Differential Engagement of Prefrontal Cortex Underlies Local Bias in Children with Autism

Background: Visuospatial information processing in Autism Spectrum Disorders (ASD) is often described in terms of an inability to “see the forest for the trees.” The Weak Central Coherence (WCC) account of ASD (Frith, 1989) posits that a detail-oriented information processing bias exists in ASD, which can manifest as strength or as weakness depending on the task at hand. For example, in social cognition tasks where strong coherence is the norm, people with autism may falter, as opposed to visual search tasks where they perform well. Differences in brain responses in frontal versus posterior regions may underlie this cognitive and behavioral profile.

Objectives: To examine the neural correlates of WCC in children with ASD in the context of visual and social processing. 

Methods: ASD (N=17) and Typically Developing (TD) (N=17) children and adolescents completed an emotion and shape judgement task in a Siemens 3.0 Tesla fMRI scanner. The task consisted of stick-figure human characters made up of geometrical shapes displaying different emotions. In the shape/local condition, participants indicated whether or not a given geometric shape (e.g., a rectangle) was present in the figure. In the emotion/global condition, participants recognized the emotion conveyed by the stick figure (e.g., sad). Whole-brain within-and-between-group activation and seed-to-voxel functional connectivity analyses were conducted in SPM12 and the CONN toolbox.  

Results: 1) Significantly increased activity in ASD children, relative to TD, in posterior regions, including the middle occipital and middle temporal areas for both emotion and shape processing; 2) ASD children showed decreased activity (ASD<TD) in right superior and middle frontal regions (RSFG/MFG) while recognizing local shapes (p< 0.01; cluster =174 voxels); 3) Significantly weaker connectivity for emotion recognition in ASD participants (ASD<TD) between left extra-striate body area (EBA) to frontal pole (FP) and MFG (p< 0.05, cluster-level FDR corrected); and 4) The ASD group was significantly faster in the local/shape processing condition (ASD mean= 2871 milliseconds, TD mean= 3381 milliseconds; t(32)= 2.41; p< 0.05), while the TD group was significantly more accurate in the emotion condition (ASD mean= 88.2%, TD mean= 95.6%; t(32)= 2.63; p< 0.05).    

Conclusions: Longer reaction times and greater recruitment of frontal regions in TD children during shape identification may suggest the need to override a global-processing bias that is not seen in ASD. Poorer accuracy of ASD children in recognizing emotions indicates that a default local bias may be a weakness in this context, and the decreased EBA-Frontal connectivity in ASD may underlie this weakness. While this local bias provided the ASD group with an advantage in faster processing of shapes, the TD group still performed the task with equal accuracy. Overall, our findings support weak central coherence in autism manifesting as a deficit and strength in social and visual tasks respectively, along with evidence for increased occipital recruitment and weaker connectivity. Clinical implications include targeting global visual processing in improving emotion recognition abilities.