Neural Code Subserving Feed-Forward Inhibition in a Collision Detection Circuit

Feed-forward inhibition shapes the activity of many neural circuits, although little is known about how (and even if) neurons mediating feed-forward inhibition encode detailed time-varying sensory information. Locusts possess an identified neuron preferentially responding to objects approaching on a collision course that plays a central role in the generation of visually-guided escape behavior. This neuron, called the Lobula Giant Movement Detector (LGMD), receives both feed-forward excitation and inhibition. While feed-forward excitation encodes a function of the angular speed of expansion during the simulated approach of an object, the information conveyed by feed-forward inhibition remains to be characterized. We identified presynaptic, inhibitory neurons to the LGMD using spike-triggered membrane potential averaging. Unexpectedly, these neurons had wide receptive fields and only a small population of them was needed to account for most of the information they conveyed to the LGMD. Together, the time-varying spike rate of these neurons accurately encoded the instantaneous angular size of an approaching object. Our results show that feed-forward inhibitory neurons convey detailed information about a stimulus parameter needed to generate escape behaviors, and thus likely play a critical role in their timing.