Trusting in or breaking with convention: towards a renaissance of principal components analysis in electrophysiology.

Event-related potential (ERP) measures have an intimate and historic association with basic neuroanatomy and neurophysiology. The sensory projection systems were originally mapped by the pioneers using the simplified evoked potential waveforms of deeply anesthetized animals, from which functional activity is easily quantified by direct amplitude and latency measurements (e.g. Marshall et al., 1937, 1941). The clinical evoked potentials and cognitive ERPs that descended from these classical studies rely on signal averaging methods (e.g. Dawson, 1954) to reveal time-locked activity that reflects the transfer and processing of information by the nervous system in awake human subjects. The variability and complexity of the resulting waveforms provide a considerably greater challenge for identifying and quantifying functional activity of interest. The construct of an ERP component (or any equivalent construct; e.g. Lehmann and Skrandies, 1984) is used to decompose and understand ERP waveforms according to both their intracerebral origin (i.e. underlying neuronal generators) and their dependence on specific experimental manipulations (e.g. Picton et al., 2000).