Lateralization of Brain During EEG Based Covert Speech Classification

Since time immemorial, researchers have attempted to decode the working of the human brain and tried to replicate its functionality. Brain-computer interfaces based on electroencephalogram (EEG) signals provide accessible and non-invasive means to capture the neuronal activity within the human brain. Towards analyzing the covert (imagined) speech, the brain activity during its production is captured through EEG signals. The ability to accurately interpret covert speech from EEG signals would be advantageous for communication by the patients suffering from speech disabilities or locked-in syndrome. On referring to the cognitive literature, we come across the Wernicke-Geschwind model that localizes the areas in human brain which are active during speech production and perception. In this paper, we present some classification results which are in support of the lateralization of the brain while analyzing the covert speech. For this study, a recently reported database which consists of EEG signals corresponding to pronounced and covert speech for six Spanish commands is used. A comparative study has been done between channels corresponding to the two hemispheres of the brain. The results indicate that there exists a laterality in brain functions and the dominant (left) hemisphere remains more active as compared to the right hemisphere during the covert speech similar to case of normal speech. This fact can be verified based on classification tasks performed on covert speech.