IIT vs Global Workspace: What the Adversarial Test Found
For the first time, rival theories of consciousness agreed in advance on experiments that could count against them. Neither came out unscathed.
What an adversarial collaboration is
In an adversarial collaboration, proponents of competing theories jointly design experiments and pre-register what each theory predicts, so that results cannot easily be explained away afterwards. The Cogitate consortium, supported by the Templeton World Charity Foundation, applied this approach to Integrated Information Theory and Global Workspace Theory.
The predictions tested
| Question | IIT predicted | GWT predicted |
|---|---|---|
| Where is conscious content represented? | Mainly posterior cortex | Prefrontal cortex involved |
| What happens while a stimulus stays visible? | Sustained posterior activity and synchrony | Ignition at onset and at offset |
| How do regions communicate? | Sustained posterior synchronisation | Long-range prefrontal–posterior exchange |
Participants viewed images for different durations while their brain activity was recorded with fMRI, MEG and intracranial electrodes in several laboratories.
What was found
The results, published in Nature in 2025, did not crown a winner. Information about conscious content was found in posterior regions, as IIT expected, and less clearly in prefrontal cortex than GWT expected; the predicted “offset ignition” of GWT was largely absent. But the sustained synchronisation in posterior cortex that IIT predicted was not clearly observed either. Each side has argued about how far the results damage its theory.
What it means
The experiment showed that theories of consciousness can be put at real empirical risk, which is progress in itself. It also showed how much interpretation still matters: both theories have core claims these experiments could not reach. Further adversarial collaborations, including tests involving higher-order and other theories, are under way.
Editorial note: this is a fast-moving area. We revise this briefing as new analyses and replies are published.
Sources
- Cogitate Consortium et al. (2025). Adversarial testing of global neuronal workspace and integrated information theories of consciousness. Nature.
- Melloni, L. et al. (2023). An adversarial collaboration protocol for testing contrasting predictions of global neuronal workspace and integrated information theory. PLOS ONE, 18(2).