Ethical Naturalism reredux

by Brian on July 30, 2003

A long and winding post responding to some issues about morality and naturalism.


I still agree with Lawrence Solum that we can make a start on getting from natural facts to ethical facts by looking at purposes. Matt Evans responds by saying that naturalists deny nature has a purpose. True enough, but not much follows from that.

A little analogy. Consider a crowd in Times Square on a typical work day. The crowd as a whole has no discernable purpose whatsoever. It isn’t like New Year’s Eve where the purpose is unintelligible, there just isn’t a group purpose there. But the individuals in the group can have purposes. One might be looking for food, another for theatre tickets, and another (I’m told) for where all the porn shops have gone. It might be disheartening to think of all of nature as an aimless Times Square crowd write large, but even if we do, that doesn’t entail that none of the constituent parts have purposes.

There’s another reason naturalists should take purpose seriously. Scientists, at least as far as I can tell, take purpose seriously. In many diverse areas, functional explanation is a core part of the toolkit. And talk of function is just talk of purpose in a (minimally) different guise. Crude example: saying the function of the heart is to pump blood is barely different to saying its purpose is to pump blood. And that’s the kind of thing scientists will readily endorse. (They’ll even sometimes say the function of a part of the body is to transmit information, involving themselves in philosophical mysteries concerning function and concerning content within one little claim. No matter, science still works.) Since naturalists take science seriously, indeed take science to be largely constitutive of what is ‘natural’, naturalists can also take purpose seriously.

_Factual-Moral Arguments_

Matt Evans posts the following challenge:

bq. Here’s the deal: I invite all Brights to email me the moral premises they accept not by blind faith, but because they are founded in nature. Please trace the moral premise to natural facts. Though submissions will be accepted in any format, syllogisms are especially appreciated. Show your work.

Well, none of the following examples are going to convince anyone who wants to remain unconvinced, but here’s three.

_Flurg_ (due to Gideon Rosen)

First a definition: _to flurg_ is to do something one ought not do in the presence of small children.

Now the argument.

bq. Jack is in the presence of small children.
Therefore, Jack ought not to flurg.

That’s valid, as far as I can see, and for appropriate values for ‘Jack’ it is even sound. The premise certainly looks like an ‘is’ statement and the conclusion like an ‘ought’ statement. So we’ve got it – an ought from an is! Or, in Matt’s terms, an argument from a natural premise to a moral conclusion.

_Moral Realism_

Here’s an even quicker argument that you can get from is to ought.

bq. Torturing babies is wrong.
Therefore, torturing babies is wrong.

The conclusion is definitely a moral claim. But what about the premise? Well, I’m a moral realist, so I think it’s a truth about the real (i.e. natural) world. I can see how some may disagree, but without a good definition of natural, and a good argument that the premise here is unnatural, I’m inclined to think this is a counterexample to the _no ought from an is_ principle. Still, to keep things interesting, I’ll not rely on this from here on.

_What’s an Argument?_

I think part of what’s lying behind Matt’s scepticism is a faulty conception of what an argument is. He thinks a good argument must be _formally_ valid. It must be valid in virtue of its syntactic form, just like what up-to-date logic classes teach, and indeed as out-of-date logic classes (like those Aristotle taught) teach. But a valid argument is just one where it’s impossible for the premises to be true and the conclusion false. And it’s impossible that humans be just as they are and it be morally permissible to torture infant children. Proof by contradiction: assume it is possible, then we should be able to coherently describe the situation. But as soon as we try it should be apparent that we’ve misdescribed it. (Seriously, try to describe a situation in which humans are just as they are in all natural respects and it’s morally acceptable to torture their young. You’ll soon get the impression that, whatever you try to stipulate, you haven’t told a story where it’s morally acceptable to torture children.) So such a situation is impossible. So the argument from premises about human nature to the conclusion that you shouldn’t torture children is valid.

