Wiki

Wiki: Consciousness theory map

Before naming a theory winner, separate prediction families from the four gates that change claim strength.

Mind Uploading Research Project

Public Page Updated: 2026-03-27 Technical / natural science only

How to use this page

Read this first to avoid getting lost

This page is a technical map for consciousness-related theories and readouts that recur across Mind-Upload. It does not ask which theory sounds most attractive. It asks what role each item actually plays: prediction family, confound-control design, perturbation benchmark, spontaneous proxy, deployability panel, or auxiliary geometry analysis.

  • The page now separates theory families, construct-validity controls, perturbation benchmarks, spontaneous proxies, multimodal clinical panels, and geometry descriptors rather than letting them blur together.
  • The same awareness-related result changes meaning depending on whether it has passed construct validity, perturbational validity, same-cohort calibration, and incremental validity.
  • No-report and criterion placement are treated as design conditions, not as bedside meters.
  • PCI / PCI-ST remains the main benchmark candidate only when stimulation, sensory-control, artifact-window, and reliability logs are exposed.
  • Resting-state complexity / criticality can correlate with PCI and still fail to substitute for it inside the same cohort.
  • Multimodal clinical panels are kept separate from theory and benchmark claims because their value turns on gain beyond behavior, site transfer, and missing-modality robustness.
Helpful next pages
Perspective Roadmap Verification
Best for
Readers who want IIT, GNWT, FEP, PCI, resting-state complexity, and multimodal consciousness readouts organized without overclaiming
Reading time
12-18 minutes
Accuracy note
The site-wide reading rule used here is synchronized to the March 27, 2026 re-audit: claim strength depends not only on which metric is named, but also on which verification gate it has passed.

Relatively clear at this stage

What we know now

  • Ferrante et al. (2025) did not produce a single winning theory of consciousness.
  • Post-perceptual processing and criterion placement can materially shift neural markers, so construct validity must be audited separately.
  • PCI / PCI-ST is a serious perturbation benchmark candidate, but sensory contamination and target-specific reliability remain explicit failure modes.
  • Resting-state complexity / criticality remains a same-cohort calibration track rather than a direct replacement for perturbation benchmarks.
  • Multimodal diagnosis / prognosis panels can outperform behavior-only baselines, but portability still depends on external-site generalization and missing-modality robustness.
  • Manifold / topology results remain auxiliary because sampling and preprocessing sensitivity still change the readout.

Still unresolved beyond this point

What we still do not know

  • It remains unresolved which theory family, if any, will supply a sufficient condition for WBE-relevant consciousness claims.
  • It remains unresolved whether any spontaneous metric will become a stable substitute for perturbation benchmarks across altered states and disorders of consciousness.
  • It remains unresolved which multimodal bundle is strong enough to travel across centres without recreating hidden calibration drift.
  • It remains unresolved whether geometry descriptors will become robust clinical indicators rather than auxiliary summaries.

Learn the basics

Check the basics in the wiki

What the wiki is for

The wiki is a learning aid. For the project's official current synthesis, success criteria, and operating rules, always return to the public pages.

Short conclusion for 2026-03-27

The current primary literature does not support naming a single winner among IIT, GNWT, FEP, PCI, criticality, or multimodal clinical panels. What it supports more strongly is a separation of roles. Theory names are prediction families. No-report and criterion placement are construct-validity controls. PCI / PCI-ST is a perturbation benchmark candidate. Resting-state complexity / criticality is a same-cohort calibration track. Multimodal clinical bundles are an incremental-validity / deployability track. Geometry and topology remain auxiliary analysis. On this site, those roles are not interchangeable.

2026-03-27 re-audit: what this page had to stop doing

The previous version already separated theory names from benchmarks, but it still left one important compression in place: construct validity, perturbational validity, same-cohort calibration, and incremental validity were too easy to read as one single ladder. That compression is not supported by the primary literature. The page also contained reference drift, including an incorrect DOI for the criterion-placement paper and an incorrect DOI for the spontaneous-versus-perturbational dissociation paper. This revision fixes both the conceptual compression and the citation integrity problem.

