Psychophysiology, 37:S54, 2000.

Optimizing principal components analysis (PCA) methodology for ERP component identification and measurement: theoretical rationale and empirical evaluation

Jürgen Kayser, Craig E. Tenke

Department of Biopsychology, New York State Psychiatric Institute, New York, NY 10032, USA

Abstract

Although PCA is widely used to determine "data-driven" ERP components, it is unclear if and how specific methodological choices may affect factor extraction. The usefulness of the extracted factors can be evaluated by specific knowledge about the variance distribution of ERPs, which are characterized by the removal of baseline activity. The variance should be small for sample points before and shortly after stimulus onset (across and within cases), but large near the end of the recording epoch and at ERP component peaks. As a covariance matrix preserves this information, it is lost with a correlation matrix that assigns equal weights to each sample point, yielding the possibility that small but systematic variations may form a factor. These considerations were confirmed with real and simulated ERPs, which were systematically analyzed by varying the type of association matrix (correlation/covariance), Varimax rotation (scaled/unscaled), and number of components extracted and rotated in temporal PCAs. Scaling covariance-based PCA factors before rotation approximated correlation-based solutions, and ultimately yielded the same coefficients (factor loadings) when all components were rotated. Depending on the data set, limiting the number of components changed the morphology of some components considerably. However, more liberal or unlimited extraction criteria did not change high-variance components extracted from the covariance matrix. Instead, their interpretability was improved by more distinctive time courses with narrow and unambiguous peaks (i.e., low secondary loadings).

Keywords: ERP methodology; temporal PCA; component measurement