1. Group agents

Groups come in a staggering variety of forms. The Republican Party, the Philadelphia Phillies, the founding fathers, Catholics, Caucasians, brunettes, vegans, and gamblers are all groups of one sort or another. These groups may be of interest for various reasons, including their role in explanations of certain social phenomena. Yet it would be premature to lump these varieties of groups together for their shared participation in social explanations. while vegans and the Republican party both figure in explanations of social phenomena, the nature of such explanations varies considerably. Plausibly, certain phenomena are best explained by the actions and intentions of the Republican party. These actions and intentions may be, at least practically speaking, irreducible to the actions and intentions of members of the party. In contrast, whatever actions and intentions are properly ascribable to vegans are plausibly so ascribable only in virtue of parallel actions and intentions of individual vegans. The distinction here is between group agents and nonagential groups. The Republican party is a group agent. The group of vegans is not a group agent, although some vegans might band together to form a group agent.

I focus here on the constitution of group agents. These entities have been the focus of much philosophical attention in recent years (List and Pettit Reference List and Pettit2011; Ritchie Reference Ritchie2015;Footnote ¹ Tuomela Reference Tuomela2013), not least because some group agents enjoy considerable social and political power. Just what group agents are and how exactly they relate to their members are thus pressing questions.

While my focus is on group agents, I do not mean to exclude the possibility that all groups are constituted by their members, a view Brian Epstein endorses (Reference Epstein2015). However, a satisfactory discussion of the constitution of groups in general would require some account of which groups deserve ontological commitment. That is a tricky question. If we commit to the existence of the group of vegans, should we likewise commit to the existence of the group of vegetarians, pescatarians, pollo-pescatarians, pollotarians, flexitarians, and so on? While questions of this sort are perhaps not unanswerable, such questions would take us far afield from the focus of this paper. Moreover, as we will see below, there is reason to think that the conditions in virtue of which some individuals constitute a group agent of a certain type differ radically from the conditions in virtue of which some individuals might constitute a nonagential group like one of those described above. Thus, I leave the articulation of a constitution view of groups in general, as opposed to a constitution view of group agents, for another time. For brevity, I will use “groups” to refer to group agents, unless otherwise stated.

2. Constitution without identity

In familiar cases, constitution as nonidentity seems to capture the relation holding between certain spatially coincident objects at lower and higher levels of reality. Consider the oft-discussed example of the statue and the piece of clay from which it is fashioned (Gibbard Reference Gibbard1975; Thomson Reference Thomson1998). The statue and piece of clay have many properties in common, including their spatial location, size, shape, and mass, but their similarities have limits. Suppose the origin of the statue is as follows. Toying around in her studio, an artist discovers that she has a piece of excess clay. Inspired by a book about the Second Punic War, she shapes the clay into a statue of Marcus Sergius. The statue came to exist after the piece came to exist. Moreover, while the piece’s existence does not require that anyone intended to create a piece of clay—we may suppose that the clay is leftover material—statues plausibly depend for their existence on the intentions of some creator. Additionally, the piece can survive changes that would destroy the statue. If she is unsatisfied with her work, the artist can destroy the statue by crushing it, but the piece will remain. Despite this, the statue is in some ways more resilient than the piece. Suppose the artist decides to depict Marcus Sergius after the war and so removes the right hand from the statue. The statue persists through this change but, plausibly, the piece is more closely tied to its matter and so the piece does not.Footnote ²

The preceding paragraph highlights several differences in properties between the statue and the piece of clay from which it is formed:

1. i. The piece, but not the statue, can survive crushing.

2. ii. The existence of the piece is more closely connected to the clay than is the existence of the statue.

3. iii. The statue, but not the piece, depends for its existence on the satisfaction of conditions external to itself.

By the indiscernibility of identicals, it follows from these differences in properties that the statue is distinct from the piece. Still, they are closely related—the former is made up of the latter. According to a popular view, the statue is constituted by, but not identical to, the piece.

3. The constitution of group agents

We now turn to the constitution view of group agents. We may begin by stating the view in its generic form:

Several authors, including Gabriel Uzquiano (Reference Uzquiano2004), Robert Wilson (Reference Wilson2007), Frank Hindriks (Reference Hindriks2013), and Epstein (Reference Epstein2015) have expressed support for CV. CV is widely acceptable, in part, because it says so little. In sections 4 and 5, I consider some more detailed and more controversial articulations of CV. Making CV precise requires, among other things, clarifying the relata in CV.

