Criticality as a determinant of integrated information Φ in human brain networks

Abstract

21Integrated information theory (IIT) postulates that consciousness arises from the cause-effect 22 structure of a system but the optimal network conditions for this structure have not been 23 elucidated. In the study, we test the hypothesis that network criticality, a dynamically balanced 24 state between a large variation of functional network configurations and a large constraint of 25 structural network configurations, is a necessary condition for the emergence of a cause-effect 26 structure that results in a large Φ, a surrogate of integrated information. We also hypothesized 27 that if the brain deviates from criticality, the cause-effect structure is obscured and Φ diminishes. 28We tested these hypotheses with a large-scale brain network model and high-density 29 electroencephalography (EEG) acquired during various levels of human consciousness during 30 general anesthesia. In the modeling study, maximal criticality coincided with maximal Φ. The 31 constraint of the structural network on the functional network is maximized in the maximal 32criticality. The EEG study demonstrated an explicit relationship between Φ, criticality, and level 33 of consciousness. Functional brain network significantly correlated with structural brain network 34 only in conscious states. The results support the hypothesis that network criticality maximizes 35 Φ. 36 37 Introduction 38Integrated information theory (IIT) proposes that consciousness equates with integrated 39 information in a system and the integrated information is maximized when integration and 40 differentiation of the systems' components are balanced. IIT proposes algorithmic methods to 41 identify the differentiated parts of a system and quantify the integrated information across the 42 parts [1-6]. Φ is a measure of complexity of the cause-effect structure of the minimum 43 information partition among all possible partitions. However, in a dynamic system such as the 44 brain, the optimal conditions under which the cause-effect structure-that is, the basis of 45 integrated information-arises has not been elucidated. In this study, we hypothesized that 46 network criticality, a balanced state between a large variation of functional network 47 configurations and a large constraint of structural network configurations, may be the basis of 48 the high Φ in conscious brains. 49Criticality was originally introduced for studying phase transition in physics, which was simply 50 defined as a balanced state between order and disorder in the activities of the elements that 51 make up a system [7,8]. This property has been observed broadly in physical and non-physical 52 systems and has been suggested as an optimal state for information storage, transmission, and 53 integration with high susceptibility to external stimuli [9][10][11][12][13]. In particular, several computational 54 modeling and empirical studies suggest that the brain dynamics associated with consciousness 55 reside near a critical state [14][15][16][17][18][19]. Furthermore, recent studies have attempted n...

Neural Correlates of the Shamanic State of Consciousness

“…We assessed criticality using the pair correlation function (PCF), which is the variance of network synchronization, reflecting network susceptibility to internal and external perturbations (e.g., from stimuli; Yoon et al, 2015 ). PCF is maximal at a critical point and gradually decreases as a system moves toward a sub- or super-critical state (Kim and Lee, 2019 ). Recently, the reliability of PCF in evaluation of the level of consciousness was supported with a computational model and EEG during general anesthesia (Kim and Lee, 2019 ).…”

Conscious Processing and the Global Neuronal Workspace Hypothesis

“…Nevertheless, IIT has served as a theoretical framework for more practical measures such as the perturbational complexity index, which has been successful in differentiating levels of consciousness in the setting of physiological, pharmacological, and pathological perturbations (Casali et al, 2013). Furthermore, some surrogate measures of  have been successfully applied to altered levels of consciousness in humans (Kim et al, 2018) and also shown to relate to network factors such as topological modularity (Kim et al, 2018) or critical dynamics (Kim and Lee, 2019).…”

Brain network motifs are markers of loss and recovery of consciousness

“…We constructed a binary adjacency matrix Aij using a threshold of 35%: if the wPLIij value of nodes i and j was within the top 35% of all wPLI values, Aij = 1; otherwise, Aij = 0. We chose a binarized approach to be coherent with the majority of studies assessing global network properties, which have also used a binarized approach to construct their graph 12,15,23,29,[68][69][70] . The 35% threshold was selected because it was previously shown, in the same dataset, to be the optimal threshold to avoid an iso- Figure 6.…”