Fig. 6: Simulations of electrode pooling.
From: Electrode pooling can boost the yield of extracellular recordings with switchable silicon probes
a Workflow: Groups of four recording sites ("tetrodes") each carry a spike train from one simulated unit, superposed with electrode noise and biological noise. Between M = 1 and 12 of these tetrodes are then pooled into a single 4-wire recording followed by the addition of common noise. The pooled signal is sorted and the resulting single-unit spike trains are matched with the ground truth spike trains from the M tetrodes. Units with an accuracy metric> 0.8 are counted as recovered successfully. b Number of units recovered as a function of the pool size, M, under various conditions of the simulation. Effects of varying different parameters. The "standard" condition serves as a reference: Spike amplitude V = 380 μV, spike rate r = 10 Hz, electrode noise Nele = 1.6 μV, common noise Ncom =  5.7 μV, biological noise Nbio = 9 μV. "lower ampl": V =  205 μV. "higher rate": r = 20 Hz. "higher bio": Nbio = 15  μV. "lower com": Ncom = 2.85 μV. Each parameter combination was simulated three times with noise and spike times resampled, error bars are mean ± SD.