Much of the motivation for CV derives from apparent similarities in the relationship between material objects and the objects they constitute and individuals and the groups to which they belong. Consider the Chicago Enforcers. The Enforcers were a team in the XFL, an American football league founded in 1999, that put forward a single season of play in 2001. Like any other team, the Enforcers had human individuals as members and was, in this sense, made up of individuals. Moreover, the team was plausibly spatially coincident with its members.Footnote ³ When the team competed in Orlando, it was able to do so in virtue of its members being there.

Still, it would be a mistake to identify the Enforcers with the members of that team. While the Enforcers were founded and disbanded in 2001, the members of the team existed both before and after that year. Although its short tenure precluded significant turnover in membership, the team could have persisted through gaining or losing members. This contrasts with the collection of individuals that made up the group at any particular time. Any change in members would produce a change in the identity of this collection. Moreover, the team, but not the collection of individuals that made it up, required for its existence that some external conditions be met. For instance, there plausibly could be no Chicago Enforcers without a collective recognition of football.

From the preceding discussion, we can extract the following differences between the Enforcers and the collection of individuals that made it up:

1. iv. The collection, but not the team, can survive disbanding.

2. v. The existence of the collection is more closely connected to the members than is the existence of the team.

3. vi. The team, but not the individuals, depends for its existence on the satisfaction of conditions external to itself.

Items iv.–vi. closely mirror the relationships between some material objects and their constituents, as described in i.–iii. These similarities lend CV its initial plausibility.

A further datum is often taken to support CV, but its relation to this thesis is uncertain. It is frequently observed that a single collection of individuals can exhaust the membership of distinct groups simultaneously. This is often taken to be analogous to the possibility of spatially coincident yet distinct material objects and thus to lend further support to the constitution view. But while there can be coincident yet distinct group agents and there can be coincident yet distinct material objects of more familiar kinds, the relationship between constituents and constituted objects differs drastically in each case. The spatially coincident objects whose relationship is best described by the constitution view generally exist, as we might say, on different levels of reality. In contrast, it is a familiar fact that a single collection of individuals can form the membership of two distinct groups at a time. Two distinct teams might be coextensive, for instance. In such a case, the same individuals apparently constitute two distinct groups, where neither group constitutes the other. This contrasts with more familiar cases of constitution, where no object constitutes distinct objects at a single level of reality.Footnote ⁴ Thus, CV straightforwardly accounts for the coincidence of the individuals and each of the teams, but not, by itself, for the coincidence of the two distinct teams. The widely accepted datum that distinct groups can have the same members at the same time does not straightforwardly support the constitution view. Indeed, insofar as existing versions of CV fail to adequately account for the possibility of coextensive yet distinct groups; the existence of such groups challenges that view. I argue below that existing articulations of CV offer no satisfactory account of the possibility of coextensive yet distinct groups.

4. Gabriel Uzquiano’s constitution view

Gabriel Uzquiano’s version of CV is modeled on accounts of the constitution of material artifacts and especially Judith Thomson’s (Reference Thomson1998) account of this relation.Footnote ⁵ Uzquiano defines the constitution of a group, G, by a set of individuals, S, at time, t, as follows:Footnote ⁶

Uzquiano develops his version of CV in explicit opposition to set views of group agents, the simplest version of which has it that groups are sets of their members.Footnote ⁷ A common complaint against such views, and one Uzquiano shares, is that set views fail to account for the ability of groups to differ in members across times and across worlds. Unsurprisingly, by way of condition (2), Uzquiano’s definition of the constitution relation emphasizes that groups, unlike sets, permit fluctuation in members.

Uzquiano’s approach retains one commitment of set views—that groups are made up of sets. Epstein (Reference Epstein2015, 145) has expressed some concerns about treating sets as a relatum in the constitution relation, while ultimately adopting the same assumption. Epstein’s concerns are well founded. Some of the plausibility of Uzquiano’s and other version of CV derives from the familiarity of cases of material constitution without identity. However, Uzquiano’s treatment of sets as one relatum in the constitution relation precludes understanding CV as a sort of material constitution. To see this, consider one commitment of Thomson’s account of material constitution. Thomson (Reference Thomson1998, 155) takes the spatial coincidence of objects to be a necessary condition on the material constitution of one object by the other. Indeed, a core reason to posit the relation of constitution is to explain how two distinct things can occupy the same space. Treating sets as one relatum in the relation posited by CV thus puts CV at odds with accounts of material constitution. Sets, after all, are not located in space, at least according to conventional views (Ritchie Reference Ritchie2013). Thus, if such conventional views are correct, sets cannot be spatially coincident with groups and so no version of CV can treat sets as a relatum in the group relation while respecting that coincidence is a relation holding between spatially coincident entities.

