Human consciousness is supported by dynamic complex patterns of brain signal coordination. A. Demertzi et al. Science Advances Feb 06 2019: Vol. 5, no. 2, eaat7603
DOI: 10.1126/sciadv.aat7603
Abstract: Adopting the framework of brain dynamics as a cornerstone of human consciousness, we determined whether dynamic signal coordination provides specific and generalizable patterns pertaining to conscious and unconscious states after brain damage. A dynamic pattern of coordinated and anticoordinated functional magnetic resonance imaging signals characterized healthy individuals and minimally conscious patients. The brains of unresponsive patients showed primarily a pattern of low interareal phase coherence mainly mediated by structural connectivity, and had smaller chances to transition between patterns. The complex pattern was further corroborated in patients with covert cognition, who could perform neuroimaging mental imagery tasks, validating this pattern’s implication in consciousness. Anesthesia increased the probability of the less complex pattern to equal levels, validating its implication in unconsciousness. Our results establish that consciousness rests on the brain’s ability to sustain rich brain dynamics and pave the way for determining specific and generalizable fingerprints of conscious and unconscious states.
Check also Chasing the Rainbow: The Non-conscious Nature of Being. David A. Oakley and Peter W. Halligan. Front. Psychol., November 14 2017. https://www.bipartisanalliance.com/2018/01/despite-compelling-subjective.html
---
INTRODUCTION
Consciousness is seemingly lost and recovered every day, from themoment we fall asleep until we wake up. Consciousness can also betransiently abolished by pharmacological agents or, more permanently,by brain injury. Each of these departures from conscious wakefulnessbrings about different changes in brain function, behavior, and neuro-chemistry. Yet, they all share a common feature: lack of reported sub-jectiveexperience(1).Finding reliable markers indicating the presence or absence of con-sciousness represents an outstanding open problem for science (2).We postulate that consciousness has specific characteristics that arebased on the temporal dynamics of ongoing brain activity and its co-ordination over distant cortical regions. Our hypothesis stems fromthe common stance of various contemporary theories which proposethat consciousness relates to a dynamic process of self-sustained,coordinated brain-scale activity assisting the tuning to a constantlyevolving environment, rather than in static descriptions of brainfunction (3–5). In that respect, neural signals combine, dissolve, re-configure, and recombine over time, allowing perception, emotion,and cognition to happen (6).The first biological evidence for a constantly active brain camefrom electroencephalographic recordings showing electrical oscilla-tions even when the participant was not performing any particulartask. More recently, brain dynamics have been characterized by thepresence of complex activity patterns, which cannot be completelyattributed to background noise (7). Experiments with functional mag-netic resonance imaging (fMRI) during normal wakefulness haveshown that the brain spontaneouslygenerates a dynamic series of con-stantly changing activity and connectivity between brain regions (8–10).This activity presents long-range temporal correlations in the sensethat signal changes exert long-term influence on future dynamics (11).This translates to a complex temporal organization of the long-rangecoupling between brain regions, with temporally correlated series oftransitions between discrete functional connectivity patterns (6). Thespatiotemporal complexity of brain dynamics contributes toward ef-ficient exchanges between neuronal populations (8), suggesting thatthe neural correlates of consciousness could be found in temporallyevolving dynamic processes, as postulated by influential theoreticalaccounts (3–5).In terms of states of consciousness, spontaneous fMRI dynamicconnectivity has been investigated in different sleep stages (11,12)and pharmacologically induced anesthesia in humans (13,14)andanimals (15,16). These studies indicate that, during physiologicallyreversible unconscious states, cortical long-range correlations are
disrupted in both space and time, anticorrelated cortical statesdisappear, and the dynamic explorations are limited to specific patternsthat are dominated by rigid functional configurations tied to the anatom-ical connectivity. Conversely, conscious wakefulness is characterizednot only by global integration, evidenced by strong long-distance inter-actions between brain regions, butalso by a dynamic exploration of arich and flexible repertoire of functional brain configurations departingfrom the anatomical constraints (15). Another important characteristicobserved predominantly during conscious wakefulness is the appear-ance of anticorrelations between the activity of different brain regions.This observation is in line with the prediction of the Global NeuronalWorkspace theory stating that different streams of information in thebrain compete for the global percolation (“ignition”) of a widespreadnetwork of regions, a phenomenon associated with conscious access.In terms of the fMRI blood oxygen level–dependent (BOLD) signal,this could manifest in the mutual inhibition of activity at different cor-tical regions, leading to anticorrelated dynamics (5).Although dynamic connectivity has been investigated in physio-logical and pharmacological unresponsiveness, currently, the altera-tions in brain connectivity dynamics associated with pathologicalunconsciousness after severe brain injury remain unknown. The studyof unresponsive brain-lesioned patients with preserved levels of vig-ilance offers unique insights into the necessary and potentially suf-ficient conditions for the capacity of sustaining conscious content. Sofar, the inference of consciousness in patients has rested on the use ofactive mental imagery neuroimaging paradigms (17) and by assess-ing the complexity of evoked (18) and spontaneous brain activity(19). Patients who successfully perform these active paradigms canno longer be considered unconscious and are thought to suffer fromcognitive-motor dissociation (20). Given that nonresponsiveness canbe associated with a variety of brain lesions, varying levels of vigilance,and covert cognition, we highlight the need to determine a commonset of features capable of accounting for the capacity to sustain con-scious experience. Given the above theoretical considerations, whichagree in the characterization of consciousness as a global, temporallyevolving process, we aimed at determining whether the dynamics ofbrain-wide coordination could provide such a set of common featuresin the form of transient patterns of connectivity that successfully gen-eralize between different forms of nonresponsiveness in patients withbrain injury.