First, divide by role

Item Role What the current primary literature supports How this site reads it
IIT Theory family / prediction generator It gives predictions about integration and causal structure, but it is not itself a clinical or engineering benchmark. Treated as a prediction family, never as a standalone pass/fail meter.
GNWT Theory family / prediction generator It gives predictions about global availability and task relevance, but its markers are sensitive to report and post-perceptual confounds. Treated as a prediction family that must pass no-report / criterion-placement control before its markers are read strongly.
FEP / Active Inference Generative-model / control family It offers a broad modeling framework for perception, action, and uncertainty, but it is not by itself a winning consciousness benchmark. Treated as a candidate implementation family that must still compete under shared verification conditions.
No-report / criterion placement Construct-validity design condition It helps separate conscious content from post-report or response-strategy effects, but it does not become a bedside meter on its own. Treated as the construct-validity gate, not as a consciousness readout.
PCI / PCI-ST Perturbation benchmark candidate Response complexity after perturbation remains one of the strongest benchmark candidates across altered states and disorders of consciousness. Treated as the main benchmark candidate only when perturbation logs, sensory control, artifact windows, and reliability logs are exposed.
Resting-state complexity / criticality Spontaneous proxy / calibration track It can track altered states and sometimes align with PCI, but dissociations within the same cohort remain real. Treated as a same-cohort calibration track, not as a drop-in replacement for perturbation benchmarks.
Multimodal clinical panel Incremental-validity / deployability track Panels combining behavior, EEG, MRI, PET, and clinical variables can improve diagnosis or prognosis beyond behavior alone. Treated as a deployability track that must beat the behavior-only baseline, survive missing-modality slices, and generalize across sites.
Manifold / topology / geometry Auxiliary descriptor These analyses can summarize structure missed by scalar metrics, but their output is sensitive to sampling, embedding, and preprocessing choices. Treated as auxiliary analysis rather than an independent pass/fail indicator.

What the primary literature cuts apart

Question What the current primary literature supports relatively strongly What it still does not justify
Theory competition Ferrante et al. (2025) showed partial support and partial falsification across IIT, GNWT, and RPT predictions rather than a single winner. No theory family can yet be promoted to the default engineering truth condition for WBE.
Construct validity Cohen et al. (2024), Fahrenfort et al. (2025), and Dellert et al. (2025) show that post-perceptual processing, response criterion, and modality-general no-report design all materially change how a marker should be read. No-report or report-independent design alone does not become a bedside consciousness meter.
Perturbational validity Casali et al. (2013) and Comolatti et al. (2019) support PCI-family metrics as serious perturbation benchmarks, while Hernandez-Pavon et al. (2023), Gogulski et al. (2024), and Biabani et al. (2024) show why stimulation conditions, sensory contamination, and target-specific reliability must be exposed. A PCI-like value without perturbation logs, sensory controls, artifact windows, and reliability evidence cannot be treated as a universal state meter.
Same-cohort calibration Maschke et al. (2024) linked spontaneous criticality with perturbational complexity during anesthesia, Casarotto et al. (2024) showed spontaneous-versus-perturbational dissociation in MCS, and Breyton et al. (2025) extended spatiotemporal complexity outside perturbation paradigms. Resting-state or passive metrics still cannot replace perturbation benchmarks without same-cohort calibration against PCI, behavior, and outcomes.
Incremental validity / deployability Rohaut et al. (2024) and Manasova et al. (2026) show that multimodal panels can improve diagnosis / prognosis and multicentre robustness when compared against behavior-only baselines. A multimodal panel is not yet portable clinical infrastructure on average accuracy alone; missing-modality robustness and site transfer still matter.
Geometry / topology Yoon et al. (2024) and recent mouse-visual-cortex topology work show that manifold and topology methods can summarize population structure and cross-population relationships. Those descriptors do not by themselves guarantee conscious-state validity, deployability, or WBE-relevant sufficiency.

The 4 gates that change claim strength on this site

Gate Minimum requirement Claim that must stop if the gate is missing
Construct validity Separate no-report from report, log criterion placement separately, and predefine failure conditions. Do not claim that a neural marker directly reads conscious content.
Perturbational validity Expose perturbation site, intensity, sham or control, sensory suppression route, artifact window, analysis window, and target-specific reliability. Do not call a PCI / PCI-ST-style value alone a stable state-level benchmark.
Same-cohort calibration Calibrate spontaneous metrics against PCI, behavior, clinical outcomes, and pipeline sensitivity inside the same cohort. Do not present resting-state complexity / criticality as a standalone bedside meter.
Incremental validity Show gain beyond the behavior-only baseline, site transfer, robustness under missing-modality conditions, and calibration error. Do not present a multimodal panel as deployable or portable on a headline accuracy alone.

