Many thanks to Krister Bykvist and Jonas Olson for their thoughtful review of my book Choosing Normative Concepts, and for their kind words. In this reply, I focus on one main theme they bring up.

As I present the scenario Alternative (p. 18), around which much of my discussion of alternative normative concepts revolves, the scenario concerns a community using concepts with the same normative roles as ours but differing in extension. In the book, I leave normative role schematic but gloss the notion in terms of action-guiding and motivation (p. 10). As I mention (pp. 38ff.), doubts can be raised regarding the notion of normative role. But the idea of something like normative role is familiar from the literature. For example, when Terence Horgan and Mark Timmons (‘Troubles for New Wave Moral Semantics: The “Open Question Argument” Revived’, 1992) describe their Moral Twin Earth thought experiment, they in effect stipulate that our normative terms and the Moral Twin Earthlings’ normative terms have the same normative roles. The stipulation of sameness of normative role is crucial in the Moral Twin Earth thought experiment: it is by virtue of the sameness in this regard that we have (or are supposed to have) the intuition that the Moral Twin Earthlings’ words must have the same reference as ours. In the setting of the discussion of my book, this stipulation is also important. It is because the possible normative concepts considered have the same normative roles as ours that they are supposed to be clearly relevant to metaethics. I raise questions of this kind: do considerations about what is good and right have the significance we normally attach to them, given that there are these alternative concepts? If the alternative concepts weren't in this way suitably related to our actual normative concepts, then these questions would not obviously be as pressing.

My main reaction to – and main objection to – Bykvist and Olson's review is that they seem to overlook this fact about how I set things up. They do say things like:

But they still seem to overlook the importance of normative role. Tellingly, ‘role’ does not occur in the text after the paragraph I quoted from. Instead they just go on to make the following point:

Here the talk of normative role has disappeared; instead the extensions of the concepts are compared. In the book, I don't often use locutions like ‘rightness concept’, and when I do it is just shorthand for: concept having the same normative role as our concept right. It is beside the point whether the alternative so-called rightness concept in an intuitive sense is a rightness concept. As for ‘changing the subject’, I think there is a fairly well-entrenched philosophical use of this locution where it is sufficient for me to change the subject when we ‘have a discussion about F-ness’ that I use ‘F’ with a different extension from you. Someone who starts talking about rightness* rather than rightness is certainly changing the subject in that sense. As Herman Cappelen (Fixing Language, 2018) has recently stressed, we are in ordinary contexts more liberal about what counts as the same topic or subject. But there is a question of what the theoretical significance of this ordinary notion of sameness of topic may be. At any rate, I don't rely on this intuitive notion of same topic. What I claim is that alternative normative concepts have a certain kind of metanormative significance when they have the same normative roles as our actual concepts. (Note: I do not say ‘and only when’ – more on this below.) This is independent of intuitive judgements about sameness of topic. The thought is this. Suppose that our concepts and some alternatives have the same normative roles but differ in extension. The extensions are then in part determined by something other than normative role. Whatever that something may be (causal connections? analytic connections to the descriptive?), it threatens not to be normatively relevant. But if it isn't, why should we take our concepts to be more authoritative than theirs?

Bykvist and Olson's disregard of the centrality of normative role to the discussion also affects what they say about my discussion of ‘superrightness’. What I say there, I say in response to an envisaged objector who emphasizes that the differences in extension between concepts with the same normative role are only minor; and whom I take therefore to hold that what is important is being in the extensions of all the concepts with the same normative role. When Bykvist and Olson discuss what I say about superrightness, they fail to take onto account this aspect of that discussion. Instead they treat my discussion simply as attempting to respond to the charge that only concepts not too different in extension from the concept expressed by our ‘right’ are relevant.

About the missing ‘and only when’, above: even though sameness of normative role plays a part when I first introduce the problem, I later loosen that restriction. I talk about concepts with somewhat different normative roles from ours; and I talk about systems of normative concepts which are not readily comparable to our concepts. If, like Bykvist and Olson, one is concerned about whether the alternative concepts are properly called rightness concepts, or whether someone using those concepts instead is ‘changing the subject’, one may be even more concerned on this score about these later versions of the problem. But again, I do not have such concerns, and I do not think they are reasonable concerns to have.