2. The Manifest Image of the Human Self

In order to explain the threat compositionalism in contemporary metaphysics poses to the manifest image of the human self we need first to say something about which aspects of the manifest image concern the human self.

Recall that the content of the manifest image consists in those propositions that would result from a pre-theoretical articulation of how things appear to us, unprejudiced by opinion about unobservable posits. The question now is which such propositions concern the human self. A sceptic might reply that no such propositions concern the human self: the so-called self is an obscure theoretical posit of metaphysics, any conception of which must be deeply historically conditioned by theology, politics, and other sociological forces. Even if there is such a thing as the self, its existence and character can hardly be read off from theoretically unprejudiced reflection on how things immediately appear to us, as one might describe the apparent colour of surfaces in one's immediate environment.

This scepticism deserves more attention that I can give it here but I believe it can be sidelined for present purposes by appeal to a thin, or minimal conception of the human self. In deference to the reflexive etymology of the word, let us use ‘self’ simply to pick out a potential reference of a use of the first person pronoun or other device of reflexive reference, abstracting away from the word's richer connotations of autonomy, ultimate value, immateriality, integrity, and so on.

Now on this minimal understanding there could be many as many kinds of selves as there are kinds of potential referents of the first person: there could be Martian, angelic, or robotic selves if such things could reflexively refer. So let us say that a more specifically human self is a potential referent of first person reference that is, as it were, local to us. A human self is a referent of ‘I’ related in some close way to a human body. Given that uses of ‘I’ are referentially coordinated with uses of ‘you’ and ‘he’ – my use of ‘I’ co-referring with your use of ‘you’ in conversation with me, and with an onlooker's use of ‘he’ – on the minimal conception of the self the content of the manifest image of the human self will simply consist in the propositions we would articulate, through theoretically unprejudiced reflection on appearances, by means of personal pronouns such as these.

It remains a difficult task to specify this class of propositions in any less general terms. As the lively debate about the richness of the content of experience attests, it is not always obvious what is obvious. Still, our readiness to use personal pronouns in response to experience of each other and ourselves strongly suggests at least a couple of basic elements of the manifest image of the human self. Where ‘here’ refers to the place I now seem to be occupying (a small desk chair as it happens):

These are not theory-laden attributions of controversial properties like freedom, or ultimate value. These simply concern the number or distribution of things we talk about with the ordinary use of personal vocabulary. These two claims plausibly articulate something central and obvious about our first-person perspective on ourselves – and also about our second- or third-personal perspectives on one another. Not only does it seem to me that there is just one human self in this chair, it would also seem to you that there is one human self in this chair.

3. The Threats of Compositionalism

On a compositionalist view of ordinary objects the metaphysically basic characterization of an ordinary object specifies the manner in which it is built up from its parts. Applied to human selves, the view is that a human self is something built up from things in a certain way, from things arranged human-wise. How does this metaphysics of the human self jeopardize the manifest image of the human self?

The threat to the manifest image of the human self arises from reflection on the conditions under which things are generated or built up from parts. Consider by analogy the case of sets. Suppose that we have three given objects: an apple, a banana, and a cherry. Let us ask, which sets may be formed with only these pieces of fruit as members? As a heuristic we can identify these sets in turn by means of imagined acts of ‘collecting’ or ‘gathering’. First one can gather up each piece of fruit singly – so there are the three singleton sets {a}, {b}, {c} – and one can gather them up in pairs – so we get three doubleton sets {a, b}, {a, c}, {b, c} – and finally we can gather all three together to make the three membered set {a, b, c}. So there are seven sets with only these pieces of fruit as members.

But now suppose an interlocutor takes the following contrary view. There are in fact only six sets with only the given fruit as members. While one can gather things up in every other way, one cannot gather up the banana and cherry to make the doubleton set {b, c}.

In response to this odd proposal one will request some explanation for the otherwise arbitrary sounding restriction on set formation. Suppose the request is answered with the assertion that set formation is restricted in this way because banana and cherry make a uniquely repulsive flavour combination (at least when unmitigated by the addition of further flavours).

This explanation seems implausible. Why? It seems implausible because the relation of co-palatability is intuitively irrelevant to whether the operation of set-formation has application to certain things. There is nothing in the nature of the operation that makes sense of this putative restriction in its application. A lack co-palatability among things does not seem to be the right sort of factor to ‘block’ the act of gathering them up into a set.

