Pluralistic Moral Reductionism



Part of the sequence: No-Nonsense Metaethics

Disputes over the definition of morality... are disputes over words which raise no really significant issues. [Of course,] lack of clarity about the meaning of words is an important source of error… My complaint is that what should be regarded as something to be got out of the way in the introduction to a work of moral philosophy has become the subject matter of almost the whole of moral philosophy...
Peter Singer

If a tree falls in the forest, and no one hears it, does it make a sound? If by 'sound' you mean 'acoustic vibrations in the air', the answer is 'Yes.' But if by 'sound' you mean an auditory experience in the brain, the answer is 'No.'

We might call this straightforward solution pluralistic sound reductionism. If people use the word 'sound' to mean different things, and people have different intuitions about the meaning of the word 'sound', then we needn't endlessly debate which definition is 'correct'.1 We can be pluralists about the meanings of 'sound'.

To facilitate communication, we can taboo and reduce: we can replace the symbol with the substance and talk about facts and anticipations, not definitions. We can avoid using the word 'sound' and instead talk about 'acoustic vibrations' or 'auditory brain experiences.'

Still, some definitions can be wrong:

Alex: If a tree falls in the forest, and no one hears it, does it make a sound?

Austere MetaAcousticist: Tell me what you mean by 'sound', and I will tell you the answer.

Alex: By 'sound' I mean 'acoustic messenger fairies flying through the ether'.

Austere MetaAcousticist: There's no such thing. Now, if you had asked me about this other definition of 'sound'...

There are other ways for words to be wrong, too. But once we admit to multiple potentially useful reductions of 'sound', it is not hard to see how we could admit to multiple useful reductions of moral terms.

Many Moral Reductionisms

Moral terms are used in a greater variety of ways than sound terms are. There is little hope of arriving at the One True Theory of Morality by analyzing common usage or by triangulating from the platitudes of folk moral discourse. But we can use stipulation, and we can taboo and reduce. We can use pluralistic moral reductionism2 (for austere metaethics, not for empathic metaethics).

Example #1:

Neuroscientist Sam Harris: Which is better? Religious totalitarianism or the Northern European welfare state?

Austere Metaethicist: What do you mean by 'better'?

Harris: By 'better' I mean 'that which tends to maximize the well-being of conscious creatures'.

Austere Metaethicist: Assuming we have similar reductions of 'well-being' and 'conscious creatures' in mind, the evidence I know of suggests that the Northern European welfare state is more likely to maximize the well-being of conscious creatures than religious totalitarianism.

Example #2:

Philosopher Peter Railton: Is capitalism the best economic system?
Austere Metaethicist: What do you mean by 'best'?
Railton: By 'best' I mean 'would be approved of by an ideally instrumentally rational and fully informed agent considering the question ‘How best to maximize the amount of non-moral goodness?' from a social point of view in which the interests of all potentially affected individuals are counted equally.
Austere Metaethicist: Assuming we agree on the meaning of 'ideally instrumentally rational' and 'fully informed' and 'agent' and 'non-moral goodness' and a few other things, the evidence I know of suggests that capitalism would not be approved of by an ideally instrumentally rational and fully informed agent considering the question ‘How best to maximize the amount of non-moral goodness?' from a social point of view in which the interests of all potentially affected individuals were counted equally.

Example #3:

Theologian Bill Craig: Ought we to give 50% of our income to efficient charities?
Austere Metaethicist: What do you mean by 'ought'?
Craig: By 'ought' I mean 'approved of by an essentially just and loving God'.
Austere Metaethicist: Your definition doesn't connect to reality. It's like talking about atom-for-atom 'indexical identity' even though the world is made of configurations and amplitudes instead of Newtonian billiard balls. Gods don't exist.

But before we get to empathic metaethics, let's examine the standard problems of metaethics using the framework of pluralistic moral reductionism.

