Elliott Ludvig
http://elliot.ludvig.ca/Home.html
Thursday 13th December 2012
Time: 4pm
B10 Basement Seminar Room
Alexandra House, 17 Queen Square, London, WC1N 3AR
Learning from Replayed Experience
In this talk, I introduce a new reinforcement-learning model of conditioning and decision-making that learns in a similar manner from both real and replayed experiences. Our replay model provides a unifying explanation for several classical conditioning phenomena that are problematic for traditional computational models, such as latent inhibition, spontaneous recovery, retrospective revaluation, and timescale invariance. Applied to decision-making, the replay model predicts that memory biases, such as the tendency to remember extremely salient events, should bias choices. In the second part of the talk, I present data from a series of risky choice experiments in people that corroborate this prediction.
Affiliation: Princeton University
More specific affiliation (if needed): Princeton Neuroscience Institute and Department of Mechanical & Aerospace Engineering