P300 brain-computer interface (BCI) is an experimental and clinical paradigm, where visual evoked potentials (VEPs) triggered by an attended stimulus are used to communicate messages to the outside world. In a typical implementation, called P300 speller, a subject looks at a display where text characters are flashing and attends to one of the characters. The attended character is detected as the one with the strongest VEP. Such a speller performs well when responses to the target and non-target stimuli are sufficiently different and, conversely, more trials are required for reliable selection when non-target stimuli cause sizeable VEPs. Although many strategies for improving the speller have been proposed over the years, a relatively simple one received little attention: reduction of visual field to diminish the contribution of non-target stimuli. To tackle this idea, we ran a pilot experiment in 10 subjects that first operated a traditional P300 speller and then wore an aperture that restricted their vision to the central field. Subjects selected text characters by looking at them. When the aperture was worn, responses to non-target stimuli were reduced in all subjects. Moreover, in 4 subjects, target-stimulus VEP changed in amplitude and/or shape. Since the aperture reduced the interference from non-targets and increased the response to target in some cases, we suggest that this approach be used to improve BCI performance. In addition to the aperture used, we argue that the removal of distractors could be achieved algorithmically instead of using an aperture. Additionally, future P300 BCIs could take an advantage of the different physiological properties of the central and peripheral visual fields. We also discuss how the proposed approach could help elucidate the mechanisms of visual processing.
bioRxiv Subject Collection: Neuroscience