Beyond the Avatar: Understanding Autistic Preferences in Immersive Learning Environments

Gunjan Kumari, Wolfgang Broll

Abstract: Immersive Learning Environments (ILEs) have shown promise in supporting learning, and in sensory, and emotional regulation in autistic individuals. However, there is a limited understanding of how different avatar characteristics influence user preferences. This study investigated the avatar preferences of autistic individuals across appearance (realistic vs. cartoon), familiarity (self, familiar, stranger), and voice type (male, female, robotic). Additionally, it examines whether preferences for autistic individuals differ between autistic individuals and autism stakeholders. A quantitative approach was employed to examine the preferences of autistic individuals for the above-mentioned avatar characteristics, in which participants (\(N = 23\)) completed an online survey, rating their preferences on a 7-point Likert scale. Significant overall preferences emerged for appearance (\(\chi ^2 = 32.89\), \(p = 0.003\)) and voice (\(\chi ^2 = 19.54\), \(p < 0.001\)), but not for familiarity (\(\chi ^2 = 5.48\), \(p = 0.065\)). Substantial between-group differences were found for realistic avatars (t-test, \(p = 0.027\), \(d = -1.08\)) and familiar avatars (\(p = 0.030\), \(d = -1.06\)). Autistic individuals preferred cartoon avatars and stranger representations, whereas stakeholders favored realistic and familiar avatars. Both groups consistently preferred human over robotic voices. The findings challenge common assumptions about autism and avatar design, revealing misalignment between autistic individuals and stakeholders. This highlights the need for user-centered approaches. The wide range of individual preferences calls for customizable, rather than one-size-fits-all, designs. The results support using universal design frameworks that allow flexible avatar personalization, promoting inclusive technology for autism education.

External IDs:doi:10.1007/978-3-032-12781-5_19