One view of ethics, highly useful from a descriptive point of view (99.99% of all people approach ethics this way), utterly useless from a prescriptive view is the following:
Ethics is a human faculty, much like the faculty of language, that has some broad outlines, but few specific details built in. It is a moderately tested scientific fact that the operation of the ethical faculty is almost exclusively sub-conscious, with order of operations being:
- determine conclusion.
- create reasons for conclusion.
My current understanding of ethics, from the study of psychopaths (ethics-free folks) is that ethics is basically a module (set of modules) in the mind/brain. Using the mechanism of fear primarily, humans learn the responses of guilt (to self) and anger (at others) when behavior doesn't match ethics.
My current understanding of the faculty of ethics suggests that the evolutionary "purpose" is to permit humans to function in groups more effectively...which is huge, given the fact that humans evolved primarily as large-group pack top- (or near-top-) predators.
Our current understanding of the faculty of ethics is that it has at least 6 relatively independent, relatively universal inborn dimensions (harm, justice, liberty, respect, sanctity, ingroupery) on which to vary (based on what is taught), and that various dimensions are impacted in importance by specific brain activity (amygdala, in particular).
One well thought through line of ethical reasoning, referred to as "ethical intuitionism" is to take these "intuitions" and construct logical conclusions from the data that are the ethical intuitions plus simple logic.
A great example of this approach to ethical reasoning is how Bryan Caplan reasons in this post.
It seems to me that he's neglected to observe that there is little normative force once you try to identify what moral intuitions are...but if you just take them as given, which they are for most people, then they can be used to construct a logically consistent ethic.
If, on the other hand, you think as I do, that much of the evolutionary "purpose" of ethics is to allow Prisoner's Dilemma predation on outgroups, while maintaining useful iterated PD interactions with clan-mates...you might believe that Caplan's approach, contrary the goal of the underlying system, is doomed to fail.