Impact of Speller Size on a Visual P300 Brain-Computer Interface (BCI)‎ System under Two Conditions of Constraint for Eye Movement

Abstract EN

The vast majority of P300-based brain-computer interface (BCI) systems are based on the well-known P300 speller presented by Farwell and Donchin for communication purposes and an alternative to people with neuromuscular disabilities, such as impaired eye movement.

The purpose of the present work is to study the effect of speller size on P300-based BCI usability, measured in terms of effectiveness, efficiency, and satisfaction under overt and covert attention conditions.

To this end, twelve participants used three speller sizes under both attentional conditions to spell 12 symbols.

The results indicated that the speller size had, in both attentional conditions, a significant influence on performance.

In both conditions (covert and overt), the best performances were obtained with the small and medium speller sizes, both being the most effective.

The speller size did not significantly affect workload on the three speller sizes.

In contrast, covert attention condition produced very high workload due to the increased resources expended to complete the task.

Regarding users’ preferences, significant differences were obtained between speller sizes.

The small speller size was considered as the most complex, the most stressful, the less comfortable, and the most tiring.

The medium speller size was always considered in the medium rank, which is the speller size that was evaluated less frequently and, for each dimension, the worst one.

In this sense, the medium and the large speller sizes were considered as the most satisfactory.

Finally, the medium speller size was the one to which the three standard dimensions were collected: high effectiveness, high efficiency, and high satisfaction.

This work demonstrates that the speller size is an important parameter to consider in improving the usability of P300 BCI for communication purposes.

The obtained results showed that using the proposed medium speller size, performance and satisfaction could be improved.