Visual comfort has long been an open problem in SSVEP research, and a textured flicker BCI offers a new way to tackle it. Classic luminance flicker drives strong, reliable brain responses. But staring at flashing white disks gets tiring fast, and it only gets worse when several targets flicker at once. A new study tests whether textured Gabor patches can match plain flicker’s signal while being more comfortable to watch.
Plain disk versus Gabor texture: how the study was set up
The researchers first ran a frequency-sweep session with 24 healthy adults. A second session followed two months later, built around a five-target BCI task. The analysis included 18 of the original participants. Both sessions used mbt’s Smarting PRO EEG system, a wireless amplifier paired with a semi-dry 8-channel cap. The team placed the electrodes over the occipital and parieto-occipital areas to efficiently capture visually evoked potentials.
Participants sat about 70 cm from a 60 Hz monitor in a dimly lit room, with no Faraday cage. Stimuli ran in PsychoPy, and the EEG was sampled at 500 Hz. The two were synchronized over LSL.
Two stimuli were compared. The first was a uniform flickering disk (plain flicker), and the second was a disk filled with 75 randomly oriented Gabor patches (textured flicker). Both flickered sinusoidally against a black screen. The only real difference was spatial texture, not overall brightness. The image below shows both stimuli across one flicker cycle, with a close-up of a single Gabor patch.
Plain disks (top) versus Gabor-textured disks (bottom), the two stimulus types behind this textured flicker BCI comparison, shown across one flicker cycle. The zoomed inset reveals that each “grain” of the Gabor texture is a small oriented grating, not a random speckle.A plain disk simply changes brightness. The Gabor texture is designed to engage contrast-sensitive retinal cells and orientation-selective cortical neurons, which is a different route into visual cortex.
A frequency sweep to find what works, a BCI to test it
In the first session (frequency-sweep), participants fixated a single flickering disk. The sweep had 13 blocks, one for each frequency, presented in random order. The frequencies ran from 3 to 12 Hz in 1 Hz steps, followed by 14, 16 and 18 Hz. The whole sweep ran twice, once for each texture. After every block, participants rated their comfort and mental fatigue from 0 to 10.
In the second session, participants used a five-target textured flicker BCI. Five disks flickered on screen, each at its own frequency, and participants looked at a cued sequence of digits, like entering a PIN. Each texture ran in the frequency band where it had done best in the first session, and the EEG was classified offline afterwards.
The image shows what a single trial looked like. A red square first cued the target, then the disks flickered for 2.2 seconds, followed by a short rest. The top row is the frequency sweep with one disk, and the bottom row is the BCI task, where all five numbered disks flickered together and participants looked at the cued one.Where does a textured flicker BCI actually win?
For a textured flicker BCI, the answer comes down to frequency. To compare the two designs, the researchers measured the signal-to-noise ratio (SNR) of the SSVEP, the brain’s EEG response to the flicker. SNR shows how strongly the brain responds at the flicker frequency compared with the background activity around it. The chart below plots SNR across all 13 frequencies.
SSVEP signal-to-noise ratio across the 3–18 Hz sweep. Gabor flicker (blue) peaks sharply in the 3–9 Hz range, while plain flicker (red) peaks near 16 Hz and takes over above 10 Hz.Between 3 and 9 Hz, textured flicker produced a much stronger brain response than plain flicker, peaking around 6 Hz. Above 10 Hz, plain flicker took over and stayed ahead up to 18 Hz, with its own peak near 16 Hz. The two curves cross at about 10 Hz.
Strength alone is not enough for a textured flicker BCI, because the response also needs to be reliable. The team therefore measured inter-trial coherence (ITC), which shows how repeatable the response is from one trial to the next. A high ITC means the brain responds with the same timing from trial to trial.
SSVEP inter-trial coherence across the 3–18 Hz sweep. It follows the SNR pattern, with Gabor flicker (blue) giving a more stable response in the 3–9 Hz range and plain flicker (red) above 10 Hz.Comfort tells a much simpler story
Unlike the brain response, participants’ ratings of visual comfort and mental fatigue did not flip with frequency. Gabor flicker was rated better than plain flicker on both measures across the sweep. Comfort ratings for plain flicker dropped steadily toward 16 Hz, while Gabor ratings stayed high. Fatigue followed a similar pattern with a smaller difference.
Visual comfort (left) and mental fatigue (right) across the frequency sweep. The Gabor line (blue) stays flat and high throughout, while the plain-flicker line (red) drifts steadily downward as frequency rises.The five-target textured flicker BCI task of the second session repeated these ratings and added peripheral distraction. The results told the same story.
Subjective ratings from the five-target BCI task: visual comfort (A), mental fatigue (B), and peripheral distraction (C). Gabor stimuli (blue) cluster toward the favorable end of every scale, in both the low- and high-frequency bands.Gabor flicker was rated more comfortable and less distracting in both frequency bands. So a textured flicker BCI keeps its comfort advantage when several targets flicker at once.
Building a textured flicker BCI: two frequency sweet spots
The real test of a textured flicker BCI came in the second session, where classification accuracy was measured in a five-target task. The results mirrored the sweep.
Classification accuracy in the five-target BCI, by texture and frequency band. Each texture performs best in its own band and loses reliability outside it.In the 5–7 Hz band, Gabor flicker reached very high accuracy, and scores stayed consistent across participants. Plain flicker in that same band was noticeably less reliable and more variable from person to person. In the 14–16 Hz band the pattern swapped, and plain flicker became the dependable choice. Both stimuli did well in their own band, so a textured flicker BCI mainly wins on user experience.
Conclusion
A low-frequency textured flicker BCI is more than a gentler alternative to plain flicker. Between 3 and 9 Hz, Gabor flicker gave a stronger, more stable brain response, was more comfortable at every frequency, and reached accuracy comparable to plain flicker in its own band. Plain flicker still has the edge above 10 Hz, so the best stimulus depends on the frequency range a BCI needs.
Read the full study here: Low-frequency textured Gabor flicker enhances neural entrainment and visual comfort for brain-computer interface control
