Autistic Traits Aren’t Linked To Brain Anatomy?

By Neuroskeptic | April 15, 2015 3:58 pm

According to a large study just published in the Journal of Autism and Developmental Disorders, there’s no correlation between brain anatomy and self-reported autistic traits.


Dutch researchers P. Cedric M. P. Koolschijn and colleagues looked at two samples of young Dutch adults: an ‘exploration’ sample of 204, and a separate ‘validation’ group of 304 individuals.

Most of the participants did not have autism. The researchers looked for associations between various aspects of brain structure and autistic traits, using the AQ questionnaire, a popular self-report measure.

Autistic traits are personality or behavior features similar to (but generally milder than) autism symptoms. For example, the first item on the AQ is “I prefer to do things with others rather than on my own.” If you disagree with that, you get a point. More points means more autistic traits.

Koolschijn et al. used VBM, vertex-based cortical thickness analysis, and diffusion weighted imaging to explore different aspects of brain grey and white matter anatomy. However, although AQ scores were weakly correlated with the volume of a few brain areas in the exploration sample, none of these correlations were confirmed in the larger validation sample, suggesting that they were just false positives caused by the large number of multiple comparisons.

The authors conclude that:

In this large exploration-validation study, we investigated if autistic traits in neurotypical adults were associated with a comprehensive series of structural brain indices. In contrast to our hypotheses, no evidence was found for any relationships between individual differences in behavior and brain anatomy.

Koolschijn et al. note that six previous published studies have looked for brain structural correlates of AQ scores in neurotypical people. Of those, four reported finding at least one significant association… but unfortunately, they all reported different (and even contradictory) things. The largest of these studies had 85 people, so the new study is far bigger.

If the AQ isn’t related to (currently measurable) differences in brain anatomy, why not? One possibility is that the AQ scale is just not a good measure of autistic traits. Koolschijn et al. note that it has shown “relatively low” correlations with non-self report measures of autism symptoms, such as the ADI and ADOS.

But could it be that autism traits aren’t associated with neuroanatomical differences because autism itself isn’t? Six months ago, a large study of brain anatomy in people diagnosed with autism found virtually no differences from neurotypical controls.

If most people with autism have entirely normal brain structure on the macroscopic scale, we might need to focus on the cellular level in order to understand the autistic brain.
ResearchBlogging.orgKoolschijn PC, Geurts HM, van der Leij AR, & Scholte HS (2015). Are Autistic Traits in the General Population Related to Global and Regional Brain Differences? Journal of Autism and Developmental Disorders PMID: 25847757

  • Hellson

    Another reason might be the specific characteristics in the subsample of people with autism who are able to complete self-report instruments. Given the large comorbidity of autism with, for example, intellectual disabilities, it is possible that the autistic traits in Koolschijn et al’s samples were relatively mild, or at least different to other populations that are diagnosed within the autism spectrum.

    Other than that, interesting paper!

    • Neuroskeptic

      Indeed, 95% of the people in this study didn’t have a diagnosis of autism – they were just general population volunteers. The idea was to look at milder traits in the population rather than full-blown symptoms. I’ve edited the post to clarify this because it was rather obscure before!

      • Hellson

        I haven’t read the study itself, just your blog post (as I am currently at home and have no access to my institutional library). The fact that the authors did not find any links of mild traits is certainly relevant, but makes me hesitant to draw conclusions about the “autistic brain”.

        • D Samuel Schwarzkopf

          The paper should be open access so you can access it at home… :)

  • Felonious Grammar

    What can this imaging show that autopsy can’t? Surely there has been some research on the cadavers of people who had been diagnoses with Autism. Yes? No?

    • Neuroskeptic

      There has been some. But the sample sizes are generally tiny, I think 5-15 per group is the norm because it’s so difficult to get well preserved cadavers.

  • Adam J Calhoun

    What are the reasons to think that there WOULD be anatomical differences between autistic and normal patients?

  • Uncle Al

    I prefer to do things with others rather than on my own.”” IQs on a committee pulling together add like ohms in parallel transistors. Put the smartless in charge to foment advocacy. Advocacy burns others’ monies to create intent. An advocate makes virtue of failure. The worse the cure the better the treatment – and the more that is required.

    The most dangerous and destructive person on Earth is somebody with an original idea who thinks he deserves compensation for having it. End autism! Or at least make sure 50% of autists are women.

    • Rickety Janes

      Wow, they let you have a computer in Super-Max, Ted?

      • Uncle Al

        Just wait until you enter a corporate environment institutionalizing the congenitally inconsequential. Just wait until your Graduate Remedial English class empowers you to read Dilbert (that is submitted rather than being fabricated).

        • Rickety Janes

          what is submitted? Dilbert? your comments are genuinely intriguing, but i am having trouble parsing your grammar

          • Uncle Al


            My first posts is public school-level text. Get a better public school. My second post is…nuanced art. fnord

          • Rickety Janes

            lol…I will do so…you have a very information-dense style; in spite of my own spiteful nature, I’m starting to enjoy it, like one gets used to resinated wine….do you have a blog or available writings?

          • Rickety Janes

            hey, i rescind my earlier sarcasm, your stuff on chirality is interesting….

  • Jetwarp

    I can;t find the difference but then again, my eyesight is aged. Where is the difference, if any?

    • Neuroskeptic

      There’s no difference… which is my graphical representation of the results of this paper!

