The spatial arrangement method of measuring similarity can capture high-dimensional, semantic structures
- Russell Richie, University of Pennsylvania, Philadelphia, Pennsylvania, United States
- Bryan White, Psychology, New Mexico State University, Las Cruces, New Mexico, United States
- Sudeep Bhatia, Department of Psychology, University of Pennsylvania, Philadelphia, Pennsylvania, United States
- Michael Hout, Psychology, New Mexico State University, Las Cruces, New Mexico, United States
AbstractDespite its centrality to cognition, similarity is expensive to measure, spurring development of techniques like the Spatial Arrangement Method (SpAM), wherein participants place items on a 2-dimensional plane such that proximity reflects similarity. While SpAM hastens similarity measurement, its suitability for higher-dimensional stimuli is unknown. In Study 1, we collected SpAM data for eight different categories composed of 20-30 words each. Participant-aggregated SpAM distances correlated strongly (r=.71) with pairwise similarity judgments, although below SpAM and pairwise judgment split-half reliabilities (r’s>.9), and cross-validation with multidimensional scaling fits at increasing dimensionalities suggested that aggregated SpAM data favored higher dimensional solutions for 7 of the 8 categories. In study 2, we showed that SpAM can recover the Big Five factor space of personality traits, and that cross-validation favors a four- or five-dimension solution on this dataset. We conclude that SpAM is an accurate and reliable method of measuring similarity for high-dimensional items.