The Drosophila antennal lobe is subdivided into multiple glomeruli, each of which represents a unique olfactory information processing channel. In each glomerulus, feedforward input from olfactory receptor neurons (ORNs) is transformed into activity of projection neurons (PNs), which represent the output. Recent investigations have indicated that lateral presynaptic inhibitory input from other glomeruli controls the gain of this transformation. Here, we address why this gain control acts "pre"-synaptically rather than "post"-synaptically. Postsynaptic inhibition could work similarly to presynaptic inhibition with regard to regulating the firing rates of PNs depending on the stimulus intensity. We investigate the differences between pre- and postsynaptic gain control in terms of odor discriminability by simulating a network model of the Drosophila antennal lobe with experimental data. We first demonstrate that only presynaptic inhibition can reproduce the type of gain control observed in experiments. We next show that presynaptic inhibition decorrelates PN responses whereas postsynaptic inhibition does not. Due to this effect, presynaptic gain control enhances the accuracy of odor discrimination by a linear decoder while its postsynaptic counterpart only diminishes it. Our results provide the reason gain control operates "pre"-synaptically but not "post"-synaptically in the Drosophila antennal lobe.
Keywords: Drosophila; antennal lobe; concentration invariance; decorrelation; gain control; odor discriminability; postsynaptic inhibition; presynaptic inhibition.