Breakdown of long-range temporal correlations in brain oscillations during general anesthesia

Highlights

• Long-range temporal correlations (LRTCs) analyzed in macaque ECoG recordings.

• Wakefulness: robust LRTCs form a posterior-anterior topographic gradient.

• Anesthesia: weak LRTCs and breakdown of the gradient.

• Weak LRTCs during anesthesia might indicate disrupted temporal integration of information.

Abstract

Consciousness has been hypothesized to emerge from complex neuronal dynamics, which prevails when brain operates in a critical state. Evidence supporting this hypothesis comes mainly from studies investigating neuronal activity on a short time-scale of seconds. However, a key aspect of criticality is presence of scale-free temporal dependencies occurring across a wide range of time-scales. Indeed, robust long-range temporal correlations (LRTCs) are found in neuronal oscillations during conscious states, but it is not known how LRTCs are affected by loss of consciousness. To further test a relation between critical dynamics and consciousness, we investigated LRTCs in electrocorticography signals recorded from four macaque monkeys during resting wakefulness and general anesthesia induced by various anesthetics (ketamine, medetomidine, or propofol). Detrended Fluctuation Analysis was used to estimate LRTCs in amplitude fluctuations (envelopes) of band-pass filtered signals. We demonstrate two main findings. First, during conscious states all lateral cortical regions are characterized by significant LRTCs of alpha-band activity (7–14 Hz). LRTCs are stronger in the eyes-open than eyes-closed state, but in both states they form a spatial gradient, with anterior brain regions exhibiting stronger LRTCs than posterior regions. Second, we observed a substantial decrease of LRTCs during loss of consciousness, the magnitude of which was associated with the baseline (i.e. pre-anesthesia) state of the brain. Specifically, brain regions characterized by strongest LRTCs during a wakeful baseline exhibited greatest decreases during anesthesia (i.e. “the rich got poorer”), which consequently disturbed the posterior-anterior gradient. Therefore, our results suggest that general anesthesia affects mainly brain areas characterized by strongest LRTCs during wakefulness, which might account for lack of capacities for extensive temporal integration during loss of consciousness.

Introduction

Brain activity, both task-related and spontaneously generated, is characterized by complex spatio-temporal dynamics. It has been proposed that understanding brain as a system operating in a critical state might provide a theoretical framework allowing comprehensive description of neuronal activity (review: Chialvo, 2007, Chialvo, 2010, Shew and Plenz, 2013, Beggs and Timme, 2012). A critical state can occur in complex systems composed of multiple interacting elements. When balanced interactions among the elements keep a system operating at the verge of order and randomness a plethora of non-trivial features can be observed (in contrast to rather trivial states of complete order and randomness). A hallmark of criticality is lack of any characteristic scale in system's activity (i.e. scale-free dynamics), which is indicated by a power-law scaling (Beggs and Timme, 2012).

First evidence for brain operating in a critical state came from studies investigating “neuronal avalanches”, which were defined as temporally consistent bursts (clusters) of activations occurring simultaneously at several brain locations. Neuronal avalanches were first investigated in LFP recordings, with activations defined based on deflections of LFP signals. Scale-free avalanches, with their size and duration following a power law, were observed both in vitro (Beggs and Plenz, 2003, Beggs and Plenz, 2004) and in vivo (Gireesh and Plenz, 2008, Petermann et al., 2009, Hahn et al., 2010). Further studies confirmed that similar scale-free dynamics can be found in avalanches defined based on non-invasively recorded MEG (Shriki et al., 2013) or fMRI data (He, 2011, Tagliazucchi et al., 2012).

But evidence for brain criticality is not limited to studies investigating “neuronal avalanches”. Another line of research focused on long-range temporal correlations (LRTCs), which are defined as a power-law decay of an autocorrelation function and thus indicate that a signal exhibits scale-free patterns in the temporal domain (in the form of 1/f spectrum). LRTCs can be thought to reflect a long “memory” of a signal and a simultaneous modulation across a range of time-scales. Robust LRTCs were found in amplitude modulation of oscillatory brain activity (Linkenkaer-Hansen et al., 2001, Linkenkaer-Hansen and Nikulin, 2004) and in temporal progression of EEG topographical voltage patterns (so called “microstates”; Van de Ville et al., 2010, Gschwind et al., 2015). Importantly, even before describing LRTCs in brain activity, their presence had been observed in behavioral patterns of humans and other species (Gilden et al., 1995, Kello et al., 2010). Recent studies demonstrate links between these various scales of analysis by showing that LRTCs in neuronal oscillations (meso-scale) are related to “neuronal avalanches” in spiking activity (micro-scale; Poil et al., 2012) and, at the same time, to temporal dependencies observed in behavior (macro-scale; Palva et al., 2013, Smit et al., 2013).

Operating in a critical regime provides functional benefits for a system, as it allows balanced propagation of external and internal perturbations and is related to maximization of information storage and transfer (Shew et al., 2009, Shew et al., 2011, Marinazzo et al., 2014). Crucially, the qualities of critical systems overlap with features postulated necessary for emergence of consciousness from brain networks (Tononi, 2004, Seth, 2009, Dehaene and Changeux, 2011, Tononi et al., 2016). Therefore, it has been hypothesized that consciousness occurs when brain operates in (or close to) a critical state, whereas changes in the state consciousness are associated with departure from criticality towards either sub- or super-critical states (Chialvo, 2007, Werner, 2013, Tagliazucchi et al., 2013, Tagliazucchi et al., 2016; Carhart-Harris et al., 2014). Yet, evidence supporting this hypothesis is so far scarce and inconclusive. Some studies investigating LFP recordings found that, indeed, neuronal avalanches loose their scale-free features during loss of consciousness suggesting a departure from the critical state (Ribeiro et al., 2010, Priesemann et al., 2013), but others reported that conscious and unconscious states do not differ in this respect (Dehghani et al., 2012). However, a main limitation of these LFP studies was sparse and selective data sampling by electrodes arrays. Conversely, by analyzing neuronal avalanches estimated from voltage imaging data (recorded from mice and characterized by superb spatial sampling and resolution) Scott et al. (2014) and Fagerholm et al. (2016) robustly demonstrated critical dynamics during wakefulness and departure from criticality under anesthesia.

However, in contrast to neuronal avalanches, neurophysiological LRTCs have not been systematically investigated in different states of consciousness. We are only aware of two fMRI studies which revealed loss of LRTCs in blood–oxygen-level dependent (BOLD) time-series during NREM sleep (Tagliazucchi et al., 2013) and propofol anesthesia (Tagliazucchi et al., 2016). Therefore, we set out to investigate LRTCs in neuronal activity during resting wakefulness and loss of consciousness caused by general anesthesia. Electrocorticographic (ECoG) data recorded from macaque monkeys were analyzed with Detrended Fluctuation Analysis (DFA; Peng et al., 1994), in order to estimate LRTCs in amplitude modulation of brain oscillatory activity (Fig. 1). We hypothesized to observe robust LRTCs during resting wakefulness and a breakdown of long-range temporal dependencies during loss of consciousness. To test these hypotheses we: first, investigated how general anesthesia affects LRTC; second, conducted detailed analyses of LRTCs' topography during wakefulness; and third, investigated topographic patterns of anesthesia-induced changes in LRTCs.