The Scorn of Humpty Dumpty (Part 2)

In a previous post, I discussed Huemer’s claim that there are only three possible antirealist positions: subjectivism, error theory, and noncognitivism. I stated that this is false, because all three positions presuppose a uniform and determinate account of the meaning of moral claims. If we abandon this assumption (the UD assumption, as Gill calls it), we open the possibility to additional antirealist positions that reject the UD assumption.

In particular, there are at least two (not mutually exclusive) possibilities:

Metaethical pluralism:

There is ineliminable variation in what people mean when they make moral claims. Some portions of moral discourse fit with cognitivism, others noncognitivism, some with realism, others with individual subjectivism or cultural relativism and so on. There simply is no single, shared metaethical standard. It may also be that in some cases people are committed to two or more conflicting metaethical positions simultaneously. When this occurs, their claims may be incoherent, giving rise to a novel form of antirealism proposed by Loeb, moral incoherentism.

Metaethical indeterminacy:

Some (perhaps most) ordinary moral claims don’t determinately commit speakers to any particular metaethical position. Moral claims could be like causal claims. Claims such as “It is going to rain tomorrow” do not require any commitments with respect to how to interpret quantum mechanics. Just the same, moral claims may not require any commitments about whether there are or are not stance-independent moral facts, and may not commit speakers one way or the other to any particular metaethical position. Perhaps these concepts are philosophical inventions philosophers mistakenly project onto folk thinking, imagining that the figure into the way people speak and think even if they don’t.

Both views raise questions about what, exactly, we are attempting to describe when accounting for “the meaning” of moral claims. Are we describing the mental states (such as the intentions) of the people making moral claims? That is, when a person says “murder is wrong,” and we hypothesize that they are expressing a propositional claim, are we claiming that the person intends to make a propositional claim? Or are we offering an externally adequate explanation for the linguistic practices of that person and other members of their linguistic community?

I call the distinction between the former, more explicit account of the mental states of speakers their philosophical stance and the latter, externalist description of their practices their commitments. This distinction roughly parallels a similar distinction Walter Sinnott-Armstrong discusses in an article that addresses Gill and Loeb’s suggestion that moral language may be far messier and more variable than commonly appreciated:

There are two ways to describe moral language. An internal project seeks to capture the psychological processes or representations that actually occur when people use moral language. However, contemporary realists and expressivists are not trying to do that. When Jackson and Pettit use networks of truisms or when Gibbard cites hyperstates, they surely know that these theoretical constructions do not reflect actual psychological entities or events. Instead, they want their theories to be externally adequate in capturing the outputs of our linguistic systems without necessarily reflecting the internal workings of that system. In this respect, their project is more like Chomskian grammar, which uses constructs without claiming psychological reality. (Sinnott-Armstrong, 2009, p. 237)

Stances and commitments approximate the internal and external projects, respectively. It’s not altogether clear what traditional antirealist positions are intended to reflect, but my general impression is that they usually taken to represent the external project moreso than the internal project. This is Sinnott-Armstrong’s impression as well. He follows this paragraph by remarking that:

Overall, then, I take moral realism and expressivism to be trying to externally describe the semantics of all standard moral language. At least, that is the debate that I want to discuss. (Sinnott-Armstrong, 2009, p. 237)

Note that this remark is made in an essay dedicated specifically to discussing Gill and Loeb. Given that all of these authors are at the center of 20th century metaethics, it would be at least a little surprising if this emphasis on an external account were wildly inaccurate.

Yet the distinction raises questions about what, exactly, an external account amounts to. If we’re not describing the mental states of the speakers, but we’re still trying to determine what their words “mean,” it seems to me as though we’re operating under the assumption that words and sentences mean things independent of the intentions of the speakers. While this may be a mainstream view among philosophers, I cannot make sense of it.

And if this is the approach we take, we might wonder whether pluralism and indeterminacy make much sense. One might think that this just isn’t how language works. Words and phrases have specific meanings (setting aside concerns about polysemous words, where a single word may have multiple distinct meanings, each of those distinct meanings doesn’t suffer from significant internal variation as well). At the very least, there must be some some central, canonical usage of a term, and other uses are at best elliptical or parasitic on the canonical use of the terms (see Gill, 2009; this depiction is largely put in terms similar to how Gill discusses the matter). Furthermore, it may seem obvious to philosophers that the range of available options for such canonical usages must conform to one of the traditional metaethical theses on offer.

If the proponents of these analyses claim that their positions are not empirical accounts of what people engaged in moral discourse mean, but rather what the terms and phrases mean independent of speaker intent, then we have to take a step back and examine a more fundamental disagreement about language and meaning.

Suppose we reject the notion that words and phrases have “meanings,” and instead suppose that what a given term (e.g., “good”) means depends on what speakers intend to mean, then the central semantic theses that cognitivism, subjectivism, and error theory may be better construed as empirical hypotheses about what people mean when they engage in moral discourse. A growing body of empirical research - research on the psychology of metaethics - studies this question by empirically investigating what nonphilosophers mean when they make moral claims, or what they take themselves to mean. There are numerous methodological limitations to these studies, but the results, so far, overwhelmingly point to metaethical pluralism and (I argue) indeterminacy. So if the central semantic theses of traditional antirealist positions are empirical claims, I take each of these theories to be false, or at least incomplete. If this isn’t what they are trying to do, then I think they rely on views in the philosophy of language that I reject, and thus reject any positions that rely on them.

This brings us to a more foundational problem with the three traditional antirealist positions. The deeper problem with Huemer’s claim is that it seems to rely on presuppositions about moral claims (and presumably language more generally) that philosophers are free to challenge, and some (myself included) do so. While these philosophers may be mistaken about their views, Huemer cannot dismiss such positions by declarative fiat. Rather, arguments would be needed to make a case that we’re left with just the three views Huemer mentions because the alternatives I’ve proposed are not possible. If Huemer has made that case somewhere, I’d like to see it (and I don’t mean this rhetorically to imply he hasn’t made such a case: I haven’t read most of his work, so maybe he’s addressed this somewhere).

Either way, Huemer is mistaken that the three positions he outlines are the only possible antirealist positions, unless Huemer could show that for reasons that fall outside of metaethics itself, e.g., considerations in the philosophy of language, conceptions of language like mine are likewise not possible. I find it unlikely Huemer has any kind of knock-down argument against my views on language, or the views of e.g., Wittengenstein or Ordinary Language Philosophers. From what I’ve seen, most such criticisms question-beggingly presuppose the very conceptions of language that I reject. For discussion on this, see Baz (2012).

My general impression is that this bizarre view of language, that words themselves mean things, and that sentences like “murder is wrong” have some kind of meaning independent of speaker intent, has entrenched itself into analytic philosophy as a kind of dogma that, while challenged on occasion, is never challenged with enough persistence or in sufficient numbers to overturn it.

In the next part in this series, we will consider Huemer’s remarks alongside his claim that there are only three possible antirealist positions I will show how, at least at first glance, it does appear that Huemer subscribes to the odd view of language wherein utterances have meanings quite independent of the intentions of the speaker, though at times his remarks suggest otherwise, as well. I suspect there is a genuine inconsistency, or slippage in Huemer’s views on the matter, but it’s hard to know. My hope is that Huemer will eventually see these posts, and that we can spark up a discussion about it. I’m extremely curious to develop a better understanding of our metaphilosophical differences, since I suspect at least some of my objections to Huemer would be resolved by a better understanding of where we differ in terms of our general method and approach to philosophy. I am also sure that Huemer clarifies his position elsewhere and I simply haven’t read it.