Descriptive and Normative Claims
Two sentences about the same factory: it released four hundred tonnes of solvent into the river, and it should not have. Everyone can check the first. The second is not settled by checking anything, and it is the kind of sentence ethics is about.
Two kinds of claim
Definition
Descriptive claim
A claim about how things are, were, or will be. It reports, predicts, or explains, and it is assessed by finding out whether the world matches it.
Definition
Normative claim
A claim about how things ought to be, or about what someone has reason to do. It does not report the world; it says something about what the world calls for.
The test that separates them is not about subject matter, since both kinds of claim can be about factories, taxes, or promises. It is about what would settle the question.
Suppose two people agree on every measurable fact about the discharge: the volume, the chemistry, the effect on the fish, the state of the law, what the company knew and when. They have exhausted the descriptive questions, and they can still disagree about whether the company did something wrong. Whatever that further disagreement is about, it is not about a fact either of them has failed to look up.
Notice also that a normative claim survives being false to the facts. "The company should have installed the filter" is not refuted by pointing out that it did not. Descriptive claims are answerable to what happened; normative claims are what we assess what happened against.
Check your understanding
Evaluation and prescription
Definition
Evaluative claim
A claim that something is good, bad, better, worse, admirable, or contemptible. It grades rather than instructs.
Definition
Ought
The word marking what a person is required or recommended to do. An ought-claim directs action, where an evaluative claim only ranks.
The two come apart in both directions, which is why they are worth keeping separate.
Something can be good without anyone being required to bring it about. It would be good if a stranger recovered from an illness, and no ought falls on you from that. And something can be required without being good in any independent sense: keeping a tedious promise to attend a tedious dinner is what you ought to do, and nothing about the evening is admirable.
Terminology varies here, and readers meeting the words in different books should be warned. Some writers use "normative" as the umbrella term covering both the evaluative and the prescriptive, which is the usage in this article. Others use "normative" and "evaluative" as near-synonyms. The distinction between grading and directing is real whichever labels a given author attaches to it.
Check your understanding
Crossing the line
There is a standing question about whether a normative conclusion can be derived from purely descriptive premises. David Hume noticed in the Treatise that writers on morality proceed for a while with the ordinary reasonings of "is" and "is not", and then, without remark, every proposition begins to be joined by "ought" or "ought not". He asked how the new relation could follow from the others, and thought no one had explained it.
What exactly he was claiming is disputed, and the intermediate tier takes up both the passage and the twentieth-century arguments built on it. The point to carry forward is smaller and hard to dispute: if an argument's premises are all descriptive and its conclusion is normative, something has been added between them, and a reader is entitled to ask what.
Objection
The distinction is an artefact of the examples. Take a word like cruel. To call an act cruel is to say something about what was done, that it inflicted suffering deliberately and without adequate reason, and calling it that also condemns it. The description and the condemnation are not two separable components; they are one thing said with one word. Most of the vocabulary people actually use to talk about conduct is like this, so a clean division into descriptive and normative claims misdescribes ordinary moral thought.
Reply
The examples are real, and philosophers call such terms thick, in contrast to thin terms like good and wrong that carry evaluation with little descriptive content. Whether the two elements of a thick term can be prised apart is a live dispute rather than a settled matter.
The distinction survives it. Even granting that cruel does both jobs at once, the two jobs remain identifiable, and the evidence is that they can be separated by anyone who wants to. A person can report that a practice satisfies every descriptive condition for cruelty and go on to deny that it is objectionable, and we understand what they have said even while thinking them badly wrong. That is not an option if the word carried only one meaning.
What the objection does establish is that the division is a division between the jobs a claim can do, not a sorting of sentences into two bins. Many sentences do both jobs, and the analysis pulls them apart rather than pretending they do not exist.
Check your understanding
What comes next
Everything in ethics rests on this distinction, and the field is organised around a further one that is easy to miss at first. Some questions are about which normative claims are true, and some are about what a normative claim is doing in the first place. The second sort is metaethics, and the last article of this tier introduces it.
Before either, the notion the whole subject leans on needs stating on its own. A normative claim says something counts in favour of something. The next article asks what that comes to.
Further reading
- David Hume, A Treatise of Human Nature (1739 to 1740), Book III, Part I, Section I, for the is-ought passage.
- Stanford Encyclopedia of Philosophy, "Metaethics", for how the questions are divided up.
- Stanford Encyclopedia of Philosophy, "Hume's Moral Philosophy", section 5, for the interpretive dispute about the passage.
Quiz yourself on this article
6 questions. Miss one and it comes back until you get it right.
Reached the end?
Mark it, and the sidebar keeps track of where you have been.