Cognitivism vs. Noncognitivism

One standard debate in metaethics is cognitivism vs. noncognitivism. Alexander Miller explains:

Consider a particular moral judgement, such as the judgement that murder is wrong. What sort of psychological state does this express? Some philosophers, called cognitivists, think that a moral judgement such as this expresses a belief.
Beliefs can be true or false: they are truth-apt, or apt to be assessed in terms of truth and falsity. So cognitivists think that moral judgements are capable of being true or false.
On the other hand, non-cognitivists think that moral judgements express non-cognitive states such as emotions or desires. Desires and emotions are not truth-apt. So moral judgements are not capable of being true or false.3

But why should we expect all people to use moral judgments like "Stealing is wrong" to express the same thing?4

Some people who say "Stealing is wrong" are really just trying to express emotions: "Stealing? Yuck!" Others use moral judgments like "Stealing is wrong" to express commands: "Don't steal!" Still others use moral judgments like "Stealing is wrong" to assert factual claims, such as "stealing is against the will of God" or "stealing is a practice that usually adds pain rather than pleasure to the world."

It may be interesting to study all such uses of moral discourse, but this post focuses on addressing cognitivists, who use moral judgments to assert factual claims. We ask: Are those claims true or false? What are their implications?

Objective vs. Subjective Morality

Is morality objective or subjective? It depends which moral reductionism you have in mind, and what you mean by 'objective' and 'subjective'.

Here are some common5 uses of the objective/subjective distinction in ethics:

  • Moral facts are objective1 if they are made true or false by mind-independent facts, otherwise they are subjective1.
  • Moral facts are objective2 if they are made true or false by facts independent of the opinions of sentient beings, otherwise they are subjective2.
  • Moral facts are objective3 if they are made true or false by facts independent of the opinions of humans, otherwise they are subjective3.

Now, consider Harris' reduction of morality to facts about the well-being of conscious creatures. His theory of morality is objective3 and objective2, because facts about well-being are independent of anyone's opinion. Even if the Nazis had won WWII and brainwashed everybody to have the opinion that torturing Jews was moral, it would remain true that torturing Jews does not increase the average well-being of conscious creatures. But Harris' theory of morality is not objective1, because facts about the well-being of conscious creatures are mind-dependent facts.

Or, consider Craig's theory of morality in terms of divine approval. His theory doesn't connect to reality, but still: is it objective or subjective? Craig's theory says that moral facts are objective3, because they don't depend on human opinion (God isn't human). But his theory doesn't say that morality is objective2 or objective1, because for him, moral facts depend on the opinion of a sentient being: God.

A warning: ambiguous terms like 'objective' and 'subjective' are attractors for sneaking in connotations. Craig himself provides an example. In his writings and public appearances, Craig insists that only God-based morality can be objective.6 What does he mean by 'objective'? On a single page,7 he uses 'objective' to mean "independent of people's opinions" (objective2) and also to mean "independent of human opinion" (objective3). I'll assume he means that only God-based morality can be objective3, because God-based morality is clearly not objective2 (Craig's God is a person, a sentient being).

And yet, Craig says that we need God in order to have objective3 morality as if this should be a big deal. But hold on. Even a moral code defined in terms of the preferences of Washoe the chimpanzee is objective3. So not only is Bill's claim that only God-based morality can be objective3 false (because Harris' moral theory is also objective3), but also it's trivially easy to come up with a moral theory that is 'objective' in Craig's (apparent) sense of the term (that is, objective3).8

Moreover, Harris' theory of morality is objective in a 'stronger' sense than Craig's theory of morality is. Harris' theory is objective3 and objective2, while Craig's theory is merely objective3. Whether he's doing it consciously or not, I wonder if Craig is using the word 'objective' to try to sneak in connotations that don't actually apply to his claims once you pay attention to what Craig actually means by the word 'objective'. If Craig told his audience that we need God for morality to be 'objective' in the same sense that morality defined in terms of the preferences of a chimpanzee is 'objective', would this still still have his desired effect on his audience? I doubt it.

