Atypical Development of Visual Processing During Adolescence in Autism

Background:  Results from the Cambridge Face Memory Task (CFMT; Duchaine & Nakayama, 2006) showed substantial development from adolescence to adulthood typically, but not in those with autism (O'Hearn et al., 2010).  The current studies test skills related to the CFMT, including those may develop late typically (e.g., holistic processing) or be impacted to autism (e.g., encoding eyes).

Objectives:  The goal of these studies was to further characterize the pattern of development on the CFMT, both behaviorally and neurophysiologically. We tested whether the findings were specific to faces, to face ‘parts’ (eyes or mouth), or to holistic processing, all of which have been suggested to differ in autism. Eye movement and neuroimaging measures were collected to provide insight into the mechanisms underlying the group differences on the CFMT.

Methods:  Behavioral studies include the immediate memory task and the part/whole task from the Let's Face It battery (LFI; Wolf et al.,2008). To date, 8 children, 14 adolescents and 18 adults with autism (IQ’s >80) have been tested on these behavioral tasks, as well as TD individuals matched on age, IQ and gender.  In the immediate memory task, participants see a face or car for 1 s, and then pick it out of three objects from a ¾ view.  In the part/whole task, participants view a whole face for 4 s, then choose the same face (or eyes or mouth) from two test items. Test items could be a whole face or part of a face, to examine holistic processing, and the change could be eyes or mouth, to examine performance with these important features. In addition, to date, 20 individuals with autism and 14 TD individuals completed the CFMT modified for neuroimaging in a fast event-related design. Cars were included in several runs, to examine the specificity of the results in terms of brain activation.

Results:  The developmental pattern evident previously on the CFMT was replicated with both faces and cars on the immediate memory task, indicating the results were not specific to faces or the CFMT. It was also evident with both parts and whole conditions of the part/whole tasks, indicating that it did not reflect holistic processing. Indeed, all ages in both groups displayed holistic processing (i.e., better performance with the whole than the parts). Eye recognition in adults with autism was the exception, and also the only condition to display development in autism from adolescence to adulthood, suggesting compensatory strategies in these adults. Preliminary neuroimaging data from 20 individuals with autism (10 adolescents) and 14 controls (7 adolescents) suggests that increases in activation may occur with age in frontal regions typically but not in autism, mimicking the behavioral pattern.

Conclusions:  The pattern evident on the CFMT is not specific to faces, features, or holistic processing. This pattern of development may generalize across a wide range of visual functions, making adolescence an crucial time for further study. Neuroimaging examination of this pattern of visual development will provide much needed insight into the developmental processes impacted by autism.