If I catch Reppert's sentiment correctly, he appears to accept - or at least allow - what atheists call `the autonomy of ethics'.1 (David O. Brink has an essay of that name in The Cambridge Companion to Atheism, ed. Michael Martin.) What he is asking is a little broader: what (uncomfortable) metaphysical commitments must naturalists/secularists make in formulating a satisfactory account?
In a trivial sense, some sort of metaphysical statement must be made whenever giving a descriptive account of ethics. But I do not think that secular thinkers find or need find such statements to be problematic. Ethical naturalism, for example, may be defined in the broadest sense as figuring out what we want in fact, and then how to get it in fact. To me, questions of the former type are, strictly speaking, subjective, even if many opinions are widely held. The latter type, the `goal-oriented' question, is often capable of uncontroversial, empirical answers.
1. I want people to live satisfying, intellectually fulfilled lives.
2. Poverty and constraints on liberty prevent the actualization of (1).
3. I should therefore work against poverty and tyranny given (1).
There are of course objections to this sort of ethical naturalism; I think that it is woefully incomplete. But it is a good start, and I see no difficult metaphysical entanglements in proposing such reasoning.
I have a more or less Humean stance on the nature of morality; I think it is rooted in sentiment. But is it not true that abstract notions - principles - appear to affect our moral thinking? I think so, but I do not think that this presents any fatal difficulty. This is because we are capable of sentimental attachment to principles. This is why I, while I would be considered a non-realist in many important respects, am nevertheless a moral universalist, in the sense that I think we can discover approximating categorical statements which usually capture our desires and interests and provide something I think we all should want: a clear basis for law, moral argumentation, and the communicating of judgments. So while "it is not contrary to reason to prefer the destruction of the world to the scratching of my finger", I am still a big proponent of the Universal Declaration of Human Rights. (Were you to ask me why I think it is that we can find approximating principles, I would start to babble about evolution and culture and other things.)
But this can only ever be an approximation: there are psychopaths, after all. So why do we get to imprison them? Because we can, based on stuff we want. (For the Aristotelians, I include widely-accepted `virtues' in that category.) I understand that this may feel deeply unsatisfactory, but I do not think that `wrong actions' are in the same category as `wrong statements'. Is it unreasonable for us to imprison the psychopath? In the sense that the root of our action is not in reason, yes, but it's equally unreasonable to not imprison him. The decision is ultimately sentimental and may even be crudely emotive at the fundamental level.
I am not and cannot be a Kantian. I think that the much-touted categorical imperative is vacuous and that trying to deny the circumstantial nature of moral judgments is flatly absurd. I am not a utilitarian: though a rough idea of `general happiness' is part of our moral thinking, I do not think that all the various virtues and desires are subsumed by a one-dimensional measure. I think most people would have a problem with organ lotteries.
I think that people want a different account because they want to still their uneasy consciences about tossing others in prison. They want to be as certain of their moral judgments and prejudices as they are of facts like gravity. They want to be correct, not merely right, about the objects of their strongest passions. In practice, there is little difference, except that those who feel self-righteous tend to be crueler.
But passion is passion, which is not a truth-finding mechanism. I prefer to take a more authentically scientific attitude to morality, which is piecemeal and allows for goal-indifference. Like the `piecemeal social engineering' advocated by Popper in The Open Society and Its Enemies, I want to ask questions like the following: "how do we arrange our institutions so as to prevent murder, supposing we should want to prevent it or not?" Properly considered, the moral demands of the skeptical ethicist are more stringent than those of the metaphysical ethicist. The former is forced to discover more and to think of morality feature of her surroundings as created by herself, and in application she will probably need to take far courageous stands and make great sacrifices to fulfill her desires. As Russell commented in his Sceptical Essays, utilitarians, in contrast to traditional moralists, must live a far more onerous life. I think that this counter-intuitive fact holds for active moral thinkers generally, even that of we skeptical types.
So that, roughly, is what I take to be a roughly correct, if overly broad-brushed, picture of morality, which is a mixture of moral skepticism, Humean considerations, and pragmatism. Nothing about this says anything about `grounding' any particular value or norm. This project is unnecessary and possibly even misguided.
What I'm not saying: I'm not a moral nihilist. We make moral distinctions, but they are sentimental (See Hume). I'm not a psychological egoist (see Hume again). I think that normal people are altruistic in some degree, although `self-interest' is a big part of our makeup. Norms and values are not `arbitrary'. Norms are like satisfactory definitions: they capture widespread intuitions and ideas; values are not switches which we flick on and off in our heads.
For pragmatic reasons, I speak in ordinary moral language. I quite regularly say things like "...so we should...", and I often refer to abstract concepts like `human flourishing' or even `purpose'. But when saying such things, it is important to remember their nature. I may say that humans should read philosophy in order to flourish intellectually - where intellectual flourishing is desirable - but I do not think that this statement is `true' in any normal sense of the term. It is not a fact about the universe or a fact about the mind of God. Rather, it is a sentiment-ridden assertion made by a natural being, and those who cannot empathize with my sentiments may think it curious, or white noise, and would not risk failing to understand some truth about the universe in ignoring it.
I think I'm in the minority with respect to the philosophical community on these stances, but that appears to apply to everyone. Apart from the linked SEP articles (and some of the citations in those articles), those interested may want to follow Russell Blackford, or more famously, Simon Blackburn. The papers on Blackburn's site are excellent starters; Hume's Enquiry Concerning the Principles of Morals is a quick, enjoyable read; Blackford's monstrous review of Harris' Moral Landscape is a good example of moral skepticism in practice. (I haven't read The Moral Landscape, but from this and other reviews, I can't pretend that I'm interested.)
1. This is supported in the comments:
Look, I made a very specific claim. I maintained that the modern secularist typically does not only reject theistic morality, but also rejects morality based on transcendent forms that we can know, and an inherent purpose for human existence. So what I had in mind was leaving theism out of it and trying to see that there have to be some "metaphysical" commitments made in order to have anything that looks like an adequate moral theory. For the purposes of this discussion it would be better if everyone forgot about the fact that I happen to be a Christian theist and addressed the problem I actually posed.
Edit 1 (7/27/2011): I altered some of the wording, but not, I think, in a way that affects the message.