III. The Constitution View: Some Elaboration

Although Baker provides an explicit and elaborate definition of constitution three times in her book (PB, 43, 95, 168), rather than recount that here I want to summarize it and convey its flavor.³

Some object, x, constitutes some object, y, at a given time, t, just if four conditions hold: (i) x and y are spatially coincident at t; (ii) x exists in conditions necessary for things of y's kind to exist; (iii) if something, z, of x's kind, exists together with the circumstances required for something of y's kind to exist, then an instance of y's kind exists that is spatially coincident with z; (iv) it is possible for x to exist without there being anything of y's kind that is spatially coincident with x. Conditions (i) and (iii) identify why x and y are not simply separate entities, while (iv) specifies why they cannot be identified.

To make this a little more concrete, consider David and Piece at a given time. (Suppose David is on display at the Metropolitan Museum of Art in New York.) David is a statue. Piece is a piece of marble with a particular size, shape, color, and composition. David and Piece occupy precisely the same space at that time—so (i) is satisfied. However, neither statues nor pieces of marble are ubiquitous. Among the conditions necessary for statues to exist are certain kinds of human practices, institutions, and intentions. For example, practices of sculpting, institutions of artisan craftsmanship, and the intentions to produce particular, meaningful works of art are all conditions necessary for the production of statues in the past and present. Of course, as cultures shift, these conditions may change, and we might think that none of them is strictly necessary for statues to exist, as evidenced by the production of statues by machines and with the intention to make as much money as possible (artistic intention be damned). But the general point is that some such conditions must hold if there are to be statues, rather than merely pieces of marble, and in general these conditions concern the nature of the world beyond the physical boundaries that David and Piece share.

So suppose that Piece is in these circumstances—so that (ii) is satisfied. Given these circumstances, to say that David is constituted by Piece is to say that any time you have a piece of marble with the properties that Piece has—its shape, size, color, composition—then you also have a statue that is spatially coincident with that piece of marble—so that (iii) is satisfied. Yet precisely because these conditions are contingent and distinctively necessary for statues, it is possible simply for a piece of marble to exist, as alike to Piece qua piece of marble as you like, without there being a statue at all, let alone David—thus, (iv) is satisfied.

To summarize: Constitution exists just when there is (complete) spatial coincidence between two entities, together with a set of distinctively contingent conditions that are metaphysically sufficient for the presence of the object constituted, given the presence of the constituent object or something very much like it. The constitution view says that just this relationship exists between persons and their bodies. They are spatially coincident, and the conditions that, in some sense, need to be “added” to a mere physical body to create a person—those needed to create a strong first-person perspective—are both distinctively contingent and metaphysically sufficient for there being persons, given the existence of bodies.

IV. Constitution and Pluralism

In contrasting constitution with other metaphysical relations, particularly with mereological relationships that hold between a thing and its parts, Baker says, “constitution is construed unambiguously as a vertical relation” (PB, 182). By this, and in light of the examples she provides and concentrates on, I take this metaphor to tie with another, one especially common in the philosophy of mind: that of higher and lower levels. Constitution is an asymmetrical relation, one that relates objects described at one level to those described at another. Since the existence of a constituted entity requires that there be some set of conditions in addition to those necessary for the existence of a constituent entity, it is natural to think of the former as an instance of a higher-level kind, and the latter as an instance of a lower-level kind.

There is nothing in the constitution view, or in the full and explicit characterization of constitution, to rule out one entity's being constituted by many different entities, or one entity's constituting many different entities. In fact, it is important to my extension of the constitution view—and, I think, to its broader applicability—that constitution relations be many-many. To put this together with the appeal to constitution as a vertical relation and the talk of levels that it relies on, this is to say that there can be many lower-level constituents for any higher-level entity that has a constituent, and many higher-level entities that any lower-level entity constitutes.⁴

For example, a person is constituted by her body, but also by an aggregate of body parts (e.g., heart, lungs, stomach), a causal network of bodily systems (the circulatory, respiratory, and digestive systems), and a particular arrangement of elementary particles. Conversely, a given body can constitute not just a person, but also a living thing, a member of a particular species or genus, and a moral or rational agent. If these are real kinds of things, then a many-many constitution relation is poised to make sense of them.

