[Methodological aspects of the analysis of the effects of drugs (diazepam and caffeine) on contingent negative variation].

The paper discusses the importance and implications of the event-related-potentials in the analysis of effects of centrally acting drugs. This is illustrated using a study on the influence of diazepam and caffeine on the contingent-negative-variation (CNV) as an example. Special emphasis is placed on methodological problems and their possible solutions. A linear robust fitting technique was used for quantification of the slow-rising, ramp-like CNV potentials as an alternative to conventional baseline-to-peak measures. Results obtained by this method were validated by a second analysis taking slow superimposed DC-shifts into account. This was achieved by defining a reference line from the data for CNV-development and -resolution. The possibilities of single-trial potential evaluation were explored by using a pattern recognition algorithm. The average CNV-waveform was used as a reference-signal to which the single-trial potentials were correlated. Since pattern recognition techniques are sensitive to morphological changes of the potentials they are adding a new aspect to event-related-potential analysis.