Maybe Evolutionary Debunking Arguments Don't Actually Work At All

Looking at a fascinating argument from Louise Hanson

Pictured: Donald Trump, a man known for prompting deep philosophical reflection and clarity of thought

Back in January, I published a piece on evolutionary challenges to moral realism and why I think they ultimately fail. In that piece, I called these sorts of arguments (which I’ll refer to as EDAs from now on) some of the most serious objections to realism out there — however, after recently stumbling across a brilliant paper by Louise Hanson from back in 2016, I think I’ve changed my mind. The paper, which you can read here, raises a fatal issue with the basic structure of an EDA that I find very compelling; in fact, I don’t really get how I could have missed it until now.

Everyone who’s interested at all in the topic of EDAs should definitely give the whole paper a thorough reading, but just to help introduce Hanson’s basic point to a broader audience here on Substack, I thought it would be fun to write a short overview of the critique she lays out. I’ll be quoting her at length in a few spots, but otherwise you should take all this as a rough paraphrase — if there’s anything that you think sounds a little off, I’m sure it’s just me glossing over some necessary details. Still, I hope what follows can at least give realists (and others on the fence) a solid starting point for pushing back on a particularly widespread and influential anti-realist objection.

---

As I wrote in my original piece, there are multiple different “versions” of the EDA out there, and each one is going to emphasize something slightly different. Some target the possibility of moral knowledge or the likelihood that our moral beliefs would be true, while others frame things more in terms of explanatory power and parsimony. Still, each version will almost certainly start out with two basic premises:

1. Natural selection is responsible for the moral judgments that we make.

2. Natural selection is not a truth-tracking process.

Depending on the specific aspect of moral realism the argument is seeking to target, both the next few premises and the ultimate conclusion might differ. But either way, the basic worry is pretty simple: If our moral judgments are, in fact, explained by a process that isn’t concerned with what’s actually true or false, then we have no good reason to think those judgments reflect any reliable mind-independent truth, nor do we have any reason to think those mind-independent truths actually played any role in their formation. And it’s easy to see why a conclusion like this would be bad news for the realist, since a moral theory isn’t particularly appealing if it can’t explain why we should have any confidence in the truth of our moral judgments or the explanatory importance of the moral facts that they supposedly relate to.

So how does Hanson’s rebuttal work? Ultimately, it all boils down to pointing out a simple but easily overlooked ambiguity in the claim that natural selection can explain our moral judgments. Her basic contention is that, understood one way, the claim is obviously true but mostly irrelevant; meanwhile, if it’s understood in a way that would actually undercut our confidence in realism, then it’s almost certainly false. Hanson refers to these two distinct interpretations, respectively, as the quantificational reading and the predicative reading.

The distinction between these two interpretations, and how that distinction ends up impacting EDAs, can be a little tricky to grasp in the abstract — but luckily, we’ve got Donald Trump here to help us make things a little more concrete. So let’s all imagine, for a brief (but painful) moment, that we’ve been assigned as journalists to cover an invite-only midterm GOP rally where he’s going to be the headline speaker. Our boss, desperate for yet another article psychoanalyzing MAGA voters, tasks us with getting an answer to one fundamental question: Why does everyone in the audience think Donald Trump is a great president?

In this case, it should be pretty easy to see that there are two distinct ways you could interpret what we’re being asked to figure out. We could understand the question, on one hand, as something like, “Why does this specific audience consist of people who believe Donald Trump is a great president, rather than consisting of people with other political views?” This is the quantificational reading of the question, and if that’s really all we want to know, then the explanation should be pretty obvious: The reason people in the audience think Donald Trump is a great president is because anyone who felt differently wouldn’t have been invited.

If, on the other hand, we’re interpreting the question in a predicative sense, then what we want is an explanation for why any particular person in the audience has the sort of belief that would get them invited in the first place — which is to say, we want to know why that guy specifically thinks Trump is a great president, and that guy, and that guy, and so on, for every individual in attendance. Here too, there are plenty of obvious answers, none of which would be particularly flattering for the people involved. (The phrase “ghoulish moron” might make an appearance once or twice.) All that matters here, though, is that these two sorts of explanations are very different, even if both can be seen as explaining the same basic state of affairs.

