Virtual reality and autism

Process preparations enabling future research and intervention practices

Abstract

Background: The overall aim of this study is to examine key parts that need to be considered when planning and conducting research using virtual reality (VR) for autistic individuals. VR is a technology that simulates realistic environments and experiences that also can be tailored to meet specific needs. VR has the possibility to mimic real-life situations integrated in a controlled, predictable, safe, and supportive space. Moreover, VR can enhance engagement and motivation for autistic individuals, and offers customizable opportunities to develop and practice essential skills. The integration of virtual reality (VR) represents an exciting frontier in autism research and intervention practices. However, the complexity represented by such research and that the evidence of the effectiveness of VR-interventions is still scarce. It is therefore a need to carefully examine auxiliary assumptions that can affect the outcomes. In this thesis, I present four studies aimed at enabling researchers and conducting process preparations for future scientific endeavors.

Objectives: The objective of this thesis is to investigate elements that are important to enable future research on autism and VR. This objective will be reached by (1) providing knowledge about the overall acceptability towards VR by autistic individuals, (2) addressing research gaps on autism, VR, and social skills, (3) evaluate whether available software create valid emotion expressions in avatars, and (4) investigating factors important for implementation of VR in schools.

Methods: The four papers presented in this doctoral thesis utilized different methods. The first study is a review of the available peer-reviewed literature on autism and VR. We qualitatively analyzed the reporting of acceptability from autistic participants. We categorized the findings as positive, negative, or inconclusive. Inconclusive meaning variations within the sample in a study. The second study is a systematic scoping review providing an extensive overview of the literature reporting on social skills interventions, involving autistic participants, in VR. We analyzed the studies and identified several research gaps. The third study is a web-based pilot-survey where we quantitatively evaluate emotions expressed by avatars. We created avatars from real person photos using an available software which automatically expressed various basic and complex emotions. Using structural equation modelling and Bayesian confirmatory factor analysis, we analyzed responses from the participants in terms of correctly perceived, difficulty level, and discrimination. The fourth study is a qualitative investigation of factors that are perceived important when implementing VR in schools. We used semi-structured interviews with eight individuals and a group of five individuals consisting of teachers, special educators, leaders, and student teachers. We used thematic analysis and identified factors that our informants deemed as important for a successful implementation.

Results: Study 1 revealed an increase of the number of studies on autism and VR, but less than half of the included studies reported acceptability. Of the studies providing acceptability information, 89% reports positive evaluations while 1% is negative. More than 70% of the studies using head-mounted displays (HMD) provide acceptability information. Of these, approximately 4 out of 5 studies showed positive evaluations and none were negative. In Study 2, a total of 49 studies were included and analyzed. We identified seven prominent research gaps. The first gap shows that less than half of the studies reported acceptability data. Further, there is a lack of diversity of methodology and demographics. Only 1:10 participants are female, there are few studies with participants having co-occurring diagnoses, few participants aged over 30, and few studies using HMDs. Additionally, there is lack of diversity of methods, meaning; few RCTs or case-control studies, and few multiple singlecase designs. Furthermore, a theoretical grounding in the intervention studies is missing. Study 3 revealed that few emotions were correctly perceived, however, the difficulty level were around average on a standard threshold. The discrimination of the images was .7, indicated by the factor loadings, meaning that it shares 49% of the variance with the latent factor. In Study 4, we identified five themes important for implementation: (1) sufficient knowledge and training, (2) an internal and external network, (3) sufficient resources, (4) positive attitudes, and (5) technological features. All participants expressed enthusiasm towards applying VR in schools.

Conclusions: The main findings of this thesis are that, first, VR is a preferable tool for many autistic individuals thus making it a technology worth developing. Second, there are research gaps that needs to be filled to justify future claims of effectiveness. Researchers should be aware of the possible fallacies connected to auxiliary assumptions as demonstrated in Study 3. Furthermore, to ensure proper implementation of VR in schools, and make probable treatment integrity from the start, valuable information about crucial factors is now available. The process preparations rendered visible by this thesis provide researchers with important knowledge for future endeavors and is a step in enabling future research on autism and VR.