Research Areas

My laboratory investigates how neural circuits use supervised learning to make predictions. Our goal is to answer two key questions about the computational algorithms implemented in the brain: (1) how is information about the past used to predict the future, (2) how are error signals used to update and improve these predictions. To provide mechanistic answers to these difficult questions, we study the function of neural circuits in mice trained to predict future events. Our experimental approach allows us to visualize and perturb the activity of cell-specific neuron populations with millisecond precision, by taking advantage of powerful new tools in electrophysiology, two-photon imaging and optogenetics. In addition, we use machine learning and build large-scale computer models and artificial neural networks to reveal general computational principles, test our hypotheses in real-world applications, and suggest novel research directions. Our work has implications for understanding and treating a wide range of motor and cognitive brain disorders that have been linked to deficits in supervised learning and prediction.