1. Introduction
The question of which entities can be related as cause and effect is a matter of some dispute. The view that the causal relata are events rather than facts has emerged as the front-runner. But talk of ‘events’ disguises a substantial rift between those who think of them as coarse-grained particulars and those who think of them as akin to states of affairs, such as Kimean fine-grained events or tropes (see, for instance, Davidson Reference Davidson1967 and Kim Reference Kim, Brand and Walton1976). Amidst this controversy, however, one thesis has gained widespread acceptance: the negative claim that substances aren't causes (notable recent exceptions are Lowe Reference Lowe2008 and Steward Reference Steward2012).
Confidence in this thesis has infiltrated other areas of philosophy. For instance, even those who accept that agents are causes agree that such causation is atypical. It is usually restricted to agents, or free agents, and is contrasted sharply with the standard model of event causation that encompasses the rest of nature. Moreover, many philosophers argue that the reliance on this mysterious form of causation makes such accounts of free will and agency untenable.
The aim of this paper is primarily that of shifting the burden: the case against substance causation is unproven. Indeed, there is a prima facie sound argument for substance causation. Therefore, opponents of substance causation must do more to defend their position. In section 2 I outline the thesis I wish to rehabilitate. In section 3 I present a simple argument for substance causation. In section 4 I present and reject the worry that this argument is guilty of equivocation. Finally, in section 5, I answer objections to the view that substances can be causes.
2. The Thesis
Something must first be said about what substances are supposed to be. I am concerned with the commonsense notion of concrete substances or objects, those that occupy space and time. I will use the term ‘substance’ to include entities such as (portions of) wine, methane, or gold, not referable to by count nouns and so not naturally thought of as objects. But paradigm instances of objects, such as people, dogs, watches, and cakes as well as micro-objects, such as electrons, neutrons, and protons are all included.
What distinguishes substance causes from rival candidates, such as events or property instances? Intuitively, substances are self-standing entities—they instantiate properties but are never themselves instantiated and are those entities upon whose existence property instances depend. Substances are usually found to participate in events. They are capable of change and can persist through it. In short, by ‘substance’ I mean only to refer to those entities that folk metaphysics classifies as substances or objects (for a more detailed characterization of our intuitive notion of substance, see Hoffman and Rosenkrantz [Reference Hoffman and Rosenkrantz1994: ch.1, §3]). This rough characterization suffices for my purposes, since it latches onto precisely the kind of entity that has preoccupied those philosophers skeptical of the notion of substance causation.
With this understanding of substance in place, the thesis in question can be stated simply:
SC: Some actual substances are causes.Footnote 1
Thus, SC commits one to the claim that there are some substances in this world (and, I want to claim too, in possible worlds like ours) that stand in the relation of causation. For ease of presentation, I assume there is only one relation of causation, but SC is not committed to this. If there is more than one relation of causation, SC would be true if substances stood in any one (or more) of those relations.
SC, then, commits us to the reality of substance causation. But we should note that it does not commit us to the possibility of substance causation without event causation. (‘Event causation’ is being construed broadly here to include both coarse-grained events and Kimean events, which are entities that are like states of affairs. Below, I will consider the two notions separately.) Although there are apparent instances of substances displaying their causal powers without any triggering events—my jumper keeps me warm; my desk prevents my computer from falling; the magnet holds the postcard to the fridge, etc.—perhaps all these cases really involve event causation too. For instance, it may be argued that there is some state of affairs or event-like entity here, such as the presence of the jumper/desk/magnet, causing the effect.
If there are no instances of substance causation without event causation, can't we simply dispense with the former? More will be said regarding this point later, but, briefly, to maintain the irreducibility of substance causation we need not insist that substance causation can occur without event causation. To take a rather hackneyed analogy, the property of being triangular cannot occur without the property of being trilateral, but that does not show that being triangular is reducible to being trilateral. Similarly, substance and event causation could be irreducible yet mutually dependent.
One might argue that defending the irreducibility of substance causation to other forms of causation requires taking a stand on the nature of substances, since substance causation may ultimately be reducible to event or trope causation if substances are just collections of events or tropes.Footnote 2 As I made clear, I offer no theory of substance here. But what we can still do, and what I aim to do in lieu of a theory of substance, is pursue the question of whether those entities we ordinarily classify as substances can stand in the relation of causation.
3. An Argument for Substance Causation
In a brief but pregnant remark, Lowe notes that substances may be thought to be the most fundamental causes; ‘it is, after all, substances, not events, that possess causal powers and liabilities’ (Reference Lowe2008: 143). If substances are the bearers of causal powers, a simple argument for substance causation suggests itself:
The Argument from Causal Powers
POWER: Some actual substances possess causal powers.
EFFICACY: If a substance possesses a causal power, then it is efficacious.
CAUSE: If a substance is efficacious, then it can be a cause.
MANIFEST: Some actual substances' causal powers are manifested.
Therefore,
SC: Some actual substances are causes.
