PCA-based decomposition, along with CSD transformation, has been a useful approach to elicit topographically distinct activation patterns to distinguish clinical groups from control subjects, as applied in schizophrenia (Kayser et al. 2006, 2010) and depression (Tenke et al. 2008, 2010). Using a MGT task, Bernat and colleagues (2011) examined the relationship between externalizing proneness and the feedback-related positivity (FRP/P3) and negativity (FRN/N2). Using PCA decomposed time-frequency measures accompanying P3 response to feedback cues revealed that feedback-locked delta-P3 activity was reduced among individuals high in externalizing proneness, whereas theta-N2 response was unrelated to the externalizing index. Begleiter and colleagues (1987b) elicited P3 amplitude differences between HR offspring of alcoholics and low-risk control subjects using PCA-derived ERP waveforms. Using a similar method in a flanker task in an alcohol administration study, Bartholow and colleagues (2003) reported that a PCA-derived frontal negativity ERP component was related to the high dose of alcohol during both correct and incorrect response trials. However, incorrect allocation of components and lack of functionally meaningful components have been cited as weaknesses with these methods (Wood and McCarthy 1984), although some solutions have been suggested to overcome these limitations (Dien et al. 2003).