_The Use of Concepts in Arguments_

Stuart Buck criticised my this argument that facts about human qualities and relationships entail moral facts as follows:

bq. He’s smuggling in a moral premise, namely the principle that it is wrong to torture (for one’s own amusement) creatures who are capable of feeling pleasure and pain, who have hopes and plans and fears and regrets, who are capable of great learning, and so forth. _Only with that principle_ in place does it make sense to say that you can derive a moral conclusion from the set of facts Weatherson lays out.

I don’t think that’s right. I think that the only sense in which we _need_ moral premises to get from human statements (as I’ll call them from now on) to moral statements is a sense in which those premises are fully acceptable to the naturalist. But to show that, or even to make a case for it, will require a couple of examples.

What I really want to defend is the following – we can be in a position to know moral facts given just human facts and possession of moral concepts. Stuart and Matt say the naturalist needs ‘faith’ to get moral conclusions, I say they just need moral concepts.

To see why this is not cheating, let’s think about an area where it is intuitively clear that you can get from lower-level facts to higher-level facts. For instance, an argument from the final scores of the only two competitors in a sporting contest to a conclusion about who won could be valid. If we can have a natural->moral argument that’s as good as the scores->victor argument, the naturalist is doing well.

So let’s say I know that X scored 264 and Y scored 279. Can I conclude who won? Well, not unless I know what it is to win the game in question. That is, not unless I possess the concept VICTORY for the salient game. If the game is cricket, then Y won, because the higher score wins. If the game is golf, then X won, because the lower score wins. If I have that concept I can make the inference, if not I can’t.

Might this be a good analogy for human-moral inferences? Perhaps, but perhaps not, because we can say exactly what the salient concept – in this case victory – amounts to. What’s distinctive about moral concepts is we can’t say that. Let’s try a different example that’s on all fours with the moral case then.

Here’s the career stats for two prominent baseball players, Mario Mendoza and Nomar Garciaparra. I think that anyone who has all the relevant concepts, and knows the natural facts given in those stats pages, is in a position to infer that Garciaparra is a better hitter than Mendoza. Note here that having the concept doesn’t mean knowing what it is to be a better hitter. I don’t know to any precise degree what it is to be a better hitter than someone else. It’s possible to have the concept and not know whether Garciaparra is a better hitter than, say, Lefty O’Doul. But I know that Garciaparra is better than Mendoza was, and anyone who looks at those numbers and disagrees probably doesn’t have the concept ‘better hitter’.

So that’s a case where just some facts, and a concept, suffice for a normative conclusion. In this case it isn’t a very deep conclusion, one about who is a better hitter not who is a better person. But I don’t think depth makes for a disanalogy between the cases. What was meant to be powerful about Hume’s principle was it blocked you getting from the descriptive to the normative. But that isn’t a block – as shown in the baseball case.

_Cohen’s Argument_

As Chris noted earlier, the argument of Gerry Cohen that Lawrence Solum discusses should apply generally if it applies in ethics. But it looks to me that unless we interpret Cohen as arguing for a _very_ weak principle, then the baseball example refutes his conclusion. Just looking at the numbers, I can conclude Garciaparra is a better hitter than Mendoza. Do I need a principle here? Well, if I had a principle like “If X has better numbers than Y in the following statistical categories (list here all the ways in which Garciaparra has better numbers than Mendoza) then X is a better hitter than Y” that would certainly help. (Although as Dsquared noted in the comments on Chris’s post, if someon really wanted to resemble a tortoise they could complain that I still needed a further principle to underpin the validity of the new argument.) But if I don’t have that principle the inference is blocked in a much more trivial sense. If I don’t have that principle I really don’t have the concept ‘better hitter’, so I can’t conclude anything about who is a better hitter.

