Depth cues, behavioural context, and natural illumination: some potential limitations of video playback techniques

By expanding on issues raised by D’Eath (1998), I address in this article three aspects of vision that are difficult to reproduce in the video- and computer-generated images used in experiments, in which images of conspecifics or of predators are replayed to animals. The lack of depth cues derived from binocular stereopsis, from accommodation, and from motion parallax may be one of the reasons why animals do not respond to video displays in the same way as they do to real conspecifics or to predators. Part of the problem is the difficulty of reproducing the closed-loop nature of natural vision in video playback experiments. Every movement an animal makes has consequences for the pattern of stimulation on its retina and this ”optic flow” in turn carries information about both the animal’s own movement and about the three-dimensional structure of the environment. A further critical issue is the behavioural context that often determines what animals attend to but that may be difficult to induce or reproduce in an experimental setting. I illustrate this point by describing some visual behaviours in fiddler crabs, in which social and spatial context define which part of the visual field a crab attends to and which visual information is used to guide behaviour. I finally mention some aspects of natural illumination that may influence how animals perceive an object or a scene: shadows, specular reflections, and polarisation reflections. Received: 23 November 1999 / Received in revised form: 9 February 2000 / Accepted: 10 February 2000