If we taboo and reduce, then the question of ”...but is it good?” is out of place. The reply is: “Yes it is, because I just told you that’s what I mean to communicate when I use the word-tool ‘good’ for this discussion. I’m not here to debate definitions; I’m here to get something done.”
I just wanted to flag that a non-reductionist moral realist (like myself) is also “not here to debate definitions”. See my post on The Importance of Implications. This is compatible with thinking well of the Open Question Argument, if we think we have an adequate grasp of some fundamental normative concept (be it ‘good’, ‘reason’, or ‘ought’—I lean towards ‘reason’, myself, such that to speak of a person’s welfare is just to talk about what a sympathetic party has reason to desire for the person’s sake).
Note that if we’re right to consider some normative concepts to be conceptually primitive (not analytically reducible to non-normative concepts) then your practice of “tabooing” all normative vocabulary actually has the effect of depriving us of the conceptual tools necessary to even talk about the normative sphere. Consequent talk of people’s (incl. God’s) desires or dispositions is simply changing the subject, on this way of looking at things.
Out of interest: Will you be arguing anywhere in this sequence against non-reductionist moral realism? Or are you simply assuming its falsity from the start, and exploring the implications from there? (Even the latter, more modest project is of course worth pursing, but I personally would be more interested in the former.) Either way, it’d be good to be clear about this. (You could then skip the silly rhetoric about how what is not “is”, must be “is not”.)
I was thinking of “fundamental” concepts as those that are most basic, and not reducible to (or built up out of) other, more basic, concepts. I do think that normative concepts are conceptually isolated, i.e. not reducible to non-normative concepts, and that’s really the more relevant feature so far as the OQA is concerned. But by ‘fundamental normative concept’ I meant a normative concept that is not reducible to any other concepts at all. They are the most basic, or bedrock, of our normative concepts.
Given the extremely poor access human beings have to the structure of their own concepts, it’s dubious that the methods of analytic philosophy can trace those structures. Moreover, concepts typically “cluster together similar things for purposes of inference” ( Yudkowsky ) and thus we can re-structure them in light of new discoveries. Concepts that are connected now might be improved by disconnecting them, or vice versa. It is not at all clear that normative concepts are not included in this (Neurath-style) boat.
I’m inclined not to write about moral non-naturalism because I’m writing this stuff for Less Wrong, where most people are physicalists
Physicalists could (like Mackie) accept the non-naturalist’s account of what it would take for something to be genuinely normative, and then simply deny that there are any such properties in reality. I’m much more sympathetic to this hard-headed “error theory” than to the more weaselly forms of naturalism.
I think many of our normative concepts fail to refer, but that a class of normative concepts often called hypothetical imperatives do refer, thanks to a rather straightforward reduction as given above. Are hypothetical imperatives not ‘genuinely normative’ in your sense of the phrase? Do you use the term ‘normative’ when talking about things other than hypothetical imperatives, and do you think those other things successfully refer?
“Hypothetical imperatives thus reveal patterns of normative inheritance. But their highlighted ‘means’ can’t inherit normative status unless the ‘end’ in question had prior normative worth. A view on which there are only hypothetical imperatives is effectively a form of normative nihilism—no more productive than an irrigation system without any water to flow through it.”
(Earlier in the post explains why hypothetical imperatives aren’t reducible to mere empirical statements of a means-ends relationship.)
I tentatively favour non-naturalist realism over non-naturalist error theory, but my purpose in my previous comment was just to flag the latter option as one that physicalists should take (very) seriously.
You know this, but for the benefit of others: Roughly, error theory consists of two steps. As Finlay puts it:
(1) Presupposition: moral judgments involve a particular kind of presupposition which is
essential to their status as moral;
(2) Error: this presupposition is irreconcilable with the way things are
Given my view of conceptual analysis, it shouldn’t be surprising that I’m not confident of some error theorists’ assertion of step 1. Is a presupposition of moral absolutism ‘essential’ to a judgment’s status as a ‘moral’ judgment? Is a presuppositional of motivational internalism ‘essential’ to a judgment’s status as a ‘moral’ judgment? I don’t know. Moral discourse (unlike carbon discourse) is so confused that I’m not too interested to assert one fine boundary line around moral terms over another.
So if someone thinks a presupposition of supernaturalism is ‘essential’ to a judgment’s status as a ‘moral’ judgment, then I will claim that supernaturalism is false. But this doesn’t make me an error theorist because I don’t necessarily agree that a presupposition of supernaturalism is ‘essential’ to a judgment’s status as a ‘moral’ judgment. I reject step 1 of error theory in this case.
Likewise, if someone thinks a presupposition of moral absolutism or motivational internalism is essential to a judgment’s status as a ‘moral’ judgment, I’ll be happy to deny both moral absolutism and motivational internalism, but I wouldn’t call myself an error theorist because I reject the claim that moral judgments (by definition, by conceptual analysis) necessarily presuppose moral absolutism or motivational internalism.
But hey, if you convince me that the presumption of motivational internalism in moral discourse is so widespread that talking about ‘morality’ without it would be like using the term ‘phlogiston’ to talk about oxygen, then I’ll be happy to call myself an error theorist, though none of my anticipations will have changed.
Hypothetical imperatives
I’ll reply to a passage from your post on hypothetical imperatives. My reply won’t make sense to those who haven’t read it:
When we affirm the first premise as a mere hypothetical imperative, we mean it in a sense that does not validate such an inference. We might add, “But of course you shouldn’t want to torture children, and so you shouldn’t take the means to this atrocious end either.”
I think this is because ‘should’ is being used in different senses. The real modus ponens is:
If you want to torture children, you should_ToTortureChildren volunteer as a babysitter.
You want to torture children.
Therefore, you should_ToTortureChildren volunteer as a babysitter.
Or at least, that is plausibly what some people mean when they assert what looks like a hypothetical imperative. Doubtless, others will appear to be meaning something else if pressed by interrogation.
Now, to respond to Sidgwick:
When (e.g.) a physician says, “If you wish to be healthy you ought to rise early,” this is not the same thing as saying “early rising is an indispensable condition of the attainment of health.” This latter proposition expresses the relation of physiological facts on which the former is founded; but it is not merely this relation of facts that the word ‘ought’ imports: it also implies the unreasonableness of adopting an end and refusing to adopt the means indispensable to its attainment.
I could capture this ‘unreasonableness’ by simply clarifying that from the standpoint of Bayesian rationality, it would be somewhat irrational to expect good health despite not rising early (or so the doctor claims).
