Modeling spectral changes following stimulation: Is there an event-related shift in 1/f activity?

Abstract

AbstractNeural activity contains oscillatory components (i.e., narrowband oscillations) and non-oscillatory components (e.g., event-related potentials [ERPs] and 1/f-like background activity). Here, surface-level EEG data was analyzed to investigate how the spectral content of neural activity below 25 Hz changes from before to after an event. We focused on changes in non-oscillatory background activity, a spectral component often assumed to be static across time in time-frequency analyses. During a simple auditory perception task (n = 46) and an auditory oddball task (n = 23), we found an apparent increase in the offset and a decrease in the slope of 1/f activity from before to after a tone. Importantly, however, these changes in non-oscillatory background activity were almost completely accounted for by the emergence of ERPs in response to the stimulus in frequencies below the alpha range (8-12 Hz). Our findings suggest that post-event spectral changes below 25 Hz can be modelled as the sum of pre-event non-oscillatory activity, the spectrum of the ERP, and an independent alpha component that is modulated in amplitude, but not elicited, by the event. Theta activity (4-8 Hz), however, was not present before the event and appeared to be phase-locked to it. The theoretical and methodological implications of our findings regarding the nature and origin of 1/f activity, and the interpretation of low-frequency activity in the time-frequency domain are discussed.