This paper appeared in Philosophical Topics 35 (2007), pp. 343-58.

Power Essentialism
David Robb
Davidson College

I am suggesting, then, that we should regard mechanical transactions as fundamental in our examination of the notion of causality in general. They are fundamental to our own interventions in the world, to our bringing about purposed changes: (we put our shoulders to the wheel, our hands to the plough, push a pen or a button, pull a lever or a trigger). Entering into them ourselves, we find in them a source of the ideas of power and force, compulsion and constraint. Ourselves apart, they include observable natural phenomena, actions or relations directly detectable in the particular case, the observation of which supplies explanations of the states they end in.[1]

I. INTRODUCTION

Press a square paperweight with a smooth, marble surface into a pancake-sized lump of soft clay. What results is a square impression. Could a circular impression have resulted instead? Mesh two gears of the same radius. When gear A turns clockwise at one rpm, it causes gear B to turn counterclockwise at the same rate. Could A, still turning at one rpm, instead cause B to turn at fifteen rpm? Place a dodecahedron on an inclined plane. The plane is steep enough, and there is sufficient friction, so that the dodecahedron rolls clunkily down the plane. Could the dodecahedron instead roll smoothly down the plane, as smoothly as, say, a sphere would?[2]

In each of these cases, we could achieve the alternative effects by changing the properties of the objects. Make the paperweight circular, and it will cause a circular impression; increase A's radius; smooth out the edges of the dodecahedron. But suppose we make no such changes, holding constant the properties of the relevant objects, including the spatial relations among them, and fixing how properties are distributed in the surrounding conditions. Could the alternative effects occur? The answer seems to be no: there is at least the appearance of necessity here, the appearance, in particular, of necessary relations among properties.

Now examples of necessary, or internal, relations among properties are not, in general, hard to find. Scarlet is brighter than navy blue, and necessarily so: this is part of the natures, or essences, of these properties. Two pieces of a jigsaw puzzle match in a way necessitated by the natures of their shapes. A size intermediate between two other sizes is essentially so. Examples could be multiplied.[3] But what's distinctive about the earlier examples, and what makes them potentially more controversial than the rest, is that the causal powers bestowed by the salient properties take center stage. If there is an appearance of necessity here, it is the appearance that these properties bestow the powers they do essentially, in virtue of their natures. And there is a tradition in philosophy that denies there is any such necessary relation between a property and its powers to produce another. According to this tradition, if properties can be said to bestow causal powers at all, they do so only contingently.

In this paper, I propose to take the examples seriously as evidence against this tradition, as evidence, in fact, for the general thesis that the powers bestowed by a property are essential to it. I'll call this thesis power essentialism (sometimes just essentialism for short).[4] I do not claim that these examples present the only, or the best, case for power essentialism, but I will defend their evidential value.[5] Along the way, I will try to clarify essentialism and distinguish it from some closely related theses.

II. PRELIMINARIES

First, a few background remarks about properties, laws, and powers are in order.

Properties. I will in what follows assume a few theses about properties. First, and most fundamentally, there are properties. They are not to be eliminated as certain brands of nominalism would have it. Properties occupy our world and are in the causal mix. Talk of properties--and of causal powers--is not merely a manner of speaking, to be paraphrased away or given a deflationary or pleonastic reading. On my favored way of fleshing out this sort of realism, properties are ways objects are, what used to be called, by Locke and others, modes. As a first approximation, a property may be compared to a wave on the ocean. The wave is real--for example, you can see it, it knocks ships over, and so on--but it is merely a way some water is. Like waves, properties are real. They too can be seen--for example, we can see colors and shapes--and they too causally impact the world. The paperweight causes the square impression in virtue of being square; the dodecahedron rolls clunkily in virtue of its shape; and so on.

Second, properties are in the objects that have them. Just as a wave is where its water is, so a property is where its object is. On this view, then, properties are where they appear to be. Another advantage of in re realism is that it brings properties into the causal mix. If a shape is, here and now, in a paperweight or a gear, it is at least easier to see, even if not self-evident, how it could play a causal role in the squishing of clay or the moving of another gear.

Third, while my own view is that properties are particulars (tropes), I hope to remain neutral on this issue here. Two prominent ontologies respecting in re realism about properties are trope-theory and the "Aristotelian" view on which properties are in re universals. Power essentialism on either ontology will say that a property bestows the causal powers it does essentially. But where the universalist claims this about a single property, the trope-theorist will claim it about a host of resembling properties: each bestows the same causal powers, and essentially so. I believe that the following discussion could be put into either ontology's favored terms.

Fourth, properties are qualities. It is not easy to say, in an informative way, what it is for a property to be a quality. Paradigmatic qualities include the phenomenal properties of conscious experience, the "what it's like" of pains, sensations, mental images, and so on. But there are nonmental qualities as well, such as shape and volume, properties that fill space.[6] The properties to be held fixed in the examples opening this paper are of this sort. Given this distribution of qualities, can the effects vary? The power essentialist says they cannot. Saying that properties are qualities leaves open whether they are also powers. While I assume that properties bestow causal powers, the exact nature of this relation is left open in the examples. And this is as it should be, since the examples are to serve as data for informing an account of this relation.

