A cellular network model of two‐dimensional movement remembrance

Abstract

This study intends to construct an engineering model that realizes the memory/recall function for movement vision as one of the visual functions of higher animals. The idea of the cell assembly is applied. In other words, a large number of neurons form a group as the basic unit of information processing in the brain, and by the collaborative operation of neurons with some individualities in the neural network, the information processing function inherent to each unit is realized. Based on such an idea, varied functions, such as signal detection, storage, and memory recall, are assigned to the operational cells. As the first step, the retina is modeled as a two-dimensional array of operational elements that detects the time course of the light stimulus intensity and outputs the detected result. Then, the pattern of light-detecting points that adapts the extracting function of the movement information to the detection of the movement direction and velocity of the visual object is defined. A model is formed as a set of operational cells that execute the spatial summation processing of each pattern signal. The temporary memory function for the movement information is formed by a serial connection of the operational cells with inherent slow signal transmission. The memory recall function is realized by a two-dimensional array of operational cells that exhibits the excited or nonexcited state, depending on the memorized information for the horizontal and vertical components. Thus, it is shown that a function is realized where the memory of the two- dimensional movement of the visual object is represented as the transition of the excitation pattern of the operational cells. © 1999 Scripta Technica, Syst Comp Jpn, 30(14): 64–73, 1999