That strategy works well for creating behaviors to deal with rotten meat and potential mates. After all, writing instructions to build an urge-circuit is relatively simple: if X, then Y. But what about creating complex higher behaviors, like playing the saxophone or Scrabble? As we've seen, higher skills are made of million-neuron chains working together with exquisite millisecond timing. The question of acquiring higher skills is really a question of design strategy. What's the best strategy for writing instructions to build a machine that can learn immensely complicated skills?
