Conte Center – Understanding Latent Cause Inference in Health and Illness Latent cause inference is the fundamental cognitive process by which we draw boundaries between situations that have different underlying rules, and therefore require separate learning and decision making. Every piece of information that enters our brain is first categorized as pertaining to something we already know about (an old latent cause) or something completely new (a new latent cause). Because latent cause inference is at the heart of perception, learning, evaluation, and action selection, alterations in this process may be at the core of a wide variety of mental health conditions. Following a computational psychiatry approach that bridges abstract symptoms to basic neurocognitive mechanisms using computational models, we propose a comprehensive investigation of latent cause inference as a proposed new transdiagnostic NIMH Research Domain Criteria (RDoC) domain. We propose a research program centered around three themes: quantifying individual differences in latent cause inference and relating them to mental health symptoms, testing for alterations of this process in clinical samples, and delineating the neural circuitry underlying latent cause inference. These themes cut across four proposed projects (P1-P4): P1 – Latent cause inference as a fundamental cognitive process will relate individual differences in latent cause inference to dimensional mental health symptom factors, and identify the neural circuitry involved in creating and reusing latent causes. P2 – Latent cause inference in compulsion will test a novel interpretation of compulsive disorders (e.g., obsessive compulsive disorder, drug misuse) as arising from over-splitting of latent causes. P3 – Latent cause inference in anxiety will investigate the interaction of latent cause inference with memory retrieval and updating, and their failure modes in anxiety disorders. P4 – Neural mechanisms underlying latent cause inference will probe neural mechanisms of latent cause inference using high-density neuronal recordings and chemogenetic manipulation of orexin in rat amygdala. The proposed projects are tightly interconnected, all employing a shared theoretical framework and computational modeling approach and relying on integration of methods and results to obtain a comprehensive understanding of latent cause inference in mental illness and in health. They will build on three Research Cores (Behavioral Testing and Clinical Assessment, Computational Modeling, and Neuroimaging). The investigators have a long and productive history of collaboration, and each is organically involved in multiple Projects and/or Cores, ensuring optimal integration of knowledge and methods. The Conte Center will allow us to conduct a multidimensional and translational characterization of a new candidate cognitive process for the RDoC matrix, tying it to mental illnesses and uncovering its neural mechanisms.

A variety of mental health conditions can be considered disorders of generalization, e.g., overgeneralization in anxiety and compulsion, and undergeneralization in schizophrenia. Latent cause inference is a fundamental cognitive process that lies at the heart of how we set generalization boundaries in perception, learning and decision making. We propose a comprehensive characterization of individual differences in latent cause inference and their relationship to self-reported symptom dimensions and clinical categories, as well as their underlying neural circuitry, proposing latent cause inference as a new RDoC dimension.