Zombies from Below
Abstract: A zombie is a creature just like a conscious being in certain respects, but wholly lacking in consciousness. In this paper, I look at zombies from the perspective of basic ontology ("from below"), taking as my starting point a trope ontology I have defended elsewhere. The consequences of this ontology for zombies are mixed. Viewed from below, one sort of zombie--the exact dispositional zombie--is impossible. A similar argument can be wielded against another sort--the exact physical zombie--but here supplementary principles are needed to get to the impossibility result. Finally, at least two sorts of zombie--the behavioural and functional zombies--escape these arguments from below.
A zombie is a creature just like a conscious being in certain respects, but wholly lacking in consciousness. Zombies have played a central role in recent theorizing about consciousness. They often appear in thought-experiments designed to elicit anti-materialist intuitions. Consider, for example, the exact physical zombie. This is a being just like me in all physical respects--let me be the conscious being in question--but not conscious. If such a being is so much as possible, then, it would seem, any theory on which consciousness is reducible to the physical is false. After all, my zombie twin is not conscious, yet shares with me any physical feature a materialist may claim as a reductive base of consciousness.
But is the zombie--whether it be an exact physical zombie or some other sort--even possible? If not, anti-materialist arguments making essential use of zombies would not get off the ground. These creatures have accordingly been scrutinized from several viewpoints to ascertain whether, and in what sense, they are possible: (1) Conceivability: Are zombies conceivable, and if so, is conceivability a reliable guide to possibility? (2) Knowledge: Is the possibility of zombies compatible with what we claim to know about consciousness? In particular, if my zombie twin is possible, do I have any way of knowing about my own conscious experiences? (3) Meaning: What do zombies mean with their psychological vocabulary, with terms such as "understands" and "consciousness"? And do such semantic facts reveal an incoherence in the notion of a zombie? (4) Evolution: If zombies are possible, does that mean consciousness is "invisible" to natural selection, and would this preclude an evolutionary explanation of the emergence of consciousness? (5) Causation: Is the possibility of zombies compatible with the causal role that consciousness seems to play with respect to our behavioural and cognitive lives?
Each of these lines of inquiry is promising, and a problem as difficult as the nature of consciousness deserves the resources of as many areas of philosophy and science as we can bring to bear on it. In this spirit, I propose in this paper to look at zombies from yet another perspective: ontology. I'll call this the view "from below", since it starts with the most basic features of our world. In particular, I'll look at zombies from the perspective of a trope ontology I have defended, albeit piecemeal, in other places. In section 2, I'll outline this ontology, focusing on those elements most relevant to the possibility of zombies. The consequences of this ontology for the zombie will be mixed. Viewed from below, one sort of zombie, the exact dispositional zombie, looks impossible (section 3). A similar argument can be used against the exact physical zombie, but here supplementary principles are needed to get to the impossibility result (section 4). Finally, at least two sorts of zombie, the behavioural and functional zombies, escape these arguments from below (section 5).
2. Sketch of a Trope Ontology
I take as my starting point the ontology of properties presented in Heil and Robb 2003. Although we do not use the term "trope" in that paper, what we defend is recognizably a trope ontology, for it has the following core features: (1) Properties are what characterize objects; they are "ways objects are". (2) Properties are in re, in the objects they characterize; they are not, say, predicates, concepts, or Platonic forms. (3) Properties are particulars; they are thus distinguished from types, whether types are predicates, universals, classes, or some other sort of entity. (4) Properties are sparse: there need not be a unique property, or even a unique kind of property, corresponding to every meaningful predicate. (5) Properties can resemble one another, either exactly or partially, and such resemblance grounds sameness of type; that is, when two objects belong to the same type, it is in virtue of their instantiating resembling properties. (6) Properties are not reducible to items in other ontological categories, such as objects or universals.
Although trope theorists do not speak with one voice, these six features are common to many trope ontologies. More important for present purposes, however, are two further principles from the Heil-Robb paper:
Identity: Every property is both qualitative and dispositional.
Simplicity: The qualitativity and dispositionality of a property are not aspects or properties of it: they are rather the property itself, differently considered.
My primary aim here is not to argue for these principles--let alone to defend a trope ontology more generally--but rather to look at their consequences for the zombie. Nevertheless, a few remarks about each are in order, both to motivate the principles within the context of a trope ontology and to forestall a few potential objections to the discussion of zombies that follows.
