Is the Locust DCMD A Looming Detector?

To a prey, a salient visual characteristic of an approaching predator is ‘looming’, a figure of average subtended visual angle increasing as some function of time. If the approach is at constant velocity, two important characteristics of the looming figure are an ever increasing velocity (i.e. acceleration) of sideways motion of edges away from the centroid of the figure, and the opposed motion of opposite edges. These characteristics make looming objects distinct from translation of single objects across the visual field, so that the visual looming characteristics may have value in warning of approaching predators. In the locust, a visual neurone, the DCMD (descending contralateral movement detector) in the ventral nerve cord forms an important input to the interneurones triggering the jump response in the locust (Pearson, Heitler & Steeves, 1980; Pearson & Robertson, 1981). Thus it is of considerable interest to determine whether the DCMD responds preferentially to looming visual stimuli.

The responses of the DCMD have been characterized quantitatively and in considerable detail (Palka, 1967; Rowell, O’Shea & Williams, 1977; Pinter, 1977, 1979). The DCMD responds vigorously to translation movements across the visual field of small objects (subtending 2–10° of visual angle at the surface of the compound eye) within the receptive field of approximately a hemisphere. The polar axis of this hemispheric visual field runs approximately lateral to and through the centre of the animal’s contralateral compound eye. Movements of large textured objects, or of the animal itself in a textured visual environment elicit only weak DCMD responses. Schlotterer (1977) measured the response of locust DCMD to looming and translating objects, and found that looming objects were more potent stimuli than translating stimuli. However, we have found that translating objects are as potent stimuli as looming objects.

We have examined this question in a series of experiments on Schistocerca americana, both male and female from the University of Washington colony, and on a male from the University of British Columbia colony. The ventral nerve cord was exposed between the first and second thoracic ganglia and gently hooked with a silver-wire electrode, insulated with Vaseline; a ground electrode was placed in the abdomen, if a the data presented here, the interstimulus interval (I SI) was 30 s, with the two initial trials of a run of one stimulus type discarded to lessen effects of habituation. (Similar results were confirmed with 60 s I SI’s.) The ipsilateral eye and the three ocelli were covered with opaque acrylic paint and the ears were waxed closed. Control experiments were run with an optical window to ascertain whether air currents produced by looming stimuli were affecting DCMD responses. In no case were aircurrent effects found.

In these experiments the animal’s compound eye was presented with looming and translating stimuli made from black paper and attached to an arm joined to the moving carriage of an X–Y recorder which was driven by function generators. The light background behind these dark stimuli was white illustration board illuminated by two 15 W d.c.-powered fluorescent lamps with reflectors, at a distance of approximately 1· 5 ft. The luminosity of this background was approximately that under daylight from an overcast sky.

In Fig 1 the spike train recorded from the DCMD is shown for a single trial of both a typical looming stimulus (parameters as in Schlotterer, 1977) and a comparable translating stimulus. The time histograms of spikes for 10 trials of these same responses are shown in Fig. 2 A, with spike frequency counts (peak and average) shown in Table 1. In the sequence of conditions the looming followed the translating stimulus, but this entire experiment, lasting for 3· 75 h, was without decrement in sensitivity to. control stimuli. It can be seen that looming is not a more potent stimulus than translation. All experiments on other animals verified this result. In addition, looming stimuli of larger range (5–90° and 6–135° over 1 s) could not be shown to be more potent than comparable translating stimuli. Table 2 shows that looming of sectors of dark discs (one-half and one-quarter) centred on the eye were as potent as looming of entire circles, and accelerated motion of edges of both fixed and expanding size was not more effective than constant-velocity motion. Thus looming is not a more potent stimulus than translation. Furthermore, the salient components of the looming stimulus such as acceleration of edges (in units of visual angle) and opposed motion of opposite edges are not especially potent DCMD stimuli. Schlotterer’s (1977) conclusion, that looming stimuli are more potent than translating stimuli, results from his choice of a rather weak translating stimulus, one whose velocity never approaches that achieved by the edges of the looming stimulus.

Examination of Fig. 2 A will show that the looming response is bimodal (as in Schlotterer, 1977), and that the later peak is larger than the earlier peak. This hiatus in activity is a robust property of the DCMD response, and is not a function of the frequency of spike firing during the first peak. Since the later peak is the larger, we examined the relative effects of looming and translation on it. These results are demonstrated in Fig. 2B and Table 3, where again no preference for looming over translations of objects can be inferred. Fig. 2B also demonstrates a possible systematic relationship between edge size and preferred velocity, but that is the subject of a separate investigation. The peak spike frequency certainly does increase with the velocity of a moving edge, but not with acceleration (Tables 2, 3). The increase of peak frequency at higher velocity may be a function of ‘lack of time’ to set up lateral inhibition (Rowell et al. 1977; Brodie, Knight & Ratliff, 1978 (in Limulus)), as can be demonstrated from quantitative models, such as the Hartline-Ratliff equations in their complete time- and space-dependent form (Brodie et al. 1978).

The form and magnitude of the DCMD response to looming and translation varies with stimulus location in the receptive field, but within any given region no preference for looming could be found. Further, no preference can be found for looming stimuli for parameters beyond those stated above, e.g., very small size expanding rapidly or very large size expanding slowly.

The existence of looming and changing-size responsive neurones is an old and intriguing question. They have been demonstrated and their properties measured in visual cortex of the cat (Regan, Beverley & Cynader, 1979). However, our results do not confirm Schlotterer’s (1977) suggestion that the locust DCMD responds preferentially to looming stimuli.