Neural Decoding as Cognitive Inference

2026-10-08

AI Summary

This study addresses the challenge of extracting stable cognitive states from highly variable neural activity, a process frequently confounded by spurious correlations. Inspired by the Bayesian brain hypothesis, this work reformulates neural decoding as a cognitive inference process constrained by the brain’s intrinsic priors. It introduces meta-neural semantic representations, integrating multimodal recording analyses with high-dimensional representational geometry reorganization to uncover cross-task consistent geometric relationships that elucidate the mechanisms underlying cognitive stability. The proposed approach successfully recovers stable cognitive states across five recording modalities and three cognitive domains. These findings empirically validate the theoretical hypothesis that cognition can be maintained without relying on fixed neural patterns, offering new insights into how the brain achieves robust cognitive function despite pervasive neural variability.

Abstract

The brain maintains stable cognition despite continuously changing neural activity. How to extract stable cognitive states from variable neural observations remains a central problem in neural decoding. Existing neural decoding methods map neural observations to predefined external labels based on the stimulus-response principle, often capturing recording-specific spurious correlations. Inspired by how the brain infers the world, and specifically by Bayesian brain theory, we recast neural decoding as cognitive inference constrained by brain-intrinsic priors, yielding high-level meta-neural semantic representations. In decoding experiments spanning five neural recording modalities and three cognitive domains (motor, perception and internal mentation), our cognitive inference method reorganized the geometry of neural observation representations, yielding meta-neural semantic representations that exhibited consistent geometric relationships across cognitive tasks and enabled the recovery of stable cognitive states from variable neural observations. Our work provides an account of how the brain maintains relatively stable cognition despite continual changes in the external environment. Cognitive stability is sustained through cognitive inference from changing neural activity, without requiring fixed neural activity patterns.

Problem

• Neural Decoding
• Cognitive Inference
• Cognitive Stability
• Bayesian Brain Theory
• Meta-neural Semantic Representations

Innovation

• Neural Decoding
• Cognitive Inference
• Bayesian Brain Theory
• Meta-neural Semantic Representations
• Representation Geometry