Dopaminergic computations for learning to make good decisions

Abstract :

We investigate neuronal circuit computations that enable us to learn and make good decisions, in particular decisions based on sensory stimuli. In the first part of this talk, I will present results from experiments in which we examined dopamine signals while mice learned to perform a sensory decision task from naïve to expert. The results provide two insights: first, the progression of behaviour within and across individuals follow dynamics of a deep reinforcement learning network that learns from multiple teaching signals. Second, dopamine signals in the dorsal striatum encode one of the network’s teaching signals. In the second part of the talk, I will touch on our ongoing, unpublished projects investigating neuronal computations underlying sensory-guided learning across dopaminergic circuits spanning the midbrain, striatum and frontal cortex.

Bio:

Armin Lak is Associate Professor of Neuroscience at the Department of Physiology, Anatomy and Genetics, University of Oxford, and a Wellcome Trust Henry Dale Fellow at St John's College. His research focuses on the neural circuits underlying learning and decision-making, with a particular emphasis on how sensory inputs and internal reward signals shape choices under uncertainty. Using electrophysiology, multi-photon imaging, optogenetics, and computational approaches in mice, his lab has made key contributions to understanding frontostriatal circuits, the function of the orbitofrontal cortex, and the influence of dopamine on perceptual and economic decisions.

Website: https://www.laklab.org/