A general method to generate artificial spike train populations matching recorded neurons

Area of Science:

• Computational Neuroscience
• Neuroscience

Background:

• Accurate modeling of neural activity is crucial for understanding brain function.
• Generating realistic artificial spike trains (ASTs) presents a significant challenge.

Purpose of the Study:

• To develop a general and flexible method for generating ASTs that replicate the statistical properties of recorded neuronal spike trains.
• To enable manipulation of rate-covariances within spike train populations for controlled experimental designs.

Main Methods:

• Computed Gaussian local rate functions from recorded spike trains to create rate templates.
• Generated ASTs as gamma-distributed processes with refractory periods, sampled from these rate templates.
• Applied algorithmic transformations (e.g., filtering, amplification, behavioral modulations) to rate templates to manipulate rate-covariances.

Main Results:

• Validated the method using surrogate and real neural data (cerebellum, cerebral cortex).
• Demonstrated that generated ASTs accurately reproduce firing rates, local/global spike time variance, and power spectra.
• Confirmed the ability to manipulate rate-covariances across different time scales.

Conclusions:

• The developed method provides a robust tool for generating biologically plausible AST populations.
• ASTs generated by this method serve as valuable inputs for dynamic clamp and multicompartmental modeling.
• Facilitates the study of synaptic integration under controlled, in vivo-like input conditions with adjustable covariances.