We belong to a species that evolved, like others, through cumulative natural selection, and our coming to appreciate this fact has changed how we think about many aspects of human life. But does our evolutionary background provide insight specifically into the status of ethics? Most of those who see deep connections here take evolutionary considerations to support a deflationary metaethics. Whether the thought is that evolution supports ethical nihilism, skepticism, subjectivist constructivism, or expressivism, it is often held to provide a basis at least for debunking ethical realism by undermining the old-fashioned idea that there are objective and knowable ethical truths—truths that are both independent of our evaluative attitudes and capable of being known by us. I will argue that this common debunking claim is mistaken: evolution does not provide grounds for deflationary metaethics, though it does raise questions that realists need to address and it helps to bring out some interesting commitments of realism.

Following recent discussions, I will focus on epistemically-oriented evolutionary arguments, with the aim of disentangling and critiquing various strands of argument to be found in the current debate. In doing so, I will abstract a bit from the actual arguments given by particular authors, but I hope to say enough to cover them all in the end.Footnote 1 Some of these strands are motivated specifically by evolutionary considerations while others turn out to be more general arguments with an evolutionary spin. So while in some cases the evolutionary angle provides a distinctive epistemic challenge, in others it adds little to already familiar ones. My aim will be to sort through these and to defuse the main arguments insofar as they claim that evolution raises distinctive epistemic problems for realism. I will argue that the most prominent lines of evolutionary argument against realism fail because they rely on strong claims about the explanation of ‘our moral beliefs’ that are not supported by the science unless it is supplemented with philosophical assumptions that are just question-begging in this context.

1 The capacity etiology argument

Consider first a simple argument that does draw essentially on facts about evolution, concerning the etiology (or causal origins) of our epistemic capacities:

  1. (1)

    Our basic mental capacities, including those we employ in moral judgment, are products of natural selection.

  2. (2)

    Natural selection would not have designed these capacities to be such as to yield cognitions that reliably track independent moral truths, even if they exist.Footnote 2

  3. (3)

    Therefore, even if independent moral truths exist, the capacities we employ in moral judgment either do not yield cognitions that reliably track them, or they do so only accidentally, through a lucky fluke—both of which raise skeptical problems.

The idea here is that as evolved creatures we have natural selection to thank for the basic mental capacities we employ in moral judgment, whether these are adaptations for moral judgment or just more general adaptations we’ve come to use also for making moral judgments. But processes of natural selection would have shaped these capacities, like any others, ultimately in whatever ways most enhanced the inclusive biological fitness of hunter-gatherers in ancestral environments, given the available genetic variation. Natural selection would simply have rewarded whatever cognitive and emotional traits caused our Pleistocene ancestors to maximize the relative representation of their genes in the gene pool over generations. Now in some cases, these traits accomplished this by promoting survival, which they did in turn through promoting reasonably accurate cognitive representation of the world. For example, they gave our ancestors capacities for tracking facts about the location, number, speed, size and color of medium-sized objects in their environment. So in those cases we can expect natural selection to have given us cognitive capacities that track independent truths about the world with reasonable accuracy (Griffiths and Wilkins In Press). But, the argument goes, this does not hold for the independent moral truths posited by realists.

Realists do not, after all, take these moral truths to have anything in particular to do with the competitive gene propagation rewarded by natural selection, but instead see them as grounded in various independent values. There might, of course, be some overlap between the behaviors prescribed by such values and the behaviors that tended to maximize a hunter-gatherer’s genetic propagation—for example, refraining from leaping off cliffs or from arbitrarily betraying coalition partners. But the point is that these choices would have been rewarded by natural selection not because of their fidelity to objective moral values, but simply because they increased biological fitness. Indeed, whenever a behavior tended to increase biological fitness but failed to accord with objective values—as in the case of selective cheating, say, or aggression toward weak outsiders—natural selection would still have favored it or its psychological underpinnings; and natural selection would not have favored psychological dispositions leading to behaviors consistent with objective values in cases where those behaviors failed to increase biological fitness.

So it is not independent moral facts as such that natural selection shaped our cognitive capacities to track, but simply facts relevant to competitive gene propagation in ancestral environments.Footnote 3 This means that even if there are independent moral truths, natural selection would not have given us capacities designed to track them as such or otherwise disposed to track them consistently. And (so the argument goes) this implies that the capacities we employ in moral judgment are either unreliable, at least with respect to tracking independent moral truths, or only accidentally reliable through a lucky fluke.

Now there is much that might be debated about the epistemic upshots of this conclusion or the prospects for dealing with them, but we can spare ourselves some trouble by noting straightaway that the argument is invalid. To make the argument valid, we could add a further premise, which seems to function as an implicit assumption in many discussions:

(2.5) If natural selection is responsible for our having certain basic cognitive capacities, which we employ in some domain of thought, then the only way for such exercises of those capacities to be non-accidentally and reliably truth-tracking in that domain would be for natural selection to have made things that way.

The problem, however, is that this assumption overlooks the alternative possibility of our taking general cognitive capacities bequeathed by natural selection and developing them in cultural contexts, through relevant forms of training within traditions of inquiry into the subject matter in question, and thus making our cognitive dispositions in the relevant domain non-accidentally reliably truth-tracking. Indeed, we have plainly done this in countless cases. So 2.5 is false.

