarios and the difficulty posed by certain non-replicable scenarios (Dyson et al., 2019). Consequently, experts have been compelled to identify supplementary intervention strategies that are more suitable and accessible. Virtual reality (VR), which has the capacity to integrate the tangible and the virtual, represents one such alternative intervention method. An increasing number of studies have demonstrated the efficacy of VR in treating children with ASD (Amaral et al., 2018; Babu et al., 2017; Bozgeyikli et al., 2016; Hocking et al., 2022; Kim et al., 2024; Maskey et al., 2019; Moon & Ke, 2023; Schmidt & Glaser, 2021; Welch et al., 2010; Yang et al., 2018).

The increase in the prevalence of ASD (Maenner et al., 2023), the fact that these individuals are typically enthusiastic about using digital media technology (Chen et al., 2019; Lahiri et al., 2013; Moore & Calvert, 2000; Parsons & Mitchell, 2002; Schmidt & Schmidt, 2008), and have strong visual memories (Bozgeyikli et al., 2018) have led to the view that digitally delivered interventions may be beneficial for individuals with autism. The aforementioned factors, in light of the technological advancements, conventional methodologies have been supplanted by digital approaches in the education of children with ASD (Mazon et al., 2019). Technology-assisted instruction and intervention (TAII) represents an evidence-based practice for individuals with ASD (Hume et al., 2021). These interventions encompass the use of more common technologies, such as computers and mobile devices, as well as more advanced technologies, including virtual reality (VR), augmented reality (AR), mixed reality (MR), and robots (Carnett et al., 2023).

Coates (1992) defined VR as electronic simulations that can interact with three-dimensional environments through head-mounted displays and wired clothing. Gold (1993) defined it as an alternative world full of computer-generated images that can react to human movements, supported by expensive hardware such as video and data gloves. However, advances in display technologies have rendered the costly and scarce devices referenced in the definitions accessible to the general public (Srivastava et al., 2019). In virtual reality (VR) systems, stimuli are presented to users in three dimensions. This is frequently achieved through the use of a computer monitor, head-mounted display (HMD), and CAVE (Cave Automatic Virtual Environment). Desktop VR, also called non-immersive VR, is the simplest form of VR, presenting users with high-resolution panoramic images via a standard desktop computer. In desktop VR, users use a mouse or keyboard to move and explore the virtual environment (Berki, 2019). HMDs are a type of virtual reality (VR) headset that provides the user with a fully immersive virtual environment. CAVE is a system that presents a three-dimensional environment using two-dimensional projected images arranged around the user, thereby creating an immersive environment. In comparison to CAVE, HMD VR and desktop VR are more convenient and provide users with more cost-effective solutions (Feng et al., 2022). The appeal of VR environments, coupled with the ability of users to interact with and control virtual objects and focus on the environment with ease, has positioned this technology as an effective tool in educational processes (Chen, 2010; Wang et al., 2022). As virtual reality technology is capable of creating a three-dimensional simulation of real-world environments (Baileson et al., 2008) and allows for the repetition, manipulation, and control of social situations (Lorenzo et al., 2016; Parsons, 2016; Siano et al., 2015), its use in the field of autism spectrum disorder (ASD) is inevitable. Given that social interaction and communication disorders are the most prevalent issues among children with ASD (APA, 2022), VR interventions have been primarily directed towards social skills training (Lorenzo et al., 2016; Yiğit & Sani-Bozkurt, 2021).

