Knowledge, Reality, and Value: Huemer’s Response, Part 4

I’m delighted to receive comments from Huemer! Thanks so much!

I have a lot to say about these comments. I will have to work many of these thoughts into future articles, so stay tuned for more formal critiques of Huemer’s stance on ethics.

I’m not sure what we’re talking about.

I’d love to get clear on what we’re talking about. In your discussion of moral statements, you offer an objection to noncognitivism. But I don’t know which moral statements you are talking about. When I talk about moral statements, I am talking about the moral statements that ordinary people make in everyday circumstances.

If the moral statements of everyday people are what you’re talking about, then why does your work, and the work of most other moral realists, rarely address actual instances of moral thought, language, and behavior? Why do you instead focus on hypothetical instances of moral discourse that are intended to reflect the real thing?

If I wanted to study religious belief, I wouldn’t think up sentences like “God exists” and then try to assess their meaning. I’d interview religious people or conduct surveys to figure out what they believe. If you want to study moral utterances, why not do the same?

Why should I think that armchair inferences about whatever snippets of imagined moral discourse philosophers consider provide decisive insights into what people actually think and mean when they engage in moral thought and deliberation? This exposes you to an enormous range of biases, and restricts your considerations only to moral utterances you are familiar with. Without empirical data, you are limited to a culturally and linguistically narrow set of personal experiences. Would people from different backgrounds share the same intuitions as you? Do they reach the same conclusions when they introspect? Do their languages operate in the same way? We can’t know the answer to any of these questions by just thinking about it. We need empirical data.

And if you are not trying to extrapolate from examples of moral utterances that happen to occur to you to what people actually think and mean in the real world, then I am not sure what you’re talking about. I do know what I am talking about: actual moral thought, language, and behavior, as it occurs in the real world.

Note that Caplan appears to interpret the dispute between cognitivism and noncognitivism to concern what ordinary people mean. He follows his quote of your remarks about cognitivism and noncognitivism by talking about what “human beings” mean when they say things.

Unfortunately for philosophers, what people mean in such real-world instances of moral decision-making is an empirical question that cannot be settled through a priori reasoning alone. Even so, I do not think we need to conduct studies to determine whether the sentences in your premises make sense—they do. But I do not think that because the particular list you provide includes both sentences that make sense and sentences that do not that this provides much linguistic evidence either for or against cognitivism. If you believe it does, I’d welcome an explanation as to how. The premises you entertain tell us little about what people actually mean when they engage in moral discourse, and they certainly do not tell us whether cognitivism is true (they are also consistent with error theory and subjectivism, though I am not sure whether you think these sentences weigh against those positions. I doubt any armchair consideration of sentences would do so, though).

All that your example sentences highlight is that a cognitivist analysis of moral claims sound natural to you and I. Yet this is already something any reasonable noncognitivst would agree with. Noncognitivists are (or should be) happy to grant that the surface structure of many moral utterances appears to be cognitivistic. But simply because the semantics of an expression superficially looks propositional does not mean that the speaker is actually making a propositional claim. There are all manner of non-propositional utterances that use the same assertoric structure but plausibly fail to express propositions:

1. “My neighbor is a jerk.”
2. “That food was terrible.”
3. “Tomorrow is going to suck.”
4. “I hate broccoli.”
5. “She is the worst person ever.”

All of these utterances make sense on a cognitivist reading. That is, a person making these claims could intend to make a truth-apt claim (perhaps a subjectivist reading is most appropriate for each of these, for instance). And the relevant terms could figure into formal arguments or various kinds of reasoning (e.g., “if my neighbor is a jerk, and all jerks are liars, then my neighbor is a liar”). The mere fact that the word “jerk” can figure into an argument does not show that when I say someone’s a jerk that I must be expressing a proposition. If you really want to know what people mean, you could start by asking them.