Once you've stipulated your use of 'objective' and 'subjective', it is often trivial to determine whether a given moral reductionism is 'objective' or 'subjective'. But what of it? What force should those words carry after you've tabooed them? Be careful not to sneak in connotations that don't belong.

Relative vs. Absolute Morality

Is morality relative or absolute? Again, it depends which moral reductionism you have in mind, and what you mean by 'relative' and 'absolute'. Again, we must be careful about sneaking in connotations.

Moore's Open Question Argument

"He's an unmarried man, but is he a bachelor?" This is a 'closed' question. The answer is obviously "Yes."

In contrast, said G.E. Moore, all questions of the type "Such and such is X, but is it good?" are open questions. It feels like you can always ask, "Yes, but is it good?" In this way, Moore resists the identification of 'morally good' with any set of natural facts. This is Moore's Open Question Argument. Because some moral reductionisms do identify 'good' or 'right' with a particular X, those reductionisms had better have an answer to Moore.

The Yudkowskian response is to point out that when cognitivists use the term 'good', their intuitive notion of 'good' is captured by a massive logical function that can't be expressed in simple statements like "maximize pleasure" or "act only in accordance with maxims you could wish to be a universal law without contradiction." Even if you think everything you want (or rather, want to want) can be realized by (say) maximizing the well-being of conscious creatures, you're wrong. Your values are more complex than that, and we can't see the structure of our values. That is why it feels like an open question remains no matter which simplistic identification of "Good = X" you choose.

The problem is not that there is no way to identify 'good' or 'right' (as used intuitively, without tabooing) with a certain X. The problem is that X is huge and complicated and we don't (yet) have access to its structure.

But that's the response to Moore after righting a wrong question - that is, when doing empathic metaethics. When doing mere pluralistic moral reductionism, Moore's argument doesn't apply. 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."9

The Is-Ought Gap

(This section rewritten for clarity.)

Many claim that you cannot infer an 'ought' statement from a series of 'is' statements. The objection comes from Hume, who said he was surprised whenever an argument made of is and is not propositions suddenly shifted to an ought or ought not claim, without explanation.10

The solution is to make explicit the bridge from 'ought' statements to 'is' statements.

Perhaps the arguer means something non-natural by 'ought', such as 'commanded by God' or 'in accord with irreducible, non-natural facts about goodness' (see Moore). If so, I would reject that premise of the argument, because I'm a reductionist. At this point, our discussion might need to shift to a debate over the merits of reductionism.

Or perhaps by 'you ought to X' the arguer means something fully natural, such as:

  • "X is obligatory (by deontic logic) if you assume axiomatic imperatives Y and Z."
  • Or: "X tends to maximizes reward signals in agents exhibiting multiple-drafts consciousness" (or, as Sam Harris more broadly puts it, "X tends to maximize well-being in conscious creatures").
  • Or: "X is what a Bayes-rational and Hubble-volume-omniscient agent would do if it was motivated to maximize the amount of non-moral goodness from a view in which the interests of all potentially affected individuals were counted equally, where 'non-moral goodness' refers to what an agent would want if it were he to contemplate its present situation from a standpoint fully and vividly informed about itself and its circumstances, and entirely free of cognitive error or lapses of instrumental rationality" (see Railton's metaethics).
  • Or: "X maximizes the complicated function that can be computed by extrapolating (in a particular way) the motivations encoded by my brain" (see CEV).
  • Or: "[insert here whatever statement, if believed, would motivate one to do X]" (see Will Sawin).

Or, the speaker may have in mind a common ought-reductionism known as the hypothetical imperative. This is an ought of the kind: "If you desire to lose weight, then you ought to consume fewer calories than your burn." (But usually, people leave off the implied if statement, and simply say "You should eat less and exercise more.")

A hypothetical imperative (as some use it) can be translated from 'ought' to 'is' in a straightforward way: "If you desire to lose weight, then you ought to consume fewer calories than you burn" translates to the claim "If you consume fewer calories than you burn, then you will (or are, ceteris paribus, more likely to) fulfill your desire to lose weight."11

Or, the speaker may be using 'ought' to communicate something only about other symbols (example: Bayes' Rule), leaving the bridge from 'ought' to 'is' to be built when the logical function represented by his use of 'ought' is plugged into a theory that refers to the world.

