When two objects carry the same category label, we tend to perceive that these objects are similar. When two objects carry different category labels, we perceive that these objects are dissimilar. How does this happen? In an attempt to clarify the effect of category labels on similarity judgments, pictures of animal tissues were presented with fictitious labels and participants judged the similarity of the pictures. The perceived similarity increased when the labels highlighted the interrelatedness of features; the effect of labels was absent when the labels did not highlight the interrelatedness of features. The results indicate that category labels help clarify inter-relationships of features, and modify our perception of similarity. When two objects carry the same category labels, people tend to think that they have features in common. When they carry different labels, people perceive that these objects have different features. How does this happen? One theory claims that verbal labels attract extra attention, and because of the attention they attract, verbal labels are particularly diagnostic in human induction (Sloutsky & Fisher, 2004). Another theory argues that category labels, unlike other feature labels, have special properties and provoke implicit assumptions about category members; as a result, category labels help integrate underlying features (Gelman & Markman, 1986; Yamauchi & Yu, 2008). In this brief article, we propose another explanation. We argue that category labels help make the interrelatedness of features transparent; in so doing, category labels modify perceived similarities. The interrelatedness among features has been known to affect various cognitive judgments, such as classification and feature inference (Goldstone, 1996). We think that category labels are special in similarity judgment because they clarify the extent to which features are inter-related. Figure 1: Sample stimuli (a. no-label, b. alphabet-label, c. cell-label conditions) Dog kidney cell Dog heart cell Dog kidney cell b. a.