But again, I’m not too keen to play the definitions game. If you state hypothetical imperatives with more intuitions about the meaning of hypothetical imperatives than I do, then you are free to explain what you mean by hypothetical imperatives and then show how they fit into the physical world. If you can’t show how they fit into the world, then you’re talking about something that doesn’t exist, or else we’ll have to replay the physicalism vs. non-physicalism debate, which is another topic.
Thanks for this reply. I share your sense that the word ‘moral’ is unhelpfully ambiguous, which is why I prefer to focus on the more general concept of the normative. I’m certainly not going to stipulate that motivational internalism is true of the normative, though it does seem plausible that there’s something irrational about someone who acknowledges that they really ought (all things considered) to phi and yet fails to do so. (I don’t doubt that it’s possible for someone to form the judgment without any corresponding motivation though, as it’s always possible for people to be irrational!)
I trust that we all have a decent pre-theoretic grasp of normativity (or “ought-ness”). The question then is whether this phenomenon that we have in mind (i) is reducible to some physical property, and (ii) actually exists.
Error theory (answering ‘no’ and ‘no’ to the two questions above) seems the most natural position for the physicalist. And it sounds like you may be happy to agree that you’re really an error theorist about normativity (as I mean it). But then I’m puzzled by what you take yourself to be doing in this series. Why even use moral/normative vocabulary at all, rather than just talking about the underlying natural properties that you really have in mind?
P.S. What work is the antecedent doing in your conditional?
If you want to torture children, you should_ToTortureChildren volunteer as a babysitter.
Why do you even need the modus ponens? Assuming that “should_ToTortureChildren” just means “what follows is an effective means to torturing children”, then isn’t the consequent just plain true regardless of what you want? (Perhaps only someone with the relevant desire will be interested in this means-ends fact, but that’s true of many unconditional facts.)
there’s something irrational about someone who acknowledges that they really ought (all things considered) to phi and yet fails to do so. (I don’t doubt that it’s possible for someone to form the judgment without any corresponding motivation though, as it’s always possible for people to be irrational!)
Right.
And it sounds like you may be happy to agree that you’re really an error theorist about normativity (as I mean it).
Unfortunately, I don’t think I’m clear about what you mean by normativity. The only source of normativity I think exists is the hypothetical imperative, which can be reduced to physics by straightforward methods such as the one I used in the original post. I’m not an error theorist about that kind of normativity.
Why even use moral/normative vocabulary at all, rather than just talking about the underlying natural properties that you really have in mind?
This is a good question. Truly, I want to get away from moral vocabulary, and be careful around normative vocabulary. But people already think about these topics in moral and normative vocabulary, which is why I’m trying to solve or dissolve (in this post and its predecessor) some of the usual ‘problems’ in this space of questions.
After that’s done, I don’t think it will be most helpful to use moral language. This is evident in the fact that in 15 episodes of my ‘morality podcast’ I’ve used almost no moral language at all.
What work is the antecedent doing in your conditional?
Not much, really. I wasn’t using the modus ponens to present an argument, but to unpack one interpretation of (some) ‘should’ discourse. Normative language, like many other kinds of language, is (when used correctly) merely a shortcut for saying something else. I can imagine a language that has no normative language at all. In that language we couldn’t say things like “If you want to torture children, you should volunteer as a babysitter” but we could say things like “If you volunteer as a babysitter you will have more opportunities to torture children.” The way I’m parsing ‘should’ in the first sentence, nothing is lost by this translation.
Of course, people use ‘should’ in a variety of ways, some of which translate into claims about things reducible to physics, others of which translate into claims about things non-reducible to physics, while still others don’t seem to translate into cognitive statements at all.
Thanks, this is helpful. I’m interested in your use of the phrase “source of normativity” in:
The only source of normativity I think exists is the hypothetical imperative
This makes it sound like there’s a new thing, normativity, that arises from some other thing (e.g. desires, or means/ends relationships). That’s a very realist way of talking.
I take it that what you really want to say something more like, “The only kind of ‘normativity’-talk that’s naturalistically reducible and hence possibly true is hypothetical imperatives—when these are understood to mean nothing more than that a certain means-end relation holds.” Is that right?
I’d then understand you as an error theorist, since “being a means-end relationship”, like “being red”, is not even in the same ballpark as what I mean by “being normative”. (It might sometimes have normative importance, but as we learn from Parfit, that’s a very different thing.)
My thought process on sources of normativity looks something like this:
People claim all sorts of justifications for ‘ought’ statements (aka normative statements). Some justify ought statements with respect to natural law or divine commands or non-natural normative properties or categorical imperatives. But those things don’t exist. The only justification of normative language that fits in my model of the universe is when people use ‘ought’ language as some kind of hypothetical imperative, which can be translated into a claim about things reducible to physics. There are many varieties of this. Many uses of ‘ought’ terms can be translated into claims about things reducible to physics. If somebody uses ‘ought’ terms to make claims about things not reducible to physics, then I am suspicious of the warrant for those claims. When interrogating about such warrants, I usually find that the only evidences on offer are pieces of folk wisdom, intuitions, and conventional linguistic practice.
People claim all sorts of justifications for ‘ought’ statements (aka normative statements).
You still seem to be conflating justification-giving properties with the property of being justified. Non-naturalists emphatically do not appeal to non-natural properties to justify our ought-claims. When explaining why you ought to give to charity, I’ll point to various natural features—that you can save a life for $500 by donating to VillageReach, etc. It’s merely the fact that these natural features are justifying, or normatively important, which is non-natural.
Sure. So what is it that makes (a) [the fact that you can save a life by donating $500 to VillageReach] normatively justifying, whereas (b) [the fact that you can save a mosquito by donating $2000 to SaveTheMosquitos] is not normatively justifying?
On my naturalist view, the fact that makes (a) but not (b) normatively justifying would be some fact about how the goal we’re discussing at the moment is saving human lives, not saving mosquito lives. That’s a natural fact. So are the facts about how the English language works and how two English speakers are using their terms.
It’s not entirely clear what you’re asking. Two possibilities, corresponding to my above distinction, are:
(1) What (perhaps more general) normatively significant feature is possessed by [saving lives for $500 each] that isn’t possessed by [saving mosquitoes for $2000 each]? This would just be to ask for one’s fully general normative theory: a utilitarian might point to the greater happiness that would result from the former option. Eventually we’ll reach bedrock (“It’s just a brute fact that happiness is good!”), at which point the only remaining question is....
