Stimulus‐induced change in long‐range temporal correlations and scaling behaviour of sensorimotor oscillations

Abstract

The human brain spontaneously generates large-scale network oscillations at around 10 and 20 Hz. The amplitude envelope of these oscillations fluctuates intermittently and was recently reported to exhibit power-law decay of the autocorrelation for hundreds of seconds. This indicates that the underlying networks are in a dynamic state resembling the self-organized critical state known to exist in many complex systems. Based on the mechanism of how correlations emerge in these systems, we hypothesized that the physiological basis of long-range power-law correlations is the buildup of a memory of past activity by a continuous modification of the network's functional connectivity by the ongoing oscillations. In this framework, exogenous perturbations of ongoing oscillations would degrade or abolish this dynamic network memory. We investigated the sensitivity of the temporal correlations in sensorimotor 10- and 20-Hz oscillations to median nerve stimulation that is known to have immediate effects on ongoing oscillations. Our results show that the amplitude fluctuations of these oscillations were effectively modulated by the somatosensory stimuli but still exhibited long-range temporal correlations and power-law scaling behaviour. The magnitude of the temporal correlations was, however, attenuated and the power-law exponents were decreased. This implies that the stimuli indeed degraded the network's memory of its past.

Pre-stimulus alpha oscillations over somatosensory cortex predict tactile misperceptions

“…A number of studies have reported quadratic rather than linear relationships between alpha power and performance on several tactile tasks (Ai & Ro, 2014;Linkenkaer-Hansen et al, 2004;Weisz et al, 2014;Zhang & Ding, 2010). Typically, in these studies hit rates are highest at an intermediate level of alpha power.…”

“…A parabolic relationship between performance and noise is a key characteristic of stochastic resonance. However, these reports in the somatosensory domain have not taken the possibility of changes in false alarm rate with changes in alpha power into account, and thus may have underestimated the downside of variations of alpha power (Ai & Ro, 2014;Linkenkaer-Hansen et al, 2004;Weisz et al, 2014;Zhang & Ding, 2010).…”

“…The majority of previous examinations of the relationship between alpha and performance have followed the following procedure: 1) trials are binned in order of increasing spectral power, typically using either a quartile or decile split 2) a mean hit rate, a "normalized detection rate", or signal detection measure is calculated for each bin 3) binned power is entered as a factor in a repeated measures ANOVA, with hit rate as the dependent variable (for examples of such approaches, see Busch et al, 2009;Jones et al, 2010;Limbach & Corballis, 2016;Linkenkaer-Hansen et al, 2004;Zhang & Ding, 2010 as the dependent variable. We also performed these analyses separately on each hemisphere rather than include it as a factor.…”

“…However, several other groups have reported a quadratic relationship between alpha and performance, with intermediate power levels associated with improved hit rates (Ai & Ro, 2014;Linkenkaer-Hansen et al, 2004;Weisz et al, 2014;Zhang & Ding, 2010).…”

“…Specifically, detection of targets improves and declines according to the power and phase of ongoing alpha oscillations over (typically) contralateral primary somatosensory cortex (Ai & Ro, 2014;Jones et al, 2010;Linkenkaer-Hansen, Nikulin, Palva, Ilmoniemi, & Palva, 2004;Weisz et al, 2014;Zhang & Ding, 2010). Alpha has been linked to cortical excitability, which may be the underlying mechanism that links observed alpha with performance (e.g.…”