We have developed a model of direction selectivity amongst distal retinotopic neurons in the fly visual system involving three elements at the level of receptor terminals, and two successive relays that are distal to wide-field summating elements. This model would account for neural arrangements not only in Diptera but across widely differing optic lobe architectures in advanced and basal arthropods. The elements of the model are based on known connections and/or activities of lamina amacrine cells, two efferent neurons (L2,T1), and columnar relays (Tm1) that map these efferents onto T5 direction and orientation selective interneurons. The present model, whose outputs are predicted by the Reichardt-Hassenstein model and are collated by wide-field tangential neurons (HS, VS), has an internal organization that is quite unlike the classic motion-detection circuit. The first level of the model (lamina) relies on differentiation of flicker from motion, without regard to direction of motion. The second level of the model relies on a spatial mapping function of retinotopic arrays of non-directional motion sensitive outputs from the lamina onto T5 neurons that occur as four elements for each visual sampling unit thereby providing a four-directional system of motion detectors.
C. M. Higgins, T. Vaneck, P. Joshi, and N. J. Strausfeld, "A Model for Directional Selectivity in an Insect based on Non-directional Motion Cells and Appropriate to Cross Phyla Comparisons," Society for Neuroscience annual meeting, San Diego, California, November 2001.