Summary: A major multi-year study directly compared two leading theories of consciousness—Integrated Information Theory (IIT) and Global Neuronal Workspace Theory (GNWT)—but concluded that neither theory alone fully accounts for conscious experience. While IIT emphasizes deep integration across brain regions and GNWT focuses on widespread broadcasting or “ignition” of information, data from 256 human participants did not conclusively support either account.
Crucially, the results highlight the role of sensory and perceptual brain regions—especially visual processing areas—rather than the prefrontal cortex as the dominant seat of conscious content. This insight could improve methods for detecting hidden or “covert” consciousness in unresponsive patients and guide future research priorities.
Key Facts:
- No definitive winner: The study did not decisively validate Integrated Information Theory or Global Neuronal Workspace Theory.
- Emphasis on sensory regions: Evidence points to perceptual and sensory cortices as central to conscious content, more so than prefrontal control regions.
- Clinical relevance: Better understanding of where conscious content is represented may aid detection of awareness in patients with severe brain injury.
Source: Allen Institute
Background and scope
After seven years of collaborative planning and data collection, researchers have released findings that confront two prominent, competing models of consciousness. Published in Nature, the work represents an adversarial, open-science collaboration designed to pit theory-driven predictions against each other using a single, large experimental framework.
Integrated Information Theory (IIT) proposes that consciousness arises from a system’s capacity to integrate information into a single unified whole; the richer and more unified the internal causal interactions, the stronger the conscious state. Global Neuronal Workspace Theory (GNWT) argues that consciousness depends on a broadcasting mechanism: when particular information is amplified across a widespread network—often associated with frontal regions—it becomes reportable and forms conscious experience.
A preregistered, theory-neutral consortium designed tests that would generate divergent predictions from the two frameworks. In 2019 the experiment ran with 256 human participants, an unusually large sample for neuroscience studies of this kind, and results from fMRI, magnetoencephalography (MEG), and intracranial electroencephalography (iEEG) were analyzed to determine where and when visual content could be decoded in the brain.
Main findings
The study identified content-specific neural signals in visual, ventrotemporal, and inferior frontal cortex. Sustained neural responses that tracked stimulus duration were strongest in occipital and lateral temporal cortices—regions associated with visual and perceptual processing. At the same time, researchers observed synchronization between early visual areas and frontal regions that reflected content-specific interactions.
These observations suggest a more central role for sensory and perceptual cortices in representing conscious content than for prefrontal regions alone. The prefrontal cortex remains important for higher cognitive functions such as reasoning and planning, but the data indicate that the substrates of visual awareness are rooted in perceptual systems.
That distinction has important clinical implications. If conscious perception depends primarily on sensory-region representations, tests that focus on those areas may be better suited to detect covert consciousness—cases where patients who appear unresponsive nevertheless retain internal awareness. Prior reports indicate such covert awareness may be present in a significant minority of severely injured patients.
How the theories fared
Neither IIT nor GNWT emerged as a complete explanation. Some findings aligned with predictions from each theory, but other key expectations were not met. For IIT, the anticipated sustained synchronization within posterior cortex—required by some formulations of the theory—was not consistently observed. For GNWT, the expected robust “ignition” or global broadcasting at stimulus offset and strong, generalized prefrontal representations of conscious dimensions were limited.
As one consortium participant noted, no single experiment could definitively refute either theory given their different assumptions and aims, and the current limits of measurement tools. Still, the study yielded valuable, high-resolution insights into when and where visual information related to conscious perception can be decoded across multiple recording modalities.
Collaborative approach
The project originated at an Allen Institute workshop in 2018 and proceeded as an adversarial collaboration: proponents of different theories worked together to design preregistered tests, reducing confirmation bias and sharpening contrasts between theoretical predictions. This method demonstrates how principled, cooperative competition can accelerate progress in cognitive neuroscience and encourage more rigorous, theory-driven research.
About this consciousness and neuroscience research news
Author: Liz Dueweke
Source: Allen Institute
Contact: Liz Dueweke – Allen Institute
Image: The image is credited to Neuroscience News
Original Research: Open access.
“Adversarial testing of global neuronal workspace and integrated information theories of consciousness” by Christof Koch et al. Nature
Abstract
Adversarial testing of global neuronal workspace and integrated information theories of consciousness
Different theoretical frameworks propose distinct mechanisms by which subjective experience arises from brain activity. Although each has accumulated supporting evidence, direct, side-by-side comparisons have been rare. This open-science adversarial collaboration juxtaposed IIT and GNWT under a theory-neutral consortium that jointly developed and preregistered experimental designs, divergent predictions, and criteria for interpretation.
Human participants (n = 256) viewed suprathreshold visual stimuli of variable durations while brain activity was recorded using functional MRI, magnetoencephalography, and intracranial EEG. The analyses revealed decodable information about conscious visual content in visual, ventrotemporal, and inferior frontal cortices. Sustained responses in occipital and lateral temporal regions tracked stimulus duration, and content-specific synchronization was observed between frontal and early visual areas.
These results are consistent with some elements of both IIT and GNWT, yet they also challenge central claims of each. For IIT, the absence of prolonged posterior-cortex synchronization conflicts with predictions that network connectivity within posterior regions alone specifies consciousness. For GNWT, the lack of widespread ignition at stimulus offset and the sparse representation of some conscious dimensions in prefrontal cortex call key aspects of the model into question. The findings similarly constrain other theories that make related predictions.
Beyond theory evaluation, this work provides a model for rigorous, collaborative, theory-driven research in cognitive neuroscience and underscores the need for quantitative frameworks to enable systematic theory testing and refinement.