Abstract
Many studies have reported that perceived shape is systematically distorted (e.g., Johnston, 1991; Tittle et al., 1995). However, we have found that distortions vary with tasks and observers (Lind et al., 2002). Furthermore, using action measures (reaching and grasping), we found that shape cannot be calibrated (Bingham, 2005; Lee et al., 2005). These results imply that shape perception is ambiguous rather than systematically distorted and that systematicity is produced by contextual variables. We investigated this hypothesis under full information conditions by manipulating relevant context.
Methods: Actual cylindrical objects with different depth to width aspect ratios were viewed (at 50 cm) in two conditions: 1) frontoparallel elliptical cross section (2D); 2) elliptical cross section in depth (3D). The task was to adjust the aspect ratio of an ellipse on a computer screen to match the cross section of a target object. Two different ranges of aspect ratio were tested: 1) Large range (LR): .5, .67, 1.0, 1.53, 1.9 and 2) Small range (SR); .67, .83, 1.0, 1.24, 1.53. Three different groups of 8 observers were tested: 1) 2D:SR -[[gt]] 3D:LR, 2) 2D:LR -[[gt]] 3D:SR, 3a) 2D:SR -[[gt]] 3D:SR and 3b) 2D:LR -[[gt]] 3D:LR.
Results: Observers performed the 2D task accurately. This provided the context. If 3D aspect ratios are ambiguous, then Os might resolve the ambiguity by assuming the same range of aspect ratios as experienced in the 2D task. Plotting judged aspect ratios against actual aspect ratios should yield the following results: Group 3: accurate slope of 1, Group 1: low slope [[lt]]1, and Group 2: high slope [[gt]]1 or slopes ordered as Group 1 [[lt]] Group 3 [[lt]] Group 2. The results showed the expected slope orders.
Conclusions: Results confirmed the hypothesis that perception of shape is ambiguous and the systematicity of distortions is a function of contextual variables.