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


Replies to Guala and Gallotti

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 159–172, ISSN (Online) 2196-9663, ISSN (Print) 2196-9655, DOI: https://doi.org/10.1515/jso-2016-0002, March 2016

Publication History

Published Online:


This article responds to comments by Francesco Guala and Mattia Gallotti on The Ant Trap. In the replies, I address the relation of new advances in cognitive science to the study of collective attitudes, clarify distinct questions we might ask about grounding and about anchoring in social ontology, defend various forms of pluralism about grounds and about anchors, and discuss the type-token distinction as it applies to social entities.

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

1 Introduction

Let me begin by expressing my deep appreciation to the commentators for their careful work and insights on the book, and to Arto Laitinen and Francesca Di Lorenzo for organizing this symposium and for their helpful comments. It is a true privilege to have colleagues like these, with whom to collaborate and to debate. I also want to take the opportunity to congratulate Natalie Gold on the birth of her daughter on the very day of our symposium in Palermo! We were all disappointed that Natalie could not participate, but nonetheless are thrilled for her and her family.

These replies are divided into four parts. First, I comment on Gallotti’s argument that The Ant Trap inappropriately overlooks recent work on social cognition in psychology and cognitive science. Then I clarify some different questions that Gallotti and Guala both discuss, and turn to the notion of “pluralism” versus “monism” regarding grounds and anchors. Interestingly, both commentators speak of these, despite the fact that I do not use these terms in the book. The problem with them is that they are somewhat ambiguous, so we need to clarify them before we can debate about them. Then I discuss “ground pluralism”: clearing up the notion is enough to dissolve the disagreement about it. Finally I discuss “anchor pluralism”. I take issue with Guala’s claim that a type-token confusion is present in the model or the book, and instead clarify exactly how we can think about types and tokens in social ontology. And I highlight the real disagreement between Guala and me; I argue that Guala has too narrow a view of institutions and other social kinds and of how we set them up. I then comment on Gallotti’s concerns about anchor pluralism.

2 Cognitive Science and Collective Attitudes

Gallotti argues that we need to pay closer attention to the analysis of collective attitudes, both for their own sake and for developing a clearer understanding of social facts more generally. In particular Gallotti points out that I spend a fair amount of attention in The Ant Trap criticizing analyses of collective attitudes, and yet focus my criticism on the work of Gilbert, Bratman, Searle, Tuomela, Pettit, and List. In the last several years, however, cognitive scientists have made strides in improving our understanding of social cognition. I do not address this cognitive science literature in The Ant Trap, a gap which Gallotti regards as detrimental to the project.

Although Gallotti is right that I do not discuss this part of the literature, I am puzzled to see him assert that The Ant Trap endorses instead a “traditional and caricatured view of ‘we-attitudes’.” After all, Chapters 14–18 are devoted to criticizing the basic assumptions that underlie the traditional views. Indeed, this is why I did not think it was necessary to address the recent work in psychology and cognitive science: they rely on the same assumptions that I criticize in traditional views.

In writing the book, I chose to focus on the older views, since they are more widely understood and better worked out. Still, it might have been useful to make it explicit how the same arguments also apply to the recent work in psychology and cognitive science, so I am pleased that Gallotti has raised the issue.

The arguments regarding collective intentions in the book are quite general. A central point I argue is this: the collective intentions of a group G do not in general supervene on mental states of members of G. Take all the facts about the minds of all the individual members of G. That includes whatever mental states you like: states of knowledge, belief, desire, intentions, plans; attitudes in the I-mode and attitudes in the we-mode; dynamic and interactive and relational attitudes; mental states individuated according to an internalist or an externalist account; or whatever else you like. Even with all of those, they are insufficient to determine the typical collective attitudes of a typical group.

The reason is that group or collective attitudes are grounded by facts having to do with lots of other factors entirely apart from group members. It is not simple to see this or argue it: it is a product of how facts about groups are grounded in general, and of the kinds of functional roles that group attitudes are anchored to play. But once you notice the possibility that group attitudes are grounded by much more than the attitudes of group members, you start seeing how ubiquitous it is. This is what I argue in the chapters on group actions and group intentions.

