Stochastic resonance in the LIF models with input or threshold noise

Area of Science:

• Computational Neuroscience
• Neural Modeling
• Stochastic Processes

Background:

• The Leaky Integrate-and-Fire (LIF) model is a fundamental tool in computational neuroscience for simulating neuron dynamics.
• Stochasticity plays a crucial role in neural function, influencing information processing and network behavior.
• Stochastic Resonance (SR) is a phenomenon where a non-linear system exhibits enhanced response to a weak periodic signal due to the presence of noise.

Purpose of the Study:

• To analyze and compare the stochastic resonance (SR) behaviors of two distinct stochastic versions of the LIF neural model.
• To investigate the impact of noise signal placement (firing threshold vs. input current) on SR.
• To explore the effect of introducing time correlation into the noise signal.

Main Methods:

• Two stochastic LIF neural models were formulated: one with noise applied to the firing threshold, the other with noise added to the input current.
• Discontinuous stepwise noise with uncorrelated and Gaussian distributed innovations was employed.
• The SR properties of both models were analyzed and compared under varying noise conditions.

Main Results:

• Both stochastic LIF models exhibit stochastic resonance (SR) phenomena.
• The placement of the noise signal significantly influences the SR characteristics of the LIF model.
• Introducing suitable time correlation into the noise signal can effectively transition the behavior from the threshold-noise model to the input-current-noise model.

Conclusions:

• The study provides a comparative analysis of noise integration strategies in stochastic LIF models concerning SR.
• Time correlation in noise acts as a bridge between different noise-handling mechanisms in neural models.
• Findings contribute to a deeper understanding of how noise influences neural information processing and the emergence of SR.