Fig. 4.

Spike-time precision and spike frequency for 3 different levels of noise in the model. A, left: spike-time precision increases with both input gain and synchronicity for recorded neurons (shaded) and the model (solid). The relationships between precision and gain are preserved in the model, although precision values are slightly higher. Middle: spike frequency increases with input gain at high input synchronicity but actually decreases for intermediate or no input synchronicity. Spike frequency is normalized by the value at a gain of 0.5. The mean spike frequencies at a gain of 0.5 for the model were 12.4 (high synchronicity), 12.7 (intermediate synchronicity), and 12.7 Hz (no synchronicity). The mean spike frequencies at a gain of 0.5 for the recordings were 10.0 (high input synchronicity), 9.0 (intermediate synchronicity), and 10.2 Hz (no synchronicity). Right: the spike-triggered average (STA) of inhibitory conductance (normalized to the mean conductance value) for high input synchronicity and a gain of 16 is similar for both the model and recorded neurons. The transient drop in STA conductance indicates that spikes were triggered by brief pauses in Purkinje cell input. B and C: the relationships between precision and output spike rate with input gain and synchronicity and the STA of inhibitory conductance are shown for additional levels of low noise (B) and high noise (C). For low noise, the mean spike frequencies at the gain of 0.5 were 12.4 (high input synchronicity), 12.6 (intermediate synchronicity), and 12.6 Hz (no synchronicity). For high noise, the mean spike frequencies at the gain of 0.5 were 12.4 (high input synchronicity), 12.7 (intermediate synchronicity), and 12.7 Hz (no synchronicity).