Evaluation of visual parameters to control a visual ERP-BCI under single-trial classification

Abstract

A brain-computer interface (BCIs) based on event-related potentials (ERPs) is a technology that provides a communication channel between a device and a user through their brain activity. These systems could be used to assist and facilitate decision making in applications such as an air traffic controller (ATC). Thus, this work attempts to be an approximation to determine whether it is possible to detect the stimulus through a single presentation of a stimulus (single-trial classification) and furthermore, to evaluate the ef-fects of the type of stimulus to be detected, or not knowing the position of the stimulus appearance in an ERP-BCI. This experiment has involved six participants in four experimental conditions. Two conditions varied only in the type of stimulus used, faces (a type of stimulus that has shown high per-formance in previous ERP-BCI proposals) versus radar planes; and two con-ditions varied in the prior knowledge of where the stimulus would appear on the screen (knowing vs. not knowing). The results suggest that the use of single-trial classification could be adequate to correctly detect the desired stimulus using and ERP-BCI. In addition, the results reveal no significant effect on either of the two factors. Therefore, it seems that radar planes may be as suitable stimuli as faces and that not knowing the location of the target stimulus is not a significant problem, at least in a standard BCI scenario without distracting stimuli. Therefore, future studies should consider these findings for the design of an ATC using an ERP-BCI for stimulus detection.