Abstract
Object-based transformations are spatial transformations in which the reference frame of an object is mentally updated relative to the egocentric and environmental reference frames. Perspective transformations are those in which the viewer's egocentric reference frame is mentally updated relative to object and environmental reference frames. The two have been differentiated based on behavioral, neuroimaging, and neuropsychological data (Zacks & Michelon, 2005). For scene stimuli, picture plane rotation appears to invoke object-based rotation to upright, even when the task appears to require taking a novel egocentric perspective within the scene (Shelton & Zacks, 2007). In a series of experiments, we explored how different references frames might be used to define “upright” for a scene stimulus. We created different combinations of congruent and incongruent reference frames by varying the orientation of the local (computer) and egocentric (participant) reference frame relative to the gravitational reference frame. We interrogated the data to see which orientation was fastest for making left/right judgments about objects in the scene (e.g., “If you came in the door, would the lamp be to your left or right?”). We used the relationship between the fastest orientation and the orientations corresponding to the three reference frames to infer the subjective upright. In Experiment 1, the congruence of any two reference frames was sufficient to define upright. In Experiments 2 and 3, the three reference frames were always in competition, and the egocentric orientation was consistently fastest. Together, the results suggest that rotated scene stimuli invoke some degree of object-based rotation to upright even when taking a perspective in the scene, and upright can be defined in multiple reference frames. This upright can be defined opportunistically when multiple frames coincide; in the absence of coincident frames, the egocentric reference frame appears to dominate.