Chapter 4: Introducing Ecumenical Expressivism

Chapter 4, ‘Introducing Ecumenical Expressivism’, packs in a lot of stimulating ideas that are central to the broader project of the book, so I’m afraid this summary will be a bit longer than ideal. Let’s start with a quick recap: Chapter 2 argued that normative thought is affective, action-guiding, and acrimonious, and that these features militate against understanding it as being purely representational. Chapter 3 made the case that the core features can’t be captured by adding a desire-like element to representational content. This points us in the direction of expressivism, the subject matter of the chapter at hand. For expressivists, normative thoughts are in the first instance constituted by desire-like states, and are as such inherently practical. This gives it an edge when it comes to accounting for the affective and action-guiding aspects of normative thought. Insofar as normative disagreement is best understood as disagreement in attitude, this also favours expressivism. The distinctive claim of ecumenical expressivism is that normative judgment is a hybrid state consisting of a desire-like element and a suitably related representational belief. In this central chapter, Ridge explains what exactly the hybrid state is, and argues that the meanings of normative terms are explained in terms of the mental states they express.

Procedural note: I wrote this critical summary on the basis of an earlier draft of the book, which was organized differently. Since I don’t think the order affects the points made below, I’ll leave it as it is.

Normative Thought

Ridge labels the desire-like component of a normative judgment the agent’s normative perspective. For Gibbard, it would consist in states of norm-acceptance, but Ridge prefers a Bratmanian picture, on which a normative perspective is a “set of relatively stable policies against accepting certain standards of deliberation” (115), in particular policies the agent is satisfied with (roughly in the sense of not having conflicting policies). Policies here are kind of general contingency plans, in particular plans about how to deliberate. (For Bratman, self-governing policies are plans about which desires to treat as providing justifying reasons, but Ridge rightly eschews appeal to unanalysed notions treating something as a reason.) So when you have a normative perspective, you plan (that is, intend) not to deliberate in certain ways. Perhaps you intend not to be the kind of person for whom the colour of another’s skin is an important consideration in deciding whether to pursue friendship. (This is not how Ridge puts it, but since I worry about the notion of a standard here (see below), I’m suggesting an alternative gloss in the spirit of the view, as I understand it.)

But surely there’s more to normative thought than settling on not deliberating in certain ways! Ridge agrees. Our normative perspectives are also constituted by plans to act and deliberate in ways that are required or highly recommended by all the standards we have not ruled out (any acceptable standards). They also involve a defeasible propensity to encourage others to do likewise. (Think back on Stevenson: “I approve of this; do so as well”.)

So that, very roughly, is the desire-like aspect of normative judgment for Ridge. The judgment itself consists of a combination of normative perspective with a relevant representational belief. What representational belief? Here we go back to Chapter 1 and the idea that any normative judgment can be understood in terms of something being highly ranked or required or recommended by contextually determined standards. Suppose Anne thinks that pleasure is good as an end. According to Ridge’s analysis, this amounts to her thinking that pleasure would be highly ranked as an end by any ultimate acceptable standard of practical reasoning. And this thought, in turn, has the following two elements (cf. 119):

a)     A normative perspective.

b)    The representational belief that pleasure would be highly ranked as an end by any admissible ultimate standard of practical reasoning.

So, roughly, you accept certain standards of practical reasoning, and believe that those standards rank pleasure highly as an end. That’s what it is to think that pleasure is good as an end. On this view, a normative judgment is a relational state in the sense that Mark Schroeder has popularized, since the content of the belief refers to the content of the normative perspective (‘admissible’ standards in b are those not ruled out in a). The core appeal of thinking of normative judgments as this sort of hybrid states is that they allow for any degree of logical complexity in the content of the belief element, which helps deal with some key objections to expressivism (see next chapter).

