Overview
Two theories dominate the scientific study of consciousness: Integrated Information Theory (IIT), developed by Giulio Tononi, and Global Workspace Theory (GWT), proposed by Bernard Baars and extended by Stanislas Dehaene. Both attempt to explain what makes a physical system conscious, but they approach the question from fundamentally different directions.
IIT starts from the phenomenology of experience — what consciousness *is* — and derives the physical requirements. GWT starts from cognitive neuroscience — how the brain processes information — and identifies the mechanisms that correlate with conscious access.
Side-by-Side Comparison
| Dimension | IIT | GWT |
|-----------|-----|-----|
| Core claim | Consciousness = integrated information (Phi) | Consciousness = global broadcast of information |
| Starting point | Phenomenological axioms | Cognitive architecture |
| Measure | Phi (Φ) — mathematical, in principle computable | No single measure; neural ignition patterns |
| Substrate | Any system with sufficient integration | Brain-like global workspace architectures |
| Hard problem | Addressed directly (identity theory) | Largely sidestepped (focuses on access) |
| AI consciousness | Possible if architecture has high Phi | Possible if global workspace is implemented |
| Key prediction | Posterior cortex is the main correlate | Prefrontal-parietal network is the main correlate |
| Cerebellum | Low consciousness (feedforward architecture) | Not part of the global workspace |
| Testability | Difficult (Phi is computationally hard) | More empirically tractable |
| Main criticism | Unfalsifiable in practice; panpsychist implications | Doesn't explain *why* broadcasting creates experience |
IIT in Detail
IIT begins with five axioms about experience (existence, composition, information, integration, exclusion) and translates them into mathematical postulates about physical systems. The central quantity, Phi (Φ), measures how much a system's parts are informationally integrated beyond what its parts can do independently. A system is conscious to the degree that it has high Phi.
This leads to a striking prediction: consciousness is not exclusive to brains. Any physical system with the right causal architecture — even a simple one — has some degree of experience. This aligns IIT with a form of panpsychism, which many scientists find counterintuitive but which IIT's proponents argue follows necessarily from the axioms.
GWT in Detail
GWT models the brain as having a "global workspace" — a shared informational hub (associated with prefrontal and parietal cortices) that integrates and broadcasts information from specialized unconscious processors. Consciousness occurs when information wins the competition for access to this workspace and is broadcast widely, making it available for verbal report, memory, and flexible behavior.
Dehaene's extension, Global Neuronal Workspace Theory (GNWT), grounds this in specific neural mechanisms: "ignition" events where sustained, reverberant activity in prefrontal-parietal networks makes information globally available. This makes GWT highly testable through neuroimaging.
The Adversarial Collaboration
The most important development in consciousness science is the ongoing adversarial collaboration between IIT and GWT. The COGITATE consortium and related Templeton-funded projects have pre-registered experiments testing the two theories' divergent predictions. Key battlegrounds include:
- Where in the brain consciousness resides (posterior cortex vs. prefrontal-parietal)
- When neural activity becomes conscious (IIT predicts earlier, GWT predicts later ignition)
- Whether content-specific experience requires global broadcast
Early results from 2023 were mixed — neither theory was fully supported, and both may need revision. This adversarial approach represents a new standard for consciousness science: theories must make divergent, testable predictions, and the data decide.
Why This Matters
The IIT vs. GWT debate is not merely academic. Whichever theory proves more accurate will shape how we assess consciousness in non-communicative patients, how we think about animal minds, and whether we ever take seriously the possibility that artificial systems could be conscious. These are among the most consequential questions in science.
