In recent years, there has been an explosion of interest in the neuroscience of autism spectrum disorder (ASD), and much of this research has targeted how activity in the autistic brain differs from that in the neurotypical brain. However, delving into this literature can be frustrating and immensely confusing; a quick PubMed search for “autism functional connectivity” returns over 300 results with titles describing connectivity in ASD in frequently contradictory terms such as “altered”, “preserved”, “disrupted”, “reduced”, “hyper”, etc. This is somewhat unsurprising given the well-known heterogeneity of ASD(s), but particularly for the purposes of improved diagnoses, interventions and general understanding of the disorder, it would be advantageous to be able to parse out some core characteristic(s) of the disorder.
As chance (and luck) would have it, this week’s speaker, Dr. Marlene Behrmann from Carnegie Mellon University, has been making quite a few headlines in the last week for doing just that. In a paper published in Nature Neuroscience on Jan. 19th, she and her co-authors propose that rather than a particular pattern of over- or under-connectivity being a distinguishing feature of autism, a core characteristic of ASD is having an “idiosyncratic brain”. In other words, whereas neurotypical adults tend to display relatively similar and stereotyped patterns of functional connectivity, adults with ASD deviate from this pattern in an inconsistent, “idiosyncratic” way.
What led Dr. Behrmann and her co-authors to this conclusion was an initial observation that compared to control brains, their ASD group as a whole demonstrated what the authors call a “regression to the mean” effect. This observation came from an analysis of resting-state fMRI scans (in which BOLD signal changes are measured in the absence of a task and taken as an indication of spontaneous activity, revealing functional networks through correlated patterns of activation) from five different datasets with a total of 73 controls and 68 ASD subjects. In areas of the brain with highly correlated activity between hemispheres (“homotopic interhemispheric connectivity”) in the control group, the ASD group demonstrated reduced connectivity; but where there was less interhemispheric connectivity in the control group, there was comparatively increased connectivity in the ASD group.
Although there are a number of possible explanations for this phenomenon, the one which Dr. Behrmann and her co-authors’ data best support is that there is substantial variance within the ASD group in terms of their patterns of functional connectivity, whereas control subjects are more similar to each other. The following schematic figure from their paper illustrates this nicely:
As one can see, when three hypothetical ASD subjects each have areas of strong (yellow/orange) and weaker (blue) interhemispheric connectivity but these areas are spatially inconsistent, their group average would be attenuated in each direction toward a mean level of connectivity.
And in fact, this is precisely what Dr. Behrmann and her co-authors observe. In each of their five datasets, individual control subjects’ patterns of homotopic interhemispheric connectivity are more correlated with each other (as demonstrated by more red/yellow in Figure 2a) whereas ASD subjects seem to be less correlated with each other (more blue; quantified in Figure 2b).
Interestingly, the degree of autistic subjects’ deviation from the control group’s mean connectivity pattern was modestly correlated with behavioral measures of ASD as indicated by Autism Diagnostic Observation Schedule (ADOS) scores. In particular, there were significant correlations between total ADOS or ADOS communication scores and distortion index (i.e., the within-subject variance of interhemispheric connectivity differences) such that subjects with more severe symptoms had more idiosyncratic connectivity profiles. Moreover, they found similarly variable patterns of functional connectivity that deviated from the more stereotyped control patterns when they also looked at heterotopic interhemispheric connectivity (connectivity between different regions across hemispheres) and intrahemispheric connectivity in each hemisphere, although these connectivity deviation measures did not correlate as strongly with the behavioral measures. Nevertheless, Dr. Behrmann and colleagues’ data support their proposal of “functional idiosyncrasy” as a signature of individuals with ASD.
At first, one might say, “Ok, so functional connectivity in autism varies by individual; that seems obvious, what’s the big deal?”. I would argue that a major implication of this study is that it provides a new framework for thinking about functional connectivity in autism. Not only does it help clarify many of the discrepancies in previous studies’ findings, it helps guide future studies by focusing not only on how two groups (i.e., ASD and control) are different from each other but also on how the within-group variances differ between groups. Instead of the highly variable nature of ASD being merely a nuisance for statistical tests, it is precisely the variability in deviation from the “control template brain” which sets it apart. This also raises some interesting questions for how “canonical” vs. “idiosyncratic” connectivity profiles develop. For instance, the authors suggest that the functional idiosyncrasies in autism could arise (at least in part) from autistic individuals’ abnormal interactions with their environment throughout development, whereas neurotypical individuals who (presumably) interact with their environment in a more typical way would have similar functional connectivity patterns reinforced. This is an interesting hypothesis that could be addressed by looking for similarly variable and deviant connectivity patterns in some other clinical populations.
This is likely to be the primary topic of Dr. Behrmann’s lecture given this paper’s recent publication and the fact that her talk is titled “Alterations in canonical cortical computations in Autism”. However, Dr. Behrmann has been extremely influential in other realms of cognitive neuroscience research as well. She also studies aspects of complex visual processing such as object recognition, mental imagery, spatial attention, face processing and more, and employs a wide range of methods including behavioral measures in brain-damaged patients, functional neuroimaging, and computational modeling (you can find links to her MANY publications on her website – definitely worth the read!). Dr. Behrmann is both a wonderful speaker and a wonderful scientist, so don’t miss her on Tuesday, January 27th at 4pm in the Center for Neural Circuits and Behavior Marilyn C. Farquhar Conference Room!
Megan Kirchgessner is a first-year Neurosciences graduate student currently rotating with Dr. Eric Halgren. When not in lab or in class, she is probably running around the Torrey Pines Gliderport and getting distracted by the sunset, listening to weird music and pretending to be indie, eating excessive amounts of carbs, alternating between watching independent films and Will Farrell comedies, or stalking the La Jolla Cove sea lions.
Hahamy, A., Behrmann, M., & Malach, R. (2015). The idiosyncratic brain: distortion of spontaneous connectivity patterns in autism spectrum disorder. Nature Neuroscience. Advanced Online Publication. doi:10.1038/nn.3919