Something similar is true of David and Piece. Piece constitutes David, but also constitutes a piece of art, a valuable artifact, and a work by Michelangelo. The conditions for the existence of each of these kinds of things are different from one another. Pieces of art need not be valuable artifacts, and although works by Michelangelo are both in the actual world, this need not have been the case—not least of all because Michelangelo might have died as a baby and never produced any art. More to the point for the constitution view, the spatial coincidence at a time of Piece and any one of these things is contingent, and the existence of a piece of marble like Piece in conditions sufficient for the existence of each of these kinds metaphysically suffices for the spatial coincidence of that piece of marble and the resultant piece of art, valuable artifact, or work by Michelangelo.

Conversely, David is constituted by Piece, but David is also constituted by an aggregate of the elementary particles on or inside Piece, and by the sum of the marble on the surface of Piece and all the marble inside the surface of Piece. Neither of these entities is strictly identical with Piece, for they exist embedded within the larger piece of marble from which Piece was sculpted, a condition in which Piece itself does not exist.

Baker herself would, I think, deny these particular claims about persons and bodies, and about David and Piece, for a reason that we will shortly see. But broader features of her view suggest ambivalence about the more general issue of whether the constitution view should countenance a many-many constitution relation.

On the one hand, Baker champions a form of ontological pluralism, according to which there are myriad kinds of things in the world. In addition to the examples we have already seen, there are many kinds of things that feature in Baker's own elaboration of the constitution view and the broader ontological view of which it is a central part. There are lumps of clay and aggregates of material particles (PB, 25), landscape paintings and carburetors (PB, 38), airliners, personal computers, anvils, and doorstops (PB, 41), deans (PB, 47), and coaches (PB, 51). On Baker's view, all of these things exist, and presumably each is a material object, and thus at least a candidate for subsumption under the constitution view. This ontological pluralism takes as its point of departure the “pragmatic realism” that has guided much of Baker's previous work, a realism that attempts to understand the everyday, lived world of objects, their properties, and their relations (PB, 22–24). If there are deans, persons, bodies, and aggregates of material particles in your ontology, and you are tempted by the thought that none of these entities is strictly identical to any other, then you might well turn to a many-many constitution relation to make sense of the relations between them. This would be one way to deflate the ontological extravagance of your pluralism (cf. persons and bodies).

On the other hand, Baker is clear that she thinks that there is a limit to just how ontologically extravagant the commitments of pragmatic realism are. This is manifest in her explication of a notion central to her explicit definition of constitution, that of a primary kind. She says:

Although Baker, along with everyone else, lacks a theory of primary kinds, she points to clear cases that help to mark the distinction between a change that creates a new primary kind and one that merely results in things of the same primary kind acquiring or losing properties. Being a husband, she claims, is not a primary-kind property, for a “world like ours except that it lacked the institution of marriage (and hence had no husbands) would not thereby have fewer individuals in it than our world” (PB, 40). An anvil can be used as a doorstop, but this is a case of one thing, an anvil, coming to acquire a property—rather than a new primary kind, doorstops, coming into existence (PB, 40–41).

Baker's general idea for distinguishing the two sorts of case is that it is only when we have the creation of “whole classes of causal properties” that we have a distinct primary kind. Thus, a person is constituted by her body, but a doorstop is not constituted by the anvil, because only in the former case do we have entities that belong to distinct primary kinds; only in the former case is there a whole class of causal properties that mark the existence of a distinct primary kind. Likewise, I suspect that Baker would see only the causal differences between persons and bodies—and not those either between persons, deans, and moral and rational agents, or between bodies, aggregates of particles, animals, and causal networks of bodily systems—as marking a boundary significant enough to call for distinct primary kinds.