However, these sentences also make sense as noncognitivist utterances conveying the non-propositional attitudes of the speaker. It does not follow that, simply because a cognitivist interpretation of these utterances “makes sense”, actual instances of these sentences in the wild, so to speak, typically express propositional claims, or that the central or primary function of such utterances is to make assertions. I, for one, do use language such as “I hate broccoli” to convey my attitudes, despite its superficially assertoric structure, and I would bet that many others do, too.

Just the same, simply because moral utterances can be articulated in a way that makes sense on a cognitivist interpretation does not mean that ordinary people actually intend to make propositional claims when they engage in moral discourse, or that they are implicitly committed to cognitivism when they make moral claims, or that they regard themselves as making propositional claims. Nor does it show that the primary or typical function of moral sentences is to express propositional claims. Finally, it does not show that we can disregard non-propositional moral utterances as aberrant or noncentral (or otherwise parasitic) on some supposedly central cognitivistic usage. In other words, analysis of your sentences cannot rule out noncognitivism.

I don’t know what sentences (4) and (5) are meant to convey. Are these supposed to show that noncognitivism does not offer a plausible account of the meaning of the relevant kinds of ordinary assertions? It seems like (4) reflects emotivism and (5) reflects prescriptivism.

Of course neither of those sentences makes sense. But neither emotivism nor prescriptivism commit one to thinking either of these sentences must make sense. Creating nonsense sentences like “John thinks that ouch” does not show that seemingly-assertoric claims about pain are propositional. So while “John thinks that ouch” does not make sense, this does:

John: “That is painful!”

This utterance is fully consistent with emotivism. And this is true, in spite of the fact that if you wanted, you could plop statements about pain into sentences and reason about it: e.g. “If it is always wrong to cause pain, then it is wrong for Alex to cause Sam pain.” That such sentences make sense does not entail that John must have been making a propositional claim after all, and that an emotivist analysis of John’s utterance does not make sense or doesn’t accurately describe what John meant.

Likewise, “I wonder whether please pass the salt” does not make sense, but what does that show? All it shows is that imperatival language does not fit in the unconventional sentence structure you created. What it does not show is that when ordinary people actually use moral language, that one of its central or primary functions is not to express prescriptions. We could imagine occurrences like this:

Alex: “I think I might sneak a cheat sheet into my final exam.”

Sam: “No! You shouldn’t do that! That would be wrong!”

A perfectly sensible translation of what Sam means is “Don’t do that!” And Sam might agree if we asked (in fact, plenty of native English speakers do favor noncognitivist interpretations of moral issues, when asked; more on this below). Why should we think Sam is confused about what she meant? Maybe we should, but the cognitivist is going to have much more of a story to tell than merely pointing at the surface structure of some moral utterances.

Another problem with your sentences is that they are presented outside of any social context. Your analysis of the meaning of these sentences exclusively stresses semantics, but ignores the role of pragmatics and the social contexts in which actual moral and nonmoral discourse occurs. Yet this is critical to determining the meaning of moral (and nonmoral) utterances, since the actual meaning of a sentence can (and often does) turn on pragmatic considerations relevant to the context in which it is uttered.

Imagine insisting that the concept of “sarcasm” made no sense, by focusing exclusively on a literal interpretation of sarcastic remarks. This would be absurd. Just the same, it is absurd to ignore the possibility that much of the meaning, purpose, and function of moral language is carried by the sociofunctional role that moral utterances play in actual social circumstances, since the meaning these utterances carry in such circumstances would be largely conveyed by pragmatics rather than semantics. And again, I must stress: the only way to know about these sociofunctional roles is to study real human moral practices, i.e., you need to conduct empirical research.

The bottom line is that it is a simple empirical possibility that, no matter how sophisticated a philosopher’s armchair critiques of noncognitivism might be, it could turn out that people use sentences like “Murder is wrong” to express commands. Indeed, they could in principle use sentences like “two plus two equals four” or “It is a propositional fact that the sky is blue” to express an emotion or a prescription. The only way to know would be to actually study what people are doing when they say things like this. All the sophisticated analyses metaethicists are writing up in journals are irrelevant if they simply do not square with how people actually think and speak about morality.