The argument does not depend on any of the details of either traditional or new views of those mental states. The problem I raise with those analyses of collective attitudes has nothing to do with which mental states are employed in the analysis or what role they play. Rather, the problem lies in the basic assumption baked into all of those theories: that we are to analyze the collective attitudes of a group in terms of properties of the group members.

To develop a fuller account of the grounds of collective attitudes, we need to look outside of the minds, bodies, and practices of group members. As I discuss in those chapters, we need to think about how groups in general function in the performance of practical activity. Collective attitudes are a part of that functional system. A crucial part of the design of groups is to lessen the cognitive burden on group members. What makes groups work over time – what allows them to perform their practical activities – is that there are non-member facts in place that do part of the work in grounding various facts about the collective, including collective attitudes.

There are many reasons traditional analyses have overlooked this. They have employed an overly simple paradigm of collectives – they tend to work on small groups, in which the members all have accommodating attitudes toward one another. They also have used simplistic pictures of how groups are constituted, and how group properties can be built in general. All these have led to the erroneous assumption (or in some cases the erroneous argument) that group attitudes must supervene on the mental states of members.

Now, the same erroneous assumption is also made by recent work in psychology and cognitive science. There is no doubt that this recent work is making strides in understanding the cognition of sociality. Advances in this field have influenced how I personally think about cognitive states: at one time, it seemed absurd to me that humans have distinctively social cognitive states. My view has changed on this. I think the jury is still out, but at this stage I regard it as more likely than not, and wonder why I ever thought it was absurd.

Yet that does not change the point. Even if we do have a complex and social landscape of mental states, they still are insufficient to ground collective attitudes. However, nuanced a picture we develop of attitudes – social, relational, dynamic, and so on – the same point remains. The reason I do not challenge the new cognitive science work is that to do so would be redundant. It is insufficient to account for group attitudes, and that is so for exactly the same reasons that I argued in connection with the traditional views.

For this reason, I am somewhat worried about a shift I see some evidence of in the field of collective intentionality: the center of gravity seems to be moving away from philosophers and toward psychologists and cognitive scientists. By profession, psychologists and cognitive scientists study the mind. It is not their job to consider the variety of extramental factors that go into grounding collective attitudes. Understanding the social dynamics of mental states is important work, but I hope that they and others will realize that mental states are only a part of the story in the building of collective attitudes.

3 Clarifying the Questions and Topics

The question I just discussed is about grounding: what are grounds of facts about collective attitudes? Gallotti also ties questions of collective attitudes to “anchor pluralism”; and Guala comments on both “ground pluralism” and “anchor pluralism.” To approach these, let me begin by clarifying some different questions on grounding and anchoring, and on the places collective attitudes can show up in the grounding–anchoring model.

3.1 Questions about Grounding

Starting with collective attitudes, we might ask two different questions about grounding. First is the question we were just discussing, What are the grounds of facts about collective attitudes? Suppose, for instance, the following fact obtains in the actual world:

  • (F1)

    The board of directors of Microsoft intends to acquire Nokia.

F1 is a particular fact about a collective intention. What grounds that fact? That is, what other facts actually obtain, that are the metaphysical explanation for F1 obtaining? Is F1 exhaustively grounded by the attitudes of the board members? If so, which attitudes? If not, by what else? Both “traditional” theorists and the cognitive scientists Gallotti cites are addressing this question. And this is also the topic of Chapters 14–16 of the book.

A different question is the converse, What do facts about collective attitudes ground? Instead of inquiring about the analysis of collective attitudes, this question asks about what facts collective attitudes are part of the analysis of. One example given by certain theorists is “sociality,” or “genuinely social groups” of a certain sort. Sociality, it is sometimes argued, is grounded by group members having certain collective attitudes.