While I share Baker's intuition that there is a distinction to be drawn between mere property change and entity creation, a proper understanding of the basis for that distinction seems to me unlikely to eliminate the need for a many-many constitution relation, either in general or in the case of persons and bodies (or, for that matter, in the case of David and Piece). We can (perhaps should) think of primary kinds as things bearing distinct causal clusters of properties and powers. But it is a central part of pragmatic realism not to privilege intrinsic properties and powers over relational properties, as Baker herself makes clear (PB, 24). So those clusters of powers can exist in part because of features of the circumstances that entities find themselves in. Statues, rivers, deans, coaches, landscape paintings, carburetors, doorstops, and anvils all possess distinctive causal powers, given not just possible circumstances but the world as it actually is. Thus, it is difficult to see a motivated way to rule them out, a priori as it were, from the set of primary kinds. Moreover, once you embrace the idea that there are many, many kinds of things, there seems little reason to insist that the special relation of constitution holds only between pairs of them. If one accepts the ontological pluralism implied by pragmatic realism, together with a generalized version of the constitution view, then one should view constitution as a many-many relation.

The existence of many-many constitution relations radicalizes the constitution view. For not only are there two things, person and body, related by constitution, where we might have thought there was one, but there are many things—person, rational agent, moral agent, bearer of mental properties, body, aggregate of body parts, living thing, causal network of bodily systems—that stand in a more complicated set of constitutive relationships to one another. Prima facie, at least any of the first four—person, rational agent, moral agent, bearer of mental properties—is constituted by at least any of the final four—body, aggregate of body parts, living thing, causal network of bodily systems. But this does not, obviously, exhaust the ontological relationships between instances of these kinds, which themselves are just examples of the myriad kinds of things that exist in a case like this. Call this the many-many problem for the constitution view.

Even if what we are, essentially, are persons, the relationship between persons and their bodies is no longer special in virtue of being unique. This is a welcome aspect of the constitution view, I want to suggest, since it further serves to defuse the binary opposition between entities that are mind-like (persons) and entities that are matter-like (bodies). But then we need some account of whether there is any reason to focus on the person-body relation, rather than any of the other constitutive relations that typically exist when we have a person. Two different (but not incompatible) responses to the many-many problem are suggested by my comments above.

The first tack, which I tend to favor, would be to embrace ontological pluralism wholeheartedly and supplement it with some kind of subjectivist or relativist account of why the categories of person and body might be thought to be special and deserving of distinct treatment. However wholehearted one's endorsement of pluralism here, clearly it needs to be tempered by a recognition of the distinction between entity creation and mere property change. Baker's own approach to marking this distinction, cast in terms of distinct causal clusters of powers, may be on the right track, but it seems to me unlikely to be sufficient in itself to complete the task. (For one, I suspect that an ineliminable reference to the duration of the cluster's existence will be necessary.)

A second response, which I suspect Baker would find more conducive, would be to attempt to limit even the appearance of ontological extravagance here by arguing that some of these entities—body, animal, and organism, for example, to take a trio that Baker herself equates—are not the names of distinct objects, while others—aggregate of body parts seems like a plausible candidate—are not governed by the constitution view because they are not themselves things or objects. The essential challenge that this sort of response must meet is to show that an ontologically less extravagant (but still robust) application of the constitution view to the kinds of things that exist in the world beyond Baker's stock examples (of persons and bodies, David and Piece) does not compromise the constitution view's position on such examples.

Both strategies of response to the many-many problem are likely to prove problematic in their own way, but here is not the place to articulate or defend that claim.⁵

Instead, I want to consider how we might frameshift the constitution view from persons and their constituents to what entities persons themselves might be thought to constitute.

V. Agency, the Mind, and Social Action

What seems to me clearly right about the constitution view, regardless of what one says about the many-many problem that I have touched on in the previous section, is that the category of person is important for thinking about the question “What am I?” Moreover, something like having a strong first-person perspective seems necessary to articulate that concept in a way that implies that human persons are special not just in being constituted by a human body, but in the sort of agency that persons, including human persons, manifest.

I should say something about how I think about agency.⁶

I take agents to be individual entities that are capable of acting in the world and that typically do so act. They are differential loci of actions. I am happy to be quite pluralistic about the kinds of agents there are in the world. There are physical agents, including elementary particles and atomic elements, everyday physical objects (such as tables and rocks), and larger and more distant objects (such as stars and tectonic plates). There are biological agents, such as proteins, genes, cells, organisms, demes, species, and clades. And there are social agents, including not only individual people, but also groups and collections of people, institutions, and perhaps even whole communities and nations. Agents are causes, but not mere causes, for they are individuals with some kind of physical boundary—or at least they are in nearly all of the examples I have cited above. (The exceptions here concern kinds of collective agents; I will have more to say about these in the next section.)

