To proceed from assumptions about an abstract theoretical set-up and from them to draw conclusions about the observable world and to test - by rough or more refined means - whether the conformity with observations is "good" enough, is indeed the time honoured procedure that all empirical sciences, including the natural sciences, have used. I shall therefore not plead guilty of heresy even if I do work with choice-theory concepts that are not invariant under a general monotonic transformation of the utility indicator.