Modeling biological memory network by an autonomous and adaptive multi-agent system.

Area of Science:

• Cognitive Science
• Computational Neuroscience
• Graph Theory

Background:

• Traditional graph models are static and lack the dynamic properties of biological neural networks.
• Existing models often rely on algorithms with a global perspective, limiting decentralized processing.
• Biological neural networks exhibit dynamic, autonomous behaviors crucial for memory.

Purpose of the Study:

• To introduce a multi-agent system (MAS) model based on graph theory for simulating brain memory processes.
• To develop a decentralized approach to information memory and retrieval.
• To emulate neuron electrophysiology and validate memory trace theory.

Main Methods:

• A multi-agent system (MAS) model was developed, integrating graph theory with adaptive learning agents.
• Memory storage was conceptualized as the management of MAS paths, with decentralized information processing.
• Agent adaptive learning was modeled using a microcircuit with a variable resistor, applying Ohm's and Kirchhoff's laws.

Main Results:

• The MAS model demonstrated decentralized dynamic information memory and simulation capabilities.
• Computer simulations validated the model's efficacy in memorizing and retrieving data.
• The model successfully avoided a global view, relying on neighborhood interactions for resource utilization.

Conclusions:

• The proposed MAS model offers a plausible neurobiological explanation for memory realization.
• The decentralized approach transforms memory storage into dynamic path management within the MAS.
• The study validates memory trace theory at a system level through computational modeling.