3.1. Three objections and three tweaks

Mental causation and connectivity. The basic idea of this paper is to identify the physical with a certain causal network. I have suggested that we think of causal networks as involving some sort of connectivity, but unless we are careful about stating the required connectivity, it might seem that a causal account is doomed. The problem is that many properties do seem to be causally relevant to the physical in some sense, without obviously being physical. But then if this causal relevance is sufficient to satisfy the connectivity condition imposed by a causal account, these properties would be deemed physical, which we cannot have.

Montero helpfully makes the point as follows: “Beliefs and desires seem to be part of the same macro-level causal network as rocks and trees while quarks and leptons do not. But neither the physicalist nor his foes think that the view that quarks and leptons are nonphysical is what we should infer from our central examples of physical objects” (Montero, Reference Montero1999, 184).Footnote ⁶ This illustrates two aspects of the challnege. First, things like beliefs seem to belong to causal networks including the physical but they are not, despite that, physical in quite the same way. The physicalist wants to maintain that the mental is nothing over and above the physical, but they may very well want to distinguish this from being properly physical the way that rocks, trees, and their properties are. And so we want a definition which allows this distinction. Second, things like leptons and quarks do not make any appearances in our everyday causal talk, yet they and their properties are surely physical. This might not seem like much of an issue, but it does restrict the ways we can address the first aspect of the challenge. For instance, we cannot simply dismiss mental causation as being supervenient causation, in case supervenient causation is what bridges the gap between everyday entities like rocks and the subatomic particles that compose them. What we need to do then is to selectively carve up the part of the causal network which we think counts as being straightforwardly physical—call this the properly physical—and differentiate it from the part which, although not being “over and above” the physical, is not properly physical—call this the parasitically physical—all while making sure that the scope of the properly physical is not inappropriately restricted. In short, we need to specify connectivity conditions that separate the properly physical from the parasitically physical while getting both extensions right. Call this the carving problem.

In order to address the carving problem, I’ll suggest we turn to an independently important problem for mental causation: the causal exclusion problem. Responses to the exclusion problem roughly partition defenders of mental causation into three camps and I will argue that each position gives us the resources to solve the carving problem.Footnote ⁷ First raised by Malcolm (Reference Malcolm1968) and since brought to prominence by Peacocke (Reference Peacocke1979, 134–9) and Kim (Reference Kim1989, Reference Kim1993), the exclusion problem for mental causation is the apparent inconsistency of four commonly held views:

The principle of causal completeness (CC) holds that the physical is a causally closed system in that one need never look beyond the physical for the cause of a physical effect. The principle of non-overdetermination (O) states, roughly, that physical effects do not have multiple, distinct, sufficient causes. The third claim (NR) encapsulates the spirit behind non-reductivist accounts of the mental. Finally, the fourth claim (ME) simply says that mental things such as our beliefs and desires can be causally relevant to physical effects. These four claims seem to be mutually inconsistent because if every physical event has a sufficient physical cause, and that physical cause is not identifiable with any mental cause, then the existence of sufficient mental causes would result in overdetermination.

To resolve the exclusion problem, one of the these four tenets must go or some way of reconciling them must be found. There are various ways that one can respond to the exclusion problem, though not every logical possibility is in practice viable. In particular, rejecting either of (CC) or (O) is effectively off the table as physicalists more or less have to stand by the causal completeness of the physical (Horgan, Reference Horgan1993, 573) and they seem similarly set on sticking with the denial of overdetermination. This means that there are three approaches to avoiding the exclusion problem which are relevant to us: denying the distinctness of the mental and the physical (NR), denying the causal efficacy of the mental (ME), and maintaining all four claims while arguing that they are not in conflict. In any case, all three options offers us a solution to the carving problem.

The first approach we’ll look at is to reject (NR) and deny that the mental is distinct from the physical. This strategy avoids concerns about overdetermination by saying that mental causes are in fact physical causes and so there are not, strictly speaking, multiple distinct sufficient causes. More precisely, the mental is held to reduce to the physical in some way such that there is an identity between the mental and physical causes of an effect and this identity means that there is really only one sufficient cause—so overdetermination is not a concern. In effect, this approach solves the exclusion problem by advocating for a reductive physicalism. Conveniently, doing so dissolves our carving problem.

