Foundations of Institutional Reality 0197657346, 9780197657348

Citation preview

Foundations of Institutional Reality

Foundations of Institutional Reality ANDREI MARMOR

Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide. Oxford is a registered trade mark of Oxford University Press in the UK and certain other countries. Published in the United States of America by Oxford University Press 198 Madison Avenue, New York, NY 10016, United States of America. © Oxford University Press 2023 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, by license, or under terms agreed with the appropriate reproduction rights organization. Inquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above. You must not circulate this work in any other form and you must impose this same condition on any acquirer. CIP data is on file at the Library of Congress ISBN 978–​0–​19–​765734–​8 DOI: 10.1093/​oso/​9780197657348.001.0001 1 3 5 7 9 8 6 4 2 Printed by Integrated Books International, United States of America

To my daughters Galia and Noga

Contents Preface 

ix

1. Institutional Facts 

1

2. Grounding and Reduction 

23

3. Grounding Social Rules 

40

4. Constitution by Rules 

62

5. Artifacts and the Limits of Error 

84

6. Rationalizing Practices 

105

7. Power-​Structuring Rules 

129

Bibliography  Index 

151 157

Preface Hardly any aspect of our daily lives is not enabled, structured, or regulated by social practices and institutions. Most people work in some institution or organization, like a university, law firm, bank, retail store, factory, or some other corporate entity; many of our recreational activities involve participation in social practices, such as playing games, engaging with sports, or enjoying arts of various kinds. We routinely use money, credit, and other institutional means of exchange, and our savings are invested in some financial institution or other. Many people participate in various forms of religious practices, often involving membership in some religious organization. We follow numerous conventional practices of etiquette and manners in dealing with acquaintances and strangers. And then, in the background of all of this, and in some sense, on top of it all, we have laws regulating our lives, determining ways in which we can relate to each other in myriad ways. Take property, for example: laws determine what counts as mine and what does not, what it practically means for something to be mine and not yours, how I can use something that is not mine or condition others’ use of something I own, and so on and so forth. And there is a lot to this “so on and so forth”: laws regulate almost every aspect of our lives, private and public. The institutional aspects of our world thoroughly shape our lives, and structure the environments in which we conduct most of our daily routines. A substantial part of the world in which we live in is a world that we, humans, collectively somehow, created ourselves. It is this collective social aspect of reality or, at least, certain basic elements of it, that I want to explore in this book, from a metaphysical perspective. I want to examine the foundations of our institutional reality, ways in which we can build it up, as it were, from relatively more simple building blocks. The construction metaphor, shared by many philosophers who write on these issues, is not accidental, of course. It suggests that the philosophical project envisioned is one of grounding or reduction. The idea is that the central aspects of institutional reality—​in particular what it is for a social practice or an institution to exist—​can be accounted for by way of reducing them to more foundational elements, such as patterns of conduct, attitudes and dispositions, functions, motivating

x Preface reasons, and products of human agency. Therefore, part of the project here is to identify what elements are more foundational than others, how to unpack this building or construction metaphor, and show the explanatory payoff we get from all this. The objective of the book is twofold: to provide a novel account of the ontological foundations of institutional facts, and to show that some important epistemic and methodological implications follow from this ontology. The first part of the book offers a detailed reductive account of institutional facts by way of metaphysical grounding. It shows that an ontology of institutional facts requires an ontology of social rules, and the latter depends on a reductive account of collective attitudes that determine the content of rules. A grounding-​reductive account of collective attitudes that comports with methodological individualism is developed in Chapter 3. The constitutive grounding relation between rules and practices, analyzed in Chapter 4, is given by a functional explanation, challenging John Searle’s influential distinction between constitutive and regulative rules; rules are constitutive of practices when they function in certain ways. The second part of the book aims to show that a number of important epistemic and methodological conclusions follow from the ontology of institutional facts. First, I demonstrate that certain types of widespread errors about the socially constructed aspects of reality are not metaphysically possible. The functionalist account of artifacts articulated in Chapter 5 is an essential step in developing this argument, demonstrating its upshot and delineating its limits. The second main methodological conclusion, explored in Chapter 6, is that a metaphysical account of institutional reality does not have to provide an explanation of the relevant social practices in terms that would rationalize the practice for the participants themselves. Thus, together, Chapters 5 and 6 purport to highlight the epistemic and methodological conclusions that follow from the grounding argument presented in Chapters 1–​4. Finally, Chapter 7 circles back to the idea of hierarchical practices, introduced in Chapter 1, arguing that basic social power-​structuring rules function to transform brute power into an elaborate normative framework, constituting authoritative institutions that are central to our institutional reality. The philosophical foundations laid out in H. L. A. Hart’s The Concept of Law and Searle’s The Construction of Social Reality inspire much of this book. Hart’s work on the foundational role of social rules, and Searle’s more ambitious project of constructing the basic elements of institutional facts from collective attitudes, functions, and constitutive rules, I take to be important

Preface  xi philosophical insights in the right direction. My main purpose here is to contribute some corrections and modifications to these projects, and add a few more steps to the argument, thereby offering a more complete picture of the metaphysical foundations of our institutional world. Helping me in this endeavor are recent works in metaphysics about grounding, collective intentions, and the nature of artifacts. I aim to show that we can use some developments in metaphysics to shed new light on old ideas and, I hope, develop them further. Throughout the book I use the law as one of the main examples of a social institution (alongside natural language and art). Why give prominence to law? First, law is undoubtedly the most complex, sophisticated, and structured social institution human cultures have created. This makes law a challenging example, but theoretically rewarding for that. Furthermore, over the last century or so, law has attracted a great deal of philosophical attention in the analytical tradition, much more so than other social institutions (with the exception of language, of course, but I will dwell on language as well). The result is that we can rely on a rich philosophical literature that aims to explain what the law is and what it would take to explain the nature of law and its constitutive elements. Not that there is a great deal of agreement on any of these issues; controversies abound, but at least we have a good roadmap of the questions in play and the relevant aspects of law that are philosophically puzzling. Finally, the lack of metaphysical clarity in philosophy of law is itself a reason for working with law as a central example. Almost every essay about the nature of law attests to the fact that metaphysical issues are lurking in the background, but most of the metaphysical concepts deployed in these debates remain unaccounted for. Legal philosophers often mention such concepts as social facts, social practices, social rules, and institutional facts. Yet none of these concepts gets explained in any detail. The fact that legal philosophy lacks an adequate metaphysics to back up the various positions in dispute makes the legal example even more rewarding to work with. Since this book is rather short, it does not warrant a long introduction. Chapter 1 introduces some of the main questions and challenges that will be addressed in the rest of the book, and outlines the roadmap for the subsequent chapters. What I would like to mention here, however, are a few points connecting this work to some of my previous work on social conventions and philosophy of law. When you write a new book on something that is closely related to other books and articles you had written before, you cannot

xii Preface simply expect your readers to be familiar with your previous work. “Copy and paste” is not a good solution either. So this leaves you with some tough choices to make. I have opted for reducing reliance on my previous work to a minimum, mentioned in footnotes here and there, but not much more. In two cases, however, this is somewhat problematic. The first has to do with the nature of social conventions. Much of the present work, as will become apparent from Chapter 1, builds on the metaphysics of social rules. I am not assuming that all social rules are conventional in nature but surely many kinds are. Yet, since I have already published a book on social conventions, I felt that it would be unnecessarily repetitive to include the main arguments about conventionality of rules in this work as well.1 Furthermore, my work on conventions was not focused on metaphysical aspects of social rules; the analysis I offered of social conventions was largely structured by concerns of practical reasoning. Therefore, I felt that I could bracket the conventionality of rules in this work. I make no attempt here to distinguish conventional from nonconventional rules, and I do not rely on my analysis of conventions to substantiate the arguments I make here. I hope that this approach works, and that the arguments deployed in these two works converge nicely, without tensions or contradictions. The second point of connection with previous work is in philosophy of law, and relates to my association with the legal positivist school of thought. So let me emphasize that it is not the purpose of the present work to argue for legal positivism. My arguments imply, at various points along the way, that some of the metaphysical commitments of standard anti-​positivist views are questionable, but these arguments are not confined to legal theory, and the lessons about theoretical constraints on grounding-​reductive accounts of our institutional reality are meant to be much broader in scope. Defending legal positivism is not my aim here. However, there is an aspect of legal philosophy that echoes a theme I have argued for in the past—​namely, the methodological point that legal positivism is best seen as a metaphysical theory about the nature of law, and a reductive one at that. But, if I may reiterate, I use the law, and theorizing about it, only as one example in this book. My ambition is to offer a reductive account of institutional reality much more broadly. Whether and to what extent the account I offer here aligns with legal positivism or not will be discussed briefly in Chapter 6, but it is rather tangential to my concerns in this project. 1 Marmor, Social Conventions.

Preface  xiii It is not part of this work, however, to offer an analysis of the philosophical foundations of the social sciences. How to go about doing science in the social realm, what would count as a scientific explanation of sociality, and similar questions are not the kind of issues I address in this book. Some might find the distinction somewhat artificial; after all, the foundations of the social sciences clearly involve metaphysical issues, and perhaps even vice versa, our views about the metaphysics in play might be influenced by methodological views about social sciences or what we would consider a scientific explanation in the social domain. I’m not denying these connections, but there is only that much that one philosophical essay can hope to cover, even if there are closely related issues in the vicinity. I am indebted to a number of friends and colleagues who have read and commented on parts of the manuscript at earlier stages, in particular to my Cornell University colleagues Emad Atiq and Alexander Kocurek; to David Enoch and participants in the Hebrew University Department of Philosophy Colloquium; and to Jeremey Waldron, Liam Murphy, and participants in the New York University Colloquium in Legal, Political, and Social Philosophy. I owe a particular debt of gratitude to the two reviewers of Oxford University Press. Their meticulous, thoughtful, and penetrating comments contributed a great deal to the content and structure of this work.

1 Institutional Facts My purpose in this chapter is to begin to clarify some of the key theoretical concepts at the foundations of institutional reality, in particular, the ideas of a social fact, an institutional fact, and a social rule. But I am hoping to achieve a bit more than that; I am hoping to show that by trying to elucidate some of these foundational building blocks of our institutional reality, we can also reveal some of the main questions and challenges we face in any attempt to account for the metaphysics of social practices. This chapter does not aim to provide detailed answers; its purpose is mostly to raise some questions and introduce the main issues that will be addressed later in the book. Philosophical interest in social ontology is wide and diverse. This book is focused on a limited number of some key aspects of it, and by the end of this chapter I am hoping to clarify what those aspects are and provide a roadmap for the issues to be discussed in subsequent chapters.

1.1.  Social Facts and Institutional Facts The institutional aspects of reality are clearly part of the much larger and much more diffuse social aspect of the world. In other words, institutional facts are a subset of social facts. It might be useful, however, to start with the larger category, with the idea of a social fact. A simple, noncircular definition of a social fact would be one that utilizes the idea of ontological dependence. We could say that a fact is of a social kind if and only if it is the kind of fact that ontologically depends on the existence of a multitude of human beings interacting with each other in certain ways. It doesn’t mean that a social fact has to be a fact about groups of human beings, as such, or about a particular type of interaction among them. The idea is that a fact is social if and only if it depends, ontologically, on the existence of human interactions. Ontological dependence is by no means a problem-​free idea. Different kinds of ontological dependence are discussed in the literature, and there is a general sense that some of them are too coarse-​grained to be useful enough Foundations of Institutional Reality. Andrei Marmor, Oxford University Press. © Oxford University Press 2023. DOI: 10.1093/​oso/​9780197657348.003.0001

2  Foundations of Institutional Reality in metaphysical inquiry.1 But for our purposes here, the most generic and hopefully the most intuitive notion of ontological dependence would do the work. The basic intuition is that we have things or facts of type A ontologically depend on things or facts of type B, when A could not have existed without B. I say “things” or “facts,” rather loosely, because for our purposes the exact relata of ontological dependence does not matter much. It is, probably, more plausible to think of ontological dependence obtaining between things that can exist (or not) in some sense. Nevertheless, I will assume here that there is a certain interchangeability between things that exist and facts about them. It might be a fact (or not) that “o is an F”; it might also be a fact (or not) that “o exists,” or that “o exists as an F.” When we discuss the idea of grounding, in the next chapter, I will have more to say by way of justifying the focus on facts and about what facts are, in the relevant sense here. Thus, the basic idea is that A depends ontologically on B if it is necessarily the case that A could not exist without B. This is a fairly coarse-​grained, modal-​existential notion of ontological dependence, which can be defined as follows: A ontologically depends for its existence on B =​df Necessarily, A exists only if B exists.2

Among the many possible complications here is a temporal issue: ontological dependence can be understood sequentially or continuously. We can talk about ontological dependence in cases where a particular thing could not have existed without some other thing that had existed before, such as the dependence of my existence on the prior existence of my biological parents, and we can talk about continuing existence, such as the continuing existence of a set depending on its members or the existence of water depending on H2O molecules. For purposes of defining what a social fact is—​that is, for saying whether facts about A are social facts or not—​ontological dependence needs to be understood in the continuous-​dependence sense. Facts of a social kind generally depend on the continuous existence of human interactions, not on some prior state in which there were humans who interacted with each other. This does not mean that the sequential or causal sense of ontological dependence is not relevant to various aspects of social reality. Some 1 See, for example, Koslicki, “Varieties of Ontological Dependence.” 2 I borrow this formulation from Tahko and Lowe, “Ontological Dependence.”

Institutional Facts  3 causal links might be necessary features of certain aspects of our sociality. But for the characterization of what social facts are, generally speaking, we need the continuous sense of dependence. A fact is of a social kind if and only if it depends on some interactions between humans. In other words, I am assuming that if x years from now there are no longer humans on our planet, at that time there will be no social facts on our planet, only a history of them. Generally speaking, ontological dependence, in this coarse-​grained, generic sense, is a modal concept. It points to a necessary connection between the existence of one kind of thing and another. The basic intuition of (continuous) modal-​existential dependence is that for facts of type A to exist, some other type of facts, B, must exist as well. Now applying this to social facts, the basic intuition I try to cash out here is that a fact is social if and only if it depends, as the kind of fact it is, on the existence of human beings interacting in certain ways. Which is to say two things: First, that a world in which there are no human beings (or, if you will, creatures sufficiently similar to us) is a world in which there are no social facts. Second, a world in which there are only Robinson Crusoes, each one totally isolated on an island with no interaction with any other, is a world in which there are no social facts. Social facts are the kind of facts that ontologically depend on the existence of human interactions. Now one might worry that this definition is both overinclusive and underinclusive. It might be thought to be overinclusive in that the definition would make, for example, many facts about morality social facts. It also renders most facts about language social facts. True, but I don’t see it as a problem. Surely our ability to communicate in a natural language is one of the most fundamental aspects of our sociality. And there is hardly any reason to doubt that many facts about moral aspects of our lives are social facts, that is, at least by way of ontological dependence.3 Bear in mind that ontological dependence does not entail reducibility or supervenience or anything of the sort. So yes, the definition of what counts as a social fact is very broad, but we will soon narrow down our interest to a subset of social facts, and that’s where details and scope may matter. The definition of social facts by way of ontological dependence on human interactions is obviously underinclusive by excluding nonhuman animals 3 A certain type of robust realism about morality (e.g., Enoch, Taking Morality Seriously) holds the view that there are certain moral truths that do not depend on human interactions or even on the existence of humans. I do not need to take sides on this particular debate. I am certainly not committed to the view that all facts about morality are social facts.