A conventional view is more plausible: if we are going to accept the existence of any entities corresponding to these notional acts of gathering then we should accept the existence of every logically possible result of such acts of gathering. That is to say, the only restrictions on set formation should relate to the avoidance of contradiction.

Now what about ordinary objects? Of course the usual compositionalist view is not that ordinary objects are sets. They have parts rather than members in the set-theoretic sense, and there is no ‘empty’ ordinary object, as there might be an empty set corresponding to the act of gathering up nothing at all. A more detailed theory could identify ordinary objects with the results of classical mereological summing, or yet some other composition operation on given things.Footnote ⁹ Still, on any such view we can ask under which conditions such things are formed, under which conditions the relevant composition operation has application.

Once again, a theoretically non-arbitrary view seems to be this: if we accept the existence of any entities corresponding to the composition operation then we should accept the existence of every logically possible application of the operation. This means that ordinary objects belong to a category that is extremely abundant in its instances. To use Lewis's notorious example, there is a composite resulting from the summing of an undetached front half of a trout in one location, and the undetached back half of a turkey in some distant location. Unlike cats or trees for example, these trout-turkeys do not normally attract our attention. But it would be arbitrary to deny that such composites are formed.Footnote ¹⁰

In pursuit of a more sparse conception of the category to which ordinary objects belong, various more restrictive views of composition have been proposed.Footnote ¹¹ Although this is not the place to defend the point fully it seems to me that any such proposal must share the basic defect of the co-palatability restriction on set-formation: it is always going to be difficult to see why the proposed restriction should have any relevance to the existence of composites (as opposed to aspects of their individual character, like salience, or integrity): it is mysterious why a composition operation on things should be sensitive to whether the putative inputs to the operation are fastened together like the parts of a material unity, or caught up in the life like the parts of a living organism. There is nothing in the nature of a composition operation as such to make sense of any such non-logical restriction, or block. To speak metaphorically, how could the operation know in advance to proceed in its application to some things only if the further thing it would go on to generate has the sort of individual character stereotypical of the objects we commonsensically recognize?

The abundant view is theoretically simple, then, but it arguably threatens our manifest image of the world in more than one way. First, we are surely ordinarily inclined to reject the claim that there are, overlapping with me in my chair right now, a massive number of kinds of macroscopic physical object of the same general category as the objects we ordinarily recognize. Second, the abundant view also jeopardizes, for any given ordinary kind of object, the claim that there are only so many instances of the kind to a region. Why? Here is a pertinent example: observe that there are millions of massively overlapping pluralities of atoms in my chair right now, differing by just a single atom. Millions of these pluralities are as properly ‘arranged human-wise’ as any other. But if the operation of composition is not restricted then it should have application respectively to each of these pluralities, generating a human self for each plurality. Thus the theoretically non-arbitrary abundant view of composition not only generates many more kinds of object than we would ordinarily countenance, it also generates a vast multiplicity of human selves in my chair right now.Footnote ¹² This is the first threat to the manifest image of the human self: a theoretically non-arbitrary view of composition threatens Sparseness.

Abundance is not the only view of composition that refuses to draw an arbitrary theoretical line. One might agree that if there are any entities corresponding to a composition operation then there are entities corresponding to every logically possible application of the operation. But, in the spirit of nominalism about abstract objects, one could deny the antecedent: one could take the view that the most economical view of reality should dispense with the entire shadowy host of massively overlapping entities putatively existing in addition to the given, non-composite, fundamental building blocks of reality.Footnote ¹³ This eliminativist view of composites is as theoretically elegant as abundance but it appears to be equally incompatible with an element of the manifest image of the human self. It threatens Existence, the claim that there is at least one human self in this chair. If human selves are ordinary objects if they are anything at all, ordinary objects are composites if they are anything at all, but there are no composites, then there is not even one human self in this chair.

So the two theoretically non-arbitrary views of composition, abundance and eliminativism, jeopardize Sparseness and Existence respectively. What are the options for those inclined to defend the manifest image of the human self?

