Dynamics of the granule cells in response to sinusoidally oscillating MF signals at 0.5 Hz.

<p>(A) Top panel, spike patterns of 1,000 out of 102,400 granule cells during a cycle of MF signal oscillation (2.0 s = 0.5 Hz). Black dots indicate spike discharges. At the beginning and end of a cycle where the firing rate of MFs is low, granule cells elicit spikes uniformly at random. At the middle of a cycle where the firing rate of MFs is high, spike patterns exhibit a variety of temporal profiles. Bottom panel, the ratio of active granule cells, which plots the number of active granule cells in each 10 ms bin divided by the total number of granule cells ( = 102,400). (B) Spike patterns of 5 representative granule cells across 100 cycles of MF signal oscillation. Black dots and grey lines in each panel show spike discharges in each cycle and the averaged spike density histogram, respectively. The granular layer modeled as a random recurrent network generates a variety of discharge patterns of granule cells, which can be used to produce a sequence of active granule-cell populations. (C) Similarity index <i>S</i>(Δ<i>t</i>) defined by Eq. (7) for the spike patterns shown in <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0033319#pone-0033319-g002" target="_blank">Figure 2A</a>. The index monotonically decreases from 1 at Δ<i>t</i> = 0 to 0.58 as Δ<i>t</i> increases, suggesting that the active granule-cell population gradually changes into another, uncorrelated population over time. Therefore, a non-recurrent sequence of active granule-cell populations is generated. (D) Reproducibility index <i>R</i>(<i>t</i>) defined by Eq. (9) for 10 pairs of spike patterns for all granule cells across two successive cycles of MF signal oscillation. The reproducibility increases towards 0.9 at the beginning of a cycle, and then linearly decreases towards 0.8, suggesting that the spike patterns of granule cells are highly reproducible across cycles.