Connecting neurobiology to computation
I am interested in what we can learn, and why.
All humans – and mice, too – easily learn some things but are stumped by other problems. What sets these learning propensities? What defines the line between easy and hard, 'natural' and 'unnatural' tasks? The answer determines not only what we can learn but who we are – we are the result of our learning algorithms.
I believe progress in neuroscience will require bridging theory-first and data-first approaches. I actively pursue both research programs.
In my theory work I study AI and artificial neural networks as model systems. Like any learning machine, ANNs extract certain generalizations from data. By studying these learning preferences in this limited context, we can build frameworks that may transfer to explain the mammalian brain.
I also work extensively with neural data. Neuroscience is now in a 'big data' era, especially in regards to single-cell anatomy. I build AI tools to analyze these datasets, working in collaboration with experimental labs obtaining single-cell neuroanatomical (projection and transcription) and functional (2p) data. I seek to discover the cell types and basic functional components of learning and computation. I operate under the hypothesis that the genome encodes canonical rules for cortical learning that modify a genetically-encoded initial scaffold of long-range connectivity.
I did my PhD with Konrad Kording at UPenn and am currently a postdoc with Tony Zador at Cold Spring Harbor Laboratory.