A Model-Free Method to Quantify Memory Utilization in Neural Point Processes.

Area of Science:

• Computational Neuroscience
• Information Theory
• Dynamical Systems

Background:

• Quantifying predictive capacity in neural systems is vital for understanding information processing.
• Information storage (IS) is a key measure but is limited to discrete-time processes.
• A continuous-time analysis method for neural data is needed.

Purpose of the Study:

• To introduce a model-free method for estimating the memory utilization rate (MUR) in continuous-time neural point processes.
• To quantify the predictive capacity stored within neural dynamics.
• To provide a reliable computational tool for neural data analysis.

Main Methods:

• Developed a method to estimate the memory utilization rate (MUR) for continuous-time neural point processes.
• Employed a surrogate data-based procedure to correct estimation bias.
• Validated the method using simulations and real neural data.

Main Results:

• The MUR method was validated in simulations of Poisson and coupled dynamical systems.
• Applied to cortical neuron cultures, MUR revealed increasing memory utilization during maturation.
• In heartbeat analysis, MUR reflected physiological stress responses.

Conclusions:

• The proposed approach provides a novel and computationally reliable tool for analyzing spike train data.
• Enables broader application of information-theoretic measures to continuous-time neural processes.
• Advances understanding of neural information processing and dynamic system evolution.