| Presentation | 2012-06-28 A Computational Model for State Value Estimation in the Brain Using Two Kinds of Temporal Differences Masanari SHINOTSUKA, Masahiko MORITA, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | The basal ganglia has been considered to be deeply involved in reinforcement learning since Schultz et al. reported that the midbrain dopamine neurons show activities in accordance with temporal differences (TD). However, existing TD-learning models of the basal ganglia seem to assume some implausible neural mechanisms. We constructed a more plausible model for state value computation by introducing two kinds of TD signals: prospective and actual. Computer simulations on a simple reward prediction task showed that the prospective TD signal varies frequently depending on the short-term history of actions and rewards, and that longer-term state values can be acquired based on the actual TD signal which follows the prospective one with a smaller variation. This model leads to the hypothesis that these two TD signals may be encoded by different dopamine systems in the midbrain. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | reinforcement learning / temporal difference / state value function / basal ganglia / dopamine neurons |
| Paper # | NC2012-5 |
| Date of Issue |
| Conference Information | |
| Committee | NC |
|---|---|
| Conference Date | 2012/6/21(1days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | |
| Chair | |
| Vice Chair | |
| Secretary | |
| Assistant | |
| 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 Computational Model for State Value Estimation in the Brain Using Two Kinds of Temporal Differences |
| Sub Title (in English) | |
| Keyword(1) | reinforcement learning |
| Keyword(2) | temporal difference |
| Keyword(3) | state value function |
| Keyword(4) | basal ganglia |
| Keyword(5) | dopamine neurons |
| 1st Author's Name | Masanari SHINOTSUKA |
| 1st Author's Affiliation | Graduate School of Systems and Information Engineering, University of Tsukuba() |
| 2nd Author's Name | Masahiko MORITA |
| 2nd Author's Affiliation | Graduate School of Systems and Information Engineering, University of Tsukuba |
| Date | 2012-06-28 |
| Paper # | NC2012-5 |
| Volume (vol) | vol.112 |
| Number (no) | 108 |
| Page | pp.pp.- |
| #Pages | 6 |
| Date of Issue | |