Vision is a learned skill. During development, sighted infants learn to understand the world around them from its visual appearance. They learn, without much feedback from outside, to extract the basic physical properties of the scene, segregating the visual image into distinct specific objects. They become sensitive to increasingly complex aspects of objects, agents, goals, and interactions between agents from the first months of age. For example, they learn to interpret hands as "movers" of objects, and they gradually learn to use details of grasp and hand-object interactions to interpret fine actions and their goals. This achievement should not be taken lightly: current unsupervised models in computer vision are still a far cry from being able to account for such human performance.  We still lack a consistent theory to describe visual function development from a computational viewpoint, and the possible bottlenecks which may limit perceptual performance. Understanding the processes underlying visual function acquisition is obviously important for the development of useful strategies for recovery from vision loss.  The vision attainable may depend on the duration of vision loss, the age at loss, the affected organ (eye vs. brain), transfer of information from other senses, and the broader cognitive and integrative abilities of affected individuals.