Valerie Voon - Model-based learning in psychiatry

preview_player
Показать описание

Our decisions are governed by a balance between prospective instrumental goal-oriented and retrospective habitual learning strategies. This arbitration is relevant dimensionally across psychiatric disorders characterized by compulsive behaviours. The balance can be influenced by stress, working memory, cognitive control and neurochemistry such as dopamine and serotonin, features highly relevant to psychiatric disorders. Neural substrates underlying goal-directed and habit learning including conventional overtraining and devaluation and sequential learning tasks overlap with rodent preclinical models. The concept of model-based learning can be extended to include Pavlovian learning processes. Together these findings highlight a role for model-based learning across psychiatric disorders and a potential target for therapeutic intervention.
Рекомендации по теме
Комментарии
Автор

More interesting question: how do you tell the difference between someone who has a fundamental learning impairment or someone who had "developed" an obstruction to learning that can be reversed?

kennethgarcia
Автор

Valerie: This is 6 years old, so I hope the terminology has changed. Just because something is habitual does not mean that it is not goal-directed. It only means that the model being used to drive the behavior is simpler or more linear. The thermodynamic efficiencies of the CNS are such that the brain will contract its use of unnecessary circuits depending on the complexity of the circumstances it must recognize to achieve the highest benefit to cost ratio while reducing risk. If the goal or reward is achieved with a minimum of "decision points" then substantial chunking of instructions can occur resulting in the lack of a more complex model and the operations can be consolidated within the most central regions of the striatum also reducing distances of neural network to traverse for activation to from Stim to response. Serotonin, Dopamine, 5HT all work towards expanding and contracting the representative ensemble of the activated model. I would have been happy to share my model with you of how this all works when I saw you in Zurich last. Your career seems to be going quite well. Continued success.

kennethgarcia