Even if the brain sensing headband could collect data the inability to extract artifact would render such data invalid. Secondly, ERPs  have to be timed with a stimuli which is not likely integrated into a consumer-grade EEG hardware.  

Have you tried using BCI2000(downloadable for free within academic institutions) to capture and process the information? When I last used it, it could be configured to work with Neurosky, Emotiv(there are papers that show Emotiv picks up P300 ERPs) and any consumer EEG system via a configuration file.