Enabling functional neural circuit simulations with distributed computing of neuromodulated plasticity.

Area of Science:

• Computational Neuroscience
• Systems Neuroscience
• Neuroplasticity

Background:

• Bridging the gap between system-level learning and synaptic plasticity is a key challenge.
• Neuromodulatory signals are increasingly recognized as crucial for synaptic plasticity and learning.
• Simulations are vital for studying the multi-scale dynamics of neural learning.

Purpose of the Study:

• To develop a general framework for implementing neuromodulated plasticity in distributed neural network simulations.
• To address the computational challenges of integrating dynamic neuromodulatory signals in large-scale models.
• To enable the study of learning across multiple spatial and temporal scales.

Main Methods:

• Developed a general, implementation-agnostic framework for neuromodulated plasticity.
• Integrated the framework into the NEST simulator for time-driven distributed simulations.
• Tested the framework's scalability with recurrent networks and neuromodulated spike-timing dependent plasticity.

Main Results:

• The framework efficiently handles neuromodulated plasticity in distributed simulations.
• Demonstrated excellent computational scaling up to 1024 processors.
• Successfully simulated a recurrent network with dynamic neuromodulation.

Conclusions:

• The developed framework provides a robust solution for large-scale simulations of neuromodulated plasticity.
• This approach facilitates the investigation of learning mechanisms across different scales in the brain.
• Enables more realistic computational models of brain function and learning.