Nonlinear learning and learning advantages in evolutionary games