BCI driven by self-induced emotions: a multi-class study

Brain Computer Interfaces (BCIs) use measurements of the voluntary brain activity for driving a communication system, by means of the activation of mental tasks. In recent literature, a novel activation paradigm, based on the self-induction of emotions, has been proposed and some classification strategies for self-induced emotions have been designed, together with a modular framework for the implementation of binary BCIs. We extended the BCI system, to manage the multi-class scenario, in order to increase the number of recognizable commands, thus improving the efficacy of the communication. The objective was to provide a correction function that would allow the increase of the accuracy, without the overhead of a verification method. A poll oriented classification algorithm was used in conjunction with a matrix based graphic interface to allow the user to communicate through three self-induced emotional states: the disgust produced by remembering a bad odor, the good sensation produced by remembering the odor of a good fragrance and a relaxing state. The proposed system was tested on a healthy subject. Preliminary results were reported and discussed.