A steady-state visual evoked potential (SSVEP) is the visual cortex's response to a repetitive visual stimulus. It appears in the EEG over occipital electrodes (O1, Oz, O2 and surrounding sites) as a peak at the stimulation frequency and its harmonics. Each target on a screen flickers at its own frequency, or frequency-and-phase pair, so detecting the dominant frequency identifies the target.
SSVEP BCIs have the highest information transfer rates of the non-invasive paradigms, need almost no user training, and work for most people. The costs are visual fatigue and a dependence on gaze, since the user has to look at the target.
Stimulus design decides the ceiling
Frequency range
Low-to-mid frequencies (roughly 6–20 Hz) give strong responses. Higher frequencies are more comfortable but weaker.
Monitor refresh
On a screen, the achievable frequencies depend on the refresh rate. Sampled sinusoidal coding allows frequencies that are not integer divisions of it.
Harmonics
Avoid targets whose harmonics coincide, for example 8 and 16 Hz.
Joint frequency-phase
Coding targets by frequency and phase allows many more targets in a narrow band, as in 40-target spellers.
Detection methods
| Method | Training data | Idea |
|---|---|---|
| PSD peak | None | Compare spectral power at each candidate frequency and its harmonics. |
| CCA | None | Correlate multichannel EEG with sine-cosine references at each frequency; pick the highest canonical correlation. |
| FBCCA | None | CCA on several sub-bands to use harmonic information, combined with weights. |
| TRCA | Calibration | Spatial filters that maximise reproducibility across trials of each target; very high accuracy with calibration data. |
| Trained classifier / CNN | Calibration | Spectral features or raw epochs into a classifier or a compact CNN. |
What BCILattice provides for SSVEP
BCILattice does not currently ship CCA, FBCCA or TRCA as dedicated SSVEP blocks. If you need those training-free detectors, the Custom Code block runs your own Python inside the pipeline. It is the one plan-gated block.
What it does provide is the trained route: epoching around stimulus onset, the PSD-based channel selection method for band-power differences, frequency-domain feature families (band power, band peak frequency, spectral entropy and more), scikit-learn and Riemannian classifiers, and compact CNNs such as EEGNet. The Pseudo-Online Evaluator reports bit rate, and benchmarks include SSVEP decoding tasks.
Report the information transfer rate
SSVEP accuracy rises with window length, so a 4-second window at 98% can be a worse BCI than a 1-second window at 90%. Information transfer rate (ITR) combines the number of targets, accuracy and selection time into bits per minute, and it is the number SSVEP systems should be compared on.
Questions
What is an SSVEP BCI?
An SSVEP BCI uses steady-state visual evoked potentials: when a user looks at a light flickering at a fixed frequency, their visual cortex responds at that frequency. Several targets flicker at different frequencies, and the BCI detects which frequency dominates the EEG.
Which electrodes are used for SSVEP?
Occipital and parieto-occipital electrodes over the visual cortex, such as O1, Oz, O2, PO3, POz and PO4.
What is CCA in SSVEP?
Canonical correlation analysis finds the correlation between multichannel EEG and sine-cosine reference signals at each candidate frequency. The frequency with the highest correlation is taken as the attended target. It needs no training data.
Does SSVEP work without looking at the target?
Poorly. SSVEP depends on gaze: covert attention to a flicker produces a much weaker response than looking at it directly.
Does BCILattice include CCA or TRCA?
Not as dedicated blocks today. They can run through the Custom Code block. BCILattice provides the trained route instead: spectral features, classical and Riemannian classifiers, and compact CNNs.
Read the reference
Related
P300 BCI and the P300 Speller Explained
When a rare target appears among frequent non-targets, the brain produces a positive deflection about 300 ms later. A P300 BCI flashes options and finds the one that evokes it.
Motor Imagery BCI: EEG Classification Explained
Imagining a movement suppresses the mu and beta rhythms over the opposite motor cortex. A motor imagery BCI detects that suppression and turns it into a command, with no movement and no external stimulus.
EEG Frequency Bands: Delta, Theta, Alpha, Beta, Gamma
EEG is described in frequency bands, from slow delta to fast gamma. The bands are a useful shorthand for a continuous spectrum, and the popular "one wave, one mental state" story oversimplifies them.
EEG Feature Extraction: Which Features to Use
A feature is a reduction of a trial to a number. The right ones depend on where the signal lives (time, frequency, space or the links between channels), and the wrong ones multiply until the model overfits.