影子在生活中处处可见，能够提供关于物体、光源、投影平面的多种信息，可以帮助我们构建立体世界。在使用影子信息之前，视觉系统必须稳定地建立影子与其投影物体间的联系。图形相似性是解决这一问题的有效线索，研究表明，物体与影子间存在欧氏几何性质差异时可能会影响影子从属关系的建立进而影响影子的加工。拓扑性质作为几何性质中特殊的一类，是最稳定的同时也是最先被加工的。物体与影子间拓扑性质差异是否以及如何影响影子的从属关系的建立呢？具体而言，如果拓扑性质差异使得被试不能建立影子的从属关系，被试是会如何表征图形呢？被试只将图形解释为类影子（符合影子特征的刺激），还是不将图形视作影子？本研究选取了洞的有无、数量两个拓扑性质，以洞的位置（拓扑相同）这一性质作为对照，对此进行探讨。

研究一初步探究拓扑性质差异对于影子从属关系建立的影响。共进行三个实验，分别探究洞的有无、数量、位置三个性质对影子加工的影响。每个实验中图形均符合光影规律，同时设置两个条件，图形中物体与影子特征匹配与否。以特征匹配为对照，探究特征不匹配时被试是否表现更好。研究二系统地探究拓扑性质差异对于影子加工的影响。共进行三个实验分别探究各性质对于影子加工的影响，每个实验设置三种图形（是否符合光影规律），对于每种图形分别设置物体与影子特征匹配与否两个条件。探究相较于特征匹配时，被试在特征不匹配时的表现，可知特征差异是否影响了从属关系的建立。同时探究特征不匹配时，相较于图形违背光影规律时，被试在图形符合光影规律时的表现，可知存在特征差异时，被试是否仍将图形视作影子。

研究结果表明，影子与物体间拓扑性质差异会干扰被试对影子的加工，而非拓扑性质差异不会。洞的有无差异稳定的产生影响，两个研究结果一致，当图形符合光影规律时，特征不匹配时反应时显著快于特征匹配时，而当图形不符合光影规律时，特征匹配与否间未观察到差异，说明被试未建立影子的从属关系，同时，对于特征不匹配的图形，被试对三种是否符合光影规律的图形反应时相近，说明此时被试未将图形视作影子。洞的数量差异同样会影响影子的加工，当图形符合光影规律时，特征匹配与否间存在差异，在意识层面比较图形信息更容易（变化觉察任务），个体在特征不匹配时反应时更快，但是在加工图形信息时（视觉搜索任务），个体反而在特征不匹配时反应时更慢，而当图形不符合光影规律时，特征匹配与否间未观察到差异，以上结果说明被试未建立影子的从属关系；同时，对于特征不匹配的图形，被试对三种是否符合光影规律的图形的反应时边缘显著，说明被试此时仍将图形视作类影子。当物体与影子拓扑性质一致仅有洞的位置有差异时，个体仍将图形解释为物体及其影子。

以上结果说明了拓扑性质的特殊性，物体与影子间拓扑性质差异会影响影子从属关系的建立，但不同拓扑性质对于图形表征的影响是不同的，当存在洞的有无差异时，被试去除了图形的影子标签，当存在洞的数量差异时，被试将图形看作类影子，即符合影子一般特征但不是临近物体的影子。本研究说明，视觉系统加工影子时不仅判断了哪部分是影子，还判断了是哪个物体投射出的影子。本研究结果再次支持了影子忽视效应，结合以往研究对影子的加工机制进行了进一步探讨，讨论了从属关系判断中自上而下以及自下而上的影响。同时，本研究与以往研究均表明光影加工是粗糙的，针对影子和镜像加工的相似性，讨论了对于场景中冗余信息的处理方式。

Shadows can be seen everywhere in life, providing a variety of information about objects, light sources, projection planes, and can help us build three-dimensional worlds. Before using shadow information, the vision system must stably establish a connection between the shadow and its projected object. Graphic similarity is an effective clue to solve this problem, and the results show that the difference in Euclidean geometric features between the object and the shadow may affect the establishment of the subordinate relationship and thus affect the processing of the shadow. Topological properties, as a special class of geometric properties, are the most stable and the first to be processed. Does and how do differences in topological properties between objects and shadows affect the establishment of shadow subordination? Specifically, if the topological properties differ so that the subject cannot establish the affiliation of the shadow, how does the participant characterize the "shadow"? Did the participant interpret the figure only as a shadow-like (stimulus that fits the characteristics of a shadow), or did they not see the figure as a shadow? In this study, the two topological properties of the presence or absence and number of holes are selected, and the position (topology is the same) of the holes is used as a comparison to discuss them.

The first study explores whether the difference in topological properties affects the establishment of shadow correspondence. The figure conforms to the law of light and shadow, and the condition of feature matched is used as a control to explore whether the participants perform better when the features are not matched. The second study systematically explores the influence of differences in topological properties on shadow processing, which sets up three kinds of figures that differs whether can be interpreted as shadows or not. Under the condition of whether it can be interpreted as a shadow, the participants' performance when the features are not matched compared with the feature matched are explored, so as to know whether the feature difference affects the establishment of the subordination relationship. At the same time, by comparing whether there is a difference between the three figures when the features are not matched, and whether the difference in features makes the participants not regard the graphics as shadows.

The results show that the topological property difference between the shadow and the object will interfere with the participant's processing of the shadow, while the non-topological property difference will not. When the figure can be seen the object and its shadow, the RT is significantly faster than when the feature is matched, and when the figure cannot or is difficult to see the crop body and its shadow, no difference is observed; and at the same time, whether the graph can be interpreted as shadows, there is no difference. The difference in the number of holes will also affect the processing of the shadow, when the figure can be seen the object and its shadow, there is a difference between feature matched or not, it is easier to compare shadow information at the level of consciousness (change perception task), individuals process faster when features not matched, but when processing shadow information (visual search task), individuals have a slower processing rate when features do not match; When the figure cannot be seen the object and its shadow, no difference is observed between the feature matched or not; At the same time, there is a marginal difference of RT between 3 figures. When the topological properties between the object and the shadow are the same, only the position of the hole is different, the individual still interprets the figure as the object and its shadow.

The above results show that the difference in topological properties between objects and shadows will affect the establishment of shadow affiliation, but different topological properties may not equal. When there is a difference in the presence or absence of holes, the participants do not regard the figure as a shadow; when there is a difference in the number of holes, the participant regards the figure as a shadow-like but does not establish a subordination. This study shows that when the visual system processes shadows, it determines not only which part is the shadow, but also which object casts the shadow. The top-down and bottom-up effects of shadow correspondence judgment are discussed, and the results of this study once again support the shadow discount effect, and further explore the processing mechanism of shadow in combination with previous studies. At the same time, this study and previous studies show that shadow processing is rough, and the processing of redundant information in the scene is discussed in view of the similarity of shadow and reflections processing.