That treating sets as relata in CV requires abandoning the conception of constitution as a relation holding between spatially coincident entities may appear an acceptable cost, because some apparent cases of constitution involve objects that are not spatially located. Consider the German national anthem.Footnote ⁸ Presently, the third stanza of Das Lied der Deutschen is the German national anthem. Previously, however, the entirety of Das Lied der Deutschen was the German national anthem. We may thus perhaps say that the German national anthem is constituted by, but not identical to, the third stanza of Das Lied der Deutschen. Let us suppose that this is a genuine case of constitution. Even on this supposition, the case provides no basis for accepting constitution without spatial coincidence. While no point in space is occupied by both the third stanza of Das Lied der Deutschen and the German national anthem, no point is occupied by one but not the other either. Thus, we may say, even if trivially, that Das Lied der Deutschen and the German national anthem are spatially coincident.Footnote ⁹ The case is therefore unlike that of the purported constitution of group agents by sets, in which the former are spatially located and the latter are not.

The case developed here against versions of CV that treat sets as the constituents of groups depends on two assumptions. First, sets—even impure sets—are not located in space. I will not defend this assumption here, but arguments for this position are available in the literature (Cook Reference Cook2013). The second assumption on which this line of objection depends is that group agents are located in space. There is ample reason to accept this second claim. Many properties we associate with group agents depend on their spatial locations.Footnote ¹⁰ Whether a military force poses a threat may depend on its spatial location. An American football team’s prospects for success may depend on its field position. Examples like these indicate that group agents are spatially located and thus, given the preceding assumption, not spatially coincident with sets. Consequently, Uzquiano’s version of CV, committed as it is to the claim that group agents are constituted by sets, abandons the notion of constitution as a relation holding between spatially coincident entities.

My criticism of Uzquiano’s version of CV has thus far seized on its treatment of sets as constituents of group agents. This need not be fatal critique, as Uzquiano’s view might be revised to exchange sets for some alternative sort of entity that is spatially located. Epstein expresses a deeper worry for Uzquiano’s view—namely that it is uninformative (Reference Epstein2015, 145–46). As Epstein points out, Uzquiano offers a definition not of the constitution relation, but of the group-constitution relation. This is a relation whose relata, by Uzquiano’s lights, can only be sets and groups. Sets have their members essentially, while groups do not. So, as Epstein notes, condition (2) in Uzquiano’s definition merely points out an implication of (1). We are thus left with an uninformative definition of the group-constitution relation.

One drawback of the lack of detail in Uzquiano’s articulation of CV is that Uzquiano’s view apparently does not account for the existence of coextensive yet distinct groups. It is widely accepted, however, that the same individuals can make up distinct groups at the same time. Indeed, both Uzquiano (Reference Uzquiano2004) and Epstein (Reference Epstein2015) take the existence of coextensive yet distinct groups to be evidence for CV and against the set view.Footnote ¹¹ To be sure, Uzquiano is quite clear that his version of CV is consistent with the existence of coextensive yet distinct groups (2004, 151), but Uzquiano says little about the conditions under which the same individuals might constitute distinct groups. What Uzquiano seems to have in mind is that just as groups differ in their modal properties from the sets that make them up and are therefore distinct entities, so might two coextensive groups differ in modal properties and therefore be distinct. But such an explanation raises a new puzzle.

The puzzle is like one familiar to proponents of material constitution. Such views have it that certain objects materially and spatially coincide but are distinct. The puzzle for proponents of material constitution is how to account for modal and sortal differences between such objects.Footnote ¹² This “grounding problem” presents a serious challenge for proponents of material constitution without identity.Footnote ¹³ Proponents of CV face an analogous worry. Call this the “group grounding problem.” Uzquiano, for instance, apparently takes the distinction between coextensive group agents to be possible due to the distinct modal properties of the group agents, but such a difference requires its own explanation.Footnote ¹⁴ The suggestion here is not that Uzquiano and other proponents of CV cannot provide such an explanation, but rather that articulations of CV lacking such an explanation are incomplete.

The present concern about Uzquiano’s version of CV points to another. In general, when some object, x, constitutes another, y, at time, t, x could have failed to constitute y at t. The question can thus be asked why, or in virtue of what, x constitutes y at t. Call this the “in virtue of what” (IVW) question. Plainly, answering this question will go at least some way toward answering the traditional grounding problem for x and y may differ in their modal properties because the condition(s) in virtue of which x would constitute y may fail to be met. Similarly, an answer to the IVW question would help to solve the group grounding problem. Some individuals may constitute distinct groups simultaneously in virtue of different sets of conditions being met.