Both of these questions are about how specific kinds of facts are grounded. We can also ask more general sorts of questions: How are social facts grounded in general? Are all social facts grounded in the same way, or according to the same template? Some theorists argue, for instance, that all social facts are grounded by individualistic facts of a certain kind.

This is the topic of Chapters 11–13. These chapters argue that typical social facts are grounded by a very heterogeneous set of facts. And they also argue that different social facts are grounded in very different ways from one another. In the book, I address the general case before the specific case of collective attitudes. I do this so that we can use some of the apparatus and insights of the general work to help clear up misconceptions about the grounding of collective attitudes.

The topic of “ground pluralism” pertains to this last question. Below I will discuss how to interpret Guala’s use of this term. In the end, I will argue that I am pluralistic about grounding in every way I can make sense of, and Guala’s worry about this is misplaced.

3.2 Questions about Anchoring

We can also ask specific questions about anchoring in connection with collective attitudes, and also questions about anchoring in general. Again there are two sides to each question. With respect to collective attitudes, there is the question about why they have the grounding conditions they do: What anchors the grounding conditions of facts about collective attitudes? One approach is that they are anchored, at least in part, by the functional role that collective intentions play in the practical activity of groups. On this approach, in order to explain why facts about collective attitudes are grounded the way they are, we need to think about the functional roles in a system of practical activity.

Again we can also ask the converse question, What role do collective attitudes play in anchoring? Are the grounding conditions for x is a dollar, for instance, anchored by collective attitudes? Searle thinks that all constitutive rules are anchored by a particular kind of collective attitude. Is this the way anchoring works in general? Are collective attitudes all there is to anchoring? Are they irrelevant to anchoring? Are they some but not all of the anchors of social kinds? These are the questions Gallotti considers in his discussion of “anchor pluralism”.

These are specific questions about the place collective attitudes play in anchoring. Here too we can also ask a more general question: How are frame principles anchored in general? Are all frame principles anchored in the same way, or according to the same template? Searle, for instance, argues that all constitutive rules (his narrow version of a frame principle) are anchored by collective acceptance of the rules themselves. This is what Guala discusses in connection with “anchor pluralism.” He argues that “institutions” are anchored in just one or two ways.

As I mentioned, I actually do not discuss ways of anchoring in the book. I do give examples of heterogeneous anchors, but defer the discussion of theories of anchoring to other places, since it is a huge topic on its own. So I find it a little surprising that both Guala and Gallotti discuss this extensively in their comments. Still, Guala is correct that I endorse anchor pluralism, properly understood, so we have a genuine disagreement. Interestingly, where Guala thinks I am too pluralistic about anchoring, Gallotti thinks I am not pluralistic enough. He worries that I rule out a role for collective attitudes in anchoring. (As I discuss below, I do not rule that out, so he should not be worried.)

4 Guala on Ground Pluralism

Guala speaks of two claims, “ground pluralism” and “anchor pluralism.” I think it is reasonable to take my views of both to be pluralistic, but it is also important to be clear just what we mean by pluralism. Oddly, Guala uses the term in one sense when applying it to grounding, and in a different sense when applying it to anchoring. So we need to sort out what is meant by ground pluralism and anchor pluralism, and see in what ways it makes sense to be pluralistic.

Guala’s characterization of grounds and anchors is clear and accurate. We can think of the anchors as setting up the way that a given social fact is grounded. So, for instance, the clergy in a society might decree that certain animals slaughtered in certain ways are kosher, or halal. The declarations of the clergy set up the conditions for a fact of the form x is kosher or x is halal. Those conditions might include things like x is not derived from a pig, or x is slaughtered in such-and-such a way. In this case, the non-pig-ness and the method of slaughter are among the grounds for being kosher or being halal. The declarations of the clergy are among the anchors. Alternatively, if it is God’s decree that sets up the conditions for being kosher, or being halal, then it is God’s decree that is the anchor. The grounds remain the same.

