A Neural Network Approach for Analyzing the Illusion of Movement in Static Images

Resumen

The purpose of this work is to analyze the illusion of movement that appears when seeing certain static images. This analysis is accomplished by using a biologically plausible neural network that learned (in a unsupervised manner) to identify the movement direction of shifting training patterns. Some of the biological features that characterizes this neural network are: intrinsic plasticity to adapt firing probability, metaplasticity to regulate synaptic weights and firing adaptation of simulated pyramidal networks. After analyzing the results, we hypothesize that the illusion is due to cinematographic perception mechanisms in the brain due to which each visual frame is renewed approximately each 100 msec. Blurring of moving object in visual frames might be interpreted by the brain as movement, the same as if we present a static blurred object.