Book Review: The Ant Trap: Rebuilding the Foundations of the Social Sciences,by Brian Epstein

1. Introduction

The Ant Trap by Brian Epstein deals with a central issue in philosophy of the social sciences: the ontological nature of the social world and the metaphysical assumptions of social explanations. Epstein’s general argument is that the social sciences are suffering from the false but widely accepted ontology of individualism and thereby from a flawed anthropocentric picture of the social world. The starting point of Epstein’s analysis is that the social sciences find themselves in a paradoxical situation. Despite the recent explosion of new data and technologies providing them with incredible potential for empirical and theoretical progress, the social sciences seem at a stalemate because they have not improved empirical performance regarding explanation and prediction. For Epstein, ontological individualism is to blame. According to Epstein, ontological individualism is erroneous because the social world cannot be regarded as made up only of individual components. In his opinion, the fact that economists were unable to predict the 2008 financial crisis appears as the disquieting and unequivocal proof of the miserable condition of the social sciences. For him, their need for fundamental reform is thus apparent and urgent.

In The Ant Trap, Epstein scrutinizes various philosophical positions resting upon the anthropocentric view and discusses the philosophical and methodological consequences (p. 6f.). He furthermore proposes an alternative view of the ontological foundation of the social world by drawing upon new tools from metaphysics. His model rests upon the distinction between frames, grounds, and anchors of social facts and the book is largely about illustrating the usefulness of this model to rethink the foundations of the social sciences. In 18 chapters, Epstein aims to entirely rebuild the foundations of the social sciences, arguing that the dominant individualist social ontology accepted by economists, sociologists, and philosophers is dangerously mistaken and must be rejected. As such, The Ant Trap is enormously ambitious, not only in the revolutionary metaphysical project that it aims to set out but also because of Epstein’s self-professed aspiration that this “abstract work on the metaphysics of groups” is decisive for practical social science (p. 182).

Epstein begins by emphasizing that the social sciences are in crisis. Scholars, he explains, have attributed the crisis to at least five different practices: (a) the assumption of perfect rationality in economics; (b) the commitment to reductionist explanatory strategies; (c) the hyper-formalization and lack of realism of economic models, that is, the fact that economic models ignore the problem that social phenomena are highly complex; (d) the failure to base the social sciences on cognitive science; and (e) the tendency to model society or economies as a whole instead of developing small and more testable models. All five, however, miss the point for Epstein, because they all assume “that the objects of the social sciences are built out of individual people much as an ant colony is built out of ants” (p. 7). Therefore, “they overlook a deeper problem” (p. 182): although individuals are obviously relevant to the social sciences, the social world cannot be reduced to a world of individuals. To Epstein, the “picture of the social world as being composed by individual people”—the “ant trap” of the book’s title—(p. 10) appears naïve and inappropriate.

To revise this picture, Epstein engages with the metaphysics and ontology of the social. He calls into question the individualist ontology in the social sciences that social theorists have been defending for centuries and that had been reinforced by prominent figures such as Max Weber, Karl Popper, Friedrich August von Hayek, and John William Neville Watkins, as well as by prominent social ontologists such as John Searle, Raimo Tuomela, and Philip Pettit. While The Ant Trap is a criticism of both these intellectual groups, it also rejects the old-fashioned holist ontology defended by thinkers such as Friedrich Hegel and Karl Marx. Epstein’s proposed model in turn is thus nothing less than a radically new metaphysics of the social world, which is highly innovative yet has to be taken, as we explain later, with a grain of salt.

The Ant Trap is composed of two parts. Part 1, titled “Foundations, Old and New,” criticizes in detail ontological individualism and proposes a new nonanthropocentric ontology for the social sciences. Part 2, titled “Groups and the Failure of Individualism,” applies the conceptual tools developed in part 1 to the analysis of groups and their dynamics.

2. Part 1: Foundations, Old and New

The beginning of the book (chapters 1 and 2) focuses on the history of social ontology and the social sciences. As Epstein points out, in the social sciences, ontological individualism—meaning here that social phenomena are made up entirely of individuals—was first supported by methodological individualists, whose approach merges the interpretative perspective (Verstehen) and the invisible-hand model of explanation. They developed their approach to challenge holism, or the idea that social phenomena must be explained in terms of sociocultural determinism. While Epstein rejects ontological individualism, he shares methodological individualists’ criticism against holism and its tendency to reify social entities and deny human autonomy. As stressed by Epstein, today, methodological individualism is not as appealing as it used to be. Most social philosophers support what they consider a new kind of individualism as opposed to traditional methodological individualism (e.g., Sawyer 2002, 2003). This new individualist approach developed as a consequence of the fact that, starting from the 1950s, methodological individualism has been criticized by analytic philosophers on the grounds that it is tantamount to reductionism (see Mandelbaum 1955; Di Iorio 2015, 2016a). Epstein points out that methodological individualism is largely regarded today as based on an impossible principle of semantic reduction of social properties to individual ones and, as a consequence, is considered to be logically mistaken. However, the accusation of reductionism leveled at methodological individualism, which led to the development of an alternative nonreductionist approach within analytic philosophy, did not call ontological individualism into question. As explained by Epstein, Steven Lukes’s work is central to understand the development of that nonreductionist approach. Lukes, one of the most important critics of methodological individualism understood as reductionism, pointed out that “methodological individualism . . . was not just a single thesis: it was two. In fact it consisted of one controversial thesis and one trivial thesis. The controversial part has become known as explanatory individualism, and the trivial part as ontological individualism” (p. 21). Like most analytic philosophers, Lukes accepts ontological individualism (understood to mean that social phenomena are fully made up of individual ones), but rejects explanatory individualism, that is, reductionism. He considers explanatory individualism impossible: “even if societies consisted of nothing more than people, it may be impractical or impossible to construct social explanations individualistically” (p. 21), that is, to provide reductionist explanations. For the last 40 years, the nature of social explanations has been widely debated, and methodological individualism has often been criticized because of its alleged commitment to reductionism (e.g., Harold Kincaid), but ontological individualism, which is related to the rejection of social “spirits” (p. 13), that is, supra-individual substances, has been regarded as unquestionably correct.