DISCUSSION
We studied the brain’s dynamic organization during consciouswakefulness and after severe brain injury leading to disorders of con-sciousness, with the aim of determining patterns of signal coordinationspecifically associated with conscious and unconscious states. Weidentified a pattern of positive and negative long-distance coordination,high modularity, with low similarity to the anatomical connectivity,
potentially relevant for the support of conscious cognition (pattern 1).We also identified a pattern of low interregional dynamic coordina-tion, low efficiency, with high similarity to anatomical connectivity,potentially specific to reduced or absent conscious processing (pattern4). With respect to pattern 1, momentary neural coalitions have beenpreviously shown to constitute a basis for complex cognitive function,with signals fluctuating between states of high and low connectivityand with more integrated states enabling faster and more accurate
performance during cognitive tasks (23). With respect to pattern 4,studies in physiological and pharmacological unconsciousnessshowed a breakdown of long-range interareal positive and negativeconnections. For example, during sleep, the presence of negative dy-namic connections disappear (11). Our results are in line with previ-ous findings in animals. As in the present study, the brain activity ofanesthetized nonhuman primates resided most frequently in a patternof low connectivity resembling the anatomy, which was sustained forlonger periods of time in comparison to more complex patterns (15).In addition, we demonstrated that network properties, such as mod-ularity, integration, distance relationship, and efficiency, increasedwith the participants’conscious state. The latter result is in line withthe hypothesis that high-efficiency patterns carry higher metaboliccosts (24), which are restricted underpathologicalunconsciousconditions (25).Our findings also align with theoretical considerations on dynamicconnectivity, suggesting that alternating patterns of correlations andanticorrelations may constitute a fundamental property of informationprocessing in the brain (26). Differentmodels of consciousness proposethat intermittent epochs of global synchronization grant segregated
and parallel network elements access to a global workspace, integratingthem serially and allowing effortful conscious cognition (27). There-fore, the transient exploration of this global workspace could permitthe brain to efficiently balance both segregated and integrated neuraldynamics and to encode globally broadcasted and therefore reportableconscious contents (3). In the absence of transient epochs of global syn-chronization, the transmission of information is expected to be rela-tively ineffective (6).With respect to patterns 2 and 3, these were not predominantlypreferred by any group in terms of occurrence probabilities and dura-tion and hence could represent transitional states (28). Pattern 3 showedthe overall positive interareal coherence. For pattern 2, the significanceof the overall negative coherence with regions of the visual networkcan only be speculated. For the moment, we suggest that it indicatesthe presence of a local coordination pattern reflecting the anatomicalorganization of the visual cortex (29).Whether the identified dynamic coordination patterns entail thepresence/absence of mental contents or cognitive function is difficultto assume without probing moment-to-moment changes in thecontents of conscious experience. Although a link has been shownbetween intrinsic connectivity networks and various behavioral tasks(30), it has been suggested that BOLD correlations need not necessar-ily reflect moment-to-moment changes in cognitive content. Instead,they may predominantly reflect processes necessary for maintainingthe stability of the brain’s functional organization (31). Also, the BOLDsignal is nonstationary (32), and some of its spontaneous fluctuationsmay not be a faithful reflection of functionally relevant brain dynamicsor the underlying nonstationarities of neural activity and coordination(10). We also acknowledge that the identification of these dynamicconfigurations required time-resolved analyses of fMRI time seriesin the scale of few seconds. It can be argued that conscious cognitionand the relevant features of our environment develop on a faster timescale of hundreds of milliseconds (3). However, the BOLD signal hasbeen shown to correlate with infraslow neurophysiological oscilla-tions, i.e., the slow cortical potential (33). The slow cortical potentialis important for large-scale information integration, hence suggest-ing that the flow of the conscious experience could be supported byprocesses at slower time scales (34). Future experiments should ad-dress a potential relationship