Abstract
Identification of human emotion from brain activities has been attracted in various research fields such as the neuromarketing. A relationship between brain activity and pleasant and unpleasant emotions has been studied using Near-infrared Spectroscopy (NIRS). In this study, we propose a method to detect the pleasant and unpleasant emotions from measurement of brain activity by NIRS using neural network. To measure brain activity under pleasant and unpleasant emotions, we used the international Affective Picture System (IAPS). The participants were 21 healthy men. Input data for the neural network were the oxy-Hb and its differential values at central part of the frontal lobe. As a result, it was shown that the method can detect emotions(‘pleasant’, ‘unpleasant’ and ‘neutral’) with the accuracy of 79% (the highest) and 65% (average). And it was found from the result that the correct rate of pleasant emotion is especially low. However, other emotions (‘unpleasant’ and ‘neutral’) can be detected with the accuracy of 97% (the highest) and 80% (average). These results show that the method using neural network can be applicable for detecting human emotions.
