Fig. 13.

Difference in the test-power between NeuroXidence and the unitary event method (UE-method). (a–c) Variation of length l of the analysis window (a), the number of trials T (b), and the spike rate r (c) from the stand parameter set (T = 50, r = 15 ap/s, S = 20, η = 3) are presented. Spike trains were modeled as a single-interaction process (SIP) based on homogenous Poisson processes. The frequency of excess joint-spike events (JSEs) beyond the chance level is given by the x-axis. (a) The length of the analysis window used for both NeuroXidence and the UE-method varied between 0.2 s, 0.4 s, and 0.8 s. (b, c) The NeuroXidence analysis window was 0.8 s while and the UE window was 0.2 s long. Rows (1–4) show the difference in test-power for test-patterns of complexity 2–5