Overview: This paper tackles a model of dyslexia the cerebellar deficit hypothesis of dyslexia. This is based around the idea that it is cerebellar cortical loops that are disrupted in dyslexia and lead to some of the behavioral deficits such as poor phonological awareness. The authors note that the cerebellum is functionally segregated with lobule VI, crus I, crus II, and lobule VIIb related to language processing and reading. They discuss some strange remediation programs which tie other cerebellar functions in to help reading like balance exercises could treat dyslexia. They present a scatter-shot of work with findings that relate cerebellar activity, structure or connectivity to reading skill or dyslexia status, however the authors argue that the cerebellar deficit hypothesis has never been explicitly tested by testing cerebellar function and connectivity in a group of children during a reading task. This paper does this with a word reading paradigm and activity level and functional connectivity analyses.
Methods: Participants - 23 children with and 23 children without dyslexia, mean age of both groups 9.7-9.8. Left handers were included (one in each group). Children with ADHD were included in both groups with no difference in symptom burden between groups. Behavioral measures - WASI for IQ, WJIII word identification and word attack, RAN, digit span. control needed > 92 on the WJIII fMRI task - participants looked in a string of either real words or fake fonts for a ‘tall feature’ what? A button press indicated whether or not there was a tall feature. FF acted as an active control where the same perceptual stuff was going on but no reading was possible. Text types were blocked. Modeling was RW > fixation for task effects and RW > FF for reading effects. Imaging acquisition - Two runs were collected which two blocks of each FF and RW. Scanner performance was lost for a participant (and they kept them in?. Acquisition is pretty normal COMPLETE CEREBELLAR COVERAGE of course. Imaging analysis - Three analysis methods were used. Univariate mean contrasts of cerebellar activity for RW > fixation and RW > FF. Functional connectivity was tested in two ways ‘background’ functional connectivity ie task based functional connectivity ala Turke Brown 2012 and gPPI. Within and between group maps were generated (what does this mean). Motion was scrubbed at .75 FD with runs with >25% frames removed were excluded, somewhat stringent except for FC analyses .75 is way high. SUIT was used for cerebellum extraction, ‘manual corrections’ were also used and then normalized into SUIT space. The cerebellum was masked into eight literature based functional regions. SVC of each region was used for ROI analyses to a set of reading cortical ROIS (martin 2015). CONN tool box was used for background FC analyses, pretty standard regressors and filtering except for .75 motion threshold still. Task effects were regressed out and L1 was done using GLM and bivariate correlation parameters (not sure where these come from or what this model looks like) gPPI was also done introducing in task regressors for RW > FF.
Results Behavior - no group differences in task behavior. Univariate activity - in controls no cerebellar activity seen for the RW > FF indicating the cerebellar activity seen in the RW > fixation is more about the visual or motor perceptual processing not reading per se. No activity seen in the dyslexia group for either condition. No differences seen between groups however either. Task based/background connectivity - cerebellar seeds were correlated with occipital regions and each other’s cerebellar homologue. Some similar patterns in dyslexics, with connectivity seen between VIIb lobules also showing connectivity to parietal lobe and superior temporal gyrus, other regions showed connection to angular gyrus and SMG. When comparing both groups dyslexics showed more connectivity between cerebellar regions and reading ROIs. gPPI analysis revealed NOTHING.
Thoughts: The authors interpret this paper in a LENGTHY discussion about the cerebellum and dyslexia literature. They actually question their own task and suggest maybe its the implicit reading that makes the effects in this paper kind of watered down and not supporting the cerebellar deficit hypothesis. I wonder why they didn’t report or look at any individual difference stuff and why they didn’t do any whole brain connectivity analyses. Or maybe they did and they were unyielding. One thing confusing is that it did look like control and dyslexic maps revealed FC differences especially related to the reading regions but the authors did not view this as support for a cerebellum deficit because they expected to see the opposite pattern I guess? I’m not sure I’d consider this a null finding… This paper could have some important cerebellar reading parcellation ideas if we dig into cerebellum for P4. I also am still a little unclear about how correlation parameters are included in an L1 model. Not sure if this is just weird SPM modeling that is different or there is something about the model that’s not clicking.