Laws. The causal powers bestowed by a property are intimately connected to the laws of nature. In particular, I assume at a minimum that for every causal power, there is a law. However, this leaves open the ontological status of laws. Two broad conceptions of laws will be most relevant in what follows: transcendent and immanent.[7]

A transcendent law is a representation, say a proposition or statement of the form All Fs are Gs. Transcendent laws are countenanced by regularity theorists, but also by some philosophers who reject the regularity theory.[8] Such laws are transcendent in the sense that they are not present in the causal relations they subsume. Transcendent laws are not in these relations but are instead about them. Immanent laws, by contrast, are in the causal network. One prominent account of immanent laws is Armstrong's.[9] On his view, a law of nature is a state of affairs consisting of various properties related by nomic necessity, what Armstrong calls "N." If F and G are the properties N relates, the law is F's standing in N to G, or as Armstrong abbreviates it, N(F, G). This state of affairs, just like F and G, is present in the causal relations it subsumes.

A transcendent theorist is likely to accuse an immanent theorist of confusing laws, which are representations, with the properties and relations in the world that make these representations laws.[10] An immanent theorist is likely to accuse a transcendent theorist of confusing laws, which are worldly items, with representations of these laws. While my sympathies are with immanent laws, I will not here try to adjudicate between these rival theories.

Powers. Powers, I'll assume, are bestowed by properties, but I hope to remain neutral on some of the central debates on the nature of powers, debates concerning, for example, how power ascriptions should be analyzed, whether powers are intrinsic, and how exactly powers and laws are related.[11] That said, I do assume a few theses about powers here.

First, in the examples opening the paper, powers are present and manifested. The paperweight has the power to cause a square impression in clay. Gear A has the power to turn gear B. The dodecahedron has the power to roll (clunkily) down the inclined plane. These are all active powers, powers an object has to produce some effect. In each case, there are also corresponding passive powers. For example, the clay has the passive power to receive a square impression from the paperweight. Power essentialism is about both active and passive powers, but my focus in what follows will be on the former.[12]

Second, in these examples, the things that have the causal powers and that therefore do the causing are objects: a paperweight, a gear, a dodecahedron. And they have their powers in virtue of their properties. But are causes the objects themselves or something else, such as events, states of affairs, or properties? I'll avoid this controversy here, and instead just assume provisionally that objects are causes, remaining optimistic that adjustments can be made, if needed, under pressure to recognize more exotic entities such as events.

Third, it is frequently pointed out that the powers bestowed by a property are, at least in the typical case, conditional in the sense that a property bestows a power, not simpliciter, but only relative to, or in the context of, other properties.[13] Such enabling properties can be internal to the causing object or external to it. The shape of a knife bestows on it the power to cut bread, but only when the knife is made of steel or some similar material; this enabling property is internal to the causing object. A sprinter can run one hundred meters in under ten seconds, but only given sufficient oxygen in the vicinity; this enabler is external.

Whether one designates a given property as power-bestowing and another as a mere enabler can vary depending on the explanatory context. In one context, or for one set of purposes, one may think of the knife's shape as power-bestowing and its being made of steel as an enabler. In another context, or for another set of purposes, these designations may be reversed. It would be a mistake, however, to think that this somehow infects causal powers (or causation, for that matter) with a kind of mind dependence. The relations among properties leading to the effect are objective, even if which property we happen to single out as power-bestowing is not.

III. POWER ESSENTIALISM AND ITS SIBLINGS

With this background in place, power essentialism can be compared to a number of related theses.

(1) The laws of nature are necessary. Essentialism's relation to (1) turns on, among other things, what a law of nature is. First take a law to be transcendent, say a generalization of the form All Fs are Gs, where the predicates here pick out properties. Given power essentialism, if such a generalization is a law, it must remain so if the properties exist. But this need not mean that the generalization is necessary, for it would fail to be (nonvacuously) true if the properties in question were not to exist. In a world without F and G, it may be true that if these properties had existed, then it would have been a law that all Fs are Gs. But this is to just to say that in the absence of these properties, this is not a law, merely that it could have been. If this is right, essentialism does not entail (1).[14]

Further complications arise if we take laws to be immanent. Suppose a law of nature is not a universal generalization, but a state of affairs, N(F, G). Here (1) will assert, not the necessary truth of laws, but their necessary existence. But if properties are contingent, the lawful states of affairs will also be contingent, and so the laws again will not be necessary, even given power essentialism. That is, suppose essentialism is true, so that if F and G exist, so must N(F, G). This is compatible with the contingency of N(F, G), for this immanent law will fail to exist in the absence of F and G. Assuming, then, that which properties exist is a contingent matter, the laws of nature are not necessary.

The upshot is that whether power essentialism entails that the laws are necessary is a substantive matter depending on how a number of controversial issues are settled. Here I present the examples as evidence for essentialism, but leave it open whether (1) is thereby supported as well.

