A neural network model based on the analogy with the immune system

Area of Science:

• Neuroscience
• Computational Biology
• Immunology

Background:

• The immune system and central nervous system share system-level similarities.
• Component-level analogies between these systems are less apparent.
• Existing neuron models do not fully capture these potential analogies.

Purpose of the Study:

• To formulate a novel neural network model based on immune and central nervous system parallels.
• To introduce a new neuron model incorporating hysteresis.
• To explore the model's potential for exhibiting properties like free will and learning.

Main Methods:

• Development of a hypothetical neuron model exhibiting single-neuron hysteresis.
• Modeling a network of N neurons using N-dimensional ordinary differential equations.
• Formulation of a conjecture for stimulus-response learning without synaptic modification.

Main Results:

• The N-neuron network model exhibits nearly 2N attractors.
• The model demonstrates a property analogous to free will.
• A learning conjecture suggests external stimuli guide the network to desired behaviors.
• A sleep model predicts changes in neuronal firing rate variance related to memory.

Conclusions:

• The proposed neuron and network model offers a new framework for understanding complex system dynamics.
• The model provides a unique perspective on learning, suggesting external guidance over synaptic plasticity.
• The research highlights a potential role for sleep in memory consolidation, with testable predictions.