In a new working paper, I develop a quasi-naturalist response to Eklund’s (2017) problem of alternative normative concepts. Here is a detailed summary of this paper.
The following scenario seems possible: a community uses concepts that play the same role in guiding individual actions and shaping social life as our normative concepts, and yet refer to something else. Here is the example that Eklund uses:
“Alternative. There is a linguistic community speaking a language much like English, except for the following differences … While their words “good,” “right,” and “ought” are associated with the same normative roles as our words “good,” “right,” and “ought,” their words aren’t coextensive with our “good,” “right,” and “ought.” So even if they are exactly right about what is “good” and “right” and what “ought” to be done, in their sense, and they seek to promote and to do what is “good” and “right” and what “ought” to be done in their sense, they do not seek to promote what is good and right and what ought to be done.” (Eklund 2017, 18)
If Alternative is possible, Eklund argues, this poses a problem for any normative realist who tries to vindicate the thought that reality itself favours certain ways of valuing and acting. This idea seems to be undermined by the alarming symmetry between our concepts and the ones employed by the imagined community: users of those alternative concepts seem to be getting things right just as much as we do. And any claim we might make about the privileged status of our concepts, e.g. that using those concepts is what we ought to do if we are to describe normative reality correctly, can be mirrored by members of the other community, using their own normative concepts and with equal justification.
This problem affects a wide range of realist views that seem to allow for concepts with the same normative role and different extensions, from naturalist theories that adopt a causal theory of reference for normative terms to standard versions of non-naturalist realism.
The most promising strategy in the face of this challenge is to avoid it in the first place, by arguing that there can be no alternative normative concepts. This means arguing that concepts that have the same normative role—that is, the same role in guiding individual deliberation and interpersonal criticism and advice—have the same reference as well. Eklund labels this thesis Referential Normativity (RN).
In this paper I argue that normative quasi-naturalism, a view that combines expressivism and naturalist realism about normativity, upholds (RN) and thereby avoids Eklund's challenge.
First, a clarification is needed. Alternative-type scenarios only pose a problem when they involve certain normative concepts, like the moral wrong or the all-things-considered ought—that is, authoritatively normative concepts. No realist should be troubled by scenarios in which, say, etiquette concepts have the same normative role but different extensions, given that such concepts are not plausibly amenable to a realist account in the first place. Realists, therefore, need to provide a metasemantic picture that supports (RN) for authoritatively normative concepts while allowing that the reference of other normative concepts can vary between communities. I argue that quasi-naturalism can accomplish this twofold task.
Normative quasi-naturalism has three main ingredients:
(1) expressivism about normative discourse;
(2) deflationism about certain notions in terms of which realism is usually stated, such as truth, fact, or description;
(3) naturalist realism about normativity, i.e. the view that normative properties are identical with, or fully constituted by, objective natural properties.
Expressivism is, broadly speaking, the view that normative claims express desire-like mental states. In this paper I rely on a hybrid version of expressivism, on which normative claims express both desire-like attitudes and corresponding representational beliefs. This view is well placed to account for the metaphysical claims of naturalist realism and can fit into a broader contextualist semantics for deontic and evaluative terms, a feature which plays an important role in properly addressing Eklund’s challenge.
Before bringing contextualism into the picture, I introduce a simple version of hybrid expressivism (Ridge 2007, Toppinen 2013) in order to show how this view can accommodate metaphysical naturalism about normativity. On this simple view, an atomic normative sentence like “Stealing is wrong” expresses (1) an attitude of disapproval of actions that have a certain property, and (2) the belief that stealing has that property.
Deflationism about truth, facts, and other related notions allows expressivists to endorse many tenets of realism. For instance, on a deflationary account, “It is true that p” is equivalent to p, and this equivalence schema fully captures the meaning of “true”. Thus, expressivists can hold that “It is true that stealing is wrong,” taking this claim to simply rehearse the first-order normative judgment that stealing is wrong. Similarly, expressivists can claim that there are normative facts, or that normative judgments describe such facts, by relying on deflationary accounts on these notions.
Moreover, expressivists can endorse claims about objective normative truths and facts, by taking such claims to express a particular kind of attitude. For instance, the claim that “It is an objective fact that stealing is wrong” will express (1) an attitude of disapproving actions that have a certain property, even when considering scenarios in which we ourselves did not disapprove of such actions, and (2) the belief that stealing has that property.
Now take the following naturalist thesis:
(A) Moral wrongness is identical with failing to maximize utility.
(A) can be understood as expressing (1) an attitude of disapproving action types that have a certain natural property, and (2) the belief that the natural property in question is identical with failing to maximize utility. More generally, this is how hybrid expressivists can accommodate metaphysical naturalism: by offloading the content of metaphysical claims about property identity or constitution onto the representational content of normative claims.
Here is now how quasi-naturalism can address Eklund’s challenge. First, we need to distinguish between Conceptual and Referential Normativity:
Conceptual Normativity (CN) Sameness of normative role entails sameness of concepts.
Referential Normativity (RN) Sameness of normative role entails sameness of reference.
If a view delivered (CN), this might seem to seem to enough to forestall Eklund's challenge, because (CN) seems to make Alternative-type scenarios impossible. However, this is too quick, because it still seems possible that different communities use the same normative concepts but with different extensions. Indeed, as mentioned before, this is precisely the type of scenario that we should expect when it comes to non-authoritatively normative concepts such as etiquette predicates. A proper defence of realism in the face of Eklund's challenge needs to distinguish such relativistic concepts from authoritatively normative concepts, for which both (CN) and (RN) must be upheld.
