An "evolutionary" argument against moral objectivism?

At least in my more optimistic moments, I lean heavily toward a belief in an objective morality, i.e. a morality that isn't true only relative to minds. (For a fuller discussion of the kind of "objectivity" in question here, see this earlier post.) However, I have long been bothered by a nagging worry, which can be expressed in the following argument:

1. If there is an objective morality that applies to human beings, then human beings must be capable of knowing that morality. (Premise)
2. If human faculties have been shaped only by random factors (e.g. mutations, genetic drift) and natural selection, then human beings are probably incapable of knowing an objective morality. (Premise)
3. Therefore, if human faculties have been shaped only by random mutations and natural selection, then there probably is no objective morality that applies to human beings. (1, 2)

It's important to realize how modest the conclusion is. The conclusion doesn't deny that an objective morality applies to human beings. It doesn't even say that an objective morality probably doesn't apply to human beings. Rather, it simply says that if human faculties have been shaped only by random factors and natural selection, then an objective morality probably doesn't apply to human beings.

Is the argument sound? It's valid, so any unsoundness must result from false premises. Are the premises plausible?

I regard premise 1 as non-negotiable: there must be some sense in which it's true. If there's no sense in which human beings can know objective morality, then either (a) there is no objective morality that applies to human beings or (b) there is such a morality, but human beings are incapable of knowing it. I regard (b) as intuitively absurd. If human beings lack any epistemic access to an objective morality, then it seems absurd to say that that human beings are answerable to that morality.

We might quibble over exactly how to interpret premise 1. If an objective morality applies to human beings, must all human beings be capable of knowing that morality? Perhaps not. Even a moral objectivist might want to leave room for morally malformed individuals—individuals who are incapable of grasping morality due to emotional trauma, bad upbringing, etc. A moral objectivist might even want to leave room for the possibility that human beings require special training in order to understand or perceive moral principles. Nonetheless, I claim that human beings must have some kind of epistemic access to any objective morality that applies to them.

Most of the resistance to the argument will probably focus on premise 2. Yet, at least at first glance, premise 2 has an undeniable plausibility. Random factors, being random, will not tend, by themselves, to produce faculties that provide epistemic access to anything—much less to objective morality. In contrast, natural selection is not random. However, natural selection doesn't care about morality; it just selects traits that promote survival and reproduction. So why would natural selection make human beings capable of knowing objective morality?

At this point, there's an obvious objection that one might feel tempted to make. Morality involves "prosocial" behaviors like kindness and generosity. And a society containing kindness and generosity is more likely to survive than one without kindness and generosity. Thus—one might argue—in selecting for traits that promote survival, wouldn't natural selection be likely to select for an awareness of morality?

There are two problems with this objection. First, it confuses knowledge with behavior. If prosocial behaviors promote survival, then natural selection might give human beings psychological tendencies to perform prosocial behaviors. But natural selection can do that without making human beings capable of knowing that those prosocial behaviors are objectively right. To know that prosocial behaviors are objectively right, human beings would need more than a tendency to perform such behaviors; indeed, they would need more than a evolutionarily-induced belief that such behaviors are objectively right: they would need (it seems) nothing less than perceptual and/or cognitive faculties capable of accessing moral properties such as rightness. And there's absolutely no reason to think that natural selection would give human beings such faculties.

Here's the second—and more important—problem with the objection: it begs the question. It assumes that kindness and generosity are objectively right. In doing so, it assumes that we already know what actions are objectively right. But that's precisely what's in question. The question is how we could possibly know what's objectively right, if our faculties have been shaped solely by random mutations and natural selection. We surely cannot answer this question in a way that presupposes knowledge of what's objectively right!

At this point, one might object that I've been focusing on only one kind of moral objectivism—specifically, a kind of extreme moral realism, according to which rightness is an intrinsic property of certain actions, analogous to the roundness of an orange or the mass of an electron. If this is what we mean by "moral objectivism", then premise 2 seems to be true: it's certainly hard to see why natural selection would give us faculties capable of detecting such a mysterious and biologically-irrelevant property. But when many philosophers say that they believe in objective morality—the objector might say—they simply mean that if I, say, see a drowning child, then I have a reason to act—namely, a reason to save the child. According to this objection, moral objectivism is a matter of objective moral reasons, and in order to know reasons, one needs nothing more than the faculty of rationality—something that natural selection clearly has bestowed upon us. So (the objection runs) premise 2 is false: natural selection has given us the ability to know objective morality.

I have two remarks about this objection. First, it may draw a false distinction between a metaethics based on moral properties and one based on moral reasons. After all, if I have (moral) reason to save the child, then saving the child is the right thing to do; and if saving the child is the right thing to do, then the act of saving the child has the property of rightness. We don't escape moral properties simply by shifting to talk of moral "reasons". (Of course, we needn't think of moral properties as analogous to natural properties such as roundness or electric charge. We could say that to have the property of rightness just is to be the action that one has (moral) reason to perform.)

Second, the objection simply fails to address the issue. Suppose we shift from talk of moral properties to talk of moral reasons. Then we face the following question: Why would natural selection give us the ability to know moral reasons? A vague appeal to "rationality" will not solve this problem. To achieve knowledge on a given subject, rationality needs some sort of "starting-point". For example, if I want to achieve knowledge about the solar system, my starting-point is astronomical observations. If I want to achieve knowledge about human psychology, one of my starting points is observation of human behavior. Now, what is my starting-point for knowledge about objective morality? It seems that the starting-point would have to be a perception or intuition of certain moral properties, principles, or reasons. But, for reasons that we've examined, there's no reason to think that natural selection would give us the ability to access such properties, principles, or reasons, since they are irrelevant to survival and reproduction.