Project 4 – Neural mechanisms of latent cause inference: the role of orexin signaling The goal of Project 4 is to identify and manipulate neural substrates of experience-dependent latent cause inference to help “unlearn” previously learned associations. This is important because unlearning of maladaptive behavior is critical in many psychotherapies; however, relapse of old learning is common. Here, we will focus on neural activity in the central nucleus of the amygdala (CeA) during exposure to new information regarding rewards as a substrate for affecting whether old information is “overwritten” by new experiences and attempt to modify this activity (and consequent behavioral choices) using the neuropeptide orexin/hypocretin system. Orexins are hypothalamic neuropeptides that regulate arousal and motivation, and play an important role in the formation of memories and in reward-mediated learning. We hypothesize that orexin -- by reactivating reward memory during devaluation or extinction -- can tilt the balance from ascribing new devaluation/extinction information to a separate latent cause (a two-cause model that does not overwrite the old association) to incorporating this information within the old latent cause (a one-cause model). We will use high density unit recordings in CeA to monitor reactivation of a reward memory as an index of an inferred latent cause. We will use orexin antagonism or chemogenetics to test a causal role of orexin signaling in latent cause inference during reward devaluation by LiCl malaise in a Pavolvian paradigm (Aim 4.1), or by extinction of lever pressing for sucrose in an instrumental paradigm (Aim 4.2). Together, these studies will identify circuitry involved in latent cause mechanisms for Pavlovian and instrumental behaviors using parallel approaches: devaluation to challenge existing latent causes (using either LiCl malaise in Aim 4.1, or extinction in Aim 4.2), and recordings from CeA as well as activation of orexin signaling to identify sites of action for modulation of latent causes. We have engaged consultants, Geoff Schoenbaum and Angela Langdon who are experts in computational and animal studies of latent cause inference, to help advise on this Project. This Project uses the Center latent cause inference framework, and specifically, the model developed by Core C. Findings from this Project will suggest new therapeutic approaches to mitigating compulsive behaviors with deleterious consequences (see P2) and reducing relapse of anxiety after extinction-based psychotherapy (see P3). Brain-wide neuroimaging findings from P1, and the findings of P2 and P3, will therefore dovetail with our investigations in P4, and all Projects will benefit from close interactions.