<p>Joint attention is essential for establishing effective social communication. However, children with Autism Spectrum Disorder (ASD) often exhibit deficits in joint attention, which hinder their social interactions with peers and adults. Traditional assessment methods predominantly rely on expert observation, lacking objective measures of children’s joint attention responses. To overcome these limitations, digital learning environments utilizing virtual reality technology and eye-tracking offer new opportunities for effectively assessing joint attention in children with ASD. This study explores the joint attention performance of children with ASD under social and non-social stimuli by tracking their eye movements in virtual environments. The study results indicate that: (1) Under social stimulus conditions alone, the joint attention abilities of children with ASD are significantly influenced by the environment, while typically developing (TD) children’s abilities remain unaffected; (2) the additional of non-social stimuli significantly enhanced the joint attention scores for both groups. The TD group responded more effectively to non-social stimuli in the VR environment; (3) using TD children’s joint attention scores under the S_social condition as a baseline, the additional of non-social stimuli helped ASD group approach the joint attention levels of TD. The conclusion suggests that VR-based multi-stimulus environments provide meaningful joint attention opportunities for children. By comparing with TD group, the study offers new insights and references for future interventions aimed at improving joint attention in children with ASD.</p>