Material Constitution as Representational Elimination

Material Constitution as Comprehensive Representation

John Dilworth, Draft only, 10/06

A broadly representational, conceptualist approach to the material constitution of midsized objects is argued for. Assuming a robustly realistic physicalism of ultimate particles etc., some account is needed as to how such real items are related to the midsized objects that we are evolutionarily conditioned to perceptually recognize. A representational solution is proposed: various similarly located, changing clusters of ultimate particles each serve to represent to perception an invariant, self-same worldly object.

In order to make this solution workable, a concept of comprehensive representation--i.e., representation of all of the properties of an object--is introduced. The solution is also unusual in its systematic integration of four areas often kept separate, namely problems in analytic ontology, issues of non-reductive physicalism, evolutionary psychology, and broad metaphysical debates about realism versus anti-realism.

There are well-known problems concerning material constitution, holding between items such as statues and the clay lumps apparently co-located with them, baseballs and their constituting physical microparticles, and so on. Nor is there even any agreement as to how constitution itself should be defined--some regard it as a symmetric relation distinct from that of part-whole composition, others treat it as an asymmetric relation closely related to that of composition, or as a kind of supervenience relation, with various other views being found as well.[1]

Also, problems concerning constitution are only one factor in a whole constellation of problems concerning the identity and existence of midsized material objects, including whether they are reducible to, or eliminable in favor of, the entities postulated by physics, whether they perdure or endure, whether vague existence is possible, problems concerning indefinitely many co-located objects, and so on.

In addition, there is a strong tendency in the literature to artificially separate four interrelated categories of problems concerning constitution. First, analytic ontologists discuss standard metaphysical problems concerning constitution, derived from particular problematic cases such as apparently co-located material items. Second, physicalists and philosophers of mind discuss problems concerning non-reductive physicalism. Third, evolutionists and cognitive scientists seek to explain the origins and mechanisms of object-related human conceptual structures. While fourth, broad metaphysical debates about realism versus anti-realism provide a more encompassing stage or setting for all of the other categories of debates.

Now from a broad explanatory point of view, it would be extremely desirable, if at all possible, to come up with a single unified explanation of material constitution, that could also provide the resources to potentially systematically clarify the relations of all four categories of issues, and which could also potentially systematically resolve all of their intersecting problems in one fell swoop. This might seem like a tall order, but all of these problems plausibly can be regarded as arising out of the very substantial differences between the entities posited by physics, such as fields or microparticles, versus the ordinary midsized material objects discussed in the special sciences and assumed by folk psychology.

Presumably, one central reason for postulating a relation of constitution is as a foundational element in an explanation of how entities at these two very different levels of description could be related to each other. My suggestion is that if the right account of the nature of constitution could be provided, then all of these problems and issues would fall into place as arising out of constitution as thus explicated. In this paper such a basic account of constitution will be outlined, that draws upon a perhaps unexpected source: representational concepts. But as is only to be expected, given the very broad scope of this undertaking, the present paper must be confined to providing some initial outlines and hypotheses, leaving most of the details to be provided in subsequent work.

1. Initial Outline

Here is an initial outline of the view to be argued for, both to provide some initial plausibility to the account, and to provide a roadmap for the succeeding sections. To begin, ontological physicalists claim that everything is physical.[2] But as applied to ordinary midsized objects such as chairs, mountains or biological organisms, such a view seems to be clearly false, in that, among other things, their identity-conditions seem to be distinct from those of the microphysical particles etc. that constitute them. Such points give rise to familiar and highly controversial issues of non-reductive physicalism, including issues as to how the ontology of the special sciences, such as geology and biology, relates to that of basic physics.[3]

A wider and more traditional dispute between broadly realistic views on the one hand, versus broadly Kantian, conceptual or irrealist views of the world on the other hand, should also be mentioned. For example, Hilary Putnam has argued at various stages of his career that standard scientific realist views, including extreme views such as that of physicalism, are fundamentally incoherent,[4] while writers such as John McDowell argue that a broadly Kantian view, in which the experienced world is thoroughly infused by human concepts, must be correct.[5]

However, there is a potentially attractive intermediate position on the battleground of realist versus non-realist views that seems not to have been adequately investigated yet, and which provides the foundation for the present account. This intermediate position would accept that everything is constituted by purely physical, mentally and conceptually independent entities, but that nevertheless the crucial relation of constitution is itself a cognitively dependent one. Or in other words, though the entities described by basic physics are mind-independent, the midsized objects that they constitute, such as mountains, cats or human minds, are not. Instead, I shall argue, our dealings with, and concepts of, such midsized objects integrally involve the representational capacities of sophisticated higher organisms such as humans. In a nutshell, it will be my view that microparticles constitute ordinary objects by comprehensively representing them (i.e., representing all of their properties), so that cognitively able perceivers of items can perceive the ordinary objects when they view the microparticles that constitute or comprehensively represent those objects. (For more on the perceptual mechanisms involved see section 6).

As an initial analogy to clarify the issue, consider a picture of a mountain, as painted by an artist. When one looks at the physical surface, one can see the represented mountain there, even though, strictly speaking, all one is looking at is a purely physical paint surface which is not itself a mountain. In such a case, the physical picture surface representationally constitutes the mountain that one can see, in that a specific configuration of physical elements on the surface of the picture represents the seen mountain. My basic suggestion will be that all cases of constitution by microphysical particles may potentially be similarly analyzed. The two main differences from more standard or typical cases of representation are that a) the constituent microparticles represent all of the properties of some midsized object, whereas in the painting case, only some limited subset of the mountain's properties are represented by the painting; and b) the represented midsized object is ontologically dependent on the existence of the painting.