One option would be to make a special exception for human selves. It might be claimed that the human self is not, like a cat, table, or tree, an ordinary material object. It is a more exotic kind of object, perhaps an immaterial soul as traditionally conceived.Footnote ¹⁴ I shall not pursue this option here, for two reasons. First, the option ignites a range of traditional worries about dualism, for example about psychophysical interaction. The second reason is this. Although there is perhaps a special Cartesian introspective appeal to the two claims about the existence and sparseness of human selves, the immaterialist option leaves untouched the compositionalist threat to parallel, but also compelling, claims about the existence and sparseness of ordinary objects, like cats, tables and trees. The idea that there are billions of cats, or tables, or trees where there seems to be just one present – or no cats, tables or trees at all – is not much less offensive to the manifest image of things than eliminativism or abundance about human selves. A more general response would be preferable.

Another response of course is to concoct a restricted view of composition, on which composites are generated under conditions corresponding roughly to the manifest image of the population of ordinary objects. It might be insisted that the fact that this position is arbitrary from the point of view of theoretical reflection on the nature of composition is just too bad. Manifest appearances defeat theoretical virtue.

However, this position is precarious because it is not clear that manifest appearances should be treated at unshakeable bedrock in this way. Of course they are our starting point, but when theoretically highly elegant views of composition have come into consideration it is not clear that the evidential force of appearances is not thereby rebutted or undercut.Footnote ¹⁵

It is a difficult question when, if at all, the theoretical virtues of a metaphysical theory defeat apparent counterevidence, and it is a large question beyond the scope of this paper. But whichever way we go on the question, a compositionalist view of ordinary objects will be stuck with an intellectual tension: we seem not to be able to find a position that simultaneously respects manifest appearances and theoretical non-arbitrariness.

4. The Nomological Conception of Ordinary Objects

I want to suggest a way forward. We can save manifest appearances, without denying that the human self is an ordinary object, and without denying either of the two theoretically non-arbitrary views of composition. The trick is to deny the presupposition that ordinary objects are composites, that ordinary objects are things fundamentally characterized in terns of their generation from parts.

We may begin by noting that it has become commonplace in recent metaphysics to draw a certain distinction among the truths about a given kind of thing. One can distinguish between those truths that reveal what things of the kind fundamentally are, and those truths that are, as it were, incidentally true – even necessarily true – of things of the kind. For example it is necessarily true that water is H₂0, and it is also necessarily true that water is such that 2 + 2 = 4. But intuitively only the former necessary truth tells us what water is.Footnote ¹⁶

Now it is true, perhaps necessarily true, that macroscopic objects of ordinary human experience have parts in various senses: trees have trunks, cats have claws, whiskers, front halves and back halves; inspect the paths they occupy more closely and one will find trajectories of cells, molecules, atoms, and even smaller constituents. But it is optional to regard these truths as the metaphysically basic truths about these kinds of objects. It is optional to explain what ordinary objects fundamentally are in terms of their composition from parts.

What is the alternative? To begin to see the alternative we should ask why it is that things like cats, trees, and humans attract our attention and interest. One could easily get the impression from reading contemporary metaphysics that we humans have simply found ourselves implanted with a random set of ‘intellectual intuitions’ about the existence of a certain restricted menu of kinds. But in fact the ordinary objects that we latch onto in perception and thought have a distinctive character: they are robust, stable, invariants, which enable prediction, manipulation, explanation, and counterfactual-supporting generalization at higher scales than fundamental physics. In other words, corresponding to the ordinary kinds that we recognize are an indefinite range of true lawlike generalizations, of biology, mechanics, folk-psychology, geology, and other special sciences. These true lawlike generalizations mention the individual activity and properties of the objects in question, rather than their composition from parts.

The proposal I am recommending is to treat these truths, lawlike truths to the effect that Ks ϕ in conditions C … rather than compositional truths to the effect that Ks have such and such parts structured in such and such ways… as the truths which reveal what Ks most fundamentally are. The metaphysically basic characterization of an ordinary kind of object states the lawlike activity of things of that kind. Call this the nomological conception of ordinary objects.

I am contrasting the nomological conception of an ordinary object as fundamentally a subject of high-level lawlike activity characteristic of its kind, with the compositionalist view that an ordinary object is fundamentally the result of the kind-characteristic application of a composition operation to some things. Why prefer the nomological conception?

One appealing, though not decisive, feature of the view is its harmonization of the metaphysics and epistemology of the macroscopic world. We do not in fact discover the existence of high-level kinds by asking under what conditions smaller things generate larger things; rather we discover them figuring in their own right in strong, simple, counterfactual supporting generalizations at larger scales than physics. On the metaphysical view that these objects just are subjects of characteristic lawlike activity, this epistemology makes perfect sense.

