Correlational structure as a determinant of goodness of complex patterns.

The stimulus sets used by Whitman and Garner (1962) to study the effect of form of redundancy on free-recall learning were arranged in 3 X 3 configurations to form single patterns. Various spatial arrangements of the 9 figures in each of the 3 sets tested several questions regarding judged goodness of form. Rating of goodness of form on a 7-point scale were made by 32 college students. Patterns composed of elements which have correlated structure in the form of simple contingencies, and which Whitman and Garner found were easily learned in free recall, are good; patterns composed of elements which have correlational structure in the form of complex interactions, and which Whitman and Garner found were extremely difficult to learn, are not. Uncertainty calculated for spatial position using 2-dimensional Cartesian coordinates as variables does not appear to be associated with pattern goodness. However, spatial arrangements of the elements exert a very strong effect on pattern goodness.