We examined the performance of human observers to identify three-dimen
sional (3D) shape from motion induced by exploratory head movements. S
ubjects categorized 3D quadratic surfaces with randomly chosen shape b
ut with a fixed amount of curvature in one of eight shape categories.
Human observers had their best performance with convex parabolic and c
oncave parabolic shapes, but had more difficulties in identifying the
hyperbolic shapes. The identification of 3D shape was not significantl
y influenced by the amount of curvature. This means that our descripti
on of 3D shape closely matches the intuitive notion of shape, and that
its use in 3D shape perception tasks is justified.