Abstract
Construction and display strategies for mapping of evoked brain potential data are reviewed. The effect of choice of reference (no change of map configuration but change of voltage level, inviting misleading interpretations), and the effect of change of baseline (change of map configuration) are discussed. Methods for the construction of reference-independent local and global map values are reviewed. Such values are necessary for data reduction over space and time (further analysis), which in turn is necessary as basis for subsequent statistics, since mapping is no data reduction. Strategies are presented for reference-independent determination of 'component latencies', and reference-independent adaptive, spatial criteria-based segmentation of the map series into stationary epochs which might be caused by different neural generator processes. The principles of spatial analysis can also be applied to recordings in few channels.