This seems to be a valid argument, with plausible-sounding premises, but it leads to a conclusion that is widely rejected. In the remainder of section 3, I consider each premise, arguing that such widespread rejection is unmotivated.
3.1 Power
Power is supported by our standard attributions of causal powers. It is the sugar that is soluble; the arsenic that is poisonous; the vase that is fragile; the electron that is negatively charged, and so on. Substances are the entities that manifest the responses constitutive of these paradigmatic causal powers. It is they that break, dissolve, etc. Perhaps other kinds of entities can also be correctly attributed causal powers, but that is no reason to reject premise one.
Philosophers do sometimes talk of the powers of properties. Ehring, for example, writes, ‘What a property is causally relevant to will depend, in part, on its causal powers’ (Reference Ehring2003: 367); and Clarke talks of ‘the properties that carry causal powers’ (Reference Clarke2003: 207). One interpretation of such statements is that they express the view (PP) that it is properties, rather than substances, that are the real bearers of causal powers. On such a view, then, properties alone are the entities to which we should attribute causal powers.
Since PP is revisionary, however, we are owed an argument for it. In crediting PP to Bird, Menzies (Reference Menzies2009: 773) suggests that arguments for dispositional essentialism are arguments for PP. But this is not so. Bird's view is that ‘fundamental properties are potentialities—they have dispositional essences’ (Reference Bird2007: 5), and to say that the fundamental properties are powers is not to say that properties have powers. If properties are powers and substances are the bearers of properties, dispositional essentialism ascribes powers to substances, not to properties. Arguments for dispositional essentialism, if sound, would thus support, rather than undermine, Power.
An alternative way of challenging Power would be to argue that causal powers are never correctly attributed to objects. Even Ryle, however, who thought that it was a mistake to look for the truthmakers of dispositional statements in observable or unobservable states of affairs (1949: 120), thought that dispositions could be truly attributed to substances. These attributions, despite not being fact-conveying, contain important information that enables us to make inferences about what would and does happen (see Ryle's discussion of dispositional statements as ‘inference-tickets’ [1949: 119]). So Power would be acceptable to him. Similarly, while there are those who think that our commonsense notion of causal power is in some way mistaken, the demand that there be some account, albeit reductive, of the many dispositional and power ascriptions within the sciences is widely accepted (see, for instance, Lewis Reference Lewis1997). Such reductive accounts of powers and dispositions do not obviously threaten Power, since they allow that causal powers can be correctly attributed to objects. On further elaboration, such views may present a challenge to the argument, but this issue will be addressed in section 4.2.
3.2 Efficacy
The claim that if an entity possesses causal powers, then it is efficacious (or, more fully, causally efficacious) seems uncontroversial. The notions of causal power and efficacy are so closely related that they are often used interchangeably. Rosen, for instance, writes,
The most widely accepted version of the Way of Negation has it that abstract objects are distinguished by their causal inefficacy. Concrete objects (whether mental or physical) have causal powers; numbers and functions and the rest make nothing happen. (2009, §3.2)
Powers can characterize objects whether or not they are manifested. A substance's being efficacious therefore does not entail that its power is manifested. But possessing a causal power does entail that an entity is efficacious. If a medicine can make you better, then it is efficacious. Medicine doesn't have to be making you better to be efficacious, but it can and, one hopes, will do so if taken properly. If, as Power claims, some actual substances possess causal powers, then those substances are efficacious.
3.3 Cause
Cause likewise sounds intuitively secure. The concept of efficacy, in our ordinary vernacular, is not clearly distinguished from that of cause. In the Oxford English Dictionary, for example, ‘efficacious’ is defined as ‘That produces, or is certain to produce, the intended or appropriate effect’ (www.oed.com/view/Entry/59732). If the medicine is efficacious, then it can make you better. There may be countervailing factors so that, in a given circumstance, the medicine fails to do this. Perhaps you do not take it correctly, or you are allergic to the medicine, or there is only a high probability of its killing the illness-inducing bacteria. However, if the circumstances are right and perhaps you are lucky, then the medicine will work. That is, if the medicine is efficacious, then it can be a cause of your recovery.
3.4 Manifest
Manifest simply allows us to move from the claim that some actual substances can be causes to the claim that they are. Not only is sugar soluble, it sometimes dissolves; not only is arsenic poisonous, it sometimes poisons; and so on.
One might object that Manifest is not innocent, since the manifestation conditions of causal powers need not obtain. We could do without Manifest by weakening SC to: Some substances in the actual world, or sufficiently similar possible worlds, are causes. But I think that this is unnecessary. If we grant that there are objects with causal powers in the actual world, it seems manifestly false to claim that none of the manifestation conditions of these powers ever do or ever have obtained throughout the earth's entire history.
4. Two Types of Efficacy
It may be objected that the argument from causal powers is, despite appearances, invalid since it equivocates over two distinct senses of ‘efficacious’. We can define the two senses as follows:
Efficacy1: An entity is efficacious1 if and only if it can be a cause.