The carving problem was motivated by the intuition that, although they might not be over and above the physical, mental entities such as beliefs are not physical in quite the same way as things like rocks are. This prompted us to search for a way of distinguishing the mental from the properly physical. But the reductive physicalist is going to maintain not just that the mental is nothing over and above the physical, but that the mental is in fact somehow identical with the physical. Thus they are not going to be sympathetic to our concerns about distinguishing the properly physical from the parasitically physical,Footnote ⁸ as doing so would conflict with their reductivist thesis. In sum, the reductivist simply denies Montero’s problem about differentiating beliefs from rocks and in doing so they relieve us of having to solve the problem.

A second approach to resolving the exclusion problem is to reject (ME). On this way of doing things, we allow that the mental and physical are distinct but overdetermination is avoided by denying that physical effects ever have mental causes. This approach allows that mental effects can have physical causes, but insists that the causal influence does not run the other way. So although it might seem like the feeling of hunger causes me to grab a hamburger, this approach has us maintain that it does not. We might concede that my feeling and my action are closely connected in that the physical cause of my feeling is also what causes me to act, but in denying (ME) one maintains that the feeling of hunger is not itself the cause. We thus escape the exclusion problem by advocating for a form of epiphenomenalism.

This approach gives us resources to solve the carving problem as it introduces an asymmetry in the causal relations between the mental and the physical. Once again, the current challenge is to carve up the causal nexus so that the non-physical, the properly physical, and the parasitically physical all come out as separate. Doing so within the framework of a causal definition of the physical means differentiating those categories in terms of their causal relations to each other. By denying the causal relevance of mental properties to physical properties, we thereby have a way of distinguishing the properly physical from the parasitically physical in terms of their causal roles: the properly physical are causally relevant to other properly physical properties and vice versa, whereas parasitically physical properties are not causally relevant to properly physical properties, even if properly physical properties are causally relevant to parasitically physical ones. Thus the epiphenomenalist response to the exclusion problem is compatible with a causal definition of the physical because we can carve off the properly physical by requiring the symmetry just described.Footnote ⁹

The third and final approach to the exclusion problem is to articulate some form of “compatibilism,” according to which the mental and the physical are distinct from each other and can both be causally relevant without causing any problematic form of overdetermination. The general approach here is to try and salvage our intuitions about the causal relevance of the mental by distinguishing different types of causal relevance (some of which may be more epistemological than metaphysical) which make our talk of mental causation legitimate. Horgan describes the general approach as follows:

Compatibilist responses to the exclusion problem thus work by distinguishing the type of causal relevance that the mental bears to the physical from the variety of causal relevance that the physical bears to the physical. They hold that mental causation does not lead to overdetermination because it is of a different—though in some sense still legitimate—form than physical causation.Footnote ¹⁰

Just as the epiphenomenalist response provided us the resources needed to differentiate the properly physical from the parasitical, so does the compatibilist. By distinguishing between mental and physical causation, the compatibilist simultaneously provides what we need in order to distinguish the properly physical from the parasitically physical. By adopting the symmetry clause we introduced to reconcile our definition with epiphenomenalism and then adding a clause which excludesFootnote ¹¹ the variety of causal relevance that compatibilists attribute to the mental, we can distinguish the properly physical from the parasitically physical. That is, although the mental is causally influenced by the physical, its only causal relevance to the physical is the special form of relevance introduced by compatibilists, and so by stipulating the exclusion of this variety of causal relevance, mental entities fail to exhibit symmetric causal relevance and hence are distinguished from the properly physical. Thus, one can maintain the causal relevance of the mental as the compatibilist does, and still be able to distinguish the properly physical from the parasitically physical within the context of a causal definition of the physical.

This concludes my response to the carving problem, though it will be worthwhile to review. Intuitively, the problem was that there seemed to be properties which are causally connected to the physical but which do not seem to be physical in quite the same way as, say, massiveness. More precisely, the issue was that if we identified the physical with a connected graph of properties, then properly physical properties would get lumped in with all sorts of other properties. So, we needed to find a way to identify the proper physical in causal terms, even though they may be causally connected to other properties in some manner. In other words, we needed to specify further structural features of the physical in order to pick out the correct subgraph. This was done by refining the connectivity required of a causal network. The manner in which this gets tweaked depends on one’s response to the causal exclusion problem, but the most demanding tweak requires that a physical property not only stand in some relation of causal relevance to another physical property, but that it both be relevant to a physical property and have some physical property relevant to it, where this does not include compatibilist-relevance. In other words, we require that the physical not only be connected, but that it be strongly connected: so that physical properties are all connected via paths in both directions. This ruled out properties that are only connected in one direction, such as mental properties would seem to be.

