EEG-powered brain computer interfaces have the potential to allow severely disabled people to operate home automation systems, wheelchairs, and other technologies. Yet, anyone who has tried playing games controlled through EEG knows that it’s not easy to get one’s mind in a specific state every time you want to. So it can be a challenge that frustration can only exacerbate, resulting in mental fatigue and poor performance of the system. Peñaloza Sanchez, a PhD student at University of Osaka in Japan, has been working on overcoming this problem by building a learning mechanism into an EEG brain-computer interface.
The system attempts to recognize user patterns and eventually takes short cuts when it notices the person trying to achieve a task done previously. The EEG system works in conjunction with environmental sensors placed around the room the user would be in, feeding info like room temperature, whether doors are open or closed, and if appliances are left on. The system tries to predict what the user wants to do based on earlier recordings, and activates a series of actions such as to navigate a wheelchair to a certain spot or to turn off the lights. In order to prevent frequent mistakes by the over eager system, there are algorithms built in that can identify so-called “error-related negativity,” or the brain’s reaction when it notices a mistake being made. When such a signal is detected during the system’s automatic activity, it’s a sign that it’s making a mistake and the system cancels its current action to try again.