Fig. 3.

SPIKE-synchronization (A), A-SPIKE-synchronization (B) and their difference (C) illustrated using five spike trains with four simple events. For the original measure (A) the small interspike intervals of spike doublets (first and second event) or bursts (third event) result in an unreasonably high demand for spike timing accuracy. With the adaptive generalization (B) for all these cases the likelihood increases that at least one of the spikes is part of a coincidence. On the other hand, if there are no doublets or bursts (last event), nothing changes (best seen in (C)). Note that the color scales differ, for better visibility we use gray-black in (A) and (B) but white-black in (C).