Another broader factor to consider is as follows. Since the current RC (representational constitution) conception of constitution is intended to be fully compatible with physicalism, it is necessary that, either potentially or actually, a fully naturalistic account must be available as to the nature of such representational capacities, and also as to how such capacities actually developed in organisms, and what positive role of such capacitites is in furthering evolutionary fitness. In broad outline, an evolutionary account should argue that there are evolutionary advantages available to creatures who happen to have acquired an ability to perceive and think of the basic physical microparticles--that constitute the physical world--as representing midsized objects (whether or not those creatures could conceptualize the issue thus).

As an initial datum, it is intuitively obvious that our ordinary concepts of midsized objects are much simpler and more manageable than the complex, and cognitively demanding, purely physical concepts which apply to basic physical microparticles and the laws governing them. Thus from an evolutionary point of view, species who accidentally happened to acquire ways of cognitively structuring the world that are much simpler than their actual physical structure--but which ways are nevertheless systematically related in transformational ways to their actual structures--would have significant evolutionary advantages over other species lacking such abilities. Also, as long as such representational abilities and uses are evolutionarily advantageous, no issue need arise as to the precise physical status of what is thus perceived.

From a physicalist point of view, all that is ontologically required of midsized objects, as thus perceived, is that they should be supervenient on physical microparticles. As for the epistemology of perceptual reference, all that is required for compatibility with physicalism is that all perceptual references to midsized objects should always involve references to collections of physical microparticles. Hence, to summarize, the conceptualization of such references in terms of ontologically dependent, representationally constituted midsized objects is a valuable evolutionary strategy for simplifying cognitive interactions with a complex physical environment. So the RC approach to constitution potentially provides an unusual kind of non-reductive physicalism, which is significantly different from more standard functionalist etc. kinds.

Another important desideratum in any physicalist account of constitution is that it should be able to explain both the scientific legitimacy, and utility, of the special sciences--geology, biology and so on--given that such sciences seem to have as their subject matter midsized objects that at best would have a dependent existential status on the RC view. All of these issues will be further discussed in section 5.

2. More Details of the Representational Theory

The concept of representation is a central cognitive concept. Though its precise analysis is a matter of controversy (e.g., Cummins, 1996), it would generally be agreed that representations of worldly items X provide information about those items X to those cognitively using them--whether they occur as internal perceptual representations of worldly objects, or as public representational objects such as diagrams or paintings.

As previously mentioned, our usual concept of representation an item X is confined to cases where only some of the properties of X are represented, such as its visual appearance from a certain angle. In such cases the relevant representations provide limited information about a few aspects or properties of object X. However, as I have shown elsewhere, representational theory can be extended to cover cases of comprehensive representation of an object X, in which all of X's properties are represented in virtue of complete information about X being provided by the representation.[6]

The apparent previous neglect of such a conceptual possibility perhaps occurred for the following reason. Perhaps it has been assumed that such a concept of comprehensive representation would merely involve a degenerate case of representation--one in which a representation A of an object X only succeeds in representing all of the properties of X in virtue of itself having all of those properties of X--and hence itself being X, or an X. Thus understood, the concept of comprehensive representation would apply only to legitimate samples or examples of Xs, which would seem to be either not representations of an X at all, or to be only a degenerate or trivial case of such.

However, I have shown (Author 7, 8) that the concept of comprehensive representation can be used so as to provide a nominalist substitute for ontological properties. Instead of taking the realist position that there are genuine properties of e.g. being a cow, so that cows instantiate cowhood, nominalists could instead regard particular cows as being so in virtue of their comprehensively representing a cow as realistically understood. Particular cows could provide complete information about a cow, in virtue of thus comprehensively representing one, without actually possessing or instantiating a property of cowhood--or any associated property such as that of having a cow's heart. Hence it is possible to comprehensively represent an object that would, if it were real, actually instantiate the property of cowhood, without the representation itself actually having to instantiate that property. This is possible because clearly it is possible to represent items that do not exist, such as unicorns and Santa Claus, by representationally providing appropriate kinds of information about them. To be sure, if nominalists are correct, there are no existent or actually instantiated properties. But this does not prevent us from having or constructing representations of instantiations of such non-existent properties--including, I have argued, comprehensive representations of them.

The current paper applies this same representational structure to the issue of the status of midsized, everyday objects. For compatibility with physicalism, a theory of midsized objects must deny that there are any ineliminable properties that such objects have of being those objects, because the existence of such properties would be inconsistent with a thoroughgoing physicalism which claims that only purely physical properties are actually instantiated. So, as in the nominalist case, there cannot be any genuine or ineliminable instantiations of a property of being a cow. But there is nothing to prevent us from having a concept of such a property of cowhood, and for it to be the case that there exist complex groups of microparticles that comprehensively represent for us a cow as thus conceived. In this manner, groups of microparticles can representationally provide, as required, full information about the immediate perceptual presence of a cow, without us having to pay the ontological price of there thereby being perceptually present to us an actual instantiation of the non-physical property of cowhood.

To be sure, it must be conceded that the resulting concept of representational constitution for midsized objects is not a purely ontological concept. The sense in which a particular group of microparticles representationally constitute a cow is one that depends on the existence of beings having appropriate conceptual structures and representationally based recognitional abilities applicable to midsized collections of microparticles such as those we conventionally call 'cows'. Nevertheless, this is not to deny that there are really are cows, but only to deny a certain realist or platonist construal of what it is for an object to be a cow in a purely physical universe. Our recognition of a certain cluster of microparticles as being a cow is correct as long as that that cluster does indeed comprehensively represent all of the properties of a cow.

