Overview – Metaethics

A level metaethics is about what moral judgements – e.g. “murder is wrong” – mean and what (if anything) makes them true or false.

The syllabus covers 5 metaethical theories. These theories can be categorised as either realist or anti-realist theories.

  • Moral realism: There are mind-independent, external moral properties and facts – e.g. “murder is wrong” is a moral fact because the act of murder has the moral property of wrongness
  • Moral anti-realism: Mind-independent moral properties and facts do not exist.

The theories can be further categorised as either cognitivist or non-cognitivist theories.

  • Cognitivism: Moral judgements express cognitive mental states – i.e. beliefs, aim to describe reality, and can be true or false
    • E.g. “Murder is wrong”
  • Non-cognitivism: Moral judgements express non-cognitive mental states, do not aim to describe reality, and are not capable of being true or false
    • E.g. “Boo! Murder!” or “Don’t torture animals!”

The following table categorises the 5 metaethical theories according to whether they are realist or cognitivist:

Moral realism Moral anti-realism

Having these two sets of categories (realism/anti-realism and cognitivism/non-cognitivism) can make it a bit confusing because there is overlap between the categories. For example, all non-cognitivist theories are anti-realist theories but not all cognitivist theories theories are realist theories.

So, it’s important to be clear on the differences between what the categories are about:

  • Realism/anti-realism is about whether or not mind-independent moral properties exist
  • Cognitivism/non-cognitivism is about what people mean when they make moral judgements

The theories are linked with each other to some degree. For example, the arguments against naturalism are the same arguments in favour of non-naturalism, and the arguments against non-naturalism are the same arguments in favour of error theory.

Realist theories

Realist metaethical theories argue that mind independent moral properties and facts exist.

So when we make moral judgements – e.g. “murder is wrong” – these statements are capable of literally being true or false in the same way “grass is green” is literally true. Murder has the property of wrongness in the same way grass has the property of greenness.

There is disagreement among realists as to what these mind-independent moral properties – e.g. ‘right’, ‘wrong’, ‘good’, ‘bad’ – actually are:

  • Ethical naturalism says moral properties are natural properties
  • Ethical non-naturalism says moral properties are non-natural properties

Both ethical naturalism and ethical non-naturalism are cognitivist theories: they agree that moral judgements express beliefs that are capable of being true or false.

Ethical naturalism

Ethical naturalism: moral properties are natural properties.

So, “murder is wrong” expresses a belief that murder is wrong – where ‘wrong’ refers to a natural property.

It’s a reductionist theory: it says moral properties can be reduced to natural properties. These properties exist and are mind-independent. Hence, ethical naturalism is a moral realist theory.

Utilitarianism is an example of a naturalist ethical theory. It says ‘good’ can be reduced to pleasure, and ‘bad’ can be reduced to pain.

Natural properties are ordinary, physical properties. Pain and pleasure, for example, are natural properties of the brain – a physical thing.

Problems for ethical naturalism:

Ethical non-naturalism

Ethical non-naturalism: moral properties are non-natural properties.

So, “murder is wrong” expresses a belief that murder is wrong – where ‘wrong’ refers to a non-natural property.

Non-natural moral properties cannot be reduced to anything simpler. They are basic.

However, these non-natural moral properties exist independently of minds. Hence, ethical non-naturalism is a moral realist theory.

G.E. Moore: Principia Ethica

The naturalistic fallacy:

Moore’s book begins with criticisms of ethical naturalism.

He invents the term ‘Naturalistic fallacy’ to describe the fallacy (i.e. bad reasoning) of equating goodness with some natural property. His main arguments for the naturalistic fallacy are:

  • Moral properties may be correlated with natural properties, but are not identical to them. For example, having a heart is correlated with also having kidneys, but hearts are not the same thing as kidneys. Similarly, happiness may often accompany morally good actions, but they are not the same thing.
  • Moore’s ‘open question’ argument: if goodness and pleasure are the same thing, it wouldn’t make sense to ask ‘is pleasure good?’, because it would basically be like asking ‘is pleasure pleasure?’. But because the question ‘is pleasure good?’ does make sense, this proves that pleasure and good are not the same thing.
    • Reply to open question argument: It makes sense to ask ‘is water H2O?’ even though they refer to the same thing. The fact that this question does make sense does not prove that water and H2O are two separate things. So how can Moore’s argument (which follows the same format) prove that goodness and pleasure are two separate things?

