A Functionally Equivalent Artificial Neural Network Model of the Prey Orientation Behavior of Waterstriders (Gerridae)

Authors

  • Michael R. Snyder

    Corresponding author
    1. Department of Psychology, University of Alberta, Edmonton, Canada
      Department of Psychology, P-220 Biological Sciences Building, University of Alberta, Edmonton, Canada, T6G 2E9. E-mail: msnyder@psych.ualberta.ca
    Search for more papers by this author

Department of Psychology, P-220 Biological Sciences Building, University of Alberta, Edmonton, Canada, T6G 2E9. E-mail: msnyder@psych.ualberta.ca

Abstract

Waterstriders, a family (Heteroptera, Gerridae) of predacious insects, orient toward the source of water surface vibrations. We describe an artificial neural network that simulates a waterstrider's discrete rotational movement towards a prey item and compare the results to published data. A back-propagation network with six input units, each corresponding to a vibration receptor on a leg of the waterstrider, and two output units corresponding to the elicited angle of rotation, was used. The network was trained with a full complement of receptors to rotate towards the point source of a surface vibration. When the network was tested with all receptors present, a linear relationship was found between the desired and obtained rotational angles. Lesioning of one or two receptors resulted in marked deviation from linearity within the angular range of detection corresponding to that of the amputated receptor(s), while amputation of three receptors resulted in the network rotating contralaterally to all vibrations originating ipsolaterally to the lesioned side. All trials produced results that corresponded qualitatively to published behavioral data.

Ancillary