Kent Berridge
http://www-personal.umich.edu/~berridge/
Wednesday 13th February 2013
Time: 4pm
Basement Seminar Room
Alexandra House, 17 Queen Square, London, WC1N 3AR
Modeling cue-triggered temptation: new and changing ‘wants’
Computational models of reinforcement learning are important in reward-addiction neuroscience, and have generally presumed motivation values triggered by a reward-related Pavlovian CS to be derived entirely from previous reward-learning experiences. In other words, CS-triggered ‘wants’ are equivalent to learned caches or acquired knowledge of reward. But recent evidence indicates that Pavlovian mesocorticostriatal brain activations also generate unlearned (i.e., novel) ‘wants’ as well as fluctuating ‘wants’ of high intensity. These may especially apply to addiction. In other words, previously learned values are not always the essence of motivation, but often simply one of several input signals transformed by mesocorticostriatal brain systems into ‘wanting’ output at the moment of CS re-encounter. New opportunities for computational modelers to capture mesocorticostriatal reward functions are offered by ‘wants’ that soar in intensity above previously learned values.