Processing of the perceived 3D-structure of objects in the human visual cortex

Successful interactions in our dynamic environment require that the visual system processes the shape, the 3D-structure and the motion of objects. Different cortical areas have been proposed to be involved in the processing of visual motion (hMT+/V5 = human middle temporal homologue), kinetic boundaries (V3B/KO = kinetic occipital), depth information (V3A) and object shape (LOC = lateral occipital complex). The goal of this study was to investigate whether these areas are selective for the 3D-structure of shapes defined by coherent motion and horizontal disparity information. To this end, we conducted two human fMRI experiments at a Siemens TRIO 3T MRI facility in which we used an event-related fMRI adaptation paradigm. In this paradigm, lower fMRI responses are observed for two identical than for two different stimuli that are presented consecutively in a trial. Adaptation across a change between two stimuli provides evidence for a common neural representation invariant to that change, while recovery from adaptation suggests neural representations selective for specific stimulus properties. We employed stimuli in which shapes were defined by the relative motion of random dots in the shape and the background. Additionally, the 3D-structure of these shapes was defined by the horizontal disparity of the random dots. In our first study, we tested whether shape and motion related areas are either selective for the 3D structure of shapes defined by horizontal disparity or for their outline. To this end, we tested four conditions: a) Identical, b) Different 3D-structure (Concave/Convex), c) Different shape outline, d) Different 3D-structure and shape outline. Recovery from adaptation was observed across changes in 3D structure in V3B/KO, hMT+/V5 and the LOC. Changes in the shape outline resulted in increased fMRI responses in hMT+/V5 and the LOC. These results suggest selectivity for 3D-shape in both shape (LOC) and motion-related areas (hMT+/V5). In a second study, we tested whether selectivity for the 3D structure of shapes can be simply accounted by differences in local disparities or by selectivity for global 3D-structure. To this end, we presented subjects with correlated and anti-correlated random dot stereograms, which contain horizontal disparity information but do not induce the perception of 3D-structure. Specifically, we tested for fMRI responses in four conditions: a) same disparity without 3D-structure b) different disparity without 3D-structure, c) same disparity and different 3D-structure b) different disparity and different 3D-structure. Selectivity for both local disparity information and global 3D-structure was observed in shape-related areas (V4v, LOC) as well as in motion-related areas (V3B/KO, hMT+/V5). In contrast, early visual areas (V1, V2, VP) showed selectivity for local disparity only. In summary, our findings provide conclusive evidence that not only shape (LOC) but also motion-related areas (hMT+/V5, V3B/KO) are involved in the selective representation and perception of the global 3D-structure of shapes