Advances in Brain, Vision, and Artificial Intelligence

Volume 4729 of the series Lecture Notes in Computer Science pp 428-437

Bounds of the Ability to Destroy Precise Coincidences by Spike Dithering

  • Antonio PazientiAffiliated withComputational Neuroscience Group, RIKEN Brain Science Institute, Wako
  • , Markus DiesmannAffiliated withComputational Neuroscience Group, RIKEN Brain Science Institute, Wako
  • , Sonja GrünAffiliated withComputational Neuroscience Group, RIKEN Brain Science Institute, WakoBernstein Center for Computational Neuroscience, Berlin

* Final gross prices may vary according to local VAT.

Get Access


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.


multi-channel recording spike train Monte-Carlo surrogate data correlation