A data-driven latent variable approach to validating the research domain criteria framework

Area of Science:

• Neuroscience
• Psychiatry
• Cognitive Science

Background:

• The Research Domain Criteria (RDoC) framework is widely used in neuroscience and psychiatry.
• Concerns exist regarding the RDoC framework's specificity and breadth in relation to brain circuitry.

Purpose of the Study:

• To address limitations of the RDoC framework by developing a more accurate model of brain circuitry.
• To propose a data-driven revision of the RDoC framework based on functional neuroimaging data.

Main Methods:

• Utilized a latent variable approach with bifactor analysis on 84 task-based fMRI (tfMRI) activation maps from 6192 participants.
• Employed internal validation with a training/held-out set and external validation using Neurosynth meta-analysis data.
• Compared the fit of a novel bifactor model against the existing RDoC framework.

Main Results:

• A bifactor model, including a task-general domain and a split cognitive systems domain, demonstrated a superior fit to tfMRI data compared to the RDoC framework.
• The arousal and regulatory systems domain was identified as underrepresented within the current RDoC structure.
• Findings provide empirical support for refining the RDoC framework.

Conclusions:

• The current RDoC framework may not optimally capture the underlying brain circuitry.
• A revised RDoC framework, informed by neuroimaging data and bifactor modeling, could enhance specificity and accuracy in psychiatric and neuroscience research.
• Future research should focus on data-driven validation and refinement of the RDoC framework.