To take just one example of how this might work, consider philosophical beliefs we might have, such as the belief that if water is H2O then this is a metaphysically necessary truth: water is H2O in all possible worlds where it exists. Now the basic mental capacities that enable us to sit around worrying about things like metaphysical modality are part of our evolutionary heritage: they didn’t appear by chance and they weren’t designed by God; they evolved through natural selection. But natural selection did not design our cognitive capacities to track truths about metaphysical necessity, which were as irrelevant to the reproductive success of Pleistocene hunter-gatherers as the realist’s moral truths would be. Hunter gatherers didn’t fill their time around the fire pondering metaphysical necessity—or if some did, it’s safe to say they weren’t the ones enjoying most of the reproductive success! Still, despite the irrelevance of obscure metaphysical facts to the natural selection etiology of our basic mental capacities, we’re able to deploy those capacities, in the cultural context of philosophical training, to think intelligently and often accurately about things like metaphysical necessity or countless other arcane topics such as differential geometry and relativistic quantum theory, the facts of which are equally irrelevant to the etiology of the capacities we use in thinking about them (cf. Railton 2010).

The first point, then, is that we don’t need natural selection to have given us cognitive capacities designed specifically to track a certain class of truths, on the model of perceptual adaptations, in order to be in a position now to track those truths non-accidentally and reliably, and to be warranted in our beliefs. Nor do we even need natural selection to have given us, as an incidental by-product of some unrelated adaptation, a ready-made, specialized capacity that happens to be attuned to the truths in question. Such a thing would indeed be as unlikely as natural selection’s coughing up the human eye as a fluke by-product of some unrelated adaptation.Footnote 4 But again we don’t need any such thing. It’s enough if natural selection has given us general cognitive capacities that we can now develop and deploy in rich cultural contexts, with training in relevant methodologies, so as to arrive at justified and accurate beliefs in that domain.Footnote 5

We needn’t worry, then, that it would take some ‘miraculous coincidence’ for our beliefs about metaphysical necessity to match the metaphysical facts simply because natural selection didn’t hand us a ready-made faculty for doing metaphysics. It’s enough if natural selection has given us general capacities we can develop and use, with philosophical training and a little help from Kripke and Putnam, to grasp sound arguments for conclusions about metaphysics. And similarly for truths in higher mathematics and physics and, the ethical realist will now add, ethics. We don’t need either scientifically dubious adaptations for tracking moral truths or improbable by-product capacities that just happened to come specially attuned to them: it’s enough if we’ve learned to use our capacities for critical reflection on our experience, and our reasoning with normative concepts, in the context of traditions of moral inquiry, to reflect accurately on how it’s good and right to live. Of course, the realist owes a plausible metaphysical and epistemological account of how this works in the case of ethics, and these quick remarks are obviously not meant to substitute for that. But that theoretical burden is nothing new, and the point is that the evolutionary considerations so far pose no special obstacle to the realist.

2 Are moral capacities specially problematic?

Defenders of the capacity etiology argument might respond as follows. It’s true that we often use our evolved cognitive capacities in ways that go far beyond anything natural selection designed them to do, discovering truths of types that had nothing to do with the ancient biological origins of those capacities. Still, this is not the case for the domain of moral beliefs and independent moral truths, if there are any. Unlike in other domains of belief, when it comes to morality we’re just out of luck if natural selection hasn’t given us ready-made capacities specially attuned to moral truths; for we cannot now develop sound moral methodologies that would enable us to track moral truths reliably, as we’ve done in so many other domains. So, the thought goes, while 2.5 is indeed false, a similar claim limited to moral beliefs is true, and that is sufficient for the debunking argument to go through.

Now that would be a problem, but why believe even this more targeted version of 2.5? Perhaps the worry is a sense that in the other examples—from metaphysics, math and physics—the reasoning deployed is in some way just an extension of a broader sort of reasoning that our cognitive capacities were designed to be accurate about, thus explaining its reliability. Facts about prime numbers may have had nothing to do with the etiology of our mathematical capacities. But those capacities were plausibly designed to yield accurate thoughts about numbers—for example, in the course of deciding whether to fight or flee from a hostile tribe. And if we add that we also evolved capacities for logical reasoning and conceptual sophistication, it’s unsurprising that we can now draw on these capacities and apply them in reflecting on numbers to reason accurately about prime numbers, eventually discovering such arcane truths as that there are infinitely many primes. Similar stories might be told for physics and metaphysics. But, the argument goes, the same is not true for morality. Moral reasoning is not similarly an extension of any reasoning our cognitive capacities were designed to do accurately. It’s just a sui generis domain of emotion-laden thinking that, if favored by natural selection, was favored not for accuracy but just for direct contributions to biological fitness. One cannot, then, appeal to other domains such as math, science and metaphysics as ‘companions in guilt’ (cf. Joyce 2006, pp. 182–184).

The main difficulty with this line of response, however, is that it’s just not true that moral reasoning cannot likewise be viewed as an extension of forms of reasoning our capacities were designed to do accurately. At a formal level, we employ the same logical and analytic abilities in moral reasoning as in other forms of reasoning.Footnote 6 And in terms of conceptual content, moral reflection and reasoning is continuous with broader evaluative and normative thinking that our cognitive capacities were plausibly designed to do accurately. Since so much of the discussion in this area is breezily speculative, we may as well hop aboard and add that it was likely important for our Pleistocene ancestors to understand the application of evaluative concepts in connection with relevant standards. They needed to make accurate evaluative judgments about good and bad dwelling places, or hunting partners, fighters, and mushrooms, and related normative judgments such as that one ought not to eat the little brown mushrooms or to fight with Big Oog. Moral judgments obviously go beyond these sorts of things, but just as in the other cases, they can be seen as an extension of such thinking. They still involve employing evaluative and normative concepts in connection with standards and ends, though now conceived as standards and ends defining what it is to live well all things considered, rather than just narrow standards of edibility or safety.

