Perceptual frames of reference and two-dimensional shape recognition: further examination of internal axes

Three experiments on the perception of simple four-sided two-dimensional shapes are reported. In the first experiment subjects were given a paper-and-pencil test in which they had to consider each of a set of shapes in turn. They were instructed to draw in what they considered to be the most salient axis of each of the shapes ie a line that they felt most naturally went with the shape. The results showed a significant tendency to draw an axis of symmetry if one was present. However, when presented with instances of a shape that was elongated but possessed no symmetries they failed to consistently draw any particular axis. A further, speeded classification, task revealed that for this shape explicit axis information appeared not to influence performance. In contrast, such information clearly affected performance with a shape that did possess a salient axis of symmetry and elongation. Indeed this axis was shown to be integral with the contour of the shape when a final classification experiment was carried out. The axis of elongation of the other shape acted as a characteristic which was separable from the contour. The results are discussed in relation to accounts of shape perception which assume that a necessary stage in the perception of shape is the derivation of an axis-based structural description.