(2) In what does the normative signifiance of [happiness] consist? That is, what is the nature of this justificatory status? What are we attributing to happiness when we claim that it is normatively justifying? This is where the non-naturalist insists that attributing normativity to a feature is not merely to attribute some natural quality to it (e.g. of “being the salient goal under discussion”—that’s not such a philosophically interesting property for something to have. E.g., I could know that a feature has this property without this having any rational significance to me at all).
(Note that it’s a yet further question whether our attributions of normativity are actually correct, i.e. whether worldly things have the normative properties that we attribute to them.)
I gather it’s this second question you had in mind, but again it’s crucial to carefully distinguish them since non-naturalist answers to the first question are obviously crazy.
On my naturalist view, the fact that makes (a) but not (b) normatively justifying would be some fact about how the goal we’re discussing at the moment is saving human lives, not saving mosquito lives.
What if you actually should be discussing saving of mosquito lives, but don’t, because humans are dumb?
I take you to mean “what would maximize Luke’s utility function” (knowing that ‘utility function’ is probably just a metaphor when talking about humans) when you say “you actually should...” Of course, my ‘utility function’ is unknown to both of us.
In that case, it would remain true in our hypothetical scenario that I should-HumanLivesAreGood donate to VillageReach (assuming they’re a good charity for saving human lives), while I should-UtilityFunctionLuke donate to SaveTheMosquitos.
(Sorry about formatting; LW comments don’t know how to use underscores, apparently.)
On my naturalist view, the fact that makes (a) but not (b) normatively justifying would be some fact about how the goal we’re discussing at the moment is saving human lives, not saving mosquito lives.
But the question then is what goal you should have. It is easy to naturalise norms inasmuch as they are hypothetical and indexed to whatever you happen to be doing.
(if you want to play chess, you should move the bishop diagonally)
The issue is how to naturalise categorical ends,the goals you should have and the
rules you should be following irrespective of what you are doing.
That’s a natural fact. So are the facts about how the English language works and how two English speakers are using their terms.
Such facts aren’t supernatural. OTOH, they fall on the analytical/apriori side of the
fence, rather than the empirical side, and that is an iimportant distinction.
What reasons are there for doubting the existence of categorical imperatives that do not equally count against the existence of hypothetical imperatives? I can understand rejecting both, I can understand accepting both, but I can’t understand treating them differently.
Depends what you mean by ‘categorical imperative’, and what normative force you think it carries. The categorical imperatives I am used to hearing about can’t be reduced into the physical world in the manner by which I reduced a hypothetical imperative into the physical world in the original post above.
I take it you want to reduce a hypothetical imperative like “If you want to stretch, then you ought to stand” into a physically-kosher causal claim like “standing is causally necessary for satisfying your desire to stretch”. Now, I’m skeptical about this reduction, simply because I don’t see how a mere causal claim could provide any normative direction whatsoever. But in any case, it seems that you could equally well reduce a categorical imperative like “Regardless of what you personally want, if you are the only one around a drowning child, then you ought to help save it from drowning” into a physically-kosher causal claim like “your help is causally necessary for the survival of the drowning child”. Both causal claims are equally physicalistic, and both seem equally relevant to their respective imperative, so both physicalistic reductions seem equally promising.
Of course, the question “why should I assume that the drowning’s child survival actually matters?” seems reasonable enough. But so does the question “why should I assume that satisfying my desire to stretch actually matters?” If the first question jeopardizes the reduction of the categorical imperative, then the second question would also jeopardize the reduction of the hypothetical imperative.
it seems that you could equally well reduce a categorical imperative like “Regardless of what you personally want, if you are the only one around a drowning child, then you ought to help save it from drowning” into a physically-kosher causal claim like “your help is causally necessary for the survival of the drowning child”
I’m not used to the term ‘categorical imperative’ being used in this way. Normally, a categorical imperative is one that holds not just regardless of what you want, but regardless of any stated ends at all. Your reduction of this categorical imperative assumes an end of ‘survival of the drowning child.’ To me, that’s what we call a hypothetical imperative, and that’s why it can be reduced in that way, by translating normativity into statements about means and ends.
I think you must be mistaken about categorical imperatives and ends.
In the Groundwork, Kant deems rational nature (‘humanity’ in us) to be an end in itself which can ground a categorical imperative, in the 2nd Critique, he deems the highest good (happiness in proportion to virtue) to be a necessary end of practical reason, and in the Metaphysics of Morals, he deems one’s own perfection and the happiness of others to be ends that are also duties. For Kant, the categorical imperative of morality is directed at ends, but it is not a mere hypothetical imperative grounded in subjective ends.
And leaving Kant aside, plenty of moral systems aspiring to categorical force have ends at their center. Classic utilitarianism picks out the greatest overall balance of pleasure over pain as the one and only end of morality. Eudaimonist virtue ethics has as its end a well-lived flourishing life for the agent. Thomist ethics says an intellectual vision of God in the afterlife is the ultimate end of humans. These systems do not traffic in merely hypothetical imperatives: they present these ends as objectively worth pursuing, regardless of the agent’s personal preferences.
But if your main point is simply that my ‘reduction’ assumes that the end ought to be pursued, and that this assumption is a form of cheating, then I agree. I’ve left the normativity unexplained and unreduced. But then in exactly the same way, your reduction of hypothetical imperatives assumes that effective means to one’s ends ought to be taken, and this assumption is also a form of cheating. You have also left the normativity unexplained and unreduced.
So I don’t see how hypothetical imperatives are any more fit for naturalistic reduction than categorical imperatives.
I’m sure I speak differently about categorical imperatives than Kant does. I haven’t read much of Kant, and I don’t regret that. In your language, what you call an “end in itself” is what I mean to pick out when I talk about an imperative that “holds… regardless of any stated [arbitrary, desired, subjective] ends at all.” I don’t really know what it means for something to be an “end in itself”. Kant’s idea seemed to be that we ought to do X regardless of what anyone wants. Your way (and perhaps Kant’s way) of talking about this is to say that we ought to do X regardless of what anyone wants because X leads to a particular “end in itself”, whatever that means.
your reduction of hypothetical imperatives assumes that effective means to one’s ends ought to be taken, and this assumption is also a form of cheating
My reduction of hypothetical imperative doesn’t assume this. It only translates ‘ought’ into a prediction about what would realize the specified end. If there’s something mysterious left over, I’m curious what you think it is and whether it is real or merely a figment of folk wisdom and linguistic practice.
So, on your use of ‘end’, an ‘end’ cannot be objective and unconditional? I think that’s a highly uncommon use of the term.
But if you go this way, it seems like it’s less of a reduction of ‘ought’ and more of a misinterpretation, like reducing ‘Santa Claus’-talk into talk about Christmas cheer, or ‘God’-talk into talk of love.