Moore argues that moral properties cannot be reduced to anything simpler, such as pain or pleasure. Moral properties are basic.

Moore’s Intuitionism:

Having argued against ethical naturalism, Moore sets out his own version of ethical non-naturalism, which he calls intuitionism.

We might ask Moore: if moral properties are non-natural properties, how do we know about them?

It’s no mystery how we know about natural properties, such as blueness, roundness, or largeness. But non-natural properties are more difficult to explain – they’re not like ordinary physical properties that can be scientifically investigated.

Moore’s response to this problem is intuition. He argues that, via the faculty of rational intuition, we can directly reflect on the truth of moral judgements such as “murder is wrong”. The truth/falsehood of such moral judgements is said to be self evident.

Problems for ethical non-naturalism:

Anti-realist theories

Anti-realist metaethical theories argue that mind independent moral properties do not exist. So, there are no such things as true moral facts.

The syllabus looks at 3 anti-realist metaethical theories:

  • Error theory says moral judgements are cognitive statements but properties don’t exist
  • Emotivism says moral judgements are non-cognitive statements that express feelings of approval or disapproval
  • Prescriptivism says moral judgements are non-cognitive statements that are intended as instructions

Error theory

Error theory: moral judgements express beliefs about the way the world is but moral properties don’t exist.

So, “murder is wrong” expresses a belief that murder is wrong – where ‘wrong’ refers to a non-existent property.

Since moral properties don’t exist according to error theory, it claims that all moral propositions – e.g. “murder is wrong” – are false.

J.L. Mackie: Ethics: Inventing Right and Wrong

Mackie’s book starts with various arguments in favour of cognitivism generally.

Moral philosophy, he argues, has tended to assume objective moral values (e.g. Plato and Kant) – i.e. that moral judgements are objectively true or false.

Not only that, ordinary language assumes cognitivism as well. To illustrate this, Mackie uses the example of someone facing the moral dilemma of whether to engage in research related to bacteriological warfare. In trying to resolve this dilemma, you don’t ask how you feel about it – you ask whether such action is wrong in itself.

The argument from queerness

Having established that moral judgements are cognitive and thus aim to be true or false, Mackie turns his attention to the nature of moral properties.

He argues that if moral properties are non-natural properties, they must be very strange (or ‘queer’, to use Mackie’s term). His reasons:

  • Epistemically queer: How do we know about these non-natural properties? Knowledge of natural (physical) properties and facts can be explained scientifically, but non-natural properties seemingly cannot
  • Metaphysically queer: These non-natural moral properties must be metaphysically unlike anything else we have experience of. If an action has the moral property of ‘good’ – e.g. giving to charity – then it must somehow have to-be-doneness built into it.

Less specifically, the argument from queerness can be thought of as the claim that moral properties are unscientific. We can’t see goodness or badness – or measure them with any scientific instruments – so what evidence do we have that they exist at all?

A related concern is how moral facts relate to natural facts. Imagine someone stealing from a shop. Then imagine someone stealing from a shop and that this action has the property of wrongness. What’s the difference between the two cases?

The queerness of moral properties, Mackie argues, is evidence that moral properties do not actually exist.

All moral judgements are false

Remember, Mackie started by arguing for cognitivism, and so believes that moral judgements aim to be true or false.

And given his second claim – that moral properties don’t exist – it follows that all moral judgements are false.

So, according to Mackie, “murder is wrong” is false, because ‘wrongness’ doesn’t exist.

Not only that, “murder is good” is also false – because ‘goodness’ doesn’t exist either.

Cognitive moral language, Mackie argues, relies on a false assumption that moral properties exist. It aims to make true/false claims, but ends up always making false claims because there are no such things as moral properties.



Emotivism: moral judgements express feelings of approval or disapproval.

So, when someone says “murder is wrong!”, what they really mean is “boo! murder!”

(By the way, “boo!” here is like when people “boo!” at a football match, not “boo!” like when you jump out at someone to scare them. This doesn’t really come across in text…)

Similarly, when someone says “giving money to charity is good”, what they are really expressing is “hooray! Giving money to charity!”

Notice how none of these attitudes are capable of being true or false. They are just expressions of approval or disapproval – not beliefs. Hence, emotivism is a non-cognitivist theory.

