A Circuit Model for Working Memory Based on Hybrid Positive and Negative-Derivative Feedback Mechanism

Area of Science:

• Cognitive Neuroscience
• Computational Neuroscience
• Neural Networks

Background:

• Working memory (WM) is crucial for learning and decision-making, supporting daily tasks like remembering codes.
• Existing WM models often rely on synapse parameters, limiting robustness.
• Understanding WM at the microscopic neural level requires advanced computational models.

Purpose of the Study:

• To design a microscopic neural network model explaining the internal mechanisms of working memory.
• To develop a more robust WM model using a hybrid feedback mechanism.
• To simulate information storage, association, updating, and forgetting at the neural circuit level.

Main Methods:

• Developed a hybrid positive and negative-derivative feedback (HPNF) model to enhance robustness against disturbances.
• Constructed a memory-storage sub-network (SET) utilizing positive feedback and negative-derivative feedback.
• Designed a storage distribution network (SDN) integrating SET for memory operations.

Main Results:

• The SET network demonstrates robustness in self-sustaining neural information.
• The SDN effectively constructs a storage distribution network at the neural circuit level.
• The model successfully simulates information storage, association, updating, and forgetting.

Conclusions:

• The proposed HPNF-based neural network model provides a robust mechanism for working memory functions.
• The model demonstrates adaptability across different individuals with minimal parameter adjustments.
• This work advances our understanding of working memory at the neural circuit level.