The 2023/2025 adversarial collaboration between proponents of Integrated Information Theory (IIT) and Global Neuronal Workspace Theory (GNWT), run by the theory-neutral Cogitate Consortium with 256 human subjects and fMRI, MEG, and intracranial EEG, confirmed some predictions of each theory while substantially challenging both: conscious content could not be linked to sustained synchronization within posterior cortex (against IIT), and the predicted prefrontal ignition at stimulus offset, as well as prefrontal representation of low-level content such as orientation, were largely absent (against GNWT). The collaboration itself listed the issues it could not settle. As of August 2026, all of those issues remain open: the data are purely correlational so the causal necessity of prefrontal cortex for consciousness is untested; the discrepant predictions are contested between reinterpretation and genuine failure; and no accepted quantitative framework integrates the multimodal evidence or tests IIT's measure of integrated information. The problem is to resolve these linked questions and thereby complete the adjudication between the two leading theories of the neural mechanisms of consciousness.

Background

Consciousness research asks how subjective experience arises from brain activity. Two leading theories make competing claims. Integrated Information Theory (IIT) holds that the substrate of experience is a network of neurons in the posterior cortex — the 'posterior hot zone' spanning early visual and temporo-parietal areas — whose integrated cause-effect structure specifies the content of experience; it predicts that conscious content should be decodable from posterior areas, actively maintained there throughout the experience, and supported by sustained short-range synchronization within the posterior cortex. The theory's core quantity is integrated information, written φ\varphi (phi), a measure of how much a system's causal structure exceeds that of its parts; φ\varphi is defined mathematically but is computationally intractable for real neural systems and has never been directly measured in the experiments designed to test the theory. Global Neuronal Workspace Theory (GNWT) instead holds that a mental content becomes conscious when it triggers a nonlinear 'ignition' — a sudden, self-sustaining burst of recurrent activity — in a global neuronal workspace distributed over prefrontal and parietal cortex, which then broadcasts the content to memory, evaluation, and report systems; it predicts prefrontal ignition at stimulus onset and offset, with content held between these events in an activity-silent ('silent') state.

To arbitrate between such theories without confirmation bias, the Templeton World Charity Foundation funded structured adversarial collaborations in which theory proponents jointly preregister the design, the divergent predictions, and the pre-approved interpretations before data collection, and an independent consortium runs the experiment. The first of these, by the Cogitate Consortium, tested three preregistered predictions of IIT and GNWT about (1) where conscious content is represented, (2) how it is maintained over time, and (3) the functional connectivity accompanying conscious perception: 256 human subjects viewed suprathreshold stimuli of variable durations while neural activity was recorded with functional magnetic resonance imaging (fMRI, which localizes activity indirectly via blood oxygenation), magnetoencephalography (MEG, millisecond-scale magnetic fields recorded outside the head), and intracranial electroencephalography (iEEG, electrodes placed directly on or in the brains of epilepsy patients, here including 472 prefrontal but only 12 V1/V2 early-visual electrodes). The outcome, published as a preprint in 2023 and in Nature in 2025, confirmed some predictions of each theory while substantially challenging both: for IIT, a lack of sustained synchronization within posterior cortex contradicts the claim that network connectivity specifies consciousness; for GNWT, there was a general lack of ignition at stimulus offset and only limited prefrontal representation of certain conscious dimensions, such as the orientation of a face. Both theories survived, and the consortium explicitly declined to issue a verdict.

The published work and its discussion leave a set of linked unresolved issues. All results are correlational, so whether prefrontal cortex is causally necessary for consciousness — rather than for reporting, attending, or acting on conscious content — remains untested; the relevant manipulations (electrical stimulation or lesions in animal models, optogenetic silencing) had not been performed within this program. The failed predictions are ambiguous: absent posterior synchronization could reflect sparse clinical sampling of early visual cortex, and absent offset ignition could reflect the fact that stimulus offsets were task-irrelevant and possibly not consciously attended. Adjacent no-report work (paradigms in which subjects do not report each stimulus, avoiding report confounds) keeps the frontal-versus-sensory dispute alive without settling it, and the field's own 2025 state-of-the-discipline review concludes that the agreed experiments did not directly test the core commitments of either theory and that a formal quantitative framework for weighing predictions across analyses and modalities is still lacking.

Problem Statement

The problem is to complete the empirical adjudication between Integrated Information Theory (IIT) and Global Neuronal Workspace Theory (GNWT) that the Cogitate adversarial test left unfinished, by resolving its linked open questions with adequate causal and measurement power:

  1. Causal necessity of prefrontal cortex. Establish, by causal intervention (electrical stimulation or lesions in animal models, optogenetic silencing, or comparable manipulations in humans), whether prefrontal cortex is causally necessary for conscious visual experience itself, or only for report, task-relevance, and abstract categorical aspects of content — the 2025 results were purely correlational.

  2. Genuine failure or methodological artifact. Determine whether the discrepant predictions — the absence of sustained synchronization within posterior cortex (challenging IIT's claim that network connectivity specifies conscious content) and the absence of prefrontal ignition at stimulus offset together with weak prefrontal representation of low-level features such as orientation (challenging GNWT) — reflect genuine failures of the theories' mechanisms or limitations of the experiment: sparse intracranial sampling of early visual cortex (12 V1/V2 versus 472 prefrontal electrodes), stimulus offsets that were task-irrelevant and possibly not consciously attended, and macro-electrode resolution insufficient for fine-grained prefrontal feature coding.

  3. Maintenance of conscious content. Explain how a fundamental feature of visual experience such as orientation is neurally maintained over the duration of a conscious percept, given that its representation in posterior cortex was weak and non-sustained and its prefrontal representation was largely absent.