(2) Nomological necessity is a species of metaphysical necessity. I will have more to say about the kinds of necessity in the next section, but for now I just note that (2), like (1), is not clearly equivalent to essentialism. Consider the earlier example of a contingent, transcendent law: The law is the generalization that all Fs are Gs, a proposition that would fail to be nonvacuously true in the absence of F and G. Such a scenario is compatible with power essentialism. But then it's compatible with essentialism that there be propositions that are nomologically but not metaphysically necessary. In particular, if F and G exist, it will be nomologically but not metaphysically necessary that all Fs are Gs. Similar points apply if a law is instead taken to be immanent. This is enough to show that, at least under certain assumptions, essentialism does not entail (2).

(3) The laws of nature are fixed by (supervene on) what properties there are. Put another way: To create the laws of nature, all God must do is create the properties; the laws are an ontological free lunch. Whether laws are transcendent or immanent, it looks as if power essentialism entails this thesis. Whether it in turn entails power essentialism, however, is another matter. Suppose that "what properties there are" in (3) means "what properties exist and how they are distributed." In that case, (3) is too weak for the essentialist, since even a regularity theorist believes that once all of the properties are distributed--past, present, and future--the laws are thereby fixed,[15] and regularity theorists are not power essentialists. They believe--if they allow talk of causal powers at all--that a given property could have bestowed different causal powers. On the other hand, if "what properties there are" means merely "what properties exist," regularity theorists would deny (3), for they allow that the laws could have been different given the same properties, so long as these properties are distributed differently. And here I believe (3) entails power essentialism given a sufficiently robust reading of "fixed by": (3) should be true, that is, because the laws flow from the natures of the properties themselves, not simply because the properties and the laws are necessarily correlated. (Suppose an omnipotent and necessary being is by nature determined to pair certain properties with certain laws. In that case, (3) would be true, but only on a reading on which the laws are not a free lunch: the laws would be merely correlated, albeit necessarily, with certain properties. This would be too weak for the essentialist.)

(4) Properties are individuated by the causal powers they bestow. According to (4), properties, either in a world or across worlds, are identical if and only if they bestow the same causal powers. But whether properties are universals or tropes, essentialism does not entail (4).[16] In general, to say that something is essentially F is not to preclude there being a distinct thing that also is essentially F. Essences are not always individuating. My aim in this paper is to present evidence for power essentialism, not for this closely related thesis. I return now to this evidence.

IV. THE WRONG SORT OF NECESSITY?

In each of the examples presented, it seems that the alternative effects--round impression, rotation at fifteen rpm, smooth rolling--could not occur when the properties of the relevant objects are held fixed. Are there good reasons to deny the evidential value of these appearances?

One potential challenge comes from a distinction between kinds of modality. I have asked whether the square paperweight could have caused a round impression, but what sort of possibility is in play here: nomological or metaphysical? I take metaphysical modality to be the most fundamental sort, possibility and necessity tout court. The nomological variety can then be defined in terms of this. For example, P is nomologically possible just in case the conjunction of P and the laws of nature (or the propositions expressing them) is metaphysically possible--that is, just plain possible.[17]

One worry, then, is that while it does appear as if the alternative effects could not occur, this is merely the appearance of nomological impossibility, that is, (metaphysical) impossibility given the actual laws of nature. Fixing these laws, it appears as if the paperweight could not have caused a round impression, and so on. But this would provide no support for power essentialism, which requires that these alternative effects be metaphysically impossible, precluded by the natures of the properties. The modal appearances here are defeated by, so to speak, being deflated: they are appearances of the wrong sort of necessity.[18]

But the metaphysical modality is what I intend to be at work in the examples. Fix on this modality and consider the examples: could a square paperweight, under the specified conditions, cause a round impression in the clay? Could gear A, turning at one rpm, cause gear B to turn at fifteen rpm? Could the dodecahedron roll smoothly down the plane? It seems not. Comparing this with the earlier noncausal examples may be useful: Scarlet could not be darker than navy blue. The shapes of the puzzle pieces essentially match. One size must be intermediate between the other two. One need only consider the properties themselves to see this: one need not first hold fixed some regularity or law-like relation among them.

Nevertheless, some may insist that this is precisely the respect in which the causal examples are distinctive: it appears as if the alternatives could not occur, but only if nomological possibility is in play; these appearances are lost if the metaphysical could is intended. Or, in what seems to me a better way of putting the point, they allow that the alternative effects do seem (metaphysically) impossible, but only given that the laws of nature are held fixed.[19] My primary response to such an objection is in the previous paragraph, but in an attempt to advance the dialectic a bit further, I offer a secondary response in the form of a dilemma: either the laws that (allegedly) need to be held fixed are transcendent or they are immanent.

