Edit to add another (better?) phrasing
I'm looking for information that addresses this question:
All prescriptive ethical theories need to be descriptive to at least some degree(1). However, a completely descriptive ethical theory is just that -- descriptive. How do you judge (or what is a framework for judging) whether a set of cases where the ethical theory prescribes different moral judgement/behaviour than is typically observed constitutes a failure of the ethical theory?
Who has explicitly addressed the problem as to when to decide to allow for a specific ethical theory to deviate from common-sense moral intuitions?
(1) By descriptive I mean describing/reproducing what some people assess as the correct moral decision. If a purported ethical theory did not have any particular relationship to the common sense use of the word ethics, we'd hardly call it an ethical theory.
Original Question
Who has explicitly addressed the problem as to when to decide to allow for a specific ethical theory to deviate from common-sense moral intuitions?
Suppose I set up a suitably elaborated theory of ethics, and I find that gratuitous murder is perfectly acceptable. This seems like a good reason to reject that theory of ethics, i.e. it seems that matching at least some basic moral intutions is a requirement of any ethical theory.
Suppose I set up a suitably elaborated theory of ethics, but it fails to correspond with typical people's responses to the Trolley Problem(s). Should that be interpreted as a failure of the theory of ethics, or as a case where peoples' typical intuitions are morally suspect?
Requiring that theories of ethics merely be able to reproduce typical moral intuitions seems unduly limiting -- it rules out the existence of prescriptive moral theories. [Note that arguments that all moral theories cannot/should-not be prescriptive would address this question.] In addition, this seems problematic in the moral test cases where there is no clear common intuition (I don't know of any such cases, but I can't rule out that they exist).
I see similar delineation problems in terms of aesthetics and epistemology, and it might arise in other fields as well. Thus someone may have tackled this as a kind of general problem, if so that is of interest too.