Measuring Autistic IQ

Share on FacebookShare on Google+Email this to someoneTweet about this on Twitter

ScienceNOW Daily News has an article on The Case of Mistaken IQ:

Health workers routinely assess autistics using a standard IQ test known as the Wechsler test. But this test requires that children understand oral commands, a trait that many autistic children have trouble with. Cognitive neuroscientist Laurent Mottron of the Hopital Riviere-des-Prairies, Montreal and colleagues noticed that autistic children did poorly on the verbal comprehension part of the Wechsler but exceedingly well on a part that tests non-verbal intelligence and reasoning.

So the researchers decided to test 30 autistic children and 30 autistic adults with a different IQ test called the Raven’s Progressive Matrices test, which is written rather than oral. Healthy children and adults performed similarly on both the Wechsler and the Raven test. But speaking autistics scored up to 30 percentile points higher on the Raven test than the Wechsler test, the researchers reported here 19 February at the annual meeting for the American Association for the Advancement of Science (which publishes ScienceNOW). “Thirty percentile points could raise a retarded person to normal or a normal one to a superintelligent one,” says Mottron.

It sounds like a bit of factor analysis could have gotten them the same result. I didn’t find an actual reference, but here’s the AAAS press release about the symposium.



  1. “It sounds like a bit of factor analysis could have gotten them the same result” 
    Would it have, though? 
    IIRC, Verbal Comprehension is one of the more g-loaded sub-tests in the Wechsler battery.

  2. Hmm. Seems to me autism is not terribly different from “cognitive impairment”, since one of the defining characteristics of autism is that one’s mental functioning (cognition) is, like, impaired. Or are autistics merely “differently cognitioned”? And what in the world would that mean? 
    Or is IQ merely a measure of potential cognitive ability? If so, how could that possibly be measured?

  3. Hi guys, a general comment: 
    this is one of the several examples of IQ limitations. On this regards I would appreciate your opinion about the following study we reported on our webblog (please visit us, we are new online!): 
    Season of birth is associated with anthropometric and neurocognitive outcomes during infancy and childhood in a general population birth cohort.  

  4. Omigod, not to gross anyone out, but I think I just came when I read Michela’s message… here is the abstract of the study she pointed us to. We already knew winter-springs are more likely to be schizophrenic, but now it seems they’re likely to be somewhat smarter as well! 
    Michela — I’m about to post an article I’ve written that shows that artistic & scientific geniuses are disproportionately winter-spring. Thank god I didn’t post it yet, since I’ve got another great study to link to! The standard epidemiological interpretation of seasonality of births is that it reflects exposure to microbes. Some could harm you (schizophrenia), but some could apparently help you too. Fascinating!

  5. There’s an undercurrent here of defining disability up by pointing at a test on which autistics perform well, but I think the idea of potential cognitive ability is exactly right. The high spatial abilities indicate high value for g, which in a normal person would be expressed in verbal abilities as well. 
    What defines autism, and by extension the autism spectrum, is not general cognitive impairment but social impairment.

  6. This is interesting, but much of it is not surprising. Having a (very) high-functioning form of autism myself, I feel as if I may be able to extrapolate a little to have an idea of what is going on here. It seems as if the central feature of the autism spectrum is not a general, all around cognitive deficit but a sometimes extreme increase in one type of intelligence at the expense of others, particularly social intelligence. I highly doubt that even low functioning autistics have any doubt that they themselves have minds, as the press release suggests, and at least farther up the spectrum there is no difficulty following someone else’s train of thought either, as long as we understand the subject matter. The lack of “theory of mind” is more a lack of intuition about what someone else may be thinking or feeling, if it is not explicitly said. This may be why I have never found sociological questions about how groups of people act very interesting, and why biological arguments for the basis of the mind are so natural to me in a way.  
    About the progressive matrices test, I can very much relate to the prominence of spatial and visual intelligence. Though I can (obviously) express myself verbally just fine, and actually started speaking a little ahead of time developmentally, my default mode of thinking has always been visual.  
    The most surprising finding to me is that normal people use a verbal strategy to solve the progressive matrices. I looked up some sample questions on the net (to see one go to, and I cannot see anything there that even vaguely resembles letters (except for the instructions). I would have assumed that everyone would use a visual/spatial strategy. 
    This probably explains why I have such a strong preference for intuitive (to me) mental models of natural phenomena over rigorous mathematical ones, as I have said before. I feel that it is much more efficient to try to understand how a biochemical pathway works, for example, by trying to visualize it working in my mind and theoretically altering parameters rather than writing equations or computer programs, which are essentially verbal formulations of the same ideas. The equations are rigorous and quantitative, whereas the visual/spatial models are qualitative, but the equations need to be simple enough to solve, which ofter requires severe approximations, and the computer programs may show the behavior of the model but not give a satisfying explanation for why it acts as it does. This brings me to a question I have been wondering the answer to for quite some time, and I think I alluded to in the post about the “end of insight”: when does each method (spatial analysis or rigorous mathematical model) actually capture the behavior of a system better, i.e. lose fewer critical features of the behavior through the inevitable lack of precision?