4.1 Interpreting Ground Pluralism

The example Guala gives of ground pluralism is the dollar and the euro both being money and yet having different grounding conditions. The fact x is a dollar is grounded by the fact x was issued by the Bureau of Engraving and Printing; while the fact x is a euro is grounded by the fact x was issued by the European Central Bank. (Guala and I agree that this is a simplification of the actual grounding conditions, but it is fine for illustration.) Guala says that I endorse a form of “ground pluralism” for the kind money.

In general, I can think of three sensible ways we might interpret “ground pluralism,” but I have a hard time making sense of which of these (if any) Guala might mean. Here are the three:

  • (GP1)

    Some social facts have heterogeneous grounds.

  • (GP2)

    Some social facts do not have unique grounds.

  • (GP3)

    Different social facts are grounded in different kinds of ways.

In my view, all of these are true. For GP1, I argue extensively in the book that social facts have very heterogeneous grounds. For instance, the fact Ruth Bader Ginsburg is a Supreme Court Justice is grounded by a variety of facts about her, about the President, about the Senate, and about her inauguration.

GP2 is also straightforward: social facts do not typically have unique grounds. This is true for pretty much all facts, and the reason is that grounding is transitive. Suppose you have a fact F that is grounded by fact G. And G in turn is grounded by fact H. Because of transitivity, F is grounded by H. So G is not the unique ground of F. H is a distinct ground.

And GP3 is also true: different social facts are grounded in very different ways. Facts about Supreme Court justices are grounded in different ways than are facts about rioting mobs, and those are grounded in different ways than are facts about kosher meats.

It seems that Guala has something different in mind. Something along the lines of the view that a single kind might not just have non-unique grounds, but also distinct and incompatible grounds. To be precise, we should talk about the grounds for facts, and about the instantiation conditions or membership conditions for kinds. So Guala’s idea seems to be that one kind – money – has different instantiation conditions in Europe and in the US.

Later on in his comments, Guala rejects this form of “ground pluralism.” The thing is, he does so for basically the same reasons I would: it is not really coherent. There is no one kind that has multiple incompatible instantiation conditions. Each kind has its own instantiation conditions. Guala mentions three different kinds – the kind dollar, the kind euro, and the kind money. And I already gave the respective instantiation conditions for dollar and euro; Guala goes on to give the instantiation conditions for money as serving as a medium of exchange, a store of value, and a unit of accounting. And Guala points out that this is merely a different kind, with its own grounds.

Now, Guala puts the point in terms of types and tokens, which I will object to just below. It is not clear what Guala means when he says that it only seems that ground pluralism is true because of a confusion of these. I will suggest that a better way for Guala to express his own view is to take money to be a particular sort of causal-role-kind and to take dollar and euro are realizer-kinds of that role-kind, or to take money and dollar as standing in a determinable/determinate relation. But either way, it does not affect the point. Money, dollar, and euro are three distinct kinds, each with its own set of instantiation conditions. All the talk of ground pluralism in a sense distinct from GP1, GP2, or GP3, is a strawman.

(I should also add that in subsequent communication, Guala pointed out that that some of the confusion arises from Searle’s formulation of constitutive rules. Searle’s rule is Bills issued by the Bureau of Engraving and Printing count as money in the US. Guala is correct to object that this makes it unclear whether Searle is giving the conditions for a bill to be a dollar or to be money in general. This is one of many reasons why I replace Searle’s formulations with more precise ones in the book.)

4.2. The Ambiguity of Certain Counterfactuals

Another argument Guala gives regarding grounding involves the interpretation of certain counterfactuals. Consider a counterfactual conditional such as If the Geneva Convention had not been enacted, then Genghis Khan would not be a war criminal. Guala points out that there are two readings of such counterfactuals. There is a natural reading on which it is false: the standards for being a war criminal are what they are – i.e. having committed certain atrocities in the context of war – across all contexts and worlds. Given the real definition of being a war criminal, we can evaluate whether someone meets those conditions in any world or context – even in ones where the “setup” facts do not obtain. This makes the counterfactual false. He would still have been a war criminal even if the world had ended in the year 1500.

