Autism Brain Scans Flawed? You Read It Here First

By Neuroskeptic | November 1, 2012 5:32 pm

According to a piece in Nature today, a major line of research about autism might be seriously flawed:

One of the most popular and widely accepted theories on the cause of autism spectrum disorders attributes the condition to disrupted connectivity between different regions of the brain.

This ‘connectivity hypothesis‘ claims that the social and cognitive abnormalities in people with autism can be explained by a dearth of connections between distant regions of the brain. Some flavours of this theory also predict more connections between nearby brain regions.

Recent studies, however, have found that when a person moves their head while undergoing functional magnetic resonance imaging (fMRI) – a method that maps how different neuroanatomical structures of the brain interact in real time, its functional connectivity – it looks like the neural activity observed in autism. That’s a sobering discovery…

So the characteristic pattern of “abnormal connectivity” in autism might not be real: it might just reflect the fact that people with autism move around more during the scan.

A sobering idea, indeed… but not an entirely new one. I suggested it over a year ago:

Head motion affects estimates of functional connectivity. The more motion, the weaker the measured connectivity in long-range networks, while shorter range connections were stronger… Disconcertingly, this is exactly what’s been proposed to happen in autism (although in fairness, not all the evidence for this comes from fMRI). This clearly doesn’t prove that the autism studies are all dodgy, but it’s an issue. People with autism, and people with almost any mental or physical disorder, on average tend to move more than healthy controls.

ResearchBlogging.orgBen Deen, and Kevin Pelphrey (2012). Perspective: Brain scans need a rethink Nature

CATEGORIZED UNDER: autism, bad neuroscience, blogging, fMRI

    don't throw away the child with the bathwater. All this shows that fMRI stinks not that the autism wiring theory is incorrect.

    To me it is a very valid theory that needs serious study. From personal experience and from interaction as a moderator on a major Asperger forum the bad connection theory feels right.

    When gazzaniga had split-brain operation patients who started talking with their right hemisphere he described pretty much that how the hemisphere expresses itself and sees the world uncannily resembles how we think and talk.

    Which to me decidely indicates there is some over expression of that hemisphere going on.

    It being a different size is not enough to explain that.

  • Anonymous

    Please see the following article for an explanation of the mechansim for these spurious correlations – contrast fixed relative to the scanner and motion correction:

    “A Simulation of the Effects of Receive Field Contrast on Motion-Corrected EPI Time Series”

  • LokaSamasta

    I have autism. The hypoconnectivity hypothesis did and does not ring true with me.

    Only two hypotheses feel right and don't conflict with what we know about autism and that's the hypernutrition hypothesis and the 'cannibal' hypothesis, neither of which have piqued the curiosity of everyday autism scientists and researchers.

  • practiCal fMRI

    Motion is – and is likely to remain for the foreseeable future – a major systematic confound for group studies by fMRI. It seems that many in the neuroimaging community are coming to that realization only belatedly. (Neuroskeptic's blog should be required reading! Thanks for being a skeptical advocate for better methods.)

    We're going to have to keep on hammering the point until the field recognizes the motion limitations. They are everywhere. The scanner is partly to blame, subject selection/grouping is partly to blame, experimental technique is partly to blame, and T2*-BOLD contrast is partly to blame. And anyone relying on correlations in resting-state fMRI data as the basis of their hypothesis is on really thin ice.

    See and links therein for more information on just one of the ways that modern scanner hardware exacerbates the problem.

  • Jon Brock

    What Petrossa said. fMRI is not and never was the appropriate tool to be looking at functional interactions in the autistic brain. This from our 2007 paper:

    “The tentative but converging fMRI and PET studies discussed above may not provide the necessary temporal resolution to fully characterise the connectivity within and between both local and large scale co-ordinated networks. Contemporary models of connectivity emphasise the role of rapid and transient integration and segregation of both local neuronal clusters or assemblies and large scale co-ordinated networks (Friston, 2000; Engel and Singer, 2001; Varela et al., 2001).

    “Recent models of cortical ‘co-ordination dynamics’ (Bressler and Kelso, 2001) claim that the transient re-organisations of coupled local neuronal clusters or assemblies are indexed by changes within a millisecond timescale of patterns in the frequency and phase of oscillatory brain activity, requiring detailed measurement of the spectral characteristics of the cortical signal within very brief time windows (Penny et al., 2002; Makeig et al., 2004).

    “Contemporary research has focussed on activity in the gamma band (30–60 Hz) which has been measured in tasks demonstrating the binding together of parts to make coherent wholes (see review, Tallon-Baudry and Bertrand, 1999). In our earlier paper (Brock et al., 2002), we predicted, therefore, that disordered connectivity would be manifested in task-specific abnormalities in gamma activation in individuals with autism.

