Abstract
Studies using a short-term priming paradigm show that attended images prime themselves and their left-right reflections, whereas ignored images prime themselves but not their reflections (Stankiewicz et al., 1998). In two experiments with 3D rendered grey-scale images we tested priming for common objects rotated in depth. In Experiment 1 objects were rotated 90 degrees in depth across the line of sight. Attended objects primed themselves in the same view and when rotated, whereas ignored images only primed themselves in their corresponding view. The effects of view and attention were additive. Experiment 2 tested priming for orientations in which surfaces and parts change from study to test view. Objects were shown in two orientations that were rotated 60 degrees in depth within the line of sight. One view was a complete side-view of the object, thus occluding old and revealing new parts/surfaces compared to the second view. Priming results were similar to experiment 1, with substantial priming for all but the ignored rotated view condition. However, the effects of attention and view interacted, with a greater cost for attended than for ignored rotated objects. These data support models of human object recognition that rely on qualitative different representations (e.g. Hummel & Stankiewicz, 1996), namely part-based or structural representations for attended and holistic or view-like representations for ignored images.