A Systematic Review of Virtual Reality Applications for Adaptive Behavior Training in Individuals with Intellectual Disabilities

Abstract

(1) Deficits in adaptive behavior significantly hinder individuals with intellectual disabilities from performing essential daily tasks and participating in community life. Although virtual reality shows promise for supporting adaptive behavior in this population, systematic reviews on this topic remain scarce. (2) Methods: Twenty-five experimental studies from the databases Web of Science, PubMed, Scopus, and ERIC, published between 2005 and 2024, were analyzed in the context of a systematic review. (3) Results: The studies revealed a significant surge in research on VR interventions for adaptive behavior in individuals with intellectual disabilities, particularly after 2021. The most frequently applied domain was practical skills, while social and conceptual skills received relatively less attention. Most studies employed high-immersion head-mounted displays as the primary technology type and adopted controller-based unimodal interaction as the dominant interaction mode. Pedagogical strategies such as ABA, structured teaching, and contextual learning are favored in interventions. (4) Conclusions: VR interventions have been increasingly applied to support adaptive behavior development in this population. However, further exploration is needed to tailor VR designs to better accommodate the individual differences and specific needs. This review synthesizes current evidence, identifies key trends and limitations, and offers guidance for future research.

1. Introduction

Intellectual disability (ID), formerly known as mental retardation, involves significant limitations in intellectual functioning. It is typically diagnosed when individuals score at least two standard deviations below the mean on standardized cognitive assessments conducted during the developmental period (American Psychiatric Association, 2013). In addition to cognitive impairment, this population also exhibits substantial deficits in adaptive functioning across conceptual, social, and practical domains. Notably, individuals with similar cognitive abilities may still show substantial differences in how well they adapt to social and daily life due to varying levels of adaptive behavior. In fact, the severity of these deficits is a critical determinant of an individual’s ability to perform everyday tasks and participate in community life (Schalock et al., 2021). Despite increased support efforts, individuals with IDs remain at a disadvantage in developing adaptive behavior, with limited access to meaningful academic and life-skill learning opportunities (de Oliveira Malaquias & Malaquias, 2016). To ensure timely and effective interventions and reduce the risk of lifelong adverse outcomes, we need to integrate advanced, evidence-based technologies in collaboration with professionals from various disciplines to better support this group.

In recent years, a growing variety of intervention strategies aimed at adaptive behavior have been applied to them. For example, applied behavior analysis (ABA) has shown success in lowering challenging behaviors, encouraging social interaction, and promoting skill development (Burns et al., 2019). Cognitive behavioral therapy (CBT) could also improve cognitive processing and emotional recognition (Dagnan & Jahoda, 2006). Other strategies, such as video modeling, contextual learning, and peer-mediated instruction, have shown promise in enhancing social skills (Walton & Ingersoll, 2013). While traditional interventions have demonstrated effectiveness, their scalability and adaptability in diverse real-world settings remain limited, highlighting the need for more flexible, technology-enabled approaches. In response to these limitations, a growing body of research has begun to explore the utility of digital technology-based interventions in this field (Kirkpatrick et al., 2022).

With the advancement of digital technologies, innovative tools such as virtual reality (VR) and computer-assisted instruction (Kurzeja et al., 2024) have increased the potential for special education and rehabilitation. Specifically, VR is an immersive interface that allows users to engage with realistic and captivating computer-generated environments (Schultheis & Rizzo, 2001). It could be carried out in homes, hospitals, and classrooms as well as in other settings (Q. Yang & Zhong, 2021). Research demonstrates that gamification components in VR applications improve user motivation, engagement, and adherence to treatment objectives (Suhendi & Murli, 2024). Therefore, exploring the use of VR for training adaptive behavior in individuals with IDs aligns with the broader trend of integrating modern technology into special education.

1.1. Adaptive Behavior

Adaptive behavior forms the foundation for individuals with IDs to manage daily life, interact with society, and achieve meaningful independence. Since its first introduction by Heber (1959) as one of the early diagnostic criteria for ID, the concept of adaptive behavior has achieved broad acceptance in clinical and educational research. In 2002, the American Association on Mental Retardation (AAMR), now known as the American Association on Intellectual and Developmental Disabilities (AAIDD), defined adaptive behavior as a collection of three domains: conceptual, social, and practical skills (Schalock et al., 2002). Specifically, conceptual skills refer to cognitive and academic abilities like language, literacy, money use, and self-determination. Social skills involve interpersonal interaction, emotional regulation, and social conformity. Practical skills focus on everyday tasks essential for independent living (Harrison & Oakland, 2015).

The Vineland Adaptive Behavior Scales Third Edition (Vineland-3) (Sparrow et al., 2020) and the Adaptive Behavior Assessment System-3 (ABAS-3) (Harrison & Oakland, 2015) are the main tools for evaluating adaptive behavior, and they both rely on feedback from clinicians, teachers, and parents. As previously mentioned, interventions to improve adaptive behavior in this population often rely on evidence-based, experiential, and context-based teaching strategies (Lin-Siegler et al., 2016). To help students identify environmental cues and react correctly, teachers sometimes use simulated instructional scenarios to educate students in recognizing environmental cues and responding appropriately (Sun & Brock, 2023). Studies have emphasized how crucial it is to expand learning opportunities into real community contexts to promote the transfer and generalization of acquired skills (Neely et al., 2018).

However, in real-world teaching environments, both classroom simulations and community-based learning are constrained by time, staffing, and resources. These constraints hinder both the consistency of training and students’ ability to apply skills in varied settings. To address these challenges, educators and researchers have increasingly turned to technological solutions. In recent years, digital technologies like VR, augmented reality (AR), and artificial intelligence (AI) have become increasingly integrated into adaptive behavior training. Researchers have begun to explore how immersive and interactive virtual environments can help improve individuals’ performance, particularly in practical and social skills areas. Studies often use specific adaptive behavior subdomains as indicators to assess the relevance and outcomes of technology-based interventions.

