Serial Spike Time Correlations Affect Probability Distribution of Joint Spike Events

Area of Science:

• Computational Neuroscience
• Systems Neuroscience
• Neuroscience

Background:

• Detecting temporally coordinated neural spiking activity is crucial for understanding cortical information processing.
• Current methods often assume independent spike generation (Poisson process), neglecting known neural dependencies like refractory periods and bursting.

Purpose of the Study:

• To investigate non-renewal processes incorporating spike-history dependence in individual neurons.
• To evaluate the impact of these dependencies on the analysis of synchronized neural events.

Main Methods:

• Utilized Monte Carlo methods to estimate coincidence count distributions.
• Modeled inter-spike interval distributions to incorporate spike-history dependence.
• Compared results with homogeneous and non-Poisson processes.

Main Results:

• Non-renewal processes significantly alter the width of the joint spike event distribution compared to Poisson models.
• These processes can result in both heavy-tailed or narrow coincidence distributions.
• Small variations in point process autostructure lead to substantial changes in coincidence distribution width.

Conclusions:

• Standard significance estimation for joint spike events may be inadequate due to overlooked spike-history dependencies.
• Accurate modeling of neural spike train autostructure is essential for reliable detection of coordinated activity.