between conscious experience, the slowcortical potential, and functional network reconfigurations measuredas with fMRI.Regardless of the implicated time scales, our analyses did not aim attracking the moment-to-moment contents of conscious experience,but at identifying brain-wide dynamic networks supporting differentglobal states of consciousness. We consider that the four-pattern modelcan account for modes of conscious and unconscious informationprocessing. Our interpretation is sustained by the additional testsfor the validity and replicability of the main results. We found thatthe complex dynamic pattern 1 presented low probabilities of appear-ing in patients under propofol anesthesia (whether they were commu-nicating or not at baseline) and that it was most likely to appear inpatients with covert cognition (i.e., patients in UWS who successfullyperformed mental imagery neuroimaging tasks). Both findings sug-gest its implication in conscious states. We also found that the pat-tern of low interareal coordination (pattern 4) uniformly presentedhigher probabilities of appearing in all anesthetized patients, regardlessof clinical diagnosis, and it was most likely to manifest in unresponsivepatients who did not perform the mental imagery neuroimaging task,supporting its relationship to absent or reduced conscious cognition.Pattern 4 remained visited by healthy controls even under typicalwakeful conditions. In the absence of experience sampling duringdata acquisition, the interpretation of this finding can only be specu-lative. On the one hand, it could be that healthy controls entered tran-sient microsleep states as a result of fluctuating levels of vigilance, afrequently observed phenomenon during resting-state experiments(35). Our experimental setup did not include simultaneous poly-somnography recordings to directly test this hypothesis. However,we derived different fMRI-based proxies to assess the presence ofmicrosleeps. First, we examined head movements time locked to theoccurrence of all coordination patterns and found no substantial as-sociations between the two variables, as could be expected if pattern 4was related to lapses in vigilance. Second, we performed a whole-braingeneral linear model (GLM) analysis, with the coordination patterntime series as regressors. We did not observe significant positive/negative BOLD signal changes associated with the onset of the differentcoordination patterns; in particular, the presence of coordination pat-tern 4 did not result in BOLD signal changes typical of microsleeps.Last, the likelihood of pattern 4 occurring over time did not positivelycorrelate with the elapsed scan time, as has been shown to occur forpatterns associated with lapses in vigilance (35,36). Once we rule outtransient loss of vigilance as the cause of the intrusion of pattern 4 inconscious wakefulness, we can speculate that the flow of consciouscognition may be separated by periods of absent or reduced effortfulinformation processing, as recently it was hypothesized that betweentwo successive self-reports, a subject may present states of reducedawareness (37). Behaviorally, this could take the form of“mindblanks”during which participants are not engaged in cognitivedemanding processes, although they remain vigilant (38). This inter-pretation is parsimonious with the observation that participantswere instructed to rest inside the scanner, without engaging in anyeffortful cognitive task. The potential role of transient lapses of aware-ness in the stream of conscious contents during healthy wakefulnessshould be addressed by future experiments.Together, our results suggest that, following loss of consciousness,coordinated brain activity is largely restricted to a positive pattern ofinterareal coherence dominated by the anatomical connections be-tween brain regions. In contrast, conscious states are characterizedby a higher prevalence of a complex configuration of interareal coor-dination that, while still constrained by brain anatomy, also deviatesfrom it and presents both positive and negative long-distance inter-actions. It did not escape us that such a complex interareal coor-dination pattern sporadically appeared in the group of unresponsivepatients. The real-time detection of this pattern and its reinforcementthough externally induced manipulations could represent a promis-ing avenue for the noninvasive restoration of consciousness. We con-clude that these patterns of transient brain signal coordination arecharacteristic of conscious and unconscious brain states, warrantingfuture research concerning their relationship to ongoing consciouscontent, and the possibility of modifying their prevalence by externalperturbations, both in healthy and pathological individuals, as well asacross species.
No comments:
Post a Comment