Start with the first horn. Imagine for a moment a much simpler objection to the first example: "Granted, it appears that the square paperweight cannot cause a circular impression, but this is only given that the shape of the paperweight bestows the power to cause a square impression. Let this fact vary, and the alternative effect no longer appears impossible." Clearly this objection is harmless. Allowing the fact in question to vary does permit the alternative effects, but the point of the examples is to discover whether it can vary.

But it seems to me that matters are not significantly different if what's allowed to vary is a representation such as a universal generalization. Let some regularity of the form All Fs are Gs be the law subsuming the causal interaction in the paperweight example. Now the objector is saying that so long as this law is held fixed, it does appear that the paperweight must result in a square impression. Remove this restriction, says the objector, and the appearance of necessity vanishes.

It must be granted that once the truth of this representation of the causal interaction is allowed to vary, one has no idea of whether the paperweight will cause a square impression. But the point of the example is to see whether the truth of this description can vary. Transcendent laws are ontologically posterior to the causal relations they subsume. All of the action is going on in the objects and their properties. At best, a transcendent law can report on this action. So one cannot allow transcendent laws to vary, nor can one hold them fixed, until one has already determined whether the causal powers bestowed by the relevant properties can vary. And determining this is the aim of the examples.

Next, take the laws to be immanent, and start again with a naive form of the objection: Suppose a law of nature is not a state of affairs involving nomic necessitation, but just another ordinary property (F) on the scene. An objector says that holding F fixed, the alternative effect seems impossible, but not if F is allowed to vary. But granting this, how has it weakened the case for power essentialism? The essentialist says it's part of the essence of the paperweight's shape that it bestow the power to cause a square impression in soft clay. But properties bestow causal powers only relative to certain enablers. The objection on its current version would at most have shown, not that the shape of the paperweight bestows its causal powers contingently, but that F is one of the enabling properties relative to which the shape bestows the powers manifested in this example.

Now switch to a more plausible immanent account of laws. Suppose a law of nature is a state of affairs having as a constituent a higher-order relation among properties, Armstrong's N. Grant the objector that the alternative effects seem impossible only if the relevant properties are given as N-related. But again, how would this weaken the evidence for power essentialism? At most the objector would have shown that the properties in question seem necessarily to bestow their causal powers, but only relative to the enabling N.

One may protest that N could not be an enabling property, since it's not a property at all, but a relation among properties. There are two potentially relevant distinctions here, one between properties and relations, the other between first-order relations, holding among objects, and second-order relations, holding among properties. But while these distinctions are important for a general ontology of properties, I don't think they are going to be as crucial for present purposes. As long as we restrict properties to what's qualitative, we can, without affecting the essentialist thesis, extend "property" to cover relations, even second-order relations among properties. After all, in some causal interactions, one property's being three feet from another is needed for the first to exercise certain causal powers. But there should be nothing in power essentialism banning this second-order relation as an enabling property. And similarly, whatever N is, I do not see any barrier to including it among the enabling properties.[20] Its distinguishing feature, if the objector is right, is that it is a universal enabling property, an enabler with respect to the causal powers bestowed by every property. But a power essentialist can accept universal enablers. For example, being in space, if it's a property at all, is plausibly a universal enabling property; an essentialist need have no objection to this.

There's another "wrong sort of necessity" objection similar to the foregoing, though now the alleged troublemaker is not a law of nature, but the wider causal background against which we are to evaluate the relevant possibilities. Consider the square paperweight again, which will cause a square impression in soft clay, but only against an external background whose character I loosely described as "the surrounding conditions," but which I cannot fully specify. After all, virtually anything might prevent a square impression: a strange field may affect the clay at the moment of contact; the spacetime region may distort in unexpected ways; and so on. So if the square paperweight must cause a square impression, the worry goes, this is not the metaphysical must, but some kind of weaker, ceteris paribus necessity. In that case, the appearance of necessity here provides no help for the essentialist, who requires necessity tout court.

This objection is structurally similar to the previous one, as are the replies to it. First, we should run the examples with the fundamental modality, not any weaker sort. And second, in any case, the properties making up the causal background should be included among the enabling properties relative to which the shape of the paperweight bestows the power in question, and similarly for the other examples. What the objector thinks of as a weaker brand of necessity is just metaphysical necessity relative to the causal background. As long as this background includes only qualities,[21] there need be nothing here to undermine the use of these examples in support of essentialism.

Granted, even if we are aware of the most salient of these background properties, some remain unknown. These unknown qualities may be collectively picked out by a ceteris paribus gesture, but this would not reveal their nature. But if we don't know precisely which properties to hold fixed, can we trust the resulting appearances? Such worries are important, but they should not be exaggerated. After all, we seem to be able to ascertain the actual causal powers bestowed by a property without being able to fully grasp the enabling background. We know that squareness actually bestows on the paperweight the power to create a square impression even if we can't fully specify all of the properties relative to which it bestows this power. Why, then, can't we exploit this grasp we have, however imperfect, on the qualitative background in ascertaining whether the shape essentially bestows the power? If our fix on the background is adequate enough for us to know what the shape of the paperweight is actually doing, it should, it seems, be enough for us to know what the shape could be doing in what are stipulated to be those same circumstances. (Consider, by analogy, the holistic background arguably necessary for the ascription of a belief. Few could specify this background in all of its detail, but our imperfect grasp of it nevertheless grounds, not just our ascriptions of a belief, but our essentialist claims about it. For example, I am not entirely sure what doxastic background is required for a subject to believe that Obama is president, but I can nevertheless tell that in the context of this background, such a belief essentially disposes a believer to infer that someone is president.)