In short: A quantificational explanation for some judgment being widespread within a group will tell us why the group is made up of people who tend to share that judgment, whereas the predicative explanation will tell us why the people in the group do, in fact, tend to share it. The ambiguity is that, if I just say “The crowd has the beliefs they do because of Donald Trump’s campaign managers,” it’s not clear whether I’m just saying that Trump’s campaign managers selected the crowd from a preexisting population of supporters, or if I’m making the much stronger claim that his campaign managers are directly responsible for the fact that those people are Trump supporters in the first place. To resolve that ambiguity, I’d have to clarify whether my claim was meant in a quantificational or predicative sense.

So what does this have to do with EDAs? As Hanson points out, a claim like “Natural selection explains why we make the moral judgments that we do” can also be read in a predicative or quantificational sense, with the two different readings giving two very different results. Quantificationally, no one (or at least no one worth taking seriously) would deny that many of our fundamental moral intuitions are fitness-enhancing, and that natural selection has played a major role in ensuring that those sorts of intuitions were passed along. So if we want to know why we exist, and not some other creatures who make very different sorts of judgments, then natural selection is a perfectly good explanation, or at least a big part of one.

On the other hand, reading the EDA’s central claim in a predicative sense commits us to a significantly stronger position: That natural selection explains why every human being, in an individual sense, makes one moral judgment instead of another. But why would anyone think this stronger claim is true? At the very least, it’s not generally the case that natural selection can explain traits in a predicative sense, since evolutionary pressures only act on traits that are already present. Quoting Hanson directly here:

When we talk about a particular trait common to members of a given species being the result of natural selection, we are saying, roughly, that individuals who had that trait or tendencies towards it had more offspring than individuals who didn’t, resulting, ultimately, in a situation where individuals with that trait existed rather than individuals who lack this trait. When we say, for example, that the long-necked-ness of giraffes is the result of natural selection, we are saying, roughly, that the ancestors of giraffes were such that, those among them who had slightly longer necks than the others tended to produce more offspring, and that this led to a situation where a greater and greater proportion of the creatures that were born had long necks, which eventually led to the long-necked giraffes that exist now existing instead of some other creatures with shorter necks. This is a claim about the process that led to there being individuals who had the trait in question, not a claim about the process that led to these individuals having this trait.

What’s more is that, even if we do want to say that natural selection can explain our individual moral judgments predicatively, it could only ever do so through a broadly quantificational process; as Hanson says, it’s not as if there’s some other way natural selection could influence our individual moral judgments, except by slowly pruning off any competing dispositions in our ancestors. So regardless of whether we end up calling that sort of process a predicative explanation or not, a more robust predicative reading — one that frames evolutionary pressures as literally “changing our minds” from one judgment to another — is pretty clearly off the table.¹

All this is to say, the best a defender of the EDA could plausibly claim is that natural selection has slowly filtered out human beings who happen to develop very different moral intuitions, until only the ones who share our moral judgments are left — not that natural selection is responsible, here and now, for the judgments being made by the ones who remain.

To see why this distinction is so devastating for the EDA, let’s imagine we’re back at that Trump rally one more time. As you sit there listening to him ramble, you start to develop the radical notion that — hear me out — people who think Trump is a great president might not have the most reliable political judgments. The question, then, is this: If we all try our best to explain why the crowd supports Trump, then what sort of explanation, predicative or quantificational, would give us good reason to think their judgments aren’t truth-tracking? Or, to put things another way: If all we can give is a quantificational explanation for why everyone in the crowd thinks Trump is a great president, is that enough by itself to show us why they shouldn’t be trusted?

Clearly, a robustly predicative explanation for their judgments absolutely could do the job. Let’s say, for example, that we do some interviews after the speech has ended and realize that 95% of the audience forms their political views solely by reading Trump’s posts on Truth Social. In that case, given the extremely controversial assumption that Trump’s posts aren’t always made with a rigorous devotion to fairness and accuracy, this sort of explanation would be a pretty solid reason to think the views of the crowd aren’t reliable: Their individual political judgments are explained by something that’s indifferent to actual political reality.

Things look very different, however, when it comes to the impact of a quantificational explanation. As we said before, the quantificational explanation for why the audience feels the way it does about Trump is just that the rally’s organizers only invited people they knew would be supportive. But this, by itself, obviously isn’t enough to show that the supporters who did get invited are wrong, since the organizers played no role in the supporters actually forming their individual judgments. Crucially, this is the case regardless of whatever moral or intellectual qualities the organizers have; even if they themselves operate with absolutely zero regard for the truth — which I assume would be the case for a Trump campaign manager, honestly — that’s no reason (by itself) to think the judgments that attendees previously formed would be similarly unreliable.

