Back

Low-Frequency Textured Gabor Flicker Enhances SSVEP Entrainment and Visual Comfort for BCI Control

GOMEL, J.; Dehais, F.; Corsi, M.-C.

2025-12-15 neuroscience
10.64898/2025.12.12.693875 bioRxiv
Show abstract

Steady-state visual evoked potentials (SSVEPs) are widely used in cognitive neuroscience and brain-computer interfaces (BCIs), but the visual discomfort induced by repetitive luminance flicker limits their usability, particularly in multi-target settings due to strong peripheral distraction. Textured flicker composed of Gabor patches has been proposed as a more comfortable alternative, but its suitability for SSVEP paradigms and its frequency-dependent impact on neural entrainment remain unclear. Here, we directly compared textured Gabor-based flicker and classical luminance flicker using a frequency sweep followed by a multi-class SSVEP BCI task. In Session 1 (N=24), we measured SSVEP signal-to-noise ratio (SNR), inter-trial coherence (ITC), and subjective comfort across 13 stimulation frequencies (3-18Hz). Gabor-based textures elicited higher SNR and ITC than plain flicker at low frequencies (3-9Hz), whereas plain flicker produced stronger and more phase-consistent responses at higher frequencies (12-18Hz), revealing a robust crossover in entrainment. Across almost all frequencies, Gabor stimuli were rated as more comfortable. Based on these results, we defined a low-frequency Gabor-optimized band (5-7Hz) and a higher plain-optimized band (14-16Hz). In Session 2 (N=18), these bands were used in a five-class offline SSVEP BCI. Classification accuracy was highest (Gabor: 95.7% at 5-7Hz; plain: 98.1% at 14-16Hz) when each stimulus type was used in its optimal band and decreased markedly when stimulus type and frequency band were mismatched. Gabor stimuli were consistently rated as more comfortable and nearly imperceptible in peripheral vision. Together, these findings establish textured Gabor flicker as a comfortable and effective alternative to luminance flicker for low-frequency SSVEP paradigms.

Published in Biocybernetics and Biomedical Engineering · not in our set (fewer than 10 published preprints to learn from) · training set

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.