On Continuous Time Learning Models