But one must not fall into the trap of thinking that a definition you've stipulated (aloud or in your head) for 'ought' must match up to your intended meaning of 'ought' (to which you don't have introspective access). In fact, I suspect it never does, which is why the conceptual analysis of 'ought' language can go in circles for centuries, and why any stipulated meaning of 'ought' is a fake utility function. To see clearly to our intuitive concept of ought, we'll have to try empathic metaethics (see below).

But whatever our intended meaning of 'ought' is, the same reasoning applies. Either our intended meaning of 'ought' refers (eventually) to the world of math and physics (in which case the is-ought gap is bridged), or else it doesn't (in which case it fails to refer).12

Moral realism vs. Anti-realism

So, does all this mean that we can embrace moral realism, or does it doom us to moral anti-realism? Again, it depends on what you mean by 'realism' and 'anti-realism'.

In a sense, pluralistic moral reductionism can be considered a robust form of moral 'realism', in the same way that pluralistic sound reductionism is a robust form of sound realism. "Yes, there really is sound, and we can locate it in reality — either as vibrations in the air or as mental auditory experiences, however you are using the term." In the same way: "Yes, there really is morality, and we can locate it in reality — either as a set of facts about the well-being of conscious creatures, or as a set of facts about what an ideally rational and perfectly informed agent would prefer, or as some other set of natural facts."

But in another sense, pluralistic moral reductionism is 'anti-realist'. It suggests that there is no One True Theory of Morality. (We use moral terms in a variety of ways, and some of those ways refer to different sets of natural facts.) And as a reductionist approach to morality, it might also leave no room for moral theories which say there are universally binding moral rules for which the universe (e.g. via a God) will hold us accountable.

What matters are the facts, not whether labels like 'realism' or 'anti-realism' apply to 'morality'.

Toward Empathic Metaethics

But pluralistic moral reductionism satisfies only a would-be austere metaethicist, not an empathic metaethicist.

Recall that when Alex asks how she can do what is right, the Austere Metaethicist replies:

Tell me what you mean by 'right', and I will tell you what is the right thing to do. If by 'right' you mean X, then Y is the right thing to do. If by 'right' you mean P, then Z is the right thing to do. But if you can't tell me what you mean by 'right', then you have failed to ask a coherent question, and no one can answer an incoherent question.

Alex may reply to the Austere Metaethicist:

Okay, I'm not sure exactly what I mean by 'right'. So how do I do what is right if I'm not sure what I mean by 'right'?

The Austere Metaethicist refuses to answer this question. The Empathic Metaethicist, however, is willing to go the extra mile. He says to Alex:

You may not know what you mean by 'right.' But let's not stop there. Here, let me come alongside you and help decode the cognitive algorithms that generated your question in the first place, and then we'll be able to answer your question. Then we can tell you what the right thing to do is.

This may seem like too much work. Would we be motivated to decode the cognitive algorithms producing Albert and Barry's use of the word 'sound'? Would we try to solve 'empathic meta-acoustics'? Probably not. We can simply taboo and reduce 'sound' and then get some work done.

But moral terms and value terms are about what we want. And unfortunately, we often don't know what we want. As such, we're unlikely to get what we really want if the world is re-engineered in accordance with our current best guess as to what we want. That's why we need to decode the cognitive algorithms that generate our questions about value and morality.

