From Spiking Neuron Models to Linear-Nonlinear Models
2011

Understanding Neuron Firing Rates with Simplified Models

publication Evidence: moderate

Author Information

Author(s): Srdjan Ostojic, Nicolas Brunel, Peter E. Latham

Hypothesis

To what extent can the input-output mapping of spiking neuron models be reduced to a simple linear-nonlinear cascade?

Conclusion

Linear-nonlinear models provide a good approximation of the firing rate dynamics of spiking neurons, particularly when background noise is not excessively weak.

Supporting Evidence

  • The study shows that linear-nonlinear models can accurately estimate firing rates in most parameter spaces.
  • The models were validated against numerical simulations of leaky integrate-and-fire and exponential integrate-and-fire neurons.
  • An adaptive timescale rate model was introduced to improve accuracy for large input signals.

Takeaway

This study looks at how neurons respond to signals and background noise, showing that simple models can often predict their behavior accurately.

Methodology

The study compares the predictions of linear-nonlinear models with numerical simulations of spiking neurons across various parameters.

Limitations

The accuracy of the models decreases with very strong input signals and weak background noise.

Digital Object Identifier (DOI)

10.1371/journal.pcbi.1001056

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication