Abstract
Rapidly looming objects are highly salient to most animal visual systems. The sensory processing of such stimuli is now well understood in birds and insects. We conducted the first analogous study in lizards, concentrating on the ecologically-realistic challenge posed by an approaching aerial predator. In an initial experiment, we presented high-resolution digital video footage of a trained raptor flying toward the camera, together with two control sequences, one in which this motion was reversed and another depicting the bird stationary on the handler's fist. Lizards fled only from the approaching predator, suggesting that direction of movement is a necessary characteristic and verifying the effectiveness of video stimuli in this context. Two additional experiments then explored the processing of motion and of morphological attributes separately. We presented a series of symmetrically-expanding disks, systematically manipulating area / time characteristics to test looming sensitivity in the absence of other cues. Lizards oriented significantly more frequently to a sequence matching the area change of the approaching predator than to any other. Comparisons show that this response was specific to an exponential increase following a period of slow change, a pattern remarkably similar to those described in other taxa. In the final experiment, we presented a range of stimulus shapes, all with identical area and movement. Lizards were most responsive to a realistic raptor silhouette. Controls allowed us to exclude the possibility that this result was attributable to looming rate, size, or the axis of asymmetric expansion. Basic parameters such as the amount of edge in a shape likely could account for some variation in the frequency of orienting, but the finding that a normally-oriented raptor was more effective than an inverted one requires us also to postulate higher-level processing. We conclude that response to an approaching aerial predator depends upon a hierarchical series of cues, including area / time profile, edge length, shape, and orientation. The integration of this information will be an important problem for future work.