It remains a substantial challenge to model psychopathology using neuroimaging data. High rates of comorbidity and heterogeneity between and within psychiatric disorders make predictive modeling especially challenging. To address this challenge, we model psychopathology using the well-established P-factor. Furthermore, psychiatric disorders do not, themselves, reflect singular neurobiological processes. Therefore, we attempt to improve our ability to predict psychopathology by transferring knowledge from individual models trained to predict elements of executive function, cognitive processes associated with psychopathology, including working-memory, set-shifting, and inhibitory control.