Efficacy2: An entity is efficacious2 if and only if it can noncausally contribute to causation.
With this distinction made explicit, it might be claimed, we can see that the argument trades on an ambiguity. For Efficacy is only true if it employs efficacy2; yet, the truth of Cause clearly requires efficacy1. There is, therefore, no valid reading of the argument from causal powers.
Since it is the case that Cause does indeed rely on efficacy1, the focus should be on whether it is true that Efficacy is only true if it employs efficacy2. That is, we need to know whether the causal efficacy of substances is limited to their being noncausal contributors to causal relations.Footnote 3 This is by no means obvious, so we require an argument for the claim. There are two main ways of arguing for this, each of which relies on a claim about what it is that gives substances their causal efficacy. They are:
EE: Substances can only be causally efficacious insofar as they participate in, or are parts of, an event.
PE: Substances can only be causally efficacious insofar as they possess certain properties.
A possible alternative one might consider is the view that although substances do have powers, it is the powers rather than the substances themselves that are causes. This response to the argument, however, can, for all intents and purposes, be treated as a form of PE. Although there are a variety of accounts on offer as to the nature of the properties with which powers are identified, it is standardly agreed that the powers of substances should be identified with the properties of their substances (see section 4.2. for further support, details, and one qualification).
Each of EE and PE might be thought to provide a reason for claiming that substances are only ever efficacious2 and for denying that substances can ever be causes. In broad outline the idea is that since substances are efficacious only in virtue of events or property instantiations, it is those events or property instances that are the causes. On some views of events, there are not really two arguments here but one. If events are ‘exemplifications by substances of properties at a time’ (Kim Reference Kim, Brand and Walton1976: 160), then causing is always done by a substance's properties, i.e., an event. If, on the other hand, events are coarse-grained particulars, changes in substances, or spatiotemporal regions, then the arguments are distinct. In what follows, unless otherwise stated, I shall assume the latter picture.
4.1 Substances, Events, and Causal Efficacy
The view that substances, although bearers of causal powers, are causally efficacious only insofar as they participate in some event, is widely held. The move from this claim to the claim that substances are never causes may be motivated in a number of ways.
4.1.1 Ellipticality
Menzies argues that although substances have causal powers (see 2009: 773), they fail to be causes because substance causal sentences are elliptical. He writes,
A sentence such as ‘The building caused the shadow’, in which the noun phrases in the cause and effect positions refer to physical objects, is grammatically well-formed. So, ordinary language clearly recognizes physical objects as causal relata. But this sentence is most naturally interpreted as elliptical for a more complex sentence such as ‘The building's obstructing the light caused the shadow to form’, in which the noun phrases denoting cause and effect are both quasi-sentential in form. Given that causal statements relating physical objects are properly seen as elliptical in this way, I shall not treat physical objects as genuine causal relata. (1989: 59–60; also see Rosen Reference Rosen and Zalta2009, §3.2)
This leaves the causal status of substances unclear, as somewhat mysteriously possessing causal powers without ever actually causing anything (Rosen allows that substances may ‘derivatively’ count as causes).
Menzies's stated reason for denying that substances can be causes is that for any substance causal sentence we can formulate an event causal sentence for which the former is elliptical. Putting aside the view that in the case of agency, no analysis of agent causal sentences in event causal terms is possible and thus granting that substance causal sentences are always elliptical for event causal sentences, it is nevertheless unclear what metaphysical conclusions should be drawn. In particular, it is not obvious why this would show that substances are not really causes.
One argument might be as follows: if every substance causal sentence s is elliptical for some event-causal sentence s*, then s and s* have the same truthmaker. In this case, s will be made true by the event cause cited by s*. Since this is so for all substance causal sentences, there is no need to posit substance causes to make true any substance causal sentences.
This inference is problematic, however, as substances usually participate in, or are parts of, events.Footnote 4 Take, for example, sentence s*, ‘Dripping poison in the king's ear caused his death’. The poison is a part of that event, and it seems clear that the poison is required for the truth of s*. It is the poison that had the power to bring about the king's death, given the right circumstances. It is needed for the effect since without the poison the king would not have died (granted the absence of backups). Given that the poison is among the truthmakers for s* (and the substance causal statement s, ‘the poison caused the king's death’), an argument is required for the claim that it is not a cause of the event.
The claim here, just to clarify, is not that because a substance is among the truthmakers of the event causal statement it should thereby count as a cause. This inference would be unsound. The point is rather that a move that proponents of EE might make—namely, that we've no need to posit substance causes since events are the truthmakers for these substance causal statements—is unjustified given that substances are themselves part of the truthmakers of event causal statements.
4.1.2 Sufficiency
The following reasoning might be thought to support the move from EE to the claim that substances can only be efficacious2. Compare situation A, in which the poison has been administered to the king, and situation B in which the poison is left in the bottle. Since the king only dies in situation A, not B, the cause is only present in A. Given that the poison is present in both, the poison cannot be the cause of the king's death, rather it is the administering of the poison that is the cause. The fact that substances have their powers in virtue of events shows that substances are not sufficient for their purported effects.
