Bias-accounting meta-analyses overcome cerebellar neglect to refine the cerebellar behavioral topography

Abstract

The cerebellum plays important roles in motor, cognitive, and emotional behaviors. Previous cerebellar coordinate-based meta-analyses and mappings have attributed different behaviors to cerebellar subareas, but an accurate behavioral topography is lacking. Here, we show overrepresentation of superior activation foci, which may be exacerbated by historical cerebellar neglect. Unequal foci distributions render the null hypothesis of standard activation likelihood estimation unsuitable. Our new method, cerebellum-specific activation-likelihood estimation (C-SALE), finds behavioral convergence beyond baseline activation rates. It does this by testing experimental foci versus null models sampled from a data-driven, biased probability distribution of finding foci at any cerebellar location. Cerebellar mappings were made across five BrainMap task domains and thirty-five subdomains, illustrating improved specificity of the new method. Twelve of forty (sub)domains reached convergence in specific cerebellar subregions, supporting dual motor representations and placing cognition in posterior-lateral regions. Repeated subsampling revealed that whereas action, language and working memory were relatively stable, other behaviors produced unstable meta-analytic maps. Lastly, meta-analytic connectivity modeling in the same debiased framework was used to reveal coactivation networks of cerebellar behavioral clusters. In sum, we created a new method for cerebellar meta-analysis that accounts for data biases and can be flexibly adapted to any part of the brain. Our findings provide a refined understanding of cerebellar involvement in human behaviors, highlighting regions for future investigation in both basic and clinical applications.