The resulting ontology is a two-tier one, in which purely physical properties are fully objective or mind-independent, while all other more informal or conceptually based properties, such as those associated with midsized objects, are broadly Kantian in nature--our conceptions of them, and recognitional capacities with respect to them, play an ineliminable role in our conceptions of their objective status. But, to repeat, this is not to deny that their objects do have an objective status. Instead it is to characterize that status as one that is ontologically dependent both on instantiations of purely physical properties by the relevant microparticles, plus on the existence of the relevant concepts and recognitional abilities in beings capable of using clusters of those microparticles as comprehensive representations of the relevant midsized objects.

As for the relation of this two-tier ontology to the issue of property realism versus nominalism, arguably the issues are independent. Suppose that nominalists are correct in thinking that there are no instantiations of platonistically conceived properties, whether by fundamental physical microparticles or anything else. Their nominalistic substitute, such as tropes or resembling particulars, could still provide the basic ontological tier of the physical reality of microparticles, clusters of which microparticles could still comprehensively represent second-tier ontological items such as ordinary midsized objects. So the present two-tier ontology is not dependent on any particular resolution of the realism versus nominalism dispute.

3. Representation and the 'Problem of the Many'

There is a notorious 'problem of the many',[7] which concerns the fact that there are an indefinitely large number of distinct precise ways in which a material object could be analyzed into relevant sets of microparticles, for instance because of apparent vagueness in its precise boundaries. Any of those sets could plausibly be regarded as having members such that they compose the object, i.e., such that they are all of its material parts; but the sets are all distinct. Either of two intolerable consequences seems to follow: either there are an infinity of material objects co-located with the relevant material object, one for each of the sets of parts, or it is somehow vaguely constituted by all of them, in spite of the paradoxicality of the concept of ontic vagueness.[8]

However, consider a representational analog of this problem, such as a large series of closely similar but distinct pictures of an actual chair C. (A computer program could easily generate such a series, e.g. in the form of digitized images, having minor differences in the relevant arrays of pixels). In such a case, it is completely unproblematic to claim that each is equally a pictorial representation of the same material chair C, in spite of the minor differences in pixel configurations holding between the pictures. In such a representational case, the relation of representation holds separately between each pixel configuration and chair C. Now consider any of the actual sets of microparticles that compete for composing the chair itself. My claim is that each of them comprehensively represents the chair, and that they do not compete with each other in so doing, just as ordinary, non-comprehensive representations of an item X do not compete with each other. So we may distinguish a relation of composition, that does involve competition, from one of representational constitution that does not, and in this way solve the problem of the many for midsized material objects.

So a representationally based account of constitution could provide a third option between the two intolerable options suggested above for constitution, namely multiple co-located material objects versus a single vague object. This third option is that of saying a) that any relevant set of microparticles only represents rather than composes the relevant chair C; and b) hence it is unproblematic to say that each of them representationally constitutes chair C, since as representations rather than collected parts of the distinct entity C, they do not ontologically compete with each other. Hence also C does not have to be regarded as somehow 'vaguely constituted' by the totality of such sets either.

Now admittedly, the relation of representation as provided by pictorial cases may seem much too weak to underwrite an account of material constitution. In such pictorial cases, the physical nature of the pictures themselves--particularly in digitized form--seems to have very little connection with the physical natures of the worldly items represented by them. In addition, pictures represent only a very limited range of properties of that which they represent, while presumably a constitution relation would involve constitution of all of the properties of a material object, not just some of them.

Also, at the same time there is a common intuition that a complete physicalist account of the universe would completely explain the physical nature of ordinary material objects, so that some intimate relation of constitution is required, in which the very material substance of objects is explained as being constituted by the relevant microparticles. A representational account of constitution would somehow have to find a way to bridge the very significant gap between these divergent characteristics. Indeed, it is likely the apparent vastness of this gap, plus the assumption that a concept of comprehensive representation would be degenerate or trivial (see the previous section) which may explain the lack of preceding representational accounts of constitution.

However, several factors, as discussed below, can serve to bridge the gap. First,

the concept of representation can also be applied in cases in which the relations of the relevant representing and represented items are much stronger than those obtaining in simple pictorial cases. For example, representations of an airplane A could be of an enormous variety, from simple pencil or computer sketches, through photographs, scale models, full-size but non-working models, and so on. Indeed, a real airplane might be used to represent plane A, such as if the director of a film about Amelia Earhart's disappearance while flying a plane A used another airplane B of the same kind to represent Earhart's plane. In this range of cases, initially only a very limited range of plane A's properties are represented, but in the final two-plane case, plane B represents all of the intrinsic properties of plane B. Clearly it is such comprehensive representational cases which are most analogous to constitution, rather than more commonplace, non-comprehensive pictorial cases. For as noted above, an intuitive requirement on constitution is that microphysical particles should constitute all of a material object's purely physical properties, rather than just some of them.[9]

Another kind of apparent representational weakness which still needs to be dealt with is as follows. It might be objected that a representational account could not explain the sense in which a constituted material object is ontologically grounded by its representing microparticles. For in normal cases of representation of B by A, even when A comprehensively represents of B, the represented item B exists independently of, and is not ontologically grounded by, the representing item A. This was the case in the Earhart two airplanes case given above, in which the relevant planes A and B are distinct material objects even though A comprehensively represented B. But clearly constitution does--at least, on broadly realistic construals of it--require ontological grounding of material objects by microparticles.

