Timezone: »

Depression: an RL formulation and a behavioural test
Quentin J Huys · Joshua T Vogelstein · Peter Dayan

Tue Dec 09 07:30 PM -- 12:00 AM (PST) @

abstract Virtually all pharmacological treatments with clinical efficacy in psychiatry have effects on neuromodulatory systems. Reinforcement learning has provided a very detailed and rich ground for the understanding of neuromodulators' effects in normal behaviour. Here, we report a formulation of very specific notions in psychiatry in reinforcement learning terms and a decision making task testing them behaviourally. We choose depression, a mood disorder with intuitive links to affective learning, as a first test case. We argue for a very simple definition of reward sensitivity and behavioural control in a goal-directed setting, design a behavioural decision making task and show that it allows related cognitive constructs to be captured in a parametric and behaviourally specific manner. Finally we show that these tasks allow classification of subjects into healthy and depressed groups based purely on a behavioural measure. We discuss the relevance of this approach to psychitric diagnosis.

Author Information

Quentin J Huys (University of Zurich / ETH Zurich)
Joshua T Vogelstein (Duke University)
Peter Dayan (Gatsby Unit, UCL)

I am Director of the Gatsby Computational Neuroscience Unit at University College London. I studied mathematics at the University of Cambridge and then did a PhD at the University of Edinburgh, specialising in associative memory and reinforcement learning. I did postdocs with Terry Sejnowski at the Salk Institute and Geoff Hinton at the University of Toronto, then became an Assistant Professor in Brain and Cognitive Science at the Massachusetts Institute of Technology before moving to UCL.

More from the Same Authors