The is-ought gap
from 1739 A Treatise of Human Nature, book III
No collection of facts about how things are, however complete, yields a conclusion about what should be done — unless a premise about what should be done was already there.
- Active
- from 1739
- Where
- Edinburgh
- Status
- In use
- Tradition
- Devices and constraints
What it is
Hume observes that writers proceed for pages in the ordinary way of reasoning, establishing what is the case, and then without warning every proposition becomes connected by ought rather than is. He asks how this new relation can be deduced from others entirely different, and says that a reason should be given.
As a device it is the demand for that reason. Where an argument moves from evidence to a recommendation, ask what evaluative premise made the move possible — and whether it was stated or smuggled.
How it works
Find the point in the argument where the conclusion becomes a recommendation. Look back for a premise that is itself evaluative. If there is one, the argument is fine and the premise is where the disagreement actually lies. If there is not, the argument has crossed the gap without a bridge.
The common outcome is not that the argument is wrong but that its real premise is unstated, and that everyone has been arguing about the evidence instead.
The form of the inference
- The premises state what is the case
- The conclusion states what ought to be done
- A conclusion containing ought requires a premise containing ought
- No such premise has been stated
- Therefore the argument is incomplete, and the missing premise is where the disagreement lies
The argument
Proposition The survey does not support the recommendation.
- The report finds that 68 per cent of respondents use the pool fewer than four times a year, and recommends that the pool close.
- Every premise offered is a finding of fact: usage rates, cost per visit, the age of the plant, and comparable facilities in nearby shires.
- The conclusion is a recommendation about what the council should do.
- No evaluative premise appears anywhere in the report. The move from these facts to that recommendation requires one — something like: a facility used by fewer than a stated proportion of ratepayers should not be funded from general revenue.
- That premise is contestable and has never been put to anyone. The council has funded the library, the two halls and the skate park on a different principle entirely, and none of them would survive this one.
Therefore The survey does not support the recommendation on its own, and the actual question before the council is whether to adopt the unstated funding principle — which is a decision, not a finding, and belongs to the councillors rather than to the report.
The figures are not disputed and need not be. The device located the point at which a factual report became a recommendation, and found that the premise doing the work was never written down.
What it is good for
Reports, evaluations and business cases that move from data to a recommendation. It is the most useful single question to ask of any evidence-based proposal.
Where it fails
Searle's counterexample has never been fully disposed of. From the fact that someone uttered certain words in certain circumstances it appears to follow that they made a promise, and from that that they ought to keep it — with no evaluative premise anywhere. Whether the institution of promising smuggles one in is still argued about.
There are also formal difficulties. Prior showed that the logical version of the principle is harder to state than it looks: from *the sun is shining* one may validly derive *either the sun is shining or you ought to lie*, which contains an ought and follows from a purely factual premise.
And it is often used as a conversation-stopper. Pointing out that a recommendation contains an evaluative premise is true of every recommendation ever made, and saying so is only useful if you go on to examine the premise.
Common misconceptions
That it means facts are irrelevant to what should be done. Facts are essential and insufficient — the argument needs both, and Hume's point is that the evaluative half must be visible.
Where it came from
A single paragraph in book III of the Treatise of 1739, almost an aside, and among the most consequential paragraphs ever written.
What became of it
Debated continuously ever since, attacked most seriously by Searle in 1964, and never dislodged as a working principle.
Lasting influence
It shapes metaethics entirely, and it is the standard objection to arguments that move from science to policy without stating a value.
Why it still matters
Extremely high. Every evidence-based recommendation you will read contains an unstated ought.
Put another way
A bridge built from one bank, stopping short of the other.
Texts
David Hume, A Treatise of Human Nature
Book III, part I, section I. The relevant passage is one paragraph at the end of the section.
Where to start
Read the paragraph. Then find the unstated ought in the next report that crosses your desk.
The surprising bit
Hume did not present it as a discovery or a principle. He offered it as a remark, said it would subvert the vulgar systems of morality, and moved on.
In their own words
This ought, or ought not, expresses some new relation or affirmation, it is necessary that it should be observed and explained
Who held it
A. N. Prior
Connections
- Influenced
- The open question
- Parallel to
- The open question
The verdict
The most useful question to ask of any recommendation, and constantly misused as a way of dismissing one.