Even though human beings are paradigmatic agents in many respects, it is also true that the kind of agency they manifest is special in that it is either unique or shared with only a small minority of the agents that there are in the physical world. There are different ways to articulate what is special here. But all the ways that I know of presuppose that human agency goes hand-in-hand with a relatively rich mental life, whether it be acting on the basis of reasons (rather than mere causes), engaging in particular forms of reasoning (means-ends, deliberative, evaluative, inferential), or having a qualitatively distinct phenomenal life that makes a causal difference to how we react to and interact in the world. And although this may be slightly more controversial, I take it to be very plausible to think that however this idea of a rich mental life is spelled out, it consists in more than merely having mental episodes of some kind or other, or being causally governed by distinctively mental, internal states.

Given that, I do not intend to challenge the constitution view's characterization of a person as an entity that has, or has a capacity for, a strong first-person perspective. However, I do want to suggest that this capacity is insufficient for any kind of agency, including distinctively human agency. For this view is what we might call an input-based or internally-based view of persons in that it identifies persons independent of their reactions to or interactions with the world beyond their own boundary. It is not acting as an embodied agent with a (strong) first-person perspective that is crucial to being a person, according to Baker, but simply having (or even possessing the capacity for having) a first-person perspective. On this view, someone who could only register thoughts such as “I am here,” “I believe I am late,” or “I want to know more” would count as a person, provided that she could wonder about those thoughts, entertain their falsity, and attempt to explore their implications.

In his book Mental Reality,⁷

Galen Strawson defended the idea that a race of sentient, intelligent creatures, the Weather Watchers, who were individuals rooted in the ground and with a range of mental states, were possible, even though they lacked even the physiology necessary for action. This is mentation as pure registration, as reflection on internal states that are either simply registered or reflections on that registration, and part of Strawson's point in introducing the Weather Watchers was to argue that behavior had no conceptual connection to the concept of the mental. Despite Baker's own externalist sympathies in the philosophy of mind, her emphasis on the first-person perspective in her account of persons places her views here closer to Strawson's than one might have expected.⁸

Suppose that we were to concede a conception of the mind that is internalist in the way that Strawson's is, and even concede that such bearers of mental states are persons. My point here is that such individuals would not be agents. Unless they can at least see themselves as capable of acting in the world, of either adjusting their position with respect to it or modifying the world itself in some way, it is difficult to credit them with the sort of agency that creatures with our sort of rich mental life take for granted. If we wish to adapt the constitution view to apply to anything like the full range of human agency, then we need to go beyond the input- or internally-based conception of persons that it employs. This is just to recognize that despite the centrality of the concept of a person (so understood) to agency, there is more to human agency than being a person.

There are several ways in which one might read this very point into the constitution view itself. For example, human persons, perhaps unlike other persons (such as angels or God), are embodied, and bodies are just what provide for the possibility of human action in the world—action-traction, as we might call it. And we can see human agency as going beyond mere personhood in Baker's discussion of “the importance of being a person.” For part of its importance is that it allows one to be both a rational and a moral agent (PB, chap. 6), where at least the latter of these involves being subject to certain kinds of norms, some of which are presumably made possible only when a person has action-traction. Personhood, as characterized by the constitution view, is necessary for the sort of agency that humans possess, but it does not seem sufficient, because it is input- or internally-based.⁹

Being a person seems crucial not just to rational agency, but to the possibility of a fully-fledged social life. More specifically, being a person creates the possibility of such a social life, one that involves what we commonly call interpersonal relations. These social relations include those of care and friendship, of love and hate, of rivalry and jealousy, of compassion and empathy, of malice and spite, of forgiveness and kindness, of respect and admiration, of revenge and defiance. Or, to express this in a way that more accurately depicts what I consider the likely developmental and evolutionary story here, persons (on an input- or internally-based conception) and social agents coexist as nonidentical but mutually reinforcing kinds of being.

