PROBLEM TO BE SOLVED: To easily estimate a detailed recognition state reflected in an activity of the deep part of the brain using an electroencephalograph, the recognition state being similar to the one obtained by using fMRI.SOLUTION: Function learning processing is executed for acquiring a state mapping function indicating the correspondence between fMRI image data by a functional nuclear magnetic resonance image method and electroencephalogram data acquired from a plurality of electrodes on the head. When the electroencephalogram data for estimation is acquired, a multi-dimensional state space is reconfigured from a time-series pattern of signals of each electrode constituting the acquired electroencephalogram data by the calculation using the state mapping function acquired by the function learning processing, and estimation processing is executed for restoring the information on the activity pattern of the entire brain. The information on the activity pattern of the entire brain acquired in the estimation processing is output.SELECTED DRAWING: Figure 2COPYRIGHT: (C)2018,JPO&INPIT【課題】fMRIを用いた場合と同様の脳の深部の活動に反映される詳細な認知状態を推定することを、脳波計を使って簡単にできるようにする。【解決手段】機能的核磁気共鳴画像法によるfMRI画像データと、頭部の複数の電極から得た脳電図データとの対応を示す状態マッピング関数を取得する関数学習処理を行う。そして、推定用の脳電図データを取得したとき、その取得した脳電図データを、関数学習処理で得た状態マッピング関数を使った演算で、脳電図データを構成する各電極の信号の時系列パターンから多次元の状態空間を再構成し、脳全体の活動パターンの情報を復元する推定処理を行う。この推定処理で得た脳全体の活動パターンの情報を出力する。【選択図】図2
