Superior single dimension relative to "exclusive or" categorization performance by a patient with category-specific visual agnosia: Empirical data and an alcove simulation

ELM, a patient with category-specific visual agnosia, was tested on a singl e-dimension categorization problem, and the "exclusive or" (XOR) categoriza tion problem. Stimuli were computer-generated shapes in which exemplars wit hin a shape Yet shared values: across two visual dimensions (curvature and thickness). In single-dimension categorization only curvature was relevant, and ELM performed as well as normal participants, In the XOR problem, cate gorization depended on being able to extract from memory values on curvatur e AND thickness for each exemplar, and ELM was significantly impaired on th is task. A computer simulation using ALCOVE(Kruschke, 1992) reproduced ELM' s behavior by changing a single (specificity) parameter related to how eas ily proximate objects within a multidimensional shape space could be disamb iguated. (C) 2000 Academic Press.