A Neuronal Noise Critique of Integrated Information Theory
arXiv Preprint Archive December 6, 2021 via arXiv
Summary
AI-generated from the abstractIntegrated Information Theory (IIT) attempts to mathematically formalize conscious experience, but its treatment of neuronal noise contradicts experimental evidence. IIT predicts that noise reduces information integration, yet data show that decision-related noise is essential for learning, visual recognition, and categorical representation. The theory must be reformulated to account for both the beneficial and detrimental roles of noise observed in the brain.
Study at a glance
| Characteristics | Theoretical or philosophical paper Peer reviewed |
|---|---|
| Topics | Philosophy of mind |
| Keywords | Q-bio.nc Neuroscience Cognitive science Neural computation |
| Key finding | IIT's stance on neuronal noise is inconsistent with experimental data showing noise is necessary for learning and visual recognition. |
Abstract
Integrated Information Theory (IIT) is an audacious attempt to pin down the abstract, phenomenological experiences of consciousness into a rigorous, mathematical framework. We show that IIT's stance in regards to neuronal noise is inconsistent with experimental data demonstrating that neuronal noise in the brain plays a critical role in learning, visual recognition, and even categorical representation. IIT predicts that entropy due to noise will reduce the information integration of a physical system, which is inconsistent with experimental data demonstrating that decision-related noise is a necessary condition for learning and visual recognition tasks. IIT must therefore be reformulated to accommodate experimental evidence showing both the successes and failures of noise.