PROJECT SUMMARY Relapse is a major hurdle for treating substance use disorder (SUD). Relapse results in part from drug-paired cues exerting strong control over drug-seeking behavior. Indeed, the reactivity to drug-paired cues is predictive of subsequent likelihood of relapse, and this vulnerability does not go away with time; the effects of drug-paired cues over behavior increases progressively with time since drug exposure. Importantly, this is not specific to drug-paired cues. Behavioral and physiological responses to reward-paired cues is also related to likelihood of relapse, and is even correlated with a propensity to develop SUDs in rodent models. This points to an underlying change in learning and behavior to reward-paired cues, which may be exacerbated by drug exposure, to make individuals with SUDs particularly vulnerable to the effect of drug-paired cues over behavior. We have recently discovered that GABAergic neurons in the lateral hypothalamus (LHGABA) neurons are important for learning about reward-paired cues. What’s more is that this region appears to actively oppose learning about non-reward related information. This suggests that LHGABA neurons bias learning towards cues that are directly related to rewards (proximal cues) and away from information that is not directly related to rewards (distal cues). Under normal circumstances this is good. We should pay more attention to information that is directly relevant to what we need right now. However, prior research has shown that drug exposure strengthens function in LH circuits. It is easy to see how this could produce a vulnerability to reward-paired cues seen in SUDs. Increases in LH function seen following drug exposure could potentiate learning and behavior directed to proximal cues (e.g. drug-paired cues), even while trying to remain abstinent, hijacking the adaptive function of LH to focus on cues that help predict rewards that likely promote survival under normal circumstances. We will test a novel reinforcement-learning theory of LH function, and how this function might be strengthened by drug exposure to produce a vulnerability to reward-paired cues. We have developed an innovative task taken from the cognitive neuroscience literature based on human participants. In tandem with optogenetics and fiber photometry of a genetically-encoded calcium sensor, this will allow us to isolate behavior and neural activity during proximal or distal predictors in the same animal. We will also uncover the nature of learning about proximal cues supported by LH function. Our preliminary data shows this involves a sensory-specific representation of reward, which allows LH to influence behavior in flexible and specific ways. Finally, we have found that methamphetamine self-administration enhances behavioral control by proximal cues in a manner that requires a specific reward representation. We will record calcium activity in LHGABA neurons during learning after exposure to drugs to test if th...