We can train computers to learn to recognize objects by giving them millions of examples with the correct answers. A human baby, on the other hand, learns to recognize many concepts and objects all by themself simply by interacting with a few examples in the real world.