So how can the Empathic Metaethicist answer Alex's question? We don't know the details yet. For example, we don't have a completed cognitive neuroscience. But we have some ideas, and we know of some open problems that may admit of progress once more people understand them. In the next few posts, we'll take our first look at empathic metaethics.13

Previous post: Conceptual Analysis and Moral Theory


1 Some have objected that the conceptual analysis argued against in Conceptual Analysis and Moral Theory is not just a battle over definitions. But a definition is "the formal statement of the meaning or significance of a word, phrase, etc.", and a conceptual analysis is (usually) a "formal statement of the meaning or significance of a word, phrase, etc." in terms of necessary and sufficient conditions. The goal of a conceptual analysis is to arrive at a definition for a term that captures our intuitions about its meaning. The process is to bash our intuitions against others' intuitions until we converge upon a set of necessary and sufficient conditions that captures them all. But consider Barry and Albert's debate over the definition of 'sound'. Why think Albert and Barry have the same concept in mind? Words mean slightly different things in different cultures, subcultures, and small communities. We develop different intuitions about their meaning based on divergent life experiences. Our intuitions differ from each other's due to the specifics of unconscious associative learning and attribution substitution heuristics. What is the point of bashing our intuitions about the meaning of terms against each other for thousands of pages, in the hopes that we'll converge on a precise set of necessary and sufficient conditions? Even if we can get Albert and Barry to agree, what happens when Susan wants to use the same term, but has slightly differing intuitions about its meaning? And, let's say we arrive at a messy set of 6 necessary and sufficient conditions for the intuitive meaning of the term. Is that going to be as useful for communication as one we consciously chose because it carved-up thingspace well? I doubt it. The IAU's definition of 'planet' is more useful than the folk-intuitions definition of 'planet'. Folk intuitions about 'planet' evolved over thousands of years and different people have different intuitions which may not always converge. In 2006, the IAU used modern astronomical knowledge to carve up thingspace in a more useful and informed way than our intuitions do.

A passage from Bertrand Russell (1953) is appropriate. Russell said that many philosophers reminded him of

the shopkeeper of whom I once asked the shortest way to Winchester. He called to a man in the back premises:

"Gentleman wants to know the shortest way to Winchester."

"Winchester?" an unseen voice replied.


"Way to Winchester?"


"Shortest way?"



He wanted to get the nature of the question clear, but took no interest in answering it. This is exactly what modern philosophy does for the earnest seeker after truth. Is it surprising that young people turn to other studies?

2 Compare also to the biologist's 'species concept pluralism' and the philosopher's 'art concept pluralism.' See Uidhir & Magnus (2011). Also see 'causal pluralism' (Godfrey-Smith, 2009; Cartwright, 2007), 'theory concept pluralism' (Magnus, 2009) and, especially, 'metaethical contextualism' (Bjornsson & Finlay, 2010) or 'metaethical pluralism' or 'metaethical ambivalence' (Joyce, 2011). Joyce quotes Lewis (1989), who wrote that some concepts of value refer to things that really exist, and some concepts don't, and what you make of this situation is largely a matter of temperament:

What to make of the situation is mainly a matter of temperament. You can bang the drum about how philosophy has uncovered a terrible secret: there are no values! ... Or you can think it better for public safety to keep quiet and hope people will go on as before. Or you can declare that there are no values, but that nevertheless it is legitimate—and not just expedient—for us to carry on with value-talk, since we can make it all go smoothly if we just give the name of value to claimants that don't quite deserve it... Or you can think it an empty question whether there are values: say what you please, speak strictly or loosely. When it comes to deserving a name, there's better and worse but who's to say how good is good enough? Or you can think it clear that the imperfect deservers of the name are good enough, but only just, and say that although there are values we are still terribly wrong about them. Or you can calmly say that value (like simultaneity) is not quite as some of us sometimes thought. Myself, I prefer the calm and conservative responses. But as far as the analysis of value goes, they're all much of a muchness.

Joyce concludes that, for example, the moral naturalist and the moral error theorist may agree with each other (when adopting each other's own language):

[Metaethical ambivalence] begins with a kind of metametaethical enlightenment. The moral naturalist espouses moral naturalism, but this espousal reflects a mature decision, by which I mean that the moral naturalist doesn't claim to have latched on to an incontrovertiblerealm of moral facts of which the skeptic is foolishly ignorant, but rather acknowledges that this moral naturalism has been achieved only via a non-mandatory piece of conceptual precisification. Likewise, the moral skeptic champions moral skepticism, but this too is a sophisticated verdict: not the simple declaration that there are no moral values and that the naturalist is gullibly uncritical, but rather a decision that recognizes that this skepticism has been earned only by making certain non-obligatory but permissible conceptual clarifications.

