Modeling multidisciplinary data in developmental neuroscience: A structured approach to integrating brain, behavior, and environmental influences

Large-scale developmental studies increasingly collect data across multiple disciplines such as neuroscience, psychology, and genetics. Combining these data offers new opportunities for understanding human development, but it also presents substantial analytical challenges. Traditional statistical approaches often struggle to handle the complexity of large numbers of diverse measurements. To address these challenges, we propose a new method for multidisciplinary data called Stacked Domain Learning (SDL). In SDL, theories from different disciplines are translated into statistical models and these are then integrated into a single overarching model. This allows researchers to evaluate the unique and combined contributions of different disciplinary perspectives, providing a foundation for developing integrative theories grounded in both theory and empirical evidence. We demonstrate SDL with an empirical example using data from the ABCD ® Study to predict delay discounting behavior in adolescents. The results highlight the relevance of cognitive abilities and sociodemographic factors and only partially support the role of brain systems in predicting delay discounting. The application to the ABCD data demonstrates that assessing the predictive ability of theories through statistical models in SDL helps to evaluate the relevance and usefulness of these theories. Altogether, SDL offers a practical tool for integrating different disciplinary perspectives into a single framework and evaluating the predictive ability of the theories involved.