Small object detection neurons in female hoverflies

Abstract

While predators such as dragonflies are dependent on visual detection of moving prey, social interactions make conspecific detection equally important for many non-predatory insects. Specialized 'acute zones' associated with target detection have evolved in several insect groups and are a prominent male-specific feature in many dipteran flies. The physiology of target selective neurons associated with these specialized eye regions has previously been described only from male flies. We show here that female hoverflies (Eristalis tenax) have several classes of neurons within the third optic ganglion (lobula) capable of detecting moving objects smaller than 1 degrees . These neurons have frontal receptive fields covering a large part of the ipsilateral world and are tuned to a broad range of target speeds and sizes. This could make them suitable for detecting targets under a range of natural conditions such as required during predator avoidance or conspecific interactions.

Figure 1

Raw responses of a hoverfly (type II) neuron tuned to small targets. (a–c) Response to a 0.8° wide black target presented at 50° s⁻¹ (stimulus period indicated by black bar) and three different vertical heights. The dotted line indicates the time the target crossed the midline and entered the contralateral visual field. (d) A control stimulus, consisting of a drifting sinusoidal grating (1 Hz, 0.1 cycles per degree). (e) A histogram of responses to 42 presentations of a target that drifts from left to right and then back again. Asterisks (^*) indicate when the target crossed the midline.

Figure 2

Height tuning for three classes of STMD neurons. Normalized responses (see §2) are shown averaged from the time the target traversed the receptive field. For STMD type I we show responses for length tuning in two parts of the receptive field—open squares denote scans at an elevation of 60°, while filled squares show the same tuning at 40° elevation. Inset shows the response to drifting sinusoidal gratings (1 Hz, 0.1 cycles per degree, n=7). Errorbars denote standard error of the mean averaged from several neurons (n=1 for type I, n=5 for type II, n=3 for type III).

Figure 3

Raw responses from a graded type III STMD neuron to (a) horizontal and (b) vertical motion of black targets on white backgrounds. (c) The response to horizontal motion when the target is white, traversing a black background.

Figure 4

Receptive fields for three classes of STMD neurons. (a, b) STMD type I and II gave purely spiking responses, while (c) STMD type III responded with spikelets, (d) riding on depolarization. The maps show data from individual recordings to target motion from left to right. The response is colour coded to show (a–c) the response spike frequency or (d) membrane potential change from baseline.

Figure 5

Speed tuning of type II and III STMD neurons, averaged from the time the target (a 0.8° dark square) was traversing the receptive field. (a) Data plotted as normalized spike rate. (b) The same data after transforming into spatial coordinates (spikes/degree). Error bars denote standard error of the mean (n=4 for type II, n=1 for type III) and the lines join the low-pass filtered mean.