Will Dorrell

University College London

I’m Will, a Ph.D. student at the Gatsby Computational Neuroscience Unit in London, currently working on biological structure learning and neural networks. I was a physics undergraduate at Cambridge University, where I did obnoxious things like running a weekly discussion group. I then moved to Harvard University, initially as a condensed matter physicist working on metamaterials, before seeing the light and becoming a theoretical neuroscientist working with Professor Cengiz Pehlevan on olfaction. Finally, thanks to a surprising burst of foresight, I spent much of the pandemic on the subtropical island of Okinawa working with Professor Erik de Schutter on bio-plausible hierarchical reinforcement learning. I am originally from Worcester, a small city in the West of England, famous for its difficult to pronounce name and eponymous sauce — which is produced less than a mile from my childhood bedroom. In my free time, I enjoy walking, talking and writing small summaries of my achievements.

Principal Investigator: Peter Latham

Fellow: Maria Yuffa

Project
Sensory systems are marvels, turning a sea of noisy inputs into useful knowledge. Olfaction is no exception, enabling navigation and memory recall from only the soup of ambient volatile chemicals. This project will explore consequences of a recent experimental pre-print. Olfaction has long been thought unique among sensory systems due to its lack of structure — higher-order olfactory neurons randomly sample from sensory input neurons. However, recent electron microscopy work has mapped the connectivity of the fly olfactory system and found something previously overlooked: certain input regions are oversampled vs random. This project will use mathematical and computational approaches to ask, ‘what role does this new-found structure play?’ More broadly, findings could be relevant to processing high-dimensional data (previous olfactory research has inspired nearest-neighbour search algorithms). Further, this project could lead to novel, and much needed, uses of connectivity data.

Advancing Research in Basic Science and MathematicsSubscribe to SCGB announcements and other foundation updates