In order to gain an understanding of the existing literature on the use of virtual reality in the context of autism spectrum disorder (ASD), we undertook an initial review of previous studies in this field. In a systematic review study, Mesa-Gresa et al. (2018) sought to elucidate the effects of virtual reality on ASD. The researchers did not impose any limitations with respect to the skills under consideration and included a total of 51 articles in their research. Mesa-Gresa et al. (2018) demonstrated that the area of social skills was the most extensively studied. In their review study, Thai and Nathan-Roberts (2018) concentrated on the role of social skills in virtual reality (VR) use. The study comprised a mere five pages and encompasses only nine articles. In their systematic review, Li et al. (2023) synthesized virtual reality studies on individuals with ASD conducted between 2010 and 2022, revealing that the most prevalent area of focus was social and emotional skills. Similarly, Yiğit and Sani-Bozkurt (2021) obtained comparable findings. The fact that social skills are the area of focus for the majority of researchers in this field demonstrates the necessity for further investigation. Liu (2023) and Kadir et al. (2023) conducted reviews of VR applications with a focus on social skills in children with ASD. However, only five studies were included in both reviews. In recent years, there has been a surge in more affordable and user-friendly VR technology, expanding its accessibility beyond specialized research settings. For instance, head-mounted displays have become increasingly cost-effective, and desktop-based systems now require minimal additional equipment. These advances indicate that earlier reviews may not fully capture the latest technological possibilities, thereby underscoring the need for a new comprehensive analysis of VR-based interventions. Unlike previous reviews, this systematic review examined 31 studies published over the past decade (2014–2024), focusing exclusively on children with ASD. This approach offers a more up-to-date and comprehensive perspective on how VR technology contributes to social skill development in childhood. In addition, different VR systems—such as CAVE, desktop VR, and HMD—were compared in terms of cost, technical requirements, and multi-user support, providing a more nuanced assessment of their impact on social skills. By also addressing topics rarely covered in earlier reviews, such as age restrictions, cybersickness, and gender balance, the methodological scope of the study is broadened. Despite limitations related to small sample sizes and issues with generalization, this review highlights the strong potential of VR-based interventions to enhance social skills in children with ASD and aims to fill the gaps noted in the existing literature. The lack of a recent comprehensive systematic review in this regard increases the importance of this study.

The objective of this systematic review is to provide a comprehensive summary of studies employing virtual reality (VR) interventions to enhance social skills in individuals with autism spectrum disorder (ASD) over the past decade. This review will evaluate the literature, including the methodology utilized, the targeted social skills, the VR technology employed, and will facilitate comparison of different studies in terms of their methods and execution procedures for those engaged in research in this field. Therefore, to guide this review and provide a clear focus, we formulated the following research questions:

A total of 634 studies were identified through a comprehensive search of the following databases: Google Scholar (n = 357), EBSCOhost (n = 211), IEEE Xplore (n = 25), ScienceDirect (n = 21), and PubMed (n = 20). Furthermore, the initial researcher conducted an additional search on artificial intelligence but was unable to identify any additional studies apart from those already identified. Additionally, two further studies were accessed through the bibliographies of the identified studies. Following the removal of duplicates (n = 336), the remaining 300 studies were then scanned independently and simultaneously by all three authors, with irrelevant studies (n = 199) subsequently eliminated.

We examined the full texts of 101 studies meeting our initial inclusion criteria. After resolving any discrepancies among the authors, we excluded 70 studies from further consideration. The reasons for exclusion were as follows: 26 studies were excluded due to incorrect results, 17 due to incorrect year of publication, 17 due to incorrect study design, 6 due to incorrect intervention, 2 due to incorrect comparator, 1 due to incorrect diagnostic population, and 1 due to incorrect VR approach. Incorrect study design referred to studies that were non-empirical or lacked an appropriate methodological framework (e.g., purely theoretical papers, commentaries). Incorrect comparator indicated that the control or comparison intervention did not align with our VR focus (e.g., comparing two non-VR methods). Incorrect results pertained to studies that did not actually measure or report outcomes related to social skills, or whose reported results were insufficiently detailed to assess effectiveness. Finally, incorrect VR approach referred to studies utilizing a mode of VR application outside our definition (e.g., a non-HMD or non-desktop approach, or a different technology that did not meet our VR criteria). The remaining 31 studies were included in the study, and data extraction was conducted. The data obtained as a result of the data extraction process are analyzed below under the following headings: diagnoses of the participants; age range and gender of the participants; social skills addressed in the study; main technology used for both hardware and software; methodology; limitations and improvements in the study; and conclusions. The data flow chart is provided in Fig. 1 for reference (Table 2).

