A general valence asymmetry in similarity: Good is more alike than bad


The density hypothesis (Unkelbach, Fiedler, Bayer, Stegmüller, & Danner, 2008) claims a general higher similarity of positive information to other positive information compared with the similarity of negative information to other negative information. This similarity asymmetry might explain valence asymmetries on all levels of cognitive processing. The available empirical evidence for this general valence asymmetry in similarity suffers from a lack of direct tests, low representativeness, and possible confounding variables (e.g., differential valence intensity, frequency, familiarity, or concreteness of positive and negative stimuli). To address these problems, Study 1 first validated the spatial arrangement method (SpAM) as a similarity measure. Using SpAM, Studies 2–6 found the proposed valence asymmetry in large, representative samples of self- and other-generated words (Studies 2a/2b), for words of consensual and idiosyncratic valence (Study 3), for words from 1 and many independent information sources (Study 4), for real-life experiences (Study 5), and for large data sets of verbal (i.e., ∼14,000 words reported by Warriner, Kuperman, & Brysbaert, 2013) and visual information (i.e., ∼1,000 pictures reported in the IAPS; Lang, Bradley, & Cuthbert, 2005; Study 6). Together, these data support a general valence asymmetry in similarity, namely that good is more alike than bad.

Journal of Experimental Psychology: Learning, Memory, and Cognition, 42, 1171-1192