Detection and assessment of near-zero delays in neuronal spiking activity

Cross-correlation histograms (CCHs) have been widely used to study the temporal relationship between pairwise recordings of neuronal signals. One interesting parameter of a CCH is the time position of the central peak which indicates delays between signals. In order to study the potential relevance of these delays which can be as small as 1 ms, it is necessary to measure them with high precision. We present a method for the estimation of the central peak's position that is based on fitting a cosine function to the CCH and show that the precision of this estimate can be tracked analytically. We validate the resulting formula by simulations and by the analysis of a sample dataset obtained from cat visual cortex. The results indicate that the time position of the center peak can be estimated with submillisecond precision. The formula allows one also to develop a test of statistical significance for differences between two sets of measurements.