Structural processing and category-specific deficits

Abstract

We evaluated the contribution of four structural dimensions (object parts, internal details, objects contours and variability of the representation), as a possible source of categorical processing differences and category-specific deficits. Importantly, these dimensions aggregate 22 different structural measures that have been proposed to describe the Snodgrass and Vanderwart (1980) picture set. Study 1 analysed the differences between the four dimensions across domains and categories. Study 2 investigated how these dimensions may contribute to the performance of two patients with category-specific deficits that have been reported previously in the literature (Farah et al., 1991). The results showed that living things were structurally more complex than non-living things, scoring higher in object parts and object contours. Regarding the variability of the representation, living things did not show much within-item diversity but did show more contour overlap and less visual similarity, the latter two qualities of living things being detrimental to object processing in a naming task. Parts, contours and variability of the representation also differentiated animals, fruits and vegetables and, to a certain degree, non-living things: animals had more parts, fruits had more object contours and non-living things had a lower variability of the representation (which was especially related to higher within-item diversity and lower contour overlap). The same three dimensions predicted patient performance. However, when structural dimensions were considered together with domain (living/non-living) and concept familiarity, only variability of the representation contributed significantly to patient performance.