Noise induced oscillations in recurrent neural networks

Area of Science:

• Computational Neuroscience
• Neural Dynamics
• Biophysics

Background:

• Additive Gaussian white noise can induce oscillations in recurrent Hodgkin-Huxley neuron models.
• Type 1 oscillations are linked to Stochastic Resonance (SR) at lower noise amplitudes, while Type 2 oscillations appear at higher amplitudes.
• The underlying mechanism for Type 2 oscillations remains unclear.

Purpose of the Study:

• To investigate the influence of temperature on Type 2 oscillations in a recurrent neural network.
• To test the hypothesis that temperature affects the period of Type 2 oscillations.
• To elucidate the mechanisms distinguishing Type 1 and Type 2 oscillations in this model.

Main Methods:

• Computer simulations of a recurrent neural network composed of four Hodgkin-Huxley (HH) neuron models.
• Each HH neuron model received Gaussian noise and sub-threshold excitatory synaptic currents.
• Analysis of inter-spike interval (ISI) histograms to detect periodicity and firing patterns.

Main Results:

• Type 1 oscillation probability peaked at specific Gaussian noise standard deviations across tested temperatures (6.3, 15.0, 25.0 °C).
• The period of Type 2 oscillations demonstrated a clear dependence on temperature.
• Distinct mechanisms were suggested for Type 1 (synaptic delay) and Type 2 (refractory periods) oscillations.

Conclusions:

• Type 1 oscillations in recurrent HH models are induced by Gaussian white noise, influenced by synaptic delays.
• Type 2 oscillation periods are temperature-dependent, likely governed by neuron refractory periods.
• This study clarifies oscillation generation mechanisms in complex neural networks.