  4. Framework and the untested IIT core. Develop the formal quantitative framework the consortium called 'urgently needed' to integrate evidence across predictions, analysis pipelines, and neuroimaging modalities (fMRI, MEG, iEEG, single units) in a way that goes beyond pass/fail tallies and is sensitive to prediction centrality, measurement error, and cross-sample consistency; and use that framework to address the gap the collaboration's own discussion records — IIT's mathematical core, integrated information (φ\varphi), was not put to experimental test in these experiments, the paper noting that the mathematical or computational core of either theory is difficult to test directly while the competing biological implementations are empirically testable — so that φ\varphi and its derived predictions are brought within reach of genuine experimental test.

All four numbered sub-questions are mandatory components of a complete answer, and the closing synthesis draws on all four rather than standing in for any of them. Where the causal-intervention evidence required by sub-question 1 has not yet been obtained, that sub-question is still resolved by assembling the intervention evidence that exists and stating what it establishes or leaves open, together with the concrete manipulation that would settle it; no new experiment needs to be performed.

A complete answer states, with the evidence that supports each claim, which of the postulated mechanisms — sustained, synchronized posterior activity specifying content versus prefrontal ignition and global broadcast — are actually present during conscious perception, which are causally necessary for it, and consequently whether IIT, GNWT, both, or neither survives as an adequate theory of the neural basis of conscious experience.

References

  1. Cogitate Consortium, Ferrante O, Gorska-Klimowska U, Henin S, Hirschhorn R, Khalaf A, Lepauvre A, Liu L, Richter D, Vidal Y, et al. (including Dehaene S, Koch C, Tononi G, Mudrik L, Melloni L). Adversarial testing of global neuronal workspace and integrated information theories of consciousness. Nature 642, 133–142 (2025). doi:10.1038/s41586-025-08888-1, https://doi.org/10.1038/s41586-025-08888-1
  2. Cogitate Consortium. An adversarial collaboration to critically evaluate theories of consciousness. bioRxiv preprint (2023). doi:10.1101/2023.06.23.546249, https://doi.org/10.1101/2023.06.23.546249
  3. Melloni L, Mudrik L, Pitts M, et al. An adversarial collaboration protocol for testing contrasting predictions of global neuronal workspace and integrated information theory. PLOS ONE 18, e0268577 (2023). doi:10.1371/journal.pone.0268577, https://doi.org/10.1371/journal.pone.0268577
  4. Naccache L, Sergent C, Dehaene S, Wang X-J, Farisco M, Changeux J-P. GNW theoretical framework and the 'adversarial testing of global neuronal workspace and integrated information theories of consciousness'. Neuroscience of Consciousness 2025, niaf037 (2025). doi:10.1093/nc/niaf037, https://doi.org/10.1093/nc/niaf037 (correction: doi:10.1093/nc/niag020)
  5. Bryant AG, Whyte CJ. A data-driven approach to identifying and evaluating connectivity-based neural correlates of conscious visual perception. Neuroscience of Consciousness 2026, niag029 (2026). doi:10.1093/nc/niag029, https://doi.org/10.1093/nc/niag029
  6. Ponce de Leon S, Yoshimi J. Integrated information theory (IIT) and the testability of the silent neuron predictions. Neuroscience of Consciousness 2026, niag037 (2026). doi:10.1093/nc/niag037, https://doi.org/10.1093/nc/niag037
  7. Gibbons M, McBride EG, et al. Protocol for testing global neuronal workspace and integrated information theories of consciousness in non-human primates and mice. PLOS ONE 21, e0342770 (2026). doi:10.1371/journal.pone.0342770, https://doi.org/10.1371/journal.pone.0342770
  8. He L, Zhang Z, Wang Y, Meng Q, Jiang Y, Xu W, Wu J, Sheng J, Ye Z, Ding X. Report-modulated prefrontal activity and consistent posterior representations during conscious visual perception. Consciousness and Cognition 142, 104075 (2026). doi:10.1016/j.concog.2026.104075, https://doi.org/10.1016/j.concog.2026.104075
  9. Bandara KH, Rowe EG, Garrido MI. Computational modelling shows evidence in support of both sensory and frontal theories of consciousness. Neuroscience of Consciousness 2025, niaf033 (2025). doi:10.1093/nc/niaf033, https://doi.org/10.1093/nc/niaf033
  10. Cleeremans A, Mudrik L, Seth AK. Consciousness science: where are we, where are we going, and what if we get there? Frontiers in Science 3, 1546279 (2025). doi:10.3389/fsci.2025.1546279, https://doi.org/10.3389/fsci.2025.1546279
  11. Seedat A, Lepauvre A, Jeschke J, Gorska-Klimowska U, Armendariz M, Bendtz K, et al. Open multi-center intracranial electroencephalography dataset with task probing conscious visual perception. Scientific Data 12, 854 (2025). doi:10.1038/s41597-025-04833-z, https://doi.org/10.1038/s41597-025-04833-z
  12. Panagiotaropoulos TI. An integrative view of the role of prefrontal cortex in consciousness. Neuron 112, 1626–1641 (2024). doi:10.1016/j.neuron.2024.04.028, https://doi.org/10.1016/j.neuron.2024.04.028
  13. Raccah O, Block N, Fox KCR. Does the prefrontal cortex play an essential role in consciousness? Insights from intracranial electrical stimulation of the human brain. The Journal of Neuroscience 41, 2076–2087 (2021). doi:10.1523/JNEUROSCI.1141-20.2020, https://doi.org/10.1523/JNEUROSCI.1141-20.2020
  14. Potgieter D. Accelerating Research on Consciousness: The Structured Adversarial Collaboration Process. PsyArXiv preprint (2024). doi:10.31234/osf.io/rdq52, https://doi.org/10.31234/osf.io/rdq52