An psychophysical and computational study on the mechanism of early vision with respect to the phenomenon of perceptual filling-in/fading

点击翻译按钮获取中文摘要

The present research project aims to understand the mechanism of early vision including surface completion and perception, by means of psychophysical and computational studies on an illusory phenomenon called "perceptual filling-in" or "perceptual fading". Major results of the project are as follows.First, the investigator built a new environment for psychophysical experiment, where visual stimuli can be presented over a large visual field and independently to both eyes, and developed computer software enabling efficient experimental execution. Utilizing this system, the investigator systematically ran a number of psychophysical experiments and examined how various stimulus parameters (or visual features) affected the perceptual filling-in/fading. Their principal results can be summarized in the following points :1.The time required for perceptual filling-in depended on the position in the visual field. In general, the time was prolonged when the target was presented in the horizontal … More direction than when it was presented in the vertical direction.2.The time for filling-in was significantly affected by the contrast of the visual stimuli when grating patterns were given to the target and surrounding regions.3.Perceptual fading could occur in a synchronized manner at two distinct regions under certain stimulus conditions, suggesting that the neural activities for two distinct areas might interact with each other.4.When an additional stimulus (flanker) was placed near the target, the target faded immediately after the flanker disappeared (In contrast, fading was rarely perceived when the flanker appeared near the target). This fading effect was much more remarkable than the conventional one: It was perceived even when the target had a higher contrast and was located at a lower eccentricity.In addition to these psychological examinations, the investigator and co-worker built a neural network model for surface completion and perception. This model consists of a number of neural units, each which represents the orientation of local element (or feature) of a 3D object. The dynamical system formed by interaction among these units brings global surface representation of the object. The result of computer simulation showed that the behavior of the neural model well corresponded to human perception of illusory surfaces.In conclusion, the present project has provided new findings suggesting that the perceptual dynamics is tightly coupled with the feature representation in early vision. Less