With Identity, Heil and I reject the traditional distinction between qualitative (or categorical) properties and dispositional properties. All properties are both. Trope theory may not make Identity inevitable, but it does make the principle more plausible than it would be otherwise. Start with the qualitativity of every property. It's not easy to say what it is for a property to be qualitative, but when a property is physical, part of its being qualitative is its filling in space. Qualities "colour" our world in a way that bare (non-qualitative) dispositions would not, if there were such. Tropes, at least the physical ones, fit the bill: a trope is in the object that has it, and this at least involves being located where the object is. A physical trope that doesn't fill in space is difficult to imagine. Granted, it may be that some tropes are not physical. But the most plausible candidates for such properties, if there are such, are phenomenal properties, and these are qualitative if any are.
As for the dispositionality of every property, this too is made more plausible by the doctrine that tropes are in re characterizers of their objects. While predicates or abstract forms may not bestow causal powers on the objects that have them, a property that's in its object, and indeed is just a way that object is, is almost certainly in the causal mix. Alexander's Dictum--"To be real is to have causal powers"--may be controversial as usually formulated. But restricted to what's in re, the dictum enjoys a much stronger intuitive appeal.
Trope theory is friendly to Identity in another way as well: an ontology of tropes undermines two influential arguments against the principle. First, the sparseness of tropes blocks any simple semantic argument for distinguishing qualitative and dispositional properties: because there is no one-to-one correspondence between predicates and tropes, there's no reason to think that the undeniable semantic differences between qualitative and dispositional predicates is mirrored by an ontological distinction between qualitative and dispositional properties. There is no easy inference, for example, from the non-synonymy of "is made of salt" and "is water-soluble" to the conclusion that this particular cube's being made of salt (that qualitative trope) isn't identical to its being water-soluble (that dispositional trope). Second, and for similar reasons, the familiar argument from multiple realizability establishes at most a distinction between dispositional and qualitative types. Perhaps any dispositional type--universal, class, concept, or predicate--is realizable in a variety of distinct categorical types, thereby blocking Identity were it formulated as a thesis about types. But if Identity is, as intended, a thesis about properties (tropes), it's unaffected by this argument.
I should note, by the way, that even if every property is dispositional, this doesn't mean that the powers a property bestows must be powers to affect physical objects. That is, it's compatible with the dispositionality of a property that it be "epiphenomenal" with respect to a given family of properties, such as the physical properties. I mention this now to forestall a potential objection to what follows. One may worry that by insisting on the dispositionality of every property, and thus of phenomenal properties, I am begging an important question against those who think zombies are possible. For it is natural to think the possibility of zombies entails that consciousness is epiphenomenal with respect to the physical, and in particular, with respect to behaviour. But at least at this point in the discussion, such epiphenomenalism is an open question. Identity requires that all properties, and so all phenomenal properties, bestow causal powers on the objects that have them, but it does not require that these be powers to affect the physical.
Turn now to Simplicity, so-called because it follows from the more general thesis that properties are "attributively simple": properties do not themselves have other properties or attributes. This is not to say that properties are simple in every way. In particular, properties can be mereologically complex: they can have other properties as parts. For example, the colour of a sphere has, as parts, the colour of its left half and the colour of its right half. The length of a stick has as parts the lengths of the stick's two halves (and, at a lower mereological level, the lengths of the stick's four quarters, and so on). I'll return to such complexity later, but for now, the point is just that even if some properties are mereologically complex, all of them are attributively simple.
Why should a trope theorist believe in Simplicity? For one thing, because tropes are sparse, there's no straightforward semantic argument for thinking that properties themselves have properties. It's certainly true that various predicates apply to any given property. A given red trope, for example, falls under the predicates "is a property", "is bright", "is a colour", "is qualitative", "is dispositional", and so on. But given that there's no one-to-one match between predicates and properties, there's no easy route from these predications to higher-order tropes. The truthmaker for "This property is bright" and the like may be just the property itself. So trope theory at least weakens one motivation for believing in properties of properties.
And indeed, once an ontology of tropes is in place, there's little motivation from any other source to believe in higher-order properties. Consider, for example, the resemblance among tropes, something that initially may appear to call for higher-order properties. Two colours, for example, may resemble each other, perhaps exactly. But what progress is made by explaining such resemblance in terms of the higher-order tropes they instantiate? Presumably the higher-order tropes can't explain the resemblance of the colours unless the higher-order tropes themselves resemble, and we are at the start of a vicious regress: the fact of resemblance is passed to (explained in terms of) resemblance at a higher level, and so on. To stop the regress, we must eventually reach tropes at some level that resemble just in virtue of themselves. But if attributively simple tropes are unavoidable, there seems to be little reason to deny this status to the first-order tropes and be done with it. In this case, at the first order, resemblance among properties is primitive: it cannot be explained in more fundamental terms.
