| 研究領域 | 人工知能と脳科学の対照と融合 |
|---|
| 研究課題/領域番号 |
19H04995
|
|---|
| 研究種目 |
新学術領域研究(研究領域提案型)
|
| 配分区分 | 補助金 |
|---|
| 審査区分 |
複合領域
|
|---|
| 研究機関 | 国立研究開発法人理化学研究所 |
|---|
研究代表者 |
ベヌッチ アンドレア 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 (50722352)
|
|---|
| 研究期間 (年度) |
2019-06-28 – 2021-03-31
|
| 研究課題ステータス |
完了 (2020年度)
|
| 配分額 *注記 |
9,880千円 (直接経費: 7,600千円、間接経費: 2,280千円)
2020年度: 4,940千円 (直接経費: 3,800千円、間接経費: 1,140千円)
2019年度: 4,940千円 (直接経費: 3,800千円、間接経費: 1,140千円)
|
|---|
| キーワード | visual psychophysics, / optical imaging / optogenetics / computational modeling / visual cortex / neural circuits / sensory perception / neural networks / behavior / Decision-making / sensory processing / neural computation / cognition / Neural circuits / aNN |
|---|
| 研究開始時の研究の概要 |
In this project we will focus on how developing cortical visual networks learn to encode an ecologically relevant mid-level visual feature: textures.
This will allow us to test hypotheses formulated in mainstream theories of perceptual learning that postulate that as juvenile animals interact with the environment, unsupervised plasticity mechanisms shape the functional connectivity between neurons based on the statistic of the stimuli. However, the rules that link neuronal structural and functional plasticity to the learning process are currently not fully understood.
|
| 研究実績の概要 |
During FY2022, our research focused on sensory processing and sensory-based decision making, encompassing various experimental and theoretical investigations. We concluded three key studies in this period. Dr. Orlandi's study employed locaNMF tensor decomposition to extract choice signals from dorsal-parietal cortical networks during a decision-making task. This investigation revealed pervasive choice signals in these networks, reflecting top-down signals for inference in sensory and sensory-to-decision processes. The study was published in Nature Communications. Dr. Benucci conducted a separate study investigating the impact of motor-related signals in the visual cortex on perceptual stabilization during self-generated movements. Using convolutional neural networks, the study demonstrated that these signals improve categorization performance, training speed, and noise robustness of classifier networks. This work, highlighting the role of self-generated movements in perceptual stability, was published in Plos Comp. Biology. Finally, Dr. Bolanos's Ph.D. research on texture processing in the mouse visual cortex has been accepted for publication pending minor revisions in Nature Communications.
|
| 現在までの達成度 (段落) |
令和2年度が最終年度であるため、記入しない。
|
| 今後の研究の推進方策 |
令和2年度が最終年度であるため、記入しない。
|