1.2. Virtual Reality

VR is a technology that combines multiple components, including stereoscopic displays, scene modeling, and natural interaction techniques. By integrating visual, auditory, and tactile stimuli, VR creates immersive environments that enable users to interact naturally through sensor-based and computer-mediated interfaces (Rizzo & Koenig, 2017). Burdea and Coiffet (1992) characterized the key features of VR with the “3Is”: immersion, interaction, and imagination. Among these, immersion and interaction are widely regarded as the most essential and broadly accepted attributes of VR (Gao et al., 2016). Immersion refers to the realism of multisensory VR environments, while interaction concerns how users engage and communicate within them.

The immersive quality of VR primarily arises from its ability to integrate diverse representational modalities and sensory stimuli within a virtual environment. VRs are typically categorized as immersive, semi-immersive, or non-immersive, depending on the level of sensory engagement they provide (Salatino et al., 2023). Non-immersive systems use standard input devices (e.g., mouse, keyboard), allowing users to interact with virtual content while remaining aware of the real world. Semi-immersive systems offer a deeper level of engagement, often using large display screens, haptic feedback equipment, and infrared cameras to create a more convincing sense of presence (Wiederhold, 2019). Fully immersive VR uses head-mounted displays (HMDs) and 3D input devices like motion controllers or haptic gloves. These tools allow users to feel deeply engaged and interact directly with virtual environments (Zhan et al., 2024).

VR interaction refers to the various methods by which users engage with virtual environments through input devices and interfaces that enhance immersion. In contrast to conventional human-computer interaction, immersive VR facilitates more natural and intuitive engagement via sensor-driven technologies. Instead of clicking or typing, users can manipulate virtual objects using gestures, motion, or even eye movements. F. Zhang et al. (2016) categorized VR interaction into 3D, gesture-based, and mobile device-based modes. Building on this, Yi et al. (2024) further distinguished between unimodal and multimodal VR interactions based on the sensory modalities and devices involved. Unimodal interaction uses a single perceptual channel, such as voice, gesture, or touch. In contrast, multimodal interaction incorporates two or more modalities simultaneously. Systems such as the HTC Vive and Noitom Hi5 exemplify multimodal interaction by integrating multiple input channels to enrich user engagement. In summary, the intuitive and sensory-rich nature of VR aligns well with the concrete thinking styles commonly observed in individuals with IDs. These features may offer unique advantages for promoting adaptive behavior, as discussed in the following section.

1.3. The Application of VR in Individuals with IDs

As a form of assistive technology (AT), VR is often implemented alongside familiar tools like iPads, text-to-speech software, and word processors, all of which collectively support accessible learning environments (Cooper et al., 2005). Given that many students with IDs may already have experience with such technologies in school, VR represents a natural extension of these supports. Al-Azawei et al. (2016) have pointed out that multimodal learning tools help learners process and output information. This prior experience not only builds a foundation of digital literacy but also plays a significant role in enhancing device acceptance. In this review, acceptance refers to users’ subjective responses to VR use, including engagement and tolerance of side effects such as discomfort. In contrast, individuals with limited exposure to AT, particularly older adults with IDs, might face additional barriers due to unfamiliarity with interactive technology (Álvarez-Aguado et al., 2025). This difference highlights the importance of considering users’ prior experience and their level of device acceptance when designing effective VR interventions.

Having established the link between AT and VR, it is essential to examine how VR-based interventions influence adaptive behavior development in this population. Some authors included specific interventions in their reviews. For example, Klavina et al. (2024) examined the use of AT to enhance practical skills for individuals with autism spectrum disorders (ASD) and IDs. They emphasized AT’s benefits for social participation and daily living independence. While their study emphasized the value of AT, VR, and mobile apps in supporting adaptive functioning, it did not offer a comprehensive analysis of the conceptual and social skill domains. Similarly, Montoya-Rodríguez et al. (2023) reviewed a VR and AR program targeting social skills, which covers one domain of adaptive behavior. However, their review included only six studies with small sample sizes, limiting the generalizability of the findings. X. Yang et al. (2024) also reviewed evidence for VR-based social skills training in individuals with ASD, though only 6 out of 14 studies were rated as high quality. However, despite these individual studies and reviews, a systematic understanding of how VR contributes to adaptive behavior across its conceptual, social, and practical domains in this group remains limited. This review seeks to address this gap.

1.4. Research Questions

Given the increasing emphasis on inclusive education and personalized interventions, this review contributes to both theoretical understanding and practical application by mapping how VR can support adaptive development among individuals with IDs. It also identifies gaps and provides guidance for future design and implementation of VR-based special education interventions.

Based on this, our review has two main aims: (1) to examine experimental studies on VR applications for individuals with IDs and (2) to explore future directions for using VR to enhance adaptive behavior. In line with these objectives, we address the following research questions:

RQ1: What are the current trends in VR-based interventions targeting adaptive behavior in individuals with IDs?

RQ2: What are the demographic and diagnostic characteristics of the research participants?

RQ3: How has VR been used across different domains of adaptive behavior?

RQ4: What types and characteristics of VR devices are employed in these interventions?

RQ5: What pedagogical strategies and outcomes have been reported in the included studies?

2. Methods

The systematic review method involves the comprehensive and structured collection of relevant research findings on a specific topic. This approach applies rigorous literature evaluation methods to screen studies that meet predefined inclusion criteria, extract relevant data, and perform either quantitative analysis or qualitative assessment. The goal is to systematically summarize the issue and draw reliable conclusions (Andrews, 2005). The study followed the PRISMA 2020 guidelines (Page et al., 2021) to ensure transparency, completeness, and methodological rigor. The literature identification, screening, eligibility assessment, and inclusion processes are illustrated in the PRISMA flow diagram (Figure 1), and the completed PRISMA checklist is included in the Supplementary Materials.

