Extensions of Expected Utility Theory and Some Limitations of Pairwise Comparisons