Visual Search And Information Integration During Scene Identification

In everyday life, people will inevitably meet a variety of complex life situations. To identify scenes is the basis of dealing with life events properly. Whether it is identified will have a direct impact on people's life. Therefore scene identification has aroused the attention of psychology researchers. They have shown common interest in how to identify scenes. There are, however, arguments with theories and researches concerned, that is, whether the scene identification is the result of bottom-up visual information input or of top-down knowledge influence. Dual-Path Model of Scene Processing, integrated process of both bottom-up and top-down, provides an explanatory framework for scene identification in general.Within the framework of Dual-Path Model, the present study sets out to investigate the information search and integration of scene identification under different times and information acquiring conditions as well as the developmental features of different age groups. The participants were teenagers aged 15 to 23 years old. The testing materials were scenes related to living safety. Some paradigms and techniques, such as study-test, moving-window technique and so on were adopted.The present study consists of two parts. The Experimental tasks and general results are as follows:In information search research, study-test, one-shot paradigm, flicker paradigm were adopted to investigate information search and the influence of object changing in scene identification process. In study-test condition, subjects had enough time to identify scenes and acquire information through eye movement. Experiment 1 investigated the influence of object changing information. Experiment 2, separating the semantic information and perceptive information by saliency map analysis technique, designed two experimental conditions to further reveal whether the guiding role of object information is influenced by semantic constraint or by perceptive features. In one-shot condition, influence of eye-movement and long-term memory was controlled effectively through rapid presentation of a series of materials. Flicker paradigm combined the study-test paradigm and one-shot paradigm. Materials were also presented rapidly as in the one-shot approach, but it could be presented repeatedly before subjects made a decision. In this way, subjects could acquire and process scene information through eye movement.In information integration research, disappearing text paradigm and moving window technique were used. Disappearing text approach provided technical support for study on time process of information acquisition from foveal. In this condition, time process of information acquisition from foveal was studied by delaying presenting time. Moreover, by controlling fixation angle, integration of information from foveal and parafoveal was studied. In experiment 5, three delayed presentation types were designed to study information acquire and integration from parafoveal in different conditions. Moving window technique was adopted to reveal information acquisition and integration from foveal by controlling viewing window size, which could effectively control influence from parafoveal. In experiment 6, three viewing window types were designed to study information acquire and integration from foveal in different conditions.The main results were as elaborated below.(1) In the process of information searching when subjects had enough time toidentify the scenes, objects had priority to guide information searching, which is related to semantic constraints. The stronger the scene semantic constraints were, the greater the guiding role of object information would be; on the contrary, the weaker the scene semantic constraint was, the smaller the guiding role of object information would be.(2) In the process of rapid scene identification, if subjects were unable to get scene information through eye movement, identification of rapid presentation in one-shot condition was actually a kind of motion perception. On the contrary, if subjects were able to acquire scene information through eye movement, rapid scene identification would be interrupted by semantic processing.(3) In the process of information acquire and integration through eye movement, when the presenting time was in 100ms, information from fixations would be influenced. Moreover, the level of integration would be lower. When the presenting time is between 100ms and 200ms, subjects would be able to acquire information from fixations and integrate information from foveal and parafoveal more effectively.(4) The integration of information from fixations was relevant to the amount of information available. When subjects got scene information from a 1-2^? visual angle, it takes more time to integrate information. With the viewing angle enlarged, the amount of information increased, and the influence of top-down knowledge strengthened, the time needed in scene information integration decreased accordingly.(5) There were age differences in scene information search and integration process. Compared with the younger group, the elder groups were more likely to be influenced by semantic constraints in the process of information searching. In the process of scene information integration, younger group's information integration level was lower, while elder group integrated information more effectively.