The Interpretation of the Component Structure of Event‐Related Brain Potentials: An Analysis of Expert Judgments

The analysis of components of the event‐related brain potential (ERP) is accomplished through standardized statistical algorithms. However, visual inspection of ERPs is commonly used to guide the selection of the analytic techniques. Thus, the definitional criteria of components employed by the investigators may interact with the choice of analysis procedures. The present study examines the criteria which investigators employ to define components of the ERP, the relative importance assigned to different criteria, and the consistency with which investigators use the definitional criteria. ERPs are simulated so as to vary systematically the amplitude, latency, duration, and electrode distribution of the P300 component. Ten experienced ERP researchers rated the similarity of 153 pairs of simulated ERPs. The ratings were analyzed by a multidimensional scaling procedure. Eight unidimensional judgments on each of the 18 ERPs were also obtained. The results suggest that ERP investigators are capable of accurately recovering the underlying dimensions of a set of simulated ERPs. Furthermore, the degree of experience in analyzing and interpreting ERPs affects the weighting structure of the underlying dimensions. These findings support the commonly held, but previously untested, belief that judgments of ERPs based on visual inspection can be both accurate and reliable