V. MUST POWERS BE HELD FIXED?

A second line of objection says that the alternative effects appear impossible only given that the rigidity and impenetrability of the relevant objects are held fixed. The paperweight will cause a square impression, gear A will turn gear B at an equal rate, and the dodecahedron will roll clunkily only given that these things are and remain rigid and impenetrable. But, and this is the crux of the worry, these are powers: To say that a material object is rigid is to say that it can't be deformed; to say that it's impenetrable is to say that it excludes other material objects from the space it occupies. If these powers must be held fixed, then, says the objector, the essentialist cannot appeal to the examples as evidence. For the point is to determine whether powers can vary when certain properties are given. If, to get the appearance of necessity, we must hold fixed some of the powers, then powers have infected the examples, and the resulting modal appearances are useless as evidence for essentialism.

The first move to make here on behalf of the essentialist is to extend the notion of an enabler to powers as well, allowing rigidity and impenetrability to be enabling powers relative to which the shape of the paperweight bestows the power to cause a square impression, and similarly for the other examples. A better way to put this reply, I think, is to say as before that only properties enable, but that (sometimes) they perform this role only when they bestow certain powers. So let Qr and Qi be properties bestowing the powers of rigidity and impenetrability in the paperweight example, and for the sake of simplicity, suppose these are the only enabling properties relevant to the case. Earlier I wanted to say that the shape of the paperweight bestows the power to cause a square impression only relative to Qr and Qi, and that this power is essential to it. Now, under pressure from the present worry, I apparently need to say that the power is enabled by Qr and Qi, but only when these properties themselves bestow the powers of rigidity and impenetrability.

But this needn't render the examples useless as evidence for essentialism. The powers bestowed by Qr and Qi are not the powers bestowed by shape. So it's not as if fixing the objects as rigid and impenetrable prejudges the issue of whether the shape could fail to produce a square impression. Under these constraints, it is still a substantive question whether a property essentially bestows the powers it does. And the modal appearances delivered by the examples can still be useful in considering this substantive thesis. An instructive, albeit imperfect, analogy is one of the earlier noncausal examples: Suppose a given size S2 is intermediate between sizes S1 and S3. This appears to be an essential feature of S2, and such appearances should be taken seriously in spite of the fact that when we consider whether S2 can vary in this respect, we hold fixed other sizes, in particular, S1 and S3.

There is a more fundamental response available to the current objection, one I'd like to explore, if only briefly. I've so far conceded that when considering the examples, we need to fix, not just Qr and Qi, but their bestowing rigidity and impenetrability. But this concession may not be needed. How do things seem if we fix only these properties, and leave open whether they bestow their respective powers? One problem here is grasping Qr and Qi themselves. What properties are they exactly?

Start with a closely similar case. Locke said we get the idea of solidity--and so the idea of body, he thought--by experiencing resistance to pressure:

The Idea of Solidity we receive by our Touch; and it arises from the resistance which we find in Body, to the entrance of any other Body into the Place it possesses, till it has left it.

If any one asks me, What this Solidity is, I send him to his Senses to inform him: Let him put a Flint, or a Foot-ball between his Hands; and then endeavour to join them, and he will know.[22]

But do these procedures--for example, trying to squash a football--give us the idea of a power, or do they give us the idea of a property (quality) bestowing that power? Between these passages, Locke remarks that with the idea of solidity we conceive a body "to fill space."[23] But space-filling is, or so I claimed earlier, a defining feature of a property, that is, a quality. And it does seem as if one is acquainted with a property when one tries to squash a football. When I try to do this, I get the impression of a property as clearly as I do when I, say, look at a shaped object. It may be that I can see the shape only in virtue of the shape's bestowing certain powers (relative to my visual system), but for all that, it is the shape itself I perceive.[24] Similarly, even if the sensation I experience when trying to squash a football is the manifestation of one of the ball's powers, I may for all that perceive the property bestowing this power.

Suppose for a moment that this is right. The present objector insists that for the alternative effects to appear impossible, one must hold fixed certain powers, such as the rigidity and impenetrability of the relevant objects. Above I conceded this, but I may not need to. Suppose that with the help of some Locke-inspired experiment, we are able to apprehend Qr and Qi, and suppose the examples are then considered while holding fixed these properties, but without taking any stand on what thereby happens to the powers. I submit that under these circumstances, we will get the same modal appearances: it will still seem to us as if the paperweight cannot cause a circular impression, that gear A cannot turn gear B at fifteen rpm, and so on. Things will appear this way only because the relevant objects are rigid and impenetrable, but their having such powers is, like the power to bestow a square impression, discerned in the examples, not loaded in beforehand. Insofar as we grasp the relevant properties of the paperweight and other objects, including Qr and Qi, it appears that they bestow their causal powers essentially.

