Abstract
Animate and inanimate objects evoke large-scale differential responses across occipito-temporal cortex; the same has been shown for objects of big and small real-world sizes. How are these two orthogonal object dimensions represented across the cortex? Images of big animals, big objects, small animals, and small objects were presented in a blocked design to 12 participants undergoing functional neuroimaging. We used a "coarse MVPA" approach to examine the large-scale spatial distribution of responses across occipito-temporal cortex. 20 spherical regions-of-interest were arrayed across parahippocampal, fusiform, lateral occipital, and medial occipital cortex. Along this continuous band of cortex, the animacy organization had 7 alternating peaks (4 for animals>objects; 3 for objects>animals), while the size organization had 3 alternating peaks (2 for big>small entities and 1 for small>big entities). The animacy and size peaks were arranged as follows: (i) medial big animal regions, (ii) adjacent big object regions (ii) adjacent lateral animal regions with no size modulation, and (iii) a small object region at the center. At the two medial extremes, we found novel animate regions, located even more medially than scene-selective regions; these were only elicited by big animals, demonstrating an interaction between the size and animacy organization. Pattern analyses revealed that the animal/object organization was substantially stronger than the big/small organization (shared animacy: r=.56; shared size: r=-.68; t(11)=10.5, p<0.001), and that objects were more differentiated by size than animals (t(11)=4.7, p<0.001). Overall these results demonstrate that (i) the medial-to-lateral responses of objects vs animals along the ventral surface are part of an even larger-scale alternating organization, and (ii) big vs small differences apply more strongly within the inanimate domain. The topography of these responses are reminiscent of horizontal and vertical meridian alternations in early visual cortex, and thus may potentially be used to define different areas of high-level object cortex.
Meeting abstract presented at VSS 2012