Proceedings of the Annual Meeting of the Cognitive Science Society

Despite its centrality to cognition, similarity is expensive tomeasure, spurring development of techniques like the SpatialArrangement Method (SpAM), wherein participants placeitems on a 2-dimensional plane such that proximity reflectssimilarity. While SpAM hastens similarity measurement, itssuitability for higher-dimensional stimuli is unknown. InStudy 1, we collected SpAM data for eight differentcategories composed of 20-30 words each. Participant-aggregated SpAM distances correlated strongly (r=.71) withpairwise similarity judgments, although below SpAM andpairwise judgment split-half reliabilities (r’s>.9), and cross-validation with multidimensional scaling fits at increasingdimensionalities suggested that aggregated SpAM datafavored higher dimensional solutions for 7 of the 8 categories.In study 2, we showed that SpAM can recover the Big Fivefactor space of personality traits, and that cross-validationfavors a four- or five-dimension solution on this dataset. Weconclude that SpAM is an accurate and reliable method ofmeasuring similarity for high-dimensional items.