Reward Maximization for Higher Cognition

dep methods are particularly helpful for comparing different architectures, theorizing about them. And i think that that's a particularly helpful part of deparl. But other than the higher carnetof function, i think that the reward maximization framework might have limitations. So we might need a future kind of evolution of deep arrel if we went to address these sorts of more complex, long term, short term, multi scale a purchas.