This is undeniably a major cultural development, requiring abstract thought and training to make the transition to moral judgments that might reliably track moral truths. But that’s equally the case for the transition from counting hyenas or estimating the path of a thrown rock to doing research on quantum electrodynamics or modal metaphysics. All of these pursuits require conceptual and analytic sophistication and the use of methodologies that take us far beyond the mental exercises that figured into the evolutionary shaping of our capacities. It’s a striking fact that we can do this so successfully—that the mental faculties that evolved by helping Pleistocene hunter-gatherers out-reproduce their peers turn out to bring with them the potentiality to penetrate truths about quantum non-locality or non-Euclidean geometry.Footnote 7 But given that this has happened, it’s hardly a greater stretch to imagine that our evolved faculties also enable us, in the right cultural contexts, to figure out that race-based voting laws or the stoning of rape victims, for example, are unjust and cruel practices, which are not part of an all-things-considered good way for a society to behave and ought to be eliminated.

We discover the evil of racist voting laws, for example, by gaining empirical knowledge about the irrelevance of race to what matters to responsible voting, and by reflecting on the significance of such facts in light of ongoing experience of human life and the possibilities of good and harm it offers us, as part of forming a conception of what it is for human beings to live well. Why should this sort of intelligent extension of evolutionarily influenced evaluative judgment be thought any more problematic in principle than parallel extensions in other domains? Indeed, debunkers such as Street, who despite rejecting realism wish to save moral knowledge by moving to a subjectivist constructivist account of moral truth, presumably agree that such extensions are possible and afford us moral knowledge. So their thought must just be that this is problematic if the moral facts are taken to be objective or independent of our contingent desires or attitudes, as realists maintain. But nothing about evolution considered so far supports any special worry along those lines. The independence or objectivity of the goodness or badness of potential partners as hunters, or of certain mushrooms as food, was no hindrance to accurate evaluative judgment about them. Similarly, there is no reason so far why the independence or objectivity of the badness of racist voting laws should be thought to pose a special obstacle to our discovering that badness, at least as far as evolutionary considerations are concerned.

Of course, there may well be special epistemic challenges to be met in accounting for moral knowledge on a robustly realist model, as many have long argued. But again the point is that considerations about the evolutionary origins of basic capacities employed in moral judgment do not pose any obvious additional obstacle. At most they impose a constraint on realists as we go forward in developing a positive moral epistemology: any such account must at least square with our best scientific understanding of the sorts of capacities evolution gave us to work with, avoiding reliance on capacities we could not plausibly have developed from such psychological materials. That is a useful result and may pose an interesting challenge for some realists, but it falls far short of debunking realism.

3 Content etiology arguments

At this point the line of objection will likely shift to worries about the etiology of the content of our moral beliefs. It’s not just that natural selection has not designed our cognitive capacities to track independent moral truths: the real problem stems from what natural selection has done in the way of shaping the very content of our moral concepts and beliefs. Our moral thinking, after all, proceeds from intuitive moral starting points that are also used in scrutinizing other moral beliefs when we reflect. But that means that unless our intuitive starting points are roughly accurate, we have a garbage in/garbage out problem.Footnote 8 And, the objection goes, we have every reason to suspect that our starting points are not accurate (in realist terms), due to deep evolutionary influences on our moral concepts and beliefs—influences that would, again, have no particular tendency to make our beliefs track moral truths. So even if there are independent moral truths, we have no reason to be confident we’re tracking them in our moral thinking: more likely, we’re just reasoning our way to more consistent garbage of one kind or another (by realist standards). Once this becomes clear to us, we’re then either stuck with skepticism (we cannot have moral knowledge even if there are independent moral truths) or we must drop the realist standard of correctness for moral belief as alignment with independent moral truths.

This broad line of argument—call it the Content Etiology Argument—can be developed in a few different ways. The most extreme begins by denying any explanatory role to independent moral truths in accounting for our moral beliefs, which latter are just the upshots of causal factors, such as evolutionary influences, operating independently of the truth of the content of the beliefs. There are at least two versions, which I’ll call the ‘Strong Explanatory Debunking Arguments’.

3.1 The strong explanatory debunking arguments

Consider first the Implausibly Lucky Coincidence Argument:

The implausibly lucky coincidence argument:

  1. (1)

    Independent moral properties and truths play no role in the explanation of our moral beliefs.

  2. (2)

    If independent moral properties and truths play no role in the explanation of our moral beliefs, then it would be an implausibly lucky coincidence if our moral beliefs turned out to represent these independent moral truths accurately.

  3. (3)

    Thus, it would be an implausibly lucky coincidence if our moral beliefs turned out to represent these independent moral truths accurately.

  4. (4)

    If correctness for moral beliefs consists in accurately representing independent moral truths, as the realist claims, then at least once we become aware of 3, and in the absence of any independent confirmation of the truth of our moral beliefs, whatever default justification we may have had for our moral beliefs is thereby defeated.

