Grasping semantic information with and without vision