The philosophers of ancient Greece were the first people to ask this question.
David Hume claimed that "All the objects of human reason or inquiry may naturally be divided into two kinds: relations of ideas and matters of fact". He is suggesting that we only have two types of knowledge:
These two options are known as HUME'S FORK. Hume once famously declared that there couldn't be any other sort of knowledge:
However, value-judgements are very difficult to fit into Hume's Fork. Take ethical statements like "it is wrong to torture someone" or "it is good to show kindness to strangers". How can we tell if these statements are true? They don't seem to be empirical judgements, because it's hard to say what would count as evidence for torture being wrong or kindness being good. Nor do they seem to be analytical, because you don't get the "wrong-ness" of torture just from its definition and it certainly isn't apparent to everyone that torture is wrong - think for a moment about the Catholic Church's Inquisition [see right] in the Middle Ages, which tortured heretics for the good of their souls!
In the early 20th century, a group of philosophers called the VIENNA CIRCLE came to very similar conclusions. THey suggested that, in order for a statement to be meaningful, it had to pass a condition called the VERIFICATION PRINCIPLE. This means that you should be able, in principle, to say how that statement could be proved to be true, either by appealing to evidence or logic. Statements that cannot be verified should be considered meaningless. This philosophy came to be known as LOGICAL POSITIVISM and the most famous positivist in British philosophy was A.J. Ayer. Ayer understood that ethical statements about torture or kindness could not be verified.
Some philosophers have responded to the problems implied by Hume's Fork and the Verification Principle by suggesting that ethical language should be understood in different ways.
A form of moral realism was put forward by the Greek philosopher Plato who suggested that all our thoughts or ideas corresponded to eternal, timeless "Forms". Plato thought that nothing in this world was truly, 100% good, but some things resemble the Form of "Good" more than others. The idea of things on earth being related to weird "Forms" existing in another dimension might seem odd, but Plato has some powerful arguments on his side. For example, we call the sky "blue" and denim jeans "blue" even though they are not the same colour: this shows we have a concept of "blue-ness" which we apply TO the things we encounter in the world. Similarly, no one may have ever seen a perfectly round circle or a perfectly straight line, but we have an idea of what circles and straight lines are - or how else could people manufacture tyres or rulers?
Plato's ideas were developed by Christian theologians and moral realism lies behind the Divine Command Theory of ethics. With this approach, goodness exists in the mind of God and human behaviour can either resemble God's goodness or deviate from it.
These approaches are very attractive, but they certainly run afoul of the Verification Principle, because neither Plato's Forms nor the mind of God can be studied with the senses or investigated empirically. Religious believers might think that you can investigate the mind of God by studying the Bible, but this would not impress the Logical Positivists, who think that God is an unverifiable and meaningless concept in the first place.
One way of avoiding the Verification Principle is to take an anti-realist approach to ethical language. From this viewpoint, "true" only means "true within a particular community of life". Anti-realists don't expect ethical language to correspond to something 'out there' that really exists; they just expect ethical statements to cohere with other ethical statements and behaviours. For a moral anti-realist, the worst criticism of a person's ethics is that they are contradictory. For example, a person might claim that lying is always wrong, then go around telling a lot of lies; or else, a person might claim that freedom of speech is an important right then support a government campaign to close down a newspaper. For an anti-realist, "true" just means "coherent, consistent, with no contradictions". If I say that murder is wrong, then go through my whole life not murdering anyone and opposing other forms of murder (like campaigning against the death penalty) then, as far as an anti-realist is concerned, it's just is true (for me) that murder is wrong.
The anti-realist position on morality ties in with subjectivism about ethics and even moral relativism. To consider this further, think about how a moral anti-realist might view the ethical behaviour of the ancient Aztecs. The Aztec civilisation in Mexico worshipped some terrifying gods and goddesses who demanded to be fed on human hearts and blood. The Aztecs believed that if the gods were not fed regularly, they would lose their power and would be unable to sustain the world. If the gods lost their power, the sun wouldn't rise and the dry land would sink into the oceans. To stop this happening. the Aztecs carried out human sacrifice on a grand scale; for instance, in 1487 over a 4-day festival the Aztecs sacrificed 80,000 people in their capital city. Victims were stretched out over a stone, their ribs broken apart and their still-beating heart cut out [see left]. Most people would find this repulsive but a moral anti-realist will conclude that mass human sacrifice was a good thing for the Aztecs, because it was consistent with their other beliefs and practices.
