Abstract: Similarity serves as a central construct in theoretical accounts of many cognitive processes, such as perception, classification, learning, memory, problem-solving and reasoning. Previous research has suggested that similarity judgment and difference judgment are inverses of one another. The philosopher James Mill said: “distinguishing differences and similarities is the same thing; a similarity being nothing but a slight difference”. That is, similarity and difference judgments are logical opposites; as similarity increases, difference will decrease to the same extent. For instance, if wolves and dogs have a similarity rating of 4 on a 1-to-5 scale, they ought to have a difference rating of 2. Furthermore, in the words of William James, “To abstract the ground of either difference or likeness (where it is not ultimate) demands an analysis of the given objects into their parts”. In other words, the underlying basis for similarity is shared attributes or constituents. For example, a sparrow and a crow are perceived to be similar because of their many common features (e.g. beak, wings). However, studies have found that similarity and difference judgments are not necessarily logical opposites. At least in certain cases, the perception of similarity is not inversely related to the perception of difference. At first, in addition to shared features or attributes, the relations among features also influence on similarity judgments, and features and relations may have different roles in similarity judgments and difference judgments, which induce no-inversion especially when both relational similarity and attributional similarity are considered. Secondly, like structural relations and attributes, thematic relations also have effect on similarity, and the thematic relations are more salient for similarity than for difference. In this case, the similarity judgment and difference judgment are also not logical opposites. Behavior studies suggested that the mechanism of this non-inversion can be explained by the contrast model, structural alignment model and the dual process model to some extent. Event related potential (ERP) studies suggested that the effect of pictures and words information on similarity judgment may be different. Furthermore, the similarity could be divided to perceptual similarity and conceptual similarity based on the dual-coding model of the picture-word information, which may be considered complementary as they were viewed in the earlier literature, rather than mutually exclusive, which reflect the essence of this cognitive process. What’s more, based on previous studies, the time course of similarity should be consist of three stages at least, that is, selective attention (N1), perceptual processing (P2) or semantic processing and integration (N400 or P600), and comparison and judgment (P3). Comparison and judgment is the core process of similarity judgment, while the former two processes are the basis of comparison and judgment. Furthermore, the dual-processes of similarity judgment can be used to inertpret some phenomenon in cognitive psychology, such as induvtive reasoning. In conclusion, previous results suggested that two complementary processes, namely, perceptual similarity and conceptual similarity, should be differentiated while exploring similarity. Future studies should further explore this question from the views of developmental psychology and cognitive neuroscience.

Key words: similarity judgment, difference judgment, structural alignment model, the dual process model, the dual-coding model

摘要： 本文阐述了相似性判断和差异性判断不对称性的相关研究。首先介绍了不对称性的发展及理论解释。其次，根据事件相关电位的研究结果，阐述了图-词信息的双编码模型对相似性判断的影响。接着分析和介绍了相似性判断的双编码过程及其时间进程。最后，作者提出有待深入研究的几个问题：（1）差异性策略在分类和推理中的应用；（2）相似性双编码过程的实证研究及其应用；（3）从发展认知神经科学的角度探讨不对称的认知神经机制。

关键词: 相似性判断, 差异性判断, 结构匹配模型, 双加工过程, 双编码模型