Issues in the application of the average reference: Review, critiques, and recommendations

In standard event-related potential (ERP) recordings, activity at the reference site affects measurements at all the "active" electrode sites. Analyses of ERP data would be improved if reference site activity could be separated out. Apart from current source density methods, which can miss deep or distributed events, the major alternative to an "indifferent" electrode reference is one that combines all active electrodes, subtracting out the activity common to all of them. The average reference has merits (such as an insensitivity to scalp currents near any single electrode) and limitations (such as a dependence on the number and locations of all the electrodes in the average). This review compares the effects of different references on the scalp topography of the auditory N1 recorded with an array of 128 electrodes. Furthermore, it shows how assumptions associated with each reference affect inferential methods such as analysis of variance and correlation. Finally, it seeks to evaluate the efficacy of the average reference in estimating the true zero potential line.