Summary Autism spectrum disorder (ASD) has diverse presentation but can be characterized by at core by a) rigid and repetitive behavior and b) social communication deficits. In recent decades, there is increasing confidence that identified genetic differences contribute to ASD in humans and a number of high confidence risk genes have been identified. These risk genes can be studied in mice. There is hope that convergent phenotypes and endophenotypes will illuminate key features of ASD (Hyman, 2014). One striking current area of convergence seen in mice with ASD risk genes is a gain of function in rotarod motor learning and alteration in neurotransmission onto spiny projection neurons (SPNs) of the striatum (Hyman, 2014). This was first observed in mice with neuroligin gene mutations (Rothwell et al., 2014), but has also been observed in mice with Tsc1 and Tsc2 gene mutations (Benthall et al., 2021). Tsc2 (with some studies of Tsc1 for comparison) will be the central focus of this proposal. Here we propose that a major diagnostic criteria for ASD observed in TSC patients - restricted, repetitive patterns of behavior - is mediated by changes in the activity of basal ganglia circuits that control the learning and updating of appropriate actions. Specifically, we hypothesize that Tsc2 haploinsufficiency leads to changes in corticalstriatal synapses and SPN activity that facilities striatal dependent learning and makes updating learning more inflexible. In Aim 1 we will use electrophysiological and structural imaging methods to test if specific connections are stronger in Tsc2 Het mice than WT. We posit based on studies in the dorsolateral striatum that corticostriatal connections onto D1R expressing SPNs (dSPNs) will be enhanced in the dorsomedial striatum (DMS). Aim 2 will focus on behavior and test in two different tasks if behavioral inflexibility in Tsc2 Het mice can be ameliorated by reducing reward probability during learning. Aim 3 will use in vivo imaging in the striatum to examine how dopamine and striatal activity may differ in Tsc2 mice under conditions that produce behavioral differences. In sum, these data will inform basic neurobiology surrounding a convergent gain of function phenotype seen in many ASD models: gain of function in striatal learning (rotarod being the most common test). We will translate this learning phenotype into more translatable behavior learning paradigm--cue guided action learning and seek the neural correlates of this gain of function. Finally, we will test a highly translatable therapeutic idea, the idea that using lower reward probability during training will ameliorate neural differences and allow for greater flexibility later.