Coherence analysis of the ongoing EEG by means of microstates of synchronous oscillations

The sensitivity of instantaneous EEG coherence for the investigation of elementary thinking processes was shown by several authors. Similarly to EP analysis the statistical validation of the results is based on the multiple repetition of the same task. This approach is not possible for complex unreproducible thinking processes of long duration as e.g., solving mathematical problems. Classical coherence analysis of long periods of the EEG suffers from the loss of temporal information. The paper presents a temporal coherence analysis of the ongoing EEG preserving the temporal information of long cognitive processes. The main point of the method is the decomposition of the whole time interval in microstates of synchronous oscillations. The calculation of instantaneous coherence for multiple electrode pairs yields in a time-dependent high-dimensional coherence vector. A segmentation algorithm dissects the whole process into time intervals with stable coherence vectors the so-called microstates of oscillations. A subsequent clustering procedure into a small number of classes results in a sequence of prototypical microstates, which may be modeled by a Markovian process. Special entropy parameters characterize the strength of concatenation of different microstates. The method was applied in order to understand the special brain functioning of mathematically highly gifted subjects.