Dramatic advances in generative models have resulted in near photographic
quality for artificially rendered faces, animals and other objects in the
natural world. In spite of such advances, a higher level understanding of
vision and imagery does not arise from exhaustively modeling an