From experience to physics

Developed by Giulio Tononi from 2004 and refined with Christof Koch and others, IIT begins with “axioms”, properties every experience seems to have: it exists for itself, it is structured, specific, unified and definite. It then asks what properties a physical system must have to account for these, its “postulates”.

What Φ measures

The answer is integrated information, written Φ (phi): roughly, how much a system’s cause-effect structure as a whole exceeds that of its parts taken separately. A system with high Φ cannot be divided without losing a lot of its causal power. IIT identifies consciousness with this maximally integrated structure. Calculating Φ exactly is computationally intractable for real brains, so experiments use proxies.

Surprising predictions

• Experience should depend mainly on posterior cortex, with its grid-like connectivity, more than on prefrontal regions.
• The cerebellum, despite having most of the brain’s neurons, should contribute little, because of its parallel, weakly integrated wiring.
• A conventional digital computer simulating a brain might have almost no consciousness, while some simple but highly integrated systems could have a little.

The 2023 controversy

In September 2023, more than a hundred researchers signed an open letter describing IIT as “pseudoscience”, citing its untestable core claims and media coverage that overstated its support. Many others replied that the label was unjustified and harmful to the field. The episode made clear how unsettled the standards for judging theories of consciousness still are.

Tests

The perturbational complexity index, inspired by IIT, distinguishes conscious and unconscious states with promising accuracy. The large adversarial test against Global Workspace Theory challenged some of IIT’s predictions about sustained posterior activity: IIT vs Global Workspace.