This concern may be addressed by noting that not all cases of representation involve representation of some item B that is ontologically independent of item A. For example, it is commonplace for artists to paint imagined landscapes or people, which paintings do represent those items, but which items, on a strict naturalistic account, would have to be explained in terms of some subset of the physical characteristics of the paintings themselves, since there are no actual external items represented by the picture. Hence the relevant landscapes etc. have only a dependent existence--dependent on the purely physical characteristics of the paintings.[10]

But such a dependent existential status is exactly what is needed to explain the sense in which ordinary material objects are existentially dependent on, or ontologically grounded by, sets of microparticles. To use again an intuitive perceptual analogy, just as one can see a lake when looking at a physical picture of it, so also one can see a midsized material object when looking at the sets of microparticles that constitute it. The difference between the two cases is that the latter cases of constitution involve comprehensive representation, whereas the former ordinary pictorial cases do not.

To be sure, even in all of these dependent existence cases, the items represented have distinct identity-conditions from those of the microparticles that represent them, as is true for any representational relations. This is a desirable feature, because it is generally accepted that constitution is a relation that holds between distinct items having different identity-conditions.[11]

A further concern is that, even if the above replies are successful, more restrictions on the concept of representation are required in an analysis of constitution. Specifically, spatio-temporal co-location is one extra factor required for constitution, since problems concerning constitution typically arise only when apparently coinciding objects, such as lumps of clay and statues, are involved. This point may be accepted by a representational view of microphysical versus material object constitution, and indeed, arguably it is automatically satisfied because of the third consideration discussed above, namely the ontological dependence of material objects on microparticles. Thus it is common ground that in typical cases of constitution, the relevant microphysical particles, and the material object they constitute, must be spatio-temporally co-located.

Another common requirement on constitution that can also be accepted by a representational view is some degree or kind of lawful correlation between the relevant microparticles and corresponding material object. If one removes an arm from a chair, then simultaneously one causes an assemblage of microparticles to change their spatial location, and vice-versa. Indeed, any account of constitution which satisfies both some such nomic covariance or supervenience requirement, and the spatio-temporal co-location requirement, is well on the way to providing at least an initially intuitively satisfying account of constitution. (Which is not to say that the particles cause any of the changes in the material object, or vice versa. The issue is, to repeat, rather one of lawful correlations, which are possibly asymmetric as in supervenience accounts). My claim is that satisfaction of all of the above requirements, including comprehensive representation by some relevant microparticles of a relevant material object, is all that is needed to explain how microparticles could constitute a material object.

4. Vagueness Representationally Defused

Problems concerning vagueness present many thorny theoretical aspects. There is the potential threat of ontic vagueness, arising from the fact that there are an indefinitely large number of distinct precise ways in which a material object could be analyzed into relevant sets of microparticles, e.g., because of apparent vagueness in its precise boundaries. Recall that any of those sets could plausibly be regarded as having members such that they compose the object, i.e., such that they are all of its material parts; but the sets are all distinct. Hence, either there are an infinity of material objects co-located with the relevant material object, one for each of the sets of parts, or the object is somehow vaguely constituted by all of them, in spite of the paradoxicality of the concept of ontic vagueness.

The alternative representational solution suggested there was to regard each of the relevant sets, not as composing the object in a part-whole way, but instead as comprehensively representing it, with the fact that each set is slightly different from the others simply being 'business as usual' in representational cases, since there can in general be many quantitatively and qualitatively distinct representations of a single represented material object X. This kind of solution will be extended in this section.

To begin, a possibly significant additional factor in resolving vagueness problems is as follows. In representational pictures, such as portraits of a famous person such as Napoleon, almost all of them possess at least one of the following two characteristics. First, the pictures represent not only Napoleon, but also some situational or background items such as his horse, his regalia, some natural or court setting, and so on. And second, most of them also are such that not all of Napoleon's body is explicitly represented in the picture--most common are pictures that explicitly show his head and upper body, but not the rest of his body. Yet of course all of these are standard cases of representations of Napoleon himself. In the first situational kind of case, the portraits are of Napoleon himself, and not of some peculiar rectangular entity that has Napoleon as one of its parts, with parts of a situational background making up the rest of its parts. In the second, 'close-up' kind of case, in which only the upper parts of Napoleon's body are explicitly shown, these are still representations of the whole Napoleon, rather than just being pictures of some of his body parts. (Also see the previous section on how representation of objects is not limited to representation of their explicit or currently available spatio-temporal aspects).

The conclusions with respect to constitution that may be drawn from these very common features of standard representations are as follows. First, the requirement from section 1 that a necessary condition of representational constitution is spatio-temporal co-location of microparticles and material object can potentially be interpreted liberally. In ordinary representational cases, as just discussed, all that would be required would be co-location of some part of the microparticles with some part of the object, i.e., a non-disjoint intersection of their respective spatio-temporal sets. This is presumably too liberal for a scientifically useful analysis of constitution, but it does make the point that a precise spatio-temporal coincidence of the two relevant spatio-temporal sets is not necessarily required for constitution.

However, this point, in concert with the previous representational considerations about vagueness, is all that is needed to destroy the potential threats posed by standard vagueness considerations. On standard analyses of vague boundaries of material objects, one specific issue is as to whether a given microparticle, located somewhere near the boundary of the object X, is or is not part of X, or whether instead it is simply indeterminate whether the microparticle is or is not part of X. Whereas on the current representational view, to begin with, no microparticles are ever parts of X itself at all, since they represent X rather than providing parts for X. So on this account, no microparticle-related vagueness issues can ever arise about the boundaries of material objects as such.