There are other ethical theories that count as anti-realist, such as A.J. Ayer's Emotivism and R.M. Hare's Prescriptivism, but we shall deal with these later.
A lot of people are uncomfortable with moral anti-realism, which seems to involve approving of human sacrifice, Nazism, cannibalism, etc as things which are "good for the community that does them". Some philosophers have investigated a realist view of ethics which might pass the Verification Principle. This is ethical naturalism. Ethical naturalism tries to prove the truth of ethical statements by referring to the real world, but does not refer to supernatural forces like Plato's Forms or God.
The most obvious version of ethical naturalism is Natural Law Theory, which claims you can base your ethical ideas on how it is natural for human beings to live. Virtue Ethics are also based on how humans would behave in a naturally balanced way.
Another sort of ethical naturalism is Evolutionary Ethics. This theory looks to Charles Darwin's theory of evolution to provide a basis for ethical language. For the evolutionist, "good" means "what an organism has evolved to do" or even "the behaviour that helps a creature survive and pass on its genes". This means it is good for humans to band together and help each other, since this is something we have evolved to do, just like lions have evolved to hunt gazelles.
Utilitarianism can also be viewed as a type of ethical naturalism. Utilitarians redefine "good" as meaning "something that brings pleasure or avoids pain". There is nothing supernatural or mysterious about this definition of good and it lends itself quite will to empirical investigation - it's quite straightforward to find out whether someone finds something pleasant or painful.
Finally, as far back as Ancient Greece, some philosophers have argued that "good" could be understood to mean "what society approves of". Again, there's nothing supernatural or mysterious about this and you can investigate this empirically by looking at society's laws. If adultery or selling alcohol or homosexuality is illegal, then it must be wrong in that society; if it's not illegal, then it must be morally acceptable. Socrates and Plato famously argued against the "Sophists" who put forward this view back in ancient Athens, but it has returned to popularity several times. The 19th century philosopher F.H. Bradley argued for a notion of duty based on the core values of society. Unfortunately, social ethics like this have no way of telling the difference between "good" and "bad" societies - after all, Nazism based itself on core values in German society in the 1920s and 30s.
Ethical naturalism has been very influential and a lot of very different ethical theories all try to get round the problems posed by Hume's Fork and the Verification Principle by appealing to some sort of naturalism. All of them involve REDEFINING goodness in one way or another:
This attempt to redefine what "good" means struck G.E. Moore [right] as suspicious and in his 1903 book Principia Ethica he exposed the flaw behind naturalistic approaches to ethics. This is the NATURALISTIC FALLACY. The naturalistic fallacy is committed whenever a philosopher attempts to prove a claim about ethics by redefining "goodness" as some sort of natural property (such as "pleasant", "more evolved", "legal", etc.). Moore is arguing here that you cannot get ethical conclusions from non-ethical premises and no amount of showing that something is pleasant, desirable, natural, balanced or legal will lead you to the conclusion that it is a moral obligation.
Moore himself admits that the naturalistic fallacy goes further than just exposing ethical naturalism as being flawed. Most forms of moral realism suffer from it and Moore makes it clear that trying to explain ethics in terms of mystical Forms or the will of God suffers from the same problem - even if you could prove that God wanted me to be kind to strangers, this does not lead to the conclusion that I have a moral obligation to be kind to strangers. You cannot go from a fact about the world to a moral value.
Moore's argument is an updating of a point originally made by David Hume, known as the Is/Ought distinction in ethics. Hume pointed out that many writers move without realising it from DESCRIPTIVE statements about the way the world is to PRESCRIPTIVE statements about the way the world ought to be. Hume thought that is was quite illogical to move from is-statements to ought-statements and, in fact, the strict separation of is-statements from ought-statements is known as HUME'S GUILLOTINE. Hume's Guillotine has posed a big problem for ethical naturalists who have spent the past couple of centuries trying (with only limited success) to argue that you can, under certain circumstances, get an is from an ought.