Epstein stresses that this nonreductionist view, based on ontological individualism (i.e., nominalism), has been often expressed in terms of emergence. Complex social phenomena can be viewed as “emergent” in the sense that they result “from individuals and the interactions between individuals” (p. 23). Emergent phenomena are characterized by global properties semantically irreducible to strictly individual properties and laws, but they do not refer to “some independent metaphysical realm” in the sense of methodological holism (p. 25). From the standpoint of an emergentist, the macro-level is considered to be a derivative of the micro-level (p. 26). As highlighted by Epstein, another key concept in the history of ontological individualism—a concept that originated in philosophy of mind—is that of multiple realizability. Taking part in the debate on emergence and reducibility, Hilary Putnam and Jerry Fodor developed multiple realizability as a refined criticism of reductionism. One goal was to criticize Nagel’s model of theory reduction. The idea behind multiple realizability is that because “high level properties are multiply realizable,” reduction is therefore blocked (p. 32). For example, because a screwdriver can be made out of different materials, it is meaningless to reduce a screwdriver to its basic components. Multiple realizability is another variant of nonreductive nominalism.

As highlighted by Epstein, in recent years, philosophers of science and philosophers of mind have regarded ontological individualism “as a claim about supervenience” (p. 23) and assumed that claim as “obviously true” (p. 23). The concept of supervenience is the most recent variant of nonreductive individualism:

Supervenience is a relation between two sets of properties. Take property A to be all the social properties and property B to be all the individualistic properties. To say A supervenes on B, then, is to say an object cannot change its A-properties without there being some accompanying change in its B-properties. Or to put it more intuitively, the B-properties fix the A-properties. Once all the individualistic properties are in place, that fixes the social properties. (p. 33)

According to the theory of supervenience, “the social properties globally supervene on the properties of individual people” and, as a consequence, the “individualistic properties exhaustively determine the social properties” (p. 34).

In chapter 3, Epstein starts to develop his criticism against ontological individualism. In his opinion, despite its popularity, ontological individualism is “false” (p. 37). Epstein does not endorse holism, but rejects all variants of nominalism, including emergentism and supervenience. According to him, the failures of ontological individualism can be shown first and foremost “intuitively” (p. 37). In Epstein’s eyes, the view that social entities are (semantically irreducible entities) composed of individuals is as misleading as arguing that organisms are composed of cells. Organisms include “a lot of extracellular material” (p. 38) such as interstitial fluid between cells, blood plasma, ocular transparent gel, bone matrix, teeth, urine, gastrointestinal fluid, and so on. As a consequence, the human body cannot be explained in terms of supervenience, which would assume that it is composed only of cells, because “the cellular facts are too limited to exhaustively determine the anatomical facts.” Similarly, the “social facts do not supervene on the individualistic ones” (p. 36). The facts at the lower level, that is, the facts about individuals, do not exhaustively determine the facts at the higher level. To illustrate this point, Epstein suggests considering facts about the Starbucks Corporation: “To be sure, the employees are critical to the operation of Starbucks. But facts about Starbucks seem also to depend on facts about the coffee, the espresso machine, the business license, and the accounting ledgers” (p. 46).

Can we argue that changing properties of Starbucks are exhaustively determined by facts about individuals? The answer, Epstein contends, is clearly no:

Suppose, for instance, there is a freak, late night power spike at a number of Starbucks outlets, causing the blenders and refrigerators to break, the ice to melt, and the milk to spoil. Suppose that event is the last straw for a financially struggling Starbucks, underinsured as it is. So, when the power spikes and its key assets melt down, its assets no longer exceed liabilities. Overnight, as the owners, employees, and accountants are asleep in their beds, Starbucks goes from being financially solvent to insolvent. (p. 46)

Epstein stresses that it is not individuals or phenomena at the “individualistic level” that explain the transition to insolvency, that is, “the social-level transition” (p. 46). The transition to insolvency “involves property and equipment, not individuals” (p. 46). People are, of course, involved in the ordinary course of Starbucks operations, but the relationship between the individualistic level and the higher level cannot be explained in terms of supervenience, because the higher level is not exhaustively determined by the individualistic level. As a consequence, the nominalism underpinning the concept of supervenience must be regarded as a wrong assumption about the nature of the social world and must be rejected as such. Epstein does not argue that ontological individualists completely deny the relevance of nonindividual factors in the social world, but he criticizes those thinkers because they never realized that the relevance of those factors is inconsistent with their nominalist standpoint and obliges them to move far beyond it.

Chapter 4 of The Ant Trap is about one of the most popular individualistic views—what Francesco Guala has labeled the “Standard Model of Social Ontology” (Guala 2007) and which is exemplified by John Searle. Like ontological individualism, the Standard Model is almost universally endorsed today in analytic philosophy. It is grounded upon a vague ontological intuition that if all individuals were to disappear, all social institutions would disappear. In this view,

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 bills in my pockets are money because we all think of them as money. The president has the powers he does because we grant him those powers. America is a nation because we think of it as such. The social world, quite generally, is the social world in virtue of our beliefs about it. (p. 50)

The Standard Model is usually regarded by its advocates as based on ontological individualism. However, Epstein argues that it is actually inconsistent with it and that ontological individualism and the Standard Model must be kept separate. In his opinion, developing a better philosophy of the social sciences requires disconnecting the Standard Model from ontological individualism and merging it with a nonanthropocentric ontology.