Hence the focus shifts to potential vagueness issues, if any, about the representing microparticles themselves. In order for vagueness problems to arise at this level it would have to be the case that e.g. increasingly inclusive sets of microparticles, parts of which are co-located with a material object X, fail at some point in their increasing size to constitute X. But in normal cases of representation such a point is never reached. For example, a picture of Napoleon in the foreground with 10,000 of his troops in the background is still a representation of Napoleon.

To be sure, such a picture represents not only Napoleon, but also his 10,000 troops. So it can readily be conceded that increasingly inclusive sets, in a constitution case, would eventually result in a set of microparticles constituting not only object X, but also one or more other objects as well. But the fact that singular constitution cases gradually enlarge into plural constitution cases does not, again, show that such larger sets fail to constitute X itself. So ordinary vagueness considerations can gain no purchase here, once refocused at the microphysical rather than material object level. Vagueness has become defused as a mere non-specificity issue, rather that remaining alive as a threatening identity issue.

A more precise analysis of one reason why standard vagueness considerations fail in representational cases would be as follows. Recall that a standard concern is that a near-boundary microparticle might, because of vagueness considerations, neither be part of X, nor not be a part of X--that it might simply be indeterminate whether it is a part of X. Thus the relation 'part of' for material objects is claimed to be potentially vague in various metaphysically threatening ways. However, in the case of representation, no such individual-particle issues can arise, because representation is an essentially holistic phenomenon. On the representational view of constitution, it is not the case that any given microparticle represents some corresponding part of the material object X. Instead, it is only sets or fusions of such microparticles that representationally constitute the object X. Hence vagueness considerations, which essentially use a compositional divide-and-conquer strategy to pose problems for material constitution analyses, simply do not apply to such holistic sets.

Nevertheless, at this point it is prudent to accommodate those who would insist on a metaphysically purely realistic and naturalistic analysis of constitution. Such a stance might involve the following two restrictions: a) co-location must be defined as precisely as possible, no loose overlaps of any kind can be allowed; and b) only those microparticle sets, all of whose members are nomically or causally relevant items with respect to a material object X can count as constituting X. Or in other words, a robustly realistic account of constitution should, on this view, involve only tightly linked sets of microparticles and objects--linked both spatio-temporally, and causally or nomically.

Critics adopting this point of view likely would then claim that vagueness considerations again become relevant, even though it is now at the microphysical rather than the material object level. Their claim might be that there will be some precise microparticles sets that do constitute an object X, others that do not, e.g. because of too much looseness in overlap, plus an all-important third category of sets for which it is indeterminate whether they constitute X or not, whether from indeterminacy in overlap or causal/nomic indeterminacy factors. Hence, such critics might claim, vagueness considerations have been reinstated in full force.

However, even if the existence of all three categories of sets of microparticles were conceded, the following five points should serve to show that the metaphysical and epistemic threats posed by vagueness issues potentially can be considerably defused, if not completely eliminated, by the representational account. First, our standard concepts of sortals, or kinds of objects, are almost exclusively oriented toward characterizing ordinary midsized material objects, such as cows or cars, rather than any relevant underlying microparticles. Banishing the explanatory burden of explaining vagueness in material objects to issues about vagueness in constitution by underlying microparticle sets is genuine progress, because such issues are Quinean 'don't-cares' for most semantic, epistemic and definitional purposes.

Second, vagueness issues are primarily a threat to intrinsically rather than relationally characterized entities. We may accept that any precise concept of an intrinsic kind should be such as to rule out the possibility of items for which it is indeterminate whether or not they are members of that intrinsically defined kind. However, there is no intrinsic vagueness whatsoever in those microparticle sets--if any--such that it is indeterminate whether or not they are related to a distinct item X by constituting it. The burden of proof is on the critics to show that such merely relational kinds of indeterminacy, if indeed they exist, are metaphysically threatening to anything like the same degree as intrinsic cases of vagueness would be.

Third, by shifting vagueness issues to the microphysical rather that the material-object level, the present account is able to reclaim the primarily ontological status of the concept of constitution. Intuitively speaking, all that is necessary for a fully objective material object X to exist is that there should be at least one precise microparticle set that positively does constitute the object X. Once its material reality--or at least, as ordinarily understood--has been thus constituted, its relations to any other microparticle sets, whether positive, indeterminate or negative, are irrelevant to its ontological status. So again, relational vagueness or indeterminacy issues at the microphysical level make no ontological difference with respect to material constitution as here representationally defined.

Fourth, whereas vagueness issues are threatening because of the all-or-nothing nature of existence, representation permits either comprehensive or non-comprehensive representation of an item X. Thus, translated into representational terms, the charge of critics would be that there could be microphysical sets for which it was indeterminate whether or not they comprehensively represent X. But even if this were conceded, the difference between fully comprehensive, versus minimally non-comprehensive, cases of representation is so slight that indeterminacy as to which category a given case falls into is not clearly problematic. On the face of it, a representational approach can replace the all-or-nothing categorizations of standard accounts with an account of 'graceful degradation' in constitution, as a series of microphysical sets gradually represent less and less of the properties of the constituted item--or represent all of them in less complete detail. Again, the burden would fall on critics to show exactly why, and in what specific respects, this would be a serious problem.