VI. ARE THE EXAMPLES TOO SIMPLE?

Finally, it may be protested that the examples opening the paper involve only simple properties such as shape and size and their crude, mechanistic powers. Such examples may have held sway in the seventeenth century, says the objector, but they are unimpressive today. Consider examples using, say, the charge or spin of an electron, and we lose whatever essentialist-friendly appearances we might have had.

But what the objector here sees as a defect in the examples, I see as their primary virtue. If power essentialism is to be tested with examples, they must be examples in which, if the relevant properties bestow their causal powers essentially, this will be discernible to us. Too often, discussions of properties and powers focus on properties whose natures are obscure to us. If, say, charge bestows the causal powers it does essentially, then it's unlikely we'd be able to tell this: the nature of charge is not evident. One should not test a thesis with examples that we know ahead of time will hide the truth (or falsity) of the thesis from us.[25]

A brief digression into Kripkean a posteriori necessities may be useful here.[26] I am not here going to investigate the extent to which the evidence for power essentialism provided by the examples is, in some interesting sense, a priori.[27] But suppose for a moment that the examples include a significant empirical component. In particular, suppose we can know only empirically the actual causal powers bestowed by shape and the like. In that case, saying that the shape of the paperweight essentially bestows the power to cause a square impression would be relevantly similar to saying that water essentially has the molecular structure H2O.

Now what is the epistemic relation between this claim about water and the more general thesis that a chemical kind has its molecular (or deep explanatory) structure essentially? Here is one answer: We discover empirically that water is in fact H2O, then draw on our knowledge, presumably a priori, of the general thesis about chemical kinds. From the empirical discovery plus the general thesis, we conclude that water is essentially H2O.[28] But on an alternative answer, the general thesis comes in later. As before, we discover empirically that water is H2O, but we then consider whether water could have had a molecular structure other than this one. Since it appears it could not, we use this and similar examples as evidence for the general thesis that a chemical kind has its molecular structure essentially.

Regardless of which procedure Kripkeans typically follow in the water case, it seems to me that in the case of power essentialism and the examples I've used to support it, we should proceed in the second way (again, assuming for the sake of discussion that the essentialist claims about shape and the like are a posteriori necessities). While it may seem clear that, in general, the molecular structure of a chemical kind is essential to it, I find no such clarity when I consider power essentialism in itself, in abstraction from any particular examples. I would not then want to reason top-down from a general essentialist thesis as in the first way, but rather bottom-up, consulting how things seem concerning the modal variability of the actual causal powers of shape and the like, then generalizing from these particular essentialist claims to the more general thesis. And I do not see how we could follow this procedure if we started with properties, such as charge, whose nature is obscure to us. Like charges repel; could they attract instead? I draw a blank in this case. If there is bottom-up evidence to be had, it is from properties we have a chance of understanding.

There is, granted, a danger in using familiar properties to test metaphysical theses such as power essentialism. It is a danger closely related to one Hume warns of in the Enquiry:

We are apt to imagine, that we could discover these effects by the mere operation of our reason, without experience. We fancy, that were we brought, on a sudden, into this world, we could at first have inferred, that one billiard-ball would communicate motion to another upon impulse; and that we needed not to have waited for the event, in order to pronounce with certainty concerning it. Such is the influence of custom, that, where it is strongest, it not only covers our natural ignorance, but even conceals itself, and seems not to take place, merely because it is found in the highest degree.[29]

Now as I mentioned above, I have not claimed that the causal powers bestowed by, say, the shape of paperweight are knowable a priori, "by the mere operation of our reason". Still, a kind of rational insight is at work in the examples--at least, it is if the essentialist is to make use of them. Even if we can know only empirically the actual causal powers bestowed by the shape, we can know that it essentially bestows those powers only via rational insight. And one may object, in the spirit of the Hume's remarks, that our alleged insight is suspect here, for we are apt to confuse it with empirical familiarity. Perhaps an everyday, macroscopic property seems to essentially bestow a set of causal powers only because it's so often been observed to do so.

Such skeptical challenges to rational insight are important, and I cannot answer them in full here.[30] But I can say that whatever force they might have should not be limited to the causal examples. Return once more to the noncausal examples from the opening section: Might a skeptic, inspired by Hume's remarks, worry that scarlet seems essentially brighter than navy blue only because it's always been observed to be so? Do the two puzzle pieces whose shapes seem necessarily to match appear so only because such shapes have been observed always to match? Have we mistaken the essence of size S2 for a more mundane empirical regularity, that things of S2 are in fact always intermediate between things of sizes S1 and S3? I take it that the skeptical challenge here is not especially impressive. Why rational insight is not threatened by the challenge in such cases is a question I cannot answer here, but that it's not threatened seems clear enough. Turning to the powers bestowed by properties, then, should matters be any different?

