A jackknife approach to quantifying single-trial correlation between covariance-based metrics undefined on a...

Area of Science:

• Neuroscience
• Computational Neuroscience
• Data Analysis

Background:

• Quantifying functional connectivity requires multiple observations.
• Neuronal correlations fluctuate moment-by-moment, impacting metrics like reaction time.
• Existing methods struggle to quantify co-fluctuations in non-single-observation metrics.

Purpose of the Study:

• To develop a method for quantifying moment-by-moment fluctuations in neuronal correlations.
• To introduce and validate jackknife correlation (JC) for analyzing dynamic functional connectivity.
• To compare JC with alternative methods for analyzing correlated neuronal activity.

Main Methods:

• Calculating neuronal correlations using leave-one-out jackknife replications.
• Correlating jackknife correlation values to assess moment-by-moment fluctuations.
• Simulating paired data and spectral correlation metrics (e.g., coherence) for validation.

Main Results:

• JC precisely recovers conventional correlation for simulated paired data.
• JC outperforms sorting-and-binning methods, avoiding parameter-induced artifacts.
• JC excels over epoch subdivision methods for spectral correlations, especially with short epochs, preserving spectral resolution.

Conclusions:

• Jackknife correlation (JC) is a robust method for quantifying dynamic functional connectivity.
• JC accurately relates moment-by-moment fluctuations in neuronal correlations without sacrificing resolution.
• The JC method is broadly applicable to any smooth metric not defined on single observations.