Autism Europe Congress 2007

I had a good time at the Autism Europe Congress 2007. :> There were many good speakers there from all aspects of Autism research and clinical practice. :p There was plenty fodder for the mind (if none for the body).


There was an Art exhibition displaying works from people with Autism. I thought this painting really matched one of the poster presentations about face processing and visual perception.

The study by Deruelle and colleagues was looking at high spatial frequency and low spatial frequency information biases in emotion and gender aspects of face processing in children with Autism Spectrum Disorders and typically developing children. High spatial frequency information in a visual image provides the details of the image; Low spatial frequency information provides the global outline (i.e. relationships between different parts of the picture) of the image. The picture above can be what a face would look like if only low spatial frequency information were given.

Hybrid faces were used. Each had a low spatial frequency filtered faces super-imposed on a high spatial frequency filtered face. The high and low spatial frequency face were either male or female, or of 2 different emotions. Each image was presented for 100ms and then participants were asked to categorize the faces they saw as male or female, one emotion or another.

Typically developing participants made more low spatial frequency choices for the gender task, more high spatial frequency choices for the emotion task. Participants with ASD showed more low spatial frequency choices in both types of task. The typically developing participants differed signficantly from the ASD participants only on the emotion task.

Participants with ASD did not show differential use of high and low spatial frequency information like the typically developing controls when switching between emotion and gender tasks. The authors suggest reduced top-down modulation on perceptual processes in participants with ASD.

People with autism have been known to show difficulties in processing emotions in faces. Could this low spatial frequency bias in hybrid faces be a clue to understanding why that is so? :p The picture is a pretty good description of this question isn't it?