A well-known example of the Standard Model is Searle’s (1995) concept of “institutional facts” as presented in his book The Constitution of Social Reality—a book about the relation between physical and social reality. According to Searle, members of a community create institutional (or social) facts in their community (e.g., universities, governments, restaurants, money) “by imposing ‘statutes’ on material objects” (Searle 1995, 51). The notions of “institutional facts” and “statutes” are strictly linked to that of “constitutive rule.” The status of money is imposed on some physical objects because of a constitutive rule. Consider paper money. Paper money presupposes that we assign a specific status to pieces of paper as material objects printed according to a specific process. Searle (1995, 28) argues that dollar bills have the following constitutive rule: “Bills issued by the Bureau of Engraving and Printing . . . count as dollars . . . in the United States.” What puts such a constitutive rule in place in a community “is that we collectively accept it” (Searle 1995, 52), that is, we all have a particular attitude toward it. In other words, Searle (1995) assumes that something is money “only because we think of it as money.” This means that social reality is ultimately based on physical reality and that social facts have collective intentionality in the sense that they presuppose a common attitude, “not just our attitudes as individuals, but as a community” (Searle 1995, 56; see also Guala 2007, 961-63).

According to Epstein, ontological individualism and the Standard Model are inconsistent with each other because of the central role played by the material world in the latter. In the Standard Model, the social world supervenes on both individual and physical properties. As a consequence, Epstein contends that the Standard Model, unlike what is argued by its advocates, cannot be regarded from a nominalist perspective. Unlike ontological individualism, it does not deny the metaphysical and explicative relevance of the physical world. According to the Standard Model, among the things that the fact being a dollar supervenes on are the properties of pieces of paper. “Ontological individualism holds that social facts supervene on facts about individuals. The standard model holds that facts about individuals set up the conditions for something to count as a social fact” (Epstein 2015, 58). In other words, in the Standard Model, “the social world is ‘brute’ facts treated in a certain way” (Epstein 2015, 58). This model is related to a realist or anti-idealist theory of the social world because it assumes that social facts are the product of “collective” rather than strictly individual attitudes and also because, from its standpoint, “the social world is not an attitude in our heads, but the actual stuff in the world to which a certain status or convention has been assigned” (pp. 58-59). Epstein rejects the Standard Model because it is based on an inconsistency and also because it fails to recognize the shortcomings of ontological individualism.

Chapters 5 and 6 of The Ant Trap deal with lexical and conceptual issues central to Epstein’s analysis. According to Epstein, advocates of the Standard Model failed to realize the inconsistency of their approach with ontological individualism because of a lack of conceptual precision. He argues that to clarify the implicit metaphysical assumptions of their approach and improve the ontological foundations of it and the social sciences more generally, it is necessary to introduce a crucial distinction between grounding and anchoring. Grounding in metaphysics is a relation between facts. Facts are propositions about the actual world. As argued by Bricker (2006), one proposition grounds another if the former is fundamental in nature and the latter supervenes on the former. For example, “the fact a million herring turned in such-and-such directions grounds the fact the school split in two” (p. 70). These two facts “are metaphysically related in the sense that the first fact ‘metaphysically makes’ the second fact the case” (p. 70).

The concept of grounding can be applied to the Standard Model. For example, it can be argued that the aim of Searle’s “constitutive rule” is to give the conditions for grounding a social fact. Epstein maintains that for Searle, “the fact Billy is a dollar is grounded by the fact Billy was issued by the Bureau of Engraving and Printing” (p. 76). The latter fact is regarded by Epstein as a “frame principle” (p. 78). A frame principle fixes the “grounding conditions” for something to be a social fact in Searle’s sense (p. 78). For example, the social fact Billy is a dollar is grounded by the frame principle (which presupposes a collective acceptance about what to count as a dollar): “Billy was issued by the Bureau of Engraving and Printing” (p. 76). In Epstein’s view, the Standard Model is also about another relation known as anchoring. This is because this model explains how the social kinds, which are the categories used in the social sciences—categories such as dollar, class division, public good, commodity, and so on—are “glued.” In other words, this model also gives us an answer about why some physical objects are regarded as belonging to the same social kind (e.g., dollar). According to the Standard Model, social kinds “are glued by our ongoing attitudes toward those kinds” (p. 81). In Searle’s theory, they are glued together by the fact “The members of the community collectively accept the constitutive rule for the kind” (p. 81). Theories within the Standard Model that are about how social kinds are “glued” can be called anchoring relations. Those theories “are theories about the ‘putting in place’ relation that holds between a set of facts and the grounding conditions for a kind—in other words, between a set of facts and a frame principle” (p. 81). The difference between grounding and anchoring refers to two different roles individuals can play in making the social world: “Social facts can be grounded by facts about people, and frame principles can be anchored by facts about people” (p. 85).

Chapter 7 of The Ant Trap aims at clarifying the grounding/anchoring distinction by providing practical examples related to the theory of law. In contrast, chapter 8 focuses on the philosophical consequences of that distinction. With the grounding/anchoring distinction in mind, Epstein analyzes from an innovative standpoint ontological individualism, the Standard Model, and the relationship between the two. He argues that within social ontology, two different kinds of individualism must be clearly distinguished: “There is an individualism about grounding,” that is, ontological individualism that “is a claim about how social facts can be grounded” (p. 101). And there is also “individualism about anchoring” (p. 87). Epstein calls the latter “anchor individualism” and points out that it is “a claim about how frame principles can be anchored” (p. 101). As stressed earlier, for Epstein, a “frame principle” (p. 78) is a fact that presupposes collective acceptance and fixes the “grounding conditions” for something to be a social fact in Searle’s sense. Anchoring individualism argues that

all frame principles . . . are exhaustively anchored by facts about individual people. Searle’s collective acceptance theory is an example. Whenever a constitutive rule is anchored in a community, it is anchored by the “we-attitudes” of individuals in the community . . . His “collective acceptance” theory restricts anchors to a very limited set of mental states of individual people. (p. 103)

In other words,

Just as ontological individualism is the generic thesis that social facts are exhaustively built out of facts about individuals, anchor individualism can be the generic thesis that the grounding conditions for the social facts are exhaustively anchored by facts about the individuals. That is, it is a claim about a relation holding between a whole set of “social-level principles” and a whole set of “individual level principles.” (p. 105)

