A novel EEG based linguistic BCI

While a human being can think coherently, physical limitations no matter how severe, should never become disabling. Thinking and cognition are performed and expressed through language, which is the most natural form of human communication. The use of covert speech tasks for BCIs has been successfully achieved for invasive and non-invasive systems. In this work, by incorporating the most recent discoveries on the spatial, temporal, and spectral signatures of word production, a novel system is designed, which is custom-build for linguistic tasks. Other than paying attention and waiting for the onset cue, this BCI requires absolutely no cognitive effort from the user and operates using automatic linguistic functions of the brain in the first 312ms post onset, which is also completely out of the control of the user and immune from inconsistencies. With four classes, this online BCI achieves classification accuracy of 82.5%. Each word produces a signature as unique as its phonetic structure, and the number of covert speech tasks used in this work is limited by computational power. We demonstrated that this BCI can successfully use wireless dry electrode EEG systems, which are becoming as capable as traditional laboratory grade systems. This frees the potential user from the confounds of the lab, facilitating real-world application. Considering that the number of words used in daily life does not exceed 2000, the number of words used by this type of novel BCI may indeed reach this number in the future, with no need to change the current system design or experimental protocol. As a promising step towards noninvasive synthetic telepathy, this system has the potential to not only help those in desperate need, but to completely change the way we communicate with our computers in the future as covert speech is much easier than any form of manual communication and control.