The rate of transient beta frequency events predicts impaired function across tasks and species

Abstract

Beta frequency oscillations (15-29Hz) are among the most prominent signatures of brain activity. Beta power is predictive of many healthy and abnormal behaviors, including perception, attention and motor action. Recent evidence shows that in non-averaged signals, beta can emerge as transient high-power "events". As such, functionally relevant differences in averaged power across time and trials can reflect accumulated changes in the number, power, duration, and/or frequency span of the events. We show for the first time that functionally relevant differences in averaged prestimulus beta power in human sensory neocortex reflects a difference in the number of high-power beta events per trial, i.e., the rate of events. Further, high power beta events close to the time of the stimulus were more likely to impair perception. This result is consistent across detection and attention tasks in human magnetoencephalography (MEG) and is conserved in local field potential (LFP) recordings of mice performing a detection task. Our findings suggest transient brain rhythms are best viewed as a "rate metric" in their impact on function, and provides a new framework for understanding and manipulating functionally relevant rhythmic events.

Regional Brain Correlates of Beta Bursts in Health and Psychosis: A Concurrent Electroencephalography and Functional Magnetic Resonance Imaging Study

“…Consistent with the Engel and Fries ( 17 ) model, Shin et al. ( 22 ) demonstrated that a transient beta burst just prior to a liminal stimulus reduced the probability of stimulus detection. In primates, Lundqvist et al.…”

“…The resulting time–frequency spectrograms were filtered with a two-dimensional Gaussian filter (standard deviations: 1 Hz/6 ms), and peaks were identified using image dilatation with a 5-by-5 structuring element consisting of ones with a center value of zero (implementation: Tony Fast; https://gist.github.com/tonyfast/d7f6212f86ee004a4d2b ). Peak values less than 6 times the median power across all time points at the peak frequency were excluded ( 22 ). An example of a burst that exceeded this threshold is shown in Figure 1C .…”

Simulations reveal that beta burst detection may inappropriately characterize the beta band

“…The tested power thresholds ranged from 0.1 to 20 in 30 equally sized steps. For each simulation the best threshold was determined by calculating the maximum correlation between mean single trial power and the area, represented as a proportion of total TFR area per trial, above the cutoff (Shin et al 2017). Following this, we found local maxima that were above the previously calculated cutoff.…”

“…Lastly, we simulated a scenario in which the power difference was caused by a transient burst rate difference; in the HP trials the p event = 0.9 and in the LP trials the p event = 0.1. Our attention centered specifically on the transient burst count, given both its significance in numerous studies (e.g., Shin et al, 2017;Little et al, 2019) and that it indicates the presence of bursts, and thus all other characteristics are dependent on it. Initially, we visualized the single-trial burst count recovered by single-trial power for both scenarios in Fig 4A . The results indicated that as the number of detected bursts increased so did single-trial power, in all scenarios.…”

Temporal evolution of beta bursts in the parkinsonian cortico-basal ganglia network

“…Oscillations in the beta frequency band have been the focus of numerous studies due to their central role in cortical processing, particularly in mediating top-down commands (Bastos et al, 2015;Engel and Fries, 2010;Michalareas et al, 2016;Richter et al, 2017), for their prominence in the motor system (Baker, 2007;Baker et al, 1997;Klostermann et al, 2007;Sanes and Donoghue, 1993) and during Parkinson's disease (PD) (Cassidy et al, 2002;Kühn et al, 2004;Lalo et al, 2008). Across different experimental conditions, changes in beta band activity have been linked with motor performance (Kühn et al, 2004;Tan et al, 2014), decision-making (Herz et al, 2018), motor-task completion (Feingold et al, 2015), tactile perception (Shin et al, 2017) and PD treatment state (Kühn et al, 2006(Kühn et al, , 2009Little et al, 2013;Tinkhauser et al, 2017aTinkhauser et al, , 2017b.…”