What is the unit of visual working memory (feature, object or location)?How are these items organized and related to one another? Ever since thepioneering work by Luck and Vogel (1997), the unit question has been acentral issue in the study of visual working memory, and also been a hottopic in cognitive science. This issue not only extends and deepens thestudy on the visual working memory capacity, but also plays an importantrole in understanding the nature of visual working memory. Based on theprevious researches done by other researchers, this article deeply studieshow the texture, shape and color of picture are stored and bound in visualworking memory. Selecting another fundamental feature of picture-textureas stimulus, we may increase the dimension and information load of picture,and control the difference of location of feature values in the samedimension with the help of the repeatability of local texture features(textons), so that we can systematically examine the five hypotheses aboutthe unit of working memory capacity and its mechanism, and get a betterunderstanding for the storage and representation of visual working memory. The first chapter of this article is literature study, in which theresearches on the feature binding problem of visual perception have beenreviewed, and the interpretations of the mechanism of the binding problemhave been compared and analyzed. The second chapter reviewed and analyzed the history and status inquo of the researches on visual working memory, and assorted thehypotheses about the unit of visual working memory capacity and itsmechanism into five classes: (1)feature-based storage mechanism;(2)feature-object dual-storage mechanism;(3)features and binding with thesame storage mechanism;(4)strong sense of binding storage mechanism；(5)feature-object dissociation storage mechanism. Based on literature review, the third chapter not only analyzed thelimits and shortcomings of previous researches, but also put forward plan to iiisolve them. Subsequently, making five experiments to validate the plan. In Experiment 1 and Experiment 2, in order to find out they are storedindependently or bound together, the storing and binding of texture andcolor as well as texture and shape has been studied separately. In Experiment 3, the storing of pictures with three dimensions (texture,color and shape) has been studied, which can give some information aboutthe relation between the information load of picture and the detail of its'storage. By controlling the dimension of picture, Experiment 4 investigatedwhether feature binding need extra resources of working memory, that is tosay whether the information load of picture influences its storage in visualworking memory. Experiment 5 selected texture constitutes two colored circles as stimulito study whether or not these two colored circles can been bound as anobject in visual working memory, which can definitely support or deny thehypothesis of strong sense of binding storage mechanism. Having generally discussed the results of the five experiments, the lastchapter made a summary of this study and come to conclusion as follows: (1)Just as color and shape, texture is stored independently in visualworking memory. (2)The pictures with texture-color, texture-shape as well as the picturewith the three dimensions are stored as a whole object in visual workingmemory. The picture with two or three dimension features is maintained inbound units during memory, and the binding of two features is automatic,which adds evidence to the hypothesis of feature-object dual-storagemechanism, while that of three features needs extra resources of memory,which supports the account of features and binding with the same storagemechanism. (3)The two colored circles in the same texture have not been chunkedinto one memory unit, which denies the model of strong sense of bind