Equivalent dipole modelling of topographic ERP-components: A developmental case study

This paper presents a re-analysis of cross-sectional event-related brain potential (ERP) data obtained in a previous study using a standard Oddball paradigm (Wijker, Molenaar and van der Molen, 1989). These data were subjected to topographic principal component analysis (PCA). For each of the three age-groups and for each of the nine pre-selected time points after stimulus onset, two PCA's were conducted (a total of 54 PCA's). One series of topographic PCA's was performed on the pooled covariance matrices whereas the other series of topographic PCA's was performed on the raw moment matrices (including the mean field). All PCA's yielded three components which explained most of the variance. The results were consistent with the non-topographic PCA findings reported by Wijker et al. (1989). The main goal of the current study, however, was to assess whether the components thus obtained represent functional units of brain response. Therefore, the PCA-derived topographic ERP components were subjected to equivalent dipole modeling. The results indicated that only in a few instances topographic component fields can be described by an equivalent dipole. The results did not show any ystematic trends across age or time. It was suggested that individual differences in structural and functional electrocortical sources might be responsible for the poor match between scalp fields of topographic PCA-derived ERP components and equivalent dipole modelling. Possibly equivalent dipole modelling precludes the pooling of data across subjects.