Nonparametric directionality measures for time series and point process data.

Area of Science:

• Neuroscience
• Signal Processing
• Biophysics

Background:

• Determining the directionality of neural signal interactions is crucial for analyzing multichannel recordings.
• Current parametric methods, often based on autoregressive models, have limitations and concerns.
• There is a need for alternative, robust approaches to assess neural signal directionality.

Purpose of the Study:

• To present a novel nonparametric approach for quantifying directionality in bivariate random processes.
• To develop a framework that decomposes correlation and coherence by interaction direction.
• To offer a method applicable to various data types, including time series, spike trains, and hybrid data.

Main Methods:

• Combines time and frequency domain analyses of bivariate data.
• Decomposes the product moment correlation coefficient into directional components.
• Decomposes coherence at each frequency into forward, reverse, and instantaneous interaction terms.

Main Results:

• Introduced a set of complementary scalar and functional measures for directional decomposition.
• Demonstrated the method's applicability to simulated cortical neuron networks (spike train data).
• Validated the approach using experimental data from muscle spindle sensory endings.

Conclusions:

• The proposed nonparametric method provides a viable alternative for assessing neural signal directionality.
• The framework enhances standard bivariate spectral and coherence analyses.
• The method is versatile and applicable to diverse neuroscience data types.