Linking Low-Level Visual Processing Strategies to Higher-Level Processing Abnormalities in Autism Spectrum Conditions

Background: There is a large literature on visual processing abnormalities in individuals with Autism Spectrum Conditions (ASC). Previous research addresses both general visual processing, and category-specific visual mechanisms. We have developed a comprehensive experimental design to investigate low-level visual processing and relate this to higher levels of cognition. The two areas of investigation in which we are concerned are spatial frequency analysis and face-specific visual processing mechanisms. Previous studies have suggested there is a high spatial frequency (HSF) bias in ASC in face identification tasks, contrary to a low spatial frequency (LSF) bias that is characteristic of control participants (Deruelle et al., 2004). We aim to investigate whether an atypical spatial processing style occurs across all image categories, or is related to the social nature of target stimulus.   

Objectives: To address how task-demands influence spatial frequency biases and whether individuals with ASC differ in their strategy under certain stimulus conditions.

Methods: Participants were presented with spatial frequency modified images of cars and faces (low and high-pass filtered images and hybrids of these) in an identification task. Experimental manipulations were made involving presentation time interval, nature of competition and orientation of image. 

Results: At the time of writing, data-collection from ASC individuals is not complete. However, experiments have been run on control populations. Results indicated a time-dependence to spatial frequency strategy selection for car target images with LSFs biased under short conditions, but HSFs dominant in longer presentation duration. On the contrary, there was a LSF bias across all conditions with face targets. This LSF bias was reduced when the face stimuli were presented inverted. 

Conclusions: A different pattern of spatial frequency preference is exhibited for face and car images in typically developing individuals. The perceptual dominance seems to be influenced not only by the physical properties of the target stimulus, but also by the higher-level interpretation of the image; this is suggested by a reduced LSF bias for inverted face stimuli. If atypical spatial frequency bias is a general characteristic of ASC one would expect this to be evident in result for both car and face images. However, if the abnormality is specific to face stimuli, one might expect differences confined to processing this image category.