4  Foundations of Institutional Reality from the domain of discussion. And some might think that this is narrow-​ minded; after all, we know that some animal species exhibit fairly sophisticated social lives. So if we talk about ontological dependence on interactions between humans, we exclude the sociality of animals. But this should be seen as a matter of theoretical convenience at this point, and nothing more. If at any point it turns out that something we say about the metaphysics of the social domain is threatened by the exclusive focus on humans, we can revisit the assumption. I don’t think that there will be a need for that, at least not in the context of this book.4 The generic, modal-​existential version of ontological dependence is too coarse-​grained to tell us much about the nature of things. For one thing, ontological dependence is a modal concept, and, as such, is not explanatory. It just tells us that there is a necessary connection between the existence of one thing and another, but it tells us nothing about why something is the way it necessarily is.5 (More on this in the next chapter.) Thus, to say that facts of type A ontologically depend on the existence of facts of type B is rarely the kind of information that conveys a great deal of philosophical insight. Second, at least in the social context, not many cases are factually controversial. It is certainly not controversial that things like law, for example, depend ontologically on the existence of humans, on the kind of interests and needs human beings have, on the ability of humans to communicate in a natural language, on our predisposition to have moral sensibilities, and on all sorts of things like that. And the same can be said of other social practices. I think that we will get closer to our target if we draw a distinction between social facts and institutional facts as a subset of the former. All institutional facts would be social facts, on this view, but not all social facts are also institutional. By the term institutional facts, I will label facts about social aspects of reality that depend on the existence of rules.6 What kind of dependence is in 4 I have great sympathy for the view that the social sciences would probably benefit a great deal from taking into account the social lives, and even forms of collective agency, in the nonhuman animal world. (see Epstein, The Ant Trap). But this book is not about the philosophical foundations of the social sciences. My interest is confined to the institutional aspects of human cultures. As I will mention in Chapter 3, the idea of collective agency is much wider in scope than the building blocks of institutional reality that I employ in this book, and potentially not confined to human agency. 5 In some of his earlier work on dependence, Fine tied the idea of ontological dependence to essentialism (“Ontological Dependence”). He suggested that A depends on B just in case B is a constituent of A’s essence. His more recent work on metaphysical grounding makes it doubtful that he still holds the earlier views (“The Question of Realism” and “Guide to Ground”). It seems that much of what Fine wrote earlier about dependence gets folded into the more recent work on grounding, distinct from ontological dependence, that is. Grounding will be discussed in detail in Chapter 2. 6 This conception of “institutional facts” is inspired by and roughly in line with Searle, The Construction of Social Reality, 27. Searle, however, called institutional facts those social facts that

Institutional Facts  5 play is something that we will need to discuss in detail later on, after we have a detailed account of metaphysical grounding and its corresponding modes of reduction (Chapter 2). Why focus on facts that depend on rules? The reason is straightforward: our main interest here is the metaphysical foundations of social practices and institutions. Practices are structured by rules. I use the word “structured” advisedly to avoid the assumption, which would be premature at this point, that practices are necessarily constituted by rules or that the rules constituting practices are necessarily of a particular, viz., constitutive, kind. For now, I hope that at least at an initial, intuitive level, we can see that there is no hope for any metaphysical explanation of facts about social practices, such as law, language, arts, money, games and sports, corporations, schools, universities, and countless other social practices, without an account of what rules are, and what it is for a rule to exist. Institutional facts are facts about practices that are, at least partly, structured or regulated by rules. This simple insight was at the core of Hart’s contribution to analytical jurisprudence. We need “the idea of a rule,” Hart said, “without which we cannot hope to elucidate even the most elementary forms of law.”7 And that of course applies to any social practice, whether legal or not. Obviously, all institutional facts are also social facts; they depend, ontologically, on human interactions. But not all social facts are institutional. There are many types of social interactions that are not structured by rules and their existence, as social facts, does not depend, ontologically or otherwise, on the existence of rules. Examples would be things like taking a walk together, cooking a meal together, having sex, having a fight, going on a trip together, or for drinks in the pub. As Gilbert, Bratman, and others have shown, doing certain things together (which Bratman called “shared agency”) requires certain types of commitments and intentions that need to mesh, as it were, sometimes intricately so.8 But there are no rules structuring such forms of shared agency.9 Even on a larger scale, however, certain types of social depend on constitutive rules. Since I will come to call into question his distinction between regulative and constitutive rules—​in particular, the idea that only constitutive rules have the function of enabling social practices (Chapter 4)—​the difference between our uses of “institutional facts” is somewhat tangential and indirect. The essential point, that institutional facts depend on the existence of social rules, is the same. 7 Hart, The Concept of Law, 78. 8 Gilbert, On Social Facts; Bratman, Shared Agency. 9 In some of these cases of shared agency, when the shared activity occurs with some regularity, social conventions may evolve that structure certain aspects of it. But, even so, the conventions are not rules that make the activity possible, as such. Nonconventional sex is still sex, and a nonconventional cocktail party may still be a cocktail party. See also Searle, The Construction of Social Reality, 88.

6  Foundations of Institutional Reality interaction do not depend on rules. People often exercise brute power over other people, sometimes millions of other people, subjugating or subordinating them in various ways. Obviously, rule over others often takes complex social interactions, but those kinds of interactions do not have to form part of rule-​structured practices (though often they do, of course).10 Thus not all social facts are institutional facts. When rules structure a certain type of activity, then we can talk about a social practice, and facts about practices I will label institutional facts, to designate their rule dependence. For the most part in this book, I will be interested in the metaphysics of institutional facts. Admittedly, the distinction between institutional and noninstitutional social facts is not always very sharp.11 Consider the example of social solidarity. Solidarity between a group of people is a complex set of attitudes people share, and presumably, a certain set of dispositions they have to help or otherwise prioritize members of their group and exhibit some special concern for them. So, on the face of it, facts about social solidarity do not look like institutional facts; they are not structured by rules. But group membership itself is often a matter of institutional facts; social rules often determine who belongs to this or that group. So there might be an institutional, that is, rule-​ governed element to facts about social solidarity, even if solidarity itself is not essentially rule structured. I am not suggesting that rules are everywhere in our social lives. But they sometimes lurk in the background, and some aspects of social facts that do not seem institutional might have some institutional aspects, even if only at the margins. The idea of a social practice is extremely broad and covers many different types. Generally, we can say that we have a social practice when we have certain types of human activities, that is, action types, structured by rules or conventions, occurring with a certain regularity over time (in a given population). That is, social practices have a certain stability; they are maintained over some period of time, though they can change of course, sometimes dramatically so. Some social practices are relatively simple, such as the practice of our greeting conventions or table manners; others are more complex, such as playing games like chess or football; and some are much more complex, 10 There is some controversy or, rather I should say, inconsistency, in the literature about the question of whether wars are institutional in the sense we discuss here. Searle seems to think that they are (The Construction of Social Reality, 89). I’m not so sure. My inclination is to suggest that only superficial and marginal aspects of wars are rule structured. In any case, surely many cases of large-​scale violence, such as ethnic cleansings or civil wars, are not structured by rules. They are social facts, of course, but not institutional. 11 Searle agrees with this (The Construction of Social Reality, 88).

Institutional Facts  7 such as using natural language, practicing a religion, or maintaining a legal system. Furthermore, in most cases the population that engages in the practice over time is not constant; membership in the practicing population is often in flux, changing over time. The size of the population that maintains a social practice varies enormously. Sometimes, as in the case of natural language or law, the population is very large. In other cases, practices and institutions can be rather small in scale, as some peculiar game played only in one small geographic region, or a club that centers around a local hero. Given how diverse and numerous social practices are, many possible distinctions between different types of practices can be drawn, but I don’t think there is a need for it here. One distinction worth mentioning, however, is between practices we can call flat, and those that are hierarchical. Many social practices we are familiar with are highly structured in that they have what Hart called secondary rules—​namely, rules determining ways of creating new rules, modifying existing ones, monitoring compliance with the rules, and determining ways of reacting to noncompliance.12 The law is certainly highly structured and hierarchical in this way, but countless other practices share this hierarchical complexity, such as sports leagues, clubs, universities, and corporations. I call them hierarchical to designate the fact that they have secondary rules—​rules structuring the creation of new rules and reactions to violations of the rules. In Chapter 7, we will see why this hierarchical structure of rules is bound to emerge in some cases and not others, and its connection to normative powers. In other words, we will see why social hierarchy depends on a hierarchical structure of rules. However, some social practices are not hierarchical, and have no secondary rules. The use of natural language is a prominent example. Many fundamental aspects of language use are rule governed, but there are no rules about creating new rules of a language, and no rules determining reactions to noncompliance. Despite the considerable complexity of linguistic rules, language is flat, not

12 Hart, The Concept of Law, 79. See Marmor, Social Conventions, 50–​52. Admittedly, the idea of secondary rules is a bit vague. According to Hart, who introduced this distinction, primary rules are rules of conduct; they apply directly to modes of conduct or behavior, like do this or don’t do that. Secondary rules take rules as their object; they are rules about rules, such as rules about ways of changing rules or enacting new ones. But this leaves some of the familiar institutional rules a bit up in the air. Practices often have rules determining specific punitive reactions to breaking the rules; they are not exactly primary rules—​at least because their scope of application is always relative to some rule—​but then not quite rules about rules either; they don’t take rules as their object of regulation. In any case, I will stipulate here that rules that function to monitor compliance with other rules and determine ways of reacting to noncompliance are secondary rules. As we will see in Chapter 7, most of the secondary rules are power-​structuring rules.

8  Foundations of Institutional Reality hierarchical. And of course, other familiar practices share this lack of secondary rules, such as practices of etiquette or manners, genres of arts, and all sorts of customary or conventional practices. Generally, however, the idea is that we have a social practice when we have a fairly distinct pattern of human activity, structured by rules, occurring with some regularity, and persisting over time in a given population. The main lesson here, and the one I want to focus on for the rest of the chapter, concerns the foundational aspect of rules. Any metaphysical explanation of institutional facts and social practices requires a metaphysics of rules. As Hart put it very simply, we need to ask: “What are rules? What does it mean to say that a rule exists?”13 Despite presenting this question at the center of his work, the truth is that Hart hasn’t quite managed to give us an answer. Hart provided an interesting and, in many ways, a fruitful explanation of what it takes to have a social rule in place, but his account of social rules depends on the idea of a rule, an idea that remains metaphysically unexplained in The Concept of Law. I need to elaborate on this problem in Hart’s theory because it reveals one of the main challenges facing any metaphysical account of institutional facts. Before we get to Hart’s theory of social rules, however, we need to explain what kind of rules are in play here. The notion of a rule is extremely broad. Some types of rules are not quite relevant to our concerns. Common to all rules is that they have some prescriptive content, a domain of application determining the relevant subjects of the rule and its circumstances of application, and an application range that occurs with some generality or regularity. But all this is very broad and vague. So let me draw a rough distinction between two general types of rules: some rules are natural, in that they are not socially constructed. In some sense we discover them in the nature of things or, perhaps better said, extrapolate the rule from the way things are. Take, for example, rules of inference in logic, norms of instrumental rationality, rules about warranted inference from factual evidence, and all sorts of practical rules of thumb like how to get from A to B or how to achieve X by doing Y. There is a sense in which we figure out the existence of the rule from the way things are. Perhaps not necessarily so; natural rules do not have to be necessary, and they can be both a priori and a posteriori (whether necessary or not). They are natural only in the sense that they are not socially



13 Hart, The Concept of Law, 8.

Institutional Facts  9 constructed.14 It’s a good rule of thumb, for example, to be very careful when cutting things with a sharp knife, or to walk slowly on slippery pavements, otherwise you might injure yourself. Such rules are not necessary—​clearly they are a posteriori, and yet, also clearly enough, not socially constructed. It’s in the nature of things, as it were, that walking too fast on slippery roads might result in an accident, so if you care about your health, avoid it. Similarly (in a very limited sense of course), it’s in the nature of things that there are certain rules of inference preserving truth from premises to conclusions. So I call these kinds of rules natural just to signify the idea that they are not socially constructed. Natural rules will remain outside the scope of our discussion. The kind of rules that will be discussed here, and in the rest of the book, are of the nonnatural kind, that is, rules that are socially constructed in one way or another. Thus, henceforth, when I use the word “rule” I mean the nonnatural kind only. Nonnatural rules, as Hart and many others recognized, come in two main forms: rules are either enacted by someone in a position of power or authority, or they are social rules, not resulting from anyone’s deliberate enactment. Clearly the latter are more foundational. Once we have an account of what social rules are, we can go on to explain what authorities are and how they can introduce new rules or modify existing ones.15 In fact, this will be the topic of Chapter 7: we will see how foundational power-​structuring rules are, and how they enable the emergence of hierarchical social practices. In short, I take it that Hart was clearly correct in regarding the idea of a social rule as foundational. He was also correct in emphasizing that rules ought to be clearly distinguished from mere regularities of behavior.16 There are many things people do with considerable regularity, such as eating dinner, sleeping at night, or having sex, but not as a matter of following a rule. Rules, as opposed to regularities of behavior, have a separate ontological aspect that needs to be explained. To say that something is done with some regularity is only to report on the recurrence of an action type, and types of action tend to recur mostly because the relevant reason for action happens to occur with some regularity. But to say that something is done by way of following a rule is to invoke something that exists over and beyond the mere regularity of 14 Examples are going to be controversial. Some rules are necessary a priori, but whether they are necessarily analytic is controversial (Kripke, Naming and Necessity). More problematically, it might be argued that analyticity is essentially semantic and therefore social in the ontological sense. 15 On the constitutive function of social norms in structuring practical authorities, see Marmor, “An Institutional Conception of Authority.” 16 Hart, The Concept of Law, 54–​59.

10  Foundations of Institutional Reality conduct. And Hart’s main intuition here, surely correct in some sense, was that when we follow a social rule we regard the rule itself as a reason for action. The rule has to be a fact that can count in favor of doing something, which is to suggest that there is some sense in which rules exist.17 This is precisely the point where the first main challenge of our foray into the metaphysics of institutional reality emerges: the challenge of explaining the ontology of social rules. What is it for a social rule to exist? So the general assumption here is that institutional facts are facts about social practices; what makes a certain type of continuous human activity form part of a social practice is the fact that the activity is structured by rules. Therefore, a metaphysical account of social practices, an account of what institutional facts are, must start with the idea of a rule. Without an ontology of social rules we will not be able to account for institutional facts. In this respect, I think that Hart had the better starting point compared with John Searle. Searle’s account of institutional facts, as we will see later (mostly in Chapter 4), took the idea of rules seriously, but without any attempt to account for the ontology of rules. Searle was mostly interested in structural or syntactical aspects of rules, and their constitutive functions, without telling us much about what it is for social rules to exist.18 The functional aspect of rules is crucially important, and will be discussed at length in Chapters 4 and 6. But the starting point needs to be ontological. I will follow Hart in this respect, and start with the question of what it is for a social rule to exist. We will not have the answer in this chapter. First I need to show why Hart’s own answer to this question is crucially incomplete, and what it would take to complete it. The completion will come in Chapter 3, following some general metaphysical groundwork presented in Chapter 2. Before we proceed, a note on usage. Here, and for the rest of our discussion, I follow Hart in talking about reasons for action in the causal-​motivating sense. Whether a rule, or anything else is, normatively speaking, a reason for action or not is beyond the concerns of the metaphysics of sociality, quite generally. We have to be careful not to turn a metaphysical account of rules

17 Kantians may think that whenever we act for reasons for action in the right way we follow a rule. I don’t have a quarrel with that, but notice that the discussion here is confined to nonnatural rules, which is not what Kantians have in mind. 18 Apart from the vague idea that rules must exhibit a certain form of social agreement or acceptance. Acceptance, as we will see later, is too vague and too coarse-​grained to ground an ontology of rules. And the idea of agreement, as Lewis (Convention) convincingly demonstrated, is certainly not going to do the job. We need rules precisely in those cases where agreements are too difficult to obtain.