ACKNOWLEDGMENTS

An early draft of this paper was completed in 2004 while I was on sabbatical from Davidson College and a visiting scholar at the University of Colorado at Boulder. I thank both institutions for their support. This paper grew out of a series of conversations with George Bealer. Some of the ideas here (especially the remarks on immanent laws in section 4) are from these conversations, though Bealer should not be held responsible for what I've done with his suggestions. I thank Jonathan Lowe, Dan Korman, and Barry Ward for detailed comments on later drafts. And finally, though I do not expect that Sydney Shoemaker and John Heil would always agree with how I've put things, it will be clear to anyone reading this paper that my intellectual debts to each philosopher are considerable.

NOTES

[1] P. F. Strawson, "Causation and Explanation" in Essays on Davidson: Actions and Events, ed. B. Vermazen and M. B. Hintikka (Oxford: Clarendon Press, 1985), 115-35, at 124.

[2] A version of the gear example is in D. H. Sanford, "Causal Necessity and Logical Necessity," Philosophical Studies 28 (1975): 103-12; "Causal Necessity and Logical Necessity (Reply to Brand and Swain)," Philosophical Studies 33 (1978): 185-94; "Causation and Intelligibility," Philosophy 69 (1994): 55-67. I believe I first heard about the rolling dodecahedron from John Heil in conversation; see also the examples discussed in his review of Peter Unger's All the Power in the World in Nous 42 (2008): 336-48. There are many other examples in the same family, such as the brick in C. A. Mace, "Mechanical and Teleological Causation," Proceedings of the Aristotelian Society, Suppl. 14 (1935): 22-45, at 42-3; the peg in C. B. Martin, "On the Need for Properties: The Road to Pythagoreanism and Back," Synthese 112 (1997): 193-231, at 220; the knife in S. Shoemaker, Identity, Cause, and Mind, Expanded Edition (Oxford: Clarendon Press, 2003), 212; the pressing hand in E. Fales, Causation and Universals (London: Routledge, 1990), 15ff.

[3] See B. Blanshard, Reason and Analysis (La Salle, IL: Open Court Publishing Company, 1962), 449-50; C. L. Elder, Real Natures and Familiar Objects (Cambridge, Mass: MIT Press, 2004), ch. 2; G. Molnar, Powers (Oxford: Oxford University Press, 2003), 206-7.

[4] One might also call it dispositional essentialism or dispositionalism, though given how I understand properties (see the next section), power essentialism is not clearly equivalent to all of the views that go under these alternative labels. In any case, views in this family are defended in, e.g., Shoemaker, Identity, Cause, and Mind, chs. 10, 18; C. Swoyer, "The Nature of Natural Laws," Australasian Journal of Philosophy 60 (1982): 203-23; C. L. Elder, "Laws, Natures, and Contingent Necessities," Philosophy and Phenomenological Research 54 (1994): 649-67; C. B. Martin and J. Heil, "The Ontological Turn," Midwest Studies in Philosophy 23 (1999), 34-60: J. Heil, From an Ontological Point of View (Oxford: Clarendon Press, 2003), ch. 11; A. Bird, "The Dispositionalist Conception of Laws," Foundations of Science 10 (2005): 353-70; J. Bigelow, B. Ellis, and C. Lierse, "The World as One of a Kind: Natural Necessity and Laws of Nature," British Journal for the Philosophy of Science 43 (1992): 371-88; B. Ellis and C. Lierse, "Dispositional Essentialism," Australasian Journal of Philosophy 72 (1994): 27-45; B. Ellis, The Philosophy of Nature: A Guide to the New Essentialism (Chesham: Acumen Publishing Limited, 2002).

[5] Whether the examples have much to tell us about causation itself--either its metaphysics or its epistemology--is not something I explore here. For discussion of such a project, see, e.g., the papers by Strawson and Sanford cited earlier.

[6] On the need for qualities to fill in space, see, e.g., J. L. Mackie, Problems from Locke (Oxford: Clarendon Press, 1976): 24ff.; S. Blackburn, "Filling in Space," Analysis 50 (1990): 62-65; P. Unger, All the Power in the World (Oxford: Oxford University Press, 2006), ch. 1. In spite of arguing persuasively for qualities, Unger declines to reify them and prefers instead to speak only of "qualitied particulars."

[7] Power essentialism may also be compatible with some version of the "lawless" view defended in S. Mumford, Laws in Nature (London: Routledge, 2004), but I do not explore this here.

[8] They include J. Carroll, Laws of Nature (Cambridge: Cambridge University Press, 1994); Ellis, The Philosophy of Nature.

[9] D. M. Armstrong, What is a Law of Nature? (Cambridge: Cambridge University Press, 1983). While I'll take this to be my representative immanent account, I should note that Armstrong is not a power essentialist.

[10] Thus, e.g., Ellis and Lierse, "Dispositional Essentialism," 39: "Laws are not things which exist in the world; they are things which are true of the world... the truthmakers for the relevant laws of nature are, we hold, just the fundamental dispositional properties."