After clarifying the difference between ontological individualism and anchor individualism in the light of new conceptual tools, Epstein focuses more in detail on the first kind of individualism, which is “almost always analyzed in terms of supervenience” (p. 106). In his opinion, this individualism, as already pointed out, is “better understood as a thesis about grounding” (p. 106). Intuitively, ontological individualism is “a claim that social facts are exhaustively determined by facts about individuals” (p. 109). However, understanding ontological individualism in these terms is mistaken, because no such determination is possible. Ontological individualism “is a claim about the grounding of all the social facts” (p. 109). It “is best understood as a claim about dependence—about the dependence of all social facts on some set of individualistic facts” (p. 109). Dependence does not involve the “full grounding relation, but only involves partial grounding” (p. 107): “F depends on G: it is necessary that if F is the case, G partially grounds F” (p. 107). Epstein argues that “ontological individualism is a claim about what grounds what, rather than a claim about what supervenes on what” (p. 110). This is precisely because supervenience involves full grounding, while ontological individualism only involves partial grounding. A social fact cannot be described in terms of supervenience because there is “something else” that makes that “social fact obtain, other than the individualistic facts” (p. 111). This means that “if we change that other thing, the social fact will change, without having changed the individualistic facts” (p. 111). Social facts are not “metaphysically built out of the individualistic ones” (p. 112).

In chapter 9, building on the distinctions between grounding and anchoring and between ground individualism and anchor individualism, Epstein suggests that we conceive in a new way the metaphysical foundations of the social sciences. According to him, ontological individualism is a theory about grounding that, unlike what is argued by the dominant approach, must be disconnected from supervenience and ultimately rejected. This is because society is not composed only of individuals, and social phenomena are not based just on individual facts. While Epstein criticizes ontological individualism, he supports anchor individualism (e.g., Searle’s idea that there are collective attitudes toward the so-called constitutive rule), but he rejects the common view that anchor individualism presupposes ontological individualism. In his opinion, the first kind of individualism must be kept distinct from the second kind because the latter is flawed and the two perspectives are inconsistent with each other. According to Epstein, anchor individualism, which acknowledges the crucial relevance of the physical world, and ontological individualism, which regards the social world as purely individual, are not just distinct strategies: “they are at odds with one another” (p. 126). Both perspectives have been very successful in the philosophy of the social sciences. This is not surprising.

To many people, both ontological individualism and anchor individualism are appealing for the same reason. Both seem to deflate worries about dualism with regard to the social world. The social world is just us both theses hold. However, the two theses deflate these worries in conflicting ways. (p. 126)

The reason why grounding should be sharply separated from anchoring is that

it makes for a better model, and that it illuminates the project of social metaphysics. It also makes it easier to see the flaws with anthropocentrism about the social world. One reason these flaws have been so hard to see is that the two different kinds of individualism have been confused with one another. (p. 127)

Epstein’s attempt to rebuild the foundation of the social sciences is based on those assumptions.

3. Part 2: Groups and the Failure of Individualism

The second part of the book (chapters 10-18) applies the conceptual tools and terminology of the first part to explore the grounding conditions of social groups and thereby propose an alternative to the anthropocentric picture. Epstein aims at putting “ontological individualism to bed” (p. 129) and, on a positive note, suggesting a set of frame principles for real social facts about groups to show that social facts are ultimately not as anthropocentric as frequently assumed (p. 129). Groups for Epstein serve as one exemplary case to study the grounding conditions of social facts. This case is exemplary in that it illustrates the strong and persistent intuition that the social world consists only of individuals. Yet, for Epstein, the failure of ontological individualism holds for groups as it does for all social facts. To show how this is the case, he adopts a far broader definition of groups than other social ontologists such as Margaret Gilbert or Michael Bratman, who define a group as “a thing constituted by and only by individual people” (p. 133). Epstein’s definition is so broad that it seems to challenge one core but counterintuitive observation made later in the book, namely, that groups can exist even without any members. Now, what is his definition?

Epstein begins by adopting the metaphysical concept of “constitution” to outline his understanding of basic properties of groups. He defines group properties in terms of a constitution relation between social facts and those facts in which they are grounded to acknowledge that groups can persist despite changes in their composition, an observation that identifying groups by sets of people as traditionally done cannot accommodate. Epstein approaches this constitution relation from the coincidence view, which holds that while two objects (say, groups and individuals) are not identical to one another, they coincide. However, constitution is still different from spatial coincidence in that the constitution relation is asymmetric and irreflexive and in that sense similar to the grounding relation. As the latter, the constitution relation “involves a kind of hierarchy of ‘fundamentality’” (p. 142). Following Frederick Doepke’s proposal about constitution, Epstein specifies this hierarchy in terms of metaphysical reasons and a metaphysical explanation, which he spells out in turn as a core ingredient of constitution. Metaphysical explanations do not rest upon causal relations but rather identify those reasons that make something the case. “Constitution is about the stuff of one thing being part of the metaphysical reason that another thing is made up of the stuff it is” (p. 149). Those things that constitute some other things thereby partly explain specific facts about those things they constitute. More specifically, Epstein specifies two conditions for x to constitute y at time t: (a) x needs to coincide with y at time t and (b) certain facts about x need to partially ground some facts about y at t (p. 149). However, while x can metaphysically explain some facts about y, there are other facts about y that it cannot explain. This aspect allows Epstein to question the commonly accepted view that all facts about a group have to be exclusively determined by facts about individual members constituting the group.