Okay, so much for a very brief sketch of the view. I have four questions or worries. First, it’s not clear what the role of the representational belief is in normative judgment. Return to the case of thinking that pleasure is good as an end. Evidently, Anne’s normative perspective consists of policies ruling out standards that rule out pleasure as an end, and plans to give pleasure positive weight as an end in practical reasoning. Question: why does that not suffice, for an expressivist in particular, for her judging that pleasure is good as an end? If all the standards she endorses rank pleasure highly, why does she also have to believe that admissible standards rank pleasure highly? After all, the distinctively normative aspect of the thought is all in the attitudes, for an expressivist. To be sure, the hybrid model makes sense when it comes to, say, thoughts of good bread. It’s plausible enough that one way in which the thought that O’Keeffe’s bread is good can be realized is that I accept only standards for choosing bread that rank crustiness highly, and believe that O’Keeffe’s bread is crusty. But what about the thoughts that are not of this type?

Second, since the expressivist project is to explain normative thought without appeal to normative content (which will be downstream from normative thought), there’s always the question of whether an expressivist account is necessary and sufficient for normative judgment. Since Ridge packs a lot into the notion of a normative perspective, the necessity question is particularly urgent for his view. Do we really have to manifest such complex intentions and beliefs in order to think, say, that we have a reason to go swimming on a hot day? This isn’t just a point about the mind-boggling psychological complexity of Ridge’s view. It is rather that it seems possible to make normative judgments that don’t involve stable policies of any sort, not to mention stable policies regarding standards of practical reasoning. Why wouldn’t it be possible?

Third, what about sufficiency? Recall that normative perspectives are essentially sets of complex conditional intentions. The question is whether combining such planning states with representational beliefs is enough for a normative judgment. Clearly, we can intend to do something without judging that we ought to do it, or have reason to do it. So why can’t we intend to (not to) deliberate in a certain way without thinking that we ought to do so, or that we ought to act in a certain way? On this point, the contrast between Gibbard’s and Ridge’s views is instructive. On Gibbard’s original view, moral judgment consists of accepting norms for guilt and anger. Although there’s room to question the sufficiency of that, it’s at least plausible that endorsing reactive attitudes is closely bound up with moral (and perhaps more broadly normative) judgment. But for Ridge, emotions are only contingently linked with normative judgment. So all the weight is borne by policies for deliberating. And on the face of it, it seems I could have a policy of giving positive weight to X in deliberation (including rejecting standards for deliberating that don’t give X positive weight) without thereby thinking that X is good or that I have reason to X (especially if we understand giving positive weight in naturalistic terms, as something a robot might do).

My fourth, and possibly related, worry about Ridge’s account is that I’m not sure if it goes deep enough. Take again Gibbard’s original view for comparison. Accepting a norm requiring X-ing is, roughly, being motivated to X and perhaps feeling bad if one doesn’t X, and being disposed to communicate this state of mind to others when speaking without openly. So all we have is attitudes and behavioural dispositions whose object is an unambiguously factual (X-ing, speaking in a certain way). This is a clearly naturalistic, non-normative account of what it is to think normative thoughts. Ridge’s view, in what seems to me to be an important contrast, is spelled out in terms of commitments whose content involves reference to what is required or recommended by a standard. To my eyes, that’s normative talk, something the expressivist needs to analyse away in terms of attitudes toward natural facts. It’s not enough to know that in making a normative judgment, we’re committed to a standard that requires X-ing. We need to know, in terms that don’t make reference to “standards” or “requiring”, what attitudes we must have toward X-ing in order to have such commitment. This is even more urgent when it comes to beliefs about standards requiring or recommending or highly ranking something – what is it to believe such a thing?

Normative Language

So much for a brief characterization of ecumenical expressivism as a theory of normative thought. But what about normative language? It’s not clear, Ridge observes, what question expressivism is trying to answer when it promises to account for the meaning of normative language. There’s two main options. A semantic theory for a language consists of two main parts: an assignment of semantic values to the terms and a syntax that tells us how to derive semantic values of wholes from the semantic values of parts and the way they’re arranged. A meta-semantic theory, in Ridge’s terms, explains why terms have the meanings they do. Mark Schroeder thinks of expressivism as a first-order semantic theory, according to which the contents (or semantic values) of normative sentences are mental states. But Ridge argues that expressivism is best understood as a view in meta-semantics instead.

What does this mean? First, expressivism is compatible with standard first-order semantics for normative sentences. So the expressivist will agree that “Abortion is wrong” is true iff abortion is wrong, or if the referent of “abortion” is in the extension of the predicate “morally wrong”. Here terms like “truth” and “referent” are used in a deflationary sense that doesn’t entail that there is, say, the property of wrongness out there. (Making this contrast is notoriously hard once you go deflationist about everything, but you get the point.) Second, what is distinctive about expressivism is the meta-semantic thesis it advances: normative claims have the meanings they do in virtue of their expressing normative judgments, which are hybrid states of the sort discussed above.

