Bounds of the ability to destroy precise coincidences by spike dithering

Abstract

Correlation analysis of neuronal spiking activity relies on the availability of distributions for assessing significance. At present, these distributions can only be created by surrogate data. A widely used surrogate, termed dithering, adds a small random offset to all spikes. Due to the biological noise, simultaneous spike emission is registered within a finite coincidence window. Established methods of counting are: (i) partitioning the temporal axis into disjunct bins and (ii) integrating the counts of precise coincidences over multiple relative temporal shifts of the two spike trains. Here, we rigorously analyze for both methods the effectiveness of dithering in destroying precise coincidences. Closed form expressions and bounds are derived for the case where the dither range equals the coincidence window. In this situation disjunct binning detects half of the original coincidences, the multiple shift method recovers three quarters. Thus, only a dither range much larger than the detection window qualifies as a generator of suitable surrogates. © 2007 Springer-Verlag Berlin Heidelberg.