I implied earlier in this section that social agency should not be conceptualized simply in terms of the actions of individual agents but also in terms of what institutional and collective agents do. Banks can foreclose on your mortgage, city councils can raise your property taxes, and Her Majesty's government can request the pleasure of your company. As economic, political, and legal entities, each of these agents can bring about effects, sometimes effects that matter a great deal to us. They act through the agency of individual persons, to be sure, but it is only as a representative of a bank, a council, or Her Majesty's government that the acts of particular persons count as foreclosing our mortgage, raising our property taxes, or imprisoning us. I shall argue in what follows that the constitution view can be readily and fruitfully adapted to shed some light on social agency, including the collective forms that it takes, and its relationship to persons.

VI. The Constitution View and the Social Domain

Recall the many-many problem facing the constitution view of the relation between persons and bodies. If bodies constitute persons, then so too do many other things, such as aggregates of body parts, causal networks of bodily systems, and living things. And persons are not the only kind of thing that bodies constitute, for they also constitute (among other things) rational and moral agents, and individual social agents—bankers, tax assessors, and policemen, for example. The many-many problem suggests that the ontological pluralism entailed by the constitution view is extravagant, and the problem is to decide whether to embrace this extravagance or to find ways to deny that it is an implication of the view. Although the many-many problem is quite general in that it applies not just to the case of persons and their bodies but in principle to any case where a pair of things stand in a relationship of constitution, there are reasons to think that the problem will be acute when persons are one of the relata in the constitution relation.

This is especially the case where a person is not the constituted entity (as in the person-body relation) but the constituent entity. Thus far, I have not explicitly considered in detail any examples in which this is so, but I hope it does not come as a complete surprise to learn that I think that social agents and persons can be viewed as standing in a relation of constitution, much as persons and bodies can.

Consider, first and perhaps most contentiously, rational agents and persons. Essential to being a rational agent is not simply having means-ends reasoning, where ends are goals typically manifest as desires, but having the ability to evaluate both that reasoning and the behavior that it results in. Rational agents are deliberative. We might require more of rational agents—perhaps they need at least some emotional life, or to have a grasp of certain basic rules of inference, or to have some minimally coherent belief-desire set. But it is hard to see how we could require less, at least if we take ourselves as paradigms of such agents.

Suppose that we consider Kim, who is both a rational agent and a person. Intuitively, rational agency, at whose heart lies the notion of deliberation, is something distinct from, and more than simply having, a strong first-person perspective. Both rational agents and persons are reflective or contemplative, but rational agents have, in addition, some kind of action-traction that is an optional extra for persons, even embodied persons. Moreover, it is plausible to think that if there is this sort of difference between rational agents and persons, then there will be a large cluster of powers that rational agents have that mere persons lack, namely, all of those powers that concern how one can and does act in the world. Just as having a strong first-person perspective creates significant powers above those that merely mental creatures (those with a weak first-person perspective) have, so too with being a rational agent and being a person. Rational agent is a primary kind, and it is a different primary kind from person.

Here is another way to come at the difference between persons and rational agents. A person has a strong first-person perspective (by stipulation), but that perspective could be directed entirely at the “input” side of her mental life: at her perceptions, her beliefs, the grounds she has for these, the connections between them. A rational agent, by contrast, in addition to all of this higher-order mental life has an action-oriented, motivational dimension to what it is like to be that agent. She not only has goals and desires, but evaluates them and strives to achieve them, or engages with the world to increase the goodness of fit between actual states and goal states. Call this practical rational agency, if you like; my point is that it is something more than merely being a person.

To see that it is something more, consider several ways in which having a first-person perspective and being a rational agent can come apart in the actual world. Individuals who feel powerless to act can withdraw from the world, and can come to see themselves as incapable of exercising effective control over their own body. In extreme cases, they dissociate themselves from all action, a dissociation that is accompanied by depression, fear, and anxiety. Cases of dementia, of extreme memory loss, or other forms of what we call “mental breakdown” can also compromise rational agency short of compromising action itself. But part of what is so disturbing about both kinds of cases is that there remains a being with a first-person perspective, a person who has become unhinged from herself as a rational agent. In general, if we think that people can fail to be rational in ways extreme enough to justify viewing them as different kinds of beings—irrational or arational agents—but remain nonetheless persons, then rational agency is something distinct from mere personhood.