Institutional Facts  11 into a normative one. Thus, from now on, talk about reasons for action is talk about motivating reasons, not normative ones.19

1.2.  The Practice Theory of Rules Hart’s explanation of what social rules are, which came to be called The Practice Theory of Rules, can be summarized in the following way: A social rule, say R, exists in a population S, Hart maintained, if and only if the following conditions obtain (henceforth: Conditions-​H): 1. Most members of S regularly conform in their behavior to the content of R, and 2.  Most members of S accept R as a rule, which means that a. for most members of S the existence of R constitutes a reason for action in accordance with R, and b. members of S tend to employ R and refer to it as grounds for exerting pressure on other members to conform to R, and as grounds for criticizing deviations from conformity to R.20

As we can see here, according to Hart the existence of a social rule consists of a pattern of conduct and a set of shared attitudes and dispositions: we have a social rule when there is a component of conduct or behavior—​viz., the regular conformity with the rule or the regularity of conduct in accordance with it—​and a complex component of “acceptance” of the rule, which consists of (1) a belief shared by the population that the existence of the rule provides them with a reason for action and (2) a shared attitude of a positive endorsement of the rule that is manifest in the disposition to invoke the rule as grounds for exerting pressure on others to comply as well, or criticizing them when they don’t. Clearly, this is a reductive account of social rules. It purports to explain what social rules are in terms of overt behavior 19 The only connection between motivating and normative reasons we can acknowledge here, without getting into unnecessary trouble, is to be willing to assume that when a fact is a motivating reason for an agent, the agent would regard the fact as a normative reason. If I take X to be a reason for my φ-​ing, I would assume that X actually counts in favor of φ-​ing. More colloquially, perhaps, we can assume that when people act for a reason, they would assume that it’s a good reason. But this assumption is not needed for the argument and nothing depends on it in the discussion to follow. Besides, it may not be as innocuous an assumption as it may seem. 20 Hart’s explanation of the nature of social rules is scattered around several places in The Concept of Law. Most of the essential points are at 82–​86.

12  Foundations of Institutional Reality in a given population, accompanied by certain attitudes and dispositions actually entertained by (most) members of that group. Notice that this is also an aggregative account because it purports to explain a social phenomenon in terms of facts about individual members of the relevant population. This feature of social explanation is called methodological individualism. The idea is that we can explain social phenomena without positing any entities beyond individual people and their causal interactions.21 If most members of a given population behave in a certain way, and share some attitudes and dispositions with respect to that behavior, then we have “the idea of a rule.” The problem is that we don’t quite get “the idea of a rule” here. Hart’s reductive account of rules seems to be strikingly circular: we have a (social) rule, Hart suggests, when, inter alia, people regard the existence of the rule as a reason for action, as something that counts in favor of doing (or avoiding) that which the rule prescribes. But this really seems like a vicious circle: we have a rule when we regard the rule as a reason for action. That may be true, but it says nothing about what rules are. Now you might think that Hart shouldn’t be too worried about this apparent circularity. What matters is not what exactly it is that people believe to exist (the rule), but the fact that they believe that it does, and have a typical set of attitudes and dispositions about it. But this won’t do. To begin with, we would still need some content for “the idea of a rule,” that thing that people believe to exist and to give them reasons for action. Nothing in Hart’s theory would allow us to see the difference between following rules, or following the commands of God, or the will of the ancestors, or, more importantly, following reasons generally regarded as good reasons.22 We can replace “R” in Conditions-​H with “God’s will,” or “natural law,” or perhaps even just something like “the right thing to do in C,” and we get the same results. So “the idea of a rule” we do not get here. You may still think that for Hart’s purposes it doesn’t matter. Even if we have not quite articulated what rules are, at least we have an answer to the question of what it is for a rule to exist—​namely, a combination of some regularity of conduct and a particular set of attitudes and dispositions accompanying that regularity shared by the relevant population. But important questions remain unanswered here. First, how can we separate the question of what constitutes the content of a rule from its conditions of existence? In 21 This will be discussed further in Chapter 3. 22 This point against the practice theory of rules was made a long time ago by Warnock, The Object of Morality, 45–​47. Warnock pointed out that Hart’s account cannot really distinguish between following a rule and following a generally or widely accepted reason for action.

Institutional Facts  13 other words, Hart may well be quite right to suggest that we have a social rule when members of a given population by and large regard the existence of the rule as something that gives them a reason for action, a reason to exert pressure on others to comply, etc. This would clearly distinguish rule-​governed behavior from mere regularities of behavior. But in order to get closer to a metaphysics of rules, we still need an account of what it is that forms the relevant content of the shared beliefs—​what is this thing that people believe to be a rule? Second, we also need to know how the relevant attitudes can be shared by a group; surely it cannot be just a matter of a happy coincidence. As we will see in Chapter 3, the challenge here is not trivial. If the existence of a rule requires certain attitudes to be shared by a population, often a rather large and diffuse group of individuals, we need an account of what this “sharing” is. Lots of people may happen to have the same attitude about something, but in a way that is too disconnected or too accidental to ground an idea of a rule. An attempt to give an account of this aspect of rules will form one of the main challenges to be discussed in Chapter 3. Finally, we need an account of the nature of the metaphysical dependence between the shared attitudes and the existence of the rule. To illustrate, consider the negation of existence. Thus, suppose you replace “R” in Conditions-​H with “God’s will.” You will get something that tells you what it is for a population to be guided by what they take to be God’s will or God’s commands. Now consider the assertion “God does not exist.” Whether the assertion is true or not, we understand what it means and, crucially, one can deny the existence of God without denying that Conditions-​H obtain in S. But now suppose we assert something similar about a rule, say R*: I tell you that despite the fact that Conditions-​H obtain about R* in population S, there is no R* in S—​R* does not exist. Now it is no longer so clear what it is that I’m saying. Can people believe that there is a rule, R*, and that R* gives them reasons for action, even if there is no R*? What would it mean for R* not to exist if people believe that it does and behave accordingly? Without knowing what kind of thing R* could be, and how much of its existence is actually constituted by shared attitudes, we cannot hope to answer this negation problem. In other words, the practice theory of rules, formulated by Conditions-​H, doesn’t quite tell us what the idea of a rule is, nor what it takes for one to exist. There are at least two elements missing from the picture: the type-​token distinction, without which rules cannot be used; and more importantly, the idea of collective intentionality. We cannot hope to explain what a social rule is,

14  Foundations of Institutional Reality and what it is for a social rule to exist, without explaining the possibility of a group of people collectively intending that P.23 Let me use a very simple linguistic rule to explain the intuitions behind these two points. Consider a rule of the form: X stands for/​signifies Y (in context C): R1: The word “red” [as a sound configuration] stands for red [the color] (in English).

There are at least two salient features present here. First, we would not be able to use such rules if we did not have the ability to think in terms of tokens being instances of types. Both the sound configuration of the word “red” and the color red are types, not tokens. When I say “red” (with my Israeli accent) and you say “red,” we utter slightly different sounds. But as long as they are tokens of the same type, we understand each other. And of course redness is a type, not a token. The word stands for the type of color, not this or that particular red. And here I venture to generalize: our ability to use rules in our social lives hugely depends on our ability to associate particular tokens with types. Without this ability we would not be able to have rules. I am not suggesting that all rules must be formulated by way of relation between types. Rules can apply to a particular thing, such as a rule saying, “Don’t touch X,” where X is a particular object. But even here, the prescription relies on a type of action, touching, and there is a range of actions that may be “touching” in the relevant sense. Furthermore, I am not suggesting that it is impossible to have a nonnatural rule formulated in a way that does not contain types. Consider, for example, a rule requiring a number of particular people (set =​a, b, c, d) to perform a particular act (token) every January 1. Maybe there are no types here, only tokens (though we may still need some type concept to account for the recurrence of a date every year). Still, the vast majority of social rules are not like that—​which is really all we need here. I don’t think that there is much by way of philosophical analysis we can add to our ability to associate tokens with the types they are tokens of. I will assume that it is essentially biological. We are constituted in ways that enable us this mental capacity. These mental capacities probably vary with different types; it is more than likely that what enables us to see different shades of red 23 This seems to be acknowledged in more recent literature. See Searle, The Construction of Social Reality, 22–​26; Gilbert, On Social Facts; and Epstein, The Ant Trap.

Institutional Facts  15 as shades of redness is very different from ways in which we can categorize particular actions as action types, etc. All I wanted to flag here is that our ability to use rules crucially depends on these mental capacities we share. There is, however, a separate and difficult question in the vicinity, about the ways in which we carve up the world, linguistically and conceptually, into types or categories. Structuralism maintains that there is truth and falsehood about such matters since the world itself, not just our representation of it, has a real structure. Our categorization is true if it “carves up reality at its joints,” and false if it doesn’t.24 Many philosophers tend to resist realism about structures, maintaining that there is no mind-​independent reality about types and categories—​reality has no real joints to discover. Types, categories, similarities, and things like that may be more or less useful, or they may reflect something that enables us to represent different aspects of the world, but structures are, essentially, in our minds, not in the world. Though my own intuition is that this debate is bound to play out differently in different areas, I will remain agnostic about metaphysical structuralism. I am flagging this issue only to set it aside.25 What matters for the metaphysics of rules is the mental capacity humans share to identify tokens with types, without which social rules, by and large, or most of them if you will, would not be possible. Let me turn to the collective-​intention element, which is the one that’s really important here. Consider again the rule that “X signifies Y in C.” What does the relation “signifies” (or “stands for” or “represents”) consist in, given that there might be no particular reason to signify Y by X? There might be a very good reason to have Y signified by something, we may need the signification for a reason, but there are no reasons to signify it the way we happen to. So when we say that X signifies Y, we must be talking about some intention—​ it is an intention we have to signify Y by X. But then it must be the intention of a collective; the rule does not express the intention of a particular person. We intend X to designate Y, We intend X to count as Y, and so on. Who is “We”? Presumably, those whose rule it is. Sometimes the group is fairly small, and sometimes it consists of the relevant population of language users. It may be worth keeping in mind at this point that social rules necessarily have a population-​relative temporal dimension. Remember that when we talk about rules here we confine our attention to what I called nonnatural

24 Sider, Writing the Book of the World. 25 Though perhaps a limited aspect of this debate will come up in the discussion of artifacts in Chapter 5.

16  Foundations of Institutional Reality rules. Natural rules, like, say, norms of instrumental rationality or practical rules about how to achieve X by doing Y, are not socially constructed. We extrapolate the rules from the way things are, in some sense. Therefore, natural rules can be universal; their existence is not necessarily tied to particular populations at a given time. Nonnatural rules, rules that are socially constructed, as it were, are rules of a given population at a given time. When we say, “It is a social rule that R,” the questions of whose rule is it?, and when?, are always relevant and should have an answer. R could be a social rule here (say, in New York City) but not there (say, in Rome), or it could be a rule of soccer fans but not opera fans, or it could have been a social rule here some time ago but no longer is. Social rules are tied, as it were, to some population whose rule it is at a given time. Of course we can come to know about rules at other places and in different times. The point here is not epistemic, it’s metaphysical. The continuous existence of a social rule depends on the existence of the population, at a given time and place, whose rule it is. This is partly what is meant by the idea that rules are socially constructed (when they are, of course). Let me return to the example of the rule determining the semantic representation function of “red.” Assume, at least for the sake of the argument, that semantic properties are properties of words and sentences, not utterances or speech acts. I know that this is controversial, but I’m interested in exemplifying how rules work, not making a point about semantics. So let me work with this picture of meaning whereby the meaning of a word is a representation type, assigning X to represent, designate, or stand for Y.26 The crucial point to realize is that semantic rules assigning, as they do, a signifying relation between sounds and types of things they signify—​or of concepts, if you think that what I call types here have to be concepts—​cannot be understood without the idea of collective intentions. The meaning of a word crucially depends on the intention to represent Y type of things by X (a distinct sound configuration). And, generally speaking, the intention is that of a collective of language users. When I intend to use a word the way that it is normally used in English, my intention is to follow the collective intention of English speakers, the intention of the relevant language users, as a group. In using the word “red,” in an ordinary conversation, I would typically intend to use the sound configuration as other English speakers intend to use it. For all I know, 26 As Salmon aptly put it, words are expression types, not tokens, and we can see semantics, generally, as a “systematic assignment of representation” (“Two Conceptions of Semantics,” 323). That is the picture I’m assuming here, but again, only for the sake of demonstration.

Institutional Facts  17 other users of English utter “red” when they intend to refer to the color red. Barring unusual circumstances, my linguistic intention is to align with the collective one. All this seems to be very much on the surface. Think about instances when you need to correct someone who is using a word incorrectly. You might say, “This is not what we mean by X” or “It’s not how we use X in this context.” We, of course, refers to users of this natural language or of a particular idiolect, as a group. None of this entails that individuals cannot deviate from the collective intention on particular conversational occasions. After all, speakers don’t always use words and expressions according to their standard or common meanings. Deviations on particular occasions of speech are quite frequent, and usually rendered unproblematic by contextual and other pragmatic features enabling hearers to grasp the content intended to be communicated by the speaker. In other words, when I say that collective intentions determine semantic representations, I do not mean to imply that speakers cannot deviate from the collective semantic intentions in particular cases. But these deviations are generally parasitic on the standard meaning of words, meanings that are determined by the collective intentions of language users.27 Before we proceed, let me insert a clarification here: the argument about collective intentions in the semantic context does not depend on anything like Wittgenstein’s alleged refutation of the possibility of private language.28 More generally, I do not intend to imply that rules, necessarily, have to be public, reflecting some collective intention. Perhaps I can create my own version of Andrei solitaire, with rules I invent, and play the game privately; there’s nothing in what I assumed here to count against the idea that my Andrei-​solitaire rules are rules, even if nobody else knows about them. My only assumption here is that social rules, which are evidently public and ontologically dependent on human interactions, cannot be constituted by way of some happy coincidence of lots of private rules. So I take the disagreement about the possibility of private language to be irrelevant to our concerns. As far as I know, nobody argues that the possibility of private language would show that natural language is not, in some sense, a collective endeavor. The 27 See, for example, Salmon, “Two Conceptions of Semantics”; Soames, “Drawing the Line between Meaning and Implicature.” 28 Wittgenstein, Philosophical Investigations, §243. Wittgenstein’s private language argument is not free of problems, of course; even sympathetic readers find some puzzles and unresolved questions about the argument. See, for example, Baker, “The private Language Argument.”

18  Foundations of Institutional Reality word “red” means what it does by virtue of the fact that a collective or group of language users intends it to signify the color red. This remains true even if it is possible for an individual to invent their own color concepts and use them on their own.29 Semantics of a natural language might seem like a special case. When rules constitute a system of representation, they obviously rely on a type-​token distinction and obviously manifest some collective intention. But not all rules are of this kind, and not all rules constitute new modes of representation or some new activity type. Countless rules simply aim to guide conduct—​rules may express a demand, or some kind of a requirement to act in this or that way or to forebear from doing this or that. Are such conduct-​regulating rules also expressions of a collective intention?30 That depends on what kind of rule it is. Remember, we distinguished between enacted rules and social rules. Rules of the former type, enacted by someone in a position of authority or power, can express the intention of the enactor, and nothing more. They don’t have to express any form of collective intention. Social rules, however, of the kind that purport to regulate conduct included, would be very difficult to understand without the idea of collective intention (or, perhaps, some other collective attitude). Let me explain why that is the case. Since social rules of conduct are often formulated by “ought” statements (or can easily be reformulated as such), one might be tempted to think that we can analyze such rules as generalized formulations of reasons for action. Take a rule of the form: R2: “All X’s who are F ought to φ in context C.”

Let us assume that R2 expresses the idea that if an X is F in context C, X has a reason (of a certain type or structure) to φ. But leaving it at that would make the idea of invoking a rule as one of pointing to a reason, which would make it very difficult to explain the difference between things we do for a reason, with some regularity perhaps, and things we do by way of following a rule.

29 Kripke’s famous argument about rule skepticisms (Wittgenstein on Rules and Private Language), cannot be discussed here. This so-​called Kripkenstein argument has generated an industry of literature over the years, for and against skepticisms about rule following, and there is no hope of resolving that issue in this work. I am just going to assume that even if Kripke is right and there is a deep philosophical puzzle about rule following, quite generally, the philosophical puzzle leaves praxis as we find it—​namely, we do follow rules in our everyday lives, on countless occasions with respect to a great variety of rules. 30 The distinction between regulative and constitutive rules will be discussed in Chapter 4.