  5. (5)

    Therefore, if the realist is right in holding that correctness for moral beliefs consists in accurately representing independent moral truths, then we’ve lost our justification for our moral beliefs once we’ve become aware of 3, lacking any independent means of confirming their truth: realism saddles us with skepticism.Footnote 9

The claim behind premise 1 is that there is plausibly a complete non-moral genealogy for all moral beliefs in the following sense: any moral belief can be fully explained—leaving out nothing of explanatory interest—entirely in terms of causal factors operating according to principles that are insensitive to independent moral properties or facts as such, even if they exist (Joyce 2006; Kitcher 2011). That is, we were caused to have our moral beliefs by factors having nothing to do with independent moral properties or facts: our moral beliefs were caused neither by such properties or facts nor by our grasping evidential relations other facts bear to them, as such, but simply by ‘morally blind’ forces of one kind or another. That’s the thought, and if this is right, then there will be problems for the realist, as the argument brings out.Footnote 10

An example from Joyce (2006, p. 179) helps to illustrate the point. Suppose you found out that your belief that Napoleon lost the Battle of Waterloo is entirely the product of a pill someone slipped you, which causes such a belief to be formed, completely independently of any historical facts; there’s also a pill that causes the belief that he won, and you just happened to have been given the ‘lost’ pill. Surely this should undermine your confidence in your belief that Napoleon lost the Battle of Waterloo. For it would be a pure coincidence if your belief happened to match the historical facts, which played no role in the etiology of your belief. But the same holds for moral beliefs on a realist picture of the moral facts if 1 is true and our moral beliefs are simply the products of causal forces that are blind to independent moral properties and facts as such.

Before critiquing this argument, consider also a related variant:

The explanatory superfluity argument:

The Explanatory Superfluity Version begins with the same first premise but then continues along a different track:

  1. (1)

    Independent moral properties and truths play no role in the explanation of our moral beliefs.

  2. (2a)

    If independent moral truths play no role in the explanation of our moral beliefs, then they are explanatorily superfluous.

  3. (3a)

    Thus, independent moral truths are explanatorily superfluous.

  4. (4a)

    If independent moral truths are explanatorily superfluous, then we should not posit them—just as we don’t posit truths about witches, given a complete and exhaustive genealogy of beliefs about witches that makes no appeal to witch truths.

  5. (5a)

    Therefore, we should at best remain agnostic about the existence of independent moral truths, and thus also about whether any of our particular moral judgments are true (at least by realist standards). We should give up any claim to moral knowledge (at least on a realist construal of it).Footnote 11

Again, consider a parallel with Napoleon beliefs. Suppose you discovered that all your beliefs about Napoleon, even regarding his very existence, were solely a product of a pill that produces beliefs independently of any Napoleon facts: as Joyce puts it, “without this pill you would never have formed any beliefs about Napoleon at all” (Joyce 2006, p. 181). Surely you would then have to give up your beliefs about Napoleon and at best remain agnostic about whether there are any (positive) Napoleon facts at all. The claim, then, is that we’re in a similar position with respect to our moral beliefs and moral truths, with various causal factors producing all our moral beliefs independently of any moral facts.

3.2 Critique of the strong debunking arguments

Now notice that one could in principle run both of these arguments without any appeal to evolution at all, if there were some independent support for premises 1 and 2. Someone might, for example, appeal simply to familiar psychological, sociological and historical causes as determining all our moral beliefs independently of any moral facts, with no mention of evolution.Footnote 12 Still, many believe that evolutionary considerations provide compelling new support for the first two premises.

Joyce (2006, Chap. 6), for example, argues that natural selection gave us our basic moral concepts—such as fairness, harm, cheating, guilt, generosity, loyalty, etc.—along with dispositions to make certain moral judgments employing them. These psychological adaptations then combine with cultural and other causal influences to produce our moral beliefs. As always, the principles by which these concepts and dispositions were bequeathed by natural selection had nothing to do with independent moral truths: they were given to us not to help us form accurate beliefs about moral truths, but simply to enhance the fitness of ancestral humans by producing adaptive behaviors. Thus, our moral beliefs can all be explained entirely by causal chains tracing back to evolutionary influences having nothing to do with independent moral truths, which latter therefore play no role in the explanation of our moral beliefs, thus giving us premise 1.

Now it should be admitted that some such debunking story is a possibility, and it will naturally be attractive to those who have already rejected realism for other reasons. But this is nothing new. We’ve always known that there are stories, evolutionary or otherwise, according to which moral properties and facts play no role in the etiology of our moral beliefs. The mere availability of some such story does not by itself actually debunk the realist alternative: it simply provides a rival account of our moral beliefs, which will succeed in its debunking ambitions only if it is actually correct. Those attracted to it will of course cite virtues such as greater parsimony (it explains our beliefs without having to appeal to real moral properties and facts), and this may contribute to their own justification for believing it, given the rest of their views and commitments. But this can hardly be expected to have offensive force against realists. Greater parsimony is a theoretical virtue only where the world is obligingly austere, and that is exactly what is at issue in this debate. The realist’s position (at least for the realist I am defending) is precisely that the world does contain objective values that figure crucially into our having at least some of the moral beliefs we hold, in which case greater parsimony in explaining our beliefs exclusively in other terms is not a virtue at all: it is just a misrepresentation.

Debunkers therefore cannot expect to gain ground against realism simply by proposing a story that, if true, would cause problems for realism, and then claim that simply because of greater parsimony we should all accept it as true and get on board with the strong debunking arguments. They must instead make a positive and non-question-begging case for the actual truth of their debunking story and so for premise 1. The question, then, is whether that case has even really begun to be made: does the scientific evidence concerning evolution really support such a strong explanatory claim as premise 1 if we don’t supplement it with an implicit prior rejection of realism to begin with? It is not hard to see, I think, that it does not, and that the Strong Evolutionary Debunking Arguments just beg the question against realism from the start.Footnote 13

Let us grant that natural selection provided ancestral humans with moral concepts and dispositions to employ them in fitness-enhancing judgments. What follows with respect to our current moral judgments? Very little, in fact. First of all, as Railton (2010) notes, paraphrasing Wittgenstein: there’s nothing preventing intelligent creatures from using crude tools to make refined ones. Even if a crude version of the concept of fairness, for example, first arose through natural selection operating on hominin psychology, nothing prevents our having refined it in cultural contexts, as we developed sophisticated reflective conceptions of fairness. And we can then use our refined concepts to make judgments quite different from those that early humans were evolutionarily shaped to make.

