The two Q-functions are trained on the same experiences (i.e. samples), but vary in their parameters