Application of stochastic and chaotic models of EEG to the detection of readiness potential onset

An attempt to detect the onset of trend-like EEG potentials which precede voluntary movements, i.e. readiness potentials (RP) on a single trial basis is described. Additivity of the background EEG activity and RP was assumed and the dynamics of recorded data was tracked through: (i) parameters of the recorded EEG and of its AR model; (ii) the bifurcation parameter of the nonlinear filter. In these ways changes in the dynamics of the underlying process including segments manifesting chaotic time series were detected. Both approaches were tested by using artificial and real experimental data. Suitable, AR model based, detectors of the RP onset were designed and promising results with detection of tent and logistic type chaotic transients of EEG were obtained