Ridge labels the broader Lockean/Gricean metasemantic framework in which meanings of linguistic items are explained by way of the mental states they express ideationalism. Words mean what they do, because they are conventionally used to express certain states of mind. What is it to ‘express’ a state of mind? Ridge sketches two options. The first, based on Wayne Davis’s work, is that a word expresses an idea if it conventionally indicates it. (Presumably Davis explains what an ‘idea’ is – on the face of it, this suggestion is highly implausible, as the basic units of communication are sentences rather than words, and they express complex, highly unconventional thoughts rather than individual ideas.) The other is that sentences express those thoughts that people who assert them are accountable for having. For example, if you assert “There’s cheese on the moon”, you’re accountable for having the belief that there’s cheese on the moon, on pain of being charged with insincerity or semantic confusion by other members of your linguistic community. Like Ridge, I find the latter account of expression more promising – but it’s not immediately clear how it links up with the core claim that meanings are explained by the mental states they express. It’s certainly compatible with non-ideationalist meta-semantic frameworks.

Ridge does, however, provide a story of how the meta-semantic explanation works – how the mental state that a sentence expresses explains its propositional content (i.e. the proposition it expresses, in a different sense of ‘express’). Importantly, propositions here cannot be understood in standard Fregean or Russellian or Lewisian sense, as abstract entities or structured entities composed of individuals and properties or sets of possible worlds. If they were, normative propositions could be the contents of standard representational beliefs, and the contents of normative claims could be explained by virtue of what they represent, as cognitivists maintain. So Ridge has to cast about for an alternative conception of propositions, and finds one in Scott Soames’s work. Soames argues that propositions are cognitive event types. On this view, a belief that X is F, say, isn’t a relation to an abstract entity (for example) with the content that X is F. Instead, it is a cognitive event that consists of predicating F of X (where predication is a primitive contentful mental act) and some sort of endorsement of this predication. Propositions are types of basic cognitive events (such as predications).

Soames’s view is an account of propositions as bearers of representational content. But Ridge maintains that this approach generalizes naturally to normative thoughts. They, too, are cognitive events of various cognitive types. Normative propositions are, then, cognitive types of a particular kind – types of practical thought. Ridge claims that “we can have normative propositions in precisely Soames’s sense. For once we allow that thinking that abortion is morally wrong is a cognitive event, we can plausibly infer that there is a corresponding cognitive event type, and so on for all such normative thoughts.” (129)

But is this really sufficient? Sure, hybrid states form cognitive types. But central to Soames’s case for identifying propositions with cognitive event types is that cognitive events of predication (in particular) are primitively representational (and have inherent truth conditions), unlike abstract objects (Fregean propositions). But hybrid states are not primitively representational – they’re not representational at all! So I don’t see how following Soames’s arguments would get us to normative propositions consisting of essentially non-cognitive event types.

Putting that worry aside, Ridge believes that this understanding of propositions allows for a distinctively expressivist metasemantics. On this view, giving the semantic content of a claim consists of “articulating the proposition it expresses in some privileged language” (130). (Is this really all that first-order semantics does?) Since normative propositions are types of hybrid states, “what it is for a proposition p to be expressed by a claim in a context of utterance is for that claim to express the thought that p, where the proposition that p just is the cognitive event type of thinking that p” (130).

There is evidently much more to be said here, but I’ll leave it for general discussion. Here’s a final thought. It seems that expressivism faces a kind of dilemma: either it is committed to non-standard, non-truth conditional semantics, or it is committed to a controversial meta-semantic framework and a highly controversial account of propositions (which does not obviously fit with the expressivist strategy). (Officially, Ridge says that ideationalism is the “most natural home” for expressivism, but it’s hard to see how the strategy would work in other metasemantic frameworks.) Either way, it seems that accounting for normative language in terms of essentially non-representational states is not good for your plausibility points, unless you belong to the small set of people who are antecedently both ideationalist and Soamesian.