During any behavior, many thousands or millions of neurons in the brain change their activity.  To undertand how the brain controls behaviors, it is important to identify the computational steps required, and determine which neural circuits perform the computations. The goal of our lab is to understand the building blocks of computation in neural circuits using, as a model, sensory-guided behaviors in mice.  In this system we can apply optical, genetic, and electrophysiological tools during behavior, and combine them with theoretical approaches for understanding circuits and networks. Determining the computational roles of neural circuits in sensory-guided behavior can shed light on neural control of more complex behaviors, and a mechanistic understanding of neural computation can provide rational treatment targets for deficits in mental disease.  Recent work has examined the function of recurrent cortical connectivity, finding that a general property of several cortical areas is strong synaptic coupling and inhibitory-stabilized network (ISN) operation, and used two-photon imaging to understand cortical changes during behavioral learning.  Work in the laboratory is funded by the NIH Intramural program and the BRAIN Initiative.