Hierarchical interactions between sensory cortices defy predictive coding.

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Neuroscience

Background:

• Perceptual experience relies on recurrent interactions between lower and higher brain regions.
• Predictive coding theory suggests feedback from higher to lower cortices suppresses predicted neuronal activity.
• The neurophysiological evidence for predictive coding in sensory cortices remains unclear.

Purpose of the Study:

• To review deviations between canonical intra- and inter-cortical interactions during perception and attention and predictions from the predictive coding framework.
• To propose alternative theories, termed BELIEF (Bayesian Encoding, Learning, Inference, and Feedback), that may better explain these circuit interactions.
• To examine how BELIEF theories account for inter-areal interactions during attentive perception.

Main Methods:

• Review of existing neurophysiological findings on intra- and inter-cortical interactions during perception and attention.
• Comparative analysis of these findings against predictions derived from the predictive coding model.
• Synthesis of alternative theoretical frameworks (BELIEF) to explain observed neural dynamics.

Main Results:

• Canonical patterns of neural interactions during perception and attention shifts often deviate from predictive coding predictions.
• Alternative BELIEF theories provide a more comprehensive account of observed inter-areal interactions in sensory processing.
• BELIEF theories effectively explain how the brain integrates sensory information and attention.

Conclusions:

• The predictive coding framework may be insufficient to fully capture the complexity of neural interactions in sensory cortices.
• Alternative BELIEF theories offer a more robust explanation for the mechanisms underlying perception and attention.
• Further research is needed to refine and validate BELIEF theories against empirical neurophysiological data.