[11] See, e.g., Molnar, Powers.

[12] The active-passive terminology is in J. Locke, An Essay Concerning Human Understanding, ed. P. H. Nidditch (Oxford: Clarendon Press, 1975), 234 (II.xxi.2). Compare Shoemaker, Identity, Cause, and Mind, 412 on the distinction between forward-looking and backward-looking powers.

[13] See, for example, the "enabling conditions" in R. Harre and E. H. Madden, Causal Powers: A Theory of Natural Necessity (Totowa, NJ: Rowman and Littlefield, 1975), 88; the "conditional powers" in Shoemaker, Identity, Cause, and Mind, 212-13; the "reciprocal disposition partners" in Martin, "On the Need for Properties," 206.

[14] Compare the "conditional" necessity that (transcendent) laws are said to have in Bigelow, Ellis, and Lierse, "The World as One of a Kind," 379; see also A. Bird, "Strong Necessitarianism: The Nomological Identity of Possible Worlds," Ratio 17 (2004), 256-76; Heil, From an Ontological Point of View, 94.

[15] Compare the Humean supervenience articulated in D. Lewis, Philosophical Papers, Vol. II (Oxford: Oxford University Press, 1986), ix-x.

[16] J. Hawthorne, "Causal Structuralism", Philosophical Perspectives 15 (2001): 361-78, at 361-62.

[17] Compare P. van Inwagen, "Modal Epistemology," Philosophical Studies 92 (1998): 67-84, at 72.

[18] A quite different "wrong sort of necessity" objection says that what necessity there is in the examples is logical or conceptual: the descriptions setting up the examples, the objection goes, analytically guarantee the standard effects. But even if there are some conceptual truths here, the more interesting necessities confronting us in the examples, and the necessities relevant to power essentialism, are de re. The descriptions setting up the examples are needed to set the stage, but then we are to focus on the properties themselves. Fixing those very properties, could the alternative effects occur? (Compare J. R. Searle, Intentionality [Cambridge: Cambridge University Press, 1983], 121.) I take it the point is familiar enough from non-causal cases. For example, while it might be analytic that scarlet is brighter than navy blue, the corresponding de re essentialist claim is more controversial and concerns the natures of the properties themselves.

[19] Compare J. Schaffer, "Quiddistic Knowledge," Philosophical Studies 123 (2005): 1-32, at 7.

[20] T. Handfield, "Armstrong and the Modal Inversion of Dispositions," Philosophical Quarterly 55 (2005): 452-61, at 458 makes what I take to be a closely related point. Yet another point in the same neighborhood--though one I find more elusive--is this: There seems to be nothing special about N; it's just another property (in our widened sense), albeit one that's a universal enabler. Independently what of Armstrong calls it, why should N be accorded the special status of necessitation such that holding it fixed in the examples drains them of their significance? David Lewis asks a similar question (but with a quite different end in view) in "New Work for a Theory of Universals," Australasian Journal of Philosophy 61 (1983): 343-77, at 366.

[21] Such a restriction will block phony properties that, for familiar reasons, might render the examples trivial. Such pretenders include not containing anything to prevent a square impression, and being such as to result in a square impression.

[22] Locke, Essay, 122-23, 126-27 (II.iv.1, II.iv.6).

[23] Locke, Essay, 123 (II.iv.2). See also the passage citer earlier from Mackie, Problems from Locke. For skepticism that Locke's experiment acquaints us with a quality (or a "categorical ground"), see A. D. Smith, "Of Primary and Secondary Qualities," Philosophical Review 99 (1990): 221-54, at 245.

[24] In general, from the fact that we can detect a property (quality) only insofar as it bestows powers to affect us, it does not follow that we cannot perceive the property itself. To think such a skeptical conclusion does follow would be to commit a fallacy closely related to what David Stove calls the "Gem" in The Plato Cult and Other Philosophical Follies (Oxford: Basil Blackwell, 1991), 148ff.

[25] I get some inspiration here from D. C. Williams, who investigates "the elements of being" by starting with the shapes and colors of lollipops, "frivolous" yet "peculiarly perspicuous" examples, in Principles of Empirical Realism (Springfield, Ill.: Charles C. Thomas, 1966), 75.

[26] S. A. Kripke, Naming and Necessity (Cambridge, Mass: Harvard University Press, 1980); the connection between Kripkean a posteriori necessities and power essentialism is discussed in Shoemaker, Identity, Cause, and Mind, ch. 18.

[27] See Sanford, "Causation and Intelligibility."

[28] A version of this answer is in A. Sidelle, Necessity, Essence, and Individuation (Ithaca, NY: Cornell University Press, 1989), ch. 2.

[29] D. Hume, An Enquiry concerning Human Understanding, ed. T. L. Beauchamp (Oxford: Oxford University Press, 1999), 110 (4.1).

[30] But see L. BonJour, In Defense of Pure Reason (Cambridge: Cambridge University Press, 1998).