[FieldTrip] Necessity of "cfg.debias" in "ft_spike_xcorr " function?

[Zhang Xiaoxing]

Hello everyone,
I use ft_spike_corr function in the fieldtrip package to do spike time 
cross-correlation. The code runs great, I got reasonable results. Still 
a detail puzzles me.

     In the cfg, one of the option is cfg.debias, with default value set 
to 'yes'. The in-line comment is
=========
%   cfg.debias = 'yes' (default) or 'no'. If 'yes', we scale the 
cross-correlogram by M/(M-abs(lags)), where M = 2*N -1 with N
=========
     With typical options of maxlag=0.1, binsize=0.002, latency=1:2, 
numerically, the debias option increase the x value (raw 
cross-correlation count) far from center by approx. 5% and decrease the 
x value in the center for approx. 5%. My question is: Why (or When) do 
we need this debias? Is there some reference to this algorithm?

Thanks a lot!



-- 
Zhang Xiaoxing
Laboratory of  Functional Neural Circuitry
Institute of Neuroscience
Center for Excellence in Brain Science and Intelligence Technology
Chinese Academy of Sciences