Study Of Crowding Effect Based On Scene Gist Extraction

Recognition of objects presented in peripheral visual field is damaged by distractors surrounding the objects,referred as crowding effect,manifesting the ability of human visual system to recognize objects in complex environments.However,the mechanism underlying crowding effect remains equivocal.Feature integration stage hypothesis indicates that both targets and distractors presented at the periphery get access to our visual system but with low-level features mis-integrated or averaged,hence targets fail to be integrated and detected as the original objects.Attention resolution hypothesis,however,suggests that the low spatial resolution of peripheral visual field limits targets to be segregated from distractors and to reach consciousness.Exploring the visual stage where crowding effect occurs help to clarify this issue.Moreover,previous studies concentrate more on investigating the crowding effect of simple stimuli such as letters or Gabor patches.Less is found to examine such effect of scenes and other more complex stimuli,especially examine how low-level or high-level information contributes to crowding effect.To address these issues,using scene recognition paradigm in conjunction with crowding technique,the current study investigated whether scene gist extraction could get through crowding and further identified the possible mechanisms of crowding effect.The current study consisted of three experiments.In experiment 1,we varied the types of flankers and the eccentricity of targets to examine the crowding effect in scene gist extraction and in which stage it occurred.Furthermore,we conducted experiment 2 to investigate whether the duration of targets together with flankers affected the crowding effect in experiment 1.In exper:iment 3,we tested whether the semantic information of the target scene surrounded by distractors could be processed and lead to semantic priming effect.Our study drew three main conclusions.In the process of extracting the gist of complex multi-feature scenes,crowding effect occurred at both low-level feature processing stage and high-level semantic information processing stage.The duration of the scene remarkably affected the performance for basis-level categorization task but showed no effect to top-level categorization task.Moreover,the crowded scene could be processed and correctly integrated in that its semantic information induced significant negative priming effect.Taken together,the current study found crowding effect during the visual process of scene gist,and the crowded scene information could be lost at different stages.Additionally,the negative priming effect induced by crowded scene failed to support the feature integration stage hypothesis.