Simplicity is just a particular instance of this more general thesis. The qualitativity and dispositionality of a trope aren't properties of the trope. If they were, then given Identity, they themselves would need to be qualitative and dispositional, and a regress threatens again. But even setting aside Identity, there is little explanatory value a trope theorist has to gain from saying that qualitativity and dispositionality are higher-order tropes. Consider, for example, the qualitativity Q of a scarlet trope S, and suppose Q were a higher-order trope of S. If Q has the same qualitative nature as S, then clearly no progress has been made in explaining S's qualitativity. But if Q has some other qualitative nature or, worse, no qualitative nature at all, how could it have anything to do with S's qualitativity? Similar remarks could be made about S's dispositionality.
Much more needs to be said about both principles, but with this sketch of the ontology in place, I'll now move on to my main topic in this paper. What do zombies look like given the ontology? Viewed from below, from the perspective of this trope ontology, do zombies look possible?
3. The Exact Dispositional Zombie
The answer to this question depends on what a zombie is stipulated to be. Earlier, I described a zombie as a creature just like a conscious being in certain respects, but wholly lacking in consciousness. But the "respects" in question can vary. For example, are they limited to intrinsic features, or do they also include extrinsic or "wide" features as well? Are they dispositional, physical, or both? And is the resemblance between my twin and me supposed to be exact or merely partial, limited to certain salient respects? The answers to these questions generate several sorts of zombie, far more than I can look at here. To simplify matters, I will take the respects of resemblance always to be intrinsic. And I will begin with just one sort of zombie, a non-conscious being who exactly resembles me dispositionally: this is the exact dispositional zombie. Starting with a description of my twin as dispositionally just like me, I will try to show, using the ontology of properties just sketched, that such a being must be conscious, and so not a zombie. I'll present the argument as a series of numbered steps.
(1) My twin and I fall under all of the same dispositional types.
This much can be stipulated, so long as we are careful to remain neutral on the ontology of types. (So, for example, (1) might be read as the relatively innocent claim that my twin and I fall under all of the same dispositional predicates.) Since the resemblance here is exact, my twin behaves as I do, and in addition shares my non-behavioural dispositions; for example, we both tend to turn red when exposed to the sun for too long. And note that we must also share our unobservable dispositions, down to the finest detail. If I am, say, disposed to undergo a very slight rearrangement of unobservable parts given the small gravitational attraction of a passing car, then my twin is disposed to undergo an exactly similar rearrangement in the same circumstances. And so on.
(2) Therefore, for every dispositional property I have, my twin has one dispositionally just like it.
This inference from (1) to a "trope isomorphism" between me and my twin would be unwarranted if the two of us were merely partially similar dispositionally--that is, if we fell under merely some of the same dispositional types. (I will return to this point in the final section.) However, if we are dispositionally indiscernible, if we fall under all of the same dispositional types, then trope isomorphism is inevitable, for anything less than such an isomorphism would have to be reflected in a dispositional difference somewhere, even if at some very fine level of detail.
(3) My phenomenal properties--that is, those properties in virtue of which I am conscious--are dispositional.
This follows immediately from Identity: Every property is dispositional; a fortiori, my phenomenal properties are. I assume here that there are phenomenal properties, a controversial assumption even within the confines of the ontology. I cannot take on eliminativism about phenomenal properties here, except to say that the primary evidence for their existence is that we are--or at least seem to be--directly acquainted with them in conscious experience. There are some who would dispute the evidential value of such seeming, but I cannot do justice to their objections here, and so will just take for granted the existence of phenomenal properties.
(4) Therefore, my twin has properties dispositionally just like my phenomenal properties.
This follows from (2) and (3).
(5) Properties that are dispositionally indiscernible are qualitatively indiscernible.
That is, if two properties exactly resemble dispositionally, then they exactly resemble qualitatively. This crucial step is supported by Simplicity, on which the dispositional resemblance of two properties is not a matter of their instantiating resembling properties, i.e. higher-order tropes. In general, when properties resemble, it's not in virtue of their properties (they have none), but just in virtue of themselves. So two properties cannot be exactly alike in one "way" but not exactly alike in another. There just are no ways in a trope's nature to allow for such variation. And so if two properties are dispositionally just the same, they are qualitatively just the same as well. Here's another way to put the point: If two properties "share" their dispositional natures (i.e., exactly resemble each other dispositionally) then they must share their qualitative natures, for on Simplicity, a property's dispositional nature is its qualitative nature, each being just the property itself.