How convincing is this argument? Not very. Take, for example, two other situations: A* the striking of the match (when oxygen is present), and B* the striking of the match (when oxygen is not present). Since the effect of the match's lighting occurs only in A*, the cause is only present in this situation. As the striking of the match is present in both, it is not a cause. What is a cause is the striking of the match in the presence of oxygen (when the match is dry, when there is no strong wind present, etc.). The previous argument, like this one, wrongly (or, at least, extremely controversially) assumes that a cause of the effect must be sufficient for the effect to occur.
4.1.3 Overdetermination
Proponents of EE may try another tactic: that of strengthening their case against SC by appeal to the principle of no overdetermination. To say that both a substance and the event of which it is a part are causes of the same effect flouts the intuitive principle that there is no systematic causal overdetermination. Since both cannot be causes of the same effect, we must choose between them and, given that substance causal sentences are elliptical, events are the obvious front-runners.
Employing the principle of no overdetermination in this context, however, is dubious. Careful formulations of this principle state that overdetermination involves two (or more) sufficient and wholly distinct causes of the same effect. Being nonidentical is insufficient for being wholly distinct. In particular, entities are not wholly distinct from their parts (cf. Lewis Reference Lewis1986: 215 and Schaffer Reference Schaffer2003). That the car got me to the party on time does not mean that the well-functioning engine, which is partly distinct from the car, cannot also be cited as a cause. But it would be perverse to say that the effect is thereby overdetermined. Again, that Ivy's birthday party made her happy does not mean that the eating of the birthday cake, opening of the presents, etc. are thereby causally redundant. Nor do we regard the effect as overdetermined. It often makes good sense to single out parts of an event as among the causes because they are particularly important to the occurrence of the effect.
If some parts of a cause are themselves causes of the same event, then a number of different causes can stand in the causal relation to the same effect. This is consistent with maintaining the binarity of the causal relation, but it does require, to borrow Oliver and Smiley's (Reference Oliver and Smiley2004) terminology, that each place in the relation have a number of different positions. Thus, the predicate ‘causes’ is multigrade at each place, C1, (and possibly also C2, C3). . . cause E1. . . . I take it, however, that this is already a commitment of the standard view that an effect can have more than one cause.
Returning to our example, poison is not wholly distinct from the event of dripping poison into the king's ear. Just as the working engine is part of the car, the poison is part of the event. Given this, the principle of no overdetermination cannot be employed and the argument fails. True, we would not ordinarily assert that both the poison and the event of dripping poison caused the king's death, but the incongruity of such an assertion is explained by the fact that it conversationally implies something false, namely, that these are two independent causes.
It might be responded that removing the substances participating in an event changes the nature of the event. A birthday party without cake, for instance, has different features from one with cake. Since the events have changed, the pattern of causal relations also changes, and therefore there is no reason to think of the cake as a cause. It is not required for the effect.
This commits us to the controversial claim that event identity is dependent upon its participating objects. It seems prima facie plausible to say that Ivy's birthday party could have occurred without a cake, and in that case, the very same event could have occurred and not caused Ivy's happiness. (Recall, fine-grained theories of events, such as Kim's, are being put aside here; focusing on coarse-grained theories of events makes this claim more plausible.) But even if this is denied, the objection just ignores the distinctness condition on a successful application of the no overdetermination principle. Compare: ‘the car got me to the party on time’ and ‘the car's working engine got me to the party on time’. And suppose, to make the analogy closer, that the criterion of identity for cars is such that their engines are essential to them, so a car would not be the same car if it had no engine or a different one. Still, it seems correct to say that ‘the car's working engine got me to the party on time’ would have been true even though the car would have been a different car without its engine.
It might be objected that the analogy is not a good one. In the car example considered, we have an object that is a proper part of another object, whereas in the party example, we have two entities from different ontological categories—events and substances. Substances cannot be causes because they belong to a different ontological category than the events in which they participate.
This final claim, however, is more a declaration of faith than an argument. Although it is true that the analogy is different in the respect mentioned, we are offered no reason to think that this difference matters to the case in point. Consider an analogous claim that might be made in the debate about causal relata. It has been argued that Davidsonian events aren't fine-grained enough to count as causes (see section 5.3 below), instead we must appeal to their property instances or Kimean events. For example, it wasn't Sylvia's smiling that made her mother laugh, it was the smile's cheekiness. Whatever we may think of this argument, it seems inadequate to claim that property instances cannot be causes just because they are of a different ontological category from the Davidsonian events to which they are attributed. So similarly here, more needs to be said to defend the move from EE to the claim that substances are not causes.
4.1.4 Parsimony
An alternative way of motivating EE is to appeal to Occam's razor. Since substance causes are not required to make either a substance causal sentence or event causal sentence true, substance causes should not be posited. We can make do with the events of which such substances are a part.