Institutional Facts  19 In other words, Hart was quite right to emphasize that we cannot begin to explain the idea of following rules without accounting for the fact that the very existence of the rule is taken by the relevant population to be a reason for action. If a rule is a reason for action then the rule itself has to be a fact that is deemed to count in favor of doing this or that. To see the point more clearly, we can contrast social rules with rules of conduct that are, indeed, nothing more than generalized formulations of reasons for action that apply. Here’s a rule that is sadly familiar from the earlier stages of the COVID-​19 pandemic: “When talking or otherwise interacting with a person who is not a member of your immediate household, keep at least six feet apart.” Alas, there was a good reason for this rule, which is to prevent infection with the virus that causes COVID-​19, as it can spread from one person to another within a few feet, but usually not more than six. So we were all told to observe this rule, and most people tried to follow it. Now of course, this is not an example of a social rule; the rule has been recommended by medical experts and state authorities (if the coronavirus hangs on for long, it may become a social rule, but let’s hope we will not get to that). I hope this example helps us see Hart’s point about the independence of rules from the reasons that apply irrespective of the rule. Once we know that the coronavirus can spread by droplets reaching up to six feet, keeping six feet apart is what we should do, regardless of the rule. One who complied with the rule because it is a rule would be acting from the wrong kind of reason. If it turned out, for example, that science got it wrong and the virus can infect at greater distance, say up to eight feet, we should keep eight feet apart; the six-​foot rule will simply no longer apply. Or vice versa, if it turns out that the virus cannot infect beyond three feet, then there is no reason to stick to much more than that, and no sense of breaking a rule by doing so. Social rules play some role in our practical reason and in our social lives, more generally, only when their import is distinct from the reasons that apply to us anyway. When we have a genuine social rule in play, there has to be a sense in which people can break the rule even if their failure to comply is justified, all things considered. Perhaps you have good reasons to fail to greet your aunt when seeing her at a cocktail party, but even if your conduct is justified, all things considered, you would have broken a rule of etiquette, the greeting convention in this case. However, if you have good reasons not to worry about contagion with the coronavirus in a given encounter, there is no rule breaking involved in failing to keep the requisite distance. The relevant reasons just don’t apply anymore.

20  Foundations of Institutional Reality All this comes to show that Hart was quite right to assume that we need an ontology of rules that is not entirely a function of the reasons for action that apply in the circumstances. Rules must capture something that is metaphysically distinct from the reasons for action they purport to point out or formulate, if you will. And this clearly indicates that rules are very closely tied to the idea of a collective intention. To say that there is a rule in S to require X’s who are F to φ in C is to point to something that is collectively intended in S. I am not suggesting here that social rules just are collective intentions. What kind of metaphysical dependence obtains between collective intentions and rules will be analyzed in Chapter 3. But I am suggesting that there is bound to be some metaphysical dependence between social rules and collective intentions. Without the idea that a collective of people, as a collective, can share an intention that certain things ought to be done (or that certain things represent or stand for something else), we cannot account for the fact that a rule exists. To say that “a rule exists in S” must refer to something that can exist (or not), and something that might exist in S but not necessarily in Si. Remember that social rules have a population-​relative temporal aspect that must be accounted for; social rules are rules of a particular collective.

1.3.  The Roadmap Let me recap briefly and explain how the book proceeds. The conclusion that emerges so far is that we cannot provide a metaphysical account of institutional facts without an ontology of rules, and we cannot provide an account of the latter without an account of the relevant collective attitudes that determine the content of rules. Before we get to the analysis I will offer, in Chapter 3, we need to begin this journey with providing the metaphysical framework that will inform the analysis offered in the rest of the book. In Chapter 2, I will try to explain why social ontology, of the kind we are interested in here, is rightly drawn to an Aristotelian framework of grounding and reduction. This book aims to show, among other things, that some of the recent literature on metaphysical grounding can help us to a better understanding of institutional facts and ways in which we can account for the building blocks of institutional reality. For readers who are familiar with the literature on grounding, most of Chapter 2 may not provide new insights, but I will argue for a novel distinction between two types of reductive explanations that will be utilized in later chapters. Since Chapter 2 engages

Institutional Facts  21 with some controversial aspects of metaphysical grounding and its relation to reduction, it may be a bit more technical than the rest of the book, but I hope not too much so. In Chapter 3, we return to the question of what makes social rules exist, as such, utilizing some of the key concepts explained in Chapter 2. Following the idea, argued here, that there is no hope for explaining what social rules are without an account of collective intentions, Chapter 3 offers such an account, and one that is compatible with methodological individualism. Thus the metaphysical grounding of collective intentions forms a major part of this chapter, accompanied by a reconstruction of Hart’s practice theory to provide a more complete account of the ontology of social rules. Chapter 4 turns to an attempt to unpack the constitutive relation between rules and practices. In this chapter I build on Searle’s account of constitutive relations in terms of practical functions, explaining in some detail the idea of function, and ways in which the functions ground the constitutive relation between rules and practices. My explanation deviates from Searle in two respects. First, I raise some doubts about Searle’s famous distinction between regulative and constitutive rules, arguing that the functional explanation, coupled with a proper construal of the practice theory of rules, is sufficient to account for the grounding relation between rules and practices. Second, I argue that this grounding relation is incomplete or partial. It takes more than rules to ground practices; a minimal level of agreement in judgments among participants about the point of the practice and their reasons for engaging in it is also practically necessary. Chapter 4 completes the basic argument of this book about the ontology of institutional facts. In the next two chapters I show that a number of epistemic and methodological conclusions follow from the ontological account. Chapter 5 turns to a fairly detailed analysis of the nature of artifacts, and for two main reasons. First, because there is considerable overlap between artifacts, in particular, intangible and massive artifacts, and institutional facts. Indeed, it is part of the argument here that social rules, and practices structured by rules, are artifacts. The second reason to dwell on the nature of artifacts in the context of our discussion, is to show that there are some group-​wide, comprehensive errors about certain aspects of artifacts that are not metaphysically possible. A population that creates and uses an artifact has some epistemic privileges about what the artifact is. This epistemic privilege will be shown to align nicely with the idea, shown in Chapters 3 and 4, that the content of social rules just is the collective intention shared by the

22  Foundations of Institutional Reality relevant population. In other words, a closer look at the nature of artifacts and their relation to institutional facts allows us to complete an argument about the epistemic implications that follow from the social ontology presented in these chapters. It allows us to see that socially constructed aspects of reality are such that they prevent the possibility of certain types of comprehensive, group-​wide errors about the nature of the things socially constructed, as opposed to other types of comprehensive errors that are possible and sometimes quite common. This last point raises another issue, however, which forms the topic of Chapter 6: Does a metaphysical-​grounding account of social practices have to provide the kind of analysis that could be endorsed by those whose practice it? I am going to answer this question in the negative. I will defend the view that a metaphysical account of institutional facts and social practices, as opposed to a hermeneutical approach, is not constrained by the need to provide an account of a practice in terms that would rationalize the practice for those who engage in it. A discussion of some of the central controversies about the nature of law will form the main example I use in this chapter, but I hope to show that its lesson generalizes to all cases. The book concludes with a discussion of one particular type of institutional facts—​namely, normative powers, and ways in which they are socially constructed. One remarkable aspect of our social reality is that human societies managed to transform a great deal of brute power exercised by people over others into an elaborate normative framework. Normative powers are institutional facts of tremendous importance, and Chapter 7 examines in some detail how and why such powers emerge, and how they constitute and structure hierarchical social practices and institutions. This chapter aims to show that basic power-​structuring social rules are essential building blocks of a significant part of our social reality. Social hierarchy is necessarily structured by basic power conferring social rules.

2 Grounding and Reduction What metaphysics should aim to accomplish is a controversial business. It is becoming commonplace these days to distinguish between two main views about the task of a metaphysical inquiry; both views have long and respectable pedigrees. According to one view, associated with Quine among others, metaphysics should address the question of what there is in the world, focusing on what kind of things exist. Do numbers, properties, universals, meanings exist? Are these the kind of things that form part of the fabric of the universe, as it were, or merely our ways of thinking about what there is? According to the second view, sometimes labeled Aristotelian, an important task of metaphysics is (also?) to tell us what kind of things are more foundational than others.1 The physical particles and forces that my desk is made of are more foundational than the desk; human beings, with their mental capacities and dispositions, etc., are more foundational than social institutions, like universities or corporations. So the assumption here is that there is some hierarchical structure in the world, whereby certain things that exist are constituted or constructed by other things that exist, and thus, at least in some sense, certain things are more foundational than others. The kind of metaphysical inquiry we seek in the social context is almost inevitably drawn to the Aristotelian framework. Quite obviously, institutional facts do not form part of the basic fabric of the universe—​to the extent that they are facts about things that exist in the world, their existence is not foundational. Laws, institutions like universities or banks, sports, the arts, commercial means of exchange and finance, to mention a few obvious examples, are what they are in virtue of other things that exist, things or facts that are more foundational. Consider, for example, one of the central questions that has preoccupied philosophy of law for over a century, the question of what legality consists in: What makes things legal, when they are? Does legality depend on morality, and if so, in what ways? To the extent that there are metaphysical issues involved here, those would hardly be questions about what 1 The contrast between these views is nicely explained by Schaffer, “On What Grounds What.”

Foundations of Institutional Reality. Andrei Marmor, Oxford University Press. © Oxford University Press 2023. DOI: 10.1093/​oso/​9780197657348.003.0002

24  Foundations of Institutional Reality really exists. I’m not aware of any position in legal philosophy aiming to show that legality is a real mind-​independent property, part of the fabric of the universe, as it were—​or, to the contrary, a position that would suggest that there are no laws, that they do not exist. It is rather uncontroversial that there is a clear sense in which laws exist, and a fairly clear sense in which they don’t. As we saw in Chapter 1, all institutional facts are social facts, and as such, ontologically dependent on interactions between human beings. It would be just too crazy to suggest that the existence of law does not depend on human interactions. But it is equally uncontroversial that there are ways in which we can talk about the existence of laws. It is a law in New York State that the maximum speed limit on interstate highways is 60 mph. It was not the law prior to early 1900s. And it is not the law in Germany.2 Thus, assume that (1) is true: (1) The law in Si at t prescribes Lx (where Lx stands for some prescriptive propositional content).

Then the following is also true: (2) There is a law in Si at t that Lx.

Now suppose the Quinean metaphysician insists that the existential operator in (2) cannot possibly be the robust ontological existence we assume about atoms and gravity and things like that. Of course that is the case; everything that is legal depends on other things. The law’s ontological dependence on all sorts of facts about human beings; our material, mental, and social conditions; and countless factors like that, cannot be doubted. Which is to say that it is not questions about existence that have preoccupied philosophy of law. The relevant metaphysical question about legal aspects of the world is one about foundations, not about existence. Laws clearly exist in some sense and clearly do not exist in some other sense. The questions here are about hierarchies of existence, about what is more foundational than something else. In other words, and I take law to be only an example here, the metaphysics of sociality is inevitably drawn to an Aristotelian framework. What we seek to explain is the building blocks of social practices, to find the 2 I am not suggesting that every proposition about the content of laws would be either true or false. But most of them are.

Grounding and Reduction  25 foundations of the kind of things that in some sense clearly exist, and in another, not less clear sense, do not exist in a foundational way. What is it for one thing to be more foundational than another? This may mean several things, actually, but a good start would be to assume that if an A aspect of X is more foundational than its B aspect, then there is a sense in which we should be able to utilize the A aspect in order to explain the existence or essential features of the B aspect. Ideally, we should be able to say that X is B in virtue of X being A. And this is the idea of grounds. A grounds B just in case that something is B in virtue of being A.3 There is a lot of grounding talk in contemporary metaphysics, but it doesn’t mean that there is a great deal of consensus on what exactly grounding is, and what it applies to. I will start with some characterizations of grounding that are widely accepted, and will later dive into a few more controversial aspects of it. So we have already tied the notion of grounding to that of explanation. The grounding relation is meant to be explanatory, like causation but unlike modal concepts such as necessity or supervenience. To say that B is necessarily A does not, by itself, explain B. Same for supervenience: if facts of type B supervene on facts of type A, it means that we cannot have any difference in B facts without some difference in the A facts. What supervenience guarantees is a kind of necessary correlation between two distinct types of facts. But supervenience, by itself, has no explanatory power. If it obtains, it calls for explanation, and leaves us wondering why the correlation is necessary and what would explain it. Grounding, however, like causation, has some explanatory power. To say that A grounds B does seem to provide an explanation, similar to the way in which causation explains something. So there is this way of thinking about grounding as something that explains the relation between things. But grounding, perhaps like causation, though this is controversial of course, is also a relation between things in the world that obtain or exist, something that would be there even if we were not around.

3 Not every Aristotelian view of metaphysics necessarily goes in the direction of grounding, e.g., Sider, Writing the Book of the World. The idea of grounding permeated contemporary metaphysics only in the last twenty years or so. Many have contributed to this endeavor, including Fine, “The Question of Realism” and “Guide to Ground”; Rosen, “Metaphysical Dependence”; Schaffer, “On What Grounds What”; Audi, “Grounding” and “A Clarification and Defense of the Notion of Grounding”; Bennett, “Construction Area (No Hard Hat Required)” and Making Things Up. Of course, grounding does not come without its critics. See, for example, Hofweber, “Ambitious, Yet Modest, Metaphysics”; and Daly, “Scepticism about Grounding.”

26  Foundations of Institutional Reality Admittedly, this dual role assigned to grounding may seem troublesome from the start. Explanation, generally speaking, is often relative to our theoretical interests and questions raised by our theorizing about this or that. What is in need of explanation often depends on prior explanations, and theoretical frameworks often determine what would count as an adequate explanation within that framework.4 But if grounding is also a worldly metaphysical relation, then it is not tied to our interests and our ways of seeing the world. Can one and the same relation be both? Probably yes, as many would think about causation, for example.5 And this leads some philosophers to suggest that grounding is a kind of metaphysical causation. When there is no causal connection between two items, but one nevertheless exists in virtue of the other, we have a grounding relation. Perhaps, but we should be careful with this idea of grounding as “metaphysical causation.” There are some similarities between causation and grounding, for sure, but saying that grounding is metaphysical causation is no more than a metaphor, and one that is not easy to unpack. So when we say that A grounds B, we say that B is what it is in virtue of it being A. (Or, X is B in virtue of X being A.) Grounding is a metaphysical relation between things or facts, explanatory, presupposing a hierarchical structure of the world in which A is more foundational than B. Therefore, it is widely assumed, and I take it rightly so, that a grounding relation between A and B must be asymmetrical, irreflexive, and transitive. Asymmetry: If A grounds B, B cannot ground A. Irreflexivity: A cannot ground itself. Transitivity: If A grounds B and B grounds C, then A grounds C.

At this point one might have hoped that we could give some simple and uncontroversial examples of grounding. Alas, simple examples, maybe, but not uncontroversial. The use of grounding is wide and varied, but in almost all cases controversial. We can say, for example, that dispositional properties of objects are grounded in their categorical features. A glass is fragile in virtue of the molecules that make it up and perhaps laws of physics and 4 I am not endorsing here a wholesale pragmatic view of explanation, certainly not in science. But it is difficult to deny that, generally speaking, explanations are not tied to our interests and modes of theorizing about the world. 5 If you hold a Humean conception of causation, would that drive you to a kind of Humean conception of grounding? Not necessarily, I presume, but I leave this, and related issues, unresolved here.