I do not intend to imply that no plausible answers to the group grounding problem and to the IVW question can be given, only that an important desideratum for any articulation of CV is that it go some way toward addressing these issues. At a minimum, these issues must be addressed if CV is to glean support from the existence of coextensive yet distinct groups. In the following section, I show that while Epstein’s account provides resources for addressing these issues, the account is implausible.

5. Brian Epstein’s constitution view

Epstein (Reference Epstein2015) follows Uzquiano in treating sets as the constituents of groups. Epstein is, however, explicitly noncommittal on this score (145), so his adoption of this element of Uzquiano’s proposal is no serious challenge to Epstein’s view. Instead, in what follows, I focus on how Epstein’s proposal serves to address the group grounding problem and the IVW question. Epstein does not explicitly advertise his articulation of CV as being able to address what I am calling the “group grounding problem” or the IVW question. Nonetheless, Epstein’s articulation of CV offers an answer to the IVW question and, thereby, a step toward solving the group grounding problem.

Unlike Uzquiano, Epstein describes the process by which some individuals might come to constitute a group. Epstein’s most detailed description involves an intramural basketball team at Tufts University. According to Epstein, the following steps are involved in the generation of such a team (2015, 184):

1. (1) The manager goes to the relevant website and clicks to create a team within an existing league.

2. (2) A form is sent from the computers at IMLeagues.com to the manager’s computer.

3. (3) The manager fills in the relevant fields on the form and clicks to submit it.

4. (4) The database at IMLeagues.com is populated with the relevant information.

5. (5) An email is automatically sent to the relevant supervisor at the athletic department.

6. (6) The supervisor then clicks, and so on, and fills in the forms to approve the team.

7. (7) The database at IMLeagues.com is populated with the relevant information.

The completion of (1–7) culminates in the generation of a roster through which individuals can sign up for the team. By signing up in this way, individuals come to be members of the team created through (1–7). The largest set of such individuals then constitutes the team (Epstein Reference Epstein2015, 191).

Epstein’s account can plainly explain the existence of coextensive yet distinct groups. Similar steps to those described in (1–7) might generate another group, and the same individuals might undergo some process—whether signing up on a roster or performing some analogous action—to come to constitute that group. The question is whether it is plausible that individuals come to constitute groups through processes of this sort.

A first concern for Epstein’s account is that the approach to constitution underlying it lends itself to unwelcome ontological implications. To see this, notice that Epstein accounts for the existence of distinct yet coextensive groups by allowing that the conditions in virtue of which some individuals constitute a particular group are conditions extrinsic to the individuals in question. Thus, for instance, it is by virtue of their appearance on a roster generated though IMLeagues that some college students constitute a Tufts University intramural baseball team. If, however, sets of individuals constitute groups distinct from themselves in virtue of the satisfaction of extrinsic conditions, then presumably individuals can come to constitute individual social agents distinct from themselves by virtue of the satisfaction of similar extrinsic conditions.Footnote ¹⁵ For instance, just as individuals can sign up to participate in team sports via IMLeagues, so too can individuals sign up to participate in individual events like boxing, log rolling, and so on. Thus, if sets of individuals constitute groups distinct from themselves by virtue of appearing on IMLeagues rosters, so too do individuals constitute individual social agents, distinct from themselves, in virtue of signing up on IMLeagues. Epstein’s version of CV accounts for the possibility of distinct yet coextensive groups precisely through the fact that distinct extrinsic conditions of the relevant kind can be satisfied simultaneously. This is as true in the individual case as the group case so, if Epstein’s proposal is correct, individuals can constitute distinct social agents at a time.

Consider an example. Suppose Sam participates in both boxing and log rolling for her university. On the approach described above, Sam thus constitutes both a boxer and a log roller, where these are each distinct from Sam and distinct from one another. The oddity of this position has been noted elsewhere.Footnote ¹⁶ We can press this concern, first, by considering the range of social roles that Sam, and individuals in general, plays. We may suppose that in addition to being a boxer and a log roller, Sam is a daughter, a sister, a university student, an A-student, a high-school graduate, a food delivery person, and so on. On the present approach, Sam constitutes many individual social agents, but she is not identical to any of these. So, wherever Sam is, there are many distinct individual social agents.