Fifth, hard-core physicalists are free to take the following line: a purely physicalist ontology of microparticles has no purely scientific explanatory need to postulate the genuine physical, ontologically independent existence of midsized material objects at all, and consequently it has no need to recognize a representational relation of constitution either. So physicalist purists can, if they wish, simply forgo the two-tier ontology defended here, and hold--in their own rigidly extreme sense--an eliminative claim that there are no midsized material objects. Hence it cannot be a fundamental problem for physicalism that, if it were to recognize the genuine existence of such midsized objects, then vagueness issues might arise about sets of microparticles representationally constituting those objects. Hence, to sum up, vagueness issues are a problem neither for a commonsense ontology of midsized objects, nor for a robust naturalist ontology of material objects, and nor either for an austere physicalist ontology, which has no need or desire to recognize such representationally constituted material objects. Hence vagueness issues are not an ontological problem at all, if we adopt the recommended representationalist account of constitution.

As a final remark on vagueness, though the account given here is in some respects a radical one, nevertheless it does conform to one standard view, according to which all vagueness is broadly semantic rather than ontological.[12] For on the present view, vagueness issues are issues about the representational powers of sets of microparticles, just as, on such a standard view, they are issues about related representational powers of linguistic items such as sentences. So the present account is still recognizably about vagueness as ordinarily understood, and hence the potential solutions provided here to vagueness problems are genuine solutions.

5. Representation, Reference and Identity

There still remain some significant barriers to an acceptance of the current RC (representational constitution) view, which will be addressed in this section. These barriers cluster round the concepts of identity, objectivity and reference as applied to the special sciences. Here is an initial sketch as to how these problems could be resolved. Originally physicalists had hoped to reduce the special sciences to physics, and hence validate their objectivity in that way. However, it is widely agreed that that approach has failed, such as because of the distinctively different identity-criteria for sets of microparticles versus those for midsized objects. Nevertheless, I would claim, the broader strategy of which reductive strategies are a species are the right way for physicalists to approach the problem. This broader strategy is that of explaining the ontological claims of the special sciences as being in some way--though not necessarily always reductively--dependent on purely physical ontological claims.

The minimum demands as to objectivity and reference for such a broader physicalist strategy could be expressed as follows. First, any genuinely scientific reference to a putatively objective and real worldly entity X, such as a midsized object, must be, or include, reference to purely physical entities. Reductive approaches, when successful, automatically satisfy that requirement, but I shall show that there is another non-reductive way to satisfy the requirement for midsized objects X. By meeting that requirement, midsized objects can achieve at least a minimal objective status as objects of reference. And second, all of the causal powers of such putative objects X must be causal powers of purely physical entities.[13]

My solution to the relevant problems is a broadly conceptualist one. What is needed is a way to drive a wedge between reference to and identity of objective entities, so that entities having distinct identity-conditions could nevertheless be such that a reference to one is also always a reference to the other. Or, to put the matter in another way, we must distinguish talk about entities A and B, which talk is about putatively distinct entities having distinct identity-conditions, from references to A and B, which physicalists require to necessarily be, or to always include, references to purely physical entities.

Here is how this distinction might arise in a broadly physicalist and naturalistic framework. If midsized objects are to have any objective status, then observation or perception of them must be, or include, observation of sets of microparticles. Or, otherwise put, concepts of observational or perceptual reference of any kind, including of observation of midsized objects, must involve in some way references to sets of microparticles. However, at the level of scientific theory, or of cognitive conceptual interpretation, data derived from such observations could be further differentiated in various ways, depending on scientific interests, or cognitive evolutionary fitness, and so on. Thus conceptualization of such observational data as being about distinct entities--depending on the interests of the observer, such as whether she is carrying out the observation as part of a physics experiment--is consistent with the relevant observational references always having been, or having included, references to microparticles.

Thus, in place of the original physicalist hope of a purely ontological reduction of one genuinely existing class of entities to another purely physical class, this suggested conceptualist alternative claims the following. First, ontologically speaking in the most austere terms, there only exist microphysical entities, so that there is no additional class of purely physical entities, such as midsized objects, about which problems could be raised as to whether they can be ontologically reduced to microphysical entities. But second, conceptual processing of information about observation of such microphysical entities, or of perceptual reference to them, may legitimately differentiate that data in interest-relative ways, which interests can include the legitimate interests of the special sciences. Thus geologists can legitimately talk about mountains, or biologists about organisms, even though perceptual or observational worldly reference by such scientists must necessarily involve reference to worldly physical sets of microparticles.

The distinction between conceptual aboutness versus observational reference will now be put to work. My claim is that, as long as every observational reference is, or includes, a reference to appropriate sets of microparticles, it can at the same time also be a reference to a mountain, or a biological organism, and so on, as comprehensively represented by those microparticles. As a simple example, suppose I make an ostensive reference to my computer monitor by pointing at it. If physicalism is correct, I must be pointing at some sets of microparticles in so doing. However, there is nothing to prevent that same ostensive reference also being a reference to a midsized object that is comprehensively represented by those particles, namely my computer monitor. This would be true just in case a) the relevant sets of microparticles representationally constitute the computer monitor, and b) my ostensive reference to the monitor, in thus referring to those microparticles, is successful in virtue of computer monitor-related concepts and representational capacities that I possess. Hence it can both be true that the monitor, in purely physical terms, is 'nothing over and above' those microparticles, while at the same time my act of ostensive reference to the particles thereby identifies a midsized object having identity-conditions distinct from those of the microparticles themselves.

