The nature of representations of visual objects remains a topic of debate. Some theories posit that such representations are relatively abstract, in that a common representation may be activated by multiple exemplars or by multiple views of the same object exemplar (see, e.g., Biederman, 1987;Biederman & Bar, 1999;E. E. Cooper, Biederman, & Hummel, 1992;Hayworth & Biederman, 2006;Hummel & Stankiewicz, 1996;Wagemans, Van Gool, & Lamote, 1996). Other theories p posit that object representations are relatively specific, in that different representations are activated by different exemplars or by different views of the same object exemp plar (see, e.g., BĂĽlthoff & Edelman, 1992;Gauthier et al., 2002;Poggio & Edelman, 1990;Tarr, 1995;Tarr, Williams, Hayward, & Gauthier, 1998;Ullman, 1996). Alternatively, both abstract and specific representations may exist along different points on a continuum in a single, unified processing system (see, e.g., Farah, 1992;Hayward & Williams, 2000;Tarr & BĂĽlthoff, 1995). Finally, abstract and specific object representations may exist in separate and dissociable neural subsystems, with an abstract subsystem operating effectively in the left hemisphere (LH) and a specific subsystem operating effectively in the right hemisphere (RH) (see, e.g., Burgund & Marsolek, 2000;Marsolek, 1995Marsolek, , 1999 Marsolek & g , , ) Burgund, 1997, 2003.The bulk of the evidence for the dissociable neural subr systems theory comes from experiments in which familiar shapes were used. This may be cause for concern, as Curby, Hayward, and Gauthier (2004) recently suggested, since abstract effects in the LH could reflect semantic processing of postvisual information that is associated with the visualshape information, rather than processing of (abstract) visual shapes only, as posited by the dissociable neural subsystems theory. A related concern is that abstract effects in the LH could reflect linguistic processing of names associated with the visual shapes, rather than processing of abstract visual-shape information. In addition, when objects are presented multiple times in an experiment, abstract ef-f f fects in the LH could reflect episodic memory for previous trials in the experiment. In this article, we report results that indicate, in line with the dissociable neural subsystems theory, that abstract processing of visual shapes in the LH does not require semantic, linguistic, or episodic information. Abstract processing of visual shapes can occur effecd tively in the LH when the stimuli are novel objects viewed for the first and only time.According to the dissociable neural subsystems theory, an abstract category subsystem and a specific exemplar subsystem operate relatively independently and in parallel. An abstract category subsystem encodes the visual category to p g(g , p p ) which an input stimulus belongs (e.g., pen vs. cup). It uses
Rice University, Houston, TexasAn ongoing debate concerns whether visual object representations are relatively abstract, relatively specific, b both abstract an...