Operating rules in Mind-Upload

Rule

  • Theory names do not pass or fail by themselves: IIT, GNWT, and FEP stay at the prediction-family level.
  • No-report and criterion placement are design controls, not meters: they raise or lower construct validity, not clinical deployability.
  • PCI requires a perturbation log: if stimulation, sensory control, artifact window, or reliability is hidden, claim strength stops early.
  • Spontaneous indicators must be calibrated, not substituted: alignment with PCI in one context does not erase dissociation in another.
  • Multimodal panels must beat behavior on the right slice: gain, calibration, site transfer, and missing-modality robustness are all part of the evidence.
  • Geometry remains auxiliary: manifold and topology analyses can enrich interpretation, but they do not independently settle consciousness claims.
How this page connects back to the public pages

Perspective uses these distinctions to explain why consciousness-related evidence must move through different gates before the wording is allowed to rise. Verification turns the same logic into an explicit checklist. Roadmap then places those gates inside the larger dependency tree of WBE measurement and verification. This page is the role map that stops those layers from collapsing into one vague idea of “better consciousness science.”

References

  1. Ferrante O, et al. Adversarial testing of global neuronal workspace and integrated information theories of consciousness. Nature. 2025. doi:10.1038/s41586-025-08888-1
  2. Cohen MA, et al. Neural signatures of visual awareness independent of postperceptual processing. Cerebral Cortex. 2024. doi:10.1093/cercor/bhae415
  3. Fahrenfort JJ, et al. Criterion placement threatens the construct validity of neural measures of consciousness. eLife. 2025. doi:10.7554/eLife.102335
  4. Dellert T, et al. Neural correlates of consciousness in an auditory no-report fMRI study. Current Biology. 2025. doi:10.1016/j.cub.2025.10.026
  5. Casali AG, et al. A theoretically based index of consciousness independent of sensory processing and behavior. Science Translational Medicine. 2013. doi:10.1126/scitranslmed.3006294
  6. Comolatti R, et al. A fast and general method to empirically estimate the complexity of brain responses to transcranial and intracranial stimulations. Brain Stimulation. 2019. doi:10.1016/j.brs.2019.05.013
  7. Hernandez-Pavon JC, et al. TMS combined with EEG: Recommendations and open issues for data collection and analysis. Brain Stimulation. 2023. doi:10.1016/j.brs.2023.02.009
  8. Gogulski J, et al. Reliability of the TMS-evoked potential in dorsolateral prefrontal cortex. Cerebral Cortex. 2024. doi:10.1093/cercor/bhae130
  9. Biabani M, et al. Characterising the contribution of auditory and somatosensory inputs to TMS-evoked potentials following stimulation of prefrontal, premotor, and parietal cortex. Imaging Neuroscience. 2024. doi:10.1162/imag_a_00349
  10. Maschke C, et al. Critical dynamics in spontaneous EEG predict anesthetic-induced loss of consciousness and perturbational complexity. Communications Biology. 2024. doi:10.1038/s42003-024-06613-8
  11. Casarotto S, et al. Dissociations between spontaneous electroencephalographic features and the perturbational complexity index in the minimally conscious state. European Journal of Neuroscience. 2024. doi:10.1111/ejn.16299
  12. Breyton M, et al. Spatiotemporal brain complexity quantifies consciousness outside of perturbation paradigms. eLife. 2025. doi:10.7554/eLife.98920
  13. Rohaut B, et al. Multimodal assessment improves neuroprognosis performance in clinically unresponsive critical-care patients with brain injury. Nature Medicine. 2024. doi:10.1038/s41591-024-03019-1
  14. Manasova D, et al. Multimodal multicentre investigation of diagnostic and prognostic markers in disorders of consciousness. Brain. 2026. doi:10.1093/brain/awaf412
  15. Yoon B, Miolane N, Osting B, Linderman SW. Tracking the topology of neural manifolds across populations. PNAS. 2024. doi:10.1073/pnas.2402628121
  16. The topological structure of population activity in mouse visual cortex encodes visual stimuli. iScience. 2024. doi:10.1016/j.isci.2024.111613