Grounding and Reduction  27 chemistry. Some would say that moral properties are grounded in nonmoral facts. Legal positivism, as we will see in greater detail below, is best understood as a claim about grounds: positivists claim, and anti-​positivists deny, that legal facts are fully grounded in institutional facts. Some might want to say that colors are grounded in the reflection of light waves from physical surfaces and the biochemistry of our eyes and brain. Grounding possibilities are almost endless. But almost all of them are philosophically controversial. However, it is precisely this nontrivial aspect of grounding that makes it potentially fruitful. When we can distinguish clearly enough between different types or categories of facts, and ask a question about the potential grounding of one type by another, we are given a powerful tool to focus our inquiry in a way that is nicely structured. For example, if there is a suggestion that facts of type A ground facts of type B, but then it turns out that, under some conditions, B would ground A (violating asymmetry), we would know that something went wrong. Similarly, if transitivity fails, we would know that either the grounding relations we assumed are incorrect or, at the very least, incomplete. A few more clarifications before we get to some controversies. First, we need the distinction between complete and partial grounds, which also bears on the question of transitivity. So the idea is this: A would be a complete ground of B when the fact that A obtains is sufficient to account for the fact that B obtains. A partially grounds B when B obtains in virtue of A, but not only A. So here’s a schematic version: Grounding: The fact that A grounds the fact that B. A consists of items (x, y, z). Complete grounds: The fact that (x, y, z) grounds the fact that B. Partial grounds: The fact that [(x, y, z) and φ] grounds the fact that B.

In both cases, facts (x, y, z) stand in a grounding relation to B. But in the second case (x, y, z) is not sufficient to ground B—​it also takes another fact, φ, to provide the complete grounds of B. Notice that, in the case of partial grounds, transitivity might fail (but not asymmetry and irreflexivity). If A partially grounds B, and B partially grounds C, it may not be the case that A grounds C.6 So there is a crucial difference between complete and partial 6 Schaffer argues that partial grounds may fail transitivity and offers a contrastive view to account for such cases (“Grounding, Transitivity, and Contrastivity”). I’m not aware of anyone who argues, however, that complete grounding relations might fail transitivity.

28  Foundations of Institutional Reality grounds and sometimes it matters which one we talk about. When it does, I will make it explicit whether I’m assuming partial or complete grounds. What kind of things in the world does grounding relate to? The answer is somewhat controversial. Most philosophers take grounding to be a relation between things or facts, but there is no general consensus on this or on what exactly we consider to be facts.7 First, we need to separate two issues here. For most friends of grounding, the relation itself is a metaphysical business, not representational or conceptual. However, when we focus on the question of the kind of items grounding takes as its relata, things become a bit more tricky. It is generally thought that grounding is a relation between things in the world, or worldly facts, not between concepts or meanings or ways of representing the world. Following Audi and others, I will assume here that worldly facts are facts about things that exist and their properties, including their relational properties to other things. Audi, however, takes this worldly conception of facts to be opposed to a “a conceptual view of facts, according to which facts will differ if they pick out an object or property via different concepts.”8 I surmise that the thought here is that if we want to maintain the grounding relation itself to be a kind of robust metaphysical relation, then the items grounding applies to have to be worldly facts, facts that would not vary according to different ways of representing them. And this makes a lot of sense, up to a point. The difficulty with the worldly conception, thus defined, is that it leaves us wondering how grounding relations can obtain between facts that are clearly mind dependent, and in ways that would be extremely difficult to identify, as the kind of facts they are, without relying on concepts and modes of representation. In the context of social ontology, this would be particularly problematic. There is, I would suggest, a grounding relation between corporate entities and laws; there is a grounding relation between laws and institutional facts; there is a grounding relation between institutional facts and social rules. None of these facts can be identified or individuated without relying on some modes of representation. Different concepts of what is a corporate entity would pick out different entities and different facts about those entities. Or, to take an example more familiar in the grounding literature, the fact that 7 Fine, for example, often talks about grounding relations between propositions (“The Question of Realism”). He explains, quite convincingly, I think, that truthmakers are not good candidates for the relata of grounding. A view that would constrain grounding to obtain only between truthmakers would be unduly restrictive (Fine, “Guide to Ground,” 43–​46). 8 Audi, “Grounding,” 103.

Grounding and Reduction  29 Chicago is a city is grounded in a whole series of other facts, maybe all the way down to the most fundamental particles and forces of physics. So we can think of the fundamental grounding facts in terms of the worldly conception. But I don’t see how the fact that is being grounded here—​that Chicago is a city—​can be grasped or individuated without relying on concepts. Are we talking about the city of Chicago only terms of buildings, streets, people, people’s possessions, interactions between people? And if we rely on concepts, like the vague and multifarious concept of a city, we must make room for the possibility that the individuation of facts might differ according to different concepts we use or different modes of representation. I am not suggesting that we need to abandon the worldly conception of facts as the appropriate relata of grounding. I think that Audi’s intuition is correct, but it wrongly focuses on the question of individuation. I think that it is hopeless to aspire for an individuation of facts that is entirely worldly, as Audi defines it. However, even if the individuation of facts or things in the world often depend on concepts or modes of representation, it does not mean that the relevant fact is a conceptual matter. Different conceptions of what a city is would pick out different items in its extension, but the items are not conceptual—​they are not facts about concepts, just facts about things that exist and relations between them, even if some of those things are mind dependent. In other words, I will also assume here, along with Audi and others, that grounding relations pertain to worldly facts, but I will assume a more permissive attitude to the individuation or categorization of facts that might be subject to a grounding account.9 The final clarification is potentially much more controversial. Most friends of metaphysical grounding explicitly avoid any suggestion that would implicate that a grounded fact doesn’t really exist. In other words, the fact that A is more foundational than B is not meant to imply that B does not exist, as such. For example, if you hold the view that dispositional properties are grounded in categorical properties, you are not thereby suggesting that there are no dispositional properties. Similarly, if you suggest that legal facts are fully grounded in certain types of institutional facts, you are not denying that 9 This, I think, is precisely how Schaffer would characterize worldly facts grounding takes as its relata (“On What Grounds What”). Therefore, the question of whether grounding is a relation between propositions or not would depend on the appropriate ontological conception of what propositions are. If and to the extent that propositions can be said to exist, there would be no objection to including propositions in the items grounding applies to; if propositions are just ways of representing that which exits, then I will assume that we should talk about grounding of the things represented, not the representations.

30  Foundations of Institutional Reality laws exist. The problem is that this idea pulls in opposite directions—​in one sense it is intuitively compelling, but in another sense quite the opposite. The kind of Aristotelian metaphysics that grounding is tied to heavily relies on the intuition that the world admits of a hierarchical structure; some things are more foundational than others. Given the fact that there must be some constitutive relations between things, that some things can build or constitute others, the idea of hierarchy seems unavoidable. And, as I said, intuitively compelling. Surely certain things are just more foundational than others. Physical aspects of the world are more foundational than social aspects of it—​at least we can agree on that. But equally plausible is that physics is more foundational than chemistry and the latter more foundational than biology. These ideas seem natural and very much in line with a scientific world view. But then the difficult question is why would this view not lead to some kind of doubt about the reality of the grounded facts?10 If B is completely grounded in A, why would we not want to say that B is somehow less real than A? Do we get the existence of B as “an ontological free lunch,” as sometimes suggested?11 Or is it the case that the Aristotelian metaphysician wants to have the cake and eat it? There are several avenues open here, and different philosophers opt for different views.12 On one view, Schaffer’s for example, existence comes cheap.13 Everything exists, even God for the atheist. The question is always: Exists as what? In what way and on what grounds does it exist? For the atheist, God exists as an idea, something that is fully grounded in peoples’ attitudes. And that seems right. But then the problem for Schaffer is that that even if everything exists it does not mean 10 One old theme that came up in relation to reduction, and would seem to resurface now with respect to grounding, is the question of whether it has, and necessarily so, an anti-​realist import (Fine, “The Question of Realism”). Michael Dummett for instance, has long wavered on this issue. In some moods he took it as obvious that if you can provide a full reduction of a given type of facts A to facts of type B, then, ipso facto, you hold an anti-​realist position about A. You would hold the view that facts of type A, as a type, do not really exist; they are actually facts of type B. But then the opposite seems equally plausible: if you are a realist about facts of type B, then by reducing A to B, you are ipso facto a realist about A. After all, you hold the view that A exists or obtains in virtue of being B and so, if and to the extent that B is real, A is real (e.g., Dummett, The Interpretation of Frege’s Philosophy, 448). Either way, I will put the question of realism aside here; an account of the realism versus anti-​realism debate is beyond the scope of the present work. 11 The expression comes from Armstrong, A World of States of Affairs, 12. 12 For some critics of ground, this idea that reality may come in degrees is precisely what makes grounding obscure and esoteric. See, for example, Hofweber, “Ambitious, Yet Modest, Metaphysics,” 268; but cf. McDaniel, The Fragmentation of Being. Another issue that perplexes both friends and foes of ground concerns the question of whether there are ungrounded facts, and if not, how to avoid an infinite regress. As Sider nicely demonstrates, there are some really tricky issues involved here (“Ground Grounded”). See also, e.g., Bennett, Making Things Up, ch. 5. These issues will be set aside here. 13 Schaffer, “On What Grounds What,” 356–​362.

Grounding and Reduction  31 that everything is equally real. And it’s not clear how you get an Aristotelian hierarchical picture of reality without some notion of degrees or levels of reality. As I mentioned earlier, there is a sense in which laws clearly exist. But that does not entail that the reality of law is somehow on par with the reality of atoms and mountains. There is a clear sense in which laws are less real than atoms or mountains, for example, they depend, ontologically, on us, humans; they need to be made, and need to be recognized by us for what they are. So I agree with Schaffer that existence comes cheap. But that does not settle the puzzle of what it is for reality to admit of hierarchy. Notice that the issue does not carve along the lines of mind dependence versus mind independence. Even within the mind-​independent aspects of the world, grounding relations are not just possible but quite likely, as the relations between physical, chemical, and biological domains might attest. And similarly, within the domains of reality that are obviously mind dependent, grounding relations are abundant. Laws are clearly mind dependent, but laws can ground other things, like corporate entities (also, obviously, mind dependent). It seems to me that there are two main ways to go here. One is just to bite the bullet and maintain, perhaps with the backing of a long Aristotelian tradition, that reality comes in degrees: some things are just more real, ontologically, as it were, than others.14 On this view, when we say that a grounding fact is more foundational than what it grounds, we are committed to the view that, though both exist, the grounding fact exists in a more robust way. More foundational means more real, so to speak. But this is not an easy bullet to bite. As critics point out, it sounds rather esoteric, not to say obscure.15 Mountains don’t seem to be any less real than atoms, even if atoms ground mountains; chemical properties of substances are not any less real than physical properties, even if the latter ground the former. The second way to go, however, is to maintain that, from an ontological perspective, there are only things that exist and various relations between them, such as grounding, causation, and modal relations. To say that A is more foundational than B is to say nothing more than pointing to a grounding relation between A and B. Talk about degrees or hierarchies of reality is just an explanatory or conceptual matter. In other words, if we agree with Schaffer that existence comes cheap, and the question is always

14 See, for example, McDaniel, who argues that there are degrees of reality (The Fragmentation of Being). 15 See, for example, Hofweber, “Ambitious, Yet Modest, Metaphysics.”

32  Foundations of Institutional Reality one about grounds—​about existing by virtue of what?—​then perhaps the idea of degrees of reality is explanatory and representational.16 Reality does not come in degrees—​things either exist or not and they may be related in this or that way. What comes in degrees, perhaps, is our conceptualization of reality (viz., of things that exist and relations between them), our ways of thinking about it. Atoms may seem metaphysically prior to mountains because they help us to an explanation of what mountains are, how they are formed, etc. (And, I presume, not vice versa.) But it doesn’t mean that atoms are more real, in the world, as it were, than mountains or anything else. In other words, we can understand the idea of metaphysical hierarchy as a function of our interests and conceptualization of the world. That seems quite plausible, actually, but then you might wonder how we can keep thinking about grounding as a metaphysical business, something that obtains even if we do not think about it, something that is not dependent on our conceptualization of the world. To see the difficulties more clearly, perhaps it would help to think about the following three theses adhered to by friends of grounding: (1) Grounding is metaphysical (not conceptual or representational), actually obtaining between worldly facts. (2) Facts that are grounded in other facts still exist, as the kind of facts they are. (3) Grounding facts are more foundational than the facts they ground.

It would seem that the main difficulty is how to explain the idea of fundamentality (viz., foundational) in (3) in a way that retains its metaphysical import, without either contradicting (1) or (2), or else lapsing into obscure notions of degrees of reality. The half-​baked solution I’m offering here is to understand fundamentality as operating at two levels. At the ontological level, fundamentality is nothing more than (1); it is just a way of expressing grounding relations. At an explanatory level, (3) is relative to our interests and modes of representation. We deem certain things more foundational than others when they help us to an explanation of the less foundational. In this sense, atoms and molecules are more foundational than mountains and valleys only because and insofar as they help us to an explanation of what mountains are and how they are formed, but not vice versa. The existence of mountains is not needed to explain what atoms and molecules are. Notice, however, that this

16 I am not suggesting here that this is Schaffer’s position.

Grounding and Reduction  33 explanatory function is compatible with the idea that atoms and molecules actually construct mountains, that there is a real metaphysical relation here. The fact that we have these two levels of thinking about fundamentality is not surprising in the grounding context; as we noted earlier, grounding is meant to be a real metaphysical relation with an explanatory power. So we have this duality from the start. Part of what underlies the picture I’m depicting here concerns another controversy about the nature of grounds, one that is closely entangled with the question of how grounding relates to reduction. Let me turn to this now. In my view, grounding should be distinguished from identity relations captured by the phrase “just is.” To say that A just is B is not, typically, a grounding relation. Naturally, if A just is B, then B just is A. Identity relations are symmetrical. To be a bachelor just is to be an unmarried man, as it is vice versa—​an unmarried man is a bachelor.17 Similarly, to say that “a square is an equilateral quadrangle” is not grounding—​an equilateral quadrangle just is a square. Not everybody agrees. Some see a much tighter connection between in virtue of and just is types of relation.18 In order to see what is at stake here, we need to consider the relations between grounding and reduction. And we need to do this anyway, since the idea of reduction will play a major role in the following chapters. Intuitively, we have a reductive explanation when we try to explain one class of facts or propositions by way of another type. Roughly, to say that P reduces to Q is to say that P can be fully explained or accounted for by Q. For now, I will leave it open whether reduction stands for a genuine metaphysical matter.19 There is, however, nothing that prevents reduction from applying to concepts: one can explain a class or set of concepts in terms of another class or set, reductively. The assumption I am making here, however, is that reduction is not confined to the conceptual or semantic realms. We can have reductive relations between worldly facts of different kinds, as between biological facts 17 One might think that this is not so simple—​after all, we can wonder whether the pope, unmarried as he is, is a bachelor or not. But this is not a serious worry; it stems from the fact that “bachelor,” like many other words, is subject to polysemy. It’s like saying to an adult male, “You are not a man!” The word “man,” like “bachelor,” can be used as a stereotype or in a variety of figurative ways. 18 I am inserting myself here in a debate between Rosen, “Metaphysical Dependence”; and Audi, “A Clarification and Defense of the Notion of Grounding.” As will become clear shortly, my view about the relations between grounding and reduction takes a middle position. However, at this point, I certainly agree with Audi and Dorr (“To Be F Is to Be G”) that “A just is B” is not a grounding relation. On how to understand reduction we may differ. 19 Rosen assumes that it is (“Metaphysical Dependence”). I am more skeptical; as I will try to show shortly, I think that reductive explanations can take concepts as their relata, and whether that would be metaphysical or not is debatable.

