Empirical evidence for independent dimensions in the visual representation of three-dimensional shape.

Three experiments investigated whether the human visual system can make independent estimations of 3-dimensional shape dimensions, as suggested by structural description models of human object recognition. Experiment 1 used a noise masking paradigm to investigate whether primary-axis curvature and aspect ratio are estimated independently. The results showed clear evidence that these 3-dimensional shape properties are estimated independently from the image. Experiment 2 investigated whether the visual system can treat any arbitrary set of independent shape dimensions independently. The dimensions used in Experiment 2 were linear combinations of aspect ratio and primary-axis curvature used in Experiment 1. Experiment 2 results showed that these 2 dimensions were not estimated independently. Using the same masking paradigm, Experiment 3 investigated whether an object's 3-dimensional shape is estimated independently of the object's viewpoint. This experiment found that shape properties are estimated independently of viewpoint.