We can turn to the characterization of constitution recounted in Section III to make the point here in terms of one relationship that holds between persons and rational agents. Suppose that Person Kim exists in the circumstances favorable for the existence of rational agents: Person Kim has deliberative capacities, a motivational set of mental states that engage her with action, and a perspective on herself as an agent. Simplifying a little further than we have already, a relation of constitution exists between Rational Agent Kim and Person Kim just if (a) Person Kim and Rational Agent Kim are spatially coincident in some circumstances; (b) these circumstances suffice for any person to be spatially coincident with a rational agent; and (c) these conditions are distinctively contingent, such that Person Kim could exist even if Rational Agent Kim does not.¹⁰

Condition (a) should simply be granted. But the set of conditions that make for rational agency, above and beyond those that make for a strong first-person perspective, are deliberative and action-oriented, and involve both specific cognitive and motivational capacities and action-traction with the world. These conditions, and thus rational agency, could fail to exist even when there are creatures with a strong first-person perspective—and so (c) is satisfied. But given that they do exist, anything that is a person is also a rational agent—and so (b) is satisfied. Thus, rational agents are constituted by persons.

One response that would seem in keeping with Baker's own discussion in Persons and Bodies (especially chapter 3) would be to maintain that person and rational agent are not distinct primary kinds, or that any conditions sufficient for one are also sufficient for the other. The idea here is that although our concepts of person and rational agent are distinct, they travel hand-in-glove, such that there are no circumstances in which persons exist but rational agents do not. Thus, persons and rational agents do not satisfy condition (iv) in my earlier characterization of constitution, and so condition (c) above.

While there are certainly ways of conceiving of the two for which this would be true, this response is an option for Baker only if she is prepared to make one of two moves: either build more action-traction into the conception of a person than does the first-person perspective view, or adopt a view of rational agency that has looser ties to action than I have suggested it has. My hunch is that either move (but particularly the former) will reduce the extension of “person,” such that both very young and very old human beings will typically fail to be persons, as will others who suffer long-term or irreparable diminishment of motivational and action-oriented aspects to their mental lives.

The relationship between moral agents and persons seems to me less controversially constitutive. I take moral agency essentially to involve interpersonal phenomena. Certainly many of the beliefs, desires, emotions, and judgments that we make in the moral domain are other-regarding, and moral agents typically operate within a series of increasingly encompassing moral communities: family, kin, local communities, cities, and so on. Moral agency lies not just in having certain internal mental states but also in acting in ways that are subject to certain kinds of normative evaluations. Plausibly, the conditions necessary for moral agency include facts about groups and communities of people.

Consider Tim, both a moral agent and a person. Moral Agent Tim and Person Tim are spatially coincident at some time—thus, (a) is satisfied. But the conditions that make for the possibility of moral agency—the existence of other people, some sense of them as people or as agents, perhaps certain social conditions—are distinct from those for the existence of persons—thus, (c) is satisfied. And these conditions, when present, metaphysically suffice for anything that is a person, including Person Tim, to be a moral agent, including Moral Agent Tim—thus, (b) is satisfied. Thus, Moral Agent Tim is constituted by Person Tim.

Least controversial with respect to satisfying the “distinct existences” clause of the definition of constitution, are cases that involve social agents whose agency lies in the roles they play in certain institutional frameworks: the bankers, the tax assessors, and the policemen I have previously mentioned. Suppose that Diane is a banker. Are there two entities, Banker Diane and the person Diane, that stand in a relation of constitution? Provided that banker is, in Baker's terminology, a primary kind, it would seem so. For suppose that the person Diane exists in the circumstances necessary for a person to be a banker: there is the institution of banking, she is employed by a bank, she is paid for her services, etc. If Banker Diane exists, then surely she is spatially coincident with the person Diane—so (a) is satisfied. Now, if there is anything that is a person that exists in just those circumstances that Diane exists in, then that thing will also be spatially coincident with a banker—so (b) is satisfied. And the distinctive contingency of those conditions makes it possible for there to be persons but no bankers (as history shows)—so (c) is satisfied. Thus, Banker Diane is constituted by the person Diane, just as that person is constituted by (among other things) Diane's body.

