What Is Supervenience? A Relation Between Levels That Avoids Saying Too Much

The standard formulation is that one family of properties supervenes on another when no two things can differ in the first without differing in the second. Two objects identical in every physical respect cannot differ mentally, on a supervenience claim about mind, and two actions identical in every natural respect cannot differ morally, on a supervenience claim about ethics. The relation is asymmetric, since the lower level can differ without the upper one differing, which allows the same mental state to be realised by different physical arrangements. It is a claim about covariation rather than about mechanism, so asserting it commits you to a pattern of dependence and not to any account of how the dependence works. That is deliberate, since the point of the notion was to state a minimal commitment shared by positions that disagree about everything else.

The attraction is that it allows several claims at once that are otherwise hard to combine:

• •Dependence, since the upper-level facts are fixed by the lower-level ones
• •Non-reduction, since the upper-level properties need not be identical to or definable from any lower-level ones
• •Multiple realisation, since the same upper-level property can be realised by many different lower-level arrangements
• •Physicalism without translation, since one can hold that everything is physical without claiming mental talk translates into physical talk
• •Moral objectivity without naturalism, since moral facts can depend on natural facts without being natural facts
• •A way to state what is wrong with certain positions, since a view that allows two identical situations to differ morally is committed to something most people reject

The notion splinters on closer examination and the distinctions turn out to matter. Weak supervenience says no two things within a single possible world can differ above without differing below, which is compatible with the relation failing across worlds and is therefore too weak to express real dependence. Strong supervenience quantifies across worlds and says nothing anywhere can differ above without differing below, which is what most claims intend. Global supervenience compares whole worlds rather than individuals, which handles properties that depend on context rather than on intrinsic features. Each version has been argued to be either too weak to capture dependence or so strong as to entail the reduction it was designed to avoid, and the literature on whether strong supervenience collapses into something reductive is substantial and unresolved. The upshot is that the notion is less neutral than it initially promised.

One genuinely useful application is as a constraint that rules things out. A view committed to the possibility of two situations identical in every natural respect while differing morally is committed to something most people find unacceptable on reflection, since it implies that a moral difference could float free of every describable feature of the cases, and pointing that out is a real argument rather than a definitional trick. The same move applies in philosophy of mind, where a position allowing two physically identical beings to differ mentally has to explain what the difference consists in, and the famous thought experiments about beings physically identical to us but lacking experience are constructed precisely to test whether that is coherent. Aesthetics uses the relation similarly. The pattern in each case is that the claim is weak enough to be widely accepted and strong enough to have consequences, which is the combination that makes a philosophical tool worth having.

The most serious criticism is that the relation describes a pattern without explaining it, which was the original virtue and is also the limitation. If mental properties covary with physical ones, something must account for that, and supervenience by itself does not say what, so a philosopher who asserts it has described the phenomenon to be explained rather than explaining it. Critics argue this leaves the notion unable to do the work assigned to it, since the interesting question was always about the nature of the dependence rather than its existence. Supporters respond that a clearly stated minimal commitment is valuable precisely because it separates what everyone should accept from what is contested, and that many disputes became clearer once the shared commitment was isolated. Both responses are reasonable, and the practical effect has been that the term remains standard vocabulary while carrying less theoretical weight than it once did.

The relation says nothing can differ at the upper level without differing at the lower one, which gives dependence without reduction and allows the same mental state to be realised many ways. It was built to state a minimal shared commitment among positions that disagree about mechanism. Its weakness is that it names a pattern requiring explanation rather than supplying one.