The research program reported here is designed to develop a procedure for assessing sensitivity to dimensional components of non-representational art. The program was based on the assumption that non-representational art can be described in terms of empirically specifiable dimensions, and that an individual's sensitivity to or his appreciation of these dimensions is indicated by his esthetic reactions. Multidimensional scaling analyses were employed to yield a provisional characterization of the dimensions employed by subjects in grouping paintings contained in 2 sets of 40 abstract works. The "distances" among paintings, expressed in terms of their loadings on various dimensions, provide a standard against which to assess a subject's estimate of their similarity. Work is currently being undertaken to determine the relationship between the accuracy of these estimates and esthetic preference. (Author)