This point remains whether the resembling properties are mereologically simple or complex, though the reasoning is a bit more complicated in the latter case. Let T1 and T2 be dispositionally indiscernible properties, and suppose these tropes are mereologically complex, so that their dispositional resemblance is grounded in the dispositional resemblance of their parts. Exact dispositional resemblance here requires exact dispositional resemblance at every mereological level, so for every part of T1, there is a part of T2 dispositionally indiscernible from its counterpart in T1. The reasoning in the previous paragraph will then show that these parts must be qualitatively just the same. (If these parts themselves have parts, then repeat the reasoning until mereologically simple properties are reached.) And so however the qualitative parts of T1 combine to yield the qualitative nature of T1, the qualitatively identical parts of T2 will yield a qualitatively identical property.
(6) Therefore, my twin has properties qualitatively just like my phenomenal properties.
This follows from (4) and (5).
(7) Therefore, my twin has phenomenal properties.
This follows from (6): Phenomenal properties are paradigmatically qualitative, and so anything qualitatively just like a phenomenal trope is itself phenomenal. There are, granted, some views on which the phenomenal nature of a property is not intrinsic to it, so that a property can be qualitatively just like a phenomenal property, yet fail to be phenomenal. This is the case on, for example, "higher-order thought" theories of consciousness. On such views, roughly, for a property to be phenomenal, it must be the subject of a thought about that very property. If what makes a given phenomenal property of mine phenomenal is that it's the subject of such a thought, then it's possible that there be a property qualitatively just like my own, yet not phenomenal, since this qualitative duplicate may not itself be the subject of higher-order thinking. But whatever the merits of the higher-order-thought theory of consciousness are, I don't think it's going to have much effect on the present argument. Any difference in higher-order thinking between me and my twin would have to be a dispositional difference. After all, it would consist in a difference of properties instantiated, all of which are, by Identity, dispositional. But my twin and I are dispositionally indiscernible, so if my phenomenal properties are the subject of higher-order thoughts, so are those of my twin. The upshot is the same: since my twin has properties qualitatively just like my phenomenal tropes and (let us suppose) all are the subject of higher-order thinking, my twin's tropes must be phenomenal.
(8) Therefore, my twin is conscious (and so not a zombie).
This is a consequence of (7), for to be conscious just is to instantiate phenomenal properties. (Note that a conclusion even stronger than (8) is warranted by the foregoing: Not only is my twin conscious, his phenomenal properties exactly match my own. That is, he is not even phenomenally "inverted" with respect to me. What's said about zombies in this paper, then, is also relevant to the possibility of phenomenal inversion, but I will not explore that topic here.)
4. The Exact Physical Zombie
Even if the above argument is sound, it rules out only one sort of zombie. Consider now a creature physically just like me, but, as before, wholly lacking in consciousness. In this case, my zombie twin and I fall under the same (intrinsic) physical types: this is the exact physical zombie. Such a zombie is arguably the sort most often discussed in the contemporary literature. The exact physical zombie and the exact dispositional zombie are easily conflated, but strictly speaking they are distinct. There is nothing in the notion of a disposition--whether it be type or trope--requiring that the physical and the dispositional coincide. Note, for example, that it's compatible with the conclusions of the previous section that all of my phenomenal properties, in spite of being dispositional, are non-physical. In this case, the exact dispositional zombie would remain impossible, but the exact physical zombie would be possible. Ruling out the latter requires more work than what's been done so far.
What, then, is needed to supplement the ontology in order to rule out the exact physical zombie? The thesis that every phenomenal property is physical would help. But while I do endorse such a psychophysical "trope monism", a direct appeal to it is not needed here, for in fact something weaker will do. Note first that each of my physical properties is dispositional: this is a consequence of Identity. Now suppose that each dispositional property is paired with a qualitative "ground". On Identity, a dispositional property is its ground, but set that aside for a moment and consider just the following:
Grounding: My phenomenal properties are among the qualities grounding my physical dispositions.
This principle falls short of identifying physical properties with their phenomenal grounds. How short it falls depends on the nature of the grounding relation, here left unspecified; perhaps it is a nomological relation, or, more broadly, a species of supervenience. In any case, however the grounding relation is spelled out, its initial characterization will need to be strong enough so that once combined with the trope ontology--and in particular, with Identity--Grounding collapses a physical disposition into its qualitative ground. This forms the basis of an argument against the exact physical zombie, as follows.
