Comparison of the sampling properties of different information quantities and of bias correction techniques. The estimates of information and entropies obtained with a number of techniques are tested on simulated data and plotted as a function of the number of generated trials per stimulus. Results were averaged over a number of repetitions of the simulation (mean value ± st. dev. over 50 simulations). We generated simulated LFPs which matched the second order statistics of LFPs recorded from primary visual cortex during visual stimulation with color movies (see Appendix A). The neural response r used to compute information was a two dimensional response array r = [r1, r2] containing the simulated LFP power at frequencies f1 = 4 Hz and f2 = 75 Hz, respectively. The power at each frequency was binned into 6 equi-populated values. A: Values of the four plug-in entropies H(R), H(R|S), Hind(R|S) and Hsh(R|S). B: Values of the bootstrap estimates of the quantities in Panel (A), computed by randomly pairing trials and stimuli. C: Mutual information values obtained with plug-in estimate (black line) and following the application of two bias correction procedures, namely QE (red line) and PT (green line). D: values of the same quantities reported in panel (C), but after further subtracting the bootstrap correction. E: Values of the information Ish(S; R) computed using the shuffling procedure with the following bias corrections: plug-in (black line); QE (red line) and PT (green line). F: values of the same quantities reported in panel (E), but after further subtracting the bootstrap correction.
Magri et al. BMC Neuroscience 2009 10:81 doi:10.1186/1471-2202-10-81