Duplicate networks and non-structural specifications. The basic idea of this paper is to identify the physical with a certain causal network and in the previous section, we saw the importance of specifying structural conditions in order to pick out just the right network. However, you might think that structural constraints are insufficient to uniquely identify the physical. For instance, you might think that whatever structural features the physical exhibits, there is some other causal network that has the exact same structure. If so, then a characterization that only gives structural constraints will treat both networks as being physical and will fail to differentiate the physical from the non-physical.

Be this as it may, so long as there are some properties that we feel safe in identifying as examples of the physical, the objection can be overcome. In particular, we might say that the physical are the elements of the causal network that not only have such and such a structure but also include certain exemplary physical properties. For instance, if we want to take massiveness as an exemplary physical property, then we might say that acceleration is a physical property because acceleration belongs to the same causal network as massiveness, while an analogous property in a structurally identical network will not be deemed physical since the analogous network does not include massiveness.

It might be objected that by including reference to an exemplary physical property, a causal account becomes circular, or perhaps relatedly, that it fails to explain why those exemplary properties are physical in the first place. The first charge is easily defused by pointing out that the account I’m advancing is schematic and that a fully fleshed-out causal account would specify what the exemplary properties are. Then, once the expression “exemplary physical property” has been replaced, all occurrences of “physical” will be excised from the definiens. The second charge is a bit more sticky, but the key is to note that our goal is not to say why physical things are physical. Rather, we want to facilitate a statement of physicalism that accords with how physicalists understand the term. Since it seems natural to think that physicalists conceive of the physical in part via relation to familiar phenomena, it should not be problematic for our definition to incorporate exemplary properties as something of a localizing mechanism.

Categorical properties, other worlds, and rigidifying the description. While including an appeal to exemplary physical properties helped to distinguish the physical network from certain related networks, one might be concerned that this fails to fully fix the reference of our account unless we commit to certain views about properties. That is, it might be objected that given a categorical or neo-Humean view of properties (cf. Armstrong, Reference Armstrong1983; Lewis, Reference Lewis1986), the appeal to exemplary properties will only ensure the correct classification of properties in this world and will do little to ensure that the extension of the physical is correct in other worlds. Indeed, if we adopt a view according to which properties are categorical and do not possess any powers necessarily—that is, all causal relevance is contingent—then it might seem like a causal account of the physical issues unpalatable predictions about other possible worlds. For instance, if there is a world where the property of, say, massiveness has no causal relevance to any other properties, then it would seem like a causal account would classify massiveness as non-physical in that world.

Thankfully, this problem can be solved by using off-the-shelf resources for rigidifying a description. For instance, if one is concerned about the prospect of physicality varying from world to world, one could modify the characterization to incorporate a rigidifying term (like Kaplan’s (Reference Kaplan, Almog, Perry and Wettstein1989) “dthat”), thereby transforming the flaccidly designating description into a rigid designator that picks out the same things in all worlds. Less technically, we could restate the causal account so as to say that a property is physical iff, in the actual world, it is in the causal network with such and such a structure and including such and such exemplary properties.

However one hopes to carry it out, the idea here is to treat the physical akin to how Putnam (Reference Putnam1975, 146–8) treated water. So while Putnam held that water was picked out by implicit ostension to that stuff in rivers and oceans, it is not the case that water is whatever is rivers and oceans. Rather, water is (according to Putnam) what fits this description in the actual world—this is why he maintains that, even on Twin Earth, water is $ {\mathrm{H}}_2\mathrm{O} $ and not XYZ. Similarly, we can say that physical properties are not just those properties which are causally related to exemplary properties in the right way, but rather that physical properties are those properties which are so related in the actual world. And thus, even a neo-Humean can use a causal account to say that massiveness is a physical property in all worlds.

3.2. The considered causal account

At the beginning of this section I suggested that we think of the physical as the elements of a certain causal network, but I said little about which network exactly. We have just seen three ways a causal account might get the extension of the physical wrong and, in doing so, we identified ways to modify a causal account that would prevent this. In a moment, I will put these together and state the considered causal account. But before I do so, it’s important to clarify that what I am stating is best understood not as a definition but a definition-scheme: an abstract structure with place-holders that can be filled in different ways. I do not pretend to have worked out every detail for a definition of the physical and it could well be elaborated in different ways, so I put forward my “definition” as a mould for other causal definitions to take.

