Bees perceive illusionary distance information from rotating spirals

  • Published on

  • View

  • Download

Embed Size (px)


  • Naturwissenschaften 81, 42-43 (1994) Springer-Verlag 1994

    Bees Perceive Illusionary Distance Information from Rotating Spirals W. H. Kirchner and J. Lengler

    Theodor-Boveri-Institut ffir Biowissenschaften der Universitfit, Lehrstuhl fiir Verhaltensphysiologie und Soziobiologie, D-97074 Wfirzburg

    Spatial vision, reconstruction of the three-dimensional world from the two- dimensional images on the retinae and estimation of distances of objects in the visual field, is in vertebrates mainly mediated by stereopsis, distance- dependent convergence of the eyes, and accommodation of the lenses. In insects, there are ra re cases, in which stereopsis is used as it was shown in the praying mantis [8]. Generally, however, the spa- tial resolution of insect vision limits the use of stereopsis to distances not longer than one or two lengths of the insect's body [2]. Insects also do not have mov- able eyes and cannot adjust the focal length of their eyes to object distances. Thus, they have to rely on mechanisms different to those used by humans and other vertebrates to estimate distances. Exner [3] and Horridge [4] postulated that motion parallax should help moving insects to extract distance information from retinal images. Recently, more and more experimental evidence has been found to support this idea. In particular, is has been shown that freely flying honeybees estimate visual distances by using the speed of lateral image motion [6, 10]. Bees can learn to distinguish otherwise similar flower-like objects by the height above a background [7, 9]. It was therefore argued that the parallax between object and background is used to estimate distances, when bees fly at small distances over objects [7, 9]. How- ever, it is also possible that the apparent expansion and contraction of objects in the visual field during approach and departure are used to estimate distances. We therefore conducted an experiment to determine whether bees use looming as a cue to depth. In our experimental approach we used rotating spirals, which remain constant in size, but show either centrifugal or centripetal movement of contrast boundaries. This experimental layout also permitted us to distinguish between two alternative mechanisms of

    perception of expansion and contrac- tion. One mechanism makes use of the temporal changes in object size during approach by comparing memorized snapshots. The other mechanism, which does not involve memory, uses detectors of retinal image motion and a mecha- nism which integrates centripetal or cen- trifugal movement in the visual field. Single, individually marked honeybees, Apis mellifera, were trained to visit an artificial meadow, consisting of seven "flowers" of different sizes, all but one at the same height above a uniform white background. One of the flowers was 5 cm higher or in other experiments 5 cm lower than the others. A sheet of Plexi- glas was used to cover this meadow. The bees were trained to find a sucrose reward on the Plexiglas above the flower which was at a different height. Positions and sizes of the flowers were frequently changed throughout the training pro- cedure and all tests, so that the only reli- able cue for the bee was the height of the rewarded target. A similar setup was used by Lehrer et al. [7]. They used uni- form black flowers, whereas the artificial flowers we used had black spiral patterns on a white background. Each bee was trained 40 times and then tested (without a reward) for height preferences. After this test the arrangement of the flowers was changed and all flowers were pre- sented at the same height, but one of them was rotated clockwise, another one counterclockwise. The other spirals remained stationary. The spirals were rotated at 200 rpm; this corresponds to an angular expansion and contraction velocity of 45% for a bee at a height of 10 cm above the spirals. The trained bees were in this setup again tested for flower preferences. Our results show that bees learn to con- sider the height of the rewarded targets (Fig. 1). Their preference of a (during the training) rewarding higher flower was significant. The choice frequency for

    the raised flower was 63+ 8% (N=11 bees, n=200 decisions, p

  • after they had been rewarded on a flower further away (lower) than the unreward- ing flowers (54.8+5.5 %, N=14, n=865, p