34  Foundations of Institutional Reality and physical facts, or between social facts and facts about individuals and interactions between them. Notice, however, that as a form of philosophical or scientific explanation, reduction is inherently ambitious; partial reduction is no reduction. A reduction of facts of type P to facts of type Q would be successful if and only if all aspects of P facts are fully accounted for by Q facts. A reductive explanation is successful only if it is complete. Reduction would fail if certain parts or aspects of the reduced class remain unexplained or unaccounted for by the facts that they are meant to be reduced to. This is clearly attested to by the enormous difficulties some reductive projects in philosophy have faced over the centuries. Think of materialism about mental phenomena. Ideally, we would have a reductive account of the mental aspects of the world if we could reduce them to the physical aspects. But this reduction turns out to be extremely difficult to accomplish. Even with sophisticated neuroscience, which can show us how mental states are potentially explicable by the biochemical processes in our brains, some of the most basic aspects of the mental remain unaccounted for. You can show how pain, for example, is the experience of some neural biochemistry, but you would still need to account for the actual sensation of the pain, what it is for me to feel it when I am in pain, and for you to understand it. Similarly, in metaethics, it would seem that all attempts to reduce moral judgments and moral evaluations to nonmoral facts face one version or another of the open-​question argument. Reduction, in short, is a tall order. Gideon Rosen observed that reduction seems to have a very close affinity to real definition.20 A real definition is a metaphysical business, not a verbal one. It takes items in the world as its objects, not linguistic expressions. For example, when we seek an answer to the question “What is knowledge?” we seek a definition of what knowledge really is, not what the words mean or how they are typically used or such. Now it’s not difficult to see that at least a certain type of reduction takes a very similar form. A real definition of X in terms of a, b, and c, can be expressed: X =​ df a, b, and c.

Which could also be expressed in terms of: To be X just is to be (a, b, and c).

20 Rosen, “Metaphysical Dependence.”

Grounding and Reduction  35 And this looks very much like reduction of X to (a, b, and c). Now, according to Rosen, a real definition is a form of (complete) grounding.21 If (a, b, and c) constitute what it is to be X, then we can also say that (a, b, and c) ground X. So we get a grounding-​reduction link here, by way of real definition. However, since we assumed that a real definition is expressible by the “just is” type of relation, on Rosen’s account we need to regard “A just is B” as expressing a grounding relation, on par with “something is A in virtue of being B.”22 And this, I think, is a considerable price to pay for the grounding-​reduction link. I am not going to argue with Rosen, however, about the grounding account of real definitions. In fact, I find it very plausible, at least in some cases. What I find problematic, however, is giving up the distinction between a grounding relation, expressed by “in virtue of ” relations, and a certain type of identity relation that is expressed by “A just is B” or, which I take to be equivalent, “To be an F just is to be G.”23 The identity of “bachelor” with “an unmarried man,” does not express a grounding relation. The relation here is symmetrical—​an unmarried man is a bachelor. There is no directionality here between something more to something less foundational. Notice that both expressions, “bachelor” and “unmarried man,” depend on the institution of marriage. Without the institution of marriage, we have neither. But once we have it, and therefore can identify individuals who are and who are not married, to be a bachelor just is to be an unmarried man, as it is vice versa. The expressions are entirely symmetrical; hence they cannot express something that is in a grounding relation. True, one might think that, even in this case, there is some constitutive relation between the elements that define what it is to be a bachelor and being one: to be a man not married constitutes, as it were, what it is to be a bachelor. But this is misleading—​to be a bachelor is simply defined, linguistically, by the negation of being married (for a man). It is not a constitutive relation any more than a dictionary definition would be. If you ask someone what is the meaning of “vixen” in English, the answer is that it is a female fox. To be a female fox does not constitute anything; it is simply

21 Rosen, “Real Definition.” 22 It is possible that Rosen himself adjusted some of his views on this from “Metaphysical Dependence” to “Real Definition.” In the earlier article, Rosen maintained both that reduction is “a metaphysical matter,” on par with grounding, but he also held the view that real definition can be expressed by a just is type of relation and that it is reduction. That was a bit confusing. In his later grounding account of real definition, Rosen expresses the relation in traditional grounding terms of in virtue of relation (“Real Definition”). 23 Along with Audi, “A Clarification and Defense of the Notion of Grounding”; and Dorr, “To Be F Is to Be G.”

36  Foundations of Institutional Reality called “vixen” in English. Similarly, for a man not to be married does not ground or constitute anything; it is just signified by the word “bachelor.”24 To be clear about this, I do not intend to argue here against Rosen’s account of real definitions. Perhaps he is right that a real definition is explicable in terms of grounding. What I am suggesting, however, is that we have good reasons to retain the distinction between just is, expressing an identity relation, and in virtue of expressing grounding relations. The latter needs to be asymmetrical. In other words, if Rosen is correct to maintain that a real definition does not violate asymmetry because there is a constitutive relation between the thing defined and the elements that define it, the relation is better expressed in terms of “in virtue of.”25 Either way, we have good reasons to keep separate identity relations from grounding relations that are not symmetrical.26 Grounding has a certain directionality that is retained even when grounding is complete. When we say that A grounds B, we suggest that A is more foundational than B, and thus in some sense explains it, which is to say that there is a direction here from the more to the less foundational. When we say that A just is B, then we might as well dispense with A; at least we could. It sometimes happens in science: we discover that A is, actually, B, and then we might drop the A from the scientific explanation and stick with B.27 Grounding, however, is not substitution. If A is grounded in B, we get a metaphysical explanation of A, not a substitution. Notice, however, that identity relations of the type “A just is B” or “to be an F just is to be G” are not necessarily analytic or a priori. Sometimes they are, as perhaps with the identity of “square” with “equilateral rectangle” or “bachelor” with “unmarried man.” But sometimes the identity relation is a posteriori: it may take scientific discovery to realize that A is actually B or that the property F is, actually, the property G.28 The discovery that (pure) water just is H2O molecules was a major discovery, not an analytical truth. In philosophy many identity relations are controversial. For example, it is certainly controversial that identity relations obtain between moral evaluations and certain mental states or dispositions, as some expressivists would maintain. In other words, the kind of identity relations expressed by A just is B or To be 24 Barring, of course, in both cases, all sorts of figurative speech associated with the terms. 25 As Rosen clearly states in “Real Definition,” 198. 26 This view is shared by, e.g., Dorr, “To Be F Is to Be G,” 43–​44; and Audi, “A Clarification and Defense of the Notion of Grounding.” 27 We don’t talk about “morning star” or “evening star” anymore; we talk about one “star,” Venus. 28 Dorr, “To Be F Is to Be G.”

Grounding and Reduction  37 an F is to be G are often anything but trivial. In fact, sometimes the distinction between identity and grounding is itself controversial—​a particular metaphysical relation might be viewed as identity by some and as grounding by others. Some, like Rosen for example, view the definition of a “square” as an “equilateral quadrangle” to stand for a grounding relation, while others view it as an identity.29 The choice would depend on the question of whether there is a hierarchical or constitutive relation between the relevant relata, and that is not always so clear. There are clear cases of each, and controversial cases in between. Also, notice that identity, unlike grounding, cannot be partial. A can be in a grounding relation to B even if A is not the complete ground of B. But an identity relation has to be complete, otherwise it is not identity—​ and that also means that identity cannot fail transitivity, whereas grounding, when not complete, can—​which is another sense in which grounding and identity come apart. Why is this distinction important? It bears on the question of how to account for reduction. The suggestion I would like to put forward is that reduction can come in two forms. Some reductive explanations take the form of a “just is” relation: A reduces to B, on this account, when A just is B.30 This is not grounding, but identity, allowing for substitution. A just is B entails that B just is A. The other type of reduction is a complete grounding: we have a reduction of facts of type A to facts of type B if and only if facts of type B fully ground facts of type A. If, for example, it is possible to give a fully reductive account of the chemical aspects of the world in terms of physical aspects, then what we have given is a complete grounding of the chemical in the physical. Similarly, if legal aspects of the world are fully reducible to institutional facts, then we would have grounded legal facts in social-​ institutional facts. In other words, on this view reduction just is complete grounds. In short, my suggestion is that a reductive explanation is ambiguous between explanation by way of an identity relation and by way of grounding. I don’t doubt that the heuristic against assuming ambiguities whenever convenient is a sound one. But this heuristic applies to entities, not theoretical concepts. Reduction is a theoretical concept, one that emerges from modes of philosophical and scientific explanations over time, spanning a great 29 See, e.g., Dorr, “To Be F Is to Be G.” 30 Audi assumes that reduction is always of the identity “just is” type of explanation (“A Clarification and Defense of the Notion of Grounding,” 110). But he gives no argument for this restriction and as far as I can see, there is none.

38  Foundations of Institutional Reality variety of contexts and methods. Therefore, it should not come as a surprise that such theoretical constructs end up being ambiguous. To suggest, as I do here, that when philosophers talk about reduction they may be talking about one of two different things is not all that suspicious; it happens all the time. Generally speaking, reduction is a mode of explanation; it is an attribute of ways of accounting for relations between things. Reduction is an explanatory concept. Both identity and grounding can be seen as a form of reduction, but they designate different kinds of metaphysical relations, subject to different constraints, and therefore also have different constraints on what kind of items they take as their objects. Grounding, as we noted above, is usually thought to apply to worldly facts. Identity relations are not constrained by that. An A just is B type of relation can apply to concepts or other modes of representation; the possible relata of identity reduction are more wide ranging. But there is one constraint, mentioned by Rosen and others, that I do take to apply to both types of reduction: they have to carry some explanatory potential. Reduction, quite generally, is a mode of explanation; it has to convey some explanatory information. Grounding does that for us, especially when the grounding relation is complete. But identity relations are informative only if they convey nontrivial information. In other words, it is not the case that every instance in which a claim can be made that A just is B, the thesis amounts to reduction.31 An identity thesis is reductive only if it conveys some information with some explanatory potential. How to distinguish nontrivial identity claims from trivial or uninformative ones is very difficult to generalize. Obviously, when the identity information is gained by learning something about the world, scientifically or otherwise, the identity reduction would be nontrivial, often very informative. But I would not confine identity reduction to that. After all, even simple analytic truths are sometimes informative or explanatory. Most reductive explanations in the realm of social ontology are about grounding, not about identity. As I mentioned earlier, metaphysical explanations in the social realm are almost inevitably drawn to an Aristotelian metaphysics, striving to explain aspects of the social world by way of more foundational facts. Thus, most reductive explanations we encounter are of the grounding type. And yet, part of what I will try to show in 31 Obviously, since everything is identical with itself, “A just is A” is not informative. At some level A and B have to be distinguishable, at least perceived or thought to have been different in some sense.

Grounding and Reduction  39 subsequent chapters is that the confusion between these two forms of reduction explains some of the confusion over the possibility of reduction in the realm of social ontology. Thus, for the sake of convenience, I will henceforth refer to reduction by identity relations as identity reduction, and by way of (complete) grounding as grounding reduction.

3 Grounding Social Rules In this chapter I want to answer Hart’s question: What is it for a social rule to exist? My purpose is to show how the idea of metaphysical grounding can help us get to an ontology of social rules in a way that complies with methodological individualism. A secondary aim of this chapter is to show that such a grounding account of the ontology of rules entails certain epistemic conclusions. In particular, it entails that a population that follows a social rule cannot be mistaken about the content of the rule they follow. Particular members of the population may be mistaken about what the rule is, of course, but it cannot be the case that most everybody gets the content of the rule wrong. The theoretical significance of this epistemic conclusion will become apparent in subsequent chapters. First, however, we need to be clear about what it is, exactly, that needs explaining. The main challenge is to specify what kind of facts need to obtain for a social rule to exist in a given population. In order to see what this challenge involves, let us recap, briefly, some of the ideas we discussed in Chapter 1. Part of what I aim to show here is that Hart’s basic intuition in response to this challenge is correct: if most people in a given population regularly conform in their behavior to what they take to be a rule, accompanied by certain attitudes and dispositions, then there is a rule. One difficulty, however, as we saw in Chapter 1, is to explicate this without assuming the point that needs to be explained. We cannot be content with the idea that there is a rule when people regard the existence of the rule as a reason for action. In order to avoid this circularity, I will start from the middle, as it were, focusing, first, on the question of what determines the content of a rule. Following my suggestion in Chapter 1, the idea that I want to develop here is that the content of social rules is constituted by collective intentions; the content of a social rule just is whatever is collectively intended by the relevant population. When we say, for example, that the word “red” stands for the color red (in English) we talk about something that is collectively intended to be the case. English speakers collectively intend to signify the color red when they utter the sound configuration “red.” Or, when we say that there is Foundations of Institutional Reality. Andrei Marmor, Oxford University Press. © Oxford University Press 2023. DOI: 10.1093/​oso/​9780197657348.003.0003

Grounding Social Rules  41 a rule to hold the fork in the left hand and the knife in the right, we talk about something that is collectively expected, intended by a collective to be the case, perhaps in an “ought” sense of “to be.” Thus, if we can give an account of collective intentions that comports with methodological individualism, we will have made considerable progress; once we have a grounding account of collective intentions, the other pieces will be easier to put into place. The next two sections take up these tasks, respectively, and in section 3.3, I will present the epistemic conclusion that follows.

3.1.  Grounding Collective Intentions In order to see the precise role of collective intentions in the ontology of social rules, it might be helpful to recap briefly what we called Hart’s Conditions-​H for the existence of a social rule: A social rule, R, exists in a population S, if and only if the following conditions obtain: 1. Most members of S regularly conform in their behavior to the content of R; and 2. Most members of S accept R as a rule, which means that a. for most members of S the existence of R constitutes a reason for action in accordance to R; and b. members of S tend to employ R and refer to it as grounds for exerting pressure on other members to conform to R, and as grounds for criticizing deviations from conformity to R.

As we noted in Chapter 1, one of the immediate problems with Conditions-​H consists in the fact that it lacks a noncircular account of what it is that people in S take to be their reason for action: What kind of fact is it? The idea of collective intentions aims to fill in this gap. The facts that members of S take to be their reason for action are facts about what is collectively intended in S; the fact that φ-​ing in C is collectively intended in S, and thus also, as will see, commonly known to be so collectively intended, is what grounds the content of the social rule in question—​it constitutes the content that is “accepted” by the members of the relevant population. Intentions and other attitudes we ascribe to groups of people, as a group, can vary enormously in type, context, complexity, and appropriate analysis.

42  Foundations of Institutional Reality For one thing, groups of people come in many forms and shapes: a football team, university, nation or ethnic group, religious congregation, corporation, legislature—​the list can go on and on. All these groups may have intentions and all sorts of other group attitudes. Corporations can even have criminal intentions. But, in many of these cases, the attribution of intention to the group relies, at least partly, on institutional facts, consisting of rules or conventions. Corporate law, for example, contains various rules determining who gets to act and to intend on behalf of a corporation. So in this case, we have an agency model determined (at least partly) ex ante by law. Or consider legislation, for example: the relevant forms of decision-​making procedures make it the case that certain intentions are attributable to the legislature in enacting a law. These rules and procedures have something to do with the voting mechanism, and are often backed by some conventions in the background, determining various ways in which the contribution of different agents may, or may not, contribute to the end result of the legislative process, including what is intended by it.1 Thus, we may have various tools, and different ones in different types of cases, for attributing attitudes to a group, as such. This is not, however, what we should be interested in here. If we want to get a metaphysical account of social rules and institutional facts, what we need is a natural type of group intention, as it were, which does not depend on institutional facts. This is what I will be referring to here by using the term collective intention. If we want to explain what a rule is by way of relying on collective intentionality, we must work with the type of collective intention that is not rule dependent. Otherwise we’re just moving in circles. Semantic intentions would be a good example. The intentions we have, as language users, in signifying X by Y are natural, as it were—​they do not depend on rules of attribution or any other institutional facts. Thus, in the way I intend to use the term here, a collective intention is a group intention that does not depend on institutional facts. I will not try to offer here a full metaphysical account of collective intentionality, however, even in this restricted sense. A complete account of collective intentionality would go far beyond the scope of this work and, in any case, is not needed. What is needed is to show that metaphysical grounding of collective intentionality in the attitudes of individuals is quite possible, and to show that some of the key assumptions that are needed to realize 1 On the connection between legislative procedures and speech acts I have elaborated in Marmor, The Language of Law, ch. 1. For a detailed account of how different procedures affect the aggregation of corporate attitudes, see List and Pettit, Group Agency, ch. 3.

