Time-resolved spectral parameterization: Recommendations and guidelines

Non-oscillatory, aperiodic activity, is an emerging signal of interest whose physiological and functional origins are just beginning to be probed for their potential link to systems-level physiology and cognition. But there are several pernicious, practical, and statistical hazards to be aware of when deploying spectral parameterization tools for disentangling oscillations and aperiodic activity over short, noisy time windows. In this primer, we provide a set of recommendations and guidelines for time-resolved and event-related spectral parameterization to help researchers meet these challenges. Specifically, we provide recommendations on how to optimize and assess model fits, how to report analysis decisions to maximize reproducibility, and what to do when spectral parameterization fails to identify the presence of oscillations. These recommendations will help guide the study of the distinct dynamics of periodic and aperiodic neural activity in cognition, behavior, and disease at fine timescales.