2.1. Search Strategy

Our study draws on data from four major academic databases: Web of Science, PubMed, Scopus, and ERIC. We used a combination of keywords and Boolean operators. The search terms included: (“virtual reality” OR “VR”) AND (“adaptive behavior” OR “adaptive functioning” OR “functional skills” OR “daily living skills” OR “conceptual skills”) AND (“intellectual disability” OR “developmental disability”). To ensure transparency and reproducibility, we adapted the search syntax for each database, incorporating controlled vocabularies and field-specific operators. For example, in PubMed, we used MeSH (Medical Subject Headings) terms such as “virtual reality” [MeSH] and “intellectual disability” [MeSH], which expand the search to include conceptually related terms and synonyms. In ERIC, we applied the Thesaurus of ERIC descriptors. For Web of Science and Scopus, truncation (“adaptive behavior*”) was used to cover variations of terms. The full search strings for each database are provided in the Supplementary Materials. We searched English-language studies published from January 2005 to December 2024 and initially identified 1874 articles on VR applications for individuals with IDs.

2.2. Eligibility Criteria

We established the following inclusion criteria: (1) The participants were individuals diagnosed with IDs, confirmed by clinical or educational criteria; (2) the research employed an experimental or quasi-experimental design; (3) VR was employed as the primary intervention tool; (4) the article was published in a peer-reviewed journal during 2005–2024; and (5) the intervention aimed to train, improve, or assess adaptive behavior (including conceptual, social, or practical skills) in individuals with IDs. The exclusion criteria were as follows: (1) The study did not include diagnosed individuals with ID; (2) the research did not explain the intervention procedures, outcome measures, or results; (3) the experimental method used non-experimental designs (e.g., theoretical papers, descriptive studies, or qualitative-only studies); and (4) the type of literature included conference proceedings, review articles, editorials, commentaries, or unpublished reports. Additionally, we did not include grey literature sources (e.g., dissertations and trial registries such as ClinicalTrials.gov) to maintain a focus on peer-reviewed evidence.

2.3. Selection of Studies

As illustrated in the PRISMA flow diagram in Figure 1, a total of 1874 records were initially identified. After removing duplicates and irrelevant records, 1535 records remained. Two reviewers independently screened the titles, abstracts, and keywords of these records, excluding irrelevant or ineligible studies, which resulted in 434 potentially eligible articles. Subsequently, the same reviewers then independently assessed the full texts of these articles based on the predefined inclusion and exclusion criteria. As a result, 25 studies were selected as the final sample for this systematic review. The inter-rater reliability was 87%, calculated using the formula proposed by Huberman and Miles (2002). Discrepancies were resolved through discussion until consensus was reached.

2.4. Data Extraction and Coding

Guided by the research questions, we developed a coding framework to extract and summarize key information from each study. For RQ1 (VR application trends), we extracted publication year, country, and research design to analyze temporal and geographical patterns. For RQ2 (participant characteristics), we coded data on sample size, gender, age, level of intellectual functioning, prior experience with AT (categorized as yes or no), and device acceptance (classified as high, moderate, or low). In line with the American Psychiatric Association (2013), ID was classified into four levels based on intelligence quotient (IQ): mild (IQ 50–69), moderate (IQ 35–49), severe (IQ 20–34), and profound (IQ below 20). For RQ3 (intervention domains), adaptive behavior was categorized into conceptual (CS), social (SS), and practical skills (PS) based on AAMR (Schalock et al., 2002). In RQ4 (VR intervention features), the type of VR device was coded based on its level of immersion: non-immersive VR (NIVR), semi-immersive VR (SIVR), and immersive VR (IVR) (Salatino et al., 2023). The interaction mode of the VR systems was further categorized as unimodal and multimodal (Yi et al., 2024). Finally, we coded the main pedagogical strategies and reported outcomes related to adaptive behavior (RQ5). No further effect size calculations or statistical analyses were performed beyond those in the original studies. Two independent reviewers coded the data using predefined categories, achieving a Cohen’s kappa of 0.78 for inter-rater reliability. Any discrepancies were resolved through discussion until consensus was reached. SPSS 29.0 and Excel were used for data analysis in this study.

2.5. Study Risk-of-Bias Assessment

Two reviewers independently evaluated the risk of bias for each study. After discussion, we used the JBI Critical Appraisal Tools (Tufanaru et al., 2020) based on study design, including the JBI Checklist for randomized controlled trials, quasi-experimental designs, and case series (Barker et al., 2023). Overall risk of bias was determined by the proportion of items rated “Yes”: ≥70% indicated low risk, 40–69% moderate risk, and <40% high risk. Inter-rater agreement was calculated using percentage agreement, and after resolving discrepancies through discussion, a total of 18 studies were classified as low risk and 7 as moderate risk of bias. No studies were assessed as having a high risk of bias.

3. Results

3.1. Current Trends

The 25 included studies span 13 countries and regions, including China, the United Kingdom, France, and others, reflecting a global interest in applying VR to support individuals with IDs. Geographically, most studies were conducted in developed Western countries. Australia led with 16% (n = 4), followed by the United States and Italy (12% each). Research activity remained stable from 2005 to 2017, with a notable rise since 2021 (see Figure 2), indicating increased scholarly attention to VR in special education. Citation analysis shows that 68% (n = 17) of the studies have fewer than 20 citations, while the remaining 32% (n = 8) have been cited more than 20 times, including 4 with over 80 citations. The most frequently cited study, published in Yalon-Chamovitz and Weiss (2008), explored VR’s feasibility in promoting leisure activities for individuals with IDs. It has received 189 citations, highlighting its academic impact.

3.2. Participants

A total of 911 individuals with IDs ages 5 to 75 participated across the included studies (see

Table 1. Basic information on included studies (n = 25).

