Visual Processing Ability Predicts Facial Affect Comprehension in Children with Autism Spectrum Disorders

Background: The research on facial affect comprehension in individuals with Autism Spectrum Disorders (ASD) varies in both methodology and outcome.  Different stimulus, participant, and design characteristics across studies make it difficult to draw definitive conclusions, but the current state of the literature suggests that the sources of these deficits are diverse and complex.  The heterogeneity of symptom presentation within the autism spectrum population adds to the convolution.  Explanations for the inconsistent presence of deficient facial affect processing in ASD have been offered, including age of participants, visual sensory/perceptual differences, and social relevance of face stimuli.  Studies that have incorporated nonsocial visual control tasks in their examination of face and facial emotion perception have inconsistently found perceptual impairments in groups with poor affect recognition.  The majority of these studies have examined group averages, which can mask individual variation in both visual perception and affect recognition.  Few studies, however, have examined the relationship of these perceptual deficits to facial affect processing directly. 

Objectives: This study examined the relative contributions of general visual processing ability, visual encoding of faces, and participant age to the prediction of facial affect comprehension in children with ASD.

Methods: General visual processing ability was assessed by the matching subtest of the Wide Range Assessment of Visual Motor Ability (WRAVMA).  Ability to visually encode faces was estimated by the immediate recognition score of the memory for faces subtest of the Children’s Memory Scale (CMS).  Facial affect comprehension was measured by the child and adult faces subtests of the Diagnostic Assessment of Nonverbal Accuracy (DANVA-II).  Participants were 64 male and 7 female children ages 5-18 years (m = 10.68; SD = 3.9) undergoing neuropsychological assessment in a pediatric neuropsychology clinic.  Analyses were planned to examine group differences relative to the standardization sample as well as symptom-level relationships within this ASD group.  One-sample t-tests against the normative means were performed on the WRAVMA, CMS, and DANVA-II measures.  Regression analyses were performed to examine the relative contribution of participant age, facial encoding ability, and general visual processing performance to the prediction of facial affect comprehension.

Results: While scores on the CMS (t(70) = -8.24, p<.001) and DANVA-II tasks (adult faces t(70) = -3.09, p = .003; child faces t(70) = -5.56, p <.001) were significantly below the mean of the standardization samples, only the mean of the CMS task approached clinical significance (absolute Z-score > 1.0).  The linear combination of predictors accounted for a significant amount of variability in affect comprehension for both adult (R² = .26, R²_adj = .22, F(2,68) = 7.71, p<.001) and child faces(R² = .17, R²_adj = .13, F(2,68) = 4.47, p=.006).  Only the partial correlations between the general visual processing task and the affect comprehension tasks were significant. 

Conclusions: While ASD group scores were not impaired in visual processing ability, regression analyses revealed that visual processing accounted for a significant proportion of the variance in facial affect comprehension.  Results highlight the need for symptom-level analyses in ASD research with relevant comparison groups.