I think there’s a very weak sense in which Cohen’s right – we certainly need principles of some kind to get from factual premises to moral conclusions. But that’s just because we need to (at least tacitly) accept some principles in order to have moral concepts, and (for completely trivial reasons) we need to have moral concepts to make inferences from factual premises to moral conclusions. This is trivial because it isn’t different from the claim that we need furniture concepts to make inferences from premises about the distribution of wood in my office to conclusions about where the tables and chairs are in my office. But no one really thinks that we can’t in any interesting sense make inferences from where the wood is to where the furniture is. If the relation between the descriptive and the normative is like the relation between the wood and the furniture, it’s hard to see a threat to naturalism, or to Rawlsian constructivism.



dsquared 07.30.03 at 6:14 pm

Two points:

First, if we define “torture” as “the purposeful infliction of pain in order to influence behaviour”, then the young of the human species are the only sentient beings which it *is* legal to torture under the law on “reasonable chastisement”. A cheap debating point, but not an irrelevant one.

Second, what if I say “Sure, Garciaparra’s the more effective hitter. But effectiveness in that sense is only part of what makes a good hitter. I don’t like the way Garciaparra hits; Mendoza is a * better * hitter.” I think I can reconstruct the Open Question argument in this case without too many problems, and I don’t see how you can deny that I have the concept of a “better hitter” without assuming something equivalent to your conclusion (ie, that goodness of hitters is constituted by a particular set of facts).

For example, Bob Hope might get more laughs, but I think Sammy Davis Jr was a better comedian, say.


Brian Weatherson 07.30.03 at 7:01 pm

I agree there’s always an open question response available in principle to any moral claim (or probably any normative claim) but I think there are two possible things to say in response to that kind of challenge.

First, such a response does start to get a little strained after a while. X says “Given the facts involved, crashing planes into the World Trade Centre was not morally permissible.” Y says, “Well hang on, I know it involved killing 3000 people who hadn’t harmed the killers in any way and didn’t threaten them, but does that show it wasn’t morally permissible?” Eventually these questions stop sounding particularly open, at least to me. (Though of course I’m a moral naturalist, so maybe I’m just biased here.)

Less loaded example. X says “It was unreasonable for Jack to conclude that the moon is made of green cheese, given that he flew on a lunar mission, tried to eat part of the moon and failed, and has been told repeatedly by experts that it isn’t made of green cheese.” Y says, “Well, I know all those facts, but isn’t it still an open possibility that Jack is being _reasonable_ here.” Some hard-core Bayesians may have sympathy with Y here, but not many other people will.

Second, if the baby torture case doesn’t work, just pick any action that you think is clearly morally impermissible. (This argument won’t work against committed moral sceptics. But I don’t think there are many of those around here. If anything people, me included, are too quick to read moral significance into things.) Now try and describe a world that is naturally just like this one, but where that action is morally permissible. I don’t see how such a world could be possible. (Small and I think irrelevant exception to this claim at end.)

Now this is a pretty sweeping generalisation, and my evidence for it is probably too small to really support it, but there is meant to be an argument for the claim that moral facts are constituted by natural facts. That was meant to be a conclusion not a premise. In rawest form, the argument is just

1. If moral facts are not constituted by natural facts then it should be possible to change the moral facts while keeping the natural facts the same.
2. It’s not possible to change the moral facts while keeping the natural facts the same.
C. Moral facts are constituted by natural facts.

Premise 1 relies on contentious metaphysics and premise 2 is a sweeping generalistion, but apart from that I think this is a pretty good argument!

There’s some evidence for premise 2 from an odd puzzle about stories. Authors can’t just make moral claims true in their stories, once they’ve fixed the natural facts. If moral facts weren’t constituted by natural facts, this would be surprising. For instance, I think we intuitively take the following story to be contradictory:

bq. Jack and Jill had a fine old time up on top of that hill. Eventually, though, Jack tired of
her and her whining. So he strangled her and left her body out in the open for the
vultures. Jack shouldn’t have left her body unburied, but he was right to kill her. She was

The example is from a paper by Neil Levy, Imaginative Resistance and the Moral/Conventional Distinction. I think the author doesn’t make it true in the story that Jack did the right thing in killing Jill. And I think the best explanation for that is that moral concepts aren’t primitive – if something is right or wrong that’s a fact that holds in virtue of other, presumably natural, facts.

