Modelling Fly Motion Vision - Computational Neuroscience: A Comprehensive Approach - page 416

Figure 14.15
Possible mechanism underlying the selectivity of FD-cells for relative motion in the
fly visual system. A The stimulus consists of a 2D random dot pattern covering both
a bar and a background (top center). The bar is moving in front of the background
to the right, stops, and to the left again. Motion is shown as an x-t plot for a center
cross section through the stimulus pattern (top right). This stimulus is processed by a
2D array of motion detectors which then feed onto the dendrites of HS-cells (bottom
left) and FD-cells (bottom right). HS-cell dendrites contact the dendrites of CH-cells
(bottom center) which in turn inhibit FD-cells in a dendro-dendritic manner. The
excitation level of the dendrite of all three tangential cells is shown as a false-color
image (scale bar bottom left) (See color insert.).
when large-field background motion is presented ( Figure 14.16) . However, before
any selective responses of FD-cells to small-field motion can be obtained from such
dendritic activity patterns, additional nonlinear operations have to be postulated like
e.g., rectification of dendritic signals and/or local spike generation [76, 77]. Never-
theless, the present experimental findings support such a biophysical mechanism to
underlie the particular response properties of FD- and CI-cells, representing a rather
unique example of how image processing is done within the dendrites of individual
neurons (Cuntz et al., in preparation).
14.4 Conclusions
Fly motion vision represents one of the best studied biological systems where the
modelling efforts summarized above went hand in hand with the experimental in-
vestigations. However, as the reader might have noticed, the models of local motion