Epstein’s goal is to show that not even simple social facts about groups—such as their existence at a particular point in time—are individualistically grounded. He wants to conclude that we can generally not assume that social facts are grounded in facts about the members constituting that group. To argue his point, Epstein uses his conceptual toolbox of grounding and anchoring and applies it to a variety of properties of social groups that social scientists investigate. Core examples are the existence, persistence, or change of groups in the context of changing circumstances (p. 136). Epstein’s point is that to appropriately model, for instance, changing social facts about the Supreme Court—such as its emergence, its age, its hierarchy, its powers, and obligations—the properties of its members will not help most of the time. Rather, we have to understand what grounds those social facts about the Supreme Court by identifying the frame principles that determine the grounding facts. Such an analysis then reveals, first, the diversity of grounding facts for facts about the Supreme Court and, second, how few social facts are indeed grounded in facts about its individual members. For example, most social facts about the Supreme Court are grounded in facts that are fully independent of the Supreme Court justices but are rather determined by facts laid down by the U.S. Constitution, which contains the rules and procedures determining the Supreme Court’s actions, preferences, jurisdiction, powers and obligations, conditions for individual membership, and facts about the U.S. Senate, among many others. Those external facts are largely independent of its individual members but have been anchored by lawyers and legislators, who have sorted out those grounding conditions for over two centuries (p. 153). This analysis is different from Searle’s account of constitutive rules in that Searle cannot acknowledge that different social facts about the same group—for example, its existence and its constitution—can be determined by distinct grounding facts. Epstein’s analysis in turn acknowledges the heterogeneous grounds of social facts because of his strong emphasis on frame principles. A frame principle is a principle that gives the grounding conditions for a social fact in all possible and not only for the actual world (p. 78). Frame principles for different social facts can take different forms and as such imply the diversity of grounding facts.

According to Epstein, considering in detail the grounding conditions for social objects also allows us to see how the social world is ontologically different from the natural world. The difference between, for instance, the Supreme Court, a social object, and ant colonies, a natural object, is precisely that facts about them are grounded differently. The specific difference lies in the way in which typical facts about social groups are grounded compared to how typical facts about ant colonies are grounded. Not only are ant colonies not characterized by complicated social facts, such as, for example, having a jurisdiction, obligations, powers, and rights; most facts about ant colonies, such as their existence, persistence, and constitution, only depend upon facts about the ants. As such, while most natural objects are intrinsically individuated, that is almost never the case with social objects. Social objects have heterogeneous grounding conditions, and which ones are anchored is highly flexible; we can anchor social facts “to have nearly arbitrary grounds” (p. 168). This makes the ontology of the natural sciences when compared to the social sciences “a walk in the park” (p. 163). That not every social fact fully depends upon a set of individualistic facts is also why supervenience fails as a concept to account for most social facts. And we also cannot speak of social facts as emergent, something he thinks we can do in the natural sciences, because we have to go beyond facts about individual components of a group to identify the full grounds of social facts of that group.

Of course, a key observation for Epstein’s project is that groups are not fully identified with the set of its members. “Not only can the membership of a group turn over again and again, but it is plausible that a group can persist even while it has no members at all” (p. 169). Groups such as the Supreme Court, for instance, are not constitutionally tied to its members. Unlike in the natural world, where a rock can be identified almost exclusively with its constituting material and its existence depends on it, a sports team or the Supreme Court, for instance, can persist even if all its members are replaced or leave for good. This issue is very closely tied to the metaphysical issue about the nature of groups. Are they concrete, if they can exist without members? Are they abstract, if they are constituted by people? (p. 170). Epstein suggests a set of conditions for identity that are independent from constitution, which is how he can identify something as a group, its change and persistence over time even without its having members. Those criteria, however, are independent from a grounding relation because those criteria do not offer grounds for facts about identity or persistence (p. 181). Rather, Epstein understands such criteria of identity to be convenient systematizations of frame principles. “They are consequences of frame principles for the existence of a group of some kind, together with frame principles for the constitution of a group of some kind” (p. 181).

Against this background, in the final four chapters (14-18), Epstein contrasts his theory with some representative positions in the social ontology literature on group attitudes, group action, and group intention. The line of argument is the same: opposing the common assumption underlying those positions, Epstein argues that because social facts have heterogeneous grounding conditions that can be anchored in multiple ways and because facts about individual members are just one set of such potential grounding conditions, properties about groups—be they their attitudes, intentions, or actions—cannot be entirely grounded individualistically. Epstein works through examples of multiple grounding patterns such as the property that a sports team is completely dressed in red at a particular time t is grounded in two kinds of facts. First, it is grounded in the fact that a group is constituted by a specific set of individual members at a particular time t and, second, the fact that the members of the set are completely dressed in red at a particular point in time, a pattern which Epstein calls perfect match (p. 204). But there are other grounding patterns for different kinds of social facts that are not perfect matches. For example, a sports team having the property of weighing 470 pounds at a specific time, is a property that does not fit the grounding pattern of a perfect match because not every member of the team weighs 470 lbs.; in fact, it is very likely that none of the members does.

Another example is a group having the property to be arranged in the shape of the letter Y (take a sports team again). Being arranged in the shape of the letter Y can also be a property of each individual member of the group. However, both properties have little to do with each other, as the property of the individual members is completely irrelevant for the property of the group. While certain intrinsic and spatial properties about the members do have to obtain for the group property to obtain—take for instance the location of each member—the property that each individual member is arranged in a shape of a Y is irrelevant for the group property. Groups can also have properties that depend on properties that are altogether different from the properties of its individual members. For example, the property that a committee is in session at a specific point in time t is grounded in the fact that the committee exists at this particular point in time and some other facts altogether, such as for example that there is a full moon at t. But it is not grounded in the property that all its members are individually in session at t. Finally, there are group properties that are independent from any properties of its members, such as, for example, that a specific sports team is two months old at a particular point in time t is grounded in the fact that the team was initiated earlier at a particular point in time t₀, that the team still exists at t and that t is two months later than t₀. The age of the sports team is thus independent of facts about its individual members.