The strange result that Sam is spatially coincident with an indefinite number of entities distinct from herself points to a broader worry with the approach to constitution underlying Epstein’s account. Ideally, an account of constitution should distinguish between cases in which some individual undergoes a change and cases in which some individual comes to constitute a new entity (Baker Reference Baker2000, 40–41; Wilson Reference Wilson2005, 55). A core worry for Epstein’s version of CV is that by answering the IVW question with an appeal to the satisfaction of extrinsic conditions, Epstein is committed to an approach to constitution that misidentifies cases of property acquisition as cases of constitution.Footnote ¹⁷

One might defend Epstein’s version of CV by noting that the present problem is not unique to his view. The problem is, in fact, part of a broader problem for certain approaches to constitution, which Wilson calls the “many-many problem” (2008). Such a strategy is effective only if there is no version of CV that avoids the present concern. I argue in section 7 that there is such a version.

The susceptibility of Epstein’s view to the present concerns points to another peculiarity of Epstein’s view—that groups may be constituted by lone individuals, as when the first individual signs up for a team, or may exist without being constituted by any individuals at all.Footnote ¹⁸ Whether such a commitment is tenable is unclear.Footnote ¹⁹ Pretheoretically, one would expect groups to be groups of something. We may press this worry by noting that groups that are not presently constituted invariably lack at least some of the causal powers we associate with groups. Epstein addresses this line of objection by arguing that we may interact with groups that are not currently constituted (2015, 169–70). For instance, even if every position in the Supreme Court was vacated, we might interact with the Supreme Court. Congress might, for instance, rescind some of the powers currently invested in the court. This reply is unconvincing. Just as the powers of the presidency can be restricted even when no one is serving as president—say, because the president has vacated the office and the vice president has not yet been sworn in—so might the powers of the Supreme Court be restricted even when no one is serving on the Supreme Court. In the former case, there is no temptation to say that while no individual is the president, there is nonetheless a president. Instead, we might conclude that we can interact with the presidency without interacting with any individual. Similarly, it seems we can interact with the Supreme Court without interacting with any group of individuals. To act on the presidency and the Supreme Court, respectively, without acting on any person or group of persons is, plausibly, to alter laws and norms surrounding the presidency and the Supreme Court. Thus, it is not clear that this case provides evidence for the view that groups may exist without being constituted by any individuals.

Epstein contends that standard treatments of groups and their properties focus excessively on the individual members of groups and their properties. To find what grounds the existence of groups, as well as other group properties, we must look beyond their individual members, or so Epstein contends (Reference Epstein2015, chap. 17). There is much to recommend the broader focus Epstein brings to debates about groups and their properties but, as the points above illustrate, Epstein’s shift of focus from individual group members to entities and events outside of the group produces implausible results. While it is highly plausible that the existence of at least some groups by individuals metaphysically depends on events and facts outside of the individuals themselves, we should not conclude that the existence of groups can be completely secured by events and facts outside of the individual group members. Such an account would overpopulate the world with groups, and the metaphysical commitments that underlie it lend themselves to the overgeneration of individual social agents. Thus, we have reason to seek a further articulation of CV that accounts for the metaphysical dependence of groups on facts and events beyond individual group members themselves without denying the indispensability of those individuals to the constitution and, thereby, the existence of the group agents in question. I propose and defend such an account in the remaining sections.

6. Material constitution

While some of the objections facing existing versions of CV might be overcome through minor adjustments to those views, other objections cut deeper. Here I propose a novel version of CV that overcomes the minor and major objections to existing versions of CV canvassed above. As we will see, the construction of an adequate version of CV benefits from attention to a distinction at the heart of more familiar cases of constitution.

Let us begin by returning to a concern that, as I noted above, applies to both Uzquiano’s and Epstein’s versions of CV. Both authors treat sets as one relatum in the relation described by CV. As we have seen, this causes problems for both accounts. The issue is not fatal to either version of CV, nor for CV itself, as long as some adequate substitute for sets is forthcoming. The proponent of CV has options, and I cite only one here. A prima facie promising strategy for the proponent of CV is to replace sets of members with fusions of members as the relatum in CV.Footnote ²⁰ Fusions, unlike sets, are concrete material objects located in space. Thus, the proponent of CV may avoid problems that arise from treating sets as a relatum in CV by replacing sets with fusions in that relation.

Here one may object that once sets are replaced with fusions, we should rethink the relation that holds between these and group agents. Katherine Hawley (Reference Hawley2017), for instance, argues that group agents are fusions of their members.Footnote ²¹ I argue below, pace Hawley, that there are genuine cases of coextensive yet distinct groups and that these pose a significant challenge to identifying groups with fusions.

