This article presents psychophysical evidence that the Kanizsa Square does not produce an 'object superiority effect' previously reported in similar Gestalt configurations. Implications of the findings for Gestalt theory are addressed.

A series of experiments was performed in which subjects indicated whether two four-dot patterns were the same, although possibly viewed from different directions, or different, paired at random. Analyses of responses times and error rates suggest that the subjects' performance in this affine matching task is based on non-accidental properties such as convexity, parallelism, collinearity, and proximity, rather than on real affine invariants such as the ratio of triangular areas.