Learning and
Emergent Computation

How does intelligent behavior sprout out of brain tissue? Our lab studies how computation, cognition and intelligent behavior emerge from large interacting networks—both biological neural circuits and artificial neural networks. We build mechanistic theories of learning, inference, and representation, and connect them to behavior and large-scale neural recordings. 

Why and why now? 

Modern AI can perform impressive feats, but we often don’t understand why it works—or when it will fail. Neuroscience faces a mirror-image challenge: we can measure neural activity at unprecedented scale, but turning that data into explanations is hard. We treat the brain and AI as two views of the same scientific problem: how do high-dimensional networks reliably compute, learn, and generalize?

How we work?

We combine Theory from statistical physics and dynamical systems (to describe collective behavior in large networks) together with Machine learning (to build controllable models of cognition by training networks on tasks) and Data analysis (to test predictions against neural activity and behavior).

Concretely, we use tools like mean-field theory and random-matrix ideas when they help, and we use trained networks as “model organisms” you can probe and manipulate.

Join us
We’re always looking for students who like crossing boundaries—math + code + neuroscience—and who enjoy turning big questions into models you can test. If you’re excited by the idea that “the whole is more than the sum of its parts,” you’ll feel at home here.