After all, one important constraint on any interpretation of any ‘ought to X’ is that it should be positive towards X as opposed to negative or neutral, in favor of some action or attitude as opposed to against it or indifferent. But a mere predictive causal claim doesn’t have any valence at all: it’s just a neutral claim about what will probably lead to what, without anything positive or negative. So any attempt to reduce oughts to predictive causal claims seems doomed to failure.
EDIT: For the record, I’m an expressivist about normativity, and I think any attempt to understand it in terms of some actual or hypothetical ontology that could serve as the truth-conditions for a descriptive belief is a mistake. The mystery, I would say, lies in a descriptive interpretation of normativity, not in normativity itself.
So, on your use of ‘end’, an ‘end’ cannot be objective and unconditional? I think that’s a highly uncommon use of the term.
No, I just should have been clearer that when I said “stated end” I meant “subjective, desired end”. As commonly used, ‘end’ includes unconditional ends, I just haven’t ever been presented with an argument that persuaded me to think that such ends exist.
one important constraint on any interpretation of any ‘ought to X’ is that it should be positive towards X as opposed to negative or neutral, in favor of some action or attitude as opposed to against it or indifferent. But a mere predictive causal claim doesn’t have any valence at all: it’s just a neutral claim about what will probably lead to what, without anything positive or negative. So any attempt to reduce oughts to predictive causal claims seems doomed to failure.
Sure, you can stick that feature into your meaning of ‘ought’ if you want. But I’m not going too deep into the conceptual analysis game with you. If you want to include positive valence in the meaning of a hypothetical ought, then we could translate like this:
“If you want to stretch, you ought to stand” → “standing will make it more likely that you fulfill your desire to stretch, and I have positive affect toward you standing so as to fulfill your desire to stretch”
As I said in my post, expressivism is fine—and certainly true of how many speakers use normative language. I happen to be particularly interested in those who use normative language cognitively, and whether their stated moral judgments are true or false, and under which conditions—which is why I’m investigating translations/reductions of normative language into natural statements that have truth conditions.
What reasons are there for doubting the existence of categorical imperatives that do not equally count against the existence of hypothetical imperatives?
The set of non-ethical categorical imperatives is non-empty.
The set of non-ethical hypothetical imperatives is non-empty.
Hypothetical imperatives include instrumental rules, you have to use X to achieve Y,
game-laying rules, etc.
The set of non-ethical categorical imperatives is non-empty.
I agree. Epistemic imperatives are categorical, but non-empty.
The set of non-ethical hypothetical imperatives is non-empty. Hypothetical imperatives include instrumental rules, you have to use X to achieve Y, game-laying rules, etc.
Right, those are examples where non-ethical hypothetical imperatives often show up.
So how does this add up to a reason to think there is a case against categorical imperatives that doesn’t equally well count against hypothetical imperatives?
My thought process on sources of normativity looks something like this:
People claim all sorts of justifications for ‘ought’ statements (aka normative statements). Some justify ought statements with respect to natural law or divine commands or non-natural normative properties or categorical imperatives.
There’s a lot of kinds of normative/”ought” statements. Some relate to games, some to
rationality, and so on. Hypothetical “ought” statements do not require any special
metaphysical apparatus to explain them, they just require rules and payoffs. Categorical imperatives are another story.
When interrogating about such warrants, I usually find that the only evidences on offer are pieces of folk wisdom, intuitions, and conventional linguistic practice.
One man’s conventional linguistic practice is another’s analytical truth.
Hypothetical “ought” statements do not require any special metaphysical apparatus to explain them, they just require rules and payoffs. Categorical imperatives are another story.
Rules and payoffs explain “ought” statements only if you assume that the rules are worth following and the payoffs worth pursuing. But if hypothetical imperatives can help themselves to such assumptions (assuming e.g. that one’s own desires ought to be satisfied), then categorical imperatives can help themselves to such assumptions (assuming e.g. that everyone’s desires ought to be satisfied, or that everyone’s happiness ought to be maximized, or that everyone ought to develop certain character traits).
Rules and payoffs explain “ought” statements only if you assume that the rules are worth following and the payoffs worth pursuing.
I don’t think so. You ought to use a hammer to drive in nails even if you don’t want
to dive in nails. Anyone who is playing chess should move the bishop diagonally.That doesn’t mean you are playing chess.
Of course those are hypothetical, and non-ethical. It might wll be the case that the only categorical imperatives are moral categorical imperatives; that. ethics is the only area where you should do things or refrain form things unconditionally.
I don’t think so. You ought to use a hammer to drive in nails even if you don’t want to dive in nails. Anyone who is playing chess should move the bishop diagonally.That doesn’t mean you are playing chess.
Again, you’re assuming that the rule ‘if you’re driving in nails, use a hammer’ is worth following, and that the rule ‘if you’re playing chess, move bishops diagonally’ is worth following. A nihilist would reject both of those rules as having any normative authority, and say that just because a game has rules it doesn’t mean that game-players ought to follow those rules, at most it means that lots and lots of rule-violations make the game go away.
I don’t think hypothetical imperatives can be reduced. The if-ought of a hypothetical imperative is a full-blooded normative claim. But you can’t reduce that to a simple if-then about cause and effect.
To see why, consider a nihilist about oughts. She recognizes the causal connections between calorie consumption/burning and weight loss. But she doesn’t accept any claim about what people ought to do, even hypothetical imperatives about people who desire weight loss. This seems perfectly coherent: she accepts causal claims, but not normative claims, and there’s no contradiction or incoherence there. But this means the causal claims she accepts are not conceptually equivalent to the normative claims she rejects.
For another way to see why, consider the causal claim “less calorie consumption and more calorie burning leads to weight loss”. This causal claim points in no normative direction. It doesn’t recommend anything, or register any approval, or send any positive or negative messages. Of course, we can take it in one direction or another, but only by combining it with separate normative claims:
Less calorie consumption and more calorie burning leads to weight loss.
People ought to take causally efficacious steps to satisfy their desires.
Therefore, if you desire to lose weight, you ought to consume less calories and burn more calories.
Premise 2 is what provides the normativity. It points us in the direction of satisfying desires. But we could easily take things in the opposite direction.
Less calorie consumption and more calorie burning leads to weight loss.
People ought to take causally efficacious steps to frustrate their desires.
Therefore, if you desire to lose weight, you ought to consume more calories and burn less calories.
Again, premise 2 is what provides the normativity. But it points us in the opposite direction, viz. the direction of frustrating desires.