For the present, let us begin to construct a version of CV that takes fusions of individuals as the constituents of groups. A significant benefit of taking groups to be constituted by concrete material entities rather than sets is that we may, on this approach, treat the constitution of groups as material constitution. While CV is a relatively recent metaphysical innovation, discussion of material constitution has a longer history and the complexities of material constitution have thus been worked out to a greater degree. Among these complexities is the possibility that, despite initial appearances, there are at least two distinct varieties of material constitution.Footnote ²² The view that material constitution may be either “compositional” or “ampliative” has been discussed and motivated in a series of papers by Robert Wilson (Reference Wilson2007; Reference Wilson2008; Reference Wilson2009). Situating the constitution of group agents within Wilson’s framework will prove useful in developing a novel articulation of CV.

The basic motivation for Wilson’s distinction between two varieties of material constitution is best brought out by way of examples. Consider two objects: an organic molecule and a statue.Footnote ²³ The former might be constituted—as is the case of a methane molecule—by four hydrogen atoms and one carbon atom, while the latter might be constituted by a piece of clay. As Wilson points out, there is a significant difference between the constitution of the former and the constitution of the latter. Wilson holds that whereas the molecule is intuitively nothing more than the atoms that make it up there is something more to the statue than the piece of clay (2007, 3). We may press this distinction by noting that the atoms, suitably arranged, could not fail to be a methane molecule. By contrast, statues plausibly require for their existence that some conditions external to themselves be met. For instance, it is plausible that statues can only be created through the intentional activity of some creator. Arguably, the existence of statues depends more generally on the existence of an “artworld” (Baker Reference Baker1999; Reference Baker2000).

Wilson uses considerations like these to motivate the distinction between compositional and ampliative constitution. According to Wilson (Reference Wilson2007), for a thing, x, or a collection of xs to compositionally or ampliatively constitute y during a time period, p, the following conditions must be met:

Wilson takes compositional and ampliative constitution to be distinguished by two additional pairs of conditions. Cases of compositional constitution satisfy the following:

As “intrinsic necessitation” makes clear, compositional constitution is a matter of the constituent(s) of an entity being arranged so as to necessitate the existence of the constituted object. Compositional constitution thus does not depend on any conditions outside of the constituent or constituents obtaining. In contrast, cases of ampliative constitution satisfy the following:

Ampliative constitution, unlike compositional constitution, requires that certain conditions external to the constituent(s) be met.

Wilson’s distinction seems to capture clear differences in cases of material constitution. The conditions that must be met for a piece of clay to constitute a statue seem to be very different than those conditions that must be met for hydrogen and carbon atoms to constitute a methane molecule. Moreover, compositional constitution seems to capture the relation holding between various materially and spatially coincident objects. It is less clear that the conditions Wilson specifies for ampliative constitution accurately capture the relation holding between any materially and spatially coincident objects. Consider again the case of the statue and the piece of clay. While the constitution of the statue by the piece of clay does require that certain external conditions be met, it also plausibly depends on the piece of clay being arranged in a certain way. Indeed, one oft-cited difference between the statue and the piece of clay is that the latter but not the former can survive squashing. Thus, it seems, “extrinsic necessitation” does not hold even in the paradigm case of the constitution of the statue by the clay. It seems instead that the constitution of the statue by the piece of clay requires the arrangement of the clay into a certain form. The case of the statue and the piece of clay suggests either that Wilson’s account of ampliative constitution ought to be revised or that there is a third variety of material constitution. For present purposes, what matters is simply that there is plausibly a sort of material constitution that demands not only that constituents be arranged in a certain way, but also that certain external conditions be satisfied. I argue in what follows that the constitution of group agents by individuals is generally of this sort.

7. Constitution and functional structure

Taking the constitution of group agents by individuals to require a certain arrangement of those individuals may seem like a nonstarter. The formation of a novel group agent does not typically involve the assembly of human individuals into any particular physical arrangement.Footnote ²⁵ Moreover, even when individuals are put into a certain physical arrangement—a human chain or a human pyramid, to take a few examples—this process does not generate a novel group agent. It would be a mistake to conclude, though, that no condition adverting to the arrangement of individuals holds of all cases of the constitution of group agents by individuals. The lesson is instead that the constitution of group agents by individuals depends on a nonphysical arrangement of individuals.

The view presented here is perhaps best clarified through comparison to an existing view. Frederick F. Schmitt (Reference Schmitt and Schmitt2003, 6) considers a view according to which “a group is an instantiation of a functional structure in which individuals play roles.”Footnote ²⁶ Katherine Ritchie (Reference Ritchie2013) elaborates and defends this structuralist view. On Ritchie’s view, groups are realizations of functional structures, where functional structures are made up of roles and functional relations between these. A baseball team, for instance, has a structure that includes roles like “pitcher” and “catcher,” and relations like “pitch-ball-to” and “return-ball-to” (268). On Ritchie’s view, a group exists just when its structure is realized.