Another advantage of the view is its capacity to sideline spurious puzzles that arise on a compositionalist view of ordinary objects. Take the so-called ‘grounding problem’ about coinciding objects like the statue and the lump of clay.Footnote ¹⁷ How, it is wondered, could two objects composed of the same physical parts structured in the same way at a time nevertheless differ in their counterfactual propensities and other properties? Indeed, on the view that an object just is something structured from parts in a certain way it can start to seem mysterious how there could be room here for two objects at all. On the nomological view that an ordinary object just is a locus of lawlike activity, we count two objects by counting two distinct paths of lawlike activity. There is one thing that would persist and continue to transmit momentum and so on through a flattening process, and one thing that would not. Conversely the latter individual would continue to transmit certain aesthetic and cultural effects through certain changes in material, where the former would not. There is nothing especially puzzling about the fact that two such paths could pass through a single plurality of low-level physical elements at a single time. The nomological generalizations definitive of the kinds in question feature in sciences at a higher level than fundamental physics. Indeed the possibility of coincident objects is arguably the flipside to multiple-realizability, another phenomenon that supports the irreducibility or ‘autonomy’ of high-level special science kinds.Footnote ¹⁸ Multiple-realizability: the possibility that Ks could be realized microphysically in different ways shows that to be a K is not just to be arranged microphysically in a certain way. Coincidence: the possibility that a K and an incompatible K* could both be microphysically arranged in the same way also shows that to be a K is not just to be arranged microphysically in a certain way.Footnote ¹⁹

For the rest of this section however I want to focus on the way in which the nomological conception of ordinary objects resolves the intellectual tension with which I closed the previous section. The puzzle of finding a view of composition that is both theoretically non-arbitrary and also consistent with the manifest image of a fairly sparse population of ordinary objects presupposes the usual compositionalist view that the ordinary objects that feature in the manifest image just are composites. If instead we suppose that these ordinary objects are fundamentally characterized not in terms of their generation from parts but instead in terms of their characteristic high-level lawlike activity then both eliminativist and abundant views of the population of composites are perfectly compatible with the manifest image of a fairly sparse population of ordinary objects. After all it is not obvious that the manifest image has anything whatsoever to say about the population of mereological sums, impure sets, Finean rigid embodiments and variable embodiments, or any other kind of thing defined in terms of composition broadly speaking. One might rather regard composites as quasi-abstract objects, plenitudinous or non-existent according to one's taste. What would be offensive to appearances is the view that there are billions of things of the same general category as cats, trees, and humans, overlapping me in my chair right now – or else that there are no members of the category at all. If the non-compositional, nomological, conception of ordinary objects is correct then the theoretically elegant views of composition do not have these consequences: a relatively sparse population of ordinary objects is compatible with the existence of billions or none of the objects of the distinct category of composite.

We can further illustrate the compatibilist nature of the nomological conception by reconsidering the ‘problem of the many’, which, recall, was generated by an abundant view of composition in combination with compositionalism about ordinary objects. On the view that a cat, say, just is the result of composition applied to a plurality of parts arranged cat-wise, we arrive at the bizarre conclusion that there are as many cats as there are pluralities of tiny parts arranged cat-wise i.e. very many cats indeed. What does the contrasting nomological view that an ordinary object is a locus of high-level lawlike activity say about this situation? On this view we are freed up to say there is a single object present whose nomological nature enables manipulation, prediction, explanation at a higher scale. As Dennett puts a similar idea, each of the slightly different pluralities of tiny parts is a fuzzy or ‘noisy’ presentation of numerically one and the same high-level predictive ‘real pattern’.Footnote ²⁰

There is a prima facie difficulty here. Let it be granted that a single high-level pattern is presented by each of the slightly different pluralities of tiny parts. It might be objected that a pattern is something universal, or type-like. But in that case one and the same high-level object would be present wherever tiny parts are arranged cat-wise. This would amount to a problem of too few cats: after all we ordinarily take it that cat-wise arranged pluralities of tiny parts sometimes do correspond to numerically distinct cats, viz. when they are in quite widely separated locations. A cat is not multiply located like a universal. On the other hand if we try to particularize the pattern by reference to particular pluralities of parts, counting token patterns by counting token pluralities of tiny parts presenting the universal pattern-type, then we are back to the problem of the many, for there are many such pluralities.