Most importantly, unless we have already rejected realism at the start, it remains a wide open possibility that these current refined employments of moral concepts often express moral truths, and that these judgments are not merely caused by ‘morally blind’ factors but are instead often guided by our recognition of the relevant moral properties and facts as such, yielding moral knowledge. Nothing in real science rules out this possibility, of course, since the sciences don’t even address it. And the point is that if this is how things in fact turn out to be, then independent moral properties and truths would indeed figure into the proper explanation of at least many of our moral beliefs after all, in which case premise 1 would be false: we wouldn’t actually have a complete, true and fully adequate non-moral genealogy of our moral beliefs, in the sense defined earlier.Footnote 14

Joyce might object that while the realist picture is possible it would require a highly unlikely coincidence: how convenient for the realist, he might complain, that concepts originally given to us independently of any role in arriving at moral truths just happen to be refinable into concepts we can use to grasp moral truths! But in fact there is nothing far-fetched here. We should fully expect natural selection operating on intelligent social creatures to give rise to proximate adaptive psychological mechanisms concerned with promoting cooperation, social harmony and stability (along with nastier ones for selective cheating, less friendly treatment of outsiders, and so on). These social concerns, even if they evolved originally in the service of genetic propagation, overlap significantly with the subject matter of morality on any plausible account (cf. Copp 2008). So it’s no surprise that natural selection might in this context give us basic versions of concepts such as fairness, cheating, wrong, and guilt, that also turn out to be refinable into concepts that figure into moral truths about proper human social relations. Given this overlap, there is no reason to look askance at these concepts just because of their evolutionary origins. If there is something wrong with our moral concepts, that will have to be shown by independent arguments revealing them to be bad concepts, as in the case of the concept of witch or vital force; it won’t be established just through generic claims about evolutionary origins.Footnote 15

I’ve claimed, then, that despite the evolutionary origins of crude versions of moral concepts we might nonetheless employ refined concepts today to arrive at and express moral knowledge. More specifically, on the realist picture I’m advocating, we hold many of our moral beliefs for good reasons, recognized as such, and not merely due to extraneous causes having nothing to do with the truth of the content of the beliefs. These reasons are at least often the very things that make the beliefs true, and it’s our grasping precisely that fact about them that explains our holding these beliefs, just as with beliefs in other domains.Footnote 16

It may help to consider a parallel with mundane, non-moral evaluative beliefs. Suppose I have a computer that crashes frequently and runs slowly, and is therefore a bad computer, as I correctly believe it to be. Why do I believe that it is a bad computer? Not simply due to extraneous factors that cause me to have this evaluative belief for reasons having nothing to do with its badness—at least not in ordinary cases. Instead, if I’m competent and informed, I believe what I do because I am aware both of its bad-making properties and of their bad-makingness: I see that it runs slowly and crashes, and knowing about computers I understand that these are bad-making properties in a computer, inconsistent with the standards of excellence for computers; I therefore judge that this computer, by virtue of possessing such properties, is a bad computer, and this judgment is based on these good reasons I am prepared to give in defending it. In such a case, then, we may say that I believe the computer is a bad computer because it is a bad computer and, being competent, I’ve grasped that evaluative fact by grasping the reasons why it’s a bad computer, as such. And that is to say that the evaluative properties and facts come straightforwardly into the ordinary explanation of my evaluative beliefs in such cases.

Now the standards for moral excellence in connection with human beings and action are obviously not fixed in the same ways that standards for artifacts are fixed, but these background differences don’t vitiate the comparison I wish to draw here. For it is entirely plausible for a realist to maintain that however exactly the moral standards are fixed, the explanation of at least many of our moral beliefs, in ordinary cases of moral knowledge, is structurally similar: we hold at least many of the moral beliefs we do because they’re true and, being morally competent, we’ve grasped that they’re true, by grasping the reasons why they’re true, as such. Indeed, this seems central to our claim to have not merely justified true moral beliefs but genuine moral understanding.

Consider, for example, our belief that it is seriously wrong to deprive girls of educational and career opportunities simply because they are girls—something practiced and sometimes enforced with brutal violence by the Pakistani Taliban. Why do we believe this? Some will no doubt claim that a fully adequate and exhaustive explanation of our belief can be given simply in terms of sociological, psychological, or historical causes utterly insensitive to any moral properties or facts. But no one with realist leanings who is serious about this moral belief should accept such a claim. Instead, we do better to begin with first-personal explanation in the case of a morally competent agent. If asked why we believe that such practices are morally wrong we will cite reasons that we take to support the truth of the belief, not merely psychological or sociological causes for it that operate independently of such reasons. Such practices, we’ll point out, are unjust, cruel, demeaning, and sexist, violating human rights and dignity by depriving these girls of central human capabilities and goods, based on arbitrary considerations. We cite these reasons as wrong-making features of these practices, against our background view of the standards of moral excellence for human beings and action. And in cases like this what is plausibly happening is that we are correctly grasping the wrong-makingness of these factors, and this is precisely what leads to our moral judgment: we believe the Talibanic practices to be wrong because they are wrong and, being morally competent, we’ve recognized this evaluative fact by grasping the reasons why they’re wrong, as such. And that is to say that the moral properties and facts come straightforwardly into the explanation of such moral beliefs. So again, if this is right then premise 1 is just false.

