1. Many of the moral intuitions that we are inclined to accept are those which evolutionary selection would favour.
2. Evolution favours our acceptance of those intuitions independently of whether they are true.
3. From (1) and (2), either: (a) The intuitions evolution favours just happen to coincide with truth, or (b) Many of our intuitions (those which evolution has favoured) have no justificatory weight.
4. (a) is hopelessly optimistic.
From (3) and (4):
5. Conclusion: (b) Many of our intuitions have no justificatory weight.
Premise (3) strikes me as somewhat problematic, since as stated the dilemma seems to be a false one. Given that evolutionary selection favors moral intuitions, and that it does so independently of whether they are true, and ruling out mere coincidence, our intuitions still could have justificatory weight if attributing such a weight to them could be justified by something that has justificatory weight. The moral realist isn't really caught in a dilemma; he faces a challenge (to give a justification of the moral intuitions in question, i.e., to show that in normal circumstances they at least probably have a tendency to be true, as far as is discernible from other things ascertained to be true), and that's a very different sort of thing. In that sense the fact that premise 1 pulls its punch -- that it talks about many rather than all -- turns out not to be so trivial as one might have thought.
But, as Alex notes, the premise that one would most likely contest is #2. The response to such a contest is that, whereas our ability to recognize the truth of factual statements increases our chances of survival, our ability to recognize the truth of value statements could not do so.
This is an interesting argument; but I have difficulty making much sense of it. Now, the likely avenue for arguing this is what we can call, following Sharon Street and others, the adaptive link alternative: that the reason such tendencies are favored is not that they are true, but that they are mechanisms linking certain kinds of circumstances to adaptive responses. It's clear that there are lots of these. For instance, one tendency I have is, having touched a very hot surface (circumstance) to withdraw my hand quickly (response); the link between circumstance and response here is adaptive in the sense that it increases my chances of reproducing (by decreasing my chances of dying).
There very clearly are value statements that, if both true and accepted as true, would contribute to the survival for both individual and species more than they would if they were false but accepted as true. Take, for instance, the claim, 'It is good for our health and well-being to be as accurate as possible about potential dangers'. If this is true, accuracy about potential dangers contributes to our survival -- since anything that is good for our health and well-being contributes to our survival -- and certainly more so than if it turned out to be false. Now, this is certainly a value statement, since 'good' is a value term; what is more, it is a second-order value statement, since 'well-being', and at least arguably 'health' and 'danger' as well, is a value term. So some intuitive tendencies to accept some kinds of value statements would be favored more if those value statements are true. And any value statements that are allowed in start ramifying; even if they are not morality in a strong sense, they are a 'lesser morality' (to use a Humean phrase), are often treated as moral statements, and quickly start connecting up to very strong moral statements with only a few suppositions.
The problem with the adaptive link response, as far as I can see, is that it's not actually a competitor explanation. For instance, I have a tendency to see middle-sized objects in my vicinity; this tendency both tracks truth and makes a massive contribution to a web of links between my various circumstances and my various responses -- a web of links that are adaptive because they increase my chances of reproducing by decreasing my chances of being killed by middle-sized objects in my vicinity and increasing my chances of finding a mate. It would be absurd, however, to say that because vision contributes to reproductive success that therefore it has no justificatory weight when it comes to the truth or falsehood of claims about middle-sized objects; for the obvious reason that it contributes to reproductive success for the very same reason it justifies claims about middle-sized objects -- namely, that it is a means of becoming informed, however fallible and limited and approximate the information may be, about middle-sized objects. Indeed, because vision has potential justificatory weight for claims about middle-sized objects, it has justificatory weight for claims about much, much more (scientific practice builds on this sort of thing). And a similar response is available to the moral realist: the reason we think that certain kinds of moral intuitions establish adaptive links between circumstances and responses is the very same reason we should think that these moral intuitions have some tendency, however fallible and limited and approximate, to truth. And, as I've noted, there are genuine value statements, however weak, that seem to fit this sort of strategy very well. The most plausible account of why we tend to treat knowledge of potential dangers as a good thing is one that admits that it is a good thing; it contributes to good states of being, which increases our chances of survival and healthy reproduction, thus establishing an adaptive link between circumstance and response. It is true, of course, that a species like us except for a tendency to believe that knowledge of potential dangers was a bad thing would be much less likely to leave offspring than we are. But what of it? That's simply a confirming reason to think that knowledge of potential dangers is a good thing. Contrary to the claims of Street and others, Bad Things really do sometimes eat you. That things can eat you is as good a sign that they are able to be Bad Things as that they are carnivores. And the moral realist just needs to take that and move on from there.
