The standard regularity analysis for spike trains in cochlear nucleus neurons evoked by tonebursts first proposed by Bourk is widely used, primarily as one of the criteria for classification of such neurons. It is shown that this procedure does not estimate quite what it is supposed to, and introduces unnecessary noise to its results due to its use of bins. Instead the desired quantities (mean and coefficient of variation of the lengths of all inter-spike intervals in progress as a function of time since stimulus onset) can all be exactly calculated directly from the spike train without the need for data binning. The implications for classification and other studies are discussed.
© 2011 Acoustical Society of America