In order to achieve this goal, I propose that quasi-naturalists rely on a more sophisticated version of hybrid expressivism that fits into a broader contextualist semantics for deontic and evaluative terms. Ridge’s (2014) view is an example of this.
Ridge relies on a Kratzer-style contextualism for deontic and evaluative terms. On this model, two parameters determine the semantic content of terms like ought and good in particular contexts of use: (a) a modal base, i.e. a set of worlds in which certain background conditions are met; and (b) an ordering source that ranks worlds in the modal base, e.g. by how well they satisfy certain standards.
On this contextualist view, for instance, “You ought to donate a large part of your income to charity” will be true, in any context of utterance, just in case “You donate a large part of your income to charity” is true in all of the words in the modal base that are highly ranked according to the ordering source. The ordering source in each context will depend on the flavour of the given deontic or evaluative claim: in some contexts, the ordering source might consist in norms of etiquette, in other contexts worlds will be ranked in accordance with French law, and so on.
On Ridge's view, authoritatively normative claims are a subset of uses of deontic and evaluative terms, for which the ordering source consists in acceptable standards of practical reasoning. For example, “You ought to donate a large part of your income to charity”, when used in an authoritatively normative sense, means that any acceptable standard of practical reasoning would recommend that you donate a large part of your income to charity. (I should mention that Ridge simply uses the word normative for this category of claims, but I am restating his view in terms of authoritative normativity because I think we should allow that domains like etiquette and law are also normative in a weaker sense: they exhibit what is sometimes called formal normativity.)
Importantly, this contextualist semantics does not entail subjectivism or relativism about normativity. Normative realists can adopt this semantic model and claim that the ordering source in authoritatively normative contexts of use consists in objectively acceptable standards of practical reasoning, or some similar objective parameter.
Ridge also wants to make good on normative objectivity, but in an expressivist framework. More precisely, he proposes that to judge that a standard is acceptable is to endorse a normative perspective that does not rule out that standard, where normative perspectives are understood as noncognitive practical stances.
We arrive thus at a hybrid expressivist view on which any authoritatively normative claim expresses a normative perspective and a corresponding representational belief. To use the same example, “You ought to donate a large part of your income to charity,” if used in an authoritatively normative sense, expresses (1) a normative perspective, and (2) the belief that donating a large part of your income to charity is highly ranked by any acceptable standard of practical reasoning.
On this picture, both authoritatively and non-authoritatively normative concepts can exhibit (CN), but only the former are also referentially normative.
Take non-authoritatively normative concepts first. The contextualist analysis of claims about etiquette, for instance, will be purely descriptive: it will refer to what is highly ranked by certain locally accepted standards of behaviour. (To be clear, this is compatible with holding that etiquette claims pragmatically convey that the speaker has certain desire-like attitudes concerning the relevant standards, at least in typical circumstances. The claim here is that such attitudes are not part of the meaning of etiquette claims, nor do they help explain said meaning.) But judgments about what is highly ranked by locally accepted standards of behaviour also have a certain normative role, at least for the typical concept user: people will standardly rely on such judgments about social practices to guide their actions, criticize others, etc. Moreover, this normative role will be roughly the same for all communities using etiquette concepts. Thus, etiquette concepts arguably exhibit (CN). At the same time, specific standards of etiquette will vary between communities, and therefore the extension of etiquette concepts will vary as well.
For authoritatively normative concepts, in contrast, expressivism delivers both (CN) and (RN). (CN) is secured by the expressivist semantics. For example, someone like Ridge might argue that judgments about wrongness have a distinctive role in guiding deliberation and interpersonal criticism because they encode normative perspectives. So both we and the imagined community in Alternative use the same concept of wrongness in virtue of this shared noncognitive content that is constitutively linked with the normative role of “wrong”.
Expressivism also helps vindicate (RN) for authoritatively normative concepts, by allowing us to treat questions about the reference of terms like ought and wrong as internal normative questions, to which normative inquiry can establish unique answers that hold true no matter how those concepts are used by some community or other.
Again, expressivism is not a form of relativism, not even when built into a contextualist framework. Ridge's hybrid expressivism provides an account of what it is to think that a standard is acceptable, not an account of what makes standards acceptable. On this view, the truth conditions of normative claims do not make reference to our normative perspectives. They simply refer to acceptable standards.
Moreover, as mentioned before, expressivists can make sense of objective acceptability—and therefore of objective wrongness, what we objectively ought to do, etc.—by identifying a special kind of attitude expressed by objectivity claims: for instance, a normative perspective that condones or rules out certain standards of practical reasoning even with respect to scenarios in which we or others adopted different normative perspectives.
By adopting such resilient normative perspectives, we can treat questions about the reference of wrong or ought as having objective answers. To answer these questions is to identify the natural features that make an action wrong, or what we ought to do all things considered, no matter how any community uses such concepts.
This is how quasi-naturalism supports the conclusion that there are no alternative authoritatively normative concepts and therefore avoids Eklund’s challenge.
At the end I address two objections: (1) Eklund's worry that even views that adopt referential normativity might not escape the problem of alternative normative concepts, because concepts that have slightly different normative roles can still be in competition in a way that undermines realism, and (2) the concern that quasi-naturalism is not a form of genuine realism, given its expressivism and its deflationary framework. (For more details, see the full paper, which I will post on this blog soon.)