Non-cognitivism is the meta-ethical view that ethical sentences do not express propositions and thus cannot be true or false (they are not truth-apt). (See also Expressivism). A noncognitivist denies the cognitivist claim that "moral judgments are capable of being objectively true, because they describe some feature of the world." If moral statements cannot be true, and if one cannot know something that is not true, noncognitivism implies that moral knowledge is impossible.
Non-cognitivism entails that non-cognitive attitudes underlie moral discourse and this discourse therefore consists of non-declarative speech acts, although accepting that its surface features may consistently and efficiently work as if moral discourse were cognitive. The point of interpreting moral claims as non-declarative speech acts is to explain what moral claims mean if they are neither true nor false (as philosophies such as logical positivism entail). Utterances like "Boo to killing!" and "Don't kill" are not candidates for truth or falsity, but have non-cognitive meaning.
While the bare term non-cognitivism usually refers to the meta-ethical position, it can also be applied in other branches of philosophy, as in theological noncognitivism.
A close cousin of emotivism, developed by R. M. Hare, is called universal prescriptivism. Prescriptivists interpret ethical statements as being universal imperatives, prescribing behavior for all to follow. "Killing is wrong" under prescriptivism becomes, "Do not murder."
Arguments for emotivism focus on what normative statements express when uttered by a speaker. A person who says that killing is wrong certainly expresses her disapproval of killing. Emotivists claim that this is all she does, that "Killing is wrong" is not a truth-apt declaration, and that the burden of evidence is on the cognitivists who want to show that in addition to expressing disapproval, the claim "Killing is wrong" is also true. Emotivists ask whether there really is evidence that killing is wrong. We have evidence that Jupiter has a magnetic field and that birds are oviparous, but as of yet, we do not seem to have found evidence of moral properties, such as "goodness". Emotivists ask why, without such evidence, we should think there is such a property. Ethical intuitionists think the evidence comes not from science or reason but from our own feelings: good deeds make us feel a certain way and bad deeds make us feel very differently. But is this enough to show that there are genuinely good and bad deeds? Emotivists think not, claiming that we do not need to postulate the existence of moral "badness" or "wrongness" to explain why considering certain deeds makes us feel disapproval; that all we really observe when we introspect are feelings of disapproval. Thus the emotivist asks why not to adopt the simple explanation and say that this is all there is; why insist that a genuine "badness" (of murder, for example) must be causing feelings, when a simpler explanation is available.
Arguments for prescriptivism, by contrast, focus on the function of normative statements. A person telling another that killing is wrong probably does not want this other person to then go off and kill someone, and may be explicitly attempting to stop him from doing so. Thus the statement "Killing is wrong," calculated to prevent someone from killing, can be described as an exhortation not to do so.
Another argument is the "embedding problem." Consider the following sentences:
Attempts to translate these sentences in an emotivist framework seem to fail (e.g. "She does not realize, 'Boo on eating meat!'"). Prescriptivist translations fare only slightly better ("She does not realize that she is not to eat meat"). Even the act of forming such a construction indicates some sort of cognition in the process.
According to some non-cognitivist points of view, these sentences simply assume the false premise that ethical statements are either true or false. They might be literally translated as:
These translations, however, seem divorced from the way people actually use language. A non-cognitivist would have to disagree with someone saying, "'Eating meat is wrong' is a false statement" (since "Eating meat is wrong" is not truth-apt at all), but may be tempted to agree with a person saying, "Eating meat is not wrong."
One might more constructively interpret these statements to describe the underlying emotional statement that they express, i.e: I disapprove/do not disapprove of eating meat, I used to, he doesn't, I do and she doesn't, etc.; however, this interpretation is closer to individualist subjectivism than to non-cognitivism proper.
A similar argument against non-cognitivism is that of ethical argument. A common argument might be, "If killing an innocent human is always wrong, and all fetuses are innocent humans, then killing a fetus is always wrong." Most people would consider such an utterance to represent an analytic proposition which is true a priori. However, if ethical statements do not represent cognitions, it seems odd to use them as premises in an argument, and even odder to assume they follow the same rules of syllogism as true propositions. However, R.M. Hare, proponent of universal prescriptivism, has argued that the rules of logical are independent of grammatical mood, and thus the same logical relations may hold between imperatives as hold between indicatives.
Many objections to non-cognitivism based on the linguistic characteristics of what purport to be moral judgments were originally raised by Peter Glassen in "The Cognitivity of Moral Judgments", published in Mind in January 1959, and in Glassen's follow-up article in the January 1963 issue of the same journal.