Effects of Stimulus Frequency on Steady-State Visual Evoked Potential-Based Brain-Computer Interfaces

Abstract

The stimulus frequency is one of the essential parameters of visual stimuli in steady-state visual evoked potentials (SSVEPs). This study aimed to explore the effects of stimulus frequency on SSVEPs by using the public dataset benchmark. After feature extraction, the SSVEP response intensity was different for the commonly used stimulus frequencies (8∼15.8 Hz), with the relatively lower frequencies of 8∼11 Hz evoking a higher intensity, and the relatively higher frequencies of 14∼16 Hz evoking a lower intensity. Then, the five stimulus frequencies (8.8, 9.8, 8.6, 8.2, and 9.2 Hz) with the highest response intensity and five stimulus frequencies (15.8, 15.6, 15, 15.4, and 15.2 Hz) with the lowest response intensity were selected as the highly sensitive and lowly sensitive frequencies. After target identification by standard canonical correlation analysis (standard CCA) and Filter bank CCA (FBCCA), both the classification accuracy and information transfer rate (ITR) showed a relatively better performance in highly sensitive frequencies than lowly sensitive frequencies. This study proved that there were effects of stimulus frequency on SSVEPs, recommending the stimulus frequencies of 8∼11 Hz in SSVEPs, especially when the number of targets is not so many.