Visual speed estimation is used by the honeybee, Apis mellifera, for a wide range of behaviors such as visual odometry, control of landing and the centering response. Although this system is well understood behaviorally, the neuronal basis for this circuit has remained elusive. Electrophysiological recordings have yet to reveal neurons present in the optic lobes clearly sensitive to speed. In addition, most computational models of speed sensitivity use directional motion detectors based on the Hassenstein-Reichardt (HR) model, sensitive to both image spatial frequency and contrast. However, numerous behavioral experiments have shown that visual speed estimation is largely independent of image spatial frequency and contrast. We have recently proposed that individual non-directional motion units show a response proportional to speed over a range of spatial frequencies. In this poster I will present a cybernetic model that reproduces the centering response and is capable o f visual odometry. In the model two arrays of non-directional motion detectors (one for each eye) are collated and the outputs are compared to produce a motor output. The biological relevance of this model is supported by the properties of the Tm1 cell present in the optic lobe of the fly.
J. P. Dyhr and C. M. Higgins, "Visual Estimation of Speed Using Non-Directional Motion Units" (poster), 3rd Gordon Conference on Neuroethology, Magdalen College, Oxford University, UK, August 2005.