This is why it is a mistake to not recognize that the meaning of ordinary moral discourse is, at its heart, an empirical matter. This is one of (if not the) central reason why your armchair method is flawed: by considering the toy sentences you present to me, and other allegedly paradigmatic instances of moral claims, you abstract these remarks away from the real-world contexts in which they occur, in effect stripping away the rich contextual factors that would flesh out the pragmatic elements of these utterances, and, incidentally, stripping them of all context that would render noncognitivism more plausible. The situation is even more dire than this. Moral philosophers tend to consider textbook instances of moral utterances. Yet these examples were not gathered in a systematic and representative way. How well do the sentences you entertain reflect the typical contexts in which moral judgments occur? To provide just one example: philosophers focus a lot on assessing moral actions and principles, but actual moral thought and language may be far more centered on assessing moral character. In this, and many other ways, moral philosophy may fail to reflect everyday moral thought and practice.

In addition to making such broad assumptions on the basis of (what you take to be) moral utterances without any context, you impose on them philosophical standards that are in the particular interests of philosophers , but not necessarily those of ordinary people (e.g. a need for internal consistency). You then presume that these utterances must share some set of metanormative characteristics, e.g. that they express assertions. But again, the moral sentences you are attempting to analyze aren’t real moral sentences. They are proxies for the real thing. And these proxies are analyzed in contexts that don’t reflect the actual circumstances in which moral thought, language, and practice take place, if they reflect any circumstances at all.

Ordinary people are not just going around saying things that they take to be true. Even when someone does express a proposition, doing so serves a purpose. Aside from philosophers, who are incredibly weird, nobody sits in their room saying “Murder is wrong” aloud, intending to convey that it is true. Most people don’t spend much time thinking about the concept of murder at all. Rather, they think about instances of murder: cases in the news, people they know or hear about, and so on.

When a person actually says something about “murder,” they aren’t typically doing the same thing you as a philosopher are doing when you entertain a decontextualized sentence like “murder is wrong,” which floats in some philosophical void. And when they entertain actual instances of murder, and talk about them, it is precisely the real-world context and concrete nature of these considerations that would lend credence to the possibility that their utterances are primarily intended to convey nonpropositional content. If emotivism captures a central feature of moral thought and language, this isn’t going to be obvious when philosophers at dinner parties consider the sentence “murder is wrong.” But it may very well capture what John means when he says “What an evil monster! I hope they catch him!” This is real moral language. “Murder is wrong” isn’t.

In short, the very act of engaging in philosophical reflection about the meaning of moral discourses over-intellectualizes moral utterances. Rather than consider an actual person’s remark about an actual murder that had personal meaning to them and about which they felt sufficiently motivated to say something, you instead consider abstract remarks about whether “murder is wrong,” as if this were the kind of utterance ordinary people engaged in.

In short, the approach you take doesn’t even involve the actual analysis of real moral sentences. That would require empirical data. Instead, it relies entirely on your simulation of the real thing. And given that you have a particular conclusion in mind, it would be surprising if the examples that occurred to you weren’t subject to motivated reasoning, the availability heuristic, and other cognitive biases that would lead you to think up examples that conveniently support your position.

If you studied moral psychology, you may recognize that the simulations you’re engaged in are a poor reflection of actual moral thought, language, and practice, though I grant that whether you’ve managed to successfully simulate actual moral sentences is an open question. Much of my follow-up to Alexander Davis (in that same thread) argues for why I think the prospects of success for this approach are grim.

