# Joey, A Perhaps Conscious Robot This is an implementation of a very simplistic reinforcement[1] based learning system. It learns how to get rewards in a square based world. See my blog about it here: http://leverlabs.io/blog20161014/ The main feature is a set of plans with scores. And a memory of past inputs and past actions. If a plan leads to a reward, it adds to its score. If a plan leads to an unexpected state, it leads to surprise and the plan is demoted, perhaps removed. On any reward, past history is made into the shortest plan that does not exist yet. Minus some plans that are too random to be useful. It is a sketch of what perhaps in a bigger setting might be a key ingredient for consciousness. [1]: https://en.wikipedia.org/wiki/Reinforcement_learning