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

Abstract

Research into the nature of 1/f-like, non-oscillatory electrophysiological activity has grown exponentially in recent years in cognitive neuroscience. The shape of this activity has been linked to the balance between excitatory and inhibitory neural circuits, which is thought to be important for information processing. However, to date, it is not known whether the presentation of a stimulus induces changes in the parameters of 1/f activity, which are separable from the emergence of event-related potentials (ERPs). Here, we analyze event-related broadband changes in scalp-recorded EEG both before and after removing ERPs to demonstrate their confounding effect, and to establish whether there are genuine stimulus-induced changes in 1/f. Using data from a passive and an active auditory task (n=23), we found that the shape of the post-event spectra differed significantly from the pre-event spectra even after removing the frequency-content of ERPs. Further, a significant portion of this difference could be accounted for by a rotational shift in 1/f activity, manifesting as an increase in low and a decrease in high frequencies. Importantly, the magnitude of this rotational shift was related to the attentional demands of the task. This change in 1/f is consistent with increased inhibition following the onset of a stimulus, and likely reflects a disruption of ongoing excitatory activity proportional to processing demands. Finally, these findings contradict the central assumption of baseline normalization strategies in time-frequency analyses, namely that background EEG activity is stationary across time. As such, they have far-reaching consequences that cut across several subfields of neuroscience. Significance statementInterest in the functional role of the 1/f-like background brain activity has been growing exponentially in neuroscience. Yet, no study to date has demonstrated a clear relationship between information processing and 1/f activity by investigating event-related effects on its parameters. Here, we demonstrate for the first time that stimuli induce rotational changes in 1/f activity, independent from the occurrence of event-related potentials. Not only do these findings suggest that the assumption of stationary background activity in the analysis of neural oscillations is untenable, but they also suggest the presence of large-scale inhibition following stimulus onset, the magnitude of which is dependent on the demands of the task. These findings have far-reaching consequences that cut across several subfields of neuroscience.