Separate neural pathways for the analysis of visual shapes and their spatial scale
Description
Abstract:
We can easily recognize the same abstract visual shapes seen at different spatial scales, or even when they are the whole, or the parts, of an object (as in the well-known Navon figures). Using a modified form of Navon Figures, we can separate in time the presentation of local and global letters. Multivariate pattern analysis (MVPA) of fMRI data shows that abstract information about the shape, and about the spatial scale, of these letters in the cerebral cortex is segregated. Spatial scale information overlaps regions previously described as place selective, whereas shape information overlaps regions selective for objects and faces. The mid-fusiform sulci split the ventral occipito-temporal cortices into a medial division that carries more information about scale, and a lateral division that carries more information about shape. Strong functional connectivity was found between adjacent patches of cortex in the caudo-rostral direction within each division, but not between adjacent patches belonging to different divisions. This suggests separate visual pathways for abstract shape and abstract scale. These findings support a model that explains several novel psychophysical and event related potential studies of feature-based attentional selection guided by scale. Finally, an experiment measuring BOLD repetition-suppression to changes of shape and/or scale found a mapping incongruent with the MVPA results, confirming that the two techniques measure different functional properties. Novel approaches are discussed that may help integrate the two techniques
Hosted by: Nancy Kanwisher