IIT-style models suggest a system's 'phi' rises when its parts are both richly interconnected and individually distinguishable, not merely large.
phi ~ integration * differentiation
(simplified Integrated Information Theory proxy)
- System nodes — elements of the modelled system whose combined state may be 'integrated'.
- Integration density — how interconnected the nodes are — IIT holds that phi requires high integration.
- Signal differentiation — how distinguishable each node's state is from every other — repetition lowers phi.
- Activation threshold — signal strength that counts as a node actively contributing to the system state.
Integrated Information Theory proposes exactly this quantity — a system's capacity to be more than the sum of its parts — as a candidate measure of consciousness.