The argument can begin along the lines of the previous section: My twin, by stipulation, falls under all of the same physical types I do, and since the resemblance is exact down to the finest detail, an inference to a trope isomorphism is permitted as before. In this case, however, exact physical resemblance is in play: for every physical property I have, my twin has one physically just like it, a trope of the same physical type.
Now bring in Grounding. My phenomenal properties ground some of my physical dispositions, and Identity collapses the qualitative "ground" of a dispositional property into the property itself. So with these principles we can conclude that my phenomenal properties are physical. Since my twin physically duplicates all of my physical properties, and since these now include my phenomenal properties, it follows that for every phenomenal property I have, my twin has one physically just like it.
But are these properties in my twin qualitatively just like my phenomenal properties, and thus phenomenal? We are still not to this conclusion, for it could turn out that physical indiscernibility is--like the behavioural and functional indiscernibility to be considered in the next section--a case of partial resemblance. The notion of a physical type (or for that matter, a physical property) used so far is too under-described to rule this out. There are at least two ways of "thickening" the notion that would complete the argument. Both attempt to fill out part of what it is to be a physical type, though neither attempts a definition. The first is:
(P1) Properties of the same physical type exactly resemble.
This is the trope theorist's version of the claim that physical types are (or stand in for) genuine "universals". Given (P1), those properties in my twin that are of the same physical type as my phenomenal properties will exactly resemble them. By Identity, exact resemblance must include qualitative resemblance, so my twin's tropes are phenomenal as well.
This route to the conclusion, it should be noted, does not depend on Simplicity, since Identity, Grounding, and (P1) suffice. There's a different completion of the argument, however, that is structurally more like the argument against the exact dispositional zombie, for it ultimately depends on Simplicity. This appeals to a thesis slightly weaker than (P1):
(P2) Properties of the same physical type exactly resemble dispositionally.
As I mentioned earlier, the physical and the dispositional needn't coincide. After all, if there are non-physical properties, they are, by Identity, dispositional. But (P2) says merely that exact dispositional resemblance is part of what it is to be of the same physical type. Combining this principle with Simplicity will yield the conclusion. Since the relevant properties in my twin are of the same physical type as my phenomenal properties, they are, by (P2), dispositionally indiscernible. But by Simplicity, as in the previous section, exact dispositional resemblance entails exact qualitative resemblance. My twin, that is, has phenomenal properties.
Clearly the route to impossibility here is less direct than it was in the previous section, for it goes through Grounding and either (P1) or (P2). And the argument is distinctive in another way: Grounding is a contingent thesis. After all, it would be false if Cartesian dualism were true, and even materialists are usually willing to concede that such dualism is at least possible. So even if the ontology is necessarily true, it cannot be used in all possible circumstances to rule out the exact physical zombie. And this is as it should be, for if Cartesian dualism were true, the exact physical zombie clearly would be possible. So the conclusion that exact physical zombies are "impossible" must be limited to those zombies duplicating conscious beings in whom Grounding is true. By contrast, the last section's argument against the exact dispositional zombie is not similarly limited, for it appeals to Identity and Simplicity without needing Grounding as a supplement, and the former two principles, if true at all, are necessarily true.
5. Other Zombies and Partial Similarity
The arguments of the previous two sections target two sorts of zombie, but there are other sorts whose possibility is compatible with the ontology, even when supplemented by some suitably modified version of Grounding.
Consider first the behavioural zombie. This is a being whose behavioural dispositions are just like mine, but who is, as before, wholly lacking in consciousness. If behaviour is here restricted to overt bodily behaviour, then my twin is partially similar to me dispositionally. That is, my twin and I fall under some of the same dispositional types, namely the behavioural ones, but not all of them. And because the dispositional resemblance is partial, an argument of the form used against the exact dispositional zombie will be of no use here.
Such an argument would, in fact, be blocked at the first inference. My twin and I fall under all of the same behavioural-dispositional types, but the trope isomorphism of section 3 won't follow from this stipulation. Indeed, one can't even infer here that my twin exactly duplicates my behavioural-dispositional properties. Consider, for example, the property I have in virtue of which I tend to wince when my skin is burned. This is, no doubt, a mereologically complex property, composed all of sorts of properties in my nervous system and elsewhere, but it is a property all the same. Now since my behavioural twin also tends to wince when his skin is burned, he has a property that, to this extent, is dispositionally similar to my own "wincing" property. But this is just a case of partial similarity. Properties often bestow some of the same powers without being dispositionally indiscernible. The molecular structures (complex properties) of two objects may dispose each to dissolve in water, yet may make one digestible to humans and the other not. Examples could be multiplied.
