# Understanding the intrinsic/extrinsic distinction

- First Online:

DOI: 10.1007/s11016-011-9549-x

- Cite this article as:
- Francescotti, R. Metascience (2012) 21: 91. doi:10.1007/s11016-011-9549-x

- 82 Downloads

The distinction between intrinsic and extrinsic properties is one of the most fundamental in the field of metaphysics, and there is no better place to learn about the difference between these two types of property than in Hoffmann-Kolss’ new book. Intuitively, a property P is an *intrinsic* property of an object *x* just in case *x*’s having P does not depend on the features of *x*’s environment, but only on what *x* is like in itself. Each of the various definitions of “intrinsic” and “extrinsic” offered in the contemporary literature aim to make this intuitive notion more precise. Hoffmann-Kolss divides these accounts into three main categories—modal, combinatorial, and relational. The first and largest part of her book is a detailed presentation and critique of various accounts that fit these three categories, culminating in her own version of the relational account that avoids problems she presents for the other analyses.

The modal accounts addressed include Lewis’ (1983) account of intrinsicality in terms of *duplication*, where an intrinsic property is characterized as one that could not possibly differ between duplicates, and the subsequent account that Lewis develops with Langton (1998), where intrinsic properties are defined as those that can be exemplified by an individual alone in the world, *independent of accompaniment*. Vallentyne’s (1997) and Yablo’s (1999) modal accounts are also reviewed.

Hoffman-Kolss argues that each of these modal accounts either misclassifies various disjunctive properties or rests on ill-defined notions. For instance, while Lewis defines intrinsicality with the notion of a duplicate, he also recognizes that the latter notion relies on the former: *x* and *y* are duplicates if and only if they share all the same intrinsic properties. Lewis appeals to the notion of a *natural* property to clarify what it is to duplicate in a way that does not rely on the concept of intrinsicality. Yet, Hoffmann-Kolss notes, the account is unsatisfactory since the controversial distinction between natural and non-natural properties is not independently defined. The Langton–Lewis analysis, on the other hand, does not adequately handle certain disjunctive properties, such as *being non*-*red and lonely or else red and accompanied*. This property intuitively seems to be extrinsic (since whether an object has it depends on whether other objects exist), but the property is classified as intrinsic by the Langton–Lewis account since it is exemplified independently of accompaniment by lonely non-red objects. They recognize this problem, conceding that disjunctive properties require special treatment in the framework of their theory. However, without clarification of what makes a property disjunctive, Hoffman-Kolss notes, “their account rests on an undefined notion and hence does not yield an adequate definition of the intrinsic/extrinsic distinction” (52–53). (It is shown that Vallentyne’s account also “ultimately founders on problems raised by disjunctive properties and has to be discarded” (62), and that the same is true of Yablo’s analysis.)

After critiquing combinational accounts, Hoffmann-Kolss argues in Sect. 3.4 of part I that “it is possible to devise a relational account which circumvents at least most of the difficulties encountered by accounts of the other two types” (80). Relational accounts of the intrinsic/extrinsic distinction appeal to the intuition that *x*’s extrinsic properties are ones that depend on *x*’s relation to wholly distinct items, i.e., to items other than *x* and *x*’s proper parts. She considers various approximations to an adequate relational account, and along the way addresses my (1999) analysis that introduces the notion of *d*-*relationality*. D-relational properties are those whose instantiation by an object *x* consists in *x*’s bearing certain relations to wholly distinct individuals, and *x* has a property P intrinsically just in case *x*’s having P consists in *x*’s having certain non-d-relational (*internal*) properties. Hoffmann-Kolss rightly notes that for this account to be acceptable, the meaning of “consists-in” will need to be made clear. I proposed that the consists-in relation may be understood in terms of *event/state identity*: *x*’s having P consists in *x*’s having Q just in case the *event* or *state* of *x*’s having P is identical with the *event* or *state* of *x*’s having Q (1999, 599). In response, she raises the following objection: For any arbitrarily chosen property P of object *x*, there is a relation *R* where *x* bears *R* to *y* just in case *x has P and y has exactly the same qualitative properties as x*. She notes (91–92) that without some suitable theory of events, one cannot justifiably deny on my account that *x*’s having P consists in *x*’s standing in relation *R* to all individuals with the same qualitative properties as *x*, in which case, one cannot justify classifying P (whatever P happens to be) as intrinsic.

One might be tempted to replace the obscure “consists in” talk with talk of *necessary equivalence*. The basic idea is that for P to be extrinsic is for there to be a relation *R* such that *necessarily* an object has P *if and only if* the object bears *R* to wholly distinct individuals. Of course, this basic idea does not help with the problem case Hoffmann-Kolss raises for my account, since necessarily *x* is made of tin if and only if *x* stands in the strange *R* relation she describes to all individuals with the same qualitative properties as *x*, in which case, being made of tin ends up being incorrectly classified as extrinsic. However, in Sect. 3.4.3 of part I, she provides a modified relational account in terms of necessary equivalence that manages to correctly classify being made of tin as intrinsic. Her analysis also provides a correct classification of various examples that threaten other accounts (relational or otherwise)—all without relying on controversial undefined notions, e.g., the *consists*-*in* relation or the unclear notions of *natural* and *disjunctive* properties.

