2022 Fiscal Year Final Research Report
Quantitative modeling of neural representation for spatial cognition under natural visual experiences
| Project/Area Number |
19K12745
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Review Section |
Basic Section 90030:Cognitive science-related
|
|---|
| Research Institution | National Institute of Information and Communications Technology |
|---|
Principal Investigator |
Wada Atsushi 国立研究開発法人情報通信研究機構, 未来ICT研究所脳情報通信融合研究センター, 主任研究員 (10418501)
|
|---|
| Project Period (FY) |
2019-04-01 – 2023-03-31
|
| Keywords | 自然視覚経験 / 空間認識 / 脳内情報表現 / 深層学習 / モデルベース解析 / fMRI / 広視野 / 両眼立体視 |
|---|
| Outline of Final Research Achievements |
Even in a complex environment surrounded by various objects, humans can instantly grasp the spatial arrangement of the environment and themselves from retinal images. To elucidate the human brain mechanism of spatial perception under such natural visual experience, we quantitatively modeled the fMRI visual response to wide-view natural video using the internal representation of FlowNet, a deep neural network (DNN) for 2D motion estimation. The results showed that the deep layers of FlowNet deep could accurately predict neural response in the mid-level dorsal visual areas, which are suggested to involve spatial cognition. These results indicate the effectiveness of our unique approach that combines wide-field stimulation and model-based fMRI analysis.
|
| Free Research Field |
脳機能イメージング，視覚神経科学，認知科学，人工知能
|
| Academic Significance and Societal Importance of the Research Achievements |
空間認識に関わるヒト脳内情報表現の定量的モデル化は，主観的な空間認識状態を脳活動から直接推定する技術に使えるため，没入型VRの臨場感・映像酔い評価や認知症による空間認識障害の発見などの応用に資する．
|