Note, then, that even given some behavioural version of Grounding, we still will not have the resources to show that my behavioural-dispositional twin must be conscious. Suppose, that is, my phenomenal qualities are among those grounding my behavioural dispositions. Even so, since my twin's behavioural-dispositional properties are partially like my own dispositionally, they will be partially like my own qualitatively. And the qualitative differences between my twin and me may be exactly what makes my twin phenomenally "dark" inside. The upshot is that for all the ontology tells us, even when supplemented by an appropriate Grounding thesis, the behavioural zombie is possible.
Similar points show that the functional zombie survives the previous arguments. A functional zombie falls under all of the same psychologically relevant dispositional types that I do. That is, my functional twin and I are dispositionally just the same in any way that would be of interest to the science of psychology, whether it be current psychology or some idealized version. Viewed from below, such zombies do not look impossible. (I am here neutral on whether they are in fact possible.) Like my behavioural twin, my functional twin is partially similar to me dispositionally. And so he will be partially similar to me qualitatively. As a consequence, the ontology leaves open whether my twin has phenomenal properties. And this remains so even given the appropriate form of Grounding. That is, even if my phenomenal qualities are among those grounding my psychological dispositions, my functional twin won't have properties duplicating these qualities. Partial qualitative resemblance is all he enjoys, since only partial dispositional resemblance is stipulated to hold. And for all the trope ontology tells us, phenomenally "dark" qualities can resemble my phenomenal properties to the appropriate degree. My functional twin, that is, need not be conscious.
It may be wondered at this point, however, whether the ontology really does warrant treating the behavioural and functional zombies so differently from their more exact cousins. The exact dispositional zombie, for example, is supposed to look impossible from below due to (among other things) Simplicity. In particular, I said that a property's dispositional nature is not an aspect of the property, but just the property itself. There's thus no room for two properties to be just alike dispositionally without being just alike qualitatively. But then how can two properties resemble behaviourally or psychologically without being exactly alike, both qualitatively and dispositionally? After all, the objection goes, by Simplicity, the behavioural-dispositional or psychological-dispositional "nature" of a property is just the property itself. So bestowing the same behavioural or psychological powers should result in dispositional (and thus qualitative) indiscernibility. Bring in the appropriate forms of Grounding, and the behavioural and psychological zombies should, like their more exact cousins, look impossible. The worry here, in short, is that Simplicity takes what initially looked like partial resemblance and collapses it into exact resemblance--rendering all of these zombies impossible.
But Simplicity is in fact tolerant of partial similarity. This is clearest if the resembling properties are mereologically complex: partial similarity could just amount to similarity of some (but not all) parts. And in particular, the partial dispositional similarity of two complex properties could just amount to the dispositional similarity of some of their parts. This is compatible with Simplicity: no appeal to dispositionality as a "higher-order trope" is needed. Simplicity also permits the partial similarity of mereologically simple properties, though in this case an explanation is harder to come by, since, as I mentioned earlier, the similarity of simple properties is primitive. Still, I can say this much: Since partial resemblance among simple properties cannot be analyzed, it cannot (a fortiori) be analyzed as exact resemblance in some particular way or other. When simple properties partially resemble, then, there's no sense in which they "share" a nature: they just partially resemble, and that's the end of it. This is why Simplicity doesn't threaten to collapse all resemblance into exact resemblance, and it's why in the end the ontology permits the behavioural and functional zombies.
 This paper expands on the brief discussion of zombies in Heil and Robb 2003, 189. An earlier and shorter version was presented at the 2007 Central Division Meeting of the American Philosophical Association in reply to Kevin Sharpe's "Tropes and the Zombie Argument". I thank Sharpe for his paper and the subsequent discussion, which prompted me to write the present article. Thanks also to Francesco Orilia and Chris Shields for comments on an earlier draft.
 The sort of consciousness in play here is phenomenal consciousness, that is, consciousness of the "what it's like" variety: see Nagel 1974, Block 1995, Chalmers 1996. An overview of the zombie literature is Kirk 2006; see also the Journal of Consciousness Studies 2:4 (1995).
 Philosophers sometimes distinguish varieties or grades of possibility, such as nomological, metaphysical, and logical possibility. In the discussion of zombies to follow, I'll always have in mind metaphysical possibility, which is perhaps more accurately called intrinsic possibility, or even just possibility simpliciter: cf. Van Inwagen 1998.
 Examples from the literature include the following, though many of these could fall into more than one category, as there are important connections among the approaches. Conceivability: Levine 2001, ch. 2, Marcus 2004; Knowledge: Block 1980, Shoemaker 1981; Meaning: Moody 1994, Thomas 1998; Evolution: Flanagan and Polger 1995; Causation: Chalmers 1996, chs. 4-5, Bailey 2006.
