Skip to content

The Entangled Loop: Emotion, interference and hierarchy in the architecture of consciousness.

Morten L. Kringelbach, Fernando E. Rosas, Ruben Laukkonen, Shamil Chandaria, Yonatan Sanz Perl, Gustavo Deco

Neuroscience and biobehavioral reviews October 27, 2026 DOI: 10.1016/j.neubiorev.2026.107063 (opens in new tab)

Study at a glance

AI-extracted from the abstract
Characteristics Theoretical or philosophical paper Peer reviewed
Keywords Thermodynamics of mind Active inference Emotion Feeling Global neuronal workspace Hedonia Hedonic hotspots Hierarchy Large language models Mental health, consciousness Quantum-like computation
Key points Proposes that consciousness and unified experience arise from an "entangled trefoil" of emotion, interference-based inference, and hierarchical orchestration, constrained by the thermodynamics of living systems remaining far from equilibrium. Argues that similar signatures in large language models are shadows of the generating architecture rather than evidence of consciousness, and that humans' abundant long-range connections make the architecture more computationally efficient but more fragile, potentially explaining vulnerability to neuropsychiatric disorders and the poor translation of animal models.

Abstract

Why does consciousness feel like something, and why is experience unified rather than fragmented? We propose that these properties arise from an entangled trefoil of emotion, interference-based inference and hierarchical orchestration, shaped by the thermodynamic constraint that living systems must remain far from equilibrium using limited energy. Emotion compresses actions into valenced signals of what matters, while a pleasure cycle schedules competing biological needs through predictive wanting and satiety organised around the true hedonic outcome of liking. Hierarchical orchestrating regions integrate and broadcast information through interference between classical coupled oscillators, shaped by the spectral gaps of the connectome driven by long-range exceptions to local cortical wiring. The architecture of the Entangled Loop consists of three inseparable strands, each continually constraining the other two, with the one operator linking computational possibilities to the arrow of time and energetic cost of running them. Evidence suggests that the same three signatures emerge independently in large language models trained on human language, but as shadows of the generating architecture rather than evidence of consciousness. The trefoil predicts a distinctive human vulnerability because our abundance of rare long-range connections promotes more computation for less energy while making the architecture more fragile. Neuropsychiatric disorders could therefore arise from perturbations entering through any strand and propagating through the others, while the more robust architecture of other species may help explain why animal models reproduce human psychiatric disorders poorly. Sustaining and restoring flexible entanglement in the Entangled Loop may provide mechanistic routes towards flourishing.

Citations in the library