Hume: Treatise of Human Nature
Arguments against cognitivism

Hume provides two arguments for the view that moral judgements are not judgements of reason – i.e. that moral judgements are non-cognitive.

(you can use these arguments to argue against any of the cognitivist theories: naturalism, non naturalism and error theory)

Moral judgements motivate action
  • Moral judgements can motivate action
  • Reason cannot motivate action
  • Therefore, moral judgements are not judgements of reason

Judgements of reason – e.g. a belief that grass is green – can’t motivate me to act in any way.

Emotions and desires, in contrast, can motivate me to act. For example, my desire to eat pizza might motivate me to seek out pizza. Moral judgements seem more like this because, Hume argues, they motivate action. For example, my belief that “murder is wrong” will motivate me not to murder.

Moral judgements are not relations of ideas or matters of fact

If you remember from epistemology, Hume argued there are just two types of judgements of reason: relations of ideas and matters of fact.

Given this, Hume argues:

  • Moral judgements are not relations of ideas
  • Moral judgements are not matters of fact
  • Therefore, moral judgements are not judgements of reason
Arguments for emotivism

Hume’s arguments have been interpreted to support emotivism – the view that moral judgements express attitudes, not facts.

The is/ought problem

Hume argues that you cannot logically derive ought statements from statements about what is.

For example, to say “you shouldn’t torture animals because it hurts them!” derives a statement about what ought to be (you shouldn’t torture animals) from a statement about what is (torturing animals causes pain). Hume says this does not – and could not – follow.

This idea is closely related to the naturalistic fallacy that Moore argues.

And another closely related idea is the fact/value distinction. For example:

  • Fact: Smith murdered Jones
  • Value: Murder is wrong

Non-cognitivists argue that facts and values are two completely separate things. Facts are true in virtue of how the world is whereas values express attitudes towards facts.

So, since ought statements cannot be reasoned from is statements, Hume argues that:

“morality is more properly felt than judg’d of.”

Comparison to Locke’s secondary qualities

Remember Locke’s distinction between primary and secondary qualities from AS philosophy:

  • Primary qualities are inherent in the object itself (e.g. size and shape)
  • Secondary qualities are more subjective: they are powers to cause sensations in humans (e.g. colour and smell)

Secondary qualities are not properties of objects themselves. Similarly, argues Hume, moral properties are not properties of actions themselves. So, moral properties are more like secondary qualities.

Hume argues that there is nothing inherent in an act of murder that makes it wrong. However, he argues, acts of murder tend to cause feelings of disapproval in humans. Moral properties – wrongness, rightness, goodness, badness – are best understood as expressions of our own feelings, hence emotivism.

A.J. Ayer: Language, Truth and Logic
The verification principle

(The verification principle also comes up in religious language)

The verification principle: a statement only has meaning if it is either:

  • An analytic truth (e.g. “a triangle has 3 sides”)
  • Empirically verifiable (e.g. “water boils at 100c”)

Any statement that does not fit these descriptions is meaningless, according to the verification principle.

Rejection of naturalism

Ayer argues that moral judgements fail the verification principle.

For example, “murder is wrong” is clearly not an analytic truth. It’s not empirically verifiable either as observations could possibly prove it to be true or false? Sure, we could prove that murder causes pain, anger, etc. – but this is not the same as proving that it’s wrong.

Hence, Ayer rejects naturalism: we can empirically verify that murder causes pain, say, but we cannot empirically verify that murder is wrong.

Rejection of non-naturalism

Ayer also argues that there is no way to empirically verify the presence of non-natural properties. Even if “murder is wrong” did possess the non-natural property of wrongness, how could we ever prove this? It’s not empirically verifiable, nor is it an analytic truth. Therefore, it fails the verification principle.

Ayer also attacks Moore’s intuitionism as failing the verification principle: without a way to decide between conflicting intuitions, there is no way to establish whether a moral judgement is true or false.

Arguments for emotivism

Since moral judgements fail the verification principle, Ayer argues they are meaningless.

Instead of expressing meaningful statements about the world, Ayer concludes that moral judgements simply express feelings of approval or disapproval and seek to evoke the same feelings in others:

“If I say to someone, “you acted wrongly in stealing that money” […] I am simply evincing my moral disapproval of it. It is as if I had said, “You stole that money,” in a peculiar tone of horror.”



Prescriptivism: moral judgements express instructions that aim to guide behaviour.

So, when someone says “murder is wrong!”, what they really mean is “don’t murder people!”