Furthermore, when surveyed, many participants do favor noncognitivist interpretations of moral issues. In fact, in Taylor Davis’s (2021) studies, he found that the majority of participants favored noncognitivism. Beebe likewise found that many participants favored noncognitivist analyses of moral issues. While it is possible that all of these people were confused, or that these studies were flawed, the empirical evidence we have suggests that many people are perfectly happy to regard noncognitivism as an acceptable account of the meaning of ordinary moral sentences. That strikes me as at least some evidence against the notion that cognitivism is somehow correct or offers a superior analysis of ordinary moral language than does noncognitivism. Yet to my knowledge I have yet to see any moral realist seriously contend with this data.

In addition, you want to demonstrate that “They do not act like interjections (like “Ouch!”), commands (like “Pass the tequila”), or any other non-assertive sentences.”

You appear here to presuppose what Michael Gill (2009) refers to as the uniformity-determinacy assumption. That is, you assume that moral statements share a uniform and determinate set of metanormative properties, such that all genuine moral statements either do or do not fit with a particular metaethical analysis.

However, as my examples above illustrate, you may simply be incorrect. The mere fact that you can identify instances of moral language that do not operate as interjections, commands, etc. does not mean that there aren’t instances that do. Nor does it show that if they do, that such usage is noncentral, aberrant, or parasitic on cognitivistic usage. It may be that noncognitivist usages of moral utterances are so central to ordinary moral thought and language that they cannot be dismissed or eliminated in a principled way.

The premises you provide only suggest that moral language can be used in a way consistent with cognitivism and that it seems natural for us to use it this way in these cases. However, I have already provided examples of utterances where nonmoral claims can have the same structure and yet plausibly fail to express assertions. This demonstrates that such structure does not by itself establish that a given type of discourse is cognitivistic. People may also use moral language in ways consistent with noncognitivism, but the methods you have opted to use aren’t adequate for determining whether this is the case, or for how frequent and central such usage is to actual moral discourse.

This is why I maintain that you would need to conduct rigorous scientific studies to find out whether cognitivism or noncognitivism offer a better (or exclusive) analysis of the meaning of moral statements.

However, I do not think a study assessing whether the sentences you have provided would be a useful way to pursue this question. This isn’t because the dispute between cognitivism and noncognitivism isn’t an empirical one. It’s because answers to these questions are not a good way to determine whether cognitivism is true.

In other words, it does not follow that if you can think of a bad empirical study that isn’t worth conducting, that therefore the issue we are dealing with isn’t best resolved via empirical data. All it reveals is that the particular study you’ve alluded to wouldn’t be a good one.

I did not say most people are moral realists. My argument is about what’s true, not what most people believe.

My objection to the claim that people are moral realists was a remark about Bryan’s commentary, not about your position. Bryan stated that “[…] almost everyone thinks that moral reasoning, unlike sports fandom, is supposed to be a search for moral truth […].” I interpreted this as a remark about ordinary people being moral realists, since this strikes me as a plausible interpretation of “search for moral truth.” Of course, it could just be a remark about most people being cognitivists; given the context, that’s certainly possible. Either way, my comments included criticisms of your methods, as well as objections to Caplan’s extrapolation of your remarks to the conclusion that “most people” think morality is a “search for moral truth.” I was perhaps not as clear as I could have been in distinguishing where my criticisms were directed: at Caplan’s remarks about your position. My apologies for not making that more clear.

However, I do take you to be making claims that most people (that are competent with the relevant terms and concepts) speak or think as cognitivists. Much of my criticism applies to that presumption as well—that is, the presumption that intuition, introspection, and armchair analysis of linguistic acts are adequate to determine what people are actually doing when they engage in moral discourse in the real world. Once again, the only way to know what people actually mean is to conduct studies.

First, notice that this is a completely different response. Lance’s first response was “Huemer is just using intuitions, so his so-called ‘evidence’ is worthless.” This new response seems to grant that the non-cognitivist actually needs to accommodate (“handle”) my evidence.

No, it is not a completely different response. First, I neither said nor think that intuitions aren’t evidence. They are evidence, and they do need to be accounted for. Note one of my initial remarks:

“Huemer is not appealing to linguistic or introspective evidence in any systematic and rigorous way: he’s simply appealing to his intuitions about how language works, and his own introspection.”

Observe that I did not say that Huemer is appealing to intuitions, and that this is objectionable, full stop. I specifically said that you are only appealing to your intuitions, rather than systematically gathering evidence about how other people think about morality. I am, after all, suggesting you conduct empirical research on other people’s intuitions about the very same subject matter. If I thought intuitions were worthless, that would not make any sense. What I object to is that you appeal exclusively to your own intuitions, as though your personal way of seeing the world is especially probative for the rest of us.

Furthermore, I immediately followed up this comment by stating “For one thing, we cannot simply ignore people who do not share the impression that moral language is generally committed to realism.” In other words, my initial remark explicitly acknowledged that we have to consider the judgments of people with conflicting intuitions, and I appealed to my own intuitions!

My point isn’t that intuitions are worthless, but that people have different intuitions about the matter in question, so just appealing to your own intuitions is not sufficient (by itself) to resolve the matter in question. At best, your intuitions give you some defeasible reason to believe whatever it is you find intuitive. But geese and ganders. If your intuitions are defeasible evidence of realism (or cognitivism, or whatever), then my intuitions are defeasible evidence to the contrary.

The same holds for introspection. Note that I did not say introspection is worthless or refer to it as evidence in scare quotes:

“I don’t find the introspective evidence any stronger. The introspections he appeals to are his own, and perhaps a handful of people who agree with him. My introspections yield a completely antirealist picture of the world, as do the introspections of at least some other moral antirealists. The evidence here isn’t one-sided.”

Again, I take introspection to be evidence, given that I implied that my conflicting introspection resulted in the evidence not being one-sided. My problem isn’t with your appeals to introspection. It’s that my introspection yields conclusions contrary to yours. So why should I think your introspection is correct, and mine is incorrect? Why should any of us consider your way of thinking to be especially probative?
Second, I don’t agree that they “handle” those well. I don’t have time to write a lengthy discussion here, but I discussed this in Ethical Intuitionism.

Whether old accounts of noncognitivism handle particular sentences well isn’t going to settle the matter decisively in favor of cognitivism anyway, so this is moot.
Your third point is an odd one, since it seems irrelevant to me. Sure, lots of noncognitivists have spent unreasonable amounts of time trying to interpret various toy cases of moral language in ways that their own position can accommodate. This is futile. I’m not defending cognitivism. I’m suggesting all of these efforts are a titanic waste of time. What are they even doing? Who cares if someone can dream up an analysis of “moral sentences” where their examples are isolated, abstracted shadows of the real thing?

To reiterate, what matters is how moral language actually functions in everyday discourse, and what people actually mean when they make moral claims. No armchair analysis could possibly demonstrate that since, as a matter of fact, nobody actually uses moral sentences to express imperatives or only convey emotions. Nor can it demonstrate how frequent and central such instances are to everyday moral practice. These are empirical questions.

It would be very strange if you considered the findings of anthropologists, linguists, and psychologists who went out to study actual moral conversations to be irrelevant. If such researchers were to find that 100% of the time people who uttered moral statements appeared to be merely expressing emotions, and then when asked, said “Yes, I only intended to express an emotion”, would you consider this irrelevant to your theorizing about metaethics? To find it irrelevant would be very strange since, if this is what ordinary people meant, then emotivism offers a correct account of the meaning of ordinary moral claims.

Of course, I do not think this is true. My point is that what matters is the actual study of actual moral language, not whatever philosophers are using as proxies for actual moral language.

So, yes, I am suggesting that you would need to go out and conduct rigorous scientific studies.

Unfortunately for realists and cognitivists, those studies are already under way, and provide little support for either.

References

Davis, T. (2021). Beyond objectivism: new methods for studying metaethical intuitions. Philosophical Psychology, 34(1), 125-153.

Gill, M. B. (2009). Indeterminacy and variability in meta-ethics. Philosophical studies, 145(2), 215-234