Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 87 of the 89 references it cites were found in a registry or an index.

Bifurcation in brain dynamics reveals a signature of conscious processing independent of report

Open access

Abstract

An outstanding challenge for consciousness research is to characterize the neural signature of conscious access independently of any decisional processes. Here we present a model-based approach that uses inter-trial variability to identify the brain dynamics associated with stimulus processing. We demonstrate that, even in the absence of any task or behavior, the electroencephalographic response to auditory stimuli shows bifurcation dynamics around 250–300 milliseconds post-stimulus. Namely, the same stimulus gives rise to late sustained activity on some trials, and not on others. This late neural activity is predictive of task-related reports, and also of reports of conscious contents that are randomly sampled during task-free listening. Source localization further suggests that task-free conscious access recruits the same neural networks as those associated with explicit report, except for frontal executive components. Studying brain dynamics through variability could thus play a key role for identifying the core signatures of conscious access, independent of report.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1038/s41467-021-21393-z

Cites

87 of 87 references verified0 citation errors · 0 not checkable

Statistics

Figures

Fig. 1 Models, behavioral results, and evoked potentials in the active sessions. A–C show the predictions of three possible models relating stimulus strength and neural activity. For each model, simulated trial-by-trial neural activity is represented as a function of the stimulation level (SNR). The corresponding distributions for each SNR are shown next to each graph with the same color-code. The bottom row shows the profiles of mean neural activity and variability of neural activity across trials (standard deviation) as predicted by each model. The profiles are averaged across 20 simulated participants who would have performed 160 trials per condition each (error bars correspond to standard error of the mean). D shows the behavioral results: identification performance, mean audibility and standard deviation of audibility across trials as a function of SNR, for each participant (faint colored lines) and averaged across all 20 participants (thick black line). Error bars represent ± standard error of the mean, n = 20 participants. The bottom row shows the distributions of audibility responses averaged across participants, for each SNR level. E Group averaged potentials evoked by the target vowels at electrode Cz and at each SNR level (n = 20 participants). F The top row shows the topographies of the event-related potentials at different time windows following targets at the highest SNR. The middle and bottom rows show the mean and variability of evoked activity as a function of SNR at these different time windows over a group of centro-parietal electrodes (highlighted on the topographies). Shaded areas correspond to ± SEM around the group average (n = 20 participants). Source data for this figure are provided as a Source data file.
Fig. 1 Models, behavioral results, and evoked potentials in the active sessions. A–C show the predictions of three possible models relating stimulus strength and neural activity. For each model, simulated trial-by-trial neural activity is represented as a function of the stimulation level (SNR). The corresponding distributions for each SNR are shown next to each graph with the same color-code. The bottom row shows the profiles of mean neural activity and variability of neural activity across trials (standard deviation) as predicted by each model. The profiles are averaged across 20 simulated participants who would have performed 160 trials per condition each (error bars correspond to standard error of the mean). D shows the behavioral results: identification performance, mean audibility and standard deviation of audibility across trials as a function of SNR, for each participant (faint colored lines) and averaged across all 20 participants (thick black line). Error bars represent ± standard error of the mean, n = 20 participants. The bottom row shows the distributions of audibility responses averaged across participants, for each SNR level. E Group averaged potentials evoked by the target vowels at electrode Cz and at each SNR level (n = 20 participants). F The top row shows the topographies of the event-related potentials at different time windows following targets at the highest SNR. The middle and bottom rows show the mean and variability of evoked activity as a function of SNR at these different time windows over a group of centro-parietal electrodes (highlighted on the topographies). Shaded areas correspond to ± SEM around the group average (n = 20 participants). Source data for this figure are provided as a Source data file.CC BY · PubMed Central
Fig. 2 Dynamics of the different processing stages during the active session (analyzed using multivariate pattern classification). A Illustration of the multivariate pattern classification analysis in 2D (two electrodes). Each point represents activity at electrodes n1 and n2 for a single trial, in three conditions: target absent (black), target present with maximal SNR (red), and target present around threshold SNR (green). The decision boundary (black line) is computed to maximize accuracy in classifying target present with maximal SNR versus target absent in a training subset of the data. The so-called projected activity for a specific test trial is defined as the distance between that trial’s point in the multivariate space and the decision boundary. The distributions of these projected activity measures for the different conditions are plotted along the decision axis (dotted line). The classifier performance was assessed using the AROC of the distribution for target present against the target absent distribution (see “Methods”). B Time-generalization analysis. The classification accuracy is shown for each combination of training time point and testing time point. Black contours indicate periods where classification deviated significantly from chance at p corrected <0.05 (two-sided non-parametric sign test76, Benjamini–Hochberg FDR correction). C Dynamics of neural activity at selected time windows (represented by squares on the top panel in (B)). The left column represents the group average topographies at each time window for the strongest SNR (−5 dB), the next column shows the group average time courses for the different SNRs, i.e., the sensitivity of the classifier trained at a given time window (indicated by the shaded area and black arrow) and tested across the whole duration of a trial. The third column shows mean projected activity as a function of SNR for each time window of interest (arbitrary unit). The last column shows the variability of projected activity across trials as a function of SNR, i.e., the standard deviation of the projected activity across trials (arbitrary unit). Error bars correspond to ± SEM around the group average (n = 20 participants). Source data for this figure are provided as a Source data file.
