Generative Predictive Codes by Multiplexed Hippocampal Neuronal Tuplets.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Cognitive Science

Background:

• The brain rapidly forms internal representations from experiences but lacks clear mechanisms for encoding multiple distinct memories without capacity limits.
• Understanding how the hippocampus (a brain region vital for memory and navigation) achieves rapid encoding of novel sequences is a key neuroscientific challenge.

Purpose of the Study:

• To elucidate the neuronal ensemble mechanisms underlying rapid encoding of multiple distinct spatial sequences in the hippocampus.
• To investigate the role of spontaneous sequential activity and predictive coding in forming and updating internal models of spatial regularities.

Main Methods:

• Development of a probabilistic statistical model to analyze hippocampal spontaneous sequential activity.
• Modeling generative predictive codes for novel spatial sequences.
• Analysis of neuronal functional connections and potentiation during navigation and experience.

Main Results:

• Identification of an internal model generating predictive codes for multiple novel spatial sequences.
• Discovery that differences between external stimuli and the internal model are encoded by novel functional connections that update the model.
• Evidence that this predictive coding relies on neuronal organization into modules of short, high-repeat sequential neuronal 'tuplets' ('neuro-codons').

Conclusions:

• Flexible multiplexing of neuronal tuplets enables hippocampal predictive codes to vastly expand representational capacity.
• This mechanism supports rapid learning and spatial navigation by initiating top-down cortical loops.
• The study proposes a novel framework for understanding memory formation and cognitive flexibility.