Peter Dayan

Peter Dayan
Royal Society 2018
EducationUniversity of Cambridge (BA)
University of Edinburgh (PhD)
Known forproposing dopamine for reward prediction error, and developing Q-learning
SpouseLi Zhaoping
AwardsRumelhart Prize (2012)
The Brain Prize (2017)
Scientific career
FieldsComputational neuroscience
Reinforcement learning
WorkplacesMax Planck Institute for Biological Cybernetics
University College London
Massachusetts Institute of Technology
Uber
University of Toronto
Salk Institute
ThesisReinforcing connectionism : learning the statistical way (1991)
David Willshaw
Websitewww.kyb.tuebingen.mpg.de/person/95844/251691

Peter Dayan FRS is a British neuroscientist and computer scientist who is director at the Max Planck Institute for Biological Cybernetics in Tübingen, Germany. He has pioneered the field of reinforcement learning (RL) where he and his colleagues proposed that dopamine signals reward prediction error, and helped develop the Q-learning algorithm. He is co-author of Theoretical Neuroscience, an influential textbook on computational neuroscience. He is also known for applying Bayesian methods from machine learning and artificial intelligence to understand neural function, and is particularly recognized for relating neurotransmitter levels to prediction errors and Bayesian uncertainties. He made contributions to unsupervised learning, including the wake-sleep algorithm for neural networks and the Helmholtz machine.