Let me be clear that my point here is not that this realist picture is obviously correct: at noted earlier, premise 1 could be true, and all our moral beliefs might be exhaustively explained simply by way of non-moral evolutionary, psychological and sociological accounts. It’s possible, for all I’ve shown, that our reasons for thinking Talibanic treatment of girls to be wrong are just post hoc rationalizations for a belief we’ve been independently caused to have, and similarly for all our moral beliefs: they might all conform to the social intuitionist model championed by Haidt (2001). Or perhaps some fit that model and the rest conform to another debunking model according to which our moral beliefs are held for the reasons we give, but our taking those considerations to be good reasons for those beliefs (e.g., our taking the cruelty of Talibanic practices to be wrong-making) is always just a mental state caused independently of their actually being good reasons, so that the fact that certain features are actually wrong-making has nothing to do with our believing that they are. This is all possible, but the relevant question is whether the debunking arguments have provided anyone who hasn’t already rejected realism for other reasons with any good reason to believe it. I claim they have not.

It’s not that these debunking models have no application. In fact, realists themselves might make liberal use of both of these debunking models to explain all manner of false moral beliefs—as with psychological or sociological explanations of what causes people to arrive at false moral beliefs about the immorality of interracial marriage or of homosexuality, or the acceptability of gender discrimination.Footnote 17 But the issue is whether such debunking models apply across the board, to all our moral beliefs, as against the realist’s factive model for at least many of our more informed, reflective moral beliefs. For that is what is needed to support the claim in premise 1 that gets the strong debunking arguments going. And there is simply no scientific support for such a sweeping claim, given that the sciences don’t even engage with the possibility that some of our moral beliefs are caused by our apprehension of moral facts as such. It’s just not true that “the empirical data concerning human evolution” support anything approaching premise 1 (Joyce 2006, p. 188).

Of course, the debunking models would be compelling across the board if there were no moral truths to be discovered, leaving us nothing but evolutionary, psychological, sociological or other similar causal factors to appeal to in explaining beliefs. But it would obviously be question-begging in the course of an argument against realism just to deny from the start that there are moral truths. So if debunkers are to avoid just begging the question against the realist they need to acknowledge the open possibility that there are moral truths of the sort realists posit. And if it turns out that there are, then there is every reason to expect that they do enter into the explanation of at least many of our moral beliefs, in just the way I’ve suggested in the case of our beliefs about the wrongness of Talibanic treatment of girls. And if those moral truths are relevantly independent of our attitudes, as the realist claims—e.g., if the wrongness of shooting girls to keep them from attending school is objective and independent of our contingent attitudes toward it—then premise 1 is false.

What, then, is the ‘best explanation’ of our moral beliefs? My point has been that this is itself just as controversial as the original issues that divide realists from antirealists. Aspiring debunkers cannot get anywhere against realists by making claims about what really explains, or best explains, our moral beliefs, where those claims are just question-begging against the realist’s picture from the start. If we are to avoid such question-begging, then we must recognize a plurality of plausible explanatory models and be open to the possibility that our moral beliefs are a mixed bag, some to be explained using the factive model I’ve advocated, others to be explained using non-factive causal models such as Haidt’s or the other debunking model described earlier, and still others using elements from both factive and debunking models.

For example, some moral beliefs, such as the belief that we should make present sacrifices to help protect distant, future generations from the harmful effects of climate change, are poor candidates for explanations appealing to evolutionary influences, and if we think that they’re true and that we know why they’re true, then the factive model is plausibly the best explanation. By contrast, in cases where evolutionary influences provide an attractive theory of error for moral beliefs that seem on reflection to be false, such as beliefs involving racism, sexism, or xenophobia, some form of debunking evolutionary explanation is most plausible. Finally, other beliefs may be overdetermined in the sense that they are to be explained partly in terms of our having grasped moral truths and partly in terms of fortuitous evolutionary influences. This is an important point that is often overlooked, leading many wrongly to suppose that the realist explanation of a belief is somehow discredited if there is a plausible evolutionary story for why we would tend to believe such a thing.

Our moral belief that we have an obligation to care for our children, for example, may be influenced, and so partly explained, by feelings of parental love stemming from a psychological adaptation that evolved for obvious Darwinian reasons. But it’s entirely compatible with this that there’s also a genuine obligation to care for one’s children, just as we think there is, and that we’ve grasped this fact through sound moral reflection. So there may be overdetermination in many cases, where there are both evolutionary influences on belief and also good reasons for holding those beliefs, which we have understood, just as with other beliefs. Indeed, for reasons given earlier we should expect there to be many cases of overlap between what evolutionary influences might lead us to believe and what there is good reason to believe (though there will also be plenty of divergence). This is not a problem for realists: as long as our believing what is true is at least partly guided by our recognizing good reasons for the belief, there will be no problem of merely lucky coincidence in our getting things right; it is no liability if in some cases this moral recognition wasn’t necessary for the belief because we would have held it anyway due to evolutionary or other causes.

3.3 Does weakening premise 1 save the debunking arguments?

These same considerations show why the aspiring debunker cannot simply weaken the first premise to avoid the question-begging problem. Suppose premise 1 were weakened to say only that:

(1′) A purely non-moral, causal account of the relevant parts of the world is sufficient to imply our having all the moral beliefs we have, without any appeal at any point to independent moral truths.

Perhaps this is all some mean by speaking of there being a “complete non-moral genealogy” of moral beliefs, which needn’t strictly deny that independent moral truths may also often figure into the etiology of some of our moral beliefs: there could be overdetermination in every case of true moral belief. This weakening would thus technically avoid begging the question against realists in the way that premise 1 did. Does this avoid the problems and save the debunking arguments?

