Here, the spatial gradient dI/dx is approximated by Proteasome structure the brightness difference dI, of the pattern, I, sampled at two neighboring image points separated by a distance, dx. Both input signals become high-pass filtered, approximating the temporal derivative, and then added together. These two quantities are then divided by each other yielding an estimate of the local image velocity (Srinivasan, 1990). This estimate

will only depend on the image velocity and not on the spatial structure of the moving pattern because the local image contrast is expressed in a steeper spatial, as well as in a steeper temporal gradient: Dividing them leads to a cancellation of image contrast. However, as attractive as the gradient model of motion detection might appear, most models that were proposed to account for biological motion detectors actually do not calculate the spatial and the temporal gradient of the moving image. They rather correlate the brightness values measured at two adjacent image points with each other after one of them has been filtered in time (correlation model, Figure 1D). Consequently, their output is not proportional to image motion but rather deviates from it in a characteristic way. In fact, this deviation Tyrosine Kinase Inhibitor Library manufacturer has been the crucial hint for researchers in motion

vision to propose exactly this type of model. The first correlation until detector was proposed

on the basis of experimental studies on the optomotor behavior of insects (Hassenstein and Reichardt, 1956, Reichardt, 1961 and Reichardt, 1987). This correlation detector is commonly referred to as the Reichardt detector (van Santen and Sperling, 1985), and has also been applied to explain motion detection in different vertebrate species including man (for review, see Borst and Egelhaaf, 1989). Such a detector consists of two mirror-symmetrical subunits. In each subunit, the signals derived from two neighboring inputs are multiplied with each other after one of them has been shifted in time by a temporal low-pass filter. The final detector response is given by the difference of the output signals. Various elaborations of the basic Reichardt model have been proposed to accommodate this motion detection scheme to perform in a species-specific way. Perhaps the simplest correlation-type movement detector has been proposed by Barlow and Levick to explain their experimental findings on DS ganglion cells in the rabbit retina (Barlow and Levick, 1965). The Barlow-Levick model (Figure 1E) is almost identical with respect to its layout but with only one subunit of the basic Reichardt model. It consists of two input lines carrying the brightness signals which are compared after one of the signals has been delayed.