While some social facts are grounded in facts about individual members, others can have nonmember grounds, which are “built deeply into the design of groups, to powerful effect” (p. 217). Psychological facts about groups such as, for example, that the U.K. Parliament intends that the Duke and Duchess of Cambridge’s first child will succeed to the throne, regardless of gender, or that a sports team intends to discipline a player for seeking a second medical opinion on his injured leg without permission (p. 198), are not just grounded by the attitudes of U.K. parliamentarians or the team members (p. 199). Now, take, for example, Bratman’s theory of group intention, which cannot accommodate for that view. Bratman takes individual intentions to be insufficient and unnecessary for group intention but holds that individuals have to intend to accommodate one another in the group’s performance of j (Epstein 2015, 201). They intend individually that the group does J, that they hold specific intentions, namely, that they intend to mesh their subplans toward achieving J, and that all group members have common knowledge of other members holding those intentions (p. 201). Epstein’s criticism now is that while Bratman does not decompose the group intention into the individual intentions, he decomposes the group property into a very similar set of individual properties, “plus a bit of knowledge” (p. 203). As such, Bratman does commit to an ontology of group attitudes that is exhaustively grounded by member attitudes. Against this background, Epstein takes up the challenge to show that the analysis of group intentions, as one apparently obvious example for anthropocentrism, fails to exemplify that psychological facts about groups are grounded in psychological facts about individuals. According to Epstein, ignoring nonmember grounding conditions leads to a narrow or even false picture of group intentions.

In the last two chapters, Epstein goes back to ask again about the nature of social groups, discussing Social Integrate Models brought forward by Margaret Gilbert, Raimo Tuomela, and Bratman. While appreciating the differences between all three accounts, Epstein details his arguments against the underlying idea of such models that all social facts about groups are grounded in the social integration of group members and their attitudes (p. 257). His argument picks out the observation that in all three accounts, groups are identified by their sociality, which is in turn grounded solely upon facts about group members, be they, for instance, shared commitments (Gilbert) or joint attitudes (Bratman) toward a common goal. The problem with Social Integrate Models is thus that they rest only on one set of grounding conditions for a group, namely, those facts about their members that determine the conditions for sociality. “What explains the existence and unity of a social group [in those models] is that the members join forces, in one way or another” (p. 259). For Epstein, this is naturally a shortcoming. According to Epstein, individual commitments can neither determine all group facts nor can group agency determine all social facts. For instance, those models are limited when applied to groups characterized by internal hierarchies or power differences such as power differentials among stockholders of a company, which cannot be accounted for by referring to shared attitudes or commitments among group members. Because those models treat “groups as if they were sophisticated ant colonies” (p. 260), they furthermore fail to identify the diversity in grounding conditions for a group’s existence, identity, and persistence that Epstein’s project is meant to reveal and illuminate. Finally, some groups—such as courts or legislatures—are not well integrated but are nevertheless unified; they intend and act. In short, for Epstein, member integration alone is neither a necessary nor a sufficient condition for group agency (p. 262).

Epstein puts forward a similar line of argument against what he calls Status Models of social ontology that have been prominently defended by John Searle, Raimo Tuomela, and Frank Hindriks. Such models ground group agency or other social facts on the idea that some substrate gets externally assigned the status of being an agent as well as other social facts or group properties. On first sight, that seems to be a step forward because those models allow for the idea that social properties about groups can be determined not only by members of the group and their properties but also by other individuals external to the group, such as members of a larger community. For instance, for something to count as a university, it is not sufficient that members of the university socially integrate but rather that a collective agent gets assigned the status of a university by outsiders, that is, people who are not part of the university (p. 268). What is more, those models acknowledge that groups for that matter can have varying substrates that get assigned group status and can thereby acknowledge a larger set of groups than Social Integration Models. However, this advantage is also the downfall of Status Models; they ultimately go down the wrong track for Epstein and we can see why when we consider how status assignment works. We do not think of groups as having agency, intentions, and taking action just because we assign social status to some substrate (p. 265). Whether or not something has intentions or agency seems to depend upon the kind of substrate to which this status is assigned. The Searlian approach assigns the status of a boundary to a set of stones via constitutive rules that are in turn collectively accepted by a community, but we cannot easily assign the same set of stones the status of a group. That would not do justice to what we consider agency to be. For Epstein, group agents—corporate or otherwise—have to realize a system of practical activity, that is, a system that also encompasses reasoning, planning, and responsibility (p. 218). This system in large parts determines their agency. How agents go about realizing their system of practical activity determines in turn how the grounding conditions are anchored. What grounding conditions we anchor by way of frame principles can vary widely, and “it is not only properties of members that go into this realization” (p. 264). “We can anchor hierarchies, jurisdictions, deliberation procedures, membership conditions, assignment of power,” and so on (p. 272). But anchoring facts are not to be mistaken for assignments of agency. Anchoring facts are rather the requirements for the frame principles that determine which facts ground properties of a group. When we understand agency as “a matter of the group satisfying the grounding conditions that are anchored by the way groups of that kind realize a system of practical activity” (p. 272), only then do we see how widely the realization of that system can vary across kinds of groups. Epstein’s framework of anchoring, framing, and grounding allows for acknowledging this diversity in grounding conditions. The assignment of status in turn does not add anything, as frame principles do not require status assignments.

4. A Set of Remarks

The Ant Trap is an enormously rich book and an important contribution to social ontology and metaphysics. It is critical in rejecting the dominant individualist ontology and ambitious in that it aims at entirely rebuilding the metaphysical foundations of social research. Its highest merit is that it is extremely innovative in its methods while at the same time it engages in a long-standing yet stagnant debate about the proper foundations of the social sciences. As such, The Ant Trap provokes us to think in fundamentally new ways about the foundations of the social sciences. Epstein admirably extends and applies a set of metaphysical tools to metaphysical and ontological questions in the social sciences, taking a fundamentally new perspective on questions in social ontology. By rejecting holism and offering a new set of arguments against nominalism, Epstein furthermore shifts the attention away from the conventional debate about individualism versus holism, as it currently persists in philosophy of the social sciences. Given the current interpretation of individualism by analytic philosophers, this debate has largely become a discussion about reductionism and anti-reductionism in the social sciences (Di Iorio 2015, 103-115; 2016a; 2016b). As such, the debate does not do justice to its original motivation as a debate that engaged methodologists, philosophers, and practicing social scientists. Furthermore, very few social scientists are in fact methodological reductionists. Epstein stresses that too little useful research has been done to understand what the social world is made of, which has important negative implications for the way in which social scientists approach the social world with their models and theories. In order to model the social world appropriately to explain, predict, and intervene in it, we have to understand what it is made of. And pointing this out is by itself a welcome contribution. However, there are at least two aspects with regard to which we find The Ant Trap wanting: first, its nonexisting engagement with scientific practice, and second, its denial of a long tradition of authors who elaborated extensively on how we can make sense of the social world while committing to some form of individualism.

First, Epstein ultimately does little to make the debate more relevant for the social sciences. Despite making a welcome contribution to the metaphysics of the social sciences, Epstein does not clearly meet his goal of making a philosophical contribution that is relevant for the practicing social scientist (cf. Ahmed 2016). The motivation for Epstein’s enterprise of identifying the grounding conditions for social facts is to ultimately improve modeling practices in the social sciences. Epstein seems to be convinced that to model something in the social sciences, such as, for example, the occurrence of the financial crisis as a social fact of the financial market, we need a solid ontological understanding of the kind of thing we are modeling. We get a proper understanding of a social fact by identifying the grounding conditions for that social fact. As Epstein argues, because those conditions are highly diverse, we base our models upon false assumptions if we do not acknowledge this diversity. However, it remains unclear what concretely this entails for the social scientist. For a start, Epstein does not discuss a single model from the social sciences to illustrate, first, what social scientists do and where their current models exactly fail and, second, how their modeling practices would improve by acknowledging the refined ontology that Epstein’s project on a new social metaphysics lays out. How does the distinction between anchoring and grounding affect our current modeling practices in macroeconomics, for example? And how would we produce models that would have better explained and predicted the financial crisis? Epstein remains short of a concrete answer to those questions.

Second, Epstein’s attempts to propose a new nonanthropocentric metaphysics for the social sciences is based on the assumption that ontological individualism, understood as the thesis that society is entirely composed and determined by individual properties, is false because of the ontological and explicative relevance of nonhuman and nonindividual factors in the social sciences. Because of that assumption, Epstein criticizes methodological individualism and the Standard Model, both of which he regards as committed to ontological individualism as defined above. As he stresses, the Standard Model, although it considers itself alternative or oppositional to methodological individualism, is built on the acceptance of the nominalist metaphysical presuppositions of the latter.

While Epstein locates the origins of all the problems affecting the social sciences in methodological individualism and its ontology and wants to get rid of the entire individualist tradition in the social sciences, he focuses more on debates that developed within analytic philosophy in the last few decades (e.g., supervenience, collective intentionality, grounding) rather than on the wider individualist tradition. One problem is that many of Epstein’s assumptions about methodological individualism and ontological individualism are historically incorrect, which weakens his conclusion that the metaphysical foundations of the social sciences must be entirely rebuilt. First, Epstein seems to neglect that not all methodological individualists supported ontological individualism. For instance, Raymond Boudon, one of the most eminent advocates of the individualist sociology, openly rejected nominalism and tended to disconnect methodological individualism from nominalism (for more details, see Boudon 1996, 278-79). Moreover, in following Lukes and the dominant positions within analytic philosophy, Epstein equates methodological individualism with the theory that social explanations have to semantically reduce social properties to individual ones. This is problematic. While there is a reductionist variant of methodological individualism that is employed by the social contract theory and some economists, there is also a nonreductionist variant. Eminent methodological individualists such as Carl Menger, Max Weber, Georg Simmel, Friedrich August von Hayek, Karl Popper, and Robert Merton openly rejected reductionism in Lukes’s sense and developed an emergentist and systemic theory of the social world based on the concept of unintended consequences of human action—a concept that rules out the possibility of reducing social properties to the agents’ mental and behavioral properties (see Boettke and Candela 2015; Boudon 1971, 2013; Bouvier 2011; Demeulenaere 2011; Di Iorio 2015, 2016a, 2016b; Di Nuoscio 2017; Jarvie 1972, 2001; Manzo 2014; Nadeau 2016; Petitot 2016). The historical inaccuracy of Epstein’s analysis of methodological individualism is also shown by the fact that, while he accepts Lukes’s interpretation of the concept in terms of explicative reductionism, he argues that, among the originators of emergentism, as opposed to reductionist individualism, must be counted thinkers such as Bernard Mandeville and Herbert Spencer (Epstein 2015, 26). However, Mandeville and Spencer, as highlighted by an abundant literature, were paradigmatic nonreductionist methodological individualists who influenced the most recent advocates of that approach and their emergentist invisible-hand model of explanation (see, for instance, Boudon and Bourricaud 1990; Di Nuoscio 2016; Dupuy and Dumoucel 1983; Hayek 1948; Laurent 1994; Petitot 2016). It is well-known that famous anti-reductionist methodological individualists such as Hayek, Boudon, and Elster counted Mandeville and members of Scottish Enlightenment among their precursors.

The most problematic aspect of The Ant Trap is the way it deals with the concept of ontological individualism, of which Epstein provides a caricatured interpretation. Epstein assumes that ontological individualism is the thesis that society supervenes on individual properties in the sense that society is entirely composed and determined by individual properties (p. 36ff.). He rejects this thesis because it neglects the ontological and explicative relevance of nonhuman factors in the social world. According to Epstein, the social world cannot solely be grounded in individual facts. However, methodological individualists never conceived ontological individualism as the simplistic and obviously mistaken view that the social world is exhaustively determined by individual properties. Following the ancient nominalist tradition, methodological individualists opposed the view called ontological holism or realism, that collective nouns such as capitalism, class, society, state, and church refer to entities that have real existence and must be treated as substances that exist independently from individuals, such as a tree or a stone (see Hayek 1952; Mises 1949; Popper 1957, [1945] 1966a, [1945] 1966b). As understood by methodological individualists, nominalism maintains that collective nouns are nothing but convenient ways of talking in the sense that social wholes are not sui generis (i.e., independent) substances that have real existence and of which the individuals and their actions are simple derivatives (see also Antiseri 2016; Dawe 1970; Di Iorio 2015, 2016a, 2016b; Di Nuoscio 2016, 2017; Hayek 1952; Laurent 1994; Nadeau 2016; Petitot 2016; Popper 1957, [1945] 1966a, [1945] 1966b; Pribram 2008). The difference between supporting the view that social entities do not exist as real substances and supporting the view that social entities are exhaustively determined by individual properties seems to escape Epstein. Methodological individualists did not defend a simplistic ontology that denied the relevance of nonhuman factors (see Bouvier 2015). They only argued that the explanations of social processes, including the analysis of the influence of physical factors on social phenomena, must avoid any hypostatization of social wholes. The problem with Epstein’s interpretation of the nominalist standpoint supported by methodological individualists is confirmed by countless examples of empirical explanations taken from their works. For example, the obvious truth that natural factors such as meteorological conditions or plant diseases can affect the price of agricultural products or cause famines has been well-known for centuries by the individualist economists (see Bouvier 2015, 574).

Ontological individualists such as Weber, Hayek, and (arguably) Popper wrote extensively about the relationship between methodological individualism and the analysis of the ontological and causal relevance of nonhuman factors in social life. According to those authors, methodological individualism assumes that nonhuman factors play a crucial role in social life, but also that the analysis of that role must be linked to an interpretative perspective (i.e., to what the German tradition calls a Verstehen approach). According to Hayek (1952, 25), the social sciences, which must be based on a strictly nominalist and individualist method, “deal . . . with the relations between men and things and the relations between man and man.” For Hayek, methodological individualism, understood as an approach committed to nominalism, does not deny the influence of physical factors on social life, but rules out any possibility of explaining that influence in “objectivist” and mechanical terms (pp. 25-27), that is, without referring to the way the individuals interpret their environment (p. 30; see also Hayek 1948, 57ff.). Similarly, according to Weber, methodological individualism assumes that social phenomena can be produced by various and concomitant causes, some of them external to the human mind. For Weber (2012, p. 50), methodological individualism “does not confine itself to the internal aspect,” that is, to mental processes, “but conceives the whole historical constellation of the external world” as potentially causally relevant (p. 50). For example, according to methodological individualism, the “empirically founded conclusions of psychopathology and the laws of psychophysics are only relevant for history in exactly the same sense as physical, meteorological and biological knowledge” (p. 53).

Put differently, methodological individualism “does not deal with the internal processes . . . of human beings for their own sake; instead, it is concerned with the “external” conditions and effects of the way in which human beings relate to the ‘world’” (p. 53). Because of its interpretative presuppositions, methodological individualism is, for Weber (p. 53), “always in a specific sense ‘anthropocentric.’” However, this is not because, as suggested by Epstein, methodological individualism denies the ontological and causal relevance of nonhuman and nonindividual factors in the social world, but precisely because, as an interpretative approach, methodological individualism attaches causal relevance to physical influences only insofar as they affect the understandable manifestations of human life. Consider “the significance of the Black Death . . . for social history, or the significance that the invading waters of the Dollart . . . had for the history of the colonization movement etc.” (p. 35). According to Weber, “both events are in absolutely no respect different from the invasion of Germany by Gustavus Adolphus . . . or the invasion of Europe by Genghis Khan” (p. 36). This is because, from the interpretative standpoint of methodological individualism, “all those events” have had effects “historically significant” (p. 36) in the sense that they affected social life from a scientifically relevant perspective.

The ontological and causal relevance of the physical world has also been discussed by another eminent methodological individualist: Karl Popper. Epstein, who criticizes Popper as the supporter of an anthropocentric ontology, seems to ignore his refined and anti-anthropocentric analysis of the causal relationship between the physical world and the social world. In discussing the relationship between mind and body and his 3 Worlds ontology, Popper ([1977] 2003, 38) rules out that causality is exclusively mental. In his opinion, the relationship between mind and body must be explained in terms of an interaction among three worlds: (a) the physical world or World 1, (b) the world of mental states or World 2, and (c) the world of culture or World 3. According to Popper ([1977] 2003, 47), there is a mutual influence among those three worlds which interact and are capable of influencing each other (Popper ([1977] 2003; see also Di Iorio (2016b) and Jarvie (1972, 2001).

In sum, The Ant Trap is surely of interest from the standpoint of the very recent debates on the ontology of the social sciences in analytic philosophy (see, for instance, Gallotti 2016; Guala 2016). Epstein provides innovative, refined, and complex conceptual tools that can be usefully applied to those debates, but, because of the reasons underlined above, from the viewpoint of a scientific practitioner and the broader history of the social sciences, his attempt to rebuild the foundations of those sciences and provide a new nonanthropocentric metaphysical framework for social research as opposed to ontological individualism is problematic. Epstein’s mistaken interpretation of ontological individualism as traditionally understood by methodological individualists such as Weber, Hayek, and Popper, whom he considers the precursors of the most recent variants of individualism, weakens his criticism of the Standard Model developed by Searle and others. The commitment of the Standard Model to ontological individualism does not necessarily imply that this model is based on a naively anthropocentric metaphysics. All depends on what is meant by ontological individualism. The same kind of criticism may be applied to Epstein’s reflections on the grounding conditions of social groups. It is historically untrue that, according to ontological individualism, group facts are exhaustively determined by individual facts. As a consequence, statements such as “groups can persist despite changes in their compositions,” “most social facts about the Supreme Court are grounded in facts that are fully independent of the Supreme Court judges,” and “the age of this sports team is independent of facts about its individual members” are consistent with ontological individualism (as understood by methodological individualists) and do not need to be accounted for on the basis of a different ontology.