Abstract

Despite the widespread use of the Research Domain Criteria (RDoC) framework in psychiatry and neuroscience, recent studies suggest that the RDoC is insufficiently specific, or excessively broad, relative to the underlying brain circuitry it seeks to elucidate, leading to potential misrepresentation of circuit-function relations. We used a latent variable approach to address this issue, specifically utilizing bifactor analysis. We examined a total of 84 whole-brain task-based fMRI (tfMRI) activation maps from 19 studies with a total of 6,192 participants. Within this set of 84 maps, a curated subset of 37 maps with a balanced representation of RDoC domains constituted the training set of our analysis, and the remaining held-out maps formed the internal validation set. Furthermore, we externally validated the factor solutions from our curated training dataset using an independent set of 36 coordinate maps sourced through Neurosynth. We used RDoC constructs as seed terms for Neurosynth topic meta-analysis. We hypothesized that if boundaries of RDoC domains warrant refinement, this would be indicated by the presence of overlapping domains or domains lacking specificity. Our findings suggest that a bifactor data-driven structure fits better with the current corpus of tfMRI data, with a general domain representing task-general patterns of brain activation. The data-driven model also proposes a different group of major domains, particularly splitting the RDoC cognitive systems domain into distinct domains. Data-driven models are useful for revising the posited circuit-function relations outlined in the current RDoC framework.