So it’s pretty clear that premise 1 has no normativity in it. It can’t be reduced to either of the two 3′s. For we cannot arrive at a 3 without a 2.
I think stating premise 2 is a little odd. It is a bit deja “Tortoise and Achilles” all over again. If there’s a norm hiding around here, it’s an “ought” portrayed by the desire.
Second, conceptual analysis (or conceptual equivalence) is not necessary for reduction. Look at reduction in the sciences for examples.
Well, I’ll acknowledge that you could change premise 2 into an inference rule. But notice that you could change either premise 2—the pro-desire-satisfaction one and the pro-desire-frustration one—into an inference rule. Indeed, you could change any normative claim into an inference rule: you could change “people who want to have gay sex ought to go see a trained Baptist minister to get cured” into an inference rule, and then validly go from “I want to have gay sex” to “I ought to go see a trained Baptist minister to get cured”. So from the fact that premise 2 could be changed into an inference rule, I don’t think anything follows that might jeopardize its status as a full-blooded normative claim.
On the second point, I thought lukeprog was discussing direct conceptual reduction. But if he wants to provide hypothetical imperatives with a synthetic reduction, he’ll need a theory of reference capable of explaining why the normative claim turns out to make reference to (and have its truth-conditions provided by) simple causal facts. And on this score, I think hypothetical imperatives and categorical moral imperatives are on an equal footing: since reductionist moral realists have a hard time with synthetic reductions, I would expect reductionist ‘instrumental realists’ to have a hard time as well.
Perhaps so, but then the normativity stems from premise 1, leaving premise 2 as non-normative as ever. But the question is whether premise 2 could be a plausible reduction basis for normative claims.
I just wanted to flag that a non-reductionist moral realist (like myself) is also “not here to debate definitions”. See my post on The Importance of Implications. This is compatible with thinking well of the Open Question Argument, if we think we have an adequate grasp of some fundamental normative concept (be it ‘good’, ‘reason’, or ‘ought’—I lean towards ‘reason’, myself, such that to speak of a person’s welfare is just to talk about what a sympathetic party has reason to desire for the person’s sake).
Note that if we’re right to consider some normative concepts to be conceptually primitive (not analytically reducible to non-normative concepts) then your practice of “tabooing” all normative vocabulary actually has the effect of depriving us of the conceptual tools necessary to even talk about the normative sphere. Consequent talk of people’s (incl. God’s) desires or dispositions is simply changing the subject, on this way of looking at things.
Out of interest: Will you be arguing anywhere in this sequence against non-reductionist moral realism? Or are you simply assuming its falsity from the start, and exploring the implications from there? (Even the latter, more modest project is of course worth pursing, but I personally would be more interested in the former.) Either way, it’d be good to be clear about this. (You could then skip the silly rhetoric about how what is not “is”, must be “is not”.)
I’m inclined not to write about moral non-naturalism because I’m writing this stuff for Less Wrong, where most people are physicalists.
What does it mean to you to say that something is a ‘fundamental normative concept’? As in… non-reducible to ‘is’ statements (in the Humean sense)?
I was thinking of “fundamental” concepts as those that are most basic, and not reducible to (or built up out of) other, more basic, concepts. I do think that normative concepts are conceptually isolated, i.e. not reducible to non-normative concepts, and that’s really the more relevant feature so far as the OQA is concerned. But by ‘fundamental normative concept’ I meant a normative concept that is not reducible to any other concepts at all. They are the most basic, or bedrock, of our normative concepts.
Given the extremely poor access human beings have to the structure of their own concepts, it’s dubious that the methods of analytic philosophy can trace those structures. Moreover, concepts typically “cluster together similar things for purposes of inference” ( Yudkowsky ) and thus we can re-structure them in light of new discoveries. Concepts that are connected now might be improved by disconnecting them, or vice versa. It is not at all clear that normative concepts are not included in this (Neurath-style) boat.
Physicalists could (like Mackie) accept the non-naturalist’s account of what it would take for something to be genuinely normative, and then simply deny that there are any such properties in reality. I’m much more sympathetic to this hard-headed “error theory” than to the more weaselly forms of naturalism.
I think many of our normative concepts fail to refer, but that a class of normative concepts often called hypothetical imperatives do refer, thanks to a rather straightforward reduction as given above. Are hypothetical imperatives not ‘genuinely normative’ in your sense of the phrase? Do you use the term ‘normative’ when talking about things other than hypothetical imperatives, and do you think those other things successfully refer?
As I argue elsewhere:
“Hypothetical imperatives thus reveal patterns of normative inheritance. But their highlighted ‘means’ can’t inherit normative status unless the ‘end’ in question had prior normative worth. A view on which there are only hypothetical imperatives is effectively a form of normative nihilism—no more productive than an irrigation system without any water to flow through it.”
(Earlier in the post explains why hypothetical imperatives aren’t reducible to mere empirical statements of a means-ends relationship.)
I tentatively favour non-naturalist realism over non-naturalist error theory, but my purpose in my previous comment was just to flag the latter option as one that physicalists should take (very) seriously.
Error theory
You know this, but for the benefit of others: Roughly, error theory consists of two steps. As Finlay puts it:
Given my view of conceptual analysis, it shouldn’t be surprising that I’m not confident of some error theorists’ assertion of step 1. Is a presupposition of moral absolutism ‘essential’ to a judgment’s status as a ‘moral’ judgment? Is a presuppositional of motivational internalism ‘essential’ to a judgment’s status as a ‘moral’ judgment? I don’t know. Moral discourse (unlike carbon discourse) is so confused that I’m not too interested to assert one fine boundary line around moral terms over another.
So if someone thinks a presupposition of supernaturalism is ‘essential’ to a judgment’s status as a ‘moral’ judgment, then I will claim that supernaturalism is false. But this doesn’t make me an error theorist because I don’t necessarily agree that a presupposition of supernaturalism is ‘essential’ to a judgment’s status as a ‘moral’ judgment. I reject step 1 of error theory in this case.
Likewise, if someone thinks a presupposition of moral absolutism or motivational internalism is essential to a judgment’s status as a ‘moral’ judgment, I’ll be happy to deny both moral absolutism and motivational internalism, but I wouldn’t call myself an error theorist because I reject the claim that moral judgments (by definition, by conceptual analysis) necessarily presuppose moral absolutism or motivational internalism.
But hey, if you convince me that the presumption of motivational internalism in moral discourse is so widespread that talking about ‘morality’ without it would be like using the term ‘phlogiston’ to talk about oxygen, then I’ll be happy to call myself an error theorist, though none of my anticipations will have changed.