The view according to which groups are realizations of structures is subject to serious objection. Pace Ritchie (Reference Ritchie2013), treating groups as realizations of structures does not allow for groups to gain or lose members across time, or to vary in structure across worlds. For multiple realizable entities or properties, realizations are precisely those things that can vary without variance in the realized entity or property. Plausibly, groups can be realized by different individuals at different times and across worlds, but then groups are not identifiable with any particular realizations of their structures.

In a more recent paper, Ritchie writes that groups are not realizations of structure, but are instead “structures that have been realized by individuals” (2019, 10).Footnote ²⁷ This view is consistent with the possibility of groups varying in members over time and across worlds but is subject to its own difficulties. Most pressingly, Ritchie takes it to be at least plausible that structures are universals (5). But, if this is the case, and groups are structures, then groups either are not located in space or, perhaps even less plausibly, are located wherever a group instantiating that structure is located.Footnote ²⁸

While it is implausible that groups are either structures or realizations of structures, functional structures have an important role to play for group agents. Individuals constitute groups of certain sorts, I suggest, in virtue of realizing certain functional structures. This point is best brought out by analogy. Just as atoms might constitute a molecule of a certain kind in virtue of realizing a certain physical structure, some individuals might constitute a group of a certain sort in virtue of realizing a certain functional structure. What this structure is depends on the sort of group in question. A baseball team plainly has a different structure, including both different roles and different relations between these roles, than an American football team. It is in virtue of exhibiting the structure of an American football team that individuals constitute an American football team, and it is in virtue of exhibiting the structure of a baseball team that individuals constitute a baseball team. This is not to say that the constitution of groups is strictly compositional. The structures in virtue of whose realization individuals constitute groups are often, if not always, the product of intentions (Ritchie Reference Ritchie2019, 7). Thus, just as there could arguably be no statues without attitudes toward art, there could be no basketball teams without there being attitudes toward basketball.

The role of group structure in the constitution of group agents is articulated in the following account:

Although SCV rejects Ritchie’s identification of groups with structures, important similarities remain. SCV is, for instance, consistent with Ritchie’s account of group membership (2019, 10).

SCV captures the conditions that lend CV its initial plausibility. First, because F can exist without instantiating S, F can exist without constituting G. Second, because group structures are often, if not always, the product of intentions, groups often, if not always, depend for their existence on the satisfaction of conditions external to themselves. Third, although SCV does not guarantee that groups can survive changes in membership, SCV allows for this possibility.Footnote ²⁹

SCV partially solves what I have called the “group grounding problem” and does so by way of answering the IVW question. Per (1), it is in virtue of instantiating different structures that some individuals might constitute either a baseball team or an American football team. Thus, the individuals could constitute one without constituting the other. However, when the individuals simultaneously constitute groups of distinct types, this is in virtue of their simultaneous realization of distinct functional structures.Footnote ³⁰ Here I claim that some individuals can realize the structure of, for instance, an American football team even when they are not playing American football—and even when they are playing baseball. This claim seems to be common to all versions of the structuralist view and so I will defend it only briefly here. Notice first that even when a group is engaged in its characteristic activities, most of the functional relations holding between nodes in the group’s structure are not exhibited. For instance, although pitchers and catchers on baseball teams are consistently linked by a return-ball-to relation, little playtime in any given baseball game is taken up with catchers returning balls to pitchers. More broadly, the functional relations holding between positions in a group’s structure are plausibly like other functional properties in that they need not be exhibited to be present.

Why does SCV only partially solve the “group grounding problem”? By itself, SCV seems only to account for the possibility of coextensive yet distinct groups of different sorts. We sometimes think, however, that the same individuals can exhaust the membership of distinct group agents of the same sort at the same time. For instance, it seems plausible on its face that the same individuals playing the same positions might constitute both an intramural and an extramural basketball team at the same time. If so, then resources beyond those explicit in SCV must be mustered to account for the possibility of distinct yet coextensive groups of the same sort. I turn to this issue in the final substantive section of this paper.

8. Distinct yet coextensive groups of the same sort

The issue facing SCV is that while the view accounts for the possibility of distinct yet coextensive groups of different sorts, it is not clear that the view accounts for distinct yet coextensive groups of the same sort.Footnote ³¹ Here I consider two responses that can be invoked to defend SCV. Determining which response is better will take us beyond the scope of this paper, and so I do not attempt to settle the question here.Footnote ³² My aim is, instead, to demonstrate that SCV has resources for dealing with the issue.

