Strategic Influences On Face Processing in Autism Spectrum Conditions, Dyslexia and Typical Development

Background: The general issues of category-specific visual processing abnormalities and atypical early visual mechanisms are fiercely debated in the autism literature. Specifically, spatial frequency preferences and the face inversion effect are relevant to the current investigation. There is a wealth of literature to suggest that atypical processing strategies in autism spectrum conditions (ASC), including abnormal spatial frequency biases, may contribute to such findings. Face processing has been the focus of much autism research and there is mounting evidence of strategic differences in ASC individuals compared to typically developing controls (Teunisse et al., 2003). There are also thought to be visual disturbances associated with dyslexia and comparisons have been drawn between visual processing in these two sets of conditions. However, the same emphasis on strategic atypicality has not been applied to dyslexia. In this study, low-level perceptual strategies are assessed in ASC and dyslexia.

Objectives: To study the low-level visual strategies selected by participants with ASC, in comparison to those diagnosed with dyslexia and typically developing individuals. The relationship between the prevalence of autistic traits in general, and the processing strategy adopted is also explored.

Methods: A computer-based identity matching experiment was run on three groups of high-functioning adult participants: those with ASC, with dyslexia and typically developing individuals. There were two experimental sessions, testing the identification of three faces and three cars separately. Stimuli were spatial frequency modified, to investigate the low-level visual preferences exhibited by the groups under different experimental conditions. The images used were full spectrum, low-pass filtered, high-pass filtered and hybrids of low- and high spatial frequency components of different images. Stimuli could be presented upright or inverted. In addition, participants completed the Autism Spectrum Quotient (AQ) and the Advanced Raven’s Matrices (APM).

Results: At the time of writing, data-collection is incomplete. However preliminary observation of control group data reveals an intriguing relationship between upright face identification performance in certain conditions and participants’ score on the AQ. At this early stage some compelling contrasts can be observed between patterns of performance in the ASC and dyslexia groups.

Conclusions: In this experiment, face identification performance is related to the degree of autistic traits exhibited by a typically developing population. The relationship observed here is specific to rapid presentation conditions and is also sensitive to the image category, which suggests that it is the execution of particular processing strategies which is related to AQ. This relationship may extend into the ASC and dyslexia groups, or a distinct pattern of results may be observed in these participants due to additional processing atypicalities that overwhelm any influence of autistic traits as measured by the AQ.