2020 Fiscal Year Annual Research Report
動的かつ不確実な環境下での適応的知覚と行動に関する研究
| Project/Area Number |
19F19809
|
|---|
| Research Institution | Institute of Physical and Chemical Research |
|---|
Principal Investigator |
豊泉 太郎 国立研究開発法人理化学研究所, 脳神経科学研究センター, チームリーダー (50547461)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
BALTIERI MANUEL 国立研究開発法人理化学研究所, 脳神経科学研究セン ター, 外国人特別研究員
|
|---|
| Project Period (FY) |
2019-11-08 – 2022-03-31
|
| Keywords | uncertainty / bayesian inference / closed-loop control |
|---|
| Outline of Annual Research Achievements |
This project tackles the ongoing need for better computational models that can describe more naturalistic behaviour in humans and other agents alike. Via the inclusion of more realistic accounts of uncertainty in closed-loop environments, we propose an interdisciplinary approach to the study of perceptual and motor processes in naturalistic conditions. This approach will integrate computational models including accounts of unpredictable variations in the world and novel experiments focused on dynamic environments.
To achieve the aforementioned goal, I spent some months reviewing the literature to better assess the state of current modelling approaches to perceptual decision making, including for example theories of predictive coding, control theory and variational free energy minimisation. After doing so, I wrote and submitted different papers explaining some of the limitations of a number of current proposals.
|
| Current Status of Research Progress |
Current Status of Research Progress
2: Research has progressed on the whole more than it was originally planned.
Reason
The literature review already produced some published works, and more technical manuscripts are in the working.
|
| Strategy for Future Research Activity |
In the upcoming months, I will continue working on current manuscripts in order to enable a better understanding of the mathematical toolkits necessary to pursue modelling of more realistic experimental setups and their consequences. This will entail establishing clearer and more direct connections to other existing frameworks, with a focus on control theoretic frameworks applied to sensorimotor control.
The outcomes of these projects will be theoretical manuscripts with technical details highlighting different assumptions and situations under which new frameworks ought to be developed, with the goal of modelling of human behaviour and perceptual decision making.
|
