Neural mechanisms underlying flexible behaviour (360G-Wellcome-219525_Z_19_Z)

I aim to understand how the brain represents the relationships between objects and events in the world, and how these representations can be generalised to allow flexible behaviour in new situations. By combining modelling with experiments in human and macaque, I will investigate these computations at both the cellular and systems level. We have developed models that abstract relational knowledge and predict detailed neuronal representations in rodent hippocampal-frontal circuitry. I will ask whether these building blocks can be extended to explain the sophisticated abstractions and inferences commonplace in primate behaviour. In primates, such abstractions compound hierarchically, and can be combined together, allowing deep inferences. To study these computations, I will develop novel tasks that are high-dimensional, yet amenable to mathematical description. I will use state-of-the-art techniques in humans, validated by high density cellular recordings in macaques, to study representations underlying abstraction, generalisation and inference in medial temporal and frontal cortices. In doing so, I will build new bridges between human and animal neuroscience, between biological and artificial intelligence, and new analytical and experimental tools for integrating across scales of neural activity. Developing such tools is important to bridge the precision of modern neuroscience to humans and thence clinical populations.