Impairments in part–whole representations of objects in two cases of integrative visual agnosia

Abstract

How complex multipart visual objects are represented perceptually remains a subject of ongoing investigation. One source of evidence that has been used to shed light on this issue comes from the study of individuals who fail to integrate disparate parts of visual objects. This study reports a series of experiments that examine the ability of two such patients with this form of agnosia (integrative agnosia; IA), S.M. and C.R., to discriminate and categorize exemplars of a rich set of novel objects, “Fribbles”, whose visual similarity (number of shared parts) and category membership (shared overall shape) can be manipulated. Both patients performed increasingly poorly as the number of parts required for differentiating one Fribble from another increased. Both patients were also impaired at determining when two Fribbles belonged in the same category, a process that relies on abstracting spatial relations between parts. C.R., the less impaired of the two, but not S.M., eventually learned to categorize the Fribbles but required substantially more training than normal perceivers. S.M.'s failure is not attributable to a problem in learning to use a label for identification nor is it obviously attributable to a visual memory deficit. Rather, the findings indicate that, although the patients may be able to represent a small number of parts independently, in order to represent multipart images, the parts need to be integrated or chunked into a coherent whole. It is this integrative process that is impaired in IA and appears to play a critical role in the normal object recognition of complex images.

Keywords:

• Integrative agnosia
• Object recognition
• Part–whole representations
• Visual agnosia
• Visual categorization

This work was supported by a grant from the National Institutes of Mental Health (MH54246) and to the Perceptual Expertise Network ( James S. McDonnell Foundation) to M.B. The authors thank S.M. and C.R. and their families for their enthusiasm and continued participation in this research. Thanks also to James Nelson and Matt Shugrue for their help in testing the patients and control participants, respectively. We thank Stacy Cho for help with manuscript preparation.

Notes

¹ Note that this scan was acquired six years after the present study but still indicates, at a minimum even after any possible recovery of these six years, the involvement of the right ventral visual cortex.

² This is a somewhat conservative criterion but is useful in clearly denoting the patients in relation to the distribution of the controls. It is also used often in interpreting neuropsychological data and thus has precedence.