Synaptic neurotransmitter concentration modulation during learning in bio-inspired spiking neural network

Area of Science:

• Neuroscience
• Computational Biology
• Artificial Intelligence

Background:

• Understanding neurotransmitter dynamics during learning is crucial for neuroscience.
• Previous methods for measuring neurotransmitter concentrations are limited by brain complexity.
• Simultaneous, large-scale measurements across multiple neurotransmitters during learning remain a challenge.

Purpose of the Study:

• To develop a computational model for analyzing neurotransmitter concentrations during learning.
• To simulate neurotransmitter dynamics using consciousness-driven plasticity metrics.
• To validate the model's biological relevance by comparing simulated values with known physiological ranges.

Main Methods:

• A novel computational model was developed using a spiking neural network.
• The model incorporates consciousness-driven plasticity metrics, release rate, reuptake rate, and degradation rate.
• Simulations were performed using MNIST and Fashion-MNIST datasets.

Main Results:

• The model simulated neurotransmitter concentrations for dopamine, norepinephrine, acetylcholine, serotonin, glutamate, and GABA.
• Simulated values for MNIST and Fashion-MNIST datasets closely align with biological values.
• Quantitative correlation analysis confirmed the model's biological alignment.

Conclusions:

• The proposed computational model offers a viable approach to studying neurotransmitter concentrations during learning.
• The model's ability to simulate biologically relevant neurotransmitter levels advances the understanding of neural communication.
• Further quantitative correlation analysis will enhance the biological validation of the model.