Fig. 2 Dynamics of the different processing stages during the active session (analyzed using multivariate pattern classification). A Illustration of the multivariate pattern classification analysis in 2D (two electrodes). Each point represents activity at electrodes n1 and n2 for a single trial, in three conditions: target absent (black), target present with maximal SNR (red), and target present around threshold SNR (green). The decision boundary (black line) is computed to maximize accuracy in classifying target present with maximal SNR versus target absent in a training subset of the data. The so-called projected activity for a specific test trial is defined as the distance between that trial’s point in the multivariate space and the decision boundary. The distributions of these projected activity measures for the different conditions are plotted along the decision axis (dotted line). The classifier performance was assessed using the AROC of the distribution for target present against the target absent distribution (see “Methods”). B Time-generalization analysis. The classification accuracy is shown for each combination of training time point and testing time point. Black contours indicate periods where classification deviated significantly from chance at p corrected <0.05 (two-sided non-parametric sign test76, Benjamini–Hochberg FDR correction). C Dynamics of neural activity at selected time windows (represented by squares on the top panel in (B)). The left column represents the group average topographies at each time window for the strongest SNR (−5 dB), the next column shows the group average time courses for the different SNRs, i.e., the sensitivity of the classifier trained at a given time window (indicated by the shaded area and black arrow) and tested across the whole duration of a trial. The third column shows mean projected activity as a function of SNR for each time window of interest (arbitrary unit). The last column shows the variability of projected activity across trials as a function of SNR, i.e., the standard deviation of the projected activity across trials (arbitrary unit). Error bars correspond to ± SEM around the group average (n = 20 participants). Source data for this figure are provided as a Source data file.CC BY · PubMed Central
Fig. 3 Correlation with behavior, Bayesian model comparison, and prediction of conscious reports in the active session. A Predicted distributions of activity across trials as a function of SNR level for unimodal non-linear and bifurcation models (left and middle) compared with distributions of multivariate neural activity observed within the 400–500 ms time window for participant 14. B Grand average evoked potentials at electrodes POz and Pz for stimuli at threshold (−7 dB), shown separately for “heard” and “not heard” trials. Shaded area shows SEM (n = 20 participants), gray dots represent time points where the difference is significant at p-corrected <0.05 (two-sided paired t-tests, FDR correction). Below are shown the corresponding topographies at various time windows. C Mean activity as a function of stimulus strength, shown separately for “heard” versus “not-heard” trials. Model predictions for the unimodal model (left) and bifurcation model (middle) are compared to the mean projected activity that was actually observed for “heard” and “not heard” trials during the 300–400 ms time window (grand average across participants, arbitrary unit, error bars represent SEM, n = 20). D Correlation coefficient between the individual mean profile (blue) or variability profile (red) of projected neural activity with the corresponding mean or variability profile in subjective audibility. The shaded areas correspond to SEM (n = 20 participants). The periods of significant correlation are denoted as thick lines of the corresponding color (one-sided t-tests, p < 0.05, FDR corrected). E Bayesian model comparison: protected exceedance probability at the group level for the unimodal non-linear, bifurcation, and null models. F Prediction of behavioral audibility from the neural projected activity under the assumption of bifurcation dynamics. Group average performance in predicting trial-by-trial audibility responses (heard versus not-heard) is shown for intermediate SNRs (−11, −9, and −7 dB). Thick dots denote periods were prediction performance was better than chance (two-sided t-tests, p < 0.05, FDR corrected). The shaded areas correspond to SEM (n = 20 participants). Source data for this figure are provided as a Source data file.
Fig. 3 Correlation with behavior, Bayesian model comparison, and prediction of conscious reports in the active session. A Predicted distributions of activity across trials as a function of SNR level for unimodal non-linear and bifurcation models (left and middle) compared with distributions of multivariate neural activity observed within the 400–500 ms time window for participant 14. B Grand average evoked potentials at electrodes POz and Pz for stimuli at threshold (−7 dB), shown separately for “heard” and “not heard” trials. Shaded area shows SEM (n = 20 participants), gray dots represent time points where the difference is significant at p-corrected <0.05 (two-sided paired t-tests, FDR correction). Below are shown the corresponding topographies at various time windows. C Mean activity as a function of stimulus strength, shown separately for “heard” versus “not-heard” trials. Model predictions for the unimodal model (left) and bifurcation model (middle) are compared to the mean projected activity that was actually observed for “heard” and “not heard” trials during the 300–400 ms time window (grand average across participants, arbitrary unit, error bars represent SEM, n = 20). D Correlation coefficient between the individual mean profile (blue) or variability profile (red) of projected neural activity with the corresponding mean or variability profile in subjective audibility. The shaded areas correspond to SEM (n = 20 participants). The periods of significant correlation are denoted as thick lines of the corresponding color (one-sided t-tests, p < 0.05, FDR corrected). E Bayesian model comparison: protected exceedance probability at the group level for the unimodal non-linear, bifurcation, and null models. F Prediction of behavioral audibility from the neural projected activity under the assumption of bifurcation dynamics. Group average performance in predicting trial-by-trial audibility responses (heard versus not-heard) is shown for intermediate SNRs (−11, −9, and −7 dB). Thick dots denote periods were prediction performance was better than chance (two-sided t-tests, p < 0.05, FDR corrected). The shaded areas correspond to SEM (n = 20 participants). Source data for this figure are provided as a Source data file.CC BY · PubMed Central