Grounding Social Rules  43 such a reductive account are defensible. Thus, let me start with some simplifying assumptions I am going to make. First, about intentions and other attitudes: intention is just one of the attitudes groups of people can share as a group. Groups can also have aspirations, hopes, expectations, desires, fears, aversions, etc., and beliefs, of course. Therefore, though it may sound strange, yet not uncommon, I will use the expression “collective intention” somewhat loosely, not referring exclusively to intentions. When the distinction between different kinds of attitudes matters, I will make that explicit. The second set of assumptions relates to the kind of groups we are talking about. I will assume that though groups of people can come in countless shapes and manners, any nonarbitrary set of people can be a group or a collective, for our purposes. In fact, it is important not to assume that the group identity in question is somehow socially salient or significant. People who speak a certain natural language or a particular idiolect would be groups, in my sense, even if there is nothing else that unites them. But notice that language users, qua users of a particular language, are not an arbitrary group relative to the activity type in question and the respective collective intentions accompanying it. They are a group qua users of the particular language in question. And the same goes for people who follow certain social rules of conduct or adhere to a social convention practiced in their community. These would be groups even if there is nothing else that makes them salient as a group or distinct from other groups. What makes a group of people nonarbitrary relative to a collective attitude they share consists in the mode of sharing the attitude, and not just having it coincidently, as it were. I will explain that below. Finally, we have to allow for the possibility that the identity of the individuals who belong to the group or make it up is not constant over time. Many kinds of groups that have collective intentionality can survive changes in membership over time.2 Intentions, beliefs, and other attitudes are states of mind. The idea that groups can have a state of mind, that is, literally have it, not figuratively or metaphorically, has not gained a great deal of support in analytical philosophy. To have such attitudes one needs to have a brain, biologically speaking, that is, an organ that can think, believe, plan, intend, etc. Groups have no brain, which is to say, groups do not have mental capacities of their own, independent of the mental capacities of the individual humans constituting the

2 See List and Pettit, Group Agency, 31.

44  Foundations of Institutional Reality group. It’s just as simple as that, I take it.3 Note, however, that the sole focus of these remarks is on ascription of attitudes to groups, not the idea of agency. Whether groups, as a group, can act in ways that comport with some notion of agency is a separate issue and one that partly depends, of course, on the appropriate conception of agency.4 Group agency, however, does not entail that the group has a mind of its own. Therefore, not surprisingly, the dominant views about the metaphysics of collective intentionality are largely reductive. Philosophers have tried to offer accounts of collective intention of different kinds by way of showing how to get an intention of a collective from some combination of individuals’ intentions and perhaps other attitudes. And by this I intend to include Searle, who claims to have the opposite view. “In my view,” Searle says, “all these efforts to reduce collective intentionality to individual intentionality fail. Collective intentionality is a biologically primitive phenomenon that cannot be reduced to or eliminated in favor of something else.”5 Thankfully, Searle does not suggest here that groups are biologically equipped with some brain or mind that enables them to form intentions. What he calls collective intentionality and takes to be primitive, biologically speaking, is our ability, qua individuals, to have the kind of intention that is appropriately expressed by the locution “We intend to . . . φ” (henceforth: We-​ intention). So we need to be careful here. The reduction Searle rejects is an attempt to explain a We-​intend individual state of mind by way of reducing it to “I intend that We . . . and I believe that you believe that . . . etc.” But Searle’s own view, cryptic as it is, still allows for a reduction of intention of a collective to intentions of the individuals comprising the collective. His view differs from others in that it takes the state of mind of We-​intend to be primitive as a 3 I wish to take no sides in the famous debate about Searle’s “Chinese Room” hypothetical over the question of whether computer algorithms are sufficiently analogous to human minds. Even if artificial minds can be minds, it would not entail that groups of individuals can have a mind of their own, independent of the minds of the individuals that make up the group. 4 List and Pettit’s definition of agency is rather minimalist, and it relies on controversial assumptions about what it takes for an entity to have representational and motivational states that they see as essential for agency (Group Agency, ch. 1). On their view, even a single-​task robot is an agent. I have serious doubts that a computer’s representational algorithm is sufficiently on par with a state of mind we call “belief ” with humans, and the same goes for what they call motivational states. But this is not my concern in this work, and my skepticism about the List and Pettit framework of agency is not part of my argument here. Similarly, I think that Epstein’s critique of the excessive anthropocentric focus of methodological individualism would not necessarily bear on the grounding ideas I try to develop here (Epstein, The Ant Trap). As I indicated earlier, collective agency is a much broader phenomenon than collective intentionality. It is, therefore, quite possible that the metaphysics of collective agency require different tools from the ones I will be defending here. 5 Searle, The Construction of Social Reality, 24, and “Social Ontology and the Philosophy of Society.”

Grounding Social Rules  45 state of mind, not reducible to some combination of other attitudes. In other words, just as individuals can have ordinary intentions of the type “I intend that . . .,” people can also have, and often do, another type of intention, the type that is expressed by saying “We intend that . . .” Both types are primitive, Searle claims, and both are states of mind of individuals.6 But even if Searle is right about this, the question of how we get intention of a collective from We-​ intentions remains open. My point here is to learn a general lesson, not to put pressure on Searle’s view. The lesson is that there is a tendency to conflate two separate questions about reduction of collective intentionality to individual attitudes. One question relates to the articulation of what it means for an individual to have a We-​intention: Can this attitude be reduced to I-​intention, coupled with other attitudes and dispositions? An attempt to carry out precisely that kind of task is demonstrated by Bratman’s account of (a very limited type of) shared agency.7 Bratman offers a reductive account of shared agency by way of reducing the We-​intention to “I intend that We,” coupled with some other attitudes and pressures of instrumental rationality (derived from his planning theory of intention).8 Searle resists this move. He thinks that We-​ intention is primitive, on par with I-​intention, not reducible.9 But then there is a separate question, and the one that we are interested in here, which is how to attribute intention to a collective.10 Is it just a happy coincidence of lots of We-​intentions? That seems highly implausible. But what we need here is an account of that, of how groups or collectives can have intentions, and one that would explain how such intentions of collectives can exist. Perhaps our ability to have irreducible We-​intentions is part of the story. Perhaps it is one of the essential building blocks of collectives having intentions. Or perhaps we can extend Bratman’s reductive account to larger contexts and get to the We-​type of intention by way of building it up from I-​intention, coupled 6 Searle, The Construction of Social Reality, 24, and “Social Ontology and the Philosophy of Society,” 10. 7 Bratman explicitly limits his analysis to “shared intentional activities of small, adult groups in the absence of asymmetric authority relations within those groups, and in which the individuals who are participants remain constant over time” (Shared Agency, 7). 8 Bratman, Shared Agency and Faces of Intention. 9 I am not entirely sure about Tuomela’s view (The Philosophy of Sociality). Sometimes he gives the impression that his view is very similar to Bratman’s, except that instead of getting the We-​ intention from I-​intention via the pressures of rationality, Tuomela assumes that we can get it by way of recognizing the group-​based reasons for forming the relevant intention. So perhaps, unlike Searle, Tuomela’s view may be more reduction friendly. 10 Bratman is actually careful to keep these two questions separate, and he has not much to say on the latter (Shared Agency, 126–​131).

46  Foundations of Institutional Reality with other attitudes and dispositions. I will take no sides in this debate. With an exception or two, I will have little to say here about the appropriate analysis of the individual attitudes involved in the kind of intentions that can be shared by collectives. My questions will focus on the dependence relations between the individual attitudes in play and attitudes that are ascribable to a collective, as such. In fact, since a reductive account is all but inevitable, the main question is what type of reduction would make sense. Can we say that an intention of a collective is simply the aggregate of the intentions of its members?11 Is the relation here one of supervenience? Or is it a grounding relation between individual intentions and the intention of the collective? And if the dependence relation is one of grounding, is it partial grounding or complete? Thus, let us assume that individuals can have a We-​type of intention, that is, an intention to intend something as a group, whether this is biologically primitive or not.12 But then we should also assume that We-​intention incorporates or, if you prefer, is coupled with, whatever common knowledge requirement may be appropriate relative to the circumstances. Obviously, in order to get a collective intention from an aggregate of individual We-​ intentions, members of the relevant group need to be aware of the corresponding We-​intentions of others in the group.13 This comes to show, pace Searle, that We-​intention, biologically “primitive” somehow or not, is a rather complex set of attitudes, not a simple intention. Furthermore, people cannot have a We-​intention without having some sense of the reference of “We.” It makes little sense to say that “I have the intention that We φ in C” if I have no idea who the We are. In order to have a We-​type of intention, the individuals in question would need to have some sense, though certainly not precise knowledge, of their group. They need to have some sense of who are those We they take themselves to belong to in having the We-​intention. One might suspect that this last condition is too strong. After all, there seem to be cases where people follow rules, and thus are presumed to have grasped the content of the rule, without having much of a sense of who the other members of the population are that follow the rule. That is certainly 11 This seems to have been Hart’s assumption (The Concept of Law), but it’s difficult to know since he said very little about this issue. 12 To fill in the kind of attitudes a We-​intention would consist of we can think of some variation on Bratman’s account or a very similar account suggested by List and Pettit (Group Agency, 33). The details don’t matter here. 13 Both Bratman (Shared Agency, 57–​59) and List and Pettit (Group Agency, 33) regard the element of common knowledge as necessary here.

Grounding Social Rules  47 true, for example, about young children learning to speak their native tongue. They just follow the rules they learn somehow without much awareness that in doing so they belong to a group of, say, English speakers. Most children won’t know for a while, at least, that there are other languages besides the one they speak. However, I don’t think that such cases pose counterexamples. They come to show that the level of awareness of group membership can vary considerably and sometimes it is very minimal indeed. Even young children, however, learning to speak their native tongue, would have some awareness that in expressing themselves they need to conform to the expectations of the adults in their vicinity. They have some sense of their group membership, qua language users, the group of adults around them that expect them to say things this way rather than that, and often correct them if they don’t. Now, these points, taken together, can be employed to account for the idea that an intention is actually shared by a group of people. Three conditions constitute the idea of a group of individuals sharing an intention: (1) The fact that members of the group intend their attitude to be a group attitude (the We-​intend element). (2) The fact that the members of the group have a sense, viz., at least partial knowledge, of the group’s identity or composition. (3) The fact that there is common awareness between members of the group that others have the same or very similar We-​intention. (Note that this common awareness condition is about others’ We-​intention, that is, people are commonly aware of others intending their intention to be a group intention.) When these three conditions are satisfied, it is appropriate to say that members of a group share a certain intention (or similar attitude), and that the relevant intention is not just a happy coincidence of individuals having a similar intention in a given set of circumstances.14 Notice that I have left open whether facts of type (1) are further reducible to I-​intentions coupled with other elements, or not. Even if Searle is correct about the irreducibility of (1), attitudes cannot be said to be shared by a group without (2) and (3) also 14 In what I think is a very similar context, Margaret Gilbert insists that there is an additional condition here of joint acceptance—​namely, the idea that individuals condition their attitude on the attitude of others. I think that this may apply to some cases of joint intentions, but, like Bratman, I am doubtful that this applies across the board. See Gilbert, On Social Facts, 373–​382; and cf. Bratman, Shared Agency, 113–​120.

48  Foundations of Institutional Reality obtaining. Furthermore, the conditions of what it takes for a group of people to share an attitude also explain what makes a group nonarbitrary relative to the intentionality that is ascribed to the group. When conditions (1) through (3) obtain, there is no longer a worry that the group was gerrymandered to fit some intention that a number of people happen to have or, vice versa, the worry that for almost any individual’s intention we might come up with a number of other individuals who have the same or very similar intentions. Suppose, for example, that it is a fact that most New Yorkers intend to have bagels for brunch on Sundays. And let us assume that most New Yorkers know this fact. It would be false to conclude that we can ascribe a collective intention to New Yorkers, as a group, to have bagels for brunch. And that is so because New Yorkers would not have the intention to have bagels for brunch as a We-​type of intention. It is not an intention they intend for the group to have, as a group. And even if some do, the common awareness condition fails. It is not the case here that New Yorkers would be aware of an intention of others to have bagels for brunch as a We-​type of intention, even if they are aware of the fact that most of their fellow New Yorkers like to have bagels for brunch. It is just something that A intends to do and B intends and C intends, etc. This type of intention is not necessarily shared by New Yorkers, even if the knowledge of others having a similar intention plays some role in forming their own. In other words, a desire for conformity that may underlie or motivate an intention, even when it motivates many in a group, does not necessarily entail that the relevant intention is shared by the group, in the requisite sense of sharing. And if the intention is not shared, then the ascription of the intention to the group, as such, would not be warranted. When the relevant attitude is shared by a group, in the requisite sense, then it is a group intention, in virtue of the fact that it is shared. Intentions and similar attitudes can be shared by a population even if they are not identical. Most of the time, in fact, we are unlikely to have identical intentions that we share. Collective intention is, surely, one of those things that can remain essentially the same even if there are some variations in the individual intentions that make it up. We can call this general phenomenon variable instantiation, as it is true of many kinds of type-​token relations. It is true of many types that their properties as a type are not super sensitive to the particular tokens that instantiate the type in different contexts. As I mentioned in Chapter 1, our mental ability to associate tokens with a type is part of what enables us to use rules in our practical reasoning, which also entails that we cannot expect any slight variation in the tokens that individuals

Grounding Social Rules  49 identify with a type to affect the essential features of the type. There will always be some differences in tokens people associate with a type. If we think that each and every difference entails a difference in what the type is, we will have as many types as individuals contemplating them. Something along these lines seems to have led List and Pettit to the conclusion that the relation between individual intentions and the collective intention is one of supervenience.15 If A supervenes on B, we cannot have a difference in the A state without some difference in the B state. But we can have some differences in B that would not necessarily affect A. Supervenience is not identity. So in this respect, supervenience seems like a plausible candidate for explicating the dependence relation between collective intention and the individual We-​intentions that make it up. The problem, however, is one mentioned in Chapter 2: supervenience is a modal relation, not an explanatory one. By itself, supervenience does not tell us anything about the kind of constitutive relation that exists between its relata. It only points to a necessary correlation between changes in one domain and changes in another. In other words, though supervenience may intuitively support a reductive explanation, it doesn’t quite provide one. In fact, it is not even clear that if A supervenes on B, it necessarily precludes the availability of a non-​reductive account of A. One can concede, for example, that color perceptions supervene on all sorts of physical properties, while still holding a Platonist/​realist conception of colors. Similarly, we need to allow for the possibility that even a robust realist about moral properties might concede that moral properties supervene on nonmoral properties or aspects of the world. Supervenience does not necessarily entail a commitment to a reductive explanation. Given all this, it is much more plausible to maintain that the relation between collective intentions and individuals’ We-​ intentions is one of grounding. Individuals’ We-​intentions ground the collective intention. More precisely, when the facts that account for a group of people sharing an intention obtain, those facts would ground the collective intentionality ascribed 15 List and Pettit, Group Agency, 65–​72. List and Pettit refer to what I call here variable instantiation of types as an instance of multiple realizability of collective intentions. I actually doubt that it is. The idea of multiple realizability was introduced by Putnam as an argument for his functionalism in philosophy of mind (“The Nature of Mental States”). The idea was that mind-​brain identity reduction is implausible because mind states (understood functionally, at least) can be realized by different biological species, by different types of brains, if you will. I don’t think that this is the kind of relation List and Pettit have in mind here, and in any case, no such thesis is needed. What I call variable instantiation of types relative to tokens, which I think is what they also have in mind, is not multiple realizability. For a detailed account and critique of the latter, see Polger and Shapiro, The Multiple Realization Book. I am grateful to a reviewer for pointing this out to me.

