In his 2004 paper “Meta-ethics and the problem of creeping minimalism,” Jamie Dreier argued that quasi-realism and realism diverge in their explanations of normative content: realists appeal to normative facts and properties when explaining what it is to have a normative belief or to make a normative claim, while expressivist explanations appeal instead to desire-like attitudes.
As I noted in my previous post, this “explanation” explanation of the divide between quasi-realism and realism seems to fit well with the new understanding of expressivism as a metasemantic view: if expressivism offers an account of why normative claims have the meanings that they do, then the alleged explanatory contrast with realism seems to come clearly into view. For instance, someone who adopts Ridge's (2014) version of expressivism and deflationism about truth, facts, etc. might agree with realists that “You ought to donate a large part of your income to charity” is true only if a certain normative fact obtains, but will deny that the meaning of this normative sentence is explained by that fact. Their metasemantic explanation will appeal instead to normative perspectives, understood as practical stances.
Now, there are some problems with Dreier's proposal. As Chrisman (2008) argued, it seems to treat all false normative claims as non-representational, because their meanings cannot easily be explained by appeal to corresponding normative entities. More recently, Simpson (2018) has pointed out that Dreier's view fails to capture the contrast betwen expressivism and other representationalist metaethical views such as error theory, and is a poor fit for cognitivist views that do not rely on facts or properties in explanations of meaning, e.g. views that appeal to propositions in a fundamental role.
For these reasons, Simpson proposes that the explanatory contrast between quasi-realism and its rivals is best stated in terms of representation: realists and other theorists claim that representational relations play a substantive role in explanations of meaning, while quasi-realists deny this. (Here, representational relations include truth, reference, aboutness, as well as representation itself.) To be sure, quasi-realists can accept, in a deflationary framework, that normative claims are truth-apt or representational. But they deny that these features of normative discourse play any significant role in explanations of meaning.
Simpson also proposes that this debate should focus on how to define representationalism about normative discourse in a way that keeps it distinct from expressivism, where representationalism includes not only realism, but also anti-realist views like error theory. I take no issue with this, but I rush to note that any plausible account of what separates expressivism and representationalism should place standard forms of non-naturalist normative realism on the representationalist side of the divide. (This is important for my arguments.)
According to Simpson, then, all representationalists views accept, and expressivism rejects, something like the following thesis:
(R) Representational relations explain attributions of meaning to normative sentences.
However, in my paper I argue that, if we adopt a deflationary account of the relevant semantic notions, (R) does not provide us with an explanatory divide between quasi-realism and any plausible notion of normative realism. Here are my arguments in broad outline.
Deflationists (including quasi-realists) do reject a metasemantic picture on which general semantic notions like truth, reference, or representation play a substantive role in explaining meaning. But this is something that they reject for all domains of discourse, including ordinary descriptive discourse. This is why we cannot find here a plausible criterion for normative realism or representationalism: the deflationists’ rejection of (R) does not allow us to draw an explanatory contrast between domains of discourse where realism holds true, e.g. our claims about tables and chairs, and areas that might be amenable to a quasi-realist account, like normative discourse. Any good answer to the problem of creeping minimalism should allow us to draw such a distinction.
At this point, someone might object that quasi-realists do accept a certain kind of representationalist explanation of meaning in some domains of discourse, namely a causal explanation that appeals to tracking relations between language and the world to explain the emergence and use of certain bits of language, e.g. our terms for tables and chairs. Moreover, expressivists reject such tracking explanations of meaning when it comes to normative expressions. Perhaps this is the sense in which expressivism is still a form of anti-realism?
I argue that tracking explanations of meaning cannot be used to draw a divide between quasi-realism and realism either. This is because standard forms of non-naturalist realism, which should be covered by any plausible criterion for normative realism, reject such tracking explanations as well, given that they take normative facts and properties to be causally inert. (I should note again that Simpson's goal is to articulate what separates expressivism from representationalism more generally, rather than just realism. But insofar as representationalism is meant to include all standard forms of realism, the objection I raise here still applies.)
Based on these arguments, I conclude that metasemantic expressivists who endorse deflationism about truth, representation, etc. should think of their view as vindicating normative realism pure and simple.