Examining Attentional Bias for Facial Features Across Development and in Autism

Background:   Critical behavioral and biological markers of a disorder that are present early in development may be transient or may be later masked by disease progression. Thus, one challenge for researchers investigating neurodevelopmental disorders is to design tasks that can be used across broad age ranges and developmental levels so as to identify and track these markers throughout development. 

Objectives: We designed and evaluated performance in a task that provides an index of attentional bias in individuals from infancy through adulthood. Specifically, we designed a paired visual comparison (PVC) version of the “bubbles” face processing task (Gosselin & Schyns, 2001; Humphreys et al., 2006).  

Methods: The final sample included 24 school-aged children (n=12 typically developing (TYP); n=12 with autism according to CBE, ADOS, and ADI-R) and 13 typically developing infants between 6 and 14 months-of-age. Stimuli were modified “bubbles” stimuli based on Gosselin and Schyns (2001). Bubbles presented on either side of fixation revealed one of three facial areas: Eyes, Mouth, or Other Facial Parts (OFPs) such as forehead, cheek, or nose. These Areas Of Interest (AOIs) were viewed in three possible pairings: Eyes vs. Mouth, Eyes vs. OFPs, and Mouth vs. OFPs. Stimuli were presented in Tobii Studio and gaze was recorded with a Tobii X120. We computed three measures of gaze for each AOI: 1) Number of trials the AOI is fixated first, as a proportion of valid trials, 2) Time to fixate the AOI, and 3) Time spent on AOI, as a proportion of time spent on screen.  

Results:  There were few overall differences between TYP children and children with autism: TYP children spent a larger proportion of time on the AOIs (p=0.015) and were faster to reach the AOIs than children with autism (p=0.012). No Group x Part interactions were observed, indicating that both groups showed similar attentional biases. Overall, the children fixed the Eyes first more often than the Mouth and OFPs (ps<0.001). They also arrived at the Eyes faster than the Mouth (p=0.012) and OFPs (p=0.001). They spent an equal proportion of time on the Eyes and Mouth, but a greater proportion of time on these areas than the OFPs (ps<0.001).

Like the school-age children, the infants fixated the Eyes first more often than the Mouth and the OFPs (ps<0.001). They spent a greater proportion of their looking time on the Eyes than the Mouth and OFPs (ps<0.001), and Mouth compared to OFPs (p=0.006). There was no significant difference between time to fixate Eyes, Mouth, or OFPs.

Conclusions:   School-age children with and without autism, and infants, show similar viewing patterns on this paired visual comparison version of the “bubbles” task. All the groups show a bias toward eyes in terms of what they fixate first, how quickly they arrive there, and proportion of viewing time, but a bias for mouths over other face parts was less clear. This task appears to be useful for measuring attentional bias to facial features across a wide range of ages and developmental abilities.