Defenders of EE, however, need to show that simplicity favors a rejection of substance causes. That substance causal statements can be replaced with event causal statements does not show that substances are not required to bring about effects. If we remove the cake from the event of Ivy's birthday party, Ivy's party may no longer make her happy. Even if events are sufficient, substances cannot be removed from the universe without changing the causal facts. So it is incumbent upon proponents of EE to spell out the causal role of substances and further motivate this employment of Occam's razor.
It may be responded that the reason substances cannot be removed from the universe without changing the causal facts is that substances are causally relevant (make a noncausal contribution to causal relations). But this still incurs an obligation to give an account of causal relevance that explains why substances fail to count as causes, given that they seem to be required in the causal nexus. The bare claim that substances are causally relevant by being constituents of the cause, yet are not themselves causes, begs the question against the proponent of the argument from causal powers.
An alternative challenge questions the claim that substances cannot be removed from the universe without altering the causal facts.Footnote 5 Consider, for example, two possible worlds: at W1 a substance flashes brightly. At W2, there is a bright flash but no substance. The causal effects of these flashes are very limited, serving only to startle a nearby rabbit. In such a case, we might argue, the substance that flashes brightly could be removed from W1 and its causal nexus would remain the same—we would have just one event, the bright flash causing a startled rabbit. Thus the above response to considerations of parsimony is ineffective.
Granted a comprehensive understanding of the causal facts of W1, however, it seems doubtful that the substance could be removed from the universe without altering the causal facts. If we consider not only the effects but also the causes of the light's flashing, then presumably the causes of the flash will differ between the two universes. Changing the event in W1 to one lacking a substance will alter the mechanism by way of which the flash came to be. Moreover, although the actual effects of the flash are identical in W1 and W2, it is plausible to claim that the effects of these flashes could have been different. Suppose, for example, that in W1 a torch causes the flash. The resulting startled rabbit could have run into the torch, thereby hurting its paw. So if the causal nexus of the universe is construed broadly to include not only actual effects but also causes and potential effects (the latter being crucial if we want our causal nexus to include facts about preventions and omissions), then it is difficult to see how removing the substance from W1 will leave the entirety of its causal nexus intact.
Even if we grant that this could be done, however, perhaps maintaining that the events are simply uncaused, SC is a claim about our actual world and nearby possible worlds. It claims that in these kinds of worlds at least some substances are sometimes causes. In other words, if we removed all of the substances from our world (and from those similar to our world), then the causal facts of our world will alter. Consequently, even if we can construct a simple scenario in which this is false, a scenario where it is possible to remove the substances from that world while leaving the causal facts intact, this does not show that substances can be removed wholesale from our universe without changing the causal facts.
A different way of motivating an application of Occam's razor against the argument from causal powers is to argue that it is ontologically costly to view substances, in addition to the events of which they are part, as causes. But this assumes that positing substance causes enlarges one's ontology. This would be true if different causal relations were required to relate entities from different ontological categories, but there seems no reason to assume this. Since ordinary causal ascriptions acknowledge a variety of causal relata, that causation can relate entities of different categories is arguably the default assumption.Footnote 6 If this assumption is correct, then substance causation would come at no significant cost beyond granting the existence of substances. Furthermore, if the argument of section 2 is correct, we can endorse SC without taking a stand on whether substances are ultimately irreducible entities.
It might be objected that allowing both substances and events to cause the same effect substantially increases the number of causal relations being posited. Instead of just one relation between an event and an effect, we have one between the event and the effect and another between the substance and the effect. Following Lewis (Reference Lewis1973a: 87), however, it is widely maintained that what matters is qualitative, rather than quantitative, parsimony. But even if this were denied, the objection assumes that multigrade predicates such as ‘cause’ or ‘fought with’, require separate instantiations of the relation in question. That is, ‘a, b, c and d cause e’ is made true by the fact that a causes e, b causes e, etc., where these are all distinct instances of the causal relation.
We may, however, deny this assumption. When we say that ‘Sylvia and Karina carried the pram’ the predicate ‘carried’ is used collectively. It is false to say that ‘Sylvia carried the pram and Karina carried the pram’ since this fails to preserve the implication that they carried the pram together. Similarly, in the cases under consideration, the predicate ‘cause’ can be read collectively; the poison and the dripping together cause the king's death. There is, then, prima facie reason to suppose that a single relation relates all the entities in question. If so, substance causation does not even violate quantitative parsimony.
But still, it may be insisted, every true causal claim supervenes on the totality of causal facts of the form, ‘event c causes event e.’ When an object O participates in an event, this can make true a causal claim of the form ‘O causes event e’, but since this is a derivative, not a fundamental, truth, even if it is true, substances are not really causes.
