Comparative Analysis of the Permutation and Multiscale Entropies for Quantification of the Brain Signal Variability in Naturalistic Scenarios

Abstract

As alternative entropy estimators, multiscale entropy (MSE) and permutation entropy (PE) are utilized for quantification of the brain function and its signal variability. In this context, their applications are primarily focused on two specific domains: (1) the effect of brain pathology on its function (2) the study of altered states of consciousness. As a result, there is a paucity of research on applicability of these measures in more naturalistic scenarios. In addition, the utility of these measures for quantification of the brain function and with respect to its signal entropy is not well studied. These shortcomings limit the interpretability of the measures when used for quantification of the brain signal entropy. The present study addresses these limitations by comparing MSE and PE with entropy of human subjects’ EEG recordings, who watched short movie clips with negative, neutral, and positive content. The contribution of the present study is threefold. First, it identifies a significant anti-correlation between MSE and entropy. In this regard, it also verifies that such an anti-correlation is stronger in the case of negative rather than positive or neutral affects. Second, it finds that MSE significantly differentiates between these three affective states. Third, it observes that the use of PE does not warrant such significant differences. These results highlight the level of association between brain’s entropy in response to affective stimuli on the one hand and its quantification in terms of MSE and PE on the other hand. This, in turn, allows for more informed conclusions on the utility of MSE and PE for the study and analysis of the brain signal variability in naturalistic scenarios.

Self-organized criticality as a framework for consciousness: A review study

“…Out of the original articles, n = 16 (44.4%) reported on human data and in n = 7 (19.4%) studies data was recorded in animals. Regarding the methods, computational models were utilized in n = 12 studies (Zare and Grigolini, 2013;Aguilera et al, 2015;Kang et al, 2017;Podobnik et al, 2017;Stramaglia et al, 2017;Abeyasinghe et al, 2018Abeyasinghe et al, , 2020Aguilera, 2019;Lee et al, 2019;Avramiea et al, 2020;Kim and Lee, 2020;Popiel et al, 2020), electrocorticography (EcoG) was assessed in n = 4 articles (Alonso et al, 2014;Solovey et al, 2015;Krzemiński et al, 2017;Varley et al, 2020b), fMRI in n = 4 studies (Liu et al, 10.3389/fpsyg.2022.911620 Frontiers in Psychology 05 frontiersin.org 2014; Tagliazucchi et al, 2014Tagliazucchi et al, , 2016Varley et al, 2020a), voltage imaging/microelectrode arrays in n = 4 studies (Fagerholm et al, 2016;Hudetz et al, 2016;Fakhraei et al, 2017;Fekete et al, 2018), EEG/MEG in n = 10 studies (Allegrini et al, 2015;Colombo et al, 2016;Irrmischer et al, 2018;Kim and Lee, 2019;Duncan et al, 2020;Dürschmid et al, 2020;Keshmiri, 2020;Fekete et al, 2021;Huels et al, 2021; and EEG and fMRI in one study (Kim M. et al, 2021; Figure 3). Notably, different analytical tools were applied in the EEG/MEG studies to assess a surrogate measure of criticality.…”