When an observer judges the orientation in depth of a trapezoidal surface, the pictorial information of the surface is often more influential than motion information. Motion information might be more effective if pictorial information is simplified: this prompts the present study. Surfaces were triangular and pictorial information resided only in the visual lengths of the surfaces. In experiment 1, monocular observers viewed during head motions of 0 to 30 cm extent. Static judgments were somewhat dependent on visual length and tended to be frontal. Contrary to predictions, moving judgments were similarly affected: only 30 cm motion elicited near-veridical perception, as in previous studies with trapezoidal surfaces, although visual length had a residual effect. Experiment 2 involved investigation of whether visual length requires prior exposure to triangular surfaces to be effective; this was found not to be the case, which argues that observers rely on internal models of triangular surfaces. Depth perception appears to balance rapidity of processing against accuracy, in a way suggesting that `direct' approaches are incomplete. Finally, it is argued that depth-from-motion simulations-influential in assertions that motion information is fully effective-depend on pictorial information.
|Publication status||Published - 1996|