  • D Samuel Schwarzkopf

    It doesn’t seem very surprising that AQ scores don’t correlate with any
    gross differences in brain structure (and I’d call anything that shows
    up in these kind of analyses a gross difference). I haven’t read the study but I wonder if their AQ range captures the extremes really? The version of AQ they used seems different to the one I know (much larger numbers).

    As a neurophysiologist I used to be extremely skeptical of questionnaires in general. I am still skeptical but I have since become a little bit more open-minded. AQ scores do seem to covary with a lot of factors. We even found some correlations in our autism study although these were completely exploratory (which I labelled as such, all without preregistration… :P) so you shouldn’t interpret them as anything but a surprising blip until somebody tests this explicitly.

    But there have been other studies that show fairly clear relationships with AQ. In our data we saw a clear bimodal distribution of AQ for the ASD group vs controls:
    Other people actually find they overlap a lot more so not sure what this means. I think the ASD samples used in different studies are probably not homogeneous.

    So my feeling is that AQ does measure something meaningful but the underlying factors aren’t really very clear. It may just be a coincidental that AQs tend to be higher in ASD. The underlying factor may differ in ASD but it isn’t actually causally linked. To me this makes a lot of sense (especially because AQ would classify a lot of my friends and colleagues as having autism :P).

    • Neuroskeptic

      They used a short version of the AQ with 29 instead of 50 questions; they also score each question as 0-3 instead of 0-1, which is why the scores are higher than normal.

      However I don’t think this is likely to matter. All of the different AQ versions correlate with each other very strongly according to the literature (even 10-item versions).

      And yes, it’s certainly true that AQ scores tend to be higher in people with ASD than in controls.

      However, the same might be true if you used an anxiety questionnaire, or an OCD questionnaire. And AQ scores might also be higher if you used (say) depressed patients, not ASD patients. It’s not clear AFAIK whether the AQ measures autism specifically or whether it’s picking up on general “mental health problems” or “low self esteem”.

      • D Samuel Schwarzkopf

        Yeah, as I said I think it doesn’t really pick up anything that actually has a causal link to autism. Which is why I’d also be very skeptical of claims that anything that is linked AQ in controls speaks in any way to the mechanisms underlying autism.

        It would be very interesting to know what AQ actually measures though. But whatever it is, I doubt it has a very pronounced structural correlate.

  • David Nassau

    That is so flawed. Just the fact that the autistic traits are self described means you have preselected the population to exclude noncommunicative people or people who cannot focus on a questionnaire, which means you have eliminated the more serious cases from the study. It says nothing about actual autism.

  • kathryn hedges

    The autistic traits in the AQ are human traits that may cluster more tightly in autistics than non-autistics, or be felt more strongly. They’re also related to a whole slew of different brain functions. Now, if I could use this study to tell well-meaning non-autistics that my autism is different than their “not liking noisy groups either” and that is why I have a disability and they don’t, that might be useful.

  • kathryn hedges

    Also, I thought the current understanding of the neuroscience of autism is that it’s related to differences in connectivity, not lesions in specific regions. These differences in connectivity vary between autistic subjects (which is why some autistics have sensory hyposensitivity and others hypersensitivity, some have or lack speech development delays, etc.) and I have the impression they’re difficult to quantify, let alone link to a specific feature of a person’s autism.

  • Pingback: Is Synesthesia A Brain Disorder? - Neuroskeptic()

  • Pingback: Do we worry too much about what we eat? - Stress Anxiety Guide()

  • Pingback: Is Synesthesia A Brain Disorder? | Trend Inside()

  • Jessica Hert

    This study is very interesting. While there may not be a direct effect in the brain, there is some genetic truth to autism. Genes in DNA have been linked to autism and their effect. Even though a parent may not have any sign of these mutated or micro deleted genes, some people diagnosed with autism so evidence in some of their genes. It is important to not rule out any genetic relation or physical evidence.

  • Pingback: Autistic Traits Aren't Linked To Brain Anatomy? - World Health Portal()

  • Franck Ramus

    We just had a very similar experience to that of Koolschijn et al., while investigating the neuroanatomical basis of dyslexia using VBM:

    Jednoróg, K., Marchewka, A., Altarelli, I., Monzalvo, K., van Ermingen-Marbach, M., Grande, M., . . . Ramus, F. (2015). How reliable are grey matter disruptions in specific reading disability across multiple countries and languages? Insights from a large-scale voxel-based morphometry study. Human Brain Mapping, 36(5), 1741-1754.

    My take on this is that there are probably some reliable neuroanatomical correlates of the cognitive phenotypes of interest (ASD traits or dyslexia or anything else), but not the sort that VBM can reveal. I mean any method that performs analyses in a common template where all individual brains have been warped. We need to study each brain very carefully in its native space, and that’s very time-consuming, that can only be partly automated.

    The only reason why almost everybody has been using VBM is that it’s very easy and fully automated, but that’s not a very good reason.

  • Sean Kamran Rowshandel

    Do VBM, vertex-based cortical thickness analysis, or diffusion weighted imaging measure the amount of synapses? Wasn’t there a theory about the autistic brain that it had too many synapses?



No brain. No gain.

About Neuroskeptic

Neuroskeptic is a British neuroscientist who takes a skeptical look at his own field, and beyond. His blog offers a look at the latest developments in neuroscience, psychiatry and psychology through a critical lens.


See More

@Neuro_Skeptic on Twitter


Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!

Collapse bottom bar