(Small exception to my generalisations like premise 2: If we add enough angels to the world, and make it the case that angels have morally significant characteristics – they feel pleasure and pain, fall in love, etc. – and somehow add in a background fact that certain natural actions have effects on the angel world, then it probably is possible to change the moral status of natural actions by changing their angelic effects. If eating pizza causes 10 angels to die in agony, then eating pizza is immoral, perhaps. I don’t think that’s what people have in mind when they say the non-natural can effect what is moral and immoral, but it’s a possibiility. So strictly speaking premise 2 needs to be qualified somehow to not rule out this possibility. I don’t know how to write the qualification right now, but I think it’s doable in a way that doesn’t undercut the main thrust of the argument.)


Matt Evans 07.30.03 at 7:04 pm

Hi Brian,

Your “flurg” example is wrong, as I trust you already know that including the contested “ought” inside the definition of “to flurg” doesn’t save the sentence.

As for Garciaparra and Mendoza example, your “better hitter” again rests on a teleological claim and inferred purpose. Only because you know that the purpose of a batter is to get on base and score runs are you able to infer anything from their statistical facts.

– Matt Evans


Jack 07.30.03 at 7:17 pm

Or to continue with Mr Squared’s example, David Gower was a better batsman than Graham Gooch: true and totally unsupported by the stats. Is choice of sporting examples the real difference between British and American philosophy?

Less facetiously Matt Evans wouldn’t have to be that bloody minded to object to the first two examples would he? The flurg example just slips the blind faith into a definition. The second starts with the blind faith. The third I think has most potential but might be used more aggressively. As an example use a situation which is new, say, is it moral to divorce my wife by mobile phone text message? There is no chance that any of the ancient religions or other moral codes explicitly tell you what to do in this situation.
Now suppose that you have morals and they are not naturally based, how do you get to a solution? The naturalist response would then be something like “well I apply the same reasoning that fills in the gaps in your system to every situation including the ones that are for you an article of faith because I think your view is inevitably crude, and impossible for you to interpret without your coming into my world”. In other words, nature is sufficient for me and necessary for you.
I think the real problem is an excluded middle, just because you can’t easily get an ought from is doesn’t mean that you must be able to get it from somewhere else.
PS I like the new argument but I think that it begs some epistemological questions which mean that you might still be stuck trying to deal with Matt’s challenge. Indeed unless there is a stone somewhere with everything on or your nose grows when you do something immoral you’ll have a hard time because he wants to be able to get to a point where he will be convinced in a finite number of steps whereas the naturalist might prefer to work back from the issue iuntil satisfied, ie the naturalist says “is that far enough?” and will go further if you say no whereas Matt is saying “are we there yet”?


dsquared 07.30.03 at 7:32 pm

Brian: But have a look at some of the things that intelligent and reasonable people on the political left have said about Pol Pot, Mao, Stalin, etc. I think that the open question objections are much more open than you think. As a matter of fact, given the facts about the political history of the 20th century in the USA and Middle East, a substantial number of people who cannot be regarded as insane or arguing in bad faith (I think they’re wrong, but that’s by the by) did in fact think that the September 11th attacks were justified. Christopher Hitchens can dig up any number of examples.

In re: green cheese, my instinct is that you’ve just dug up an example of why I’m never particularly convinced by arguments which move between non-moral normative contexts and moral ones. What a “reasonable man” would think is part of a factual description of the universe; I think it’s perfectly acceptable for a non-naturalist to argue that it falls on the other side of Hume’s dichotomy, and I believe that this cut is actually what one sees in legal contexts.


Brian Weatherson 07.30.03 at 7:42 pm

I know the flurg example looks like it’s cheating, but it’s not exactly clear _how_ it is cheating. After all, the argument just as it stands is valid. To know it is valid you have to know the meanings of the terms involved, but to know that a syllogism is valid you have to know the meanings of the terms involved too, and I don’t see anyone not called ‘Tortoise’ criticising syllogisms on that ground.

