Principal component analysis of event-related potentials: simulation studies demonstrate misallocation of variance across components.

Simulated event-related potential (ERP) components were used to investigate the ability of principal component analysis (PCA), Varimax rotation and univariate analysis of variance (ANOVA) to reconstruct component wave shapes, to allocate variance correctly across components, and to identify the correct locus of simulated experimental treatments. The simulated ERPs consisted of 800 randomly weighted combinations of three 64-point components, corresponding to a 2 X 2 X 10 repeated-measures design with 20 subjects. Covariance PCAs, Varimax rotations and univariate ANOVAs were performed on each of 400 such simulations, 100 with no effect of any experimental treatment and 100 each with main effects on each of the 3 components. Eight hundred additional simulations were performed to investigate the effects of systematic variations in the size of the experimental treatments and the number of subjects per experiment. The wave shapes of the simulated components were reconstructed reasonably well, although not completely, by the rotated principal component (PC) loadings. However, comparison of rotated PC scores with the random weights used to generate the simulated ERPs indicated that PCA incorrectly allocated variance across overlapping components, producing dramatic increases in type I error (the largest in excess of 80%) for ANOVAs on one component when the true treatment effect was on another. Although these results should not be overgeneralized, they clearly demonstrate that the PCA-Varimax-ANOVA strategy can incorrectly distribute variance across components, resulting in serious misinterpretation of treatment effects. Additional simulation studies are needed to determine the generality of the variance misallocation problem; pending the outcome of such studies, results obtained with the PCA-Varimax-ANOVA strategy should be interpreted cautiously.

[1]  S Sutton,et al.  Emitted P300 and slow wave event-related potentials in guessing and detection tasks. , 1980, Electroencephalography and clinical neurophysiology.

[2]  Walter J Freeman,et al.  Measurement of Cortical Evoked Potentials by Decomposition of Their Wave Form , 1979 .

[3]  Marwan A. Dabdoub.  Interpretation of Principal Components , 1978 .

[4]  Daniel S. Ruchkin,et al.  The Late Positive Complex , 1984, Annals of the New York Academy of Sciences.

[5]  L. Rogers,et al.  Comparative factor analysis models for an empirical study of EEG data. , 1983, The International journal of neuroscience.

[6]  Daniel S. Ruchkin,et al.  Principles of Neurobiological Signal Analysis , 1976 .

[7]  D. G. Wastell,et al.  On the independence of P300 and the CNV: A short critique of the principal components analysis of Donchin et al. (1975) , 1979, Biological Psychology.

[8]  D. G. Wastell,et al.  PCA and varimax rotation: Some comments on Rösler and Manzey , 1981, Biological Psychology.

[9]  Dietrich Manzey,et al.  Principal components and varimax-rotated components in event-related potential research: Some remarks on their interpretation , 1981, Biological Psychology.

[10]  A. Van Rotterdam,et al.  Limitations and Difficulties in Signal Processing by Means of the Principal-Components Analysis , 1970 .

[11]  D Friedman,et al.  Multiple late positive potentials in two visual discrimination tasks. , 1981, Psychophysiology.

[12]  H. Kaiser.  The varimax criterion for analytic rotation in factor analysis , 1958 .

[13]  C. C. Wood,et al.  Anatomical and Physiological Substrates of Event‐Related Potentials , 1984, Annals of the New York Academy of Sciences.

[14]  E. Donchin,et al.  On the independence of the CNV and the P300 components of the human averaged evoked potential. , 1975, Electroencephalography and clinical neurophysiology.

[15]  S. Curry,et al.  Hemisphere Differences in Event Related Potentials and CNV’s Associated with Monaural Stimuli and Lateralized Motor Responses , 1979 .

[16]  E. Donchin.  Multivariate analysis of event-related potential data: A tutorial review , 1978 .

[17]  W. Hays.  Statistics for psychologists , 1963 .

[18]  Richard J. Harris.  A primer of multivariate statistics , 1975 .

[19]  Donald O. Walter,et al.  Mass action in the nervous system , 1975 .

[20]  D. O. Wastell,et al.  On the correlated nature of evoked brain activity: Biophysical and statistical considerations , 1981, Biological Psychology.