One solution to Hume's Guillotine and to the Naturalistic Fallacy is to deny that ethical statements refer in any way to the outside world. Philosophers who take this approach, which is really a type of moral anti-realism, might claim that ethical statements are not telling us anything about the outside world (laws, pain, God, whatever) but just tell us something about the speaker and his feelings or preferences. This general approach to ethics is termed NON-COGNITIVISM. Non-cognitive ethical theories claim that ethical statements aren't related to the outside world and cannot be proved to be true or false by using evidence.
A.J. Ayer was associated with the Vienna Circle and their Verification Principle and agreed that ethical statements could not be verified. For Ayer, when a person says that something is "good" they cannot be telling you any information about the thing itself; what they are doing is telling you about their own feelings or attitudes. This view is called EMOTIVISM. Sometimes it is nicknamed the "Yay! Boo!" approach to ethics, because it is saying that all ethical statements boil down to simple expressions of approval ("Yay!") or disapproval ("Boo!"). So, saying that torture is wrong really just amounts to "Torture - Boo!!!!" and saying kindness is good adds up to "Yay - kindness!!!".
A similar theory was put forward by R.M. Hare in his 1952 book The Language of Morals. Hare suggests that ethical statements really boil down to instructions. When someone says that murder is wrong, what they really mean is "Don't commit murder". Similarly, saying that kindness to strangers is good amounts to "Be kind to strangers". Because Hare says ethical language is all about giving people instructions, his theory is called PRESCRIPTIVISM. There are some similarities between Prescriptivism and preference utilitarianism and Hare's theories have been praised by Peter Singer. There is a tendency these days to treat Prescriptivism as a type of Emotivism rather than a separate theory.
Both of these theories seem to avoid Hume's Guillotine because they suggest that any ethical statement that seems to be an "ought" is really just an "is" in disguise! They also avoid committing the Naturalistic Fallacy, but possibly at a great cost.
G.E. Moore struck a blow against naturalism in ethics and went on to put forward his own view of non-naturalistic ethics, which is sometimes put in the catergory of Moral Intuitionism. Moore starts off by arguing that "goodness" is indefinable. He does this through the OPEN QUESTION ARGUMENT. Moore says that naturalists always try to redefine "goodness" in terms of something else, such as pleasure, but you can always turn round and ask them why is pleasure good? This is why "goodness" is an open question - you can never give an answer to "what is goodness?" that doesn't invite the counter-question "OK, but why is that good?". Moore in fact argues that goodness is a quality rather like "yellow". You cannot define what "yellow" is, though you can list many things that possess yellow-ness. This is an important comparison because Moore is showing there are plenty of other qualities in the world that are indefinable but quite ordinary, so why shouldn't goodness be one of them?
So, if goodness is indefinable, how do we recognise it? Moore suggests we have an INTUITIVE awareness of goodness when we come across it. Intuition would seem to be a third type of knowledge, besides empirical and analytical. Intuitive knowledge resembles analytical knowledge in that it is self-evident (at least, to mature, sensitive people) but it's like empirical knowledge in that it corresponds to things in the external world.
Intuitionism makes three claims:
Moore suggests that the basic moral principles are self-evident truths - things that are known and require no further proof or justification. However, applying these self-evident truths to concrete actions does require further information. It's never self-evident what we ought to do in a specific situation. This is how moral disagreement arises. Arriving at the self-evident principles of morality requires reflection and intellectual maturity. These principles have to be tested through careful examination, to uncover implications that clash with our intuitions. For example, the Terrorist Dilemma produces a clash between our intuition that torture is wrong and a competing intuition that the lives of many innocents take precedence over the wellbeing of one prisoner.
Intuitionism is very appealing because it leaves our idea of right and wrong unchanged - unlike many naturalistic ethical theories which end up redefining ethics in a way that makes them no longer resemble what we thought was right and wrong in the first place. In this way, Intuitionism follows Wittgenstein's instruction that philosophy should "leave everything as it was". However, Intuitionism, despite its initial plausibility, has some problems: