[State-dependent processing of serial signals].

The relation between EEG patterns and performance, which was found experimentally, gives evidence for a state dependent information processing. The power distribution of the EEG one second before the stimulus onset determines the state measured by the used data acquisition system. To solve the on-line identification of the state four different approaches are proposed and compared. The two modes of triggering the stimulus onset labeled by alpha wave state and non alpha wave state led to significantly different reaction times. The proposed stochastic recurrence equation can be used to model the scanning process performed by 17 subjects in solving the task of searching the first number in a string of characters. The model is based on the decomposition of the measured reaction times according to the assumed elementary processes.