A 40-class SSVEP speller dataset using beta-range frequency stimulation is released to support low-fatigue BCI research.1
Beta-range SSVEP is directly applicable to communication BCIs where user fatigue is a practical bottleneck.11
Weekly enrichment (2026-07-20)
The dataset (Kim et al., Scientific Data 2025, vol. 12, article 1751) provides a 40-class SSVEP speller collected from 40 healthy participants (25 male, 15 female, aged 20–35).2
Stimuli span the beta range 14.0–21.8 Hz in 0.2 Hz steps using joint frequency-phase modulation (JFPM) with 0.5π phase differences between adjacent targets, chosen because beta activity is less susceptible to visual fatigue than alpha/theta rhythms.23
EEG was recorded with a BioSemi ActiveTwo system on 31 central-to-occipital scalp channels (plus 2 mastoid references, 33 total) at a 1024 Hz sampling rate.23
Each subject completed 6 sessions of 40 trials, with a per-trial structure of 1.5 s rest, 0.5 s cue, and 5.0 s stimulation, alongside pre/post resting-state recordings under eyes-open and eyes-closed conditions.23
Subjective fatigue ratings plus EEG band-power analysis (delta 1–4, theta 4–8, alpha 8–12, beta 12–30, gamma 30–50 Hz) confirmed beta stimulation minimizes fatigue; fatigue mainly raised occipital delta/theta/alpha power while beta and gamma stayed relatively stable.2
Classification baselines included zero-training stdCCA and FB-CCA plus calibration-based IT-CCA, TRCA, and EEGNet; the training-based methods used 6-fold cross-validation per subject with a 1:5 test-to-training split across the 6 sessions.2
On efficiency, stdCCA achieved the highest information transfer rate (ITR) at a 3-second window and FB-CCA at 2 s, while IT-CCA, TRCA, and EEGNet peaked at a 1-second window, indicating strong real-time detection potential.2
The data are openly released on Figshare and packaged for reuse in the MOABB benchmarking framework as Kim2025BetaRange under a CC BY 4.0 license.3
Neural Noise