Each of our senses feeds the brain with information from millions of receptor cells changing their activities at near millisecond resolution in every waking moment of our life. Discovering statistical structure in this massive high-dimensional sensory input (i.e. statistical learning) is a formidable challenge that brains have evolved to solve in a seemingly highly efficient manner. Statistical learning in the brain spans an enormous range of complexity, from associating two frequently co-occurring tastes to developing object representations that allow us to parse cluttered visual scenes without supervision and to naturally classifying a novel piece of music as belonging to a particular style based on experience with a variety of music over time. In recent years neuroscience has made great strides in characterizing the basic cellular and molecular plasticity mechanisms underpinning the learning of simple pairwise associations (i.e. second-order correlations). Both the conceptual frameworks and the experimental approaches that underlie these studies, however, fail to extend, by a large margin, to the ecologically relevant spatial and temporal scales of higher-order correlations. Major insights are therefore necessary to lay the theoretical foundations as well as to develop appropriate experimental approaches (behavioral paradigms and neural recording strategies) to study statistical learning in the brain at scale. The goal of our program will be to ignite these advances by bringing together researchers from Theoretical Physics, Computer Science, Cognitive Science / Psychology and Neurobiology. The specific aims of the program are:

•To identify new theoretical frameworks for statistical learning and experimental strategies to test these theories

•To link neurobiological plasticity mechanisms directly to behaviorally and computationally-defined statistical learning

•To lay out a roadmap for future developments, both in theory and in experiments, that will be critical for making progress in understanding the neurobiology of statistical learning