Color-concept associations reveal a new conceptual space.

People are capable of using simple perceptual features such as color to signify complex meanings (perceptual semantics), an important aspect of nonverbal communication. This paper considers the nature and structure of the mental representations that support these abilities, using color semantics as a key example. Prior work has shown that people can judge the strength of association between essentially any color and any concept. Across four experiments we show that such associations express a low-rank conceptual space, color semantic space, whose dimensions can be well-characterized using color space regression. The color semantic space is distinct from classical connotative (affective) and denotative (literal) semantic spaces and from perceptual color space. Further, color semantic relations help to predict human judgments of word-meaning similarity across a diverse set of concepts-even when such judgments are not explicitly about color. The methodology we develop is general, offering a means of evaluating the mental structures underlying other aspects of nonverbal communication. Together, the work suggests that color-concept associations are encoded within a previously unrecognized conceptual representation space, and opens the possibility that human thought may rely on multiple different kinds of conceptual spaces-with implications both for cognitive science generally and for understanding the use of color in domains such as visual communication.