...The enlightened moral naturalist doesn't merely (grudgingly) admit that the skeptic is warranted in his or her views, but is able to adopt the skeptical position in order to gain the insights that come from recognizing that we live in a world without values. And the enlightened moral skeptic goes beyond (grudgingly) conceding that moral naturalism is reasonable, but is capable of assuming that perspective in order to gain whatever benefits come from enjoying epistemic access to a realm of moral facts.

3 Miller (2003), p. 3.

4 I changed the example moral judgment from "murder is wrong" to "stealing is wrong" because the former invites confusion. 'Murder' often means wrongful killing.

5 Also see Jacobs (2002), starting on p. 2.

6 The first premise of one of his favorite arguments for God's existence is "If God does not exist, objective moral values and duties do not exist."

7 Craig (2010), p. 11.

8 It's also possible that Craig intended a different sense of objective than the ones explicitly given in his article. Perhaps he meant objective4: "morality is objective4 if it is not grounded in the opinion of non-divine persons."

9 Also see Moral Reductionism and Moore's Open Question Argument.

10 Hume (1739), p. 469. The famous paragraph is:

In every system of morality, which I have hitherto met with, I have always remarked, that the author proceeds for some time in the ordinary ways of reasoning, and establishes the being of a God, or makes observations concerning human affairs; when all of a sudden I am surprised to find, that instead of the usual copulations of propositions, is, and is not, I meet with no proposition that is not connected with an ought, or an ought not. This change is imperceptible; but is however, of the last consequence. For as this ought, or ought not, expresses some new relation or affirmation, 'tis necessary that it should be observed and explained; and at the same time that a reason should be given; for what seems altogether inconceivable, how this new relation can be a deduction from others, which are entirely different from it.

11 For more on reducing certain kinds of normative statements, see Finlay (2010).

12 Assuming reductionism is true. If reductionism is false, then of course there are problems for pluralistic moral reductionism as a theory of austere (but not empathic) metaethics. The clarifications in the last three paragraphs of this section are due to discussions with Wei Dai and Vladimir Nesov.

13 My thanks to Steve Rayhawk and Will Newsome for their feedback on early drafts of this post.


Bjornsson & Finlay (2010). Metaethical contextualism defended. Ethics, 121: 7-36.

Craig (2010). Five Arguments for God. The Gospel Coalition.

Cartwright (2007).Hunting Causes and Using Them: Approaches in Philosophy and Economics. Cambridge University Press.

Godfrey-Smith (2009). Causal pluralism. In Beebee, Hitchcock, & Menzies (eds.), The Oxford Handbook of Causation (pp. 326-337). Oxford University Press.

Hume (1739). A Treatise on Human Nature. John Noon.

Finlay (2010). Normativity, Necessity and Tense: A Recipe for Homebaked Normativity. In Shafer-Landau (ed.), Oxford Studies in Metaethics 5 (pp. 57-85). Oxford University Press.

Jacobs (2002). Dimensions of Moral Theory. Wiley-Blackwell.

Joyce (2011).Metaethical pluralism: How both moral naturalism and moral skepticism may be permissible positions. In Nuccetelli & Seay (eds.), Ethical Naturalism: Current Debates. Cambridge University Press.

Lewis (1989). Dispositional theories of value. Part II. Proceedings of the Aristotelian Society, supplementary vol. 63: 113-137.

Magnus (2009). What species can teach us about theory.

Miller (2003). An Introduction to Contemporary Metaethics. Polity.

Russell (1953). The cult of common usage. British Journal for the Philosophy of Science, 12: 305-306.

Uidhir & Magnus (2011). Art concept pluralism. Metaphilosophy, 42: 83-97.