 Primarily in Heil and Robb 2003, but also in Robb 1997, 2001, 2005.
 We use "property", as I will do for the most part in this paper, occasionally switching to "trope". I consider the terms to be interchangeable.
 Useful overviews of trope theory can be found in Armstrong 1989, ch. 6, Campbell 1981, 1990, and Maurin 2002.
 These principles can be found in Heil and Robb 2003, 184-5. I've reworded both slightly and given them new names.
 Cf. Blackburn 1990, Martin 1997, 222.
 See Kim 1993, 348 ff.
 See Prior et al. 1982.
 See Heil 1999. In Robb 1997, I make the similar point that the multiple realizability of the mental in the physical shows at most that mental types are not physical; cf. Kim 1993, ch. 16.
 One route from the possibility of zombies to epiphenomenalism goes through the causal closure of the physical: see Stoljar 2001b and, for remarks on closure, Kim 2005, chs. 1-2. Chalmers (1996, 150ff.) explores but does not endorse the zombie-epiphenomenalism connection. In any case, I suspect this connection is most often asserted by opponents of zombies: see Perry 2001, ch. 4, Bailey 2006.
 Cf. Armstrong's (1978, 68-71) structural universals. For more on the mereological complexity of some properties, see Robb 2005.
 When the resembling properties are mereologically complex, their resemblance can be explained in terms of the resemblance among their parts. But this buck cannot be passed indefinitely: at some point, resemblance will have to "ground out" in the primitive resemblance among mereologically simple properties.
 For a partial defence, see Heil and Robb 2003, and for more a more detailed defence, Heil 2003, 2005. See also Martin 1997 and Martin's contribution to Armstrong et al. 1996.
 Cf. Güzeldere 1995.
 This is a significant simplification on some views. For example, Dretske's (1995) brand of representationalism about consciousness allows a non-conscious creature just like me in every intrinsic physical respect, but it prohibits a non-conscious creature just like me in every intrinsic and extrinsic (in particular, historical) respect.
 I here consider only forward-looking dispositional types, but my twin and I also fall under the same backward-looking dispositional types. On this distinction, see Shoemaker 2003, 412.
 See e.g. Rey 1997, 305-7.
 When T1 is a phenomenal property, these qualitative parts may be what Chalmers 1996, 127 calls "protophenomenal properties", properties that, while not themselves phenomenal, combine to produce phenomenal properties.
 See e.g. Rosenthal 1997.
 Robb 1997, Heil and Robb 2003. Whether this sort of trope monism should count as physicalism is a question I will not take on here. Note in any case that even if trope monism were assumed, ruling out the exact physical zombie would not be trivial, for we would still need (P1) or (P2) (see below).
 Such relations are usually defined in terms of types (or families of types), but they have trope versions as well. On the varieties of supervenience, see Kim 1993, ch. 4. Chalmers 1996, ch. 4, endorses a nomological version of Grounding, albeit with respect to physical and phenomenal types, not tropes.
 For a useful recent discussion of the nature of the physical, see Stoljar 2001a, and for scepticism about finding an adequate definition, Daly 1998.
 A related point, presented primarily in semantic terms, is in Braddon-Mitchell 2003.
 I'll take partial similarity here to be mere partial similarity, similarity falling short of indiscernibility. And I'll think of a behavioural zombie as a mere behavioural zombie. That is, I will not count an exact dispositional zombie as a behavioural zombie, even though the former shares all of my behavioural dispositions. I don't think much turns on this decision. If instead the exact dispositional zombie is included as a version of the behavioural zombie, the main point can just be rephrased: In the ontology, there's nothing in the notion of a behavioural zombie as such requiring that it be conscious, even if some "overqualified" versions of it--namely, the exact dispositional versions--must be conscious. (Similar remarks apply to the functional zombie considered next.)
 For further discussion of partial similarity, see Heil 2003, ch. 14.
 (P1) or (P2) would be of use here if it were stipulated that my twin's behavioural-dispositional properties are of the same physical type as my own. But then my twin would be a different sort of zombie: he includes more in his description than the standard behavioural zombie.
 Cf. Heil 2003, 247-8.
[Thanks to philpapers.org for providing many of these links.]
Armstrong, D. M. (1978). A Theory of Universals: Universals and Scientific Realism, Volume II. Cambridge: Cambridge University Press.
Armstrong, D. M. (1989). Universals: An Opinionated Introduction. Boulder: Westview Press.
