6 - Defining ‘intrinsic’ (with Rae Langton)  pp. 116-132

Defining ‘intrinsic’ (with Rae Langton)

By David Lewis

Image View Previous Chapter Next Chapter


Jaegwon Kim defined an intrinsic property, in effect, as a property that could belong to something that did not coexist with any contingent object wholly distinct from itself. Call such an object lonely or unaccompanied; and call an object accompanied iff it does coexist with some contingent object wholly distinct from itself. So an intrinsic property in the sense of Kim's definition is a property compatible with loneliness; in other words, a property that does not imply accompaniment.

David Lewis objected that loneliness itself is a property that could belong to something lonely, yet it is not an intrinsic property. He concluded that Kim's proposal failed. He also conjectured that nothing resembling Kim's definition would work, and if we want to define ‘intrinsic’ we had best try something altogether different.


That sweepingly negative judgement was premature. Though Kim's definition does indeed fail, a definition in much the same style may succeed.

First step. One intuitive idea is that an intrinsic property can be had by a thing whether it is lonelyor whether it is accompanied. It is compatible with either; it implies neither.

Second step. Another intuitive idea is that, although an intrinsic property is compatible with loneliness, a thing's being lonely is not what makes the thing have that property. Lacking the property also is compatible with loneliness. And likewise with accompaniment: if a property is intrinsic, being accompanied is not what makes something have that property.