Guala points out, however, that there is also a natural reading on which the counterfactual is true: Khan is a war criminal according to certain standards that were set up by the Geneva Convention. So if there were no Geneva Convention, those standards would not be in place, and he would not be a war criminal.

I actually discuss this in Chapter 9, so will only comment on it briefly. I agree that there are two readings, and I point out in that chapter that the grounding–anchoring distinction is an excellent way to make sense of those readings. The way a counterfactual is often understood is that the antecedent (the Geneva Convention was not enacted) moves us to a different possible world, the nearest one where the antecedent is true. Then we evaluate whether the consequent also holds in that world. According to the grounding–anchoring model, there are actually two things that the antecedent might do. It might shift us to a different world, but still in the same frame. (That is, where the grounding conditions for being a war criminal are anchored by the Geneva Conventions.) Or it might shift us to a different frame, where the grounding conditions are different. (A similar point regarding counterfactuals is made by Einheuser 2006.)

Now, as I also note in Chapter 9, this explanation of the counterfactual is not decisive evidence in favor of the grounding–anchoring model. The same ambiguity that Guala notices can be accommodated in many different models. Still, if Guala is right about this ambiguity, the grounding–anchoring model explains it in a natural way.

5 Gallotti and Guala on Anchor Pluralism

As applied to anchoring, the interpretation of pluralism is clearer, as are the commentators’ disagreements with me. Here too, there are three sensible interpretations, and both Guala and Gallotti clearly mean the third:

  • (AP1)

    The grounding conditions for some social facts have heterogeneous anchors.

  • (AP2)

    The grounding conditions for some social facts do not have unique anchors.

  • (AP3)

    The grounding conditions for different social facts are anchored in different kinds of ways.

I regard all three of these are true. Again, I do not actually argue these theses in the book, but my views on them are straightforward.

With respect to AP1, I do give several examples in the book of frame principles with heterogeneous anchors. In Chapter 7, for instance, I discuss the anchoring of laws. Laws are a particular kind of frame principle; for instance, if x killed a person with deliberately premediated malice aforethought, then x is a first degree murderer. In the United States, laws like this are anchored by a variety of facts, including the enacting of statutes, the interpretations of judges, the record in the legal code, and more.

AP2 is straightforward for a similar reason that GP2 is. Take a given set of anchors A. Typically, the facts in A will have grounds; call those grounds H. The idea of grounding is that in terms of the work they do, they are basically redundant with A. In other words, the following principle is plausible: Suppose that A anchors frame principle R, and H grounds A. Then H also fully anchors frame principle R. (Note that I am not saying that anchoring is transitive; the point is just that if H grounds A, then A is effectively replaceable by H in anchoring R.)

AP3 is the claim that Guala and I most clearly disagree on. Guala is right to understand me as claiming that there are many kinds of “glue that hold social kinds together.” That is, there are different ways of anchoring the frame principles for different social kinds. Kinds like kosher, first degree murderer, group agent, chair, dollar, corporation, and so on each have their own frame principles, and those frame principles are anchored very differently from one another. This is what Guala disagrees with.

To me, there is a close connection between AP3 and AP1. Once we see how heterogeneous the anchoring facts are for certain common frame principles, it becomes obvious that there is not just one way we anchor kinds.

Apparently, this is not as obvious to everyone else! Guala is correct that “anchor pluralism” in the sense of AP3 is radical and far-reaching, at least when set against the prevailing theories in social ontology. John Searle, for one, is an anchor monist in this sense. Searle argues that there is precisely one way for constitutive rules (again, his simple version of a frame principle) to be anchored: namely, by collective acceptance of that rule. Guala endorses a different version, which is something close to anchor monism.

5.1 Guala on Tokens and Types

Guala denies anchor pluralism with the argument that The Ant Trap overlooks the distinction between tokens and types. To be sure, it would be a serious flaw if the book did this. But here I have to disagree entirely. Guala is not correct that the book overlooks the distinction, nor is the way he wants to draw the distinction particularly fruitful.

