Purchase this article with an account.
Marius Cătălin Iordan, Michelle R. Greene, Diane M. Beck, Li Fei-Fei; Typicality Sharpens Object Representations in Object-Selective Cortex. Journal of Vision 2013;13(9):127. doi: 10.1167/13.9.127.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The purpose of categorization is to identify generalizable classes of objects whose members can be treated equivalently. Within a category, however, some exemplars are more representative of that concept than other members of the same category (Rosch 1973, Rosch & Mervis 1975). This typicality effect usually manifests as increased speed of recognition, as well as lower error rates for verifying category membership of the more typical item. Despite these behavioral effects, little is known about how typicality influences the neural representation of objects from the same category. To address this question, we performed an fMRI experiment in which participants were shown color photographs from 128 subordinate-level object categories grouped into 16 basic-level categories (4 species of animals, 4 types of plants, 4 transportation modalities, and 4 classes of musical instruments). Typicality for each subordinate within its basic category was assessed behaviorally. We analyzed neural responses in early visual areas and object-, scene-, and face-selective areas: V1, V2, VP, hV4, LOC, TOS, PPA, RSC, FFA. For each brain area, we computed separate similarity matrices (Kriegeskorte et al. 2008) for the most and least prototypical halves of the category set. In V1, V2, and LOC, the most typical exemplars from a basic category had a more similar representation and were more distinct from prototypes of other basic categories (using a category boundary effect measure) than the least typical exemplars. Furthermore, a subsequent analysis showed that in LOC, but not in early visual areas, the most typical exemplars also correlated better than the least typical with the average response elicited by other exemplars. Our results suggest that neural representation differs for typical and less typical object exemplars. More specifically, typicality may be correlated to neural distance between categories in LOC, with highly typical members maximizing dissimilarity to instances of other categories.
Meeting abstract presented at VSS 2013
This PDF is available to Subscribers Only