Presentation 1997/5/23
A Theory on Positive and Negative Reinforcement in Learning and Action Selection
Yasuhiro SAGAWA,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Midbrain dopamine (DA) neurons are thought to have effects on processes of learning and action selection. Neural network models have been proposed, which reproduce responses of DA neurons and consider these responses as reinforcement signals for learning and action selection. In this article, the reinforcement signal considered not only in these models but also in some other models of classical conditioning is divided into positive and negative reinforcement. Then, the relationship between quantities of these opposite reinforcements is lead theoretically. The theory means that an amount of positive reinforcement and that of negative one are equivalent during periods which satisfy a certain boundary condition.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) learning / action selection / positive reinforcement / negative reinforcement / dopamine
Paper # NC97-9
Date of Issue

Conference Information
Committee NC
Conference Date 1997/5/23(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Vice Chair

Paper Information
Registration To Neurocomputing (NC)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) A Theory on Positive and Negative Reinforcement in Learning and Action Selection
Sub Title (in English)
Keyword(1) learning
Keyword(2) action selection
Keyword(3) positive reinforcement
Keyword(4) negative reinforcement
Keyword(5) dopamine
1st Author's Name Yasuhiro SAGAWA
1st Author's Affiliation Department of Biophysical Engineering, Faculty of Engineering Science, Osaka University()
Date 1997/5/23
Paper # NC97-9
Volume (vol) vol.97
Number (no) 69
Page pp.pp.-
#Pages 6
Date of Issue