EEG topography recognition by neural network

It is proposed that neural networks can be used to recognize electroencephalographic (EEG) information and so create a new man-machine interface for data input. Existing interfaces require the user to express the data, i.e. the user thinks of a datum and then has to perform either a mechanical action (keypress) or a vocal act. It is shown that just prior to the act of expressing a datum the readiness potential (RP) of the user´s brain contains some information about the datum. A simple test was conducted in which the RPs of one subject were analyzed. For five Japanese syllables, high recognition rates were obtained from a trained neural network