Hypothetical imperatives
I’ll reply to a passage from your post on hypothetical imperatives. My reply won’t make sense to those who haven’t read it:
I think this is because ‘should’ is being used in different senses. The real modus ponens is:
If you want to torture children, you should_ToTortureChildren volunteer as a babysitter.
You want to torture children.
Therefore, you should_ToTortureChildren volunteer as a babysitter.
Or at least, that is plausibly what some people mean when they assert what looks like a hypothetical imperative. Doubtless, others will appear to be meaning something else if pressed by interrogation.
Now, to respond to Sidgwick:
I could capture this ‘unreasonableness’ by simply clarifying that from the standpoint of Bayesian rationality, it would be somewhat irrational to expect good health despite not rising early (or so the doctor claims).
But again, I’m not too keen to play the definitions game. If you state hypothetical imperatives with more intuitions about the meaning of hypothetical imperatives than I do, then you are free to explain what you mean by hypothetical imperatives and then show how they fit into the physical world. If you can’t show how they fit into the world, then you’re talking about something that doesn’t exist, or else we’ll have to replay the physicalism vs. non-physicalism debate, which is another topic.
Right?
Thanks for this reply. I share your sense that the word ‘moral’ is unhelpfully ambiguous, which is why I prefer to focus on the more general concept of the normative. I’m certainly not going to stipulate that motivational internalism is true of the normative, though it does seem plausible that there’s something irrational about someone who acknowledges that they really ought (all things considered) to phi and yet fails to do so. (I don’t doubt that it’s possible for someone to form the judgment without any corresponding motivation though, as it’s always possible for people to be irrational!)
I trust that we all have a decent pre-theoretic grasp of normativity (or “ought-ness”). The question then is whether this phenomenon that we have in mind (i) is reducible to some physical property, and (ii) actually exists.
Error theory (answering ‘no’ and ‘no’ to the two questions above) seems the most natural position for the physicalist. And it sounds like you may be happy to agree that you’re really an error theorist about normativity (as I mean it). But then I’m puzzled by what you take yourself to be doing in this series. Why even use moral/normative vocabulary at all, rather than just talking about the underlying natural properties that you really have in mind?
P.S. What work is the antecedent doing in your conditional?
Why do you even need the modus ponens? Assuming that “should_ToTortureChildren” just means “what follows is an effective means to torturing children”, then isn’t the consequent just plain true regardless of what you want? (Perhaps only someone with the relevant desire will be interested in this means-ends fact, but that’s true of many unconditional facts.)
Right.
Unfortunately, I don’t think I’m clear about what you mean by normativity. The only source of normativity I think exists is the hypothetical imperative, which can be reduced to physics by straightforward methods such as the one I used in the original post. I’m not an error theorist about that kind of normativity.
This is a good question. Truly, I want to get away from moral vocabulary, and be careful around normative vocabulary. But people already think about these topics in moral and normative vocabulary, which is why I’m trying to solve or dissolve (in this post and its predecessor) some of the usual ‘problems’ in this space of questions.
After that’s done, I don’t think it will be most helpful to use moral language. This is evident in the fact that in 15 episodes of my ‘morality podcast’ I’ve used almost no moral language at all.
Not much, really. I wasn’t using the modus ponens to present an argument, but to unpack one interpretation of (some) ‘should’ discourse. Normative language, like many other kinds of language, is (when used correctly) merely a shortcut for saying something else. I can imagine a language that has no normative language at all. In that language we couldn’t say things like “If you want to torture children, you should volunteer as a babysitter” but we could say things like “If you volunteer as a babysitter you will have more opportunities to torture children.” The way I’m parsing ‘should’ in the first sentence, nothing is lost by this translation.
Of course, people use ‘should’ in a variety of ways, some of which translate into claims about things reducible to physics, others of which translate into claims about things non-reducible to physics, while still others don’t seem to translate into cognitive statements at all.
Thanks, this is helpful. I’m interested in your use of the phrase “source of normativity” in:
This makes it sound like there’s a new thing, normativity, that arises from some other thing (e.g. desires, or means/ends relationships). That’s a very realist way of talking.
I take it that what you really want to say something more like, “The only kind of ‘normativity’-talk that’s naturalistically reducible and hence possibly true is hypothetical imperatives—when these are understood to mean nothing more than that a certain means-end relation holds.” Is that right?
I’d then understand you as an error theorist, since “being a means-end relationship”, like “being red”, is not even in the same ballpark as what I mean by “being normative”. (It might sometimes have normative importance, but as we learn from Parfit, that’s a very different thing.)
My thought process on sources of normativity looks something like this:
People claim all sorts of justifications for ‘ought’ statements (aka normative statements). Some justify ought statements with respect to natural law or divine commands or non-natural normative properties or categorical imperatives. But those things don’t exist. The only justification of normative language that fits in my model of the universe is when people use ‘ought’ language as some kind of hypothetical imperative, which can be translated into a claim about things reducible to physics. There are many varieties of this. Many uses of ‘ought’ terms can be translated into claims about things reducible to physics. If somebody uses ‘ought’ terms to make claims about things not reducible to physics, then I am suspicious of the warrant for those claims. When interrogating about such warrants, I usually find that the only evidences on offer are pieces of folk wisdom, intuitions, and conventional linguistic practice.
You still seem to be conflating justification-giving properties with the property of being justified. Non-naturalists emphatically do not appeal to non-natural properties to justify our ought-claims. When explaining why you ought to give to charity, I’ll point to various natural features—that you can save a life for $500 by donating to VillageReach, etc. It’s merely the fact that these natural features are justifying, or normatively important, which is non-natural.
Sure. So what is it that makes (a) [the fact that you can save a life by donating $500 to VillageReach] normatively justifying, whereas (b) [the fact that you can save a mosquito by donating $2000 to SaveTheMosquitos] is not normatively justifying?
On my naturalist view, the fact that makes (a) but not (b) normatively justifying would be some fact about how the goal we’re discussing at the moment is saving human lives, not saving mosquito lives. That’s a natural fact. So are the facts about how the English language works and how two English speakers are using their terms.
It’s not entirely clear what you’re asking. Two possibilities, corresponding to my above distinction, are:
(1) What (perhaps more general) normatively significant feature is possessed by [saving lives for $500 each] that isn’t possessed by [saving mosquitoes for $2000 each]? This would just be to ask for one’s fully general normative theory: a utilitarian might point to the greater happiness that would result from the former option. Eventually we’ll reach bedrock (“It’s just a brute fact that happiness is good!”), at which point the only remaining question is....
