In a provocative opinion piece, Ayzenberg and Behrmann (A+B) [1] present evidence, primarily from neuroimaging and deep neural networks (DNNs), for the idea that the global shape of objects is computed in the dorsal visual stream and is then transferred to the ventral stream where it contributes to object identification. In our commentary on their article, we challenged this proposal, arguing that, if global shape were computed in the dorsal stream, then one would expect patients with large bilateral lesions of the dorsal stream to be unable to discriminate between objects on the basis of their global shape [2].