Supplementary Figure 6: Quasi-steady-state responses of our models.

a) Responses of model PNs after 2 s of stimulation, with and without tonic presynaptic inhibition. In this model, ORN-to-PN synapses have both fast and slow components. (This model is equivalent to the one shown in Figure 8, but with tonic inhibition fixed at the pre-odor level, instead of dynamic inhibition.) Here, simulated stimuli were 2-s steps of steady ORN firing rates from zero to a range of levels (1-200 spikes/s). The time-averaged PN membrane potential was measured over the window from 1 s to 2 s after the onset of the stimulus, and was plotted as a function of presynaptic firing rate. As shown in previous theoretical studies (refs. 7,8) the postsynaptic response saturates rapidly as a function of presynaptic firing rate, and adding increasing amounts of tonic presynaptic inhibition causes the postsynaptic response to saturate at a higher presynaptic firing rate. Small amounts of presynaptic inhibition have a more profound effect on the response to low presynaptic firing rates than on the response to high presynaptic firing rates.

b) Model responses after 2 s of stimulation (calculated as in panel a), as a function of presynaptic firing rate for the different model components. The fast component of excitation at ORN-to-PN synapses has a relatively small dynamic range for encoding tonic changes in presynaptic firing rate, while the slow component shows a larger dynamic range. A model with both fast and slow components shows a larger dynamic range but saturates at a low presynaptic firing rate. Both tonic and dynamic inhibition cause the postsynaptic response to saturate more gradually as a function of presynaptic firing rate. Here dynamic inhibition was modeled as in Figure 8, and tonic inhibition was fixed at the pre-odor level of inhibition in the dynamic model.