The point of section 4.1 has not been to prove the widespread consensus false, but rather to show that it lacks adequate support. Compare: ‘Cars help us get to places.’ Since engines are parts of cars, this makes it true that ‘engines help us get to places’. But as this is a derivative and not a fundamental truth, ‘engines don't really help us get to places.’ There is something problematic with both pieces of reasoning. My diagnosis is that, in both cases, they misuse the principle of overdetermination. Those who wish to show that the argument from causal powers trades on an ambiguity owe us something better. Plugging the gap with supervenience won't suffice since it is well established that, in lieu of further argument, supervenience claims fail to establish that one set of facts is ontologically prior to or dependent upon another. To borrow McLauglin's example (Reference McLauglin, Savello and Yalcin1995), being F supervenes upon not being F, since there can be no change in an entity's F properties without a change in its not-F properties, but this doesn't show that something's being F is ontologically dependent on its being not-F or that an entity's being F is in some way derivative upon or reducible to its being not-F. Therefore, even if we grant, for the sake of argument, that substance causal facts supervene on event causal facts, this still doesn't show that substance causal facts are ontologically dependent or somehow ‘derivative’ upon event causal facts. The challenge for those who wish to argue that the argument from causal powers is invalid is to show why substances can't be causes in their own right.
4.2 Substances, Property Instances, and Causal Efficacy
The proponent of PE claims that substances can only be causally efficacious insofar as they possess certain properties and, therefore, are only ever efficacious2. This is for the reason that it is the property instances themselves that enter into causal relations. Unlike PP, discussed in section 3.1, this view accepts that substances are the bearers of causal powers and so are causally efficacious. But it maintains that when something happens, it is always a substance's properties that do the causal work.
This is probably the dominant view, and we see one significant motivation for it in accounts of dispositions or causal powers. (This is not to say that theories of powers are per se antagonistic to substance causation, as I argue below.) Although there are a variety of accounts on offer as to the nature of the properties with which powers are identified, it is standardly agreed that they will be properties of their substances.Footnote 7 At the very least, such accounts claim that power or dispositional statements are true, in part, in virtue of the properties possessed by the substance instantiating the disposition.
For example, Fara writes, ‘N is disposed to M when C’ is true iff N has an intrinsic property in virtue of which it Ms when C.’ (2005: 70). Armstrong argues that ‘the truthmakers for. . . counterfactuals should be sought in purely categorical properties of the [bearer]’ (1996: 16). Similarly, Prior, Pargetter, and Jackson (Reference Prior, Pargetter and Jackson1982) and, following them, Lewis (Reference Lewis1997: 149) all talk about the disposition's ‘causal basis’, characterizing this in terms of the property or property complex that is ‘responsible for its being such that “if. . ., then. . .’’’ (Prior, Pargetter, and Jackson Reference Prior, Pargetter and Jackson1982: 252).
Such accounts posit a noncausal relation of dependence between a substance's properties and its powers. Perhaps in combination with other entities, the former make dispositional ascriptions true. One model for this is suggested by Prior, Pargetter, and Jackson:
By a ‘causal basis’ we mean the property or property-complex of the object that, together with the first member of the pair—the antecedent circumstances—is the causally operative sufficient condition for the manifestation in the case of ‘surefire’ dispositions, and in the case of probabilistic dispositions is causally sufficient for the relevant chance of the manifestation. (1982: 251)
Properties of the substance make the dispositional ascription true because when the substance is in the right circumstances, a property or property complex causes the event to happen. On this view, which Lewis also endorses, ‘The breaking presumably was caused; and caused jointly by the striking and by some property B of the glass’ (Lewis Reference Lewis1997: 149). Because the substances’ causal powers are had in virtue of their properties, the substance's role as cause is usurped. Substances are powerful only insofar as they instantiate causally powerful properties standing in causal relations. So substances never actually cause anything. They are efficacious2 but not efficacious1.
The claim that substances have causal powers in virtue of their properties is supported by the fact that qualitatively similar substances have similar (intrinsic) causal powers. Duplicate substances behave in similar ways in similar circumstances, and therefore it is reasonable to postulate a relation of dependence between their causal powers and their properties.Footnote 8
It does not follow, however, that substances are only ever efficacious2. For PE is consistent with SC. That is, the proponent of the argument from causal powers may consistently claim both that substances can be causally efficacious only insofar as they possess certain properties and also that some actual substances are causes.
According to this combination of views, although the substance's causal powers are rightly thought of as standing in a relation of dependence to its properties, this is not understood in terms of those properties causing particular event manifestations. The function of properties is not to stand in causal relations, but rather to ‘enable’ their bearers to be causes. They are what make substances powerful, but it is the substance that does the causing. Properties, by contrast, are causally relevant by being those entities in virtue of which substances have the powers they do.
By analogy, a drink is tasty in virtue of its ingredients: its gin, tonic, lemon, and ice. There is no reason to think that any of these ingredients are tasty independently of the drink. The drink's tastiness noncausally depends on its ingredients. Similarly, substances are powerful entities that enter into causal relations. They exert powers in virtue of their properties. Their properties are what make substances powerful, causally efficacious entities that cause things. But properties are not themselves powerful entities that cause things. Properties are the kingmakers, substance is king.Footnote 9
Although the move from the claim that objects have causal powers in virtue of their properties to the claim that those properties do the causing, is often made without further argument, it is not legitimate. Many philosophers, such as Armstrong and Lewis, also accept that substances have causal powers partly in virtue of the laws of nature. But these laws are not thereby regarded as causes of the manifestations of these powers—they have a different role to play. Similarly, that a substance is powerful in virtue of its properties does not show that substances do not cause events when their powers are manifested.