Armstrong, D. M., Martin, C. B. and Place, U. T. (1996). Dispositions: A Debate, T. Crane ed., London: Routledge.
Bailey, A. (2006). Zombies, Epiphenomenalism, and Physicalist Theories of Consciousness. Canadian Journal of Philosophy, 36, 481-510.
Blackburn, S. (1990). Filling In Space. Analysis, 50, 62-5.
Block, N. (1980). Are Absent Qualia Impossible?. The Philosophical Review, 89, 257-74.
Block, N. (1995). On a Confusion About a Function of Consciousness. Behavioral and Brain Sciences, 18, 227-87.
Braddon-Mitchell, D. (2003). Qualia and Analytical Conditionals. Journal of Philosophy, 100, 111-35.
Campbell, K. (1981). The Metaphysic of Abstract Particulars. Midwest Studies in Philosophy, 6, 477-88.
Campbell, K. (1990). Abstract Particulars. Oxford: Blackwell.
Chalmers, D. (1996). The Conscious Mind. Oxford: Oxford University Press.
Daly, C. (1998). What Are Physical Properties?. Pacific Philosophical Quarterly, 79, 196-217.
Dretske, F. (1995). Naturalizing the Mind. Cambridge, MA: MIT Press.
Flanagan, O. and Polger, T. (1995). Zombies and the Function of Consciousness. Journal of Consciousness Studies, 2, 313-21.
Güzeldere, G. (1995). Varieties of Zombiehood. Journal of Consciousness Studies, 2, 326-33.
Heil, J. (1999). Multiple Realizability. American Philosophical Quarterly, 36, 189-208.
Heil, J. (2003). From an Ontological Point of View. Oxford: Clarendon Press.
Heil, J. (2005). Dispositions. Synthese, 144, 343-56.
Heil, J. and Robb, D. (2003). Mental Properties. American Philosophical Quarterly, 40, 175-96.
Kim, J. (1993). Supervenience and Mind. Cambridge: Cambridge University Press.
Kim, J. (2005). Physicalism, or Something Near Enough. Princeton: Princeton University Press.
Kirk, R. (2006). Zombies. The Stanford Encyclopedia of Philosophy, ed., E. Zalta.
Levine, J. (2001). Purple Haze: The Puzzle of Consciousness. Oxford: Oxford University Press.
Marcus, E. (2004). Why Zombies Are Inconceivable. Australasian Journal of Philosophy, 82, 477-90.
Martin, C. B. (1997). On the Need for Properties: The Road to Pythagoreanism and Back. Synthese, 112, 193-231.
Maurin, A.-S. (2002). If Tropes. Dordrecht: Kluwer.
Moody, T. (1994). Conversations With Zombies. Journal of Consciousness Studies, 1, 196-200.
Nagel, T. (1974). What Is It Like to Be a Bat? The Philosophical Review, 83, 435-50.
Perry, J. (2001). Knowledge, Possibility, and Consciousness. Cambridge, MA: MIT Press.
Prior, E., Pargetter, R. and Jackson, F. (1982). Three Theses about Dispositions. American Philosophical Quarterly, 19, 251-7.
Rey, G. (1997). Contemporary Philosophy of Mind. Oxford: Blackwell.
Robb, D. (1997). The Properties of Mental Causation. The Philosophical Quarterly, 47, 178-94.
Robb, D. (2001). Reply to Noordhof on Mental Causation. The Philosophical Quarterly, 51, 90-4.
Robb, D. (2005). Qualitative Unity and the Bundle Theory. The Monist, 88, 466-92.
Rosenthal, D. M. (1997). A Theory of Consciousness. The Nature of Consciousness, eds. N. Block et al., Cambridge, MA: MIT Press, 729-53.
Shoemaker, S. (1981). Absent Qualia are Impossible--a Reply to Block. The Philosophical Review, 4, 581-99.
Shoemaker, S. (2003). Identity, Cause, and Mind, Expanded Edition. Oxford: Clarendon Press.
Stoljar, D. (2001a). Two Conceptions of the Physical. Philosophy and Phenomenological Research, 62, 253-81.
Stoljar, D. (2001b). The Conceivability Argument and Two Conceptions of the Physical. Philosophical Perspectives, 15, 393-413.
Thomas, N. (1998). Zombie Killer. Toward a Science of Consciousness II, eds. S. Hameroff et al., Cambridge, MA: MIT Press, 171-7.
Van Inwagen, P. (1998). Modal Epistemology. Philosophical Studies, 92, 67-84.