Doubting these doubts--a reply to Collet

In reply to Collet (1989) it is argued that principal component analysis (PCA) of event-related potentials is not invalidated or disproved by the arguments of his comment. The main points of this reply are as follows. First, since Collet's analysis was based on the correlation matrix only, it cannot disprove assumptions of PCA which do not constrain the correlation (or covariance) matrix. Second, his model utilizes results of PCA which invalidate the comparison and parameter count. Third, his model does not allow specific predictions and does not lead to real data reduction, which reverses Collet's argument based on the principle of parsimony. Fourth, his model is less general than PCA, as it could apply (if at all) to slow brain potentials only.