None of this shows the falsity of the claims that substances are merely efficacious2 and that it is properties only that enter into causal relations. That is not the purpose of this section; rather, this section is to show that we have not been given sufficient reason to accept the claim that substances are merely efficacious2 as true. That is, we have not yet seen a compelling argument from PE to the claim that substances are never efficacious1.
More than this, the combination of SC with PE certainly merits consideration, as it offers an intuitive view of the relationship between substances and their powers. Consider, for example, a magnet's power to attract other substances. The magnet has a small, continuous power or influence upon its surroundings, even though it may not be obviously manifesting its power by moving ferromagnetic materials. If we grant that it is the magnet that has this power, it is counterintuitive to claim that when a paper clip is placed nearby and is caused to move, it is the properties of the magnet that are doing the causing. If substances have powers, there is no obvious reason to think that when a power is manifested, its properties are doing the causal work. There is a recalcitrant intuition that ‘in our world the pushing and shoving and forcing are done by things—elementary particles and aggregates of them’ (Bennett Reference Bennett1988: 22). The combination of SC and PE gives us a way to preserve that intuition.
5. Objections to Substance Causation
It is often thought that substance causation is deeply problematic. If the notion of substance causation is incoherent, then even if the equivocation objection discussed in section 4 is not compelling, there must be something wrong with the argument from causal powers. In this section I defend SC against objections, thereby rejecting this line of thought.
5.1 Substances and Time
An influential objection to the very idea of substance causation is put forward by Broad ([1934]1952). He writes,
Now it is surely quite evident that, if the beginning of a certain process at a certain time is determined at all, its total cause must contain as an essential factor another event or process which enters into the moment from which the determined event or process issues. I see no prima facie objection to there being events that are not completely determined. But, in so far as an event is determined, an essential factor in its total cause must be other events. How could an event possibly be determined to happen at a certain date if its total cause contained no factor to which the notion of date has any application? And how can the notion of date have any application to anything that is not an event? (1952: 215)
This objection, however, is ineffective against SC. SC only claims that there are some substances, in this world, that are causes. It does not commit us to the claim that substance causation can occur independently of event causation (cf. section 2). It may be that every instance of substance causation is accompanied by some instance of event causation triggering its power. If this is right, then Broad's claim that ‘its total cause must contain as an essential factor another event or process which enters into the moment from which the determined event or process issues’ (1952: 215) would be true. We would have a reason why it is that this substance is a cause at this particular moment of its existence, rather than at some other moment of its existence or for the entirety of its existence. But that there is event causation here fails to show that events are the only causes and that substances’ role as cause is thereby excluded.
Suppose, however, that we did want to maintain that a substance could be a cause without any accompanying event. Surely then Broad's objection would bite? In a modified version of Broad's objection, Clarke writes,
Effects are caused to occur at times, and it might be argued that this can be so only if their causes likewise occur at times—only, that is, if their causes are directly in time in the way in which events are but substances are not. (2003: 201)
But although substances and events are differently related to time, they are both ‘in’ time, as both exist during periods of time. In either case, we can pick out a period in the history of the event or substance and call it the cause at that time. So we can say that the game of musical bumps broke my teapot (in virtue of the game's properties at t), and the cake caused the children to squeal with delight (in virtue of the cake's properties at t). In light of this, we might suggest that a similar treatment can be given of both (coarse-grained) events and substances: both are causes in virtue of their properties. Although I think that this proposal has something to recommend it (see section 5.3), the further claim about events is clearly not required for the defense of SC.
Clarke rejects the proposal that a substance may cause an effect in virtue of having a property at a certain time, commenting, ‘The accommodation, however, comes perilously close to acknowledging that it is the substance's having the property at a time in question that is the cause.’ (2003: 202) But while the views sound similar, they are not equivalent. To say that an effect occurred ‘in virtue of’ the laws does not thereby make the laws causes of the effect. Similarly, saying that an effect occurred ‘in virtue of’ a substance having a property at a time does not render the having of that property the cause. If we endorse both SC and PE, we may insist that the ‘in virtue of’ relation holding between substances and their powers need not be understood in terms of what the substance's properties cause. Rather, properties are responsible for the fact that substances cause things.
Perhaps an opponent will argue that this combination of views really just amounts to a version of causation via property instances (or event causation with events construed finely, e.g., Kimean events or tropes). Substances count as merely ‘derivative’ causes on this view since their causal status derives from properties.