    Besides gamma band activity, the role of other frequencies in the spatiotemporal dynamics of the brain are emerging as equally important. Von Stein and Sarnthein (2000) propose that different frequencies are a measure of the scale of integration, with gamma associated with local synchronization, beta with integration between neighbouring areas and alpha/theta with long-range co-ordination.”

    Rippon, G., Brock, J., Brown, C.C., & Boucher, J. (2007). Disordered connectivity in the autistic brain: challenges for the ‘new psychophysiology’. International Journal of Psychophysiology.

  • Michael Forbes Wilcox

    Wow! I never understood how the “long-distance” deficiency could be reconciled with other knowledge of (slow but) relatively more myelination in autistic brains. This should promote better long-distance signaling, not worse.

    Spoken as an amateur neuroscientist who is struggling to understand his own autistic brain.

  • nathanww

    While that's certainly a big experimental design concern, it doesn't sink the theory of global hypoconnectivity/local hyperconnectivity in autims–there's lots of converging evidence from behavioral and EEG studies.

  • jonathan

    I thought head movement was usually controlled for in these experiments and they threw out data with too many artifacts in some studies.


    @ LokaSamasta

    Me too. And it's not so much hypo, hyper connectivity imho but just different connectivity.

    I reasoned it as follows:
    A genetic cause makes the white matter develop differently during gestation, which in turn causes grey matter to develop differently. The neural feedback makes neural pathways develop along different lines than the average

    As a consequence all sorts of system anomalies appear (neurotransmitter blances, hormonal balances), which in turn get reinforced via environmental feedback.

    It's open for debate if this is in fact an evolutionary trend where the brain develops away from the primate undercarriage since practically all disorders have emotional suppression in common pointign to letting go of the limbic system as primary driver.

    Anyway, i hope it is so humanity can really be human instead of just mere primates that talk

    Imo this lays at the foundation of most neurological disorders, such as PD's,Schizophrenia, true autism etc.

  • Neuroskeptic

    jonathan: It is, but the concern is that current methods don't correct for it quite enough.

  • DS


    I would go a step further and suggest that current methods introduce all sorts of motion related systematic error in their attempt to correct for motion effects.

  • DS

    It would appear that the best way to deal with motion contamination of one's fMRI data is to pronounce the data uncontaminated by virtue of your lab's efforts to avoid contamination.

    > sarcasm off <

  • practiCal fMRI

    Received this tweet today from Peter Bandettini:

    Multi-echo is a solution to fMRI motion problems as discussed by Pelphrey

    This is Prantik Kundu et al's multiecho work that characterizes BOLD (or T2*) dependent changes from other changes. Christian Schwarzbauer had another dual echo approach in NeuroImage recently. I do like the general approach of using what amounts to an extensive “navigator signal” to mitigate some of the motion effects, but I also recognize that the temporal costs will dissuade many from adopting the methods.

    I'll see if I can do a review of the multiecho methods presented to date on my blog, some time twixt now and next spring's conference season. I'd hope there will be many sessions on motion confounds at the ISMRM and HBM. Might be useful to have a review of current options before then.

  • Neuroskeptic

    Thanks – Bandettini makes an interesting point – would be great if you blogged about that, as I don't quite understand it…

  • DS

    Any imaging based method of motion correction that uses data of the same or similar resolution to that of the typical fMRI data set is bound to fail at eliminating enough motion artifact to matter to fMRI.

  • DS

    I think that we will eventually be forced to use prospective motion correction with the motion data coming from some sensors fixed relative to the skull. Others have suggested fixing sensors to the maxillary teeth. Excellent place to start. Best to do it now before we generate another 10 years of data.

  • Anonymous

    Just a young student here: I understand that fMRI has its faults, but why is nobody talking about post-mortem studies? They're not that many, but they all support the hyper-local / hypo-long distance connectivity. Or were they also moving? :) Maria

  • Anonymous

    Autism is a dysfunctional
    vascular disease of the colon,
    radiographics of postmortem
    tissue were done in europe
    in the 60's, the mesentary
    arteries all profoundly
    hypovascularized and dysfunctional,
    this causes the major tissue
    problems in the gut

    • Gabrielle Wolf-Stahl

      nope- my kid has NO GASTROINTESTINAL issues…many Autistics do not…please do not confuse health issues that SOME Autistics have with a cause…

  • Emily Morson

    Motion artifacts could explain why studies find increased short-range and decreased long-range connectivity in autism, but what about the recent study showing decreased short-range connectivity as well? Would motion also decrease the appearance of short-range connectivity, or is something else going on here?



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