Historically EEG task time-frequency measures were assumed to reflect brain oscillations, but recent evidence indicates that aperiodic activity can confound accurate measurement of neural oscillations. Here, we capitalize on recent tools for parameterizing oscillations in time-frequency maps. This Registered Report will provide empirical evidence on which features of task-induced EEG time-frequency power are oscillatory and which are aperiodic.
Brain oscillations, or rhythms, coordinate communication across distributed brain networks. These rhythms provide a foundation for the brain network interactions required for cognition. Oscillations coexist with non-rhythmic background aperiodic activity that forms a characteristic 1/f pattern in power spectra. Aperiodic brain activity is associated with cognition and can confound the detection of oscillations. In this Registered Report, we applied time-resolved spectral parameterization to EEG recordings during two common cognitive tasks. Neural dynamics recorded during many cognitive paradigms show similar patterns, including synchronization of mediofrontal theta (4–8 Hz) and desynchronization of posterior alpha (9–13 Hz) and central beta (15–30 Hz). Our results indicate that common task time-frequency signatures, including mediofrontal theta synchronization and parietal alpha desynchronization, can be attributed primarily to neural oscillatory phenomena. Intriguingly, we uncover evidence of stimulus-locked aperiodic power changes, which are responsive to the need for cognitive control and to reinforcement processing. Furthermore, aperiodic power correlated strongly with non-baseline-corrected total power estimates, and whereas oscillatory power correlated strongly with portions of baseline-corrected power estimates, it failed to correlate with other portions of baseline-corrected power. Finally, after baseline correction, aperiodic correlations with TF power remain high. These results indicate two primary outcomes. First, task TF signatures in theta and alpha bands reflect primarily parameterized oscillations. Second, aperiodic activity is time-dependent during cognitive processing, and these dynamics are not accounted for by baseline correction.