Search

Principal components analysis (PCA) has attracted increasing interest as a tool for facilitating analysis of high-density event-related potential (ERP) data. While every researcher is exposed to this statistical procedure in graduate school, its complexities are rarely covered in depth and hence researchers are often not conversant with its subtleties. Furthermore, application to ERP datasets involves unique aspects that would not be covered in a general statistics course. This tutorial seeks to provide guidance on the decisions involved in applying PCA to ERPs and their consequences, using the ERP PCA Toolkit to illustrate the analysis process on a novelty oddball dataset.

Principal components analysis (PCA) can facilitate analysis of event-related potential (ERP) components. Geomin, Oblimin, Varimax, Promax, and Infomax (independent components analysis) were compared using a simulated data set. Kappa settings for Oblimin and Promax were also systematically compared. Finally, the rotations were also analyzed in a two-step PCA procedure, including a contrast between spatiotemporal and temporospatial procedures. Promax was found to give the best overall results for temporal PCA, and Infomax was found to give the best overall results for spatial PCA. The current practice of kappa values of 3 or 4 for Promax and 0 for Oblimin was supported. Source analysis was meaningfully improved by temporal Promax PCA over the conventional windowed difference wave approach (from a median 32.9 mm error to 6.7 mm). It was also found that temporospatial PCA produced modestly improved results over spatiotemporal PCA.

Independent components analysis (ICA) and principal components analysis (PCA) are methods used to analyze event-related potential (ERP) and functional imaging (fMRI) data. In the present study, ICA and PCA were directly compared by applying them to simulated ERP datasets. Specifically, PCA was used to generate a subspace of the dataset followed by the application of PCA Promax or ICA Infomax rotations. The simulated datasets were composed of real background EEG activity plus two ERP simulated components. The results suggest that Promax is most effective for temporal analysis, whereas Infomax is most effective for spatial analysis. Failed analyses were examined and used to devise potential diagnostic strategies for both rotations. Finally, the results also showed that decomposition of subject averages yield better results than of grand averages across subjects., (2006 Wiley-Liss, Inc.)