If what is meant by ‘derivative cause' is just that the entity is a cause in virtue of further entities, then substances are derivative causes on this picture. But this rendering of ‘derivative’ is consistent with SC. Compare the claim that an event/property instance is a cause of e (partly) in virtue of the laws. Adherents of the view that events/property instances are causes will not accept that this robs them of their status as genuine causes. Similarly, that a drink is tasty in virtue of its ingredients does not render it any less tasty. Whenever any entity is labeled as a ‘derivative cause’, the onus is on the labeler to say what is meant by ‘derivative’ and why the entities in question are not simply causes. Until this is done for substances, SC remains unscathed.
5.2 Substances and Probabilities
Clarke raises another objection to the possibility of substance causation, taking it to be significant, if not decisive. He writes,
it is an idea of some plausibility that causes have something to do with objective probabilities. And it might be put forward that any cause must be of a category of individuals which can affect the probabilities of their effects before those effects occur. . . . This claim would seem to rule out substances as causes. If a substance directly causes an event, it brings the probability of that effect to 1, but only when the effect occurs. (2003: 203)
But, intuitively, substances can affect the probabilities of their effects. For example, the snake raises the probability of my panicking. Again, one can object that it is the presence of the snake, a state of affairs, not any substance, that raises the probability of the effect. But that it is a snake seems crucial to the probability of the outcome. Had it been a puppy instead, this would have significantly lowered the probability of my panicking. So the snake does seem to have ‘something to do with objective probabilities’ (Clarke Reference Clarke2003: 203). The combination of SC and PE indicates how substances can affect probabilities. The substance, at t1, raises the probability of e at t2 since the substance, in virtue of its properties at t1, is poised and able to cause e at t2 (given certain background conditions). For instance, the snake, in virtue of its properties at t1, raises the probability of my panicking at t2 since it can cause panic in certain animals, of which I am one.
Clarke counters this kind of response, arguing that if a substance acquires a property P, where P has a tendency to produce events of a certain type, then it ‘seems that s’s acquisition of P at t—the occurrence of the event—raises the probability of the later effect, and there is nothing left for s (in virtue of acquiring P) to do in this regard’ (2003: 204). But in an event such as dripping poison in the king's ear, the poison does seem to raise the probability of the king's death, and so there is something for the substance to do. In section 4.1 above, I offered reasons to think that it is legitimate to single out objects from the events in which they participate and to consider what causal contribution (which could be construed in terms of probability-raising) they make. Thus, Clarke's response fails.
5.3 The Causal Difference Argument
Substances, like Davidsonian events, are multipropertied entities that can be redescribed in many different ways. This makes them vulnerable to the causal difference objection since it seems that there can be causal differences without any corresponding difference in the causal relata. What caused the scale's pointer to move to the 150 grams mark was not the pear, one might argue, but rather the pear's weight. For if the very same pear had been a different weight—if I had taken a bite out of it earlier, for example—it would not have caused the scale's pointer to move to 150 grams. Substances, it seems, are not fine-grained enough to be causes.
Defenders of substance causation can employ Davidson's response here (see Davidson Reference Davidson1967: 697–98; other defenses are possible, for instance, Schaffer Reference Schaffer2005, but I suspect that Davidson's is on the right lines). Although ‘the weight of the pear’ provides a better causal explanation of the pointer's moving, this is not because something other than the pear caused the pointer to move since the fact that the pear's weight is not mentioned does not remove it from the pear in question. If this response is going to be persuasive, however, we need some explanation of why just citing the pear makes a bad causal explanation.
Defenders of fine-grained relata can answer that it is the weight of the pear that is the cause. The substance causal statement makes a bad explanation because it only gestures at the true, fine-grained cause.Footnote 10 This view respects the intuition that the weight of the pear makes for an objectively better causal explanation of the pointer's moving than, say, its greenness. The statement is a good explanation because it correctly tracks objective dependency relations. But the possibility of combining SC with PE illustrates that we needn't conclude that these are causal dependency relations. We can preserve the objectivity of causal explanations by saying that substances (and perhaps also coarse-grained events) are causes in virtue of their properties, where this ‘in virtue of’ relation picks out an objective relation of noncausal dependence between the properties and the causes in question. Good causal explanations are those that illuminate which properties the substances are causes in virtue of, bad causal explanations do not. If the pear was only a cause on some occasion because it had a certain weight, then a good causal explanation should cite that fact. This doesn't undermine the claim that it is substances (or coarse-grained events) that are the causes, but it does make clear that, on this occasion, they could only function as causes because they instantiated certain properties. This view, then, offers a proponent of coarse-grained causal relata a unified analysis of what makes for a good causal explanation.Footnote 11
6. Conclusion
In the dispute over which entities can be related via causation, substance causation is often ruled out prematurely. In the argument from causal powers we have an intuitive line of reasoning that purports to show that substances are causes. As I have shown, the premises of this argument are not as easy to reject as it might initially seem. Furthermore, proposed objections to substance causation do not suffice to undermine it. The burden of proof, then, is on those wishing to deny that substance causation is possible.
