A Virtual Reality Study of Complex Social-Attention In School-Aged Children with Autism

Background:

Social attention theory has contributed to clinical advances for preschool children with autism, but has had less impact on school-aged children. Social-attention paradigms with preschool children measure a child’s capacity to coordinate attention with one social partner to an external object or event. In school-aged children, however, social-attention often involves coordinating attention with multiple social partners, regarding referents to internal thoughts and representations, as well as external objects/events.

Objectives:

To address this measurement challenge we have begun to use virtual reality applications to study social-attention in school-aged children with autism.   

Methods:

Our first study examined social-attention in groups of younger 8- to 11-year-old and older 12-to 16-year-old children with higher functioning autism and matched controls (total N = 40). Participants wore a head-mounted virtual displays system and were presented with a 3D virtual classroom populated by 9 avatar “peers”. They verbally respond to questions about themselves, while trying to looking at all the avatars. To see all of these, the participants needed to turn 80 degrees left and right from midline. They also needed to look behind the front avatars to fixate on the two avatars at the rear of the room. Social attention was defined as the total number of looks to the 9 avatar peers’ head regions. In the No Cue Condition participants were instructed simply to look at, and talk to all the avatars. In a Cued Condition avatars faded if they did not receive attention, but became solid again if fixated.

Results:

Consistent with other recent research (e.g. O’Hearn et al. 2010) results revealed the later onset of a developmental disturbance in autism. There was evidence of a Diagnostic Group X Age Group interaction such that there was no difference between Younger HFA and Control groups on social attention defined as the total frequency of looks to social avatars (106.8 vs. 98.3 looks respectively), but there was a significant difference on social attention between the Older HFA and Control groups, F (1,14)  = 6.28, eta² = .31 (105.4 vs. 144.5 looks respectively). The older controls also displayed reliably more looks to avatars than younger controls (p < .025), but this cross-sectional developmental effect was not evident in the autism sample. Fortunately, the social attention of all children was malleable, including those with autism, and improved with fade cues (p < .009 HFA, p < .03 Controls). Thus, cued practice with this task could help to offset adolescent delays in this type of social-attention ability in children with autism.

Conclusions:  

Consistent with the hypothesized pivotal nature of social attention for learning (Mundy et al. 2009), poorer social-attention in was strongly associated with parent reports of learning problems in school in both groups (r = -.64 to - .65, ps < .002). Additional analyses indicated that variance in ADHD symptoms and IQ significantly moderated social attention in the HFA sample. The implications of these data for social attention theory, the developmental course of expression of autism, and the utility of virtual reality research paradigms will be discussed.