Effects of adding a second reinforcement alternative: implications for Herrnstein's interpretation of r(e).