William Dorrell

About

William Dorrell is an incoming research fellow at the Kempner Institute at Harvard University.

Research Focus

William Dorrell is interested in brains’ (and neural networks’) algorithms and how they manifest in neural activity. Towards this end, he builds normative theories which describe, optimally, how a given computation should be reflected in neural firing rates, or in activities in an artificial neural network. These normative theories allow puzzling features of neural behaviour to be understood as optimal, and the use of neural measurements as evidence of a particular computation. Dorrell has used these theories to study prefrontal, entorhinal, and retinal representations. During his fellowship he hopes to use these approaches to understand how the brain can do clever things like reason, tap rhythms, and play board games. More broadly, these tools will hopefully provide a cohesive mathematical framework to understand and infer the implementation of computations in neurons, biological or artificial.