In neuroscience, advances in recording technologies have enabled unprecedented access to large populations of neurons over learning. Simultaneously, the surge of progress in AI has inspired new insight into how artificial neural networks learn. However, we still lack the mathematical tools necessary to develop theoretical principles of learning and computation that are squarely rooted in neural data. Bridging this gap will be essential both to uncover how the brain controls complex behaviours, and to inspire new forms of brain-inspired artificial intelligence.
