Exploration of Visual Stimuli
Within our Analogue Evolutionary Brain-Computer Interfaces EPSRC
Project (EP/F033818/1) we explored a number of protocols for visual stimulation to be used for the purpose of generating P300 waves in our analogue BCI mice.
We started with testing variations of the traditional oddball protocol (which involves the presentation of rare target stimuli intermixed randomly within sequences of frequent non-target stimuli), which we had used successfully in previous research. We tested the standard flashing stimuli, stimuli that moved and stimuli that changed color. We also tested two stimulus presentation rates (100ms and 200ms). However, results
indicated that luminosity changes, the standard method of
stimulation used in visual P300 BCI protocols, did provided the best
performance of the variations we tested.
We then turned our attention to a completely novel idea: turning the sequence of stimulus presentation used in our protocols from random to periodic. This was a big risk, since the psychophysiology literature on event-related potentials, and P300 waves in particular, was clear on this: P300 are elicited by rare and unexpected target stimuli to which subjects pay active attention. So, in principle periodic sequences should not work. While it was true that some of the combinations of stimulation sequence and user task we attempted were disappointing, we found one (where periodic stimulation sequences are combined with a colour naming task) which produced far superior results than anything we had used before, eliciting very robust P300s on a large number of electrodes on the scalp, which make P300s much more easily recognisable by our analogue mouse software. The net result was a significant improvement in information transfer rate from the user's mind to the computer.