Abstract
According to several theoretical perspectives, the spatial structure of an object is represented with respect to the object's most important axis—the object principal axis (e.g., Marr & Nishihara, 1978). For objects with an elongated or a symmetrical shape (e.g., a pen, a wine glass), available evidence suggests that the principal axis may correspond to the shape's axis of elongation or symmetry (e.g., Quinlan & Humphreys, 1993; Sekuler & Swimmer, 2000). For objects with more complex shapes, less is understood about how the principal axis is defined. We examined a class of real-world objects consisting of an elongated base and a protruding part—that is, objects with shapes resembling the capital letter 'L' (e.g., a hatchet with a large protruding blade; see Fig.2 of the supplementary materials). We asked whether the principal axis was better aligned with the object's main elongation axis (e.g., the hatchet's handle), or with an axis that spanned the global contour of the object (here referred to as the object 'contour axis'). The paradigm we used involved analyzing the pattern of reflection errors participants made in an orientation recall task. Participants were asked to report the orientation of an object they previously saw. In trials where a reflection error was made, we identified the object's axis of reflection (Fig.1). This experimental paradigm was built on the previous finding (e.g., Gregory & McCloskey, 2010) that when participants made an orientation reflection error, they tended to reflect the object across its internal principal axis. Our results suggested that the principal axes of the 'L'-shaped objects were better aligned with the object's axis of elongation (Fig.4) than with their contour axes. Overall, this finding sheds light on how individual parts within an object may interact to shape the representation of the object's global spatial structure.
Meeting abstract presented at VSS 2017