I am a research scientist at the McGovern Institute for Brain Research at MIT. My long-term research goal is to uncover the neuronal circuit mechanisms behind sensory-motor behaviors that lead to cognition.
Addressing such a challenging topic requires building good hypotheses, constructing efficient models to formulate these hypotheses, and implementation of advanced technologies required for measurements and manipulations of neural circuits in behaving animals. In my work, I try to combine three lines of inquiry: first, study the behavior in the most ethologically relevant way possible in the laboratory combined with its cellular, circuit and computational aspects. Second, building methods and technologies that enable us to record and manipulate the brain during behavior, and tools for the analysis of complex behavior. Third, developing computational models with testable predictions to provide engineering level descriptions of how the brain learns and generates complex behaviors.
I work at the boundary of experimental neuroscience and machine learning: the same question — what makes a brain able to learn a new sequence of actions — read from recordings in behaving animals and from the models we build to explain them.
That boundary has become most of the work, and more of my time now goes to computation than it used to. Three threads have converged: network models constrained by experimental data, machine-readable structure for the evidence itself, and dynamical-systems methods carried outside biology altogether. The projects below are where that is happening.