The nomological conception of an ordinary object as a locus or path of high-level nomological activity suggests a solution to this difficulty. Corresponding to each of the massively overlapping pluralities is exactly the same set of predictions and causal consequences for the instantiation of high-level special science properties along a spatiotemporal trajectory. Corresponding to each of the massively overlapping pluralities is the same raised chance, for a given subsequent region of time and space, that something will be breathing, purring, pouncing, mouse-killing, at that region. This contrasts with the case of two cat-wise arranged pluralities presenting spatially widely separated cats: we will find that corresponding to these pluralities are distinct predictive and causal consequences for characteristic high-level activity going on at given subsequent regions of time and space. While this is not the place for a more detailed development of the picture, the basic proposal is to count token ‘real patterns’ by distinct spatiotemporal paths of high-level lawlike activity. On the nomological view, then, multiple overlapping pluralities can correspond to a single ordinary object, in line with the manifest image.Footnote ²¹ We are free to hold in addition that the many overlapping cat-wise arranged pluralities correspond to many composites – many things, which, like sets, or mereological sums, are fundamentally characterized by applications of some compositional operation to the respective pluralities. But on the nomological view none of these things is a cat. So there is no clash with the manifest image of a relatively sparse population of cats.

That is a major advantage of the nomological view. But isn't compositionalism about ordinary objects the metaphysical view properly respectful of the scientific image? Isn't Sellars obviously right to say that ‘man as he appears to the theoretical physicist [is] a swirl of physical particles, forces, and fields’? In fact what Sellars says here is quite dubious. Man arguably does not appear to the theoretical physicist at all. The terms of theoretical physics denote the large-scale structure of the cosmos and the finest structure of matter. Man does not get a mention. It is philosophy, not physics, which makes the metaphysical claim that man just is a composite generated from the subject matter of physics. The sciences that do directly concern the ordinary objects of human experience are not fundamental physics but the high-level special sciences: biology, folk-psychology, anthropology, and so on. The nomological conception of ordinary objects as loci of activity governed by the laws of these high-level special sciences, and not compositionalism, is the metaphysical view most naturally suggested by the scientific image of such objects.

What about the arbitrariness worry for conveniently sparse ontologies of the macroscopic world? Recall the general worry about restricted views of composition: it seems theoretically arbitrary to suppose that the generation of composites should be restricted in a way that corresponds to the manifest image. Does a similar worry afflict the nomological conception of ordinary objects? No. If an ordinary object just is a locus of high-level nomological activity then there is nothing arbitrary about the view that the population of this category is relatively sparse. It is an empirically discoverable fact that there is a relatively sparse set of kinds that feature in the explanatory lawlike generalizations of the special sciences. Things like trout-turkeys with a bizarre scattered gerrymandered nature do not figure in the simple, strong, explanatory generalizations of the high-level special sciences.

Perhaps from the point of the view of the philosopher's armchair there remains something puzzling here. After all there seems to be no a priori scrutable reason for reality to have the sparse high-level nomological structure that it does (or indeed for reality to have any high-level nomological structure at all: as Fodor puts it bluntly – not himself expecting an answer any time soon – ‘why is there anything except physics?’Footnote ²² ) But this species of ‘arbitrariness’ is just an aspect of the contingency of the empirically given structure of the universe. The claim that there are in fact certain high-level laws rather than other conceivable such laws seems no more theoretically objectionable than the claim that the universe in fact has certain laws of physics, or the claim that the universe in fact had certain initial conditions. That's just the way of the empirical world.

5. The Personal Identity Debate

Let us review the situation. We identified two elements of the manifest image of the human self, Existence and Sparseness. Given a theoretically non-arbitrary view of composition – abundance or eliminativism – it is hard to hold onto these elements of the manifest image on the supposition that the human self is a composite. The proposed way forward is not to adopt the view that the human self is some extraordinary object, but instead to deny the general conception of ordinary objects as composites. The recommended alternative is to view ordinary objects as subjects of characteristic high-level lawlike activity. This position is compatible with either non-arbitrary view of the existence of composites.

So far, however, the discussion has proceeded at a general level, without close attention to the application of the nomological conception of ordinary objects to the case of human selves in particular. A convenient way to fill in the picture of the human self is to ask the questions characteristic of the traditional personal identity debate. This will occupy the remainder of the paper.