In this systematic review, we examined 31 studies published over the past decade (2014–2024) that investigated the effectiveness of virtual reality (VR)-based social skills interventions for children with autism spectrum disorder (ASD). The results indicate that the majority of studies were published in 2018 (n = 6) and 2022 (n = 8), suggesting a growing interest in VR research in recent years. The findings further reveal that most studies focused on children aged 6–15 (n = 599), while there was limited participation from those under 3 years or over 15 years of age. Companies providing VR applications have recommended that VR technology not be used by individuals under the ages of 12 or 13 (Chen et al., 2019). For instance, the minimum age is specified as 10 for the Meta Quest (Meta, 2024), and 13 for the Apple Vision Pro (Apple, 2024). View-Master VR, developed exclusively for children, sets the lowest usage age at 7. These age-related restrictions, combined with the complexities of diagnosis and intervention processes in early childhood, may explain why research predominantly involves the 6–15 age group. Meanwhile, older adolescents and adults with ASD have different educational and therapeutic needs but appear to be underrepresented in VR studies, underscoring the importance of adapting both equipment and content for different age groups in future research.

A notable gender disparity was observed across the studies, with 485 male participants compared to 113 female participants. In addition to the established finding that ASD is more frequently diagnosed in males (Maenner et al., 2023), some evidence suggests that girls may be overlooked in the diagnostic process due to “camouflaging” of symptoms (D’Mello et al., 2022; Hiller et al., 2016). Further research is therefore needed on how to tailor VR interventions to girls’ unique profiles. This gender imbalance could limit the generalizability of current findings and serve as a critical consideration for future research.

Social and communicative deficits in ASD can be grouped into three categories—social-emotional deficits, nonverbal communication deficits, and difficulties in establishing and maintaining relationships (APA, 2022). The studies included in this review addressed all three categories. Social interaction with virtual avatars or peers was the most prominently targeted skill area, wherein various VR-based scenarios were presented to participants, and their ability to engage in social interaction was subsequently assessed. Compared to other disability groups, individuals with ASD experience greater challenges in comprehending and responding to emotional expressions (Black et al., 2017). However, emotional recognition in children with ASD is generally inconsistent, partly due to substantial differences in sample sizes, tasks, and participant characteristics (Lievore et al., 2023). The studies reviewed here that involved emotion recognition and discrimination (Didehbani et al., 2016; Frolli et al., 2022; Ip et al., 2016, 2018; Lorenzo et al., 2016; Tsai et al., 2020; van Pelt et al., 2022) also exhibited variations in outcomes. For example, while participants in the studies by Didehbani et al. (2016), Tsai et al. (2020), and van Pelt et al. (2022) showed improvements in emotion recognition following VR-based interventions, Frolli et al. (2022) reported no difference in acquisition times for emotion recognition when comparing traditional methods and VR. Ip et al. (2016) found positive differences in pre- and post-test emotion recognition scores, whereas a subsequent study (Ip et al., 2018) reported no significant improvement. These discrepancies may stem from differences in participant profiles or the specific VR scenarios employed. Future research could address such inconsistencies by standardizing emotion recognition protocols, enhancing avatar realism, and examining diverse participant groups (e.g., level 1 ASD, those with comorbid diagnoses).

Among the various VR systems employed for ASD interventions—CAVE, desktop, and head-mounted display (HMD)—each offers distinct features. The choice of system is particularly important for practitioners. In this review, desktop-based systems emerged as the most frequently used, a finding consistent with Dechsling et al. (2021). However, a systematic review by Yiğit and Sani-Bozkurt (2021) reported desktop-based systems as the least used. While the studies included here suggest that various VR systems can enhance the social skills of individuals with ASD, the question of which system is most suitable for teaching social skills remains open. Lorenzo et al. (2016) found that immersive environments may elicit more appropriate emotional behaviors compared to desktop settings, and other studies (Chen et al., 2019; Herrero & Lorenzo, 2020) similarly support the utility of immersive VR systems. Yet, CAVE systems require more equipment than HMD or desktop VR, potentially escalating costs. Notably, only four of the studies (Ip et al., 2016, 2018; Lorenzo et al., 2016; Tsai et al., 2020) used CAVE, indicating limited comparative data across multiple VR formats. Additionally, Halabi et al. (2017) applied three VR types to both children with ASD and typically developing peers and found the highest satisfaction levels with CAVE systems, as well as shorter response times in children with ASD. Cooperative VR environments (supporting multiple users simultaneously) are also believed to be effective for social skills training.