It does not. First, no realist should accept even 1’: for if realism is true, and if our moral beliefs are largely correct, then 1’ itself posits a truly remarkable coincidence: somehow a motley set of causal factors, all operating according to principles utterly insensitive to independent moral truths as such, just happen to have conspired to push us in just the right ways as to cause us to hold all of the true moral beliefs we hold. There is no plausibility in such a claim. As before, then, only those who have already rejected realism for other reasons should find 1’ compelling. So weakening the first premise in this way will not help the argument to gain traction against anyone who hasn’t already accepted its antirealist conclusion.

Second, weakening 1 to 1’ takes the fangs out of the debunking argument. Since the possibility of overdetermination is left open, in all such cases moral truths would still have an important explanatory role to play in a fully adequate account of our moral beliefs that leaves out nothing of explanatory importance, and so they aren’t explanatorily superfluous after all; nor again would there be any problem with lucky coincidence, given the real work moral truths would still be doing in guiding our beliefs. Some may claim that if sufficient non-moral explanations are available then it is obviously better just to go with those and jettison moral factive explanations, which simply ‘aren’t needed’. But again, while that is an understandable position for someone who is already an antirealist to take, it’s hardly something a realist can be expected to find compelling up front.

As a realist I claim that the moral, factive explanation of my belief that I have obligations to my children is absolutely needed if we want a true and fully adequate account of it. And I’m not going to be convinced otherwise without first being convinced either that my reasons for thinking I have such obligations aren’t really good ones after all or that somehow my taking them to be good reasons was in no way responsive to their actually being good reasons, despite how things seem. But debunkers have shown neither, and so they have not given realists any good reason to accept the great cost of giving up our fundamental understanding of why we believe what we do about what we take to be objective moral matters. If that is the price for the greater parsimony they’re selling, then I for one am not buying.

3.4 A specifically evolutionary debunking argument

Finally, let me turn to a debunking argument (based most closely on Street 2006, 2008) that begins with an explicit appeal to evolutionary influences on the content of our moral beliefs and does not strictly rely on either 1 or 1′:

  1. (1*)

    “Evolutionary forces have played a tremendous role in shaping the content of human evaluative attitudes,” so that “our system of evaluative judgments is thoroughly saturated with evolutionary influence,” through the pervasive shaping of our psychological dispositions by natural selection in ancestral environments (Street 2006, pp. 109, 114).

  2. (2*)

    This shaping would have been based simply on what promoted differential reproductive success by ancestral humans, rather than on successful tracking of independent moral truths.

  3. (3*)

    It would therefore be an implausibly lucky coincidence if the large subset of our moral beliefs that reflect evolutionary influence (which is the bulk of them) turned out to represent independent moral truths accurately.

  4. (4*)

    And even if we also have some true moral beliefs arrived at independently of evolutionary influences and due instead to our grasping of moral truths as such (as the realist might claim), they’re swamped by beliefs shaped by evolutionary influences and we have no way of sorting them out, given our reliance on some moral beliefs in critically evaluating others.

  5. (5*)

    Thus, if correctness for moral beliefs consists in accurately representing independent moral truths, as the realist claims, then at least once we become aware of 3* and 4*, and in the absence of any independent confirmation of the truth of our moral beliefs, whatever default justification we may have had for our moral beliefs is thereby defeated.

  6. (6*)

    Therefore, if the realist is right in holding that correctness for moral beliefs consists in accurately representing independent moral truths, then we’ve lost our justification for our moral beliefs once we’ve become aware of 3* and 4*, lacking any independent means of confirming the truth of our moral beliefs: realism saddles us with skepticism.Footnote 18

This argument needn’t deny that any of our moral beliefs are due to our having grasped independent moral truths as such, or insist that all of our moral beliefs can be sufficiently explained without appeal at any point to such truths. Still, it remains deeply problematic. To begin with, we once again have no compelling reason to accept the very first premise, 1*, at least in any form strong enough to do the needed work in a debunking argument.

Even if we grant that evolution gave our ancestors dispositions that influenced the content of their judgments, nothing follows about how deeply or widely this influence pervades our current moral beliefs. As we’ve seen, unless we have already rejected moral realism, it remains an open possibility that at least many of our moral beliefs—like many of our beliefs in other domains—have been arrived at quite independently of evolutionary influences. Many of our moral beliefs may spring from cumulative, ongoing experience of life and value, and on critical reflection on this experience to discover moral truths. We may have come, for example, through a wide variety of emotionally laden human interactions informed by decent moral training, to grasp the dignity of persons and the tragedy of their being deprived of basic human capabilities, and thus come to understand what is wrong with the oppressive practices mentioned earlier, and so to have true moral beliefs about them. The sciences cannot tell us how many of our beliefs fall into this category, how many are overdetermined, and how many merely reflect blind evolutionary influences. Everything depends on whether or to what extent we’ve been able to engage with independent values in the course of our lives, informing our moral beliefs with that ongoing evaluative experience—a matter about which the sciences are silent. If we have done so, then there is no garbage in / garbage out problem: we needn’t accept Street’s claim of pervasive evolutionary influence on our entire fund of moral intuitions, or anything even close to that (Street 2006, p. 124).Footnote 19