The Darwinian Dilemmaist would, I suspect, think that somewhere in this the moral realist is begging the question. Three points in particular seem likely to be made.
(1) Coincidence. How does the moral realist explain the fact that the tendencies toward beliefs that are true just happen to coincide with the adaptive links between circumstances and responses? But as noted, for at least some evaluative intuitions there seems to be no mystery and no coincidence; and given those, we can work our way up.
(2) Contrary Predispositions. How does the moral realist deal with the fact that we seem to have contrary predispositions -- tendencies to make bad evaluative judgments? But this seems no more to be a problem than the tendency of our senses sometimes to misinform us. Tendency to truth is not the same as infallible possession of it. Indeed, it is not even close. Our senses establish a tendency to make true judgments about objects that we sense, as well as sensible objects that we don't, and, under certain conditions, nonsensible objects as well. But this tendency is merely a tendency; and it's arguably nowhere as strong a tendency as we tend to think it is. What we do is reason things through, look at alternative explanations, gather additional information, in order to compensate for any weakness or limitation in the tendency. Further, it's clear that in the case of evaluations, especially moral evaluations, bad judgments can usually be traced to incompleteness. Thus, racist whites judge that whites are more morally valuable than blacks because they simply don't consider all the moral issues on the line in such a judgment. It's as if someone, having taken a look around the room, concluded that everything that existed at all was in that room, rather like Douglas Adams's man who rules the universe. But the fact that we often make judgments on incomplete information doesn't shed any light on whether our materials for judgment are any good.
(3) Bruteness. How does the moral realist explain why we don't tend to make other judgments (e.g., that infanticide is a good thing) other than by simply appealing to their falsehood as a brute fact? But the moral realist doesn't appeal to their falsehood as a brute fact -- he is perfectly able to give all the adaptive reasons the Darwinian Dilemmaist is able to give. The only difference is that whereas the Darwinian Dilemmaist tries to impose a false dilemma, the moral realist jumps through its horns. Consider a similar sort of argument for claims about middle-sized objects. The universe of logically possible claims about middle-sized objects is utterly immense; and many of these logically possible claims are things we have no particular tendency toward. For instance, I have no particular tendency to judge that this white, papery-looking block in front of me is a supernova. And the reason, I would be inclined to say, is that it pretty clearly is not a supernova, since it pretty clearly is a book. It would be nonsense to say that this is merely appealing to the falsehood of the claim 'This block is a supernova' as a brute fact. For one thing, I have multiple lines of evidence and reasoning at my beck and call to argue that this is not a supernova but is, in fact, the Proceedings of the American Philosophical Association, which is a very different thing. Likewise, I can point out that a supernova is more interesting than the Proceedings of the American Philosophical Association. But I don't have to appeal to this as a brute fact; I can point out all sorts of reasons to think it so.
One reason I can do this is that particular judgments are merely particular judgments, but tendencies to make good judgments cover entire regions of judgment. If I take any particular judgment, such as 'This thing that I see is not a supernova', the adaptive link discussion tells us nothing about it, for the obvious reason that if I have genuine tendencies (however limited, fallible, or approximative) to make good factual judgments about middle-sized objects in my vicinity, all that I need to make this particular judgment is for this thing, falling within my field of vision, actually not to be a supernova. Similarly, the Darwinian Dilemmaist seems to be engaging in a confusion of levels when arguing against the moral realist. There are actually two issues:
(1) what explains our tendencies or predispositions to make a certain kind of judgment about X
(2) what explains the fact that this particular kind of judgment about X is a good one.
The moral realist has a (2)-explanation for certain kinds of value judgments: that there really are values of a certain kind. He then points out that this goes some way toward a (1)-explanation, because some of these values are clearly adaptively relevant if true, and from these we can rationally expand our field of discussion to encompass all sorts of good moral judgments. The Darwinian Dilemmaist responds that we have a (1)-explanation, namely, that some of the values are adaptively relevant and that therefore there is no need for (2)-explanation. But we know in fact that this is not generally the case; for instance, no one would accept a Darwinian Dilemma for visual realism, because (1)-explanations don't usually tell us much of anything relevant to (2)-explanations. It is true that the moral realist needs multiple lines of evidence suggesting that the tendency is not a bad one and that some of the resulting judgments are certainly good ones -- as the visual realist definitely has. But this sort of confirmation is all the moral realist needs. And moral realists don't ignore this point; they don't simply point to the predispositions and then say, 'That's that'. They elaborate, give reasons for thinking the predisposition tends in the right direction here but not, due to unusual circumstances, there, and so forth. And that's what really would need to be examined to see whether moral realism is a cup that can hold water.