Neural Noise

Joint alpha power and phase in visual sensitivity

3 min read

  • Alpha-band oscillations (8–14 Hz) in parietal and sensory cortices influence perception; power and phase have been studied separately despite a common pulsed-inhibition account.1
  • EEG work shows that a joint alpha power–phase dynamic shapes visual sensitivity.1 1

Gardner updates

  • tRNS modulates prestimulus alpha and beta power and improves visual perception (Communications Biology). 2

  • tRNS modulation of prestimulus alpha and beta power enhances visual perception (Communications Biology), relevant to non-invasive neuromodulation and neurofeedback. 2

Weekly enrichment (2026-07-20)

  • The preprint “A joint alpha power–phase dynamic shapes visual sensitivity” is by Henry Beale, Jason Mattingley, and Anthony Harris (Queensland Brain Institute, The University of Queensland), posted to bioRxiv on 20 February 2026 (doi:10.64898/2026.02.19.706926) under a CC-BY-NC 4.0 license.3
  • The EEG study tested 8 observers across 4 sessions each (32 sessions per task) in separate detection and discrimination tasks, viewing brief 8 ms oriented gratings at contrast levels adaptively set per session to probe threshold.3
  • Prestimulus alpha (8–14 Hz) power and phase were estimated on a single-trial basis from a 500 ms pre-stimulus window using a spectral-parameterisation algorithm that separates oscillations from aperiodic background activity, after using ICA to isolate a posterior contralateral stimulus-responsive source.3
  • Higher prestimulus alpha power reduced the ~200 ms negative evoked peak in both tasks (detection: β = −0.066, t(19345) = 2.62, P = .009; discrimination: β = −0.106, t(19345) = 2.72, P = .007), consistent with alpha suppressing early evoked potentials.3
  • The authors built a generative signal detection model that folds single-trial power and phase into one scalar “inhibition” value (phasic weighting m ∈ [0,1]) and compared four candidate mechanisms — response gain, contrast gain, noise, and criterion modulation.3
  • Models were fit hierarchically over the 32 sessions with Hamiltonian Monte Carlo (No-U-Turn sampler) in PyMC v4, with all R-hat values below 1.01; κ was fixed at 5, m used a Beta(1,1) prior, and the other modulation terms used Normal(0,2) priors.3
  • The model revealed response-gain modulation of visual sensitivity in both tasks and an additional baseline/criterion (perceptual bias) shift only in detection, indicating divisive suppression of sensory evidence.3
  • Critically, alpha power suppressed sensitivity more strongly than phase — there was no phasic return to unsuppressed performance — which the authors argue is not captured by a strong “pulsed-inhibition” account of alpha.3
  • The finding contrasts with a 2025 eLife study (6 observers, 6020 trials each detecting near-threshold Gabors) that attributed alpha-phase effects on d′ to reduced internal noise and sharper spatial-frequency/orientation tuning at the optimal phase, underscoring debate over the mechanism of alpha-band perceptual modulation relevant to closed-loop and neurofeedback BCI.4
  • The classic pulsed-inhibition framework holds that when alpha power is high its phase gates awareness — targets at one phase reach consciousness while those ~50 ms away (opposite phase) do not within the ~100 ms alpha cycle — a model these new joint results partially revise.5

Footnotes

  1. https://www.biorxiv.org/content/10.64898/2026.02.19.706926v1?rss=1 ↩ ↩2 ↩3

  2. https://news.google.com/rss/articles/CBMiX0FVX3lxTE1oR2NBSXRZQ1RNcmhzTlhJa3l5SUxIdnh6a3pOTnVQMVV0RHdVTTVWX3hmRFExbHpsdXRpQm55SV9rR0ZrOVNoSy1RYjkwRjltTXBGV0YzRXpXSlZDaV9N?oc=5 ↩ ↩2

  3. https://www.biorxiv.org/content/10.64898/2026.02.19.706926v1.full-text ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8

  4. https://elifesciences.org/articles/110000 ↩

  5. https://pmc.ncbi.nlm.nih.gov/articles/PMC3132674/ ↩

Graph View