In 1996, Francesca Happe published an intriguing paper in which she reported that children with autism were less susceptible to visual illusions than non-autistic kids, suggesting in turn that the way they perceive the world is fundamentally different. However, subsequent studies have provided somewhat mixed results. Some have also found immunity to illusions; others have found that people with autism are just as susceptible as those not on the spectrum. It's not clear why the results are so contradictory, but it may be down to differences in the way the tests have been administered in different studies, or the fact that different illusions have been tested, or a combination of both.
In a new study, Peter Mitchell and colleagues tested a group of adolescents and adults with ASD on the Shepard illusion. The pictures were presented on a computer and the participants' task was to adjust the size of the right hand table by pressing the arrow keys until they thought it was the same shape as the one on the left.
The people in the ASD group were, in fact, susceptible to the illusion. In the example above, they would make the right hand table longer and thinner than it needed to be. However, the illusion effect was considerably weaker than in a group of non-autistic people from the same age range.
One possible scenario is that there are two subgroups of people with ASD - some who are susceptible to the illusion and others who are not. Mitchell et al. claim there was no evidence for different subgroups in their data. Nevertheless, it's quite noticeable that the error bars on their graph are much larger for the ASD group, indicating a much wider spread of scores. It would be interesting to know just how many of the ASD individuals showed an illusion effect that was markedly reduced and how many showed 'normal' susceptibility.
The results are particularly interesting because, in a previous study using a very similar task, the same researchers found that people with autism showed normal sensitivity to other types of illusion, such as the Ebbinghaus illusion (below). Unfortunately, they haven't yet given both kinds of illusions to the same people with ASD using this task. However, the results to date suggest that there may be something specific about the way the Shepard illusion works that means (some) people with autism are less affected.
Mitchell et al.'s argument is that people with autism make less use of 'top-down' knowledge. That is, their perception is less guided by knowledge of what it is they are viewing. In the Shepard illusion, the relevant 'knowledge' is the 'fact' that the 2D images have depth, which leads to the perceptual 'squashing' effects.
However, there is a small problem with this argument. Mitchell et al. also included a second version of the test, in which the table legs were removed, so the participants just saw two parallelograms. But even in this parallelogram version of the test, when the participants didn't know the objects had depth, it was still the case that (a) both groups showed an illusion effect; and (b) the effect was smaller in the autism group. One possibility is that, having seen the tables version of the test, everyone interpreted the parallelograms as table tops too. This would be easy to check in future experiments.
Overall, it does seem that there are important differences in the way that (at least some) people with autism perceive the world. In some situations these differences might be advantageous. Indeed, Mitchell et al. speculate that the reason why a disproportionate number of people with autism are savant artists is that they are not afflicted by the 'curse of knowledge'. They are able to draw exactly what they see.
However, there may also be down sides. Presumably, the reason we are susceptible to visual illusions is that using our knowledge of the world to infer the 3D properties of objects based on 2D retinal images is adaptive - it's useful more often than not. Visual illusions are the price we occasionally pay for being able to make sense of our sensory environment. It follows, therefore, that an interesting direction for future research is to explore the links between susceptibility to illusions and the sensory difficulties that affect many people with autism.
Reference:
Mitchell, P., Mottron, L., Soulieres, I., & Ropar, D. (2010). Susceptibility to the Shepard illusion in participants with autism: reduced top-down influences within perception? Autism Research, 3, 113-119.