The same thing is true, Hanson argues, when it comes to the EDA: If the most we can say is that natural selection happened to preserve a species-wide disposition towards certain moral judgments once those judgments developed, then that by itself doesn’t tell us anything particularly meaningful about their overall reliability — even once we acknowledge, as we should, that natural selection isn’t a truth-tracking process. I’ll quote Hanson at length here:

Read quantificationally, [the EDA] doesn’t entail that individuals only have the moral beliefs they have because they are fitness-enhancing; it entails only that people who have these moral beliefs, whatever their reasons for believing them in fact are, have tended to live longer and have more offspring, so that over time, more and more people exist who have them… [A]ll this is entirely compatible with the picture that moral realists want to endorse — that moral facts are mind-independent features of reality that we have an ability to discern. Beyond the dispositions being heritable and the beliefs being fitness-enhancing, [the EDA’s] claim, in the quantificational sense, doesn’t say anything about what these dispositions that underlie our moral beliefs are. Many kinds of dispositions could play this role, and the [EDA] read quantificationally does nothing to rule out what the realist wants to say, which is that the disposition that plays this role is an ability to discern moral facts.

Or, to put it another way: Every moral realist believes that human beings end up with moral knowledge somehow, whether through intuition, or by empirical observation, or something else — but just because that knowledge ends up being advantageous in an evolutionary sense doesn’t mean the realist has to think those evolutionary pressures themselves are responsible for it. Ironically, this is equally true for the anti-realist who thinks our moral judgments reflect nothing but fact-free subjective stances; even if natural selection can explain why a propensity for those stances is widespread among human beings, it’s still the propensity natural selection preserved that properly explains those judgments, not whatever process was responsible for preserving them.

So many anti-realists miss this point, I think, because they’re implicitly assuming that the sort of moral knowledge realists are talking about doesn’t really exist. And if you grant that assumption, then the EDA will make perfect sense: Without the possibility that we ever actually learn moral truths, all natural selection would ever have to work with is a revolving collection of randomly arising whims — and since we all agree that natural selection isn’t truth-tracking, the chance of it somehow wringing out a bunch of true judgments from random dispositional fluctuations is basically zero.

However, this way of thinking very obviously begs the question against moral realism, which necessarily involves the belief that (at least some) widespread moral judgments really do come from making contact with objective, mind-independent moral facts. If you think that sort of picture is reasonable, then the truth of the moral judgments in question doesn’t need to come from natural selection — those judgments will already be justified by whatever independently truth-tracking ability the realist has in mind, and natural selection just explains why that ability became widespread over time.

Of course, this doesn’t mean that moral realists are totally off the hook when it comes to defending the possibility of moral knowledge in the first place; without a plausible account for how it is that we as individuals gain moral knowledge, and an explanation for why that process counts as legitimately truth-tracking, anti-realists are right to say that moral realism is dead in the water. But that sort of individual account will have nothing whatsoever to do with natural selection, and to the degree that it is plausible, the further fact that natural selection preserved the capacity being described won’t matter one way or another. To quote Hanson at length one more time:

Again, it is often objected that realists owe — and are unable to supply — a convincing positive story about how [moral knowledge is obtained]. But again, EDAs purport to do something different from that kind of objection: They claim to show that whatever view realists hold about how it is that we discern moral truths, evolutionary considerations show that we can’t be doing that thing. If I’m right, this argument fails to establish this. Even if moral knowledge is deeply implausible on the realist picture — as well it might be! — the claim that our moral beliefs were selected for doesn’t make it more implausible.

So, in closing: If you’re an anti-realist, and you’re confident that our moral judgments are best explained by a purely psychological account, then feel free to make your case! There are all sorts of legitimate reasons for doubting the realist picture, and it could always end up being the case that anti-realism does ultimately win out on the merits. But just be aware that, whatever the problems with realism might be, our evolutionary history very clearly isn’t one of them.

1. I sense this might be a major sticking point for some people, so if you’re skeptical, please make sure to check out section 4 of Hanson’s paper, where she works through the idea in more detail.