Symbolic dynamics of event-related brain potentials

Area of Science:

• Neuroscience
• Dynamical Systems Theory
• Information Theory

Background:

• Traditional event-related potential (ERP) analysis of electroencephalograms (EEG) has limitations in detecting subtle changes.
• Nonstationary and noisy multivariate EEG time series require advanced analytical methods.

Purpose of the Study:

• To apply symbolic dynamics techniques to estimate event-related brain potentials (ERPs) from EEG data.
• To develop and validate novel complexity measures for analyzing EEG dynamics.
• To compare the proposed methods with traditional ERP analysis.

Main Methods:

• Symbolic dynamics (word statistics, complexity measures) applied to multivariate EEG time series.
• Statistical mechanics approach based on the Frobenius-Perron equation.
• Development of time-dependent complexity measures using running cylinder sets.
• Validation through simulations of stochastic dynamical systems and comparison with traditional ERP analysis.

Main Results:

• The symbolic dynamics approach successfully estimates ERP components.
• Novel complexity measures distinguish between different experimental conditions and simulated data.
• Qualitative changes in EEG, undetectable by traditional methods, were revealed.

Conclusions:

• Symbolic dynamics offers a powerful alternative for ERP analysis, surpassing traditional techniques.
• ERPs should be conceptualized within dynamical systems and information theory frameworks.
• This approach enhances the detection of subtle EEG alterations and brain dynamics.