Authors	Participants Characteristics	Prior Experience	Device Acceptance	Study Purpose	Design and Domains	Devices	Pedagogical Strategies	Outcomes
Burke et al. (2021)	N = 153 (111M, 42F) Mean Age: 21; IQ: Mild and moderate	N/A	N/A	To evaluate the ViTA system for enhancing job interview skills	Single-group pre-post test; PS	NIVR (Computer-based VR)	Structured curriculum, behavioral skill breakdown, timely feedback mechanism	Improved interview skills in individuals with IDs.
Butti et al. (2024)	N = 28 (21M, 7F) Age: 7–25; IQ: Mild	N	Moderate	To examine the effects of VR-Spirit on social prediction and neuro-psychological outcomes	Randomized controlled trials; SS	NIVR (Computer-based VR)	Error-feedback learning; observational learning	Improved social prediction abilities.
Capallera et al. (2023)	N = 18 (11M, 7F) Age: 12–16; IQ: Mild and moderate	N/A	High	To assess VR training for public transport use and its real-world applicability	Single-group pre-post test; PS	IVR (HTC Vive Pro)	Learning by doing	Effective real-world transfer and generalization of learned skills.
Yalon-Chamovitz and Weiss (2008)	N = 33 (23M, 10F) Age: 20–39; IQ: Moderate and severe	Y	High	To explore VR feasibility for leisure activities for individuals with IDs	Randomized controlled trials; SS	NIVR (Xtreme video capture VR system)	N/A	VR provided engaging physical exercise.
Cherix et al. (2020)	N = 15; Age: 9–18 IQ: Mild to moderate	N/A	High	To evaluate VR usability as a learning tool for young individuals with IDs	Single-group pre-post test; PS	IVR (HTC Vive Pro)	Phased learning process; contextual generalization training	VR effectively facilitated street-crossing skills.
Cheung et al. (2022)	N = 145 (80M, 65F), Age: 20–72; IQ: Moderate and severe	N	High	To evaluate VR life skills training effects on self-efficacy, memory, cognition, and behavior	Randomized controlled trials; PS	IVR (HTC Vive Pro)	Based on constructivism theory; contextual feedback mechanism	VR significantly improved cooking and cleaning skills.
Passig (2009)	N = 58; Age: 9–21; IQ: Mild to moderate	N/A	N/A	To investigate VR’s effect on time perception in adolescents with IDs	Randomized controlled trials; CS	IVR (HMD)	N/A	Participants with mild disabilities showed better temporal order perception than those with moderate disabilities.
Franze et al. (2024)	N = 36 (20M, 16F) Age: 20–75; IQ: Mild	N/A	High	Compare IVR training with non-immersive virtual environments in improving real-world skills	Randomized controlled trials; PS	IVR (Meta Quest 2)	Spaced repetition	Participants showed significant improvement in waste management skills.
Giachero et al. (2021)	N = 14 Age: 26–67; IQ: Mild	N/A	N/A	Evaluate the effectiveness of VR in teaching horticultural skills	Single-group pre-post test; PS	SIVR (NeuroVR 2.0)	Phased learning process	VR with movement observation improved procedural learning.
Hong et al. (2021)	N = 21; Age: 18–50 IQ: Mild	N/A	High	Examine the effectiveness of VR-based coffee skills training	Mix design; PS	IVR (HMD)	Structured learning	VR facilitated vocational training for those with IDs.
Jakubow et al. (2024)	N = 3, Age: 15–17 IQ: Mild	N	Moderate	Assess the effectiveness of NIVR in teaching food preparation skills to middle school students	Multi-baseline single-case study; PS	NIVR (Computer-based VR)	Error correction, maintenance training, skill generalization	Non-immersive VR was an effective intervention for teaching daily life skills.
de Oliveira Malaquias et al. (2013)	N = 15 (7M, 8F) Age: 7–22; IQ < 70	Y	High	To develop and validate serious games for students with IDs	Quasi-experimental mixed design; CS	NIVR (Computer-based VR)	Phased learning process and contextual learning	Significant post-test gains in sequencing, classification, and spatial orientation skills.
Mengue-Topio et al. (2011)	N = 18 (12M, 6F) Age: 22–29; IQ: Mild	N/A	N/A	To examine wayfinding skills in adults with IDs, focusing on path learning and shortcut performance	Randomized controlled trials; PS	NIVR (Computer-based VR)	N/A	Participants learned routes but struggled with survey knowledge.
Michalski et al. (2023)	N = 32 (20M, 12F) Age: 19–74; IQ: Mild	N	Low	To assess the ability to perform basic tasks in VR	Single-group pre-post test; PS	IVR (Oculus Quest 1)	Goal setting and a timely feedback mechanism	VR effectively enhanced real-world skills.
Michalski et al. (2022)	N = 16; Mean Age: 25 IQ: Mild	N/A	Low	To investigate VR feasibility and benefits in learning for individuals with Down syndrome	Randomized controlled trials; SS	IVR (Oculus Quest1)	N/A	VR painting experiences significantly improved learners’ overall behavior.
Mills et al. (2023)	N = 31; Age: 21–61 IQ: Moderate	N/A	Moderate	To explore the impact of IVR sensory room experiences on individuals with IDs	Single-group pre-post test; SS	IVR (HMD)	N/A	VR sensory rooms effectively reduced anxiety, depression, and sensory processing difficulties.
N’Kaoua et al. (2019)	N = 46 (36M, 10F) Age: 21–44; IQ: Mild	N/A	N/A	To evaluate the effectiveness of three-wayfinding aids for individuals with IDs	Randomized controlled trials; PS	NIVR (Computer-based VR)	Three stages of spatial cognition: surface, route, and surveying knowledge	Intervention effects were weaker than for peers without disabilities.
Panerai et al. (2018)	N = 16 Age: 15–48; IQ: 8 mild, 8 moderate	Y	N/A	To assess the feasibility and verify the effectiveness of a remote home-based rehabilitation	Single-group pre-post test; PS	IVR (Unity 3D)	Verbal reinforcement, task analysis, and total task chaining	VR training effects were effectively generalized to real-world environments.
Purser et al. (2015)	N = 108; Age: 5–11 IQ: N/P	N/A	N/A	To investigate navigation and path learning in individuals with Down syndrome and Williams syndrome	Randomized controlled trials; PS	NIVR (Computer-based VR)	Modeling and prompting	VR-based navigation training proved effective for them.
Shin et al. (2024)	N = 18 (12M, 6F) Age: 15–50; IQ: Mild	N	Low	To examine whether VR-based vocational training offers real-world transferability	Randomized controlled trials; PS	IVR (HTC Vive Pro)	Modeling; contextual learning	VR facilitated vocational training for individuals with IDs.
Simoni et al. (2023)	N = 1 (1M) Age: 18; IQ: Mild	N/A	Low	To explore gamified experiential learning using VR systems and tools	Single-group pre-post test; PS	IVR (HTC Vive Pro)	Structured learning	Participants improved grocery shopping skills in virtual and real settings.
Tam et al. (2005)	N = 16 (8M, 8F) Age: 17–23; IQ: Mild and moderate	N/A	N/A	To compare the effectiveness of NIVR and traditional methods in supermarket shopping skills	Randomized controlled trials; PS	NIVR (Computer-based VR)	Structured process design; feedback reinforcement mechanism	Significant improvement in shopping skills; non-immersive 2D VR matched traditional methods.
Trigueiro et al. (2024)	N = 15 (10M, 5F) Age: 18–35; IQ: Mild and moderate	Y	High	To investigate VR cognitive training effects on working memory, attention, and inhibition in young individuals with IDs	Single-group pre-post test; CS	IVR (Meta Quest 2)	N/A	VR-based cognitive training effectively enhanced cognitive abilities in young individuals with IDs.
X. Wang et al. (2023)	N = 49 (33M, 16F) Age: 18–35; IQ: Moderate (n = 34), Severe (n = 15)	N/A	Moderate	To evaluate adolescents’ acceptance of HMDs and immersion level in VR	Mixed design; CS	IVR (HTC Vive Pro)	N/A	Most participants reported a positive and immersive experience.
Tianwu et al. (2016)	N = 6; Age: N\P IQ: Mild	N/A	N/A	To develop a simplified VR-based travel training system	Mixed design; PS	NIVR	Phased learning process; contextual generalization training	VR effectively improved travel skills.

