Bernardo Sabatini Lab

In the first few years of life, humans tremendously expand their behavioral repertoire and gain the ability to engage in complex, learned, and reward-driven actions. Similarly, within a few weeks after birth mice can perform sophisticated spatial navigation, forage independently for food, and engage in reward reinforcement learning. 

We seek to uncover the mechanisms of synapse and circuit plasticity that permit new behaviors to be learned and refined. We are interested in both the developmental changes that occur after birth that make learning possible as well as in the circuit changes that are triggered by the process of learning.  We examine how perturbations of these processes contribute to neuropsychiatric disorders with disordered learning or action selection.

Understanding learning and computation in the brain requires both experimental and theoretical approaches.  On the experimental side, we rely heavily on optical and electrophysiological methods to both monitor and perturb synapses, neural activity, and plasticity-related intracellular pathways.  We develop new technologies as needed for these studies. On the computation side, we employ a tight back-and-forth between computational neurobiology and machine learning.  We are studying the contribution of dendritic processing to neural computation, the nature of "biologically plausible" learning learns used by the brain, and  features of computation and learning in the brain that may have the capacity to enhance the power of artificial neural networks.

The lab is supported by the Howard Hughes Medical Institute and is comprised of an experimental group in the Department of Neurobiology at Harvard Medical School and a computational/machine learning group at the Kempner Institute for the Study of Natural and Artificial Intelligence at Harvard University.