Abstract

Theories of visual object representation often assume that object shapes are represented with respect to a set of internal axes. However, few studies have examined how object axes are defined. Research has suggested that the axes of a simple elongated object (e.g., pen) are defined by the object's elongation (e.g., Sekuler & Swimmer, 2000). But what about the axes of objects that have a side protrusion (e.g., hatchet), or those that possess two equally-elongated parts (e.g., boomerang)? We probed the axes underlying representations of real-world objects in three shape categories: I-shaped, L-shaped, and V-shaped (Fig. 1). We adopted a novel task paradigm that involved analyzing participants' errors in an orientation recall task. Participants viewed an object at a random orientation, and later attempted to reproduce the object's orientation. Previous studies have shown that errors in this task often take the form of reflections across an object axis (e.g., Gregory & McCloskey, 2010). By identifying the reflection axes in participants' reflection errors (Fig.2), we inferred the axes for different shapes of objects. For all three types of stimuli, we consistently observed a pair of perpendicular reflection axes related to salient geometric properties of the objects (Fig.3). For 'I' objects, the axes were defined according to shape elongation, consistent with previous findings. For 'L' objects, the axes were defined by the objects' elongated part (e.g., hatchet's handle), independent of the protrusion (hatchet's blade). For 'V' objects, the axes were defined according to the line that partitions the objects into symmetrical halves. We discuss the implications of these findings for proposals concerning object axis representation. We suggest that our results are not adequately explained in terms of medial axes, or axes defined by the global object outline. Rather, we argue that axes are defined to allow perspicuous representation of the internal arrangement of object parts. Meeting abstract presented at VSS 2018

Full Text
Paper version not known

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call