"Pseudoscientific" Theory Correctly Predicts Location of Consciousness

A new round of brain-imaging tests is forcing a rare, concrete confrontation between two rival theories of consciousness—Integrated Information Theory (IIT) and Global Neuronal Workspace (GNW)—and the results point to a partial win for both, with neither theory landing cleanly.

IIT, associated with neuroscientist Julia Toni, ties consciousness to a single quantity often written as “phi” (Φ): the more integrated connections a system has, the more conscious it is. That framing implies that even simple systems—like electronic circuits—could be “a little bit conscious,” pushing IIT toward panpsychism. Toni has also pursued practical applications, including a patent for using IIT in a “consciousness beta,” and reporting has linked IIT proponents to efforts to translate the theory into clinical settings.

The controversy around IIT isn’t just philosophical. Critics argue that Φ is effectively impossible to calculate for real brains, because it would require evaluating connections across all possible partitions of a system. That computational burden forces researchers to simplify the theory, which critics say makes its predictions ambiguous. In 2023, more than 100 researchers published an open letter asserting that, given IIT’s panpsychic commitments, the theory as a whole still lacks empirical testability and therefore earns the “pseudoscience” label. A later commentary in Nature Neuroscience echoed the concern, arguing that IIT’s core claims are untestable even in principle.

IIT supporters counter that the theory is objective and testable, and that labeling it pseudoscientific reflects a deeper dispute about the dominant “computational functionalist” paradigm. The debate also raises a broader scientific question: is it fair to demand that every foundational axiom be directly testable? Some argue that physics often works by deriving testable consequences from assumptions that themselves aren’t measured directly.

Against that backdrop, researchers recently extracted predictions from IIT and GNW and tested them in human participants. GNW predicts that conscious access depends on rapid, front-of-brain activity—especially involving the prefrontal cortex—broadcasting signals across the rest of the brain. IIT instead predicts that conscious perception should correlate with steady, complex, tightly connected activity in the back of the brain.

The strongest signals tied to conscious perception appeared in the back of the brain, aligning with IIT’s localization claim that consciousness is generated in posterior processing. But IIT’s more specific expectation—unusually complex, tightly integrated patterns—did not show up as predicted. GNW’s signature front-burst was also largely absent, failing its central timing prediction. Still, GNW did get something important right: conscious perception recruited more brain areas than unconscious processing.

Taken together, the findings suggest that consciousness may hinge more on sensory signal processing in posterior regions than on executive-style control from the prefrontal cortex. Just as importantly, the exchange signals that consciousness research is moving from abstract disputes toward measurable, falsifiable predictions—even if the dream of reducing consciousness to a single number remains contentious.