). Excluding one study without age data, six focused on minors, four on mixed-age groups, and fifteen on adults. This finding reflects a research emphasis on adults, with limited focus on younger populations. Additionally, nineteen studies (79%) were found to involve individuals with mild to moderate IDs, reflecting a research focus on those with less severe impairments. Given the variation in participants’ prior AT (including VR) experience and acceptance levels of virtual devices, we further considered the potential impact of participant heterogeneity on intervention outcomes.

Regarding prior experience, four studies (16%) explicitly reported that participants had previous exposure to electronic devices, while five studies (20%) stated that subjects had no such experience. Most of the studies (n = 16, 64%) did not address this aspect, and none examined how participants’ prior technology experience influenced their acceptance of VR devices or intervention outcomes. Nevertheless, all studies consistently provided pre-experimental training on device operation and experimental procedures. It should be noted that researchers tended to focus more on participants’ familiarity with task content than on prior VR experience. For instance, vocational training interventions often required participants to possess foundational vocational skills, and supermarket shopping simulations stressed the need for basic computational abilities.

Concerning device acceptance, eight studies (32%) reported high acceptance of VR devices, with no adverse reactions observed. However, four studies (16%) noted mild dizziness or discomfort during use, which improved over time. Notably, nine studies (36%) did not report on participant acceptance or discomfort, leaving this aspect underexplored. Furthermore, some studies, like Simoni et al. (2023), suggested that participants’ acceptance of VR devices had no significant impact on intervention outcomes, while others (Cheung et al., 2022) indicated that higher acceptance might enhance sustained engagement. However, due to inconsistent findings, the relationship between user acceptance and intervention results remains inconclusive and warrants further investigation. We also found that five studies employed assessment tools or questionnaires to measure acceptance, while six studies documented qualitative feedback on participant engagement through observation. Interpretation should also be cautious due to the lack of formal usability assessments in most studies. These findings call for more nuanced, systematic analysis of participant heterogeneity to support personalized VR design.

3.3. Application Areas

An analysis of adaptive behavior domains in individuals with IDs reveals that practical skills account for 68% (n = 17) of the studies, followed by social and conceptual skills, also at 16% (n = 4), as summarized in

Table 2. The distribution of VR applications in the domain of adaptive behavior.

Adaptive Behavior Domain	Category	No. of Articles	Proportions
Conceptual Skills	Sequential time perception (n = 1); mathematical logic (n = 1); cognitive training (n = 2)	4	16%
Social Skills	Leisure activities (n = 2); classroom behavior (n = 1); anxiety processing (n = 1)	4	16%
Practical Skills	Vocational skills (n = 3); travel and mobility skills (n = 7); daily living skills (n = 7)	17	68%

. Studies on practical skills mainly addressed vocational tasks, mobility, and daily living, with a strong emphasis on the latter two. This concentration suggests a prevailing focus on developing functional independence through VR-based interventions. In contrast, studies addressing social skills concentrated on social adaptation training, which included leisure activities, classroom behavior, and anxiety processing. Conceptual skills, represented by the same number of studies as social skills, primarily emphasize sequential time perception, mathematical logic, and cognitive training. Overall, practical skills emerge as the predominant application area. The relatively limited number of studies on social and conceptual domains indicates a narrower research scope and practical implementation focus. This imbalance highlights the need for future research to broaden the range of training goals across all areas of adaptive behavior.

