Measurement of visual cluster.

The cluster continuum is described, along with how to measure degree of cluster and how to generate stimuli that vary along the continuum. Five experiments were performed on cluster discrimination and rating, numerosity estimation, pattern discriminability, and random generation. Difference thresholds were obtained for discrimination between random and clustered patterns, and ratings exhibited a power function with respect to cluster values of the stimuli. Numerosity estimates decreased with increasing degree of cluster, and random patterns were easier to discriminate than clustered ones. When subjects were asked to arrange items so that they appeared random, they erred on the side of regularity. The cluster continuum was discussed in relation to the distribution of living organisms and to the ecological approach of Gibson. The cluster metric was compared with the measures used in the study of response organization in free recall.