{"pk":27755,"title":"A Casual Model Approach to Dynamic Control","subtitle":null,"abstract":"Acting effectively in the world requires learning and control-\nling dynamic systems, that is, systems involving feedback re-\nlations among continuous variables that vary in real time. We\nintroduce a novel class of dynamic control environments us-\ning Ornstein-Uhlenbeck processes connected in causal Markov\ngraphs that allow us to systematically test people’s ability to\nlearn and control various dynamic systems. We find that per-\nformance varied across a range of test environments, roughly\nmatching with complexity defined by a set of models trained\non the task (an optimal model, a deep Reinforcement Learning\nagent, and a PID controller). The testbed of dynamic envi-\nronments and class of models introduced in this paper lay the\ngroundwork for the systematic study of people’s ability to con-\ntrol complex dynamic systems.","language":"eng","license":{"name":"","short_name":"","text":null,"url":""},"keywords":[{"word":"Dynamic control"},{"word":"Casual learning"},{"word":"Dynamic decision making"},{"word":"Reinforcement Learning"},{"word":"control theory"}],"section":"Publication-based-Talks","is_remote":true,"remote_url":"https://escholarship.org/uc/item/9v04k296","frozenauthors":[{"first_name":"Zachary","middle_name":"J","last_name":"Davis","name_suffix":"","institution":"New York University","department":""},{"first_name":"Neil","middle_name":"R","last_name":"Bramley","name_suffix":"","institution":"New York University","department":""},{"first_name":"Bob","middle_name":"","last_name":"Rehder","name_suffix":"","institution":"New York University","department":""},{"first_name":"Todd","middle_name":"","last_name":"Gureckis","name_suffix":"","institution":"New York University","department":""}],"date_submitted":null,"date_accepted":null,"date_published":"2018-01-01T18:00:00Z","render_galley":null,"galleys":[{"label":"PDF","type":"pdf","path":"https://journalpub.escholarship.org/cognitivesciencesociety/article/27755/galley/17395/download/"}]}