The personal identity debate is centrally concerned with two questions. First: what are we? Second: under what conditions do we persist? These questions can be understood as asking, in the material mode, about the nature and persistence conditions of whatever things we refer to using our personal pronouns. That is to say, they ask about the nature and persistence conditions of human selves.

First consider the nature of human selves. On the nomological conception of ordinary objects, human selves are fundamentally subjects of lawlike activity characteristic of their kind. The question now is which are the special sciences of human selves? Where are the relevant lawlike generalizations about the activity of human selves to be found?

In its way the position in the personal identity debate known as animalism suggests an answer to this question.Footnote ²³ On the animalist view that we are fundamentally biological organisms of a certain kind, the lawlike generalizations of biology might be thought to be those definitive of our nature. We human selves are fundamentally subjects of activity distinctively governed by the lawlike generalizations of biology: we are things that grow, metabolize, reproduce, and so on.

An opposing neo-Lockean psychological view of our nature can be seen as instead emphasizing folk psychology, or the disciplines of cognitive science.Footnote ²⁴ We human selves are fundamentally subjects of activity governed by the lawlike generalizations of these special sciences in particular.

However there is no evident rationale for singling out any one of these special sciences as especially definitive of human selves. I would recommend instead what might neutrally be called a naturalist position in the personal identity debate. According to this view, human selves, like many other ordinary objects, are fundamentally subjects of characteristic activity governed by laws from a whole range of the special sciences. The imagery of the university campus of separate buildings each housing a department with its own proprietary objects of study can obscure the fact that a single object may fall under the laws of multiple special sciences. For example artifacts like statues or tools are at once fruitful objects of study of material science, art history, perceptual psychology, and ethology. Likewise the human self is at once the fruitful object of study of biology, ecology, psychology, anthropology, sociology, history, and so on. This inclusive naturalism about the nature of human selves is the least arbitrary way of applying the general nomological conception of ordinary objects to the case of human selves.

However, recent contributors to the personal identity debate have tended to insist upon an exclusive decision between biological and psychological conceptions of our nature. Why? The answer probably has something to do with the seminal influence of the second question, the question about our persistence. Consider the following plausible claim about our persistence: each one of us was once a mindless fetus. It follows, observes the animalist against the neo-Lockean, that psychological continuity cannot be necessary for our persistence. On other hand it is plausible that each one of us could be whittled down to a thinking cerebrum in a vat, without the usual continuity of biological capacities for breathing, metabolism, etc. So, observes the neo-Lockean against the animalist in turn, biological continuity cannot be necessary for our persistence. One gets the impression, then, that intuitive verdicts about persistence force an awkward choice between a psychological and biological theory of our nature. One must take a side.

What does naturalism about human selves say about our persistence? If we think that a human self is fundamentally a subject of activity governed by the lawlike generalizations of a range of special sciences then it seem to me that a cluster theory is the obvious companion view about our persistence. On this view we associate with human selves a range, or cluster, of activities over time, a range of ways of transmitting high-level causal influence along a path, corresponding to the range of lawlike generalizations of the special sciences with application to human selves. This range will include, as subset, both psychological continuities and metabolic biological continuities. But there is no theoretical point in claiming, as the animalist and neo-Lockean both do, that one of these subsets is individually necessary for human persistence. Instead we can say that a human self persists just in case there is a sufficient continuity of characteristic activity along a path, without prejudice as to whether it is comes from the psychological or narrowly biological elements of the cluster. So this view is in fact compatible with both of the plausible verdicts about persistence just mentioned: we can take the view that in both the fetus and the cerebrum in a vat cases, enough of the cluster of lawlike activity definitive of human selves is preserved for these to count as cases of human persistence. In this respect naturalism is more attractive than both animalist and neo-Lockean views of our persistence.Footnote ²⁵

This is no more than a sketch of the naturalist view of the nature and persistence of human selves. But I hope to have said enough to indicate why its prospects are bright. The view promises to sidestep a debate between two dominant but problematically narrow positions in the debate about our nature and persistence. The view flows naturally from a general conception of ordinary objects. This general conception, as we have seen, preserves the manifest image without the arbitrariness that afflicts compositionalist attempts to do the same. It is in this theoretical framework that we should hope find a satisfying place for the self in metaphysics.