Investigation and modeling of neural mechanism of context prediction

2015 Fiscal Year Final Research Report

• Project/Area Number: 26730124
• Research Category: Grant-in-Aid for Young Scientists (B)
• Allocation Type: Multi-year Fund
• Research Field: Intelligent informatics
• Research Institution: Okinawa Institute of Science and Technology Graduate University
• Principal Investigator: FUNAMIZU Akihiro 沖縄科学技術大学院大学, 神経計算ユニット, 研究員 (20724397)
• Research Collaborator: KUHN Bernd 沖縄科学技術大学院大学, 光学ニューロイメージングユニット, 准教授 (90599557)
• Project Period (FY): 2014-04-01 – 2016-03-31
• Keywords: ベイズ推定 / 大脳新皮質 / 二光子顕微鏡 / 頭頂葉 / 強化学習 / 神経科学 / 脳科学 / 計算論

Outline of Final Research Achievements

This study proposed a hypothesis that neocortex employs Bayesian inference to predict current context. Bayesian inference is widely used in robotics and information science for context prediction. This hypothesis was published as a review paper. Neocortex has a six-layered column structure. Our Bayesian hypothesis proposed that cortical layers 1, 2/3, and 5 implement prior, likelihood, and posterior in the Bayesian inference, respectively. Top-down signals from other cortical areas or thalamus send prediction of future state (state transition), while bottom-up signals send sensory inputs to a cortical column. We tested our Bayesian-cortex hypothesis with mice using a behavioral training and two-photon microscopy which imaged population neuronal activity of posterior parietal cortex.

Free Research Field

神経工学