What Variables Produce Similarity Grouping?

Abstract

The effectiveness of several configurational or textural variables in producing similarity grouping was investigated by measuring the time it took Ss to locate the disparate quadrant in a circular array. The 'figure' and 'ground' quadrants differed in (1) slope of elements, (2) linearity of elements, or (3) the way in which the same slopes were combined into angles. Differences in slope produced excellent grouping; differences in linearity were less effective; and differences in arrangement of slopes were generally quite ineffective. Grouping was also dependent on orientation of the whole array: horizontals and verticals gave better grouping than diagonals, and even arrangement of slopes gave moderately effective grouping when angles with horizontal and vertical axes of symmetry were opposed to each other. Conditions of head tilt were used to determine whether the grouping effects were dependent on retinal or gravitational orientation. For the angles, only retinal orientation seemed important; but for line segments and paired dots varying in slope, both retinal and gravitational orientation had significant effects. The results disconfirm the hypothesis that grouping occurs entirely on a projection level, independently of constancy mechanisms. It is suggested that grouping and segregation depend on identities and differences between descriptors which in turn represent relationships between the stimulus array and an internal Cartesian reference system. Similarity is one of the most important of Wertheimer's principles of perceptual grouping, but only recently, with the work of Julesz and Beck, has any appreciable curiosity been displayed about the kinds of similarity that are effective in this respect.' It should, of course, be understood that grouping always implies the complementary process of segregation. Grouped items must be at once similar to one another and different from extraneous items; in analysis-of-variance terminology, within-group variance must be Received for publication September 22, 1969. The research was supported in part by the Air Force Office of Scientific Research (Grant 973-66) and in part by the Advanced Research Projects Agency of the Department of Defense (Contract F44620-67C-0099). 1 M. Wertheimer, Untersuchungen zur Lehre von der Gestalt, Psychol. Forsch., 4, 1923, 301-350; B. Julesz, Texture and visual perception, Sci. Amer., 212, 1965, 38-48; J. Beck, Perceptual grouping produced by line figures, Percept. & Psychophys., 2, 1967, 491-495.