Is-Ought Problem
The is-ought problem is Hume’s observation that writers on morality slide without warning from descriptive claims (“is”) to normative claims (“ought”), as if the move were innocent (see the SEP entry). It isn’t. You can’t get an evaluative conclusion purely out of factual premises.
Why is this a useful diagnostic?
If you can’t validly cross from
istoought, then any moral argument that doesn’t include a normative premise is smuggling one. That’s diagnostic for half the bad arguments in the wild.
Hume, Treatise 3.1.1.27
In every system of morality… at some point [authors] make an unremarked transition from premises linked by ‘is’ to conclusions linked by ‘ought’… which seems altogether inconceivable.
Hume asks how this move was made and what licenses it. If it can’t be explained, it “subverts all vulgar systems of morality” because their conclusions don’t follow from their premises.
Three readings (per SEP), since Hume himself never commits to one:
- Logical: no valid inference takes you from
is-premises toought-conclusions. 20th-century dominant reading (R.M. Hare), sometimes called Hume’s Law - Epistemic: reason alone can’t discover moral properties; you need sentiment to introduce them
- Anti-realist / non-cognitivist: moral utterances aren’t fact-stating at all. They’re expressions of feeling
SEP notes that Hume assumes rather than argues that factual premises are non-evaluative; the move depends on that assumption.
The is-ought gap is a consequence of Hume’s deeper commitment: reason alone cannot motivate action; only passions can. If moral judgments move us to act, and reason can’t move us, moral judgments aren’t products of reason alone. They require sentiment. So the gap isn’t a random logical observation; it’s the meta-ethical sibling of Hume’s point about causation: in both cases, something we thought reason was doing turns out to be psychological.
Naturalistic fallacy vs. is-ought
These get conflated. They’re different:
Claim Area Core point Is-Ought (Hume) Inference You can’t validly derive oughtfromisaloneNaturalistic Fallacy (Moore) Semantics You can’t define moral terms like “good” using natural terms Moore’s is about reduction: “good” isn’t synonymous with “pleasure-producing” or “survival-enhancing,” as his open question argument tries to show. Hume’s is about the structure of moral arguments. Related but not the same, and appeal to nature is a third thing (a rhetorical move, not a meta-ethical thesis).
Main objections:
- Searle’s promising argument: from “Jones said ‘I promise to pay Smith five dollars’” + constitutive rules of promising, you can derive “Jones ought to pay Smith.” The bridge is institutional facts, facts that are partly normative by their nature
- MacIntyre, functional concepts: “This is a watch” + “a watch is for keeping time” entails “this watch ought to keep time.” When a concept embeds a function, norms come along for free. Hume’s gap might only apply to purely non-functional descriptions
Neither settles the debate. SEP calls the is-ought passage the single most interpretively controversial bit of Hume.
My take
This problem is actually load-bearing for how I read bad arguments in the wild. The appeal to nature is really just a noisy instance of ignoring the is-ought gap: “X is natural, therefore X is good” skips the normative premise entirely. Same with half the moves I’ve seen in PHIL145: if you can’t find the evaluative premise, there isn’t one, and the argument is smuggling.
Where I think Hume’s Law overreaches: Searle and MacIntyre are right that once you concede normative content anywhere in the premises (in constitutive rules, in functional concepts, in definitions of roles) the gap is trivially bridged. You don’t get ethics from pure physics, but you rarely argue from pure physics. Actual moral arguments run on premises about purposes, commitments, and institutions, all of which are already partly normative.
So the useful form of Hume’s Law for me is: if every premise is purely descriptive, no normative conclusion follows, so find the smuggled ought premise. That’s diagnostic, not paralyzing.