50  Foundations of Institutional Reality to the group. Given the features of grounding we explored in the previous chapter, the idea that a grounding relation obtains between individuals’ intentions (and other attitudes) and the relevant collective intention is eminently plausible. Surely there is a clear sense in which we can say that if a collective or group shares an intention, as a group, it would have it in virtue of the attitudes of the individuals who make up the group. There is, quite obviously, a sense in which the individuals who make up the collective, with their attitudes, dispositions, etc., are more foundational than the collective’s intention. The hierarchical structure of reality that grounding presupposes seems particularly apt in this case. Furthermore, remember that the hierarchical relation between facts that ground something and the thing grounded does not entail eliminativism about the grounded aspect of reality. In this respect, the grounding picture nicely supports List and Pettit’s intuition that methodological individualism does not necessarily lead to eliminativism about collective attitudes and collective agency. And this is quite important, actually. As List and Pettit nicely explain, the views about the metaphysics of collectives have been caught for a long time between two extremes, neither one quite satisfactory.16 On one view, the “emergentist tradition,” as List and Pettit call it, a collective can have a mind of its own, independent of the minds and attitudes of the individuals who make it up. On the opposite view, collectives have absolutely no reality on their own; their entire reality, so to speak, consists of facts about individuals and their respective attitudes and dispositions.17 Seeing the metaphysical dependence relations between individuals’ We-​ intentions and collective intention in terms of grounding enables us to avoid this sharp dichotomy. We do not have to assume anything so obscure and mysterious as the idea that collectives have minds of their own, while we can still retain a very plausible aspect of the intuition driving the emergentist tradition. It is the intuition that there is a sense in which collective intentions exist, as such, as intentions of a collective. Grounding, as we saw in detail in 16 List and Pettit, Group Agency, ch. 3. 17 This dichotomy is not meant to exhaust the field; some views about the irreducibility of the social domain are not quite in line with what Pettit and List call the emergentist view, at least not in an ontological sense. See, for example, Ruben, The Metaphysics of the Social World. Ruben rejects methodological individualism on holistic grounds, and on grounds that are generally hostile to reductionism in this area. Taking up Ruben’s arguments against methodological individualism, however, is not something I can undertake in this work, as it would require going into details of various metaphysical issues not related to the present work. I’m hoping that readers will find, however, that more recent work in metaphysics, particularly on grounding and shared agency, can alleviate some of the worries that Ruben alluded to.

Grounding Social Rules  51 the previous chapter, does not entail that the things grounded do not exist. It only entails that the grounded is less foundational, perhaps even less real in some sense, as it is what it is in virtue of something else. Collective intentions exist, in virtue of the existence of individuals with their respective We-​ intentions, shared in the appropriate sense. The latter are, quite obviously, more foundational. For friends of grounds, at least, the idea that collective intentions are grounded in individuals’ attitudes seems rather obvious. The additional question is, however, whether the grounding relations that obtain here are complete or only partial. My sense is that if a relevant attitude is shared by members of a group, in the appropriate sense of sharing defined above, there is no obstacle to maintaining that the grounding relation is complete. What would be a source of doubt? Well, some might argue that people’s reasons for having and sharing the We-​intention in question must also figure in the grounding account of it.18 After all, people do not share We-​type of intentions for no reason. In many cases the reasons concern coordination functions, in others the reasons have something to do with things people value or appreciate; there are many kinds of reasons for people to share We-​intentions as a group. And those reasons would normally explain why the intention is shared by the group, as opposed to cases in which it only comes as a happy coincidence. Since it seems plausible to assume that the reasons for sharing an intention explain it as the kind of shared intention it is, would it not follow that such reasons need to be part of the facts that ground a collective intention? Perhaps the answer is yes, but some complications will be discussed in Chapter 4. Either way, as long as we bear in mind that the reasons in play are motivating reasons, not normative ones, the grounding account would not violate methodological individualism by including them. It would still be the case that we can get an account of collective intentionality by way of building it up from attitudes of individuals. A collective can share an intention, and thus have the intention as a collective, in virtue of the attitudes, and perhaps motivating reasons, of the members of the collective.

18 Though not expressed in terms of grounding, it is the view argued for by Tuomela, The Philosophy of Sociality.

52  Foundations of Institutional Reality

3.2.  The Conduct Element The metaphysical reality of collective intentions grounded in facts about sharing We-​intentions of individuals gives us a metaphysical account of the content of social rules. It tells us what it is that a rule is about, what it says. When we say that people believe the rule to exist, and to give them a reason for action, part of what we must mean is that they believe there is a certain collective intention in place. Collective intentions certainly do not give us the whole picture about what it is for a social rule to exist. In other words, it is not the case that social rules just are collective intentions, far from it. Social rules exist only if there is a pattern of conduct in the relevant population of complying with the content of the rule; there has to be an element of conduct or behavior, exhibited with some regularity, for a rule to exist. And this would seem to suggest that now, equipped with a reductive account of collective intentionality, we can return to Hart’s practice theory of rules and see whether it contains the tools we need to complete the picture. I think the answer is basically yes, provided that we understand the practice theory in terms of grounding reduction and not as a form of identity reduction. Remember that the main problem with Conditions-​H was the concern about circularity. If we take Conditions-​H to aim for something like a real definition of social rules, the definition would seem to be viciously circular. To say that there is a social rule when people believe that there is a social rule and take that to be a reason for their actions is not a good definition. Introducing the idea of collective intentions as the kind of facts that provide the content of rules goes some way toward resolving the circularity problem, but we need one more step here. And the step is to move away from seeing Conditions-​H in terms of a simple identity reduction. Conditions-​H (suitably modified) can be understood as providing us with a grounding account of social rules. Facts about social rules are, on this view, grounded in facts about patterns of behavior exhibited by individuals in the relevant population, coupled with a set of complex attitudes and dispositions motivating that behavior. These two main elements, that is, facts about individuals overtly behaving in a certain way and having certain attitudes about that behavior, ground the existence of a social rule. And since both elements—​conduct and attitudes—​are the kind of things that have a spatial-​temporal existence, we can see how the existence of social rules is contingent in the requisite sense, essentially tied to particular populations relative to a time frame. So here’s a

Grounding Social Rules  53 more schematic version of what emerges so far, taking a simple rule of conduct as our model: Suppose it is a fact that here is a social rule, R, in Si at time t, according to which all X’s who are F (Fx) ought to φ in circumstances C. “(Fx) ought to φ in C” is the prescriptive content of R.

If my arguments so far are correct, then we can say that the prescriptive content of the rule just is the collective intention shared by members of Si in the appropriate sense. So now we can see that the existence of R is grounded in facts about something very close to Conditions-​H obtaining in Si with respect to R, viz.: (1) Most members of Si regularly conform in their behavior to R’s prescriptive content. (2) Most members of Si commonly know that R’s content is collectively intended in Si and regard this fact as giving them a reason to conform and reason to exert pressure on others in Si to conform as well criticize them when they don’t. I hope we can see that there is no circularity here. We get the existence of social rules from facts that ground the collective intention, plus facts that ground the instantiation of those collective intentions in actual behavior accompanied by a typical set of attitudes and attitudinal dispositions. When we have the combination of these two sets of grounding relations, then we have an account of what it takes for a social rule to exist. There is one more complication here, however, with respect to the conduct element. A regularity of actual conformity with the content of a rule is an essential element—​without it there is no social rule in place. If there is a certain population, Si, in which people by and large do not exhibit a regularity of behavior in accordance with a rule, R, then according to the practice theory, R is not a social rule in Si. In other words, facts about the existence of a social rule are partly grounded in people’s actual conformity with the rule.19 Now, 19 In their account of social conventions, both Gilbert (On Social Facts, 347) and Millikan (“Language Conventions Made Simple,” 170) raised some doubts about regularity of conduct as a necessary condition for a convention to exist, pace Lewis (Convention). I responded to these concerns, at least specifically about conventions, in Marmor, Social Conventions, 3–​4. The text that follows here is also, in a way, a response to those doubts, since both Gilbert and Millikan base their doubts on examples of conventions that had been practiced before but ceased to be practiced widely.

54  Foundations of Institutional Reality the complication arises from the fact that the level of conformity with a rule is something that admits of degrees, and there would seem to be no way of determining what level of minimal conformity is enough. There would seem to be no facts about that, which might entail the rather uncomfortable conclusion that there are circumstances in which there is no answer to the question of whether a social rule exists or not. To see the problem here, consider the case of a social rule that used to be in place somewhere, widely and regularly followed by the relevant population, but then gradually ceased to be practiced over the years, up to the point where now almost nobody follows the rule. Now let’s put some rough figures on this: suppose at time t1, say, seventy years ago, most everybody in Si exhibited regular conformity with R; at t2, thirty years later, only about half the population in Si regularly complied with R; at t3, ten years later, only about 15 percent of the population in Si conforms to the rule; and now, at t4, almost nobody does. At the two extremes, we know the answer: R was a social rule in Si at t1, and clearly no longer is a social rule in Si at t4. But what about t2 and t3? Would it not be rather strange to suggest that there is a social rule in a given population if only a small proportion of the population exhibits a regularity of behavior according to the rule? More importantly, the problem is that it is difficult to say what kind of facts would determine the answer. How much of a regular conformity with a presumed rule is enough conformity and what renders it enough? It might be tempting to avail ourselves of the idea of vagueness here. After all, the example of a gradual dwindling of conformity with a social rule looks like a standard sorites sequence. Take a heap of sand and start removing one grain at a time. At some point there will be no heap left, but there is no way of telling what that point is. There is no point at which there is still a heap of sand but if we remove one more grain of sand, it is no longer a heap. So perhaps the problem of minimal conformity with a social rule is just an example of a sorites sequence, stemming from the vagueness of “conformity with a rule.” This may be correct, at least at the semantic level, but it does not really solve the problem. The question we are concerned with is whether a social rule exists or not; existence is not something that easily lends itself to vagueness. Talk about borderline cases of existence is something that would normally raise philosophical eyebrows, and rightly so, I take it. But I don’t think we are forced to that. The key to the solution here is to remember that the existence of a social rule is always relative to a given population at a given time, coupled with

Grounding Social Rules  55 the point I mentioned earlier, that group identity need not be socially salient: what makes a group of people count as a population that follows a rule is the rule following itself, regardless of any other feature that members of the group might have in common. Once we have a group of individuals, large or small, sharing a collective intention in the appropriate sense and exhibiting regular conformity with that intention in rule-​following behavior, we get the existence of a social rule—​it is the social rule of the population that actually follows the rule. In other words, there is no need to determine whether, in our example above, a 15 percent level of conformity with R is enough for R to be a social rule. It is a social rule for those 15 percent who regularly conform to R; they are the relevant population whose social rule it is. The fact that they are only a small percentage of some larger social group is not relevant. And that is true even if the larger group, perhaps salient as a group in some sense, is the group of people who previously followed R. If you worry at this point about the arbitrariness of defining the relevant population that follows a social rule in terms of actual compliance, regardless of any other features groups may have, remember that the conditions of what counts as sharing a collective intention that determines the content of the rule are enough to show that the relevant group is not arbitrary. What renders groups sharing an intention nonarbitrarily is determined by the conditions of sharing the intention. If those conditions obtain with respect to the content of a rule, R, and those who share the intention regularly comply with R, then we have a social rule in place. One last worry here: you might suspect that we just kicked the can down the road. What if only a handful of people follow a rule, according to the conditions we outlined above. Is that a social rule? Does it make sense to say that we have a social rule in place if only four or five people actually practice it? Here I am happy to avail myself of the idea of vagueness, not about the existence of social rules but about the question of what makes them social. We have no reason to doubt that a single individual can make themselves a rule (maybe they just assume that there is one) and follow it. We would not call it a social rule, of course. Now take a couple, or a larger family, living together, making themselves a domestic rule to φ in C, and assume they regularly conform to it. There is a rule here, for sure, but again, we would not say that it is a social rule, even if Conditions-​H, suitably modified, apply with respect to this family rule. And then you can think of a rule that is practiced by a larger family, or a clan, and so on. What counts as a social rule, distinct from other, similar cases, is probably vague.

56  Foundations of Institutional Reality It is not puzzling that the “social” label we attach to some rules and not others is vague. The vagueness has something to do with our theoretical interest in social rules and the functions they serve in building our institutional reality. In Chapter 1, I mentioned the distinction, assumed by Hart and many others, between social rules and enacted rules, rules that come about by way of being issued or directed by someone in a position of authority or power. Hart assumed, and for good reasons, as I will elaborate in greater detail in Chapter 7, that social rules are more foundational than enacted rules. Once we have a metaphysical account of social rules, we can show how enacted rules can have the institutional significance they do. But there is another reason to be interested in social rules, as such, besides their foundational role, pertaining to the very limited role of agreement in the emergence and continuous sustainment of rules. In his account of social conventions, David Lewis was particularly concerned about that.20 In fact, it was the main motivation for developing his theory of convention to show how social conventions can emerge without the need for agreement between the relevant parties.21 Lewis’s particular interests in this aspect of conventions is not our immediate concern here. But the general idea is very relevant. We know that most social rules, conventional or not, could not have come about by way of agreement between the relevant parties, and mostly because of the large number of people involved (and/​or their inability to communicate with each other). Furthermore, even if a rule may have initially emerged by way of an agreement, it often remains a social rule followed by many others regardless of the initial agreement that brought the rule into existence. Of course, many rules emerge by agreements between the relevant parties, and they are normally sustained by the agreement, as long as it lasts. The example of a domestic rule that I mentioned earlier would be a case in point. A small group of people can agree to follow a rule and then, as long as they follow it in light of that agreement, it is their rule, so to speak, a rule that exists in virtue of the agreement (and compliance) that sustains it. However, given the serious constraints on how agreements can be obtained, it is evident that we will not make enough progress in the 20 Lewis, Convention. 21 Lewis claimed that he was motivated by answering Quine’s doubts about the conventionality of language (Convention). Quine claimed that language cannot be conventional because conventions are agreements and it would take a language to agree on anything. Lewis’s coordination theory of conventions shows nicely how conventions can emerge without an agreement. As I have argued, this element of conventions can be extended to other cases, where conventions do not function to solve recurrent coordination problems (Marmor, Social Conventions).

Grounding Social Rules  57 metaphysics of institutional facts by focusing on rules that are tied too closely to general agreements.22 Part of our theoretical interest in social rules that ground our institutional reality consists in the fact that we need to account for the existence of social rules that have emerged, and are being sustained, without any agreement between the relevant parties. Lewis’s coordination theory of conventions gives one important answer to the question of how this is possible. In previous work, I have argued for the possibility of other, alternative explanations as well.23 But that is not my point here. My point is to emphasize that the idea of a social rule, that is, of what makes rules social, is to capture a very broad and diverse phenomenon, for reasons that are partly theoretical and explanatory. Therefore, it should come as no surprise that the social aspect of rules is vague and admits of borderline cases. We have clear cases of social rules, clear cases of rules constituted by agreement or enactment, and, quite possibly, borderline cases in between. But once again, keep in mind that it is not the existence of a rule that is vague here, only the classification of various rules as social or not.

3.3.  Epistemic Privilege The grounding ontology of social rules we explored here entails an important epistemic conclusion, and one that will play a crucial role in the arguments explored in Chapters 5 and 6. The conclusion is that when we have a population following a social rule, when there is a social rule in place, the population that follows the rule cannot be mistaken about the content of the rule they follow. The intuition here is, I hope, clear enough: if the content of a social rule is constituted by the collective attitudes of the population that follows it, then it is simply not possible for the relevant population to be mistaken about what the content of the rule is.24 Particular members of the group may be mistaken about what is collectively intended by the group, of course, but it just cannot be the case that most everybody in the group gets the content of 22 It is quite remarkable, or perhaps puzzling, that Searle talks about things accepted by “collective agreement” and the role of agreement in sustaining social functions, without any attempt to explain how agreements can be obtained when large and diffuse populations are concerned (The Construction of Social Reality, 39). Clearly Searle has some other, more loose sense of agreement in mind, but he never quite explains what that is. 23 Marmor, Social Conventions, chs. 2 and 3. 24 As I will show in Chapter 5, there are other aspects of a social rule, besides content, that an entire population of rule followers can be mistaken about. My remarks here are confined to the question of what is the prescriptive content of a rule, nothing beyