Jump to ContentJump to Main Navigation
Show Summary Details


Journal of Social Ontology

Editor-in-Chief: Schmid, Hans Bernhard

Managing Editor: Thonhauser, Gerhard

Ed. by Hindriks, Frank / Ikäheimo, Heikki / Laitinen, Arto / Mikkola, Mari / Salice, Alessandro / Schweikard, David

Open Access
See all formats and pricing


Précis of The Ant Trap

Rebuilding the Foundations of the Social Sciences

Brian Epstein1

1Department of Philosophy, Tufts University, Medford, MA, USA, e-mail:

©2016, Brian Epstein, published by De Gruyter. This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License. (CC BY-NC-ND 3.0)

Citation Information: Journal of Social Ontology. Volume 2, Issue 1, Pages 125–134, ISSN (Online) 2196-9663, ISSN (Print) 2196-9655, DOI: https://doi.org/10.1515/jso-2016-0001, March 2016

Publication History

Published Online:


This article summarizes The Ant Trap: Rebuilding the Foundations of the Social Sciences. The book develops a new model for social ontology, applies it to groups and collective intentionality, and criticizes various forms of individualism. Part One of the book presents two traditional approaches to social ontology and unifies them into the “grounding–anchoring model” for the building of the social world. Part Two shows that individualism is mistaken even for basic facts about groups of people, challenges prevailing views of group intention and action, and illustrates how to approach facts about groups in general.

Keywords: Social ontology; Metaphysics; Anchoring; Collective intention; Individualism; Groups; Agency; Social science

1 Introduction

If we want to improve the social sciences, it is crucial that we understand the objects and phenomena we study. Social ontology matters to social science: this is the conviction with which the book is written. The aim of The Ant Trap is to debunk longstanding and widespread errors in the study of social ontology, and to rebuild their foundations with a more modern and comprehensive model.

Oddly, it is sometimes overlooked that social ontology is a subfield of metaphysics. In the last 40 years, there have been enormous strides in metaphysics, but so far the social ontology literature has paid them little attention. I regard this as a serious error. In fact, the tools being developed in metaphysics are of great utility in untangling problems in social ontology. And there are equally important benefits flowing in the other direction as well. The field of metaphysics overall can profit enormously from a serious treatment of social ontology.

Much of the metaphysics literature can seem obscure and impenetrable, and the social ontology literature is so broad and diverse that it can be hard to get a handle on it. An important purpose of the book is to get practitioners talking on both sides – to present crucial parts of metaphysics for people who study society, and to present the problems of the nature of the social world in an organized way, so that theorists can see how various investigations fit together and approach them systematically.

The overall project of the book is to develop a new model for social ontology and apply it to central problems. The principal constructive aims are:

  1. To clarify a variety of distinct projects in social ontology that have been confused with one another. In particular, to clear up how to fit together varieties of individualism, theories of the building blocks of the social world, theories of convention, collective intentionality, and so on.

  2. To unify these into a model for the building of the social world. This is the “grounding–anchoring” model, which provides a general framework for social ontology.

  3. To show how to go about investigating the grounds for social facts. I focus on an important simple case: groups of people. What facts about groups are there, and what do they depend on?

  4. To develop a new and systematic treatment of group action and group intention (also called shared intention or collective intention). I argue against prevailing theories, in which group intention is exhaustively determined by the attitudes of group members. And replace it with a more general and less stylized approach.

  5. To criticize individualism in social ontology, and show that prevailing theories take an excessively “anthropocentric” approach to the social world.

  6. To introduce and refine tools in metaphysics and social ontology. In addition to grounding and anchoring, I discuss social facts, frames, dependence, ontological individualism, constitution, criteria of identity, and other tools that prove to be of enormous help in both fields.

My deep hope in the book is to help contribute to reworking the field of social ontology on new and better foundations. I find that the prevailing models in the field are tapped out, built on faulty assumptions and misleading simplifications. Young theorists in particular should not assume that the best way of modeling social phenomena is to model individual people interacting with one another. Or that John Searle’s theory of social construction is the best option out there. Or that we should understand collective intentions in terms of Margaret Gilbert’s joint commitments or Michael Bratman’s interlocking attitudes of group members. Or that social facts can be broken down into a homogeneous set of building blocks, such as individual mental states as economists often assume, or into practices, as sociologists like Pierre Bourdieu and Anthony Giddens do. Social ontology is important enough that it warrants a rigorous and serious treatment. When we do that, we find that much of the gospel we were raised on does not hold up, and that there are alternatives that the prevailing theories do not conceive of.

The Ant Trap is divided in two parts. Part One is titled “Foundations, Old and New.” The aim of this part is to present the two distinct treatments of the building of the social world, show that they are distinct, and develop a unified model for the building of the social world, the grounding–anchoring model.

Part Two is “Groups and the Failure of Individualism.” This part of the book focuses on grounding facts about groups of people. The aims of Part Two are (1) to show that individualism is mistaken even for basic facts about groups of people; (2) to demonstrate that group intention and action can depend on more than their members, and to challenge prevailing views; and (3) to illustrate how to approach the grounds of social facts in general.

Though the book is not terribly long, it covers a lot of ground. I have found that readers and commentators often focus exclusively on the first part, and in particular on the idea of “anchoring.” It is true that the grounding–anchoring model is one of the centerpieces of the book, but I hope it does not completely overshadow the second half. For people working on collective intention, judgment, and attitudes, Part Two may be even more relevant than Part One is.

2 Part One: Foundations, Old and New

Part One of the book reworks social ontology and its foundations more generally. Chapters 1–4 present the background of the prevailing approaches to social ontology, starting with a little history and leading up to the present consensus. Chapters 5–9 develop a unified model for social ontology – the grounding and anchoring model – and use it to clarify problems and views in social ontology.

2.1 Two Prevailing Approaches

Chapters 1 and 2 discuss the debate between individualism and holism and the social sciences. Theorists have long worried about “spirits” in social explanations, and have wanted to put the metaphysics of the social world on a solid footing. The chapters discuss the evolution of the debate in the late twentieth century. A good deal of the debate has been muddied by the confusion of questions about methodology and explanation in the social sciences with questions about ontology. The contemporary consensus is that explanatory individualism is questionable: it is not clear that social explanations can or should be given in terms of individuals. But that ontological individualism (typically understood as a thesis about the supervenience of social properties on individualistic ones) is straightforwardly true.

Chapter 3 raises doubts about this consensus. Ontological individualism is a stronger thesis than many people have realized. To illustrate this, the chapter examines the failure of a similar thesis in cell biology and how it misled practitioners. And then it shows that the same lessons apply to social theory. The chapter does not conclusively argue against ontological individualism (that comes later), but clarifies its connection to the practice of social science and gives intuitive reasons for doubting it.

Chapter 4 turns to a completely different approach to social ontology, and a competing consensus about how the social world is built. That is what some have called the “standard model of social ontology,” of which John Searle is the most prominent representative. The idea is that the social world is a kind of projection of our thoughts, or attitudes, onto the world. We, as a community, make the social world by thinking of it in a particular way. The chapter describes Searle’s theory of institutional facts, constitutive rules, and the assignment of status through collective acceptance. It also describes Hume’s theory of social convention, another example of this model. The chapter describes the tension between the “standard model” and the consensus view on ontological individualism.

Thus we have two competing models for the building of the social world. The two models take different social entities as paradigmatic. The “ontological individualist” uses examples like crowds, bazaars, and flows of commuters. That is, social objects that are composed of people. The “standard model” uses examples like dollars and parking spaces. That is, social objects that seem to involve assigning a status onto a physical object. How are these related?

2.2 Anchoring and Grounding

Chapter 5 is a quick primer on basic metaphysical tools that will be useful for social ontology. It presents facts and propositions, possibility, properties and relations, and social facts and social kinds. Then it introduces the idea of grounding. The chapter’s aim is to give a reasonably precise way to make and assess claims about the nature of the social world. Apart from minor exceptions, it sticks to the standard interpretation of the standard tools.

Chapter 6 presents anchoring and the grounding–anchoring model of how social facts are built. It explains how the two competing traditions about the nature of the social world can be unified. At the core of the model is the idea that there are different ways of “setting up” how social facts are grounded. We need to distinguish the facts that set up these ways of grounding from the facts that do the grounding. For example, there are certain conditions for a piece of meat to be kosher. If it comes from a pig, for instance, it is not kosher. Then there are the facts that set up, or “anchor,” those conditions. Perhaps these conditions are a product of agreement by religious authorities; perhaps a product of social acceptance; perhaps a product of social practices; or perhaps a product of divine command. Whatever these anchors are, they are the things that put in place the conditions for grounding a fact of the form x is kosher.

The basic elements of the model are represented graphically in Figure 1. A frame is a universe of possible worlds, and frame principles articulate how social facts are grounded in the frame. For instance, in our frame, there are frame principles articulating how a fact of the form x is kosher are grounded.

Fig. 1:

The grounding–anchoring model.

Within a frame are worlds – the actual world, and the other possible worlds. The frame principles apply to all worlds in the frame: we can look around actual and possible worlds to investigate whether various social facts obtain in those worlds. In any particular world, a social fact might or might not obtain, depending on whether there are appropriate grounds in that world. For instance, there might be a pig in some place in a given world. That grounds the fact that there is an unkosher animal in that place in that world.

Frame principles are very general, and do not have to take any canonical form. (In Chapter 11, I talk about the wide variety of forms frame principles take.) In a sense, the idea of a frame principle can be seen as a generalization of John Searle’s notion of a “constitutive rule.” And his theory of collective acceptance can be seen as a particular theory of anchoring. Searle argues that all constitutive rules have a single simple form (X counts as Y in C), and that a constitutive rule is always and only anchored by collective acceptance of the rule itself. But these limited conceptions of the form of constitutive rules and how they are put in place are radically inadequate.

Chapter 7 uses theories of law and legality to illuminate the grounding–anchoring model. In particular, it explains how laws are best understood as a kind of frame principle. This helps both clarify the model and also the nature of law. The main focus of the chapter is H.L.A. Hart’s theory of law, and his distinction between primary rules and secondary rules. Each of these can be understood as a kind of frame principle, with frames nested inside of other frames. Interestingly, we can then use the model to clarify some longstanding debates between Hart and Ronald Dworkin regarding the nature of law.

In Chapter 8, I clarify two distinct claims about the relation of individuals to the social world, which are often conflated. One claim is ontological individualism: this is a claim about how social facts can be grounded. A distinct claim is “anchor individualism”: this is a claim about how frame principles can be anchored.

Both ontological individualism and anchor individualism are false. Formulating them precisely (and charitably) is an important step to showing this. I argue that ontological individualism is false. Anchor individualism is also false, but arguing that is beyond the scope of the book; it will require a more thorough examination of the varieties of anchoring.

Chapter 9 wraps up the discussion of grounding and anchoring by addressing a common worry: why is not anchoring just a species of grounding? Instead of separating the anchors and grounds for a social fact, maybe social facts are grounded by a conjunction of what I call the anchors and what I call the grounds. In this chapter, I give a series of arguments against this “conjunctivist” view. The key problem with conjunctivism is that it gets the grounding conditions wrong for social facts. And the central piece of evidence that conjunctivists use – a kind of argument about counterfactuals – turns out to be no evidence at all. Anchoring is a distinct metaphysical relation.

3 Part Two: Groups and the Failure of Individualism

This part of the book focuses on grounding facts about groups of people. It argues that individualism is mistaken even for basic facts about groups of people, challenges the dominant views of group intention and action, and discusses how to investigate grounding in general. Chapters 10–13 talk about groups and facts about them in general, showing how we can work though the grounds for certain facts about groups, and then Chapters 14–18 deal with group intentions and group agency.

3.1 Groups and How to Ground Facts About Them

What is a group? There is a growing literature nowadays trying to identify the mathematical object that groups are: are they sets? Classes? Collections? Structures? I find these unhelpful. There are many mathematical models of a group, but I see no reason to identify a group with this sort of object. Instead, in Chapter 10 I begin with a broad and generic characterization of groups: they are entities constituted by and only by people. A benefit of this characterization is that it allows us to make sense of important features of groups – in particular, that they can change their memberships, and that there can be distinct groups that have identical memberships. To make sense of this, I discuss how to understand constitution, and also refine some views in the literature.

Chapter 11 begins the main topic of this part. What kinds of facts about groups are there, and how are they grounded? What are some techniques for investigating these? Answering these questions helps us approach group intention and action: after all, facts about group intention and action are among the facts about a group. But this chapter leads off with a simpler case. I consider facts about one particular group: the Supreme Court. I illustrate how to work through a few examples, including facts about the existence, membership, and powers of the Court. With these examples, it is easy to see that (1) there is a long list of facts relevant to social theory, not just the limited range often treated in social ontology, and (2) facts about groups are typically grounded by a variety of heterogeneous facts, not just by facts about the group members or even facts about people at all.

Chapter 12 has two complementary goals. One is to make sense of something puzzling: how can it make sense for a group to persist, and especially through times when it has no members? (This could, for instance, happen with the Supreme Court if all the members resigned at once and then were replaced.) How can we identify a group over time, if not through its members? The second goal is to clarify and generalize an important tool of metaphysics: the idea of a criterion of identity. Metaphysicians often use criteria of identity to talk about persistence. But it turns out that in their standard formulation, they cannot handle cases like this. In this chapter, I develop a general notion, which I call a “cross-identifying criterion,” and apply it to groups.

A key reason I discuss criteria of identity here is that we need it for Chapter 13. That chapter is like an advanced version of Chapter 11. Instead of showing how we look into the grounds of simple facts, it moves to some thornier facts we might encounter in building models in the social sciences. For instance, we often want to track facts about a particular group over time, or across different possibilities. Or there may be several groups of a given kind, like a number of teams of some sort, where we need to track the properties of each even as members come and go. Using the tools from the previous chapter, in Chapter 13 I show how to attack the grounding of these more complex facts.

3.2 Group Intentions and Actions

With this experience working with facts about groups, in Chapter 14 I turn to group intentions and actions. It is almost universally held among people who study group attitudes that they are exhaustively determined by the attitudes of the members of the group. This is either tacitly assumed or explicitly argued by Michael Bratman, Christian List, Philip Pettit, and many others. The aim of this chapter is start undercutting this assumption. When we look at group properties in general, there are a great many ways they are determined. Sometimes it does require that members have a property of a certain kind in order for the group to have properties of that kind. But a great many properties do not work this way. With some properties of a group, the members are actually completely irrelevant. This chapter describes several different patterns of how group properties are related to member properties.

Chapter 15 then attacks group action head on. It argues that the actions taken by a group can depend on much more than the actions taken by the group members. This is shown with three different kinds of examples. First are cases involving groups with hierarchies, divisions of labor, or structures of power, where different members make unequal contributions to group action. If the positions in the hierarchy depend on facts about non-members, this implies that the actions taken by the group may likewise depend on non-members. Second are cases involving direct constraints on group action, such as conditions on when the group is in session, its jurisdiction, and its powers. And the third cases involve membership constraints in political systems.

This last set of cases has practical implications for a topic in political science: the phenomenon of electoral control. In certain situations, it is not the legislators, but the electorate, who ontologically determine the actions of the legislature. This observation has the potential to open a new route to modeling legislative action.

In some ways, group action is the easy case. Group intention is a little subtler, in order to show that it is often determined by more than the attitudes of the members. This is the argument of Chapter 16. It starts from the idea that the model for the practical activity of groups is the practical activity of individuals. The functional roles that group intention plays for groups are analogous to the roles that individual intention plays for people. And intention is part of a system of practical activity, involving intention, planning, deliberation, and action. This means that the factors that determine group action are tied to those that determine group intention. Extending the cases from Chapter 15, I show that factors entirely external to the members can be among the ontological grounds of the intentions of a group.

Together, these arguments show that we need to develop a new approach to modeling group intention and other attitudes. Prevailing models focus only on group members, and argue about infinite nuances regarding just which properties of members aggregate in which ways into group attitudes. But for all their nuance, these models ignore whole swaths of factors that may be far more important for determining the attitudes of real-world groups.

The last two chapters of the book discuss the main theories of social groups in the literature. What is it for a group to be an agent? One approach takes genuine social groups to be “social integrates”; that is, for the members to have certain attitudes, beliefs, or commitments toward one another and toward the group as a whole. Representatives of this view include Margaret Gilbert, Michael Bratman, and Philip Pettit. Chapter 17 presents these respective theories, and shows that they are built on unfortunate and arbitrary limitations. In many cases, social integration is neither necessary nor sufficient for group agency.

A second approach to group agents is the “status model.” Representatives of this include John Searle, Raimo Tuomela, and Frank Hindriks. On this view, a group can be an agent in virtue of having a particular status projected on it. Chapter 18 presents these theories, and argues that it does not make sense to see agency as projected onto a group. For a group to be an agent, it must realize a system of practical activity. Not all the work of doing this must fall on the shoulders of group members: a group can be an agent even without much social integration, if there are other structures in place to do the work. But status assignments do not do anything to make a group into an agent.

Behind both of these approaches is the same mistaken individualistic assumption: if a group takes action, or has an attitude, that must be fully implemented via its members. So the social integrate model puts all responsibility for group agency on the members. The status model recognizes that many groups (like corporate boards) can be agents while failing to meet anything like the requirements set out by the social integrate theorists. So they infer that the only way such groups can be agents is by having it projected upon them. Yet these seem like the two available alternatives only because they overlook all the other factors surrounding groups – infrastructures, external constraints, membership requirements, existence conditions, assignment of positions of power, and so on – which do not depend on the members and yet play key roles in their performance of practical activities. Once we abandon the dogma that facts about groups are exhaustively determined by facts about their members, groups and their attitudes take on a whole new look.

Comments (0)

Please log in or register to comment.