First, the proponent of SCV may argue that there is more to a group’s structure than the functional relations holding between its members. Ritchie (Reference Ritchie2013; Reference Ritchie2019) distinguishes between internal and external structures. Functional relations are internal with respect to a group just when these relations are strictly between members of that group. But, arguably, kinds of groups are individuated by both internal and external structures. For instance, one might think that whether a team is an intramural basketball team depends on whether it is embedded in an intramural league.Footnote ³³ We might then distinguish between two basketball teams, one intramural and one extramural, based on their relations to entities outside of the team. On this approach, the two teams are of different kinds, but the same sort.

On the present proposal, a group’s identity is fixed by its relations to broader structures. This approach is consistent with the commonsense notion that the same individuals might constitute distinct groups of the same sort at the same time. However, this approach threatens to overpopulate the world with groups. As I noted in section 5, a single individual may play distinct social roles. One individual might be both a chess player in a chess league and a boxer in a boxing association, for instance. Yet there is little temptation to argue that the human individual constitutes both a chess player and a boxer, where these entities are distinct. Hawley (Reference Hawley2017) presses this point forcefully, arguing that the apparent inappropriateness of relationally individuating individuals cuts against the plausibility of relationally individuating groups. Hawley argues on this basis that the possibility of coextensive yet distinct groups need not be accommodated. If Hawley is right, then the inability to account for such groups is no challenge to her own mereological approach to groups. The key datum that precludes identifying groups with fusions of individuals is the possibility of coextensive yet distinct groups but, if Hawley is correct, there is no such datum. Hawley’s case against the existence of coextensive yet distinct groups leans heavily on the analogy between groups and individuals that play distinct social roles. To refute Hawley’s case for the mereological view, we must thus confront that analogy.

There is a core difference between individuals that play distinct social roles and some groups that do the same. To see this, start by considering a case where the analogy is strongest. Borrowing Epstein’s (Reference Epstein2015) example, Hawley (Reference Hawley2017) considers the Massachusetts Department of Transportation (MassDOT) and the Massachusetts Bay Transportation Authority (MBTA). The two committees, we may suppose, seem to meet at different times, have different responsibilities, and exhibit different powers. Hawley notes that the committees are, in these respects, like Boris Johnson who, for much of 2015, was both mayor of London and MP for Uxbridge and South Ruislip. Johnson had different powers and carried out different duties in his distinct capacities. Hawley thinks it would be a mistake to conclude, by appeal to the indiscernibility of identiticals, that Johnson then constituted a mayor and an MP, where these are distinct. Similarly, Hawley concludes, it would be a mistake to maintain the distinctness of the MassDOT and the MBTA based on their distinct duties, powers, and so on. Instead, Hawley argues, a single group plays different roles in different contexts.

Hawley’s reasoning provides grounds for doubt about coextensive yet distinct groups of the same sort. It is plausible enough that the same individuals cannot constitute distinct basketball teams, one intramural and one extramural, for instance. Instead, we might say that the same team plays in different leagues. However, Hawley’s claim is less plausible when applied to groups of different sorts. It is less plausible, for instance, that a basketball team and a football team, comprised of the same individuals, are the same team. The reason is that the two teams differ not only in the structures in which they are embedded, but also in their internal structures. By contrast, MassDOT and MBTA, we may suppose, exhibit isomorphic internal structures but bear different relations to external entities. By appealing to structures, we arrive at a middle ground between the positions staked out by Epstein and Hawley. There may be coextensive yet distinct groups of different sorts, but not of the same sort. This view accounts for the fact that Johnson does not constitute distinct individuals—lone individuals cannot exhibit different internal structures.

One might object that if we acknowledge that our practice of distinguishing between coextensive groups of the same sort is mistaken, we should do the same for our practice of distinguishing between coextensive groups of different sorts. Perhaps a basketball team and a football team comprised of the same individuals are, despite appearances, the same team. I cannot comment on this possibility at length here, but I think that SCV draws our attention to the importance that structures play for groups. The properties we associate with any group agent will be attributable in large part to the structure that group has. The present objection threatens to obscure the importance of structure and, in the absence of compelling reason to accept this cost, should be set aside.

9. Concluding remarks

CV has much to recommend it, but existing versions of that view are subject to serious objection. I have sketched a structural-constitution articulation of CV that treats the constitution of group agents as a special case of material constitution. The structural-constitution view accounts for the facts typically taken to motivate CV, while avoiding objections to existing versions of CV.