Preliminary Evidence Suggests Specific Impairments in Explicitly-Evoked Social Inferences in Adults and Children with Autism

Thursday, May 11, 2017: 12:00 PM-1:40 PM
Golden Gate Ballroom (Marriott Marquis Hotel)
L. A. Harrison1,2,3, R. P. Spunt3, E. Kilroy1,2, A. Concha2, E. J. Goo2, C. Butera2, R. Adolphs3 and L. Aziz-Zadeh1,2, (1)Brain and Creativity Institute, Dornsife College of Letters, Arts and Sciences, University of Southern California, Los Angeles, CA, (2)USC Mrs. T.H. Chan Division of Occupational Science and Occupational Therapy, University of Southern California, Los Angeles, CA, (3)Division of Humanities and Social Sciences, California Institute of Technology, Pasadena, CA
Background: Difficulties inferring others mental states may underlie social interaction deficits in Autism Spectrum Disorders (ASD). Behavioral studies (Chevallier et al., 2012) suggest that social inference deficits in autism are selective for implicit tasks. When cues are made explicit, people with autism may in fact be able to draw social inferences, but may fail to do so in everyday situations due to other attentional or perceptual shortcomings. Thus therapy may better target these different social cognitive processes.
Objectives:  This study had two objectives. The first was to test whether social inference is intact in autism in an explicitly-cued task. If so, we then sought to determine whether this generalizes across two classes of social stimuli: hands and faces. The second objective was to corroborate behavioral findings with preliminary neuroimaging findings to explore whether behavioral similarities or differences between ASD individuals and matched controls are reflected in neural processing.

Methods: Behavioral responses were collected in 20 children (10 ASD) and 41 adults (22 ASD). Neuroimaging data were collected in 16 children (6 ASD) and 23 adults (2 ASD). All ASD subjects’ diagnoses were confirmed with the ADOS. The Level of Inference (LOI) task (Spunt & Adolphs, 2014) tested explicitly-cued social inference. The LOI task asks subjects to make yes/no responses to how (low-LOI) and why (high-LOI) questions about pictures of people performing actions. Half of the stimuli featured faces; half hands. Responses were evaluated with respect to normative responses. All imaging data were collected in a 3T scanner. Adults performed the LOI task, while children completed a mentalizing task in which they “thought why” an actor performed an observed action in one of three conditions (emotional and nonemotional expressions, and hand actions).

Results: In the behavioral task, for Why-Hand trials, independent-samples t-tests showed that the percent correct for controls was significantly higher than for ASD subjects for both children (p=0.019) and adults (p=0.002). No other group effects were observed. Child neuroimaging preliminary results showed largely similar patterns across ASD subjects and controls in whole brain analyses. ROI analyses (independent samples t-tests) showed reduced left IFG activity (p=0.039) in the ASD group for non-emotional faces. ASD subjects tended to have less activation (p=0.078) in the right TPJ; in controls, LOI Why-Hand accuracy correlated with rTPJ responses to emotional stimuli (r=0.798, p=0.005) . Adult behavioral performance and brain activation maps were comparable to controls.

Conclusions: With the exception of hand stimuli, preliminary data indicate that both children and adults with autism can accurately perform tasks of explicitly-cued social inference. Small neural activation differences were observed in children, and will be further investigated with larger samples. Future work will explore which compensatory mechanisms might preserve explicit inferences about faces: stimuli with ambiguous facial expressions will test against the hypothesis that valence matching preserves Why-Face accuracy; and LOI and NEPSY-Theory of Mind accuracy will be compared. Our results indicate that to better understand everyday functioning in autism, the field should explore under what conditions implicit and explicit social inferences are impaired.