Kanai's GMW framework uses control theory to detect consciousness in neural recordings

Ryota Kanai, a prominent consciousness researcher, has released a new theoretical framework designed to give Global Workspace Theory (GWT) a formal, testable foundation. The paper, published on arXiv (2608.15926), argues that GWT lacks a precise criterion for identifying the subnetwork that broadcasts information to enable conscious access. Kanai proposes the Global Mediation Workspace (GMW), a control-theoretic model that treats a candidate subnetwork as an open system embedded in the larger neural network. The framework defines a "mediation signature" using reachability (how the rest of the network drives the candidate), observability (how candidate states affect the rest), and a boundary Hankel operator that captures internal modes linking input and output. This signature quantifies mediation capacity, input-output alignment, effective dimensionality, and routed source-target breadth.

Kanai tested GMW on synthetic benchmarks to distinguish a planted differentiated mediator from dense hubs, one-sided receivers or broadcasters, and split read/write aggregates with no common internal route. He also applied the signature to ECoG recordings from four macaques under ketamine anesthesia. The results showed that input-output alignment was reduced during unconsciousness, while potential capacity increased. This suggests unconscious states may be characterized by decoupled or misaligned mediation, even when overall capacity remains. The GMW framework is highly technical, involving trajectory-conditioned differential operators, finite-amplitude response profiles, and state-dependent coalitions for nonlinear extensions. Yet its core contribution is practical: it offers a formal, measurable way to locate candidate global workspaces in neural recordings, potentially enabling clearer tests of consciousness theories against empirical data.