Topographic analysis of event-related potentials (ERPs) has been used for segmentation of brain activity into microstates for a long time. But little is known about dynamics of topographic changes. This work shows that this dynamics is rather complex: ERPs consist of relatively long periods of stable topography (about 50 ms long) and relatively short unstable periods interspersed with them (about 20 ms long). A mathematical algorithm used for estimation of a degree of ERP topography instability is given. It is based on calculation of successive difference between normalized topographical vectors.
The data obtained in a brain-computer interface experiment (“spelling device”) were analyzed using this method. Main differences between target and nontarget conditions were revealed in the time range of P300 wave of ERPs elicited in this spatial attention task. Stable periods revealed were closely related to the principal ERP waves. Magnitude of the ERP vector was larger for the target condition as compared to the nontarget one in time intervals of N1, P3a, P3b and N400 stable periods, and maximum duration of instability was observed between N1 and P3a waves.