When you instruct someone to do something – e.g. “shut the door” – you are not expressing a belief that is capable of being true or false. Hence, emotivism is a non-cognitivist theory.

R.M. Hare: The Language of Morals
Imperatives and value judgements

Hare says there are two types of prescriptive meaning:

  • Imperatives: tell someone to do something – e.g. “shut the door” or “go away”
  • Value judgements: praise or criticise something – e.g. “that’s good” or “that’s bad”

Hare says emotivism misses the force of moral judgements. They don’t just express what you feel and try to get others to feel the same (as Ayer claims), they’re also prescriptions for what you ought to do. They are imperatives.

He continues that ‘good’ and ‘bad’ are more than just descriptions. We assume a set of standards – e.g. a ‘good’ teacher is one who can explain ideas clearly and keep the students interested – and then praise or criticise according to these standards. When we say a person is ‘good’ in the moral sense we use the general standards that apply to someone as a person.

However, Hare argues, these standards have to be adopted by us. There is nothing about the facts themselves that establish one set of standards as correct.

Moral judgements are both value judgements and imperatives.


Hare describes the key features of the word ‘good’:

  • It is used to praise and provide guidance for how to act
  • It assumes a set of standards
  • If two things are identical, they must both be good or bad. It wouldn’t make sense for one thing to be good and an identical thing to be bad

Problem: moral argument and reasoning

(note: the following argument applies to non-cognitivist theories (i.e. prescriptivism and emotivism) but not cognitivist theories (e.g. error theory))

Non-cognitivism appears to be at odds with how we typically use moral judgements.

We often use moral judgements as part of moral reasoning. For example:

  1. If murder is wrong, then you shouldn’t murder people
  2. Murder is wrong
  3. Therefore, you shouldn’t murder people

This seems like a sound argument. The conclusion follows logically from the premises. And if the premises are true then the conclusion must also be true.

But if non-cognitivism is true, the argument above isn’t actually sound. Prescriptivists, for example, say that what people really mean when they say “murder is wrong” is something like “don’t murder people!”

So, the argument above is something more like:

  1. If “don’t murder people!”, then you shouldn’t murder people
  2. “Don’t murder people!”
  3. Therefore, you shouldn’t murder people

But this argument isn’t sound, because the premises aren’t true.

In fact, it’s a key claim of the non-cognitivist theories emotivism and prescriptivism that moral judgements cannot be true or false. Premise 2, for example, is neither true nor false.

So, if non-cognitivism is true, the first argument is not a sound argument. But it clearly is a sound argument. Therefore, non-cognitivism is false.


(note: the following problems apply to all anti-realist theories (i.e. error theory, prescriptivism and emotivism))

Moral nihilism

If moral anti-realism is true, it can be argued that this leads to moral nihilism: the view that no actions are inherently wrong. There’s nothing true about moral judgements such as “murder is wrong”. This then raises the question of why anyone should bother to be moral at all.

Non-cognitivists can respond that just because there’s no inherent right or wrong, people still have moral attitudes and feelings. And the realisation that moral values are just expressions of feelings doesn’t mean we should (or could) stop having these moral feelings.

It’s also somewhat self-defeating to be a moral nihilist according to non-cognitivism. After all, living as if there are no moral values is itself an expression of a certain attitude or feeling.

Cognitivist anti-realist theories (i.e. error theory) have a harder time responding to the charge of moral nihilism. One response could be to just accept the charge of moral nihilism and argue that, though undesirable, this doesn’t make error theory any less true.

There may also be practical reasons to behave as if some moral judgements are true. For example, if you were always stealing from your friends, chances are they wouldn’t remain friends with you for very long.

Moral progress

Our moral values have changed over time. For example, it was considered morally acceptable to keep slaves back in the time of Plato but it’s not today.

If we accept that such changes are examples of moral progress, then we can make an argument along these lines:

  1. If moral anti-realism is true, then there would be no moral progress
  2. But there has been moral progress
  3. Therefore moral anti-realism is false

This is a somewhat question-begging argument though. The second premise essentially assumes the conclusion. Why should the anti-realist accept there’s been objective moral progress when he doesn’t accept the existence of objective morality in the first place?

However, we can define moral progress in less question-begging ways. For example, we could argue that our morality has become more consistent over time, or that we have adapted our moral values in response to greater knowledge of the facts.

<<<Applied ethics