Supplementary Figure 5: The slow decay in the odor responses of disinhibited PNs is primarily due to synaptic dynamics, with a minor contribution from ORN firing rate dynamics.

a) ORN firing rates evoked by a dense randomly fluctuating stimulus (valve open 50% of the time). Data represent the average of several recordings from VM7 ORNs. These ORN firing rates decay slowly over time. b) Mean PN membrane potential responses to the same odor stimuli (n = 7 PNs in glomerulus VM7). This is a subset of the data shown in Figure 1g, focusing here on VM7 PNs only. c) Mean responses of the same cells following blockade of inhibition. This should represent the purely feedforward response of the PN. Note that when ORN responses have reached steady-state, PN responses are still slowly decaying. Thus, much of the slow decay in PN responses cannot be inherited from ORN firing rates, and likely arises from the dynamics of ORN-to-PN synapses. This slow decay is normally masked by inhibition. d) Output of the single component model without inhibition (same as in Figure 1). Note the slow decay in ORN firing rates is not transmitted to the PNs, because at high presynaptic firing rates, PNs are relatively insensitive to small changes in presynaptic rate. Instead, PN response dynamics are dominated by the dynamics of synaptic transmission. e) Output of a two-component model without inhibition (acting on the ORN spike rates shown in panel a). Parameters of the fast component were the same as in Figure 2 (fit to EPSCs in IMI). Parameters of the slow component were adjusted to maximize the fit between model output and the disinhibited PN membrane potential shown in panel c. This model captures the slow decay in PN odor responses when inhibition is blocked. This model was used in Figures 7 and 8.