Before going into nuances, the basic distinction between types and tokens is simple. Tokens are spatiotemporal particulars, and types are abstract categories or kinds into which tokens fall. On the line following this sentence are five printed letters:

  • a a b a b

There are five tokens of letters in the above line. How many types? Well, there are three tokens of one type, i.e. tokens of the letter ‘a’, and two of a different type, i.e. tokens of the letter ‘b’. But there is also the type Roman letter, of which all five of these are tokens. And there is the type printed mark, of which the five are also tokens. There are many types of which these are tokens.

To apply this to social kinds: there is a token of a dollar in my wallet. It is a spatiotemporal particular, a bill that has mass and volume, and that I can touch, tear, or fold. It is a token of the type dollar, and it is also a token of the type money. Both dollar and money are types. Dollar is not a spatiotemporal particular any more than money is. There are also other types that the bill in my pocket is a token of. For instance, the type dollar printed in St. Louis. And the type dollar with serial number starting with the letter F. And the type dollar printed in 1993. And there are other subtypes of the type dollar as well, which the bill in my pocket is not a token of, such as dollar coin.

Likewise for marriage. My marriage ceremony took place on July 13, 2008, in New York. That was an event, a spatiotemporal particular. It was a token of the type marriage ceremony as codified in US law, and a token of the type marriage ceremony. My marriage as a whole, I suppose, might also be regarded as a particular that began in 2008 and continues to the present. If so, it is a token of the type marriage as codified in US law, as well as a token of the type marriage.

This is about as straightforward as we can make the type-token distinction. But Guala wants to change things up. He introduces a distinction between “institution types” and “institution tokens.” Dollar, he claims, is an institution token, while money is an institution type. Marriage-as-codified-in-Italian-law is an institution token, while marriage is an institution type. What is the reason? According to Guala, dollar and marriage-as-codified-in-Italian-law are historically located: they “have a beginning, a geographical location, and inevitably will come to an end”. He suggests that we should understand these much as Ghiselin and Hull understand species: that is, as individuals in virtue of their historical contingency.

On this understanding, I suppose that the dollar bill in my pocket would have to be considered a spatiotemporal part of a huge extended individual, which consists of all dollars. So the bill in my pocket, on his view, is not a token of the type dollar. Instead, it is a spatiotemporal part of the giant object. Likewise, my marriage is a spatiotemporal part of the huge individual consisting of all US marriages.

There are some obvious shortcomings to this way of thinking. Not only is it counterintuitive, but it also makes a hash of all the canonical examples of the type-token distinction. The letter ‘a,’ for instance, has historical origins and is contingent. So it is not a type on Guala’s view, and he must dispense with the ‘a a b a b’ example I gave earlier (which is basically the sort of example that motivated Peirce’s original distinction between types and tokens in the first place). For that matter, it is not even clear what Guala means by “contingent” as applied to institutions. Why is marriage itself not contingent? Is even the existence of humans – or anything at all – not a contingent matter? In that case, perhaps marriage is also a token institution, not a type. On Guala’s characterization, are there any types at all?

But even these shortcomings are minor in comparison to the fundamental puzzle about Guala’s move: why think that a type (or a kind or a universal) cannot have historical instantiation conditions? After all, types can have spatial instantiation conditions (such as being shaped in a certain way) so why not historical ones? They can, of course: it is fine for marriage-as-codified-in-Italian-law to be historically placed, and yet nonetheless to be a type.

In the book, I do not overlook the distinction, but find it coarser than I need. The deep problem with Guala’s focus on the type-token distinction is that it misleads him to categorize institutional kinds into just two broad sorts. Without this misleading application of the distinction, we would not be tempted down this path. It is better to speak of various kinds of kinds, and instead of tokens to speak of objects that are instances or members of those kinds. In Chapter 11, for instance, I speak about a particular group (the Supreme Court), and then in Chapter 13, about kinds of groups (intramural basketball teams) and about also particular teams. This terminology is clearer and more generally applicable.