Another way to put the point being made is as follows. Cases of perceptual or observational reference involve broadly causal processes of interaction between perceptual or scientific-instrument systems and worldly items. But such interactive, extensional kinds of reference are at too crude or unconceptualized a level for distinctions between microparticles versus midsized objects to become relevant. If I touch my monitor in ostensively referring to it, I indifferently touch either the monitor, or the microparticle sets, or both. Hence it is ontologically harmless to claim that genuine observational reference to midsized objects is possible, since that claim is compatible even with the strictest kind of physicalist account of the relevant interactions between humans and the world.

So to summarize, since midsized objects both have identity conditions distinct from those for sets of microparticles, and since genuine observational references to them can be made as just outlined, there is no threat to the current RC (representational constitution) analysis of constitution from the reference versus identity-conditions distinction being made, since the relation of constitution does indeed hold between distinct entities having different identity-conditions, both of which entities can be observationally referred to.

As for the additional causal requirement given above on physicalist explanations--that all causal properties of any entities must be purely physical properties--it is clear that midsized material objects in general cannot meet that requirement, because of their identity-conditions. For example, all such objects, whether mountains, organisms etc., are such that they can lose or gain some atoms while remaining the same object. Animals can eat, and hence increase their mass, while winds can carry away surface dust from mountains and hence reduce their mass. However, the causal power or property of being capable of having a variable mass cannot be a purely physical property, for the following reasons.

When atoms combine to form a molecule, the causal powers of the molecule are distinct from those of each of those atoms considered individually--e.g. the molecule has a greater mass than any of its atoms--and hence the compound is an entity distinct from each of its composing atoms. Similarly, a purely physicalist view of a mountain would be of an aggregate of molecules such that, every time the composition of the aggregate changes, the aggregate itself changes into, or is replaced by, a distinct aggregate that has distinct causal powers from those of the previous aggregates. It follows that the concept of a self-identical mountain, that nevertheless is capable of changing its composition and hence its mass, is not a physically acceptable concept, and hence austere physicalists must eliminate mountains, organisms etc as such from their purely physical ontology. Nevertheless, this does not prevent the rest of us from adopting the two-tier ontology defended here, which provides an ontologically secure, even though conceptually and recognitionally based, foundation for midsized objects.

6. The Perceptual Mechanisms Underpinning Recognition of Changing Objects.

Section 5 showed that genuinely physical, self-identical objects could not have a variable mass, or vary in any other causally relevant properties. But if possible, as an ancillary task, it would be very desirable to have an explanation as to exactly how observation or perception is able to produce the cognitive impression of there being midsized objects which can change their properties, while yet preserving their identity--and why it is specifically representation that is involved in the cognitive process. As for the issue of why we do this, it will be argued below that the relevant representational procedure is cognitively unavoidable during normal perceptual processing. At the same time, in broader terms, it may be assumed that the procedure, along with the perceptual processes that support it, is evolutionarily fitness-enhancing for species which adopt both, though that specific issue will not be further investigated here.

The basic idea to be used is as follows. I have argued elsewhere[14] that perceptual processing of any kind is a cognitively demanding task, because raw sensory data combines information both about physically real items, and about the causal environment in which those items and perceivers are located. First, the raw sensory shapes produced on the retina by items will be different for different viewing angles of the same item. Second, their raw sensory apparent color is a function both of their actual color, and of the environmental or aspectual lighting conditions which illuminate their surfaces. But since arguably shapes and colors exhaust the content of raw sensory data, all such data is ambiguous in this two-factor way.

Hence my claim is that correspondingly, perceptual processing of such item-related raw sensory data must inevitably involve two distinctive kinds of content. Somehow perception must cognitively separate out the environmental or aspectual content from the item-related content, since each makes its own distinctive contribution to raw sensory data. Even if we have no interest in environmental or aspectual factors as such, we cannot obtain reliable item-related information without perceptual processes that can, in some reliable way, separate out such item-related information from the irrelevant aspectual information.

Now so far, we have only been discussing internal processing of information about real worldly items such as sets of microparticles. The processing of item-related information is clearly representational, since it involves separating out informational content about the worldly items from information about their environment, but there is no suggestion so far that the worldly items themselves have any representational capacities which are exploited during perception of those items.

However, two examples will now be given, that will show how, from a perceptual point of view, a perceptual system could not easily distinguish the above task--namely that of separating out two categories of information about two genuinely physically distinct kinds of worldly entities--from another task of distinguishing physically distinct worldly items that happen to be related.

Here is the first example. A perceptual system must, in its efforts to separate out item-related content from aspectual content, be able to deal with common cases, such as when an item X is partially obscured by something in front of it. Or in other words, the item content extraction system--or item recognition system--must be flexible enough to be able to recognize the same X even though parts of X are not visible. However, suppose that the obscured part of X is then physically removed from X. Naturally, the recognition system will still identify the new 'X minus' entity--X minus the removed part--as being X, even though strictly this is now false. Indeed, the remaining chunk of X, if large enough, will still be identified as X, even when it is moved into full view, because of the same built-in flexibility of the recognition system.

Thus the perceptually based judgment that the object X still survives, even though it has undergone change by losing one of its parts, is, as far as perception alone goes, based on nothing more than the required flexibility of recognition needed to recognize unchanged items under widely varying worldly circumstances. Indeed, arguably our whole folk ontology of changing but enduring objects is based on nothing more than such inevitable by-products of evolutionarily required kinds of perceptual flexibility.