The point here is quite general. It applies to agents defined in terms of their employment (bankers, presidents, miners), their family roles (father, sister, cousin), their living and social conditions (neighbor, volunteer, coach), and the economic and legal institutions they are subject to (debtor, criminal, citizen). The crucial issue, in each case, is whether each of these socially defined roles carries with it a distinctive cluster of causal properties and powers sufficient to warrant considering it a distinct primary kind. In keeping with my earlier remarks about ontological extravagance at the end of Section IV, I have no doubt that this is not always true. But the real issue seems to be whether it is ever true; for if it is, then it seems that at least certain kinds of individual social agents are constituted by persons. Certainly there are circumstances where occupying some of these social roles makes a massive difference to one's place in the causal net. Presidents of powerful countries can bring about massive changes in the world; being a father at least sometimes makes a large difference to what one cares about, how one spends one's time and money, and one's broader moral perspective; and there is a battery of statistics about criminals qua criminals, tracking causal regularities about limitations to their employment prospects, their likelihood of being imprisoned, and their longevity.

Thus, while I share the intuition that such forms of social agency need not be causally impactful enough to create entities of a new primary kind, there also seem to be cases where it is hard to deny that they meet this criterion. In each case, the constituent entity is a person or, recognizing constitution as a many-many relation, a person is one of the entities that constitute any individual social agent. Persons are important because lots of the kinds of things we are, the kinds of things in virtue of which we make a differential causal impact on the world, are constituted by persons in the very same sense articulated by the constitution view of persons.

One response we saw to the claim that persons constitute rational agents claimed that persons and rational agents were not distinct primary kinds. A second response, one that would allow one to address the many-many problem more generally and the ontological extravagance that it seems to pose, would be to adopt a modified view of constitution as holding not only when two things are of distinct primary kinds but when they occur at distinct ontological levels. This would allow one to rest less heavily on the notion of a primary kind, or on particular construals of concepts such as person and rational agent, but would place a corresponding burden on the metaphor of levels. Either of these responses could be used to buttress the intuition that while persons may be distinctly existing entities, social agents are simply determinate forms that such entities may take, what philosophers and linguists might call phase-sortals of the category “person,” much as “teenager” and “child” are.

Applied to persons, this further modification of the constitution view could grant that while person, rational agent, moral agent, and bearer of mental properties are distinct primary kinds, these occur “at the same level,” and thus there can be no constitution relations between them. (The same could be said of body, aggregate of body parts, living thing, and causal network of bodily systems.) Constitution would remain a many-many relation that applies to the social domain, but social agents (including rational and moral agents) would be kinds of person, not kinds constituted by persons. This would preserve the intuition, which I think Baker herself shares, that there is something fundamental about the division between persons and bodies that is not shared by any of the other distinctions that can be drawn “within a level.”

There are three reasons for caution here. The first is the burden that this view places on talk of levels, talk that despite (or is it because of?) its ubiquity is seldom recognized as metaphorical, or cast in more precise terms. (The burden here may be greater for Baker herself than for others, since she is resistant to incorporating part-whole notions into the constitution view; mereology, the study of such relations, has provided the resources for most of the literal accounts of levels-speak.)¹¹

The second reason for caution is the generality (and conversely, the ad hocness) of this modification. It would have to apply to David (as a statue, a work by Michelangelo, a piece of art, a valuable artifact) and Piece, as well as the other examples of constitution that Baker mentions (e.g., dollars and pieces of paper; genes and strands of DNA), and would have to provide a principled account of the relationships between entities at any given level. Finally, this kind of restriction on the application of the constitution view runs the danger of undermining the view's application to its paradigm case, that of persons and bodies. For if persons and social agents are better thought of as standing in the relation of determinable to determinate, or primary kind to phase-sortal of that kind, rather than in a relation of constitution, then the very same might be argued to be true of body and person.

VII. Collective Social Agents and the Constitution View

Individual social agents are persons. Indeed, if what I have suggested up until the final two paragraphs in the previous section is on track, then persons stand in the particularly intimate relation of constitution to at least some social agents. In the final substantive section of this essay, I want to consider collective and institutional agents—banks, city councils, and Her Majesty's government—and the issue of whether the constitution view might be used to shed some light on both their actions and their place in the causal order of things.

Collective and institutional agents are something other than simply aggregates of individual social agents. Like all agents, they are knots in the causal net, loci of action, and there is some intuitive sense in which they are “higher-level” entities than individual persons. But they are not themselves persons (the phrase “corporate person” notwithstanding). Persons are, however, important to understanding collective and institutional social agency in a number of ways. Consider just two.

