Item

Abstract

The output of a motion detector depends on the direction of pattern motion, relative to the axis defined by its two input elements. Usually a cosine shaped directional sensitivity characteristics is assumed: The response is strongest for pattern motion along the detector axis, reversing its sign for motion in opposite direction; for motion perpendicular to the detector axis it is expected to be zero, with intermediate values for oblique motion directions. However, geometric considerations show that this expectation is by no means trivial. When a periodic pattern moves along a direction a relative to the detector axis, the spatiotemporal intensity distribution along the detector axis can be described by its apparent spatial wavelength and its apparent velocity, which both vary with 1/cosα. In consequence, a motion detector depending exclusively on the apparent velocity of the stimulus would respond strongest to gratings moving perpendicular to the detector axis, whereas motion along the detector axis would yield the smallest response in such a detector. The response of a motion detector of the correlation type, on the other hand, is determined by the ratio between velocity and wavelength, the temporal frequency, which is not influenced by the direction of pattern motion, and by the ratio between sampling base and the wavelength. The latter feature leads to a cosine-shaped directional characteristics for large pattern wavelengths. However, for smaller wavelengths specific deviations from a simple harmonic are expected. These expectations were confirmed by the simulation of an elementary motion detector model and extended for a slightly more elaborated model. Representing a biological motion detecting unit of the correlation type, the directional characteristics of the H1 interneuron in the fly's brain was investigated electrophysiologically, and compared to the simulations.