The "DQN" (Deep Q Network) stuff from Google DeepMind has states. (And actions that transition from state to state.) This comes from Reinforcement Learning th

The "DQN" (Deep Q Network) stuff from Google DeepMind has states. (And actions that transition from state to state.) This comes from Reinforcement Learning theory. (The Q-Learner from Reinforcement Learning is the "Q" in Deep Q Networks.) [doi:10.1038/nature14236] Suri and Shultz argue that dopamine in the mammalian brain follows the "reward prediction error" from Reinforcement Learning [doi:10.1016/S0306-4522(98)00697-6] (Indeed the DQN paper mentions dopamine in the very first paragraph.) Because of this, I am very excited about DQN. (I do think that it's only a building block towards building a self-aware brain, though.)