3.4. Intervention Devices

An analysis of immersion levels for intervention devices revealed that immersive VR devices were the most commonly used (n = 14; 56%). Among these, the HTC Vive Pro was predominant (n = 9), followed by the Meta Quest 2 (n = 2), Oculus Quest 1 (n = 2), and the VIVE Pro Eye HMD (n = 1), which features integrated eye tracking. Non-immersive VR devices made up 28% of studies (n = 7) and mainly included desktop-based systems like 2D non-immersive flat-panel VR training software, Virtools 5.0, 3D Vidia Virtools, and the ViTA virtual intervention tool. Semi-immersive devices were the least utilized (n = 4; 16%), including platforms such as NeuroVR 2.0, Kinect wearable devices, and the Xtreme video capture VR system. These findings suggest a preference for immersive systems, which likely reflects their potential to support more interactive and realistic training experiences. In terms of interaction modes, unimodal interaction was the most prevalent (n = 20; 80%), including controller-based (n = 9), keyboard (n = 6), gesture-based interaction via wearable devices (n = 3), and head gaze interaction (n = 1). Multimodal interaction was used less frequently (n = 5; 20%), and included combinations such as keyboard with voice guidance (n = 1), controller with eye tracking (n = 2), and controller with head tracking (n = 2). The prevalence of unimodal systems suggests that simplicity and ease of implementation may be prioritized over more complex multimodal approaches, which may require greater technological sophistication and user training.

As detailed in

Table 3. Device types and interactive modes in each domain.

Domains	Device Types	Interactive Modes
Conceptual Skills (n = 4)	Immersive devices (n = 3), Non-immersive devices (n = 1)	Unimodal interaction (n = 4)	Controller-based interaction (n = 2), keyboard interaction (n = 1), head tracking (n = 1)
Social Skills (n = 4)	Immersive devices (n = 2), Non-immersive devices (n = 1), Semi-immersive devices (n = 1)	Unimodal interaction (n = 3)	Controller-based interaction (n = 2), Motion-based interaction (n = 2)
		Multimodal interaction (n = 1)
Practical Skills (n = 17)	Immersive devices (n = 9), Non-immersive devices (n = 5), Semi-immersive devices (n = 3)	Unimodal interaction (n = 13)	Controller-based interaction (n = 7), Keyboard interaction (n = 5), Motion-based interaction (n = 1),
		Multimodal interaction (n = 4)	Controller + head tracking (n = 2), Controller + eye tracking (n = 2)

, immersive VR devices were applied across all three adaptive behavior domains, indicating their versatility and wide applicability. Non-immersive devices were primarily used in practical skill training, relying on keyboard or mouse input, and were less commonly used for conceptual or social skill development. Semi-immersive systems appeared in both social and practical domains but were relatively infrequent overall. In terms of interaction types by domain, conceptual skill training relied exclusively on unimodal interfaces. Social skill interventions primarily used unimodal modes (n = 3), with only one study employing a multimodal setup combining motion input and multisensory feedback. Practical skill training showed the most varied use of interaction modes, including the highest occurrence of multimodal systems (n = 4), such as controllers with eye-tracking or head-tracking combinations. This suggests that more complex functional tasks, such as those in practical domains, may benefit from richer interaction modalities. In conclusion, immersive VR devices are the most frequently used, and interaction modes are largely unimodal, with practical skills showing more domain-specific variation. These trends may reflect both technological constraints and the nature of targeted behavioral outcomes.

3.5. Pedagogical Strategies and Outcomes

To better understand the intervention process and inform personalized VR design, we analyzed the pedagogical strategies employed. Behavioral principles based on ABA were the most frequently applied (n = 8, 32%), typically involving immediate reinforcement, repetitive practice, and positive feedback to promote learning. Structured learning approaches emphasizing systematic instruction were used in 20% of studies (n = 5), mainly targeting stepwise skill acquisition. Contextual learning grounded in real-life scenarios appeared in 12% of studies (n = 3), supporting skill generalization. These foundational approaches guide the instructional design of VR-based interventions. Personalized strategies such as leveled training goals (n = 7, 28%) and cognitive load management (n = 6, 24%) reflected efforts to tailor interventions to individual learner needs. However, only some studies detailed how these strategies translated into specific VR design features, such as adaptive feedback or graduated task difficulty. This variability suggests that while instructional strategies broadly informed VR interventions, the operationalization of personalized and scaffolded supports was varied. Overall, the diverse pedagogical approaches highlight a need to align instructional design more explicitly with the heterogeneity of learners with IDs to maximize intervention effectiveness and engagement.

An analysis of research designs and outcome measures across the 25 included studies revealed that 11 employed randomized controlled trials (RCTs). Most of these studies reported that participants in the VR intervention groups showed significantly greater improvements than those in control groups. Several authors also provided interpretive insights into these results. For example, Tam et al. (2005) proposed that VR improves learning by offering consistent feedback and situational simulations while emphasizing the necessity for larger samples to confirm these effects. Passig (2009) found a more substantial impact of VR on adolescents with mild developmental delays compared to those with moderate delays. Nine studies adopted a single-group pre-post design and reported significant improvements in the targeted skill domains after VR interventions. One study employed a multiple-baseline across-participants design (Jakubow et al., 2024) and found that non-immersive VR effectively supported life skills training. Four studies used mixed-method designs (de Oliveira Malaquias et al., 2013; Tianwu et al., 2016; Hong et al., 2021; X. Wang et al., 2023), combining questionnaires with observational data. These studies agreed that the immersive and interactive features of VR helped boost learner engagement. Three studies examined whether the skills acquired in virtual settings could transfer to real-life contexts and found promising evidence of generalization. In summary, most of the reviewed studies reported improvements in adaptive behavior following VR interventions, as described in their original findings.

