Autism and the optical gestalt
Can you read this? The first lines are the hardest. So if you have trouble, scroll down a bit. Once you have the hang of it, you probably will have no problem reading it from the start.
This is a great example of perception as interpretation. We are not cameras and tape recorders. In the case of vision, we use the optical gestalt rather than information about pixel by pixel intensity. There are some who have difficulty with the gestalt but are great at the details. In particular, it has been said that individuals with Autism Spectrum Disorder (ASD) follow local clues over global cues when interpreting a visual scene. To take an example from a nice recent paper from Nancy Kanwisher’s group (http://web.mit.edu/…/media/pdfs/Koldewyn.JADD2013.pdf), imagine seeing a triangle made up of lots of little squares. At a global level, this is a triangle but it is made up of squares and so at a local level, this would match to a square. If one is asked whether the triangle made of lots of little squares matches to a triangle or a square, most people pick the triangle – the global choice. But individuals with ASD often pick the local choice (matching to the small squares that make up the big triangle). In the experiment, the likelihood that an individual picked local over global increased with the severity of his or her autistic symptoms.
You can try this out for yourself. Look at the picture below:
Now which of the two pictures below does the one above match?
Kanwisher and her colleagues then asked whether children with ASD who see local, actually cannot see global. In other words, are they matching a sample to the local cues because they cannot put together the global picture? To examine this question, children were asked to match samples to either the local (what is this shape made up of?) or global (what’s the big shape?) cue. They were then given a choice that was either congruent (eg. a triangle made of triangles) or incongruent (a triangle made of squares). All children were faster at matching to congruent than to incongruent choices. Think of being asked to match XXXX to either BLUE or PINK. Of course, you can do this. But it akes you a moment to ignore the semantic information and choose the work pink over blue (this is the Stroop effect btw). And you would be much faster at matching XXXX to congruent choices: PINK or BLUE. The comparison of reaction times between matching to a congruent or a incongruent choice tells us about how someone is processing the information (and is the basis for tests of implicit bias….but we digress).
Let’s get back to ASD and global vs local processing. Well, it turns out that ASD children take longer to match to incongruent choices, global or local, than to congruent choices. There is no difference that is based on local or global incongruence. They are seeing the incongruity at both local and global levels. And in this respect, the performance of ASD children is not different from that of typical children (except in one respect: they are slower across the board in all reaction times). This finding tells us that ASD children can see the global picture. BUT they don’t.
Kanwisher and colleagues interpret their results as a disinclination rather than a disability or inability to see globally. I will offer a different take on this, one that is far less cognitive. Maybe ASD childen simply don’t make the eye movements that are needed to look at the whole picture. There are many reports of differences between the eye movements of ASD and typical individuals. Below, I sketched potential differences in eye movements (the red lines) that could bias perception toward global (left) or local (right) cues.
The fact that ASD individuals can but don’t interpret optical scenes globally reminds me of the finding that individuals with damage to the amygdala can but don’t look at people’s eyes. without looking at eyes, fear is a difficult facial expression to interpret. Indeed patients with amygdala-damage have a problem with recognizing fearful faces. When told to look at the eyes, the patients do fine at recognizing fearful faces, but shortly after receiving the instructions, they appear to forget to look at the eyes and revert to their more typical facial scanning mode that fixates on the mouth.
One possibility is that both of these perceptual differences in fact stem from a motor difference, a difference in the way that people scan a scene.
I should say that I tend toward the local myself and this is what makes it hard for me to understand cartoons. There is a Far Side cartoon that epitomizes this. “Mad scientists” are standing around in their white coats and one is being urged to drink a smoking potion. “Drink it, drink it”. The dry humor was lost on me because the scientist who was supposed to drink the potion did not have a mouth drawn in. My understanding got hooked up on this small detail. Local processing may be over-represented in the scientific personality…