With that being said, if we take the basic idea advanced at the beginning of Section 3 and construe the physical as the elements in a causal network, but (1) add that they are the elements in a network that is strongly connected by non-compatibilist causal relevance, (2) incorporate an appeal to exemplary physical properties, and (3) rigidify our description, then we get the following: a property is physical iff, in the actual world, it belongs to the causal network that is strongly connected by non-compatibilist causal relevance and that includes the exemplary physical properties. Effectively, this definition works by taking some sure instances of the physical and then extending itself to include all properties that are bi-directionally connected to these exemplary entities via a chain of properties, each of which is connected to its neighbour via non-compatibilist causal relevance. So physical properties are those properties that are of mediate causal relevance to exemplary physical properties and vice versa.

For example, we might say that, taking massiveness as an exemplary property, acceleration and magnetism both come out as physical because massiveness is of mediate causal relevance to both properties and both properties are of mediate causal relevance to massiveness. On the other hand, if we have a property that has no causal relevance (like vorpality), then it is not physical, and even if we have a property that seems to exhibit some causal relevance, such as feeling hungry, it will not come out as properly physical if that causal relevance is asymmetric or only of the compatibilist’s variety.

Availing myself of graphical representation to clarify the matter, our account pictures the physical as the elements of a directed graph where the vertices are properties, and the edge relation is causal relevance. It says that if we take a graph whose vertices are all properties and where the edge relation is given by their actual relations of causal relevance, the properly physical are those properties in the strong component containing exemplary physical properties. By a strong component, I mean the largest subgraph $ G=\left\langle V,E\right\rangle $ such that for any vertices $ a,b\in V $ , there is a path from $ a $ to $ b $ and a path from $ b $ to $ a $ . The requirement of bi-directional connection reflects our discussion of mental causation and requiring the physical to be the largest such subgraph simply ensures that our account is well-defined. Having the physical include exemplary properties as a subgraph is how we differentiated the physical from other structurally similar graphs (i.e., strong components), and having the edge relation be given by the actual relations of causal relevance reflects our decision to rigidify the description.

Given all this, physicalism may be stated as the thesis that this component and the vertices which follow from it constitute the whole graph. For example, consider Figure 1. If the vertices in Figure 1 represented all instantiated properties and the edge relation was actual causal relevance, then taking $ b $ as an exemplary physical property, we’d have that $ b,c, $ and $ f $ were properly physical, that $ d $ was parasitically physical, and that the existence of $ a $ or $ e $ would render physicalism false. Though this remains schematic, I hope it demonstrates how a causal account can facilitate a careful statement of physicalism and that it can do so while overcoming the triviality problems that plagued physics-based accounts.

Figure 1. A causal network as a directed graph.

4.1. Misclassification of causally integrated entities

One concern for a causal account of the physical is that it seems at least conceivable that there are some non-physical entities that exhibit symmetric causal relevance to the physical. So, for instance, it might be conceivable that ghosts possess some property that exhibits symmetric causal relevance to the physical despite not itself being physical. Perhaps less radically, it seems conceivable that mental properties exhibit bi-directional causal relevance to the physical—indeed, this would seem to be what interactionist dualists maintain. But it’s absurd for non-physical properties to be physical, and so if a causal account says that such a thing is conceivable, a causal account must be wrong.

Though there is indeed a sense in which seemingly non-physical properties could be conceived as causally integrated, the argument as it is presented above comes too fast and threatens to elide two separate arguments. In order to address this objection, we need to disentangle the related arguments, which we will do by keeping in mind the difference between talk of the non-physical and talk of the mental. As we’ll see, the plausibility and upshot of the argument depend greatly on whether we are talking about the possibility of causally integrated mental entities or causally integrated non-physical ones. I will argue that neither disambiguation is both plausible and problematic.

