Biomimetic model of corticostriatal micro-assemblies discovers a neural code

Area of Science:

• Neuroscience
• Computational Biology
• Cognitive Science

Background:

• Linking low-level neural activity (spiking, field potentials, biochemistry) with high-level cognition (decision-making, working memory) remains a significant challenge.
• Existing computational models often struggle to bridge this physiological-cognitive gap mechanistically.

Purpose of the Study:

• To introduce a mechanistically accurate multi-scale model that directly simulates neural physiology.
• To demonstrate how emergent neural and cognitive phenomena arise from this simulated physiology.
• To validate the model's predictions against experimental macaque data.

Main Methods:

• Development of a multi-scale computational model incorporating spiking, field potentials, and synaptic plasticity.
• Simulation of neural and cognitive processes including working memory, decision-making, and categorization.
• Validation of model outputs against extensive, previously unseen experimental macaque data.

Main Results:

• The model successfully generated simulated physiology (spiking, fields, phase synchronies, synaptic change).
• Emergent cognitive functions such as working memory, decision-making, and categorization were directly produced.
• A novel neural code, "incongruent neurons," was discovered, predicting erroneous behaviors and subsequently confirmed in empirical data.
• The model demonstrated predictive power by linking computational decision/reinforcement signals with neurobiological spiking/field codes.

Conclusions:

• The developed biomimetic model provides a direct and predictive link between neural activity and cognitive functions.
• This multi-scale approach advances our understanding of how low-level neural mechanisms give rise to complex cognitive abilities.
• The discovery of "incongruent neurons" offers new insights into the neural basis of errors and behavioral prediction.