Special education is a highly diverse field, requiring multiple research designs (Odom et al., 2005). Alongside the four frequently used approaches—descriptive, correlational, experimental, and qualitative—mixed-method studies have become increasingly popular (Cook & Cook, 2016). Methodologically, most studies in this review employed experimental or quasi-experimental designs, yet only nine included a control group (Amat et al., 2021; Kourtesis et al., 2023; van Pelt et al., 2022). The absence of control groups limits the capacity to conclusively compare VR interventions with alternative methods. Small sample sizes (n = 16 studies) and difficulties in generalization (n = 8 studies) were also commonly reported. The challenges that children with ASD face in generalizing skills (Yerys et al., 2009) are closely tied to how closely VR environments mimic real-life conditions (Rajendran, 2013). Hence, factors like high-resolution visuals, avatar realism, and scenarios reflective of daily life contexts could be critical for success in future studies.

Moreover, cybersickness was mentioned in only one study (Kourtesis et al., 2023), suggesting that this issue is often overlooked in VR research. This gap may indicate a lack of consistent protocols or measurement tools for safeguarding participant well-being. Symptoms such as nausea and dizziness could pose (Davis et al., 2014; Stauffert et al., 2020) particular challenges for children with ASD, who often have heightened sensory sensitivities. Although recent hardware and software aim to minimize these effects by reducing latency and jitter (Farmani & Teather, 2020), more sensitive measurement and reporting tools are needed to fully address cybersickness in VR studies.

In conclusion, the reviewed studies suggest that VR-based interventions can serve as a valuable tool for enhancing social skills in children with ASD. Nevertheless, several factors—small sample sizes, absence of control groups, gender imbalances, and cybersickness—limit the scope and robustness of current findings. Future research should incorporate more diverse samples in terms of age and gender, conduct comparative analyses of different VR systems, and employ controlled, longitudinal designs (including follow-up measurements) to yield more generalizable results. Such efforts will help clarify how VR technology can sustainably contribute to the social skills development of children with ASD.

In conclusion, this study examined the effects of using virtual reality (VR) technology in social skills training by synthesizing the findings of studies conducted over the past decade, including VR interventions for social skills of individuals with autism spectrum disorder (ASD). The findings indicated that VR-based interventions can facilitate the creation of a secure environment in which individuals with ASD can engage in the practice of social skills in a manner that is both safer and more comfortable. Furthermore, these interventions have the potential to enhance participation by providing participants with tasks that are motivating and engaging. The use of VR technology is becoming increasingly commonplace, with new equipment being introduced on a daily basis. Nevertheless, it is imperative that experts exercise caution when utilizing VR technology, particularly in individuals with ASD. In light of the findings and cautions articulated by the research sponsors, it is evident that this technology is appropriate for individuals above the age of 12 or 13. All varieties of VR have exhibited promise in the field of social skills instruction. It is established that desktop-based systems are the most cost-effective option; however, users may prefer CAVE-style VR systems. Further research is required to elucidate the distinctions and consequences of the various types of VR, namely HMD, CAVE, and desktop-based systems. It is recommended that future studies address the limitations of existing research, including the use of small sample sizes and the lack of generalizability. Additionally, there is a need for more studies on gender balance in VR research. Overall, the evidence suggests that VR technology has the potential to enhance the social skills of individuals with ASD, but further research is necessary to confirm this hypothesis.