Optimizing an Assistive Brain Computer Interface that uses Auditory Attention

Brain Computer Interfaces (BCIs) are a technology that allows for the operation of systems using consciously controllable aspects of brain activity. BCI technology provides new possibilities for methods of assistive communication where no other interventions are possible. A common feature utilized in BCIs is the natural electrophysiological response to salient or “oddball” stimuli. Oddball stimuli evoke very specific patterns in EEG (electroencephalography) and can be used in conjunction with a system where attention to a particular stimuli yields a desired outcome. To date, oddball stimuli have been used for visual-specific tasks, although they have been shown to be equally effective for auditory attention tasks. While visual communication BCI technology is effective, there are a few conditions in which visual attention is neither practical or possible as a mode of evoking a response to a stimulus, such as locked-in syndrome - a condition typically resulting in a being fully conscious and aware without the ability to move or speak. This work uses oddball auditory stimuli to optimize the classification accuracy of an Auditory Brain Computer Interfaces (ABCI) . The ABCI allows individuals to operate a system by directing their attention to one of two auditory streams repeating the phrases “Yes” or “No”. These two audio streams are all played in a dichotic listening format, where the stimulus onset of one stream occurs at the moment the other stimulus is completed. In its most basic format, The ABCIs used in this paradigm allow users to answer yes or no questions by attuning their listening attention to either one auditory stimulus or the other. Current accuracy in intended response classification for ABCIs can be as high as a 77% mean accuracy. It remains to be determined if individuals have the capacity to improve this accuracy, or which factors can be altered to achieve any improvements. This research evaluates the speed of stimuli repetition and its effect on classification accuracy. The outcome of determining optimal stimuli repetition speeds would be finding features of stimulus presentation that yield the highest accuracy in correctly classifying stimulus prompt from user EEG. This study determined optimal parameters for running an ABCI as means of communicating with individuals unable to use other forms of assistive communication technology. The results of this study indicate a need for continued research in making optimal non-invasive alternative communication technologies as well as making progress in finding the potential ceiling in accuracy that an Auditory Brain Computer Interface can have in interpreting brain activity for the intended action of the user.