4. Discussion

This literature review systematically summarizes the findings of 25 articles according to our research question. Based on the reviewed evidence, VR interventions appear to offer meaningful support for improving adaptive behavior in this population. In the following discussion, we will explore emerging trends and outline potential directions for future research and practice in this field.

4.1. Current Research Status and Development Trends

Over the past two decades, research on VR interventions to improve adaptive behavior in individuals with IDs has shown a consistently increasing trend, with a notable surge in publications since 2021. This growth reflects a rising global interest in integrating inclusive education, intelligent rehabilitation, and artificial intelligence technologies within special education (M. Zhang et al., 2022). Most studies come from developed countries in Europe and North America, likely due to their well-established special education systems, great investment in technology innovation, and strong interdisciplinary research foundations (Al Farsi et al., 2021). In contrast, although developing countries like China have begun exploring this field, their general research output remains limited (Fu & Ji, 2023), which highlights considerable potential for future development. At the same time, despite the increased number of publications, the scholarly influence of previous research remains relatively low. Specifically, 68% of the reviewed articles have been cited less than 20 times, suggesting a demand for high-quality and high-impact research. To promote theoretical advancement and practical application, future studies should emphasize deeper conceptual frameworks and employ more rigorous empirical designs.

4.2. Characteristics of Participants

Participants in the included studies ranged in age from 5 to 75, reflecting considerable variation in developmental stages and learning needs. While both children and adults were represented, the majority of studies centered on adult participants, potentially limiting generalizability to younger populations. As previously discussed, prior experience with AT, including VR, may moderate the effectiveness of VR interventions (Álvarez-Aguado et al., 2025). Nevertheless, this factor was largely overlooked, with 64% of studies failing to detail prior experience or incorporate it into inclusion criteria. Only the study by Yalon-Chamovitz and Weiss (2008) mentioned that participants were required to have prior VR experience, while three other studies specified familiarity with electronic devices such as computers and tablets (de Oliveira Malaquias et al., 2013; Panerai et al., 2018; Trigueiro et al., 2024). Regarding participant acceptance, Trigueiro et al. (2024) suggested that greater acceptance of VR may promote engagement and intervention efficacy, though empirical validation remains limited. Few studies have examined device acceptance as an independent variable, with most attributing the outcomes directly to the VR training itself (Simoni et al., 2023; Butti et al., 2024). Some scholars have argued that execution challenges stem more from cognitive limitations rather than challenges in operating the devices, like difficulties in understanding task instructions or locating target items (Hong et al., 2021). Thus, we could also understand why the majority of research has concentrated on individuals with mild to moderate intellectual impairments. This underscores the necessity of optimizing device design to improve usability and reduce cognitive and operational demands, thereby enhancing intervention equity. Future research should systematically examine how factors, such as age, severity level, prior experience, and device acceptance, influence intervention outcomes and optimize VR interventions for individuals requiring intensive support.

4.3. Applying VR in Different Domains of Adaptive Behavior

Our study revealed that VR interventions have mainly targeted practical skills, reflecting a preference for functional and applicable domains. Practical skills are concrete, observable, and easy to measure, making them ideal for VR-based interventions that utilize task simulation and interactive feedback (T. Wang et al., 2013). This trend has become particularly pronounced since 2020, reflecting the growing emphasis on enhancing self-care and independent living skills among individuals with IDs (Walton & Ingersoll, 2013). In contrast, although studies on social skills account for only 16% of the reviewed literature, our analysis indicates that most of them have been published since 2022. This suggests increasing research interest in complex skills like social interaction and behavior management. VR’s immersive, low-risk, and controllable environments offer valuable opportunities to practice social skills that are difficult to train in real-life settings (Nabors et al., 2020). Although still fewer in number, studies on social skills show strong potential and deserve more research attention.

Similarly, research on conceptual skill interventions remains relatively limited. Most existing studies have primarily focused on cognitive training, such as attention, memory, and rote learning. Complex conceptual skills, such as abstract thinking, language comprehension, and financial literacy, remain underexplored. This limitation arises, on the one hand, from the inherent challenges of simulating abstract cognitive processes in a virtual environment. On the other hand, it reflects current difficulties in VR content design, particularly in the areas of instructional strategy and cognitive transfer. Future studies should explore ways to translate abstract concepts into concrete experiences using gamification, multimodal interaction, and adaptive systems. These strategies could improve the effectiveness of conceptual skill training (Kim et al., 2020). In summary, VR interventions have established a solid foundation in the domain of practical skills, while the application of VR to social and conceptual skill development remains at its early stage with substantial research potential. To support adaptive behavior more fully, future research should refine VR’s scope, technological tools, and instructional design.

4.4. Application of Virtual Devices

Our analysis reveals that immersive HMDs, such as HTC Vive Pro, VIVE Pro Eye, and Oculus Quest, are widely used in VR-based interventions for individuals with IDs. This reflects a growing trend toward high-immersion technology in adaptive behavior training. The primary advantage of IVR lies in transforming learning experiences from flat, passive, and unidirectional formats to three-dimensional, active, and interactive environments. This experiential environment is effective for individuals with IDs who benefit from context-rich learning experiences. Since the introduction of advanced devices like the Oculus Rift and HTC Vive in 2016, immersive VR systems have increasingly integrated motion tracking, spatial positioning, and sensory feedback to simulate real-world interactions (Waterfield et al., 2024). In contrast, non-immersive and semi-immersive systems offer lower interactivity, limiting their utility in contexts requiring behavioral engagement, motor coordination, or embodied learning. However, when choosing VR devices, we need to consider practical and pedagogical factors. For instance, extended use of immersive headsets could lead to sensory overload or motion sickness (Didehbani et al., 2016). Additionally, device cost, comfort, accessibility, and adaptability to the needs of each person should also be carefully evaluated (Radianti et al., 2020). To ensure meaningful learning outcomes, Parong and Mayer (2021) suggested aligning device features with learning goals, user capabilities, and contextual constraints. In other words, we should further explore the personalization of VR devices, not only in terms of setting tiered and leveled goals and adjusting difficulty but also in terms of providing multidimensional adaptations, including physiological and cognitive adjustments, as well as individual differences.