Fig. 4 Comparison of neural activity in the active and passive sessions. A Group averaged topographies at various time windows following stimulus presentation in the active and passive sessions at −5 dB. Group averaged evoked potentials at Cz for the different SNRs in the active and the passive sessions. B Classification scores (AUC) for the cross-classification analysis, at an SNR of −5 dB. Regions where classification was significantly different from chance are outlined in black (two-sided non-parametric sign test76, p < 0.05, FDR-corrected). C Source reconstruction at 330 ms post-stimulus. The activation maps show the t-values of a paired t-test across subjects between mean activations for stimulus present at −5 dB versus stimulus absent, in the active or passive sessions (left and middle) and between mean activations for the active versus passive session for stimulus present at −5 dB (right). Only significant activations are shown (p < 0.05, FDR corrected). D Group averaged time course of source reconstruction in three regions of interest, shown for the active or passive sessions (SNR = −5 dB as above). Thick gray dots indicate periods of significant difference (two-sided paired t-tests, p < 0.05, FDR corrected). Source data for this figure are provided as a Source data file.
Fig. 4 Comparison of neural activity in the active and passive sessions. A Group averaged topographies at various time windows following stimulus presentation in the active and passive sessions at −5 dB. Group averaged evoked potentials at Cz for the different SNRs in the active and the passive sessions. B Classification scores (AUC) for the cross-classification analysis, at an SNR of −5 dB. Regions where classification was significantly different from chance are outlined in black (two-sided non-parametric sign test76, p < 0.05, FDR-corrected). C Source reconstruction at 330 ms post-stimulus. The activation maps show the t-values of a paired t-test across subjects between mean activations for stimulus present at −5 dB versus stimulus absent, in the active or passive sessions (left and middle) and between mean activations for the active versus passive session for stimulus present at −5 dB (right). Only significant activations are shown (p < 0.05, FDR corrected). D Group averaged time course of source reconstruction in three regions of interest, shown for the active or passive sessions (SNR = −5 dB as above). Thick gray dots indicate periods of significant difference (two-sided paired t-tests, p < 0.05, FDR corrected). Source data for this figure are provided as a Source data file.CC BY · PubMed Central
Fig. 5 Neural dynamics for passive sessions. A The top row shows the group averaged topographies at different time windows following targets played at −5 dB SNR. The middle and bottom rows show the mean and variability of evoked potentials as a function of SNR, for different time windows and over a group of temporal electrodes highlighted on the topographies. Shaded areas are SEMs (n = 20 participants, from which 10 at −3 dB). B Correlation coefficient between the individual mean profiles (blue) or variability profiles (red) of projected activity in the passive session with the individual mean profiles or variability profiles of behavioral audibility during the active session (shifted by 1 or 2 SNR levels). The shaded areas are SEMs (n = 20 participants). Periods of significant correlation are denoted in thick lines of the corresponding color (one-sided t-tests, p < 0.05, FDR corrected). C Prediction of mind-wandering content from the neural projected activity under the assumption of bifurcation dynamics. The graph shows the group average performance in predicting, trial by trial, whether the participant responded that the sound was on their mind at the moment of the probe or not. All SNRs are included in this analysis. The shaded areas are SEMs (n = 20 participants). Thick dots denote periods were prediction performance was better than chance (two-sided t-tests, p < 0.05, FDR corrected). Source data for this figure are provided as a Source data file.
Fig. 5 Neural dynamics for passive sessions. A The top row shows the group averaged topographies at different time windows following targets played at −5 dB SNR. The middle and bottom rows show the mean and variability of evoked potentials as a function of SNR, for different time windows and over a group of temporal electrodes highlighted on the topographies. Shaded areas are SEMs (n = 20 participants, from which 10 at −3 dB). B Correlation coefficient between the individual mean profiles (blue) or variability profiles (red) of projected activity in the passive session with the individual mean profiles or variability profiles of behavioral audibility during the active session (shifted by 1 or 2 SNR levels). The shaded areas are SEMs (n = 20 participants). Periods of significant correlation are denoted in thick lines of the corresponding color (one-sided t-tests, p < 0.05, FDR corrected). C Prediction of mind-wandering content from the neural projected activity under the assumption of bifurcation dynamics. The graph shows the group average performance in predicting, trial by trial, whether the participant responded that the sound was on their mind at the moment of the probe or not. All SNRs are included in this analysis. The shaded areas are SEMs (n = 20 participants). Thick dots denote periods were prediction performance was better than chance (two-sided t-tests, p < 0.05, FDR corrected). Source data for this figure are provided as a Source data file.CC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 89 references

