Global Workspace Theory (Baars, Dehaene) proposes that specialized modules — vision, memory, language, emotion — compete unconsciously, and consciousness is what happens when one module's activity wins enough support to "ignite": broadcast globally and briefly synchronize the whole network.
module: a_i(t+dt) = a_i + noise − decay·a_i + integration·mean(a_j≠i)
ignition: if max(a_i) > threshold → broadcast winner to all modules
Φ* ≈ integration · synchrony · (ignited fraction) (a simplified proxy, not literal IIT Φ)
- Cognitive modules — how many specialized processors compete for global access.
- Integration strength — how strongly modules influence each other; low values isolate them (like a split-brain or under anesthesia).
- Ignition threshold — how much activation a module needs before it "wins" and broadcasts — a real construct from Stanislas Dehaene's ignition model of conscious access.
- Arousal / noise — background random drive; too low and nothing ever ignites (deep sleep), too high and ignition becomes chaotic.
Real-world relevance: this ignition-and-broadcast pattern is what EEG/fMRI studies actually look for to distinguish conscious perception from unconscious processing — a sudden, wide, synchronized burst versus a signal that stays local and fades.