(2) In what does the normative signifiance of [happiness] consist? That is, what is the nature of this justificatory status? What are we attributing to happiness when we claim that it is normatively justifying? This is where the non-naturalist insists that attributing normativity to a feature is not merely to attribute some natural quality to it (e.g. of “being the salient goal under discussion”—that’s not such a philosophically interesting property for something to have. E.g., I could know that a feature has this property without this having any rational significance to me at all).
(Note that it’s a yet further question whether our attributions of normativity are actually correct, i.e. whether worldly things have the normative properties that we attribute to them.)
I gather it’s this second question you had in mind, but again it’s crucial to carefully distinguish them since non-naturalist answers to the first question are obviously crazy.
Yup. I’m asking question (2). Thanks again for your clarifying remarks.
What if you actually should be discussing saving of mosquito lives, but don’t, because humans are dumb?
I think this is a change of subject, but… what do you mean by ‘actually should’?
No idea.
I take you to mean “what would maximize Luke’s utility function” (knowing that ‘utility function’ is probably just a metaphor when talking about humans) when you say “you actually should...” Of course, my ‘utility function’ is unknown to both of us.
In that case, it would remain true in our hypothetical scenario that I should-HumanLivesAreGood donate to VillageReach (assuming they’re a good charity for saving human lives), while I should-UtilityFunctionLuke donate to SaveTheMosquitos.
(Sorry about formatting; LW comments don’t know how to use underscores, apparently.)
But the question then is what goal you should have. It is easy to naturalise norms inasmuch as they are hypothetical and indexed to whatever you happen to be doing. (if you want to play chess, you should move the bishop diagonally) The issue is how to naturalise categorical ends,the goals you should have and the rules you should be following irrespective of what you are doing.
Such facts aren’t supernatural. OTOH, they fall on the analytical/apriori side of the fence, rather than the empirical side, and that is an iimportant distinction.
What reasons are there for doubting the existence of categorical imperatives that do not equally count against the existence of hypothetical imperatives? I can understand rejecting both, I can understand accepting both, but I can’t understand treating them differently.
Depends what you mean by ‘categorical imperative’, and what normative force you think it carries. The categorical imperatives I am used to hearing about can’t be reduced into the physical world in the manner by which I reduced a hypothetical imperative into the physical world in the original post above.
I take it you want to reduce a hypothetical imperative like “If you want to stretch, then you ought to stand” into a physically-kosher causal claim like “standing is causally necessary for satisfying your desire to stretch”. Now, I’m skeptical about this reduction, simply because I don’t see how a mere causal claim could provide any normative direction whatsoever. But in any case, it seems that you could equally well reduce a categorical imperative like “Regardless of what you personally want, if you are the only one around a drowning child, then you ought to help save it from drowning” into a physically-kosher causal claim like “your help is causally necessary for the survival of the drowning child”. Both causal claims are equally physicalistic, and both seem equally relevant to their respective imperative, so both physicalistic reductions seem equally promising.
Of course, the question “why should I assume that the drowning’s child survival actually matters?” seems reasonable enough. But so does the question “why should I assume that satisfying my desire to stretch actually matters?” If the first question jeopardizes the reduction of the categorical imperative, then the second question would also jeopardize the reduction of the hypothetical imperative.
I’m not used to the term ‘categorical imperative’ being used in this way. Normally, a categorical imperative is one that holds not just regardless of what you want, but regardless of any stated ends at all. Your reduction of this categorical imperative assumes an end of ‘survival of the drowning child.’ To me, that’s what we call a hypothetical imperative, and that’s why it can be reduced in that way, by translating normativity into statements about means and ends.
I think you must be mistaken about categorical imperatives and ends.
In the Groundwork, Kant deems rational nature (‘humanity’ in us) to be an end in itself which can ground a categorical imperative, in the 2nd Critique, he deems the highest good (happiness in proportion to virtue) to be a necessary end of practical reason, and in the Metaphysics of Morals, he deems one’s own perfection and the happiness of others to be ends that are also duties. For Kant, the categorical imperative of morality is directed at ends, but it is not a mere hypothetical imperative grounded in subjective ends.
And leaving Kant aside, plenty of moral systems aspiring to categorical force have ends at their center. Classic utilitarianism picks out the greatest overall balance of pleasure over pain as the one and only end of morality. Eudaimonist virtue ethics has as its end a well-lived flourishing life for the agent. Thomist ethics says an intellectual vision of God in the afterlife is the ultimate end of humans. These systems do not traffic in merely hypothetical imperatives: they present these ends as objectively worth pursuing, regardless of the agent’s personal preferences.
But if your main point is simply that my ‘reduction’ assumes that the end ought to be pursued, and that this assumption is a form of cheating, then I agree. I’ve left the normativity unexplained and unreduced. But then in exactly the same way, your reduction of hypothetical imperatives assumes that effective means to one’s ends ought to be taken, and this assumption is also a form of cheating. You have also left the normativity unexplained and unreduced.
So I don’t see how hypothetical imperatives are any more fit for naturalistic reduction than categorical imperatives.
I’m sure I speak differently about categorical imperatives than Kant does. I haven’t read much of Kant, and I don’t regret that. In your language, what you call an “end in itself” is what I mean to pick out when I talk about an imperative that “holds… regardless of any stated [arbitrary, desired, subjective] ends at all.” I don’t really know what it means for something to be an “end in itself”. Kant’s idea seemed to be that we ought to do X regardless of what anyone wants. Your way (and perhaps Kant’s way) of talking about this is to say that we ought to do X regardless of what anyone wants because X leads to a particular “end in itself”, whatever that means.
My reduction of hypothetical imperative doesn’t assume this. It only translates ‘ought’ into a prediction about what would realize the specified end. If there’s something mysterious left over, I’m curious what you think it is and whether it is real or merely a figment of folk wisdom and linguistic practice.
So, on your use of ‘end’, an ‘end’ cannot be objective and unconditional? I think that’s a highly uncommon use of the term.
But if you go this way, it seems like it’s less of a reduction of ‘ought’ and more of a misinterpretation, like reducing ‘Santa Claus’-talk into talk about Christmas cheer, or ‘God’-talk into talk of love.
After all, one important constraint on any interpretation of any ‘ought to X’ is that it should be positive towards X as opposed to negative or neutral, in favor of some action or attitude as opposed to against it or indifferent. But a mere predictive causal claim doesn’t have any valence at all: it’s just a neutral claim about what will probably lead to what, without anything positive or negative. So any attempt to reduce oughts to predictive causal claims seems doomed to failure.