So suppose we run the argument in terms of mental properties. It’s hard to deny the conceivability of such properties being causally integrated, but it’s not so obvious that this is problematic for a causal account of the physical. While mental properties are often thought of as being non-physical, it’s not clear that they are so by definition (Dowell, Reference Dowell2006, 45; Stoljar, Reference Stoljar, Zalta and Nodelman2024, Sections 4.4–4.5). This is a crucial difference between talk of non-physical properties and mental ones: non-physical properties are definitionally not physical, while mental properties are not obviously in definitional conflict with the physical. In fact, it seems entirely plausible that the mental and physical are independently defined concepts—indeed, this would presumably be the case if both were defined via a positive comparison to paradigmatic examples. But then this would leave it as an open possibility that some properties are both physical and mental, and consequently, such overlap would not be the absurdity the current objection takes it to be. So if the objection is framed in terms of causally integrated mental properties, it must at least be supplemented with arguments to the effect that the mental and physical are mutually exclusive.

Wilson (Reference Wilson2006, Section 2.3) attempts to provide such arguments by appeal to historical considerations. Wilson argues that since physicalism is the successor to materialism and since materialism was opposed to dualism, definitional separation of the mental and physical is desirable in order to “preserve materialism’s traditional incompatibility with its dualist rivals” (Wilson, Reference Wilson2006, 85).Footnote ¹² At a high level, Wilson thus argues that we must keep the mental and physical separate in order to explain physicalism’s position in philosophical debate. But while there may be something to be said for understanding physicalism in terms of its dialectic position, we should not confuse dialectic with perceptions of a historical debate and we should question both the historical picture Wilson presupposes as well as the implications she draws from it.

To start, it’s important to note that by appealing to a historical opposition between materialists and dualists, Wilson takes for granted not just that there are two ways of describing historical thinkers but that there was a well-defined, bipartite division of philosophers and, moreover, that this division was effected by some substantial disagreement. And yet we could quite reasonably question both of these presuppositions. Thus, one might question the very idea that historical thinkers can be so neatly divided into two groups. Instead, you might well think of materialism and dualism as we now tend to think of rationalism and empiricism: as rough labels for describing certain tendencies that thinkers might exhibit to varying degrees (Markie & Folescu, Reference Markie, Folescu, Zalta and Nodelman2023). But in this case, it’s not clear that we should think of materialism as necessarily precluding dualism, let alone that physicalism should. Now, you might grant that there is a well-defined class of materialists and a well-defined class of dualists and that they all opposed each other, but you could still think that this opposition was accidental to their views: you might think that the perceived disagreement between these two camps was either imagined, tangential to the views they represented, or simply mistaken. But in this case, it would seem the opposition to dualism was inessential to materialism and so being a successor to materialism would not require preserving the opposition. In general, Wilson’s appeal to the history of philosophy requires that the history of philosophy be understood in terms of well-defined groups that inherently oppose each other, and so if you think this is an oversimplification in any way, then Wilson’s historical argument should not motivate a definitional separation of the mental and the physical.

But even if you allow that there was an inherent, well-defined disagreement between historical materialists and historical dualists, you should question why the views descended from these ought to inherit the opposition. While materialism and dualism may have had a substantive disagreement, you might think that as these two views evolved, their successors developed in ways that erased the disagreement. Seeing as how the considerations that motivate modern physicalists are not exactly the same as the considerations that motivated materialists and how these views are articulated in terms of different concepts, you can imagine how time might have redrawn the boundaries of the dispute. Indeed, it seems the only way to preserve all the oppositions that materialism was committed to would be to advocate for materialism. Conversely, it would seem that whenever a view changes, there must be some change in what it opposes. And so it should not be a fatal blow if a causal account of the physical renders physicalism consistent with interactionist dualism: while there may or may not have been an inherent incompatibility between materialism and interactionist dualism, modern physicalists should not be beholden to preserve this grudge. Rather, a causal account of the physical might make us ask whether continuing this disagreement would be anything more than a verbal dispute.

So the conceivability of causally integrated mental properties need not spell the end of a causal account, but surely the conceivability of causally integrated non-physical properties would. Be this as it may, it’s not clear that such properties are conceivable, at least not as such. Not only is it hard to come up with a sure-fire example of a non-physical property that one could conceive of as causally interacting with physical things, but even if one does have such an example in mind, it’s questionable that conceiving of its causal interaction is consistent. Indeed, it might seem like in doing so, one is conceiving of a certain property as if it were physical—just as I might take a red apple and imagine that same apple as if it were blue. Now, I do not want to beg the question and say that in conceiving of something as being causally integrated one is thereby conceiving of it as physical; the point is merely that it is not so obvious that causal integration of non-physical properties is genuinely conceivable as such. Conceiving is a slippery thing and we should be modest in what we claim to conceive—in particular, we should question whether we can conceive of the causal integration of a non-physical property qua non-physical property.

