PROJECT SUMMARY This exploratory project between CNH and Georgetown University leverages the DC-IDDRC infrastructure and its Neuroimaging, Neurobehavioral and Clinical Translational Cores to test mechanistic hypotheses about individual differences in the ability to transfer learned knowledge to novel settings in Autism Spectrum Disorders (ASD). The history of ASD intervention is rife with poor real-world outcomes and high heterogeneity in generalization success (1). One known contributing factor is executive dysfunction, particularly behavioral inflexibility (2, 3). Understanding this nexus of learning and executive function (4) likely holds the key to resolving the generalization challenge in ASD, but it has received little attention. The proposed project aims to elucidate the association between learning and flexibility by testing whether intervening to promote flexible behavior in ASD changes learning and associated neural mechanisms. The scientific premise of the proposed study is that flexible use of learned concepts depends on generating prototypes, whereas learning tuned to individual exemplars promotes specificity (5, 6). Current models of concept learning (7) have used computational modeling of individual generalization performance and model-based functional magnetic resonance imaging (m-fMRI) to attribute prototype-generation to ventral medial prefrontal cortex (vmPFC) and exemplar-biased learning to the medial temporal lobes (MTL) (8). We propose that variability in prototype/exemplar learning mechanisms is associated with behavioral flexibility and explains differences in adaptive and treatment outcomes. We employ a longitudinal case-controlled design in 54 14-18 year old youth with ASD at 3 time-points 8 months apart, each including m-fMRI during category learning and behavioral measurement of executive and adaptive function. Aim 1 tests the hypothesis that individual variation in learning biases (prototype/exemplar) and their neural correlates predicts behavioral flexibility (Time1) and is stable over time (Time2). Aim 2 tests plasticity of learning mechanisms induced by a cognitive-behavioral intervention for flexibility (Unstuck-and-On-Target) that targets development of prototypical knowledge (9). Intervention will strengthen prototype learning, and associated vmPFC involvement will be associated with better behavioral response to intervention. Aim 3 tests hypothesis about intervention-induced plasticity of intrinsic functional connectivity. Stronger resting- state functional connectivity between MTL and vmPFC specifically and network connectivity of the MTL subsystem of the default mode network (10) will be associated with prototype learning and intervention response. Our approach is novel, methodologically in the use of individualized characterization of learning mechanisms, and theoretically in unifying learning and executive function to explain mechanisms of treatment response and heterogeneity in treatment outcome in ASD. Fi...