Some computational and statistical techniques that can be used in the analysis of event-related potential (ERP) data are demonstrated. The techniques are fairly elementary but go one step further than do simple area measurement or peak picking, which are most often used in ERP analysis. Both amplitude and latency measurement techniques are considered. Principal components analysis (peA) and methods for electromyographic onset determination are presented in detail, and Woodyfiltering is discussed briefly. The techniques are introduced in a nontechnical, tutorial review style. One and the same existing data set is presented, to which the techniques are applied, and practical guidelines for their use are given. The methods are demonstrated against a background of theoretical notions that are related to the definition of ERP components.The electroencephalogram (EEG) is the record of the potential differences between electrodes placed on the (human) scalp. EEG records can be classified into two categories: the spontaneous or background EEG and event-related potentials (ERPs). The background EEG consists of periodic voltage fluctuations, which can be classified according to their frequency content-for instance, alpha (8-12 Hz), beta (13-30 Hz), gamma (31-50 Hz), delta (0.5-4 Hz), and theta (5-7 Hz). ERPs are discrete waveforms that are associated with an event. They are hidden within the background EEG and are visualized by averaging discrete EEG epochs that are synchronized to meaningful task events, such as the presentation ofa stimulus or the occurrence ofa response. Since the background EEG is not assumed to have a temporal relationship to the synchronization event, averaging of discrete EEG epochs will, when a sufficiently large number of epochs (trials) are averaged, result in its attenuation. The activity that does have a temporal relation to the event-the ERP-will then be enhanced in relation to the background EEG and thus become visible. If plotted in time, ERPs appear as a series of positive and negative deflections, which are thought to be the manifestation of underlying ERP components.This paper was prepared while the author was supported by NWO Grant 575-270-082B, awarded to M. W van der Molen and C. H. M. Brunia. The section on principal components analysis benefited from discussions with Marcel Croon of the Department of Statistics, Tilburg University. The useful suggestions ofMartin Eimer, Emanuel Donchin, and two anonymous reviewers are gratefully acknowledged. Correspondence concerning this paper should be addressed to G. 1. M. van Boxtel, Department of Psychology (P607), Tilburg University, P.O. Box 90153, 5000 LE Tilburg, The Netherlands (e-mail: g.j.m.vboxtel@kub.nl).
87ERP components may be exogenous or endogenousthat is, evoked by events that are extrinsic or intrinsic to the nervous system. The exogenous ERP components occur near the eliciting event, and the time relation to that event is very stable. When a stimulus is presented, for instance, the exogenous ERP components have an early onset l...