Still, it is not clear that her account is to be preferred. By replacing the *consists*-*in* relation with *necessary equivalence*, the account fails where various *universally essential* properties are concerned. Many universally essential properties, e.g., *being such that 2* + *3* = *5*, seem to be extrinsic, but her account incorrectly classifies them as intrinsic. Early in her book, she mentions that she will exclude universally non-contingent properties from the scope of her discussion, claiming that “whether or not an individual *x* has a non-contingent property of this kind neither depends on what *x* is like, nor on the environment of *x*, but rather on the logical structure of the *property itself*” (22). But this reason for excluding universally essential properties is suspect. Being such that 2 + 3 = 5 does depend on *x*’s environment since it depends on the nature of objects (albeit necessary objects) distinct from *x* itself. Also, while being such that 2 + 3 = 5 is a *mere Cambridge* property of any contingent object, having nothing to do with what the object is like in itself, any acceptable account of extrinsicality should apply to the property and classify it as extrinsic, especially since it is extrinsic in the strongest possible sense. (Indeed, she devotes Sect. 5.2 to showing how her relational account can be extended to cover mere Cambridge properties.) An account that does not apply to universally essential properties, it seems, is not obviously superior to one that fails to apply to disjunctive properties or relies on certain undefined notions.

and its ¬P counterpart, (ii)-(c). These conditions seem to have the result that in cases where P fails to be an extrinsic property of(i)-(c) there are two individuals

z_{1}andz_{2}which both inhabit some possible worldwand instantiatePand an individualyalso inhabitingw, such thatz_{1}stands inRtoy, whereasz_{2}does not,

*x*by not meeting these conditions, P’s failing to extrinsic—thereby being intrinsic—depends on the condition of items (z

_{1}and z

_{2}) distinct from

*x*.

I would also like to add that the problem she presents for my account is not especially threatening. Take any property, P, of *x* and suppose that relation *R* is such that *x* bears *R* to *y* just in case *x has P and y has the same qualitative properties as x*. Her complaint is that without some suitable theory of events, we cannot justifiably deny that *x*’s *having P* is the same event as *x*’s *bearing R to all items with the same qualitative properties as x*, in which case, we cannot justify classifying P as intrinsic (whatever P happens to be). However, even without a theory of events, it seems we can still be quite sure that these are not quite the same events and that no plausible account of events would count them as the same. Perhaps the notion of an event is in itself problematic enough that no adequate analysis should rely on it. But even if this were true, the content of my account could quite easily be captured in other terms—identity of propositions, facts, states of affairs, or whatever one finds most tolerable. I agree with Hoffmann-Kolss that a relational approach is the way to go in defining the intrinsic/extrinsic distinction, but for the reasons I have given it seems doubtful that her analysis is the best way to go.

Despite potential objections to her relational account, the first part of the book is an impeccable presentation of the debates and concerns that arise when trying to define the intrinsic/extrinsic distinction. The second and shorter part of the book has two aims: (i) to present a notion of supervenience that is applicable to extrinsic higher-level properties, and (ii) to show that there are dispositional properties that must be classified as extrinsic. The motivation for (i) is the widespread belief that the ontological relation between higher-level and lower-level properties can be understood in terms of the supervenience relation together with the fact that many everyday descriptions and explanations of events appeal to extrinsic higher-level properties. Since the extrinsic properties of an individual do not supervene on its intrinsic properties, the supervenience base itself will need to include extrinsic properties. Yet, not just any extrinsic properties can be allowed for otherwise the supervenience thesis collapses into a *global* version (which allows *individuals* in the same possible world to differ in terms of supervening properties without differing with respect to the relevant subvening properties). The trick is to find a way to include in the supervenience base only those extrinsic properties that are relevant to the instantiation of the higher-level property. In Sect. 2.3.2 of part II, Hoffmann-Kolss develops the elegant idea of *property*-*dependent supervenience* to do just that.

In addition to explicating a brand of supervenience which does indeed achieve the goal of aptly applying to extrinsic properties, the thorough presentation and careful critique in part II of her book is highly beneficial to anyone wishing to survey the complex terrain of the literature on supervenience. It is also a great merit of part II that it clearly exposes the importance of the intrinsic/extrinsic distinction to understanding how the supervenience relation should be used to describe the ontological dependence of higher-level properties on lower-level properties. So despite the concerns I raise above about her own relational account, *The Metaphysics of Extrinsic Properties*, with the insightful discussion of the intrinsic/extrinsic distinction and its consequences for the supervenience debate, is a welcome and valuable contribution to the field of metaphysics.