6 ideas
Koch argues that experience is not a function a system performs but the causal power a system has upon itself. A system is conscious to the degree that its present state constrains its own past and future states in a way that cannot be reduced to independent parts. Consciousness is therefore a property of physical causal structure, not of information processing or behavior.
Phi quantifies how much a system's cause-effect structure is lost when it is cut into its least-interdependent parts. High phi requires both differentiation, meaning many distinguishable states, and integration, meaning the parts cannot be separated without destroying the whole's causal repertoire. The maximum of phi picks out which system and which grain is conscious; the specific shape of that structure determines what the experience is like.
IIT begins from features of experience taken as self-evident: it exists intrinsically, is structured, specific, unified, and definite. Each axiom is then translated into a requirement that a physical substrate must meet, such as intrinsic causal power for existence and irreducibility for unity. The method reasons from phenomenology to mechanism instead of trying to derive experience from neurons upward.
The cerebellum holds roughly 69 billion of the brain's 86 billion neurons, yet people born without one or who lose it have motor deficits and little or no change in conscious experience. Its architecture is largely parallel feedforward modules with little recurrent integration. Koch uses this to show that neuron count and computational load do not produce consciousness; integrated architecture does, and the posterior cortical 'hot zone' fits that requirement.
Because consciousness depends on a substrate's intrinsic causal structure, a conventional von Neumann computer running a perfect brain simulation would have near-zero phi. Intelligence and experience come apart: machines may match human behavior while feeling nothing, unless built on neuromorphic hardware with integrated causal structure.
The perturbational complexity index measures consciousness by delivering a magnetic pulse to the cortex (zap) and compressing the resulting EEG echo (zip). A conscious brain gives a complex response that is both widespread and differentiated, while deep sleep and anesthesia give simple, local, or stereotyped responses. The method can detect covert consciousness in behaviorally unresponsive patients, which has consequences for decisions about brain-injured people.