First, and perhaps most obviously, such social agency presupposes the existence of persons in order to act at all, not just as role-fillers but as cognitive agents who are able to plan, to make decisions, to inquire, to communicate, to judge, to set goals, to evaluate outcomes, to make estimates, to readjust schedules, to comfort and console. It is not simply bodies that fill these roles, but creatures with a first-person perspective, for many of the skills that these roles require are those of persons.

Second, and less obviously, we often use a kind of cognitive metaphor in describing the activities of collective social agents, in effect treating them as cognitive agents in their own right.¹²

Corporations reach decisions, governments distrust one another, and the school band refuses your generous offer to play the harmonica with them. If we build enough into the idea of having a strong first-person perspective, such that an entity must have second-order states directed at a phenomenal life, then organizations, institutions, and even whole cultures are seldom, if ever, treated as persons, even if they are treated as if they were cognitive agents of some kind. The cognitive metaphor, at least as it is applied in the social sciences, is never complete, and tends to focus on intuitively more cognitive and behaviorally grounded psychological states and traits. Yet what happens in such cases is that collective social agents are personified, at least in part, just as are many biological agents—selfish genes, cell recognition, immune defense as self-defense—when the cognitive metaphor is used in the biological sciences.¹³

One reason that collective social agents do not seem like apt relata for the relation of constitution, however, is that they are not physically bounded entities. As such, they seem unlikely candidates for satisfying the first condition of constitution: spatial coincidence. Boards of directors, trade unions, philosophy classes, families, and the welders in a factory are or can be agents of some kind, but they are not continuous, spatially bounded, physical agents, and thus cannot be spatially coincident with entities that have these features.

What I want to suggest instead is that such social agents are what I shall call agency coincident with the collections of persons that belong to them, whose members are (typically but not exclusively) the means through which they act. If this is correct, then I think we can defend an analogue of the constitution view relating collective social agents and such persons.

Two objects are spatially coincident at t just if they occupy precisely the same parts of space at t. Two agents are agency coincident at t just if they undertake precisely the same actions at t. To be agency coincident simpliciter is to be agency coincident at all times (cf. spatial coincidence). Collective social agents can act, of course, through the agency of just one or more individual agents, as when a president acts on behalf of a nation to declare war on another country (or even on a phenomenon, such as “terror”); those agents need not even be part of the collective social agent, as when a lawyer acts on behalf of shareholders to remove a wayward director. Yet this is compatible with (say) the shareholders and the collection of individual shareholders being agency coincident, so long as the action of the lawyer is that of the shareholders just if it is also that of the collection of individual shareholders. (And the action of the president is an action of the nation just if it is an action of the collection of individuals that belong to the nation.) Since there are many social mechanisms, institutions, and practices that authorize individuals and small groups of individuals to act on behalf of many individuals considered as members of a collective social agent, it would be a mistake to focus exclusively on unanimity, consensus, or even majority decision as our model for how collective social agents act.

To make the case that collective social agents are agency coincident with the collections of individuals that belong to them and that they represent is a large task that I do not propose to undertake here. But I would not raise it as a possibility if I thought that it had no prima facie plausibility. What I do want to argue for, though, is the claim that if agency coincidence were established, then by showing how collective social agents and individuals (for short) or persons satisfied the metaphysical sufficiency and contingency conditions of the constitution view, we would have shown how to adapt the constitution view of persons to make sense not only of individual agency but of collective social agency.

It is relatively easy to show that collective social agents and persons do satisfy these conditions. Consider the collection or group of persons who belong to or are represented by a given collective social agent at a given time. Clearly, that group of persons could exist without that social agent existing. Each member of the board of directors could exist without the social institutions presupposed by the existence of the board itself, or the mechanisms making just those individuals members of the board. But if we have a group of persons, and those conditions are in place—the corporation exists, those persons have been appointed to the board, etc.—then we must also have the corresponding collective social agent, in this case, a board of directors. As in the standard cases of constitution, this is guaranteed by the nature of these conditions; thus, taking an agent that is a kind of constituent—a group of people—and adding these conditions is metaphysically sufficient to create a collective social agent.