Chapter 1: Locating Normative Thought and Discourse

First off, thanks to everyone for signing up and reading along.  I hope you find this short summary useful.


Ridge begins by drawing on Geach’s important claim that the word “good” is an “incomplete predicate”.  Roughly, this is the view that nothing is good simpliciter, rather, actions, character traits, objects and states of affairs are always good in some way, and Ridge believes that this semantic blank is filled in by the context of use.  As such, he holds that the meanings of normative terms are “systematically context-sensitive”. 

As I see it, a central tension animates this chapter: how do we account for the (rather obvious) context-sensitivity of key normative terms without resorting to the postulation of brute semantic ambiguity? 

So long as each term has a Kaplanian character, or a “systematic function from contexts to contents”, then we can avoid this result.  This leads us to the most important concept in Ridge’s first-order semantic theory, that of a standard, which is meant to serve as the concept that will provide semantic unity to normative terms. Beginning with evaluatives like ‘good’, Ridge claims that “To call something good is to characterize it as being such that any standard of a certain kind would rank it highly”, where context fixes precisely which standard is being applied to the object in question. By way of illustration, he refers to functional kinds (such as toasters, knives, automobiles) which we “naturally” associate with certain standards.

He also provides a similar semantic theory for directive terms such as “ought” and “must”.  To say that we ought to/must X is just to say that “any standard of contextually specified kind S would, relative to a contextually specified set of background information or facts B, recommend/require X.” 

Here, there is a little more structure than there is for ‘good’, since the terms ought and must are semantically associated with the concepts of recommending and requiring, respectively. Moreover, here, the semantics calls for the context to fix two distinct elements: the evaluative standard itself and a set of background information, information which “limit[s] the space of possible worlds to those consistent with the background evidence or circumstances.”

In fact, there are more elements, since he unifies his account of modals by claiming that standards themselves are a species of the genus mode of ranking. Thus, at a very abstract level, which encompasses both epistemic and practical uses, modals like ‘ought’ to P and ‘must P’ have the following semantics:

Any mode of ranking of kind R would, relative to a contextually specified set of background information or facts B, rank P sufficiently highly.

What counts as ‘sufficiently highly’?  That, too, is determined by context.  I want to note at this point that the poor old context is being called upon to do a fair bit of work in fixing the meanings of normative sentences.  More on this shortly.

Ridge moves to the concept of a reason, and he follows much recent work in claiming that reasons are facts which make either actions or natural phenomena intelligible. He (unsurprisingly) deploys the concept of a standard in order to cash out the ways in which practical reasons render things intelligible:

[Fact F is a reason for A to Φ in C iff]: Fact F explains why any standard of a certain sort for Ψ-ing would assign positive weight to A’s Φ-ing in C.

Now, it’s not immediately clear what this sort of explanation will look like. However, he does provide one example which, I think, can illustrate how this is supposed to work.  A chess expert says:

“the fact that your queen can be skewered to your king is a reason to move your queen.”

Presumably, any good standard for chess-playing will assign positive weight to actions which protect your queen, ceteris paribus. So, we can take as contextually fixed a standard that ranks chess moves in order of how likely they are to lead to victory. We then see the fact:  [A’s queen can be skewered to A’s king], and this fact will play a central role in an explanation for why moving A’s queen will be highly ranked by that standard.  Thus, the fact becomes a reason (along, presumably, with any other facts that are relevant to the explanation). 

Ridge concludes by claiming that his work in this chapter has provided an illustration of how we should locate the normative: by giving a first-order semantics for key normative terms.  We discover, in particular, that normative uses of these terms involve semantic reference to an acceptable standard of practical reasoning. 

A critic might register familiar worries about the metaphysical status of “acceptable standards of practical reasoning”, and about what the word ‘acceptable’ is pointing to. However, Ridge has a response: such debates are really meta-semantic, and that familiar metaethical debates can be reborn as meta-semantic debates.  We have discovered that normative semantics refers to acceptable standards, but that does not settle the question of what acceptable standards are. 

Perhaps in fairness to this critic, however, we could say that, in spite of what Ridge seems to suggest, first-order semantics doesn’t actually locate the normative, rather, it tells us what we are looking for.

I want to close on a more critical note: If a semantics is supposed to generate truth-evaluable propositions via some kind of specifiable algorithm that takes us from natural language sentences to literal content, then unless Ridge tells us how context fixes such things as evaluative standards or action-rankings, then we do not actually have a semantics on the table.  This conclusion might seem unfair, but compare Kaplan’s semantics for indexicals, which is (appropriately) context-sensitive but which actually gives us the function that maps a word (for example, ‘I’) to determinate literal content (the speaker). 

Moreover, this heavy reliance on context to fix contents might undermine what I take to be the central aim of the chapter.  If the truth-conditional content of normative sentences is actually fixed by contextual factors in a relatively unsystematic way, we are (effectively) postulating brute ambiguity in the semantics of these terms.  After all, if I decide to lose the chess game, my reasons change, what counts as a "good move" changes, and what I ought to do changes, because the contextually relevant standard is suddenly very different.  An outside observer will watch my next move and say "that was a bad move", and I'll say, "no, it wasn't" and we might be talking right past one another, since our normative terms might refer to very different standards. The worry is that this might happen a lot, and the apparent semantic unity provided by the concept of a 'standard' does nothing on its own to prevent it. What is needed is a determinate story about how context helps to fix standards, a story that will guarantee that most of the time we are fixing on (roughly) the same standards.

Thus endeth the short summary and critique.  What say you all?