Oscillatory burst
Averaged spectra suggest sustained rhythms, but single-trial time-frequency analysis shows that many neural rhythms, beta in motor cortex most clearly, occur as brief, intermittent bursts lasting a few cycles rather than as continuous oscillations. Trial-averaged band power that looks like a smooth, sustained rhythm can be the blurred sum of a few high-amplitude bursts whose timing varies from trial to trial.
This distinction matters for both physiology and BCI. If beta suppression before movement reflects a drop in burst rate rather than a graded change in a continuous oscillation, then single-trial burst timing, not average band power, may be the more informative and more decodable feature. Detecting bursts requires thresholding a time-frequency amplitude and defining onset, duration and rate, choices that strongly shape the reported statistics and that trial averaging hides entirely.
Beta-band amplitude in sensorimotor cortex, examined trial by trial, appears as discrete events of roughly 100 to 150 ms; the classic smooth pre-movement beta decrease is largely a fall in how often these events occur.
Single-trial bursts underlie what trial averaging renders as a smooth rhythm.
Trial averaging is the trap: it turns intermittent, variable-timing bursts into an apparently continuous rhythm and hides the very structure that carries information.