Gravar-mail: Probability learning as a function of momentary reinforcement probability