I believe that a conditioned behavior is a kind of "concept" (or concept-like data object). Aaron implied that a person (or an AGI entity) can combine two conditioned behaviors if one could be litely morphed to meet the situational criteria of the other. I said yeah, sure but now you are saying that the "concepts" one possesses can be used to modify one's own behavior. Once you have taken that road RL is no longer the primary method of learning. Self-directed psychological motivation and complex learning have to be included in the learning toolbox. Of course we have behaviors modified by external rewards but we are now way beyond the limited scope of what is typically meant by Reinforcement Learning. Jim Bromer
------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657 Powered by Listbox: http://www.listbox.com
