We are using a computational neuroscience approach to identify precise, objective and quantifiable markers of autism spectrum disorders (ASD) in physiological, behavioral and neural processing. To this end, we are combining state-of-the-art techniques, including model-based functional magnetic resonance imaging (fMRI) (Van Horn and Pelphrey, 2015) and eye-tracking. Computational neuroscience can inform the diagnosis and treatment of autism by identifying separate brain networks that are associated with ASD. Our studies emphasize social learning (learning about others) throughout adolescence and in the transition from adolescence to adulthood. The long-term goal of this research is to detect biological predictors of treatment response that help us tailor treatments to optimize individual outcomes.
Using a computational neuroscience approach, we pinpointed abnormalities in functional brain activity of autistic individuals that are present in brain areas closely connected to language processing (Venkataraman et al., 2015).
In a study testing how adolescents predict the preferences of others, we found that adolescents relied on the same principles to update incorrect predictions as did adults. Adolescents, however, made more incorrect predictions and updated prediction errors more slowly. These developmental differences in social learning were reflected in the extent to which brain regions encoded predictions and prediction errors in the two age groups (Rosenblau et al., 2017).
We are currently expanding our research agenda by studying non-social learning and habituation in autistic children. A recent grant by the Simons Foundation will fund efforts to explore the potential of computational models to successfully predict treatment outcomes of autistic children.
Rosenblau G, Korn CW, Pelphrey KA (2017) A computational account of optimizing social predictions reveals that adolescents are conservative learners in social contexts. J Neurosci.
Van Horn JD, Pelphrey KA (2015) Neuroimaging of the developing brain. Brain Imaging Behav 9:1–4 Available at: https://doi.org/10.1007/s11682-015-9365-9.
Venkataraman A, Duncan JS, Yang DYJ, Pelphrey KA (2015) An unbiased Bayesian approach to functional connectomics implicates social-communication networks in autism. NeuroImage Clin.