Sweet and sour taste classification using EEG based brain computer interface

This work presents an EEG based brain computer interface for differentiating between sweet and sour tastes. For this purpose, eight channels EEG was recorded from ten healthy subjects when they hold the tastants in their mouth. Different features extracted from the signals were kurtosis, skewness, energy and wavelet entropy. Extracted features are classified using a linear discriminant classifier. The results show that energy and wavelet entropy were able to classify the tastes with greater than 98% accuracy while the other two features barely gives the 60% accuracy. Analysis was also carried out to evaluate the best time interval after the stimulus was given. It was found that the best discriminatory response in the EEG signal based on the extracted features was between 20-30 s after the stimulus.