EDIT: For the record, I’m an expressivist about normativity, and I think any attempt to understand it in terms of some actual or hypothetical ontology that could serve as the truth-conditions for a descriptive belief is a mistake. The mystery, I would say, lies in a descriptive interpretation of normativity, not in normativity itself.
No, I just should have been clearer that when I said “stated end” I meant “subjective, desired end”. As commonly used, ‘end’ includes unconditional ends, I just haven’t ever been presented with an argument that persuaded me to think that such ends exist.
Sure, you can stick that feature into your meaning of ‘ought’ if you want. But I’m not going too deep into the conceptual analysis game with you. If you want to include positive valence in the meaning of a hypothetical ought, then we could translate like this:
“If you want to stretch, you ought to stand” → “standing will make it more likely that you fulfill your desire to stretch, and I have positive affect toward you standing so as to fulfill your desire to stretch”
As I said in my post, expressivism is fine—and certainly true of how many speakers use normative language. I happen to be particularly interested in those who use normative language cognitively, and whether their stated moral judgments are true or false, and under which conditions—which is why I’m investigating translations/reductions of normative language into natural statements that have truth conditions.
What do you think human wellbeing is for,then?
The set of non-ethical categorical imperatives is non-empty. The set of non-ethical hypothetical imperatives is non-empty. Hypothetical imperatives include instrumental rules, you have to use X to achieve Y, game-laying rules, etc.
How exactly does this answer the question?
I agree. Epistemic imperatives are categorical, but non-empty.
Right, those are examples where non-ethical hypothetical imperatives often show up.
So how does this add up to a reason to think there is a case against categorical imperatives that doesn’t equally well count against hypothetical imperatives?
My thought process on sources of normativity looks something like this:
There’s a lot of kinds of normative/”ought” statements. Some relate to games, some to rationality, and so on. Hypothetical “ought” statements do not require any special metaphysical apparatus to explain them, they just require rules and payoffs. Categorical imperatives are another story.
One man’s conventional linguistic practice is another’s analytical truth.
Rules and payoffs explain “ought” statements only if you assume that the rules are worth following and the payoffs worth pursuing. But if hypothetical imperatives can help themselves to such assumptions (assuming e.g. that one’s own desires ought to be satisfied), then categorical imperatives can help themselves to such assumptions (assuming e.g. that everyone’s desires ought to be satisfied, or that everyone’s happiness ought to be maximized, or that everyone ought to develop certain character traits).
I don’t think so. You ought to use a hammer to drive in nails even if you don’t want to dive in nails. Anyone who is playing chess should move the bishop diagonally.That doesn’t mean you are playing chess.
Of course those are hypothetical, and non-ethical. It might wll be the case that the only categorical imperatives are moral categorical imperatives; that. ethics is the only area where you should do things or refrain form things unconditionally.
Again, you’re assuming that the rule ‘if you’re driving in nails, use a hammer’ is worth following, and that the rule ‘if you’re playing chess, move bishops diagonally’ is worth following. A nihilist would reject both of those rules as having any normative authority, and say that just because a game has rules it doesn’t mean that game-players ought to follow those rules, at most it means that lots and lots of rule-violations make the game go away.
I don’t think hypothetical imperatives can be reduced. The if-ought of a hypothetical imperative is a full-blooded normative claim. But you can’t reduce that to a simple if-then about cause and effect.
To see why, consider a nihilist about oughts. She recognizes the causal connections between calorie consumption/burning and weight loss. But she doesn’t accept any claim about what people ought to do, even hypothetical imperatives about people who desire weight loss. This seems perfectly coherent: she accepts causal claims, but not normative claims, and there’s no contradiction or incoherence there. But this means the causal claims she accepts are not conceptually equivalent to the normative claims she rejects.
For another way to see why, consider the causal claim “less calorie consumption and more calorie burning leads to weight loss”. This causal claim points in no normative direction. It doesn’t recommend anything, or register any approval, or send any positive or negative messages. Of course, we can take it in one direction or another, but only by combining it with separate normative claims:
Less calorie consumption and more calorie burning leads to weight loss.
People ought to take causally efficacious steps to satisfy their desires.
Therefore, if you desire to lose weight, you ought to consume less calories and burn more calories.
Premise 2 is what provides the normativity. It points us in the direction of satisfying desires. But we could easily take things in the opposite direction.
Less calorie consumption and more calorie burning leads to weight loss.
People ought to take causally efficacious steps to frustrate their desires.
Therefore, if you desire to lose weight, you ought to consume more calories and burn less calories.
Again, premise 2 is what provides the normativity. But it points us in the opposite direction, viz. the direction of frustrating desires.
So it’s pretty clear that premise 1 has no normativity in it. It can’t be reduced to either of the two 3′s. For we cannot arrive at a 3 without a 2.
I think stating premise 2 is a little odd. It is a bit deja “Tortoise and Achilles” all over again. If there’s a norm hiding around here, it’s an “ought” portrayed by the desire.
Second, conceptual analysis (or conceptual equivalence) is not necessary for reduction. Look at reduction in the sciences for examples.
Well, I’ll acknowledge that you could change premise 2 into an inference rule. But notice that you could change either premise 2—the pro-desire-satisfaction one and the pro-desire-frustration one—into an inference rule. Indeed, you could change any normative claim into an inference rule: you could change “people who want to have gay sex ought to go see a trained Baptist minister to get cured” into an inference rule, and then validly go from “I want to have gay sex” to “I ought to go see a trained Baptist minister to get cured”. So from the fact that premise 2 could be changed into an inference rule, I don’t think anything follows that might jeopardize its status as a full-blooded normative claim.
On the second point, I thought lukeprog was discussing direct conceptual reduction. But if he wants to provide hypothetical imperatives with a synthetic reduction, he’ll need a theory of reference capable of explaining why the normative claim turns out to make reference to (and have its truth-conditions provided by) simple causal facts. And on this score, I think hypothetical imperatives and categorical moral imperatives are on an equal footing: since reductionist moral realists have a hard time with synthetic reductions, I would expect reductionist ‘instrumental realists’ to have a hard time as well.
For what it’s worth, I think what’s really being inferred by the advice-giver is:
1 Granting your (advisee’s) starting point, you ought to lose weight.
2 Less calorie consumption and more calorie burning leads to weight loss.
3 Therefore you ought to consume less and burn more calories.
The advisee’s desire portrays the starting-point as a truth.
Perhaps so, but then the normativity stems from premise 1, leaving premise 2 as non-normative as ever. But the question is whether premise 2 could be a plausible reduction basis for normative claims.