Again, it is not my purpose here to try to develop a positive realist account of how this works: there could be many different versions depending on the details of the realist view. For the moment what matters is just that nothing about evolution shows that no plausible account can be given: it remains an open possibility that many of our moral beliefs have been guided by our apprehension of real values through informed ethical experience and reflection, as I’ve suggested, so that at most we have a challenge to realists to provide an account of how this works. This is far from a debunking and is really nothing new, except insofar as there is the further constraint mentioned earlier to ensure any positive account is consistent with what we know about the raw materials we have had to work with thanks to evolution.Footnote 20

Realists should therefore reject 1*, just as we did 1 and 1′. This is not to deny that there were significant evolutionary influences on the moral belief-forming dispositions of Pleistocene humans, or even that some such influences have affected some current moral beliefs to some extent. That much should be granted, and to the extent that natural selection did shape some of our current moral beliefs, it will incidentally have pushed some of them in the right direction and others in the wrong direction. This will involve a certain amount of distortion (e.g., tendencies toward tribalism and xenophobia), as well as a certain amount of incidental positive reinforcement (e.g., tendencies to care for our families and friends). But the distortion here is just another example of an old problem: we’ve always known, given the amount of moral disagreement, that in addition to whatever true moral beliefs we have, there’s plenty of ‘noise’—false moral beliefs attributable to all sorts of cultural, sociological and psychological factors. To be told that some of this noise is also due to evolutionary influences is interesting but adds nothing really new to the old problem of how we can have justified moral beliefs despite all the noise.

There would, of course, be a special problem if we granted that the evolutionary distortion has been overwhelming. But again, we needn’t do so, both because (i) the science doesn’t tell us how pervasive the evolutionary influence on our current beliefs is, and (ii) even where there is such influence we have no reason to suppose it’s vastly more incidentally distorting than it is incidentally supporting, for reasons already brought out in the response to Joyce.Footnote 21 Perhaps debunkers think there’s something about evolutionary noise that causes special problems: evolutionary influence on our moral beliefs is somehow so distorting and difficult to expose that it undermines our ability to develop and employ reflective techniques to home in on independent moral truths—as claimed in 4*. That worry, however, is unfounded, because the situation is actually the opposite. When it comes to evolutionary distortion, the very same Darwinian considerations that raise these worries equally tell us what to be on the lookout for.Footnote 22 We might expect, for example, an overemphasis on loyalty toward one’s in-group and underemphasis on duties toward distant strangers; we might look for xenophobia, gender inequalities, obsession with ‘purity’, male control of female sexuality, hierarchical social structures, or excessive conformity to local norms and traditions.

Moral beliefs along these lines can thus be flagged as potentially reflecting evolutionary distortion and thus subjected to particular scrutiny, exactly as reflective people have always done in the past, whatever the source of the distortion. If such beliefs fail to stand up to critical scrutiny in open, informed moral discourse, we may reject them and thereby make progress. Awareness of evolution may thus give some people reason to rethink their commitment to blind loyalty to their clan or ethnic group, for example, in combination with moral doubts raised by informed scrutiny. But nothing about evolution gives us any serious reason to question our knowledge that it is morally wrong to torment people with disabilities for amusement, or to do nothing to address climate change—even if we take these to be real, independent moral truths.

4 Conclusion

The primary lesson to take away from these debates is that as realists develop a positive moral epistemology, they are constrained by the need to take seriously the fact that natural selection, operating on principles having nothing to do with independent moral truths, gave us the raw psychological materials that we are somehow able now to develop, train, and deploy to arrive at moral knowledge. As I’ve emphasized, this is not different in principle from what we find in other domains. But depending on the details of the proposed realist moral epistemology there may be some genuine challenges here, and this should be acknowledged.

There will be no special difficulty if ethical conclusions can all be derived simply through intelligent reflection and reasoning, as with mathematical facts, or through this together with empirical information, as with scientific facts. But if we think, as I myself do, that emotions play a significant epistemic role in grasping ethical facts, then we have a further theoretical commitment: we must hold that natural selection happened also to provide us with the specific emotional raw materials necessary in order for us to be able to train our emotional dispositions, in cultural contexts, to become appropriately sensitive to independent moral properties and facts as such. If proper appreciation of human dignity, for example, requires the right kinds of emotional development and training, rather than just familiar extensions of reasoning as in mathematics, then our ability to gain moral knowledge related to human dignity depends on our having been given the emotional potentialities and dispositions that were necessary for us to be able to develop characters capable of engaging with such moral properties as dignity. And the challenge some will no doubt raise here is that this may seem just ‘too convenient’ to be plausible. Why should natural selection operating on hunter gatherers, according to genetically oriented principles indifferent to objective values as such, have given us emotional raw materials that turn out to be capable of being developed in ways that put us in touch with independent values of the sort posited by realists (especially in the case of realists who take these values to be non-natural)? Aren’t we right back to the problem of implausible coincidence?

This is an interesting challenge that needs to be taken seriously, and it has to do specifically with our evolutionary background rather than being just an evolutionary spin on more general challenges. Though I cannot spell this out here, I believe this challenge can be met, partly along the same lines as the response to Joyce. It is also important to emphasize both (i) how much work is done by our deliberate shaping and training of our emotional capacities in the course of decent moral upbringing (so that evolution needn’t have given us anything close to reliable dispositions from the start) and (ii) the fact that had evolution given us very different basic emotional potentialities the moral facts themselves would equally have been relevantly different, so that there is no ‘miraculous coincidence’ that evolution gave us just the right capacities needed to appreciate independent moral facts that would have obtained even if we were emotionally very different creatures from what we are.Footnote 23 In any case, the legitimate challenge that remains does not support the antirealist conclusion “that things are good, valuable, and required ultimately because we take them to be” (Street 2008, p. 225), and neither do any of the allegedly debunking arguments we have examined