Abstract
Atypicalities in the development of regions within the core and extended face-processing network have been implicated in the development of social symptoms for individuals with autism (Schultz, 2005; Scherf et al., 2014). As a result, the functional organization among these regions may also be impacted. Uddin, Supekar, & Menon (2013) proposed a developmental model, suggesting that adolescence may be a time of functional under-connectivity in neural networks of individuals with autism. To investigate whether the face-processing network exhibits such functional under-connectivity in autism, the current study examined functional connectivity within the face processing networks of 14 adolescents with high functioning autism (HFA) and 14 typically developing (TD) adolescents (13 to 18 years). The fMRI task consisted of a 1-back memory task while viewing multiple visual categories including: human faces, animal faces, and common objects. Regions in the face-processing network were defined at the group level separately for TDs and HFAs, and then fit to each participant’s individual activation. For connectivity, the best-fit model for the 12 regions was assessed for each group separately using unified structural equation modeling. We computed graph theory metrics based on the connection weights for each individual participant. The HFA adolescents had significantly higher clustering coefficients and global efficiency (p < .05), denoting more direct connections between regions. Following this, there was also a trend for the HFA group to have a greater number of edges or connections between regions. These results suggest that task-related functional connectivity between individual regions in the face-processing network of HFA adolescents is largely over-connected compared to TD adolescents. These results converge with our additional findings that TD adults with weaker face recognition abilities also have over-connected networks compared to those with stronger abilities. Together, these findings suggest that over-connected, redundant networks may interfere with proficient face-recognition behavior.
Meeting abstract presented at VSS 2015