5.2. The Real Dispute with Guala

Guala wants to divide institutions or institutional kinds into two sorts: those that are wholly ahistorical and functionally defined, and instances of those that are historically and spatially located. (Making use of a different terminology, we might understand these as causal-role-functional-kinds and realizer-kinds).

It is not that these categories are empty or useless. The problem is that there are far more social and institutional kinds than just these two. And even the paradigmatic examples Guala discusses do not fit into these two.

I will not discuss the example of money: I doubt that the simple three-part functional definition Guala gives (and that one finds in some Econ 101 textbooks) is correct, but the reasons would take us too far afield. But marriage is better: I think it is more straightforward to see the problems with marriage as having a generic causal-role functional definition.

Guala’s own proposal is that the functional role of marriage institutions is to solve a particular type of coordination problem: a problem having to do with procreation, education, economic cooperation, and emotional support (see Guala 2016). He holds that the various institutions of marriage in each culture are solutions to this problem. (Again, it would help generalize this view to think of these as realizer-kinds.)

Only, is it plausible to think that there is a single coordination problem that all these cultural institutions are universally aiming at solving? No one has managed to give even a remotely plausible story about the functions of marriage; I would venture to say that the reason no one has specified exactly what problem regarding procreation, education, and so on is being solved is that there simply is no single one. Marriage does not have one generic function, or solve a common set of problems. There are problems that people share across some cultures, and others share across other cultures, and some that are more historically embedded or specific to local circumstances. I do think we can talk about marriage cross-culturally, but it is folly to imagine that we can leave the actual historical and cultural particulars aside and still capture it. To put it in terms that an “externalist” might: a good part of what individuates the general institution of marriage on planet earth is the actual tokens and practices of actual marriages. Take those away, and we are not even characterizing a generic institution.

Notice that this disagreement with Guala has nothing to do with types and tokens. The disagreement is that he wants to fit social kinds into two levels: causal-role-kinds of a particular sort (i.e. kinds whose realizers are kinds that solve a particular coordination problem), and realizer-kinds. And he thinks that all sorts of disarray stems from confusing one level for another, or from failing to see that all our institutions have this structure.

I agree that there may be some causal-role-kinds of that particular sort. Perhaps everyone driving on the right side of the road versus everyone driving on the left fits this model: there is a simple coordination problem for which we have two available solutions. But most institutions and social kinds are not like this. Even among the functional kinds, there are many sorts, not just causal-role-functional ones. And among the causal-role-functional ones, there are many more than just those that can plausibly be construed as solving coordination problems.

If there is one moral to The Ant Trap, it is that social ontologists should stop force-fitting all social kinds and institutions and phenomena into a single type or into simple dichotomies. There are many kinds of social facts. There are many ways different social facts are grounded. And there are many ways those grounding conditions get set up.

5.3 Gallotti on Anchor Pluralism

Like Guala, Gallotti is concerned with claims about “anchor pluralism.” Only, Gallotti seems to think that I am not pluralistic enough. Gallotti argues that my approach to collective attitudes leads me to overlook the potentially important role collective intentionality does play in anchoring features of the social world. He argues that I should not rule out a potential role for collective attitudes in anchoring, which he thinks I do.

But here I suspect that Gallotti ascribes to me a thesis I do not hold. Again, as I mentioned earlier, The Ant Trap does not discuss how anchoring works in general, so I do not actually make a claim either way in the book on this point. But contrary to Gallotti’s worry, I am completely open to the possibility that collective attitudes may be among the anchors of some frame principles. I do believe that collective attitudes are not the only anchors of frame principles. Lots of features of the world go into anchoring our social kinds, including features of the environment and more. I also do suspect that collective attitudes are not necessary for anchoring: that is, that the anchors of at least some frame principles include no collective attitudes. But neither of these points excludes a role for collective intentions in anchoring.


  • Einheuser, I. (2006): “Counterconventional Conditionals”. In: Philosophical Studies 127. No. 3, p. 459–482.

  • Guala, F. (2016): Understanding Institutions: The Science and Philosophy of Living Together. Princeton: Princeton University Press.

Comments (0)

Please log in or register to comment.