Neuroscience and AI

Biological brains remain the most capable and energy-efficient intelligences we know of. Our Neuro-AI research asks what principles from neuroscience — sparse coding, predictive processing, plasticity and neuromodulation — can make artificial systems more robust and efficient.

We study the two-way street between the fields: using AI to model and interpret neural data, and using neuroscience to inspire new architectures that learn continuously rather than freezing after training.

This work underpins our longer-term AI Brain programme.