Now, while I endorse the foregoing, I recognize that not everyone will find these arguments compelling. But even if this is the case, it should not mean entirely giving up on a causal account of the physical: though perhaps imperfect in its present form, a causal account might still be the best option on the table. Indeed, if one considers the possibility of causally integrated entities to be the Achilles’ heel of a causal account, it’s important to note that they are just as much an issue for physics-based accounts. Indeed, if there were some property that was so integrated into our causal networks as to be causally indistinguishable from the physical, then it would seem like we should be able to run tests on, perform experiments about, and generally study it in a scientific manner. But then it’s not clear why it would not fall under the purview of an ideal physics, and why it would not be considered physical by a physics-based account. Physics-based accounts may well have responses to this, but of the ones advanced in the literature (claiming that the mental and physical aren’t incompatible (Dowell, Reference Dowell2006), adding a No-Fundamental Mentality clause (Wilson, Reference Wilson2006)), all are available to a causal account as well. So, at the very least, causal accounts seem to be no worse off than the leading alternative.

4.2. Dowell’s counterfactual argument

In this final subsection, I want to return to an argument that we saw earlier in the paper and which might seem to pose a problem for my view. Recall that when surveying the literature in Section 2, we looked at definitions which worked by directly or indirectly identifying properties that were supposed to define the physical and then we switched to looking at how definitions could appeal to physics to solve the problems of those views. This transition follows the typical development of the dialectic in the literature, which is motivated in part by the failure of definitions that simply list objects or properties and in part by an argument that this failure will generalize. Specifically, Dowell argues that, for any given property, we can imagine physics discovering some object without it, but despite that, we would not say physics disproved physicalism. According to Dowell, physicalism should not be disprovable by physics. The upshot of this is supposed to be that an appeal to physics is the only way to prevent an unacceptable separation of physics and the physical.

Now, drawing this conclusion would rule out the definition of the physical I’ve proposed here, as it makes no appeal to physics. However, Dowell’s argument should be considerably less compelling when we consider a causal definition of the physical. Under the current definition, a property could come out as non-physical in two ways. One way would be because the property was not causally relevant to exemplary physical properties. But in this case, it’s not clear how a physicist could detect it: if it eludes all microscopes, cloud chambers, and whatnot, how do we expect a physicist to discover it? While the holistic nature of theory confirmation might make this, in principle, possible, it’s hard to imagine what theoretical work could be done by a property that is, in principle, incapable of causally interacting with the properties the theory otherwise concerns.Footnote ¹³ So in this case, physics would not disprove physicalism as Dowell is concerned it might. The second way a property might not count as physical under my definition is if it were causally relevant to exemplary physical properties but was itself beyond the causal reach of exemplary physical properties. In this case, I think we might just say that physics did discover that physicalism is wrong. If it turned out that there was some unmoved mover which causally determined our world from the outside, then it might just be time to give up on physicalism. Indeed, it would seem to violate the principle of causal completeness that physicists hold dear. In sum, anything outside of the causal definition would either not be posited by physics or would, in fact, warrant the rejection of physicalism. In one case, the state of affairs Dowell describes would not obtain, in the other case, I maintain that my definition gives the correct description. Either way, the causal definition withstands Dowell’s counterfactual argument.

The significance of this is not just that my account has been saved from an objection, but a gap in the dialectic has been uncovered. The transition to definitions of the physical in terms of physics comes very quickly in the debate, and it comes too quickly. Whether or not the particular definition I’ve advanced works, there is room for definitions that make no reference to physics. The appeal to physics was implemented to give us some flexibility and insurance against our inability to exactly identify the properties that characterize the physical, but this flexibility and insurance can be obtained in other ways. Definitions which, rather than simply listing properties held by physical entities, identify the inter-relations which unify the physical do exactly this. Not only do such definitions avoid the problem that no properties obviously define the physical, but they can also avoid the stronger counterfactual argument Dowell makes about physics proving physicalism wrong. This means one can consistently occupy a place in logical space between definitions that list properties and definitions that appeal to physics. I hope that in this paper I’ve made that seem like more than just a logical possibility.