Next, as to why representation by worldly items must be involved in such cases. Recall from section 1 the analogy of standard representations, such as a painting of a mountain. Arguably perception of such paintings triggers the same recognitional capacities as would be triggered by actual mountains. In such cases, the relevant perception has the mountain-related content that it does because the physical surface of the painting represents the mountain--which, recall, need not have any independent existence. Now my claim is that any cases in which such non-standard or non-veridical kinds of recognitional triggering occur will have exactly the same kind of explanatory structure, in which the perceptual triggering is explained by the worldly item functioning to represent, or functioning as a representation of, what the person perceives. Or in other words, it is legitimate to explain such cases in terms of one standard and well-supported view of representation by physical items X, namely that physical item X represents Y just in case perception of the visible physical surface configurations of X triggers Y-related recognitional capacities in the perceiver.[15]

Here now is the second example, of what could be called 'serial recognition'. It is often assumed that personal identity, and other such controversial cases of identity, must have some deep metaphysical underpinning that yet remains to be discovered. However, perceptual recognition cases strongly suggest that such cases can mainly be explained in relatively superficial cognitive terms, as follows. The parents, relatives etc. of a new-born baby X will, from day to day, recognize the growing person as being the same X, because their recognitional basis for identification of X is based on their recent perceptual experiences with X. The perceptual flexibility already discussed will take care of any changes as the baby grows. Now, over time, that recognitional basis shifts, as X grows up, but at any time X will still be thus recognizable in terms of the parents' current recognitional basis. Thus, even if it is true that every seven years or so, most of the cells in a human body are completely replaced, the parents will continue to identify the current physical group of cells as being X--even though those cells only comprehensively represent X--because of the continuously updated recognitional triggering that such groups of cells serially provide to the parents. It is only when a drastic change of state occurs, such as if the current group of cells were to die, that they become incapable of causing X-related recognitional triggering in perceivers of them, and hence cease to represent X. Thus metaphysicians can have a perceptually based, deflationary account of personal identity in purely recognitional and representational terms.

7. Summary

Because of the synoptic scope of this paper, a brief summary may be in order. The basic idea on which the paper is based is that material constitution is conceptually dependent on human perceptual mechanisms and recognitional abilities. As discussed in section 6, the necessary flexibility required to recognize the same worldly physical item in widely varying circumstances has, as an inevitable result, a corresponding identification of related but different physical items as being the same midsized, changing object. Arguably, the best account we have available as to how this process works is a representational one, according to which such distinct physical items X each comprehensively represent the same midsized object Y, just in case perception of the visible physical surface configurations of each item X triggers Y-related recognitional capacities in the perceiver. As a consequence, on the present account, such related physical items X constitute a midsized object Y by comprehensively representing Y.

Also this account is, strictly speaking, an ontologically dependent one, in that perception of the relevant midsized object Y is ontologically dependent on perception of the relevant physical microparticles X that represent Y. Nevertheless, since constitution cases involve, on the present account, comprehensive representation of the properties of ordinary objects--so that potentially complete information about object Y is provided by microparticles X--the ontological integrity of midsized objects as potentially providing epistemically complete information about themselves to observers of them is preserved, in spite of their not being strictly reducible to purely physical entities.

More broadly, the paper has argued that an adequate account of constitution must somehow simultaneously satisfy demands from no less than four different areas of inquiry: analytic ontology, philosophy of mind issues about non-reductive physicalism, evolutionary theory plus cognitive science, and wider metaphysical issues about realism versus anti-realism as well. The arguments given in the paper have attempted to address some issues in all of these areas, though clearly the brief arguments given here are primarily arguments as to the initial viability or promise of such an integrative research program, rather than providing fully defended, finished results.

Nevertheless, some initial results are at least suggestive. The current RC (representational constitution) approach offers a novel way to deal with 'the problem of the many' (section 3), explain identity through change in a deflationary way (section 6), potentially defuse vagueness issues (section 4), and generally to provide a plausible account of how the special sciences relate to physics. More broadly, a moderate realist, partly conceptualist position is espoused, which could provide a convincing way for physicalists to advance their cause, while yet acknowledging, in a non-damaging way, some of the insights of non-reductive physicalist views.

References

Bickle, John 1998. Psychoneural Reduction. Cambridge, MA: MIT Press.

Cummins, Robert 1996. Representations, Targets and Attitudes. Cambridge, MA: MIT Press.

Fodor, Jerry 1975. The Language of Thought. New York: Thomas Crowell.

Geach, Peter 1980. Reference and Generality 3rd Edn. Ithaca, NY: Cornell University Press.

Gillett, Carl & Barry Loewer 2001. Physicalism and its Discontents. Cambridge: Cambridge University Press.

Lopes, Dominic 1996. Understanding Pictures. Oxford: Oxford University Press.

McDowell, John 1994. Mind and World. Cambridge, MA: Harvard University Press.

Putnam, Hilary 1981. Reason, Truth and History. Cambridge: Cambridge University Press.

Rea, Michael 1997. Material Constitution. New York: Rowman and Littlefield.

Schier, Flint 1986. Deeper into Pictures: An Essay on Pictorial Representation. New York: Cambridge University Press.

Shoemaker, Sydney 2001. Realization and Mental Causation. In Gillett&Loewer 2001: 74-98.

Unger, Peter 1980. The Problem of the Many. Midwest Studies in Philosophy 5: 411-467.

Weatherson, Brian 2003. Many Many Problems. Philosophical Quarterly 53: 481-501.

Weatherson, Brian 2004. The Problem of the Many. The Stanford Encyclopedia of Philosophy (Fall 2004 Edition), Edward N. Zalta (ed.), URL = <http://plato.stanford.edu/archives/fall2004/entries/problem-of-many/>.

Wilson, Jessica 1999. How Superduper Does a Physicalist Supervenience Need to Be? Philosophical Quarterly 49: 33-52.

Notes