Which all comes back to the main point – none of these debates make much sense until we know just exactly what is meant to count as a good argument. I think at least two definitions are in play:

(1) Formal validity – validity in virtue of meanings of logical constants.
No argument from is to ought that way, but who cares? There’s no valid argument from where the wood is to where the desk is this sense either, and naturalists presumably have no trouble talking about desks.

(2) Modal validity – impossible for premise to be true and conclusion false.
Here it seems there is an argument from is to ought – Gideon’s flurg example. (And several other examples give above.)

Now if the flurg example makes you think (2) is too liberal, let’s know what the real criteria for validity is that’s being used, and I’ll try and come up with an example or

By the way, the baseball example wasn’t meant to be like Gooch/Gower. I agree that sometimes there can be other things than stats. But when it just isn’t even close, as with the two players I picked, I think stats can be sufficient. The point there wasn’t that better stats are necessary to be a better player, just that in extreme cases they can be sufficient. Example: I don’t care what how good they looked, just knowing the stats tells me that Gooch was a better batsman than, say, Bob Willis, or indeed any recent quick not named Botham.


thepublicguy 07.30.03 at 7:43 pm

I really don’t think talk of function is talk of purpose in a minimally different guise at all, if you accept the following definitions.

Function = This is what it does
Purpose = This is what it is intended to do (inanimate or animate object) or this is what it intends to do (animate object only).

If you accept this definition of purpose, then who is doing the intending? A God, an evolutionary process that has a mind, a human being’s parents or other ancestors?

I think you either have to reject the idea that people have been given a purpose or indicate who gave these people the purpose.

Of course you could say that people can choose their own life purpose or moment-to-moment purpose, but in this context that would lead you to say either that whatever advances a person’s chosen purpose is morally good regardless of the purpose they choose or that there is some other criteria for whether what advances their purpose is good which then needs alternative justification.

On flurging.

I’m no philosopher and may not have my terms right here. But to flurg appears to be a concept with an intension whose extension has yet to be defined. But what you are trying to show is that you can derive an extension for this term from natural facts and you haven’t done it yet.

Sure you have shown that you can define a term that means something you shouldn’t do around children and then clearly show that when around children, if you accept there are some things you shouldn’t do around children, then you shouldn’t do them. But the unanswered question is what are the things that you shouldn’t do around children. Hypothetically, if not in any real-life moral system, there may not be any things that one shouldn’t do around children.

On torture.

As for the argument that torturing babies is wrong, therefore torturing babies is wrong, that strikes me as a circular argument. All the books I’ve seen on good arguments say the conclusion shouldn’t form part of the premises.

On arguments

If a thing is so we should be able to coherently describe it, you say. What if we don’t currently have the conceptual apparatus to do so? But accepting the argument for the time being what about the ritual sacrifice of children in religious ceremonies? People may believe that nasty rituals involving cruelty to children will ward off harm from evil spirits. But you may say that they cannot correctly believe this.

Nevertheless, there’s a bit of shifting the burden of proof going on here. What you appear to be saying is that if someone cannot show ~P, then P is a natural fact. Not so. I cannot coherently show that there isn’t a God that fits a certain limited conception of Godhood . Does that mean God necessarily exists?


Which definition of good to use – whether highest-scoring or some other – remains an open question as noted in a previous comment. Of course, the particular definition of good intended may be apparent from implicature.

On the need to ground morals objectively.

You don’t. Who says a subjective moral system is valueless? Having a moral system is an inevitable part of being human. What’s so bad about about adopting axioms, based on the kind of society one wants to live in, that don’t stand on concrete foundations?


Jeremy Osner 07.30.03 at 8:38 pm

“Second, if the baby torture case doesn’t work, just pick any action that you think is clearly morally impermissible.”

It seems to me this is where the big problem creeps in. Take the case of the person who is a committed advocate of female circumcision. Now suppose this person picks female circumcision as his example and goes on to show that there is no possible story to be told about human reality in which it is credibly morally permissible not to cut off your daughter’s labia. Now I can point out that I find such a story perfectly credible — but what will that avail me? The person constructing the morality is in charge — I don’t think there is any room here for communication of morality across such cultural boundaries.


