Visual processing problems on the autistic spectrum disorder: what are they, anyway?

I discovered yesterday that there are a large number of people investigating how sensory processing problems contribute to ASD--and specifically, how people on the spectrum tend to be more detail-oriented rather than globally/holistically-oriented. My first reactions: 1) Wow, that's so cool, people are actually studying this! 2) Looks like there's a lot of evidence supporting my hypotheses. Cool! Then I started really looking at the results of these studies.

This shouldn't have surprised me, but it turns out that "not seeing the forest for the trees" can mean a lot of different things, and the implications are very different. (Just like "difficulty understanding other people's mental states" can mean a lot of different things, with different implications). A few entries ago, I raised the possibility that people on the autistic spectrum might be detail-oriented in the sense that they literally do not see a context, only a set of details. They see so many details that scanning becomes impossible. My prediction, in other words, was that at least some ASD people might function like Dr. Gray. One could interpret the body language study as confirming this by showing that ASD people are seeing a lot of separate motions of different parts of the body rather than one coherent, meaningful gesture.

The problem is that some of the studies that are supposed to support ASD people's detail-centric processing suggest that ASD people are phenomenal scanners. O'Riordan et al1 gave neurotypical kids and kids with ASD two classic psychology tasks: a "feature search" task and a "conjunction search" task. (Those of you who know what these are should skip to the next paragraph). A "feature search" is when people are asked to find one object in a set that differs in only one important characteristic. For instance, one might be asked to find a green square in a set of red squares, or a green square in a set of green circles. A "conjunction search" is when multiple traits are present that could be shared by the object you're trying to find. One might be asked to find a red circle among red squares and green circles, for instance. For most people, the different object seems to "pop out" during a feature search. You don't have to look at every single item to find something that is different in only one respect. So neurotypical people take about the same time to do a feature search no matter how many items are in the display. On the other hand, a conjunction search requires one to look at every single item in the display to make sure that it is different in BOTH important respects. Thus, neurotypical people take more time to do the search as the number of items in the display increases. Notice that this task is basically a test of scanning under different degrees of stimulus complexity.

Search tasks like this are one of several common methods that seem to be used to examine the visual perception abilities of kids with ASD. O'Riordan* found the usual pattern for neurotypical kids: basically, feature detection is easy and conjunctive search is hard, with a linear time increase as the number of items increases. Autistic kids performed the same as neurotypical kids on the feature search, but they were significantly faster on the conjunction task. More precisely, "the ASD group was significantly faster than the control group when searching the largest display of 25 items but not when searching 15-item or 5-item displays" (display sizes were 5, 15, and 25 items). A study that cites the O'Riordan study2 says that the ASD group was nearly twice as fast in the 25 item condition!

This means that ASD kids are actually MUCH better scanners than neurotypical kids. That seems to rule out the possibility that ASD kids are hyperfocused Dr. Gray types. These results would seem to suggest that, if ASD kids knew what details to look for in social & practical situations, they would actually pick up meaningful detail much faster (and probably as accurately) as neurotypical kids.

Now wait a second, I thought. Maybe the problem is that these tasks are so circumscribed compared to the real world, which has countless 1000s of meaningful visual stimuli to parse. In this study, the rules of the game were pretty clear (find the one thing that's different), and there were only 2 items in the conjunction. In the real world, the parsing rules are loose or even nonexistent, and rarely stated, while there are literally countless conjoint visual stimuli to sift through. So maybe ASD kids are more skillful in clearly circumscribed tasks where they can put their detail abilities to use, but the real world is just too complex for them.

There's a problem with that interpretation, though. Notice that conjunction searches are more complex than feature searches. So you'd expect that even in the lab, ASD kids would be better at feature searches than neurotypical kids, and worse at conjunction searches--but that's exactly what you don't find.

I think a study would have to be done with much larger numbers of conjoint features in order to clarify the issue. After all, there's a bigger difference between real world-level complexity and lab conjunction searches than there is between conjunction searches and feature searches. But at the very least, the evidence suggests that a scanning problem is unlikely.

As I read more, the question becomes: we might know what it doesn't mean to see the forest for the trees...but what does it mean, and what do the studies actually predict? And is there anything that the studies aren't looking at that maybe they should be? Stay tuned for further thoughts...

*Others seem to have found this too, according to the following overviews of the topic...
Behrman, Marlene, Thomas, Cibu & Humprheys, Kate, Seeing it differently: visual processing in autism. Trends in Cognitive Sciences Volume 10, Issue 6, June 2006, Pages 258-264
Dakin & Frith. Vagaries of Visual Perception in Autism. Neuron, Vol. 48, 497–507, November 3, 2005
1 O'Riordan, Michelle A., Plaisted, Kate C., Driver, Jon, Baron-Cohen, S. Superior visual search in autism. Journal of Experimental Psychology: Human Perception and Performance, 0096-1523, 2001, Vol. 27, Issue 3. O'Riordan also did a follow-up study in 2004, which is next on my list of articles to read.
2 Dakin & Frith. Vagaries of Visual Perception in Autism. Neuron, Vol. 48, 497–507, November 3, 2005