Over the past century, great debate has ensued regarding the fundamental properties of emotions. The idea that 2 properties-valence and arousal-are critical for emotion and psychologically irreducible has had substantial staying power in the literature. In the present report, we examine whether a third dimension-reflecting the social properties of emotion-might arise if stimuli high in that dimension (i.e., "theoretically social emotions") were included in the task and, or, if social information was primed. We used a similarity-rating task to evaluate whether a dimension representing the "socialness" of emotion-the extent to which emotions are associated with social contexts-might arise as the result of inclusion of emotion words that are theorized to represent "social emotions." In Study 1, we assessed the dimensional structure of 41 different emotion terms (of which 41% were "social emotions") based on pairwise similarity ratings of a subset of the emotion terms. In Study 2, we tested whether priming social information before and during the similarity rating task would shift the emergent dimensional structure of emotion words. Results of multidimensional scaling across both studies indicated that the structure of emotion is best described by 2 dimensions-valence and arousal-and was not influenced by the priming of social information. Contrary to predictions, evidence did not emerge for a third dimension corresponding to socialness, nor any other property. (PsycInfo Database Record (c) 2020 APA, all rights reserved).