This　paper　presents　a　bionic　autonomous　learning　method　based　on　operant　conditioning　automata,　and　designs　a　cognitive　model　of　Skinner　operant　conditioning　automata.　The　model　adds　a　unit　of　behavior　propensity　to　generate　learning　mechanism　of　＇reaction-strengthening＇,　and　simulates　nature　biotic　randomly　compliance.　A　recursive　learning　algorithm　is　proposed,　which　is　to　simulate　biotic　operant　conditioning　mechanism　and　let　it　have　bionic　self-organization　function　including　self-learning　function　and　self-adaptive　function.　Finally,　the　model　is　proved　to　have　the　bionic　self-learning　ability　and　can　be　used　to　descript,　simulate,　and　design　various　self-organizing　systems.