In terms of interaction modes, we found that most VR interventions employed unimodal systems, primarily handheld controllers, which are precise and user-friendly (Yi et al., 2024). In contrast, fewer studies applied multimodal systems, which could supplement controller input with head or eye tracking to enhance immersion and engagement. For example, the HTC Vive Pro helps individuals with IDs navigate virtual social environments by spatial positioning and controller manipulation, thereby facilitating the development of conceptual skills (Passig, 2009). Similarly, the VIVE Pro Eye incorporates eye-tracking technology to monitor attention allocation, which could be instrumental in assessing social skills (Shin et al., 2024). These findings reveal that the majority of VR interventions are based on single-sensory modalities, with limited exploration of multiple sensory interactions. This confirms previous findings (Tao et al., 2022), suggesting that limited use of multimodal systems stems from high demands on hardware, system design, and users’ cognitive capacity. For these individuals, excessive complexity can raise cognitive load, hindering usability and outcomes. Therefore, future work should explore the relationship between disability severity, cognitive characteristics, and interaction complexity to optimize adaptive VR designs.

4.5. Pedagogical Strategies in VR Interventions

We conducted an in-depth analysis of the pedagogical strategies employed within these interventions. The review revealed that strategies based on ABA and structured learning were commonly utilized in VR interventions. For example, Tam et al. (2005) applied structured teaching in a VR-based barista training program. In this intervention, complex tasks such as controller operation and espresso extraction were decomposed into simple, sequential sub-procedures. This process-oriented design exemplifies how instructional strategies can inform the hierarchical structuring of VR content. Behavioral support based on ABA principles, such as prompting, modeling, repetition, and reinforcement (Cheung et al., 2022; Simoni et al., 2023), was widely adopted and often operationalized through real-time feedback systems and reward-based mechanisms embedded in the VR environment. However, the extent to which pedagogical strategies were systematically integrated into VR system features varied considerably. Feedback mechanisms were implemented in many interventions, yet the depth of personalization, such as adjusting task difficulty, pacing, or prompt levels based on learner performance, was inconsistently addressed. Scaffolding strategies, including graduated assistance or adaptive guidance, were rarely described in detail. This indicates that dynamic instructional support received limited attention in most VR system designs (X. Wang et al., 2023). In addition, seven of the included studies did not clearly describe the pedagogical strategies they used. As a result, it was difficult to determine how these interventions addressed the diverse needs of learners with IDs. These findings highlight the need for clearer alignment between pedagogical goals and VR design. Future research should focus on incorporating adaptive mechanisms that address individual differences in cognition, behavior, and communication. Features like personalized feedback, scalable scaffolding, and goal-based content can help VR systems better support diverse learners with IDs.

5. Conclusions

Our research is the first systematic study on using VR to improve adaptive behavior in individuals with IDs. It reviews studies from 2005 to 2024 to explore how VR has been applied in the training of adaptive behavior. Based on the five research questions, the key findings are as follows: (1) There has been a notable increase in VR interventions for adaptive behaviors, particularly after 2021, highlighting a growing interest in this area. (2) Research has primarily focused on adults with mild to moderate IDs. Some studies also considered factors like prior VR experience and device acceptance. (3) VR has been more frequently applied in interventions for practical skills training, while applications for social and conceptual skills remain limited. (4) High-immersion HMDs, such as the HTC Vive Pro, are the most commonly used devices. The predominant interaction mode is unimodal, with controller-based input being the most widely adopted. (5) Pedagogical strategies like ABA, structured teaching, and contextual learning are commonly used in VR interventions. However, personalized VR design remains limited, suggesting a key area for future exploration. Overall, our study synthesizes the existing literature, providing valuable insights into the current state of VR applications in adaptive behavior training. While the evidence suggests that VR may enhance adaptive behaviors, further validation and personalized approaches are necessary to fully realize its potential. This review offers important directions for future research and technological development in this field.

The study still has limitations. First, the small sample sizes in many of the included studies limit the generalizability of the findings. Second, some studies lack detailed information on intervention duration, device parameters, and other critical factors, making it difficult to assess the effectiveness of different VR intervention models. Third, the inclusion criteria were based on specific methodological designs, which may have biased the findings with more structured, better-resourced, or more likely to report positive outcomes. Additionally, the search was limited to English-language studies, which may have excluded relevant research published in other languages and could contribute to potential language bias.

To further advance research in this field, several key directions should be considered. First, as most current studies focus on adults with mild to moderate IDs, future research should broaden the scope to include individuals with severe and multiple disabilities and explore the integration of VR with medical rehabilitation. Second, to mitigate potential bias and increase generalizability, it is important to incorporate a broader range of methodologies and data sources, including non-English studies. Third, VR applications for social and conceptual skill training remain limited despite the critical role of these domains in adaptive functioning. Investigating how VR can be effectively implemented in special education classrooms and social learning contexts is essential. Lastly, VR interventions should be more closely aligned with pedagogical principles and personalized learning needs. This calls for closer collaboration across educational technology, special education, and cognitive psychology to develop more adaptive and inclusive environments.

Despite the promising progress, more robust and well-documented intervention studies remain urgently needed. This need is especially pronounced in under-represented contexts such as non-Western and low-resource settings, where technological and implementation challenges differ significantly. Addressing these gaps will strengthen the evidence base and enhance the inclusivity and applicability of VR interventions for adaptive behavior training.