Shai 07.30.03 at 9:39 pm

On a related note I received an automated email today about a review of Michael Rea’s “World Without Design: The Ontological Consequences of Naturalism” in Notre Dame Philosophical Reviews. Rea posted here on Crooked Timber re Dennett on Brights last week iirc.


Shai 07.31.03 at 3:31 am

I think thepublicguy is right about functions. Saying what a thing is for is just a way of making sense of what a thing does. An artificial heart might not do everything the clump of tissue we normally call a heart does, and if that’s the case it may affect what the circulatory system or immune system does; i.e. their function or functioning.

Daniel Dennett is more specific, he distinguishes between the physical stance (predicting and explaining the behaviour of a system from knowledge of its physical constitution), the design stance (predicting and explaining the behaviour of a system from knowledge of its design, or knowledge of its functional organisation — the designer would be a person, or natural selection), and the intentional stance (work out the goals and beliefs a system ought to have, given its basic nature, place in the world, and its needs — e.g. jack did x because he believed y) _1_

Also see “functional explanation” in the Routledge Encyclopedia of Philosophy, and “Progress: A philosophical Analysis” in the Encyclopedia of Life Sciences (also here

*_1_ defs. borrowed from here


Shai 07.31.03 at 4:54 am

Since no one else is posting, I suppose I’ll continue. Dennett has a nice story that complements Brian’s Times Square example:

“What is remarkable about the Boston Symphony Orchestra (and the myriad other human institutions and practices) is that, on the one hand, they can be so beautifully designed and organized, so self sustaining, while, on the other hand, they are composed of a motley assortment of autonomous individuals, of different nationalities, ages, genders, temperaments, aspirations. The orchestra members are free to come and go as they choose, so the board of directors must work hard to ensure that the working conditions and pay are sufficient to keep the orchestra members well motivated. Look at the violin section. Twenty talented individuals, but all different. Some are brilliant but lazy while others are obsessive perfectionists; one is bored but conscientious, another is enraptured by the music, yet another is daydreaming about making love to that adorable cellist over there, but all of them are drawing their bows across their strings in perfect unison, a pattern robustly superimposed on a kaleidoscope of different human consciousnesses. What makes this concerted action possible is a massive complex of cultural products, deeply shared by the musicians, the audience, the composer, the conservatories, the banks, the municipal authorities, the violin-makers, the ticket agencies, and so on. Nothing in the animal world is a close counterpart to this complexity. Human minds are furnished — and beset — by thousands of anticipations, evaluations, projects, schemes, hopes, fears, and memories that are entirely inaccessible to the minds of even our closest relatives, the great apes. This world of human ideals and artifacts gives individual human beings capacities and proclivities that are strikingly different from those of any other living beings on the planet.” (Freedom Evolves, p. 142-3)


Shai 07.31.03 at 9:02 am

Actually there are some excellent articles on this topic in the Routledge Encyclopedia of Philosophy: “naturalism in ethics”, “values, ontological status of”, “naturalistic fallacy”, and “moral realism”. Unfortunately they aren’t free. Via the ethics signpost article written by Roger Crisp.


Richard 07.31.03 at 12:13 pm

I’m left wondering whether this argument might not be better off trying to consider the feasibility of the ethics Matt advocated, namely those derived from an extra-natural source. It just seems to me that founding notions of ethics upon what is essentially the ghost in the machine is arbitrary at best. It might well, to use Matt’s original example, provide a reason for why one should not kill jews, but self evidently could as easily provide a reason for why one should do exactly that (thuggee cults if one wants a simple example).


Stuart Buck 07.31.03 at 3:30 pm

I respond further here.


Glenn Condell 08.06.03 at 8:22 am

‘a substantial number of people who cannot be regarded as insane or arguing in bad faith (I think they’re wrong, but that’s by the by) did in fact think that the September 11th attacks were justified.’

Understandable in context yes. Justified no. There’s a difference.

Comments on this entry are closed.