Research works printed but not linked: 1

  1. Luck, S. J. An Introduction to the event-related Potential Technique. Second Edition, pp. x, 406 (The MIT Press, 2014).

Cited by

See all 227

Metadata

Authors
Sergent, C., Corazzol, M., Labouret, G., Stockart, F., Wexler, M., King, J.-R., Meyniel, F., & Pressnitzer, D.
Affiliations
  • Centre National de la Recherche Scientifique (FR)
  • Université Paris-Sud (FR)
Corresponding author
Sergent, Claire
Venue
Nature Communications
Published
2021-02-19
Volume
12
Issue
1
Pages
1149-1149
ISSN
2041-1723 (ISSN-L)
OpenAlex
W3131701397
Cited by
200
References
89
Citations by year
2026: 29 · 2025: 41 · 2024: 57 · 2023: 38 · 2022: 26 · 2021: 9
FWCI
17.09
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature Communications (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • HAL (submittedVersion, other-oa)
  • RePEc: Research Papers in Economics (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Funders
Agence Nationale de la Recherche (ANR-17-CE37-0004-01, ANR-17-EURE-0017, ANR-10-IDEX-0001-02)
Topics
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Neural dynamics and brain function
  • Social Sciences › Arts and Humanities › Philosophy › Philosophy and Theoretical Science
  • Social Sciences › Arts and Humanities › Philosophy › Free Will and Agency
Keywords
conscious access, bifurcation dynamics, trial-to-trial variability, EEG, source localization, model-based analysis
MeSH terms
Cognitive Neuroscience, Acoustic Stimulation, Adolescent, Adult, Auditory Perception, Behavior, Brain, Consciousness, Electroencephalography, Female, Humans, Male, Visual Perception, Young Adult

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Bifurcation in brain dynamics reveals a signature of conscious processing independent of report. https://app.cochat.ai/papers/10.1038/s41467-021-21393-z
The paper
Sergent, C., Corazzol, M., Labouret, G., Stockart, F., Wexler, M., King, J.-R., Meyniel, F., & Pressnitzer, D. (2021). Bifurcation in brain dynamics reveals a signature of conscious processing independent of report. Nature Communications. https://doi.org/10.1038/s41467-021-21393-z

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.