Perception of wheel-generated motions.

Data are presented on an old and familiar Gestalt demonstration--perceiving wheel-generated motions--in which the perceived motions of a rolling wheel are shown not to be obviously derived from the motions of the parts. The history of study of this phenomenon is presented, and contradictions in the literature are noted. The focus for experimentation is on the contrasting approaches found in Johansson's perceptual vector analysis and Wallach's arguments for the priority of object-relative displacement in the extraction of invariants. Johansson's approach asserts that common vectors are extracted from moving events first, whereas Wallach asserts that the motion of objects relative to each other is first. These two approaches yield different predictions about what ought to be seen when different configurations are viewed in rotation. In five experiments viewers rated how wheellike the movement of various point-light systems attached to a rolling wheel appeared to be. Results support Wallach's views over Johansson's. Viewer judgments of goodness in wheellike motion correspond highly with a mathematical description of the parameters of cycloidal motion for the geometric center of any system of lights on a rolling wheel. This specification can be made only after the extraction of object-relative displacement information. Number of lights and order of symmetry influence viewer judgments to a much lesser degree, and placement of a light at the wheel's center matters not at all.