This　paper　presents　autonomous　operant　conditioning　automata　AOC.　It　involves　a　discrete　computer　model　described　self-automatic　machines,　including:　set　of　actions,　the　state　set;　＇condition　-　action＇　rule　set,　the　observed　state　transition,　as　well　as　operant　conditioning　learning　law.　And,　it　defines　the　behavior　entropy　based　on　the　orientation　value　of　the　state　of　AOC.　AOC　provides　a　recursive　run　the　program.　AOC　simulates　biological　operant　conditioning　mechanism,　which　has　a　bionic　self-organizing　feature,　including　self-learning　and　adaptive　capabilities,　can　be　used　for　description,　simulation,　design　of　a　variety　of　self-organizing　system.　Based　on　the　model　this　paper　designs　a　bionic　self-learning　control　method　and　uses　it　to　simulate　Skinner　animal　experiments　to　prove　that　AOC　can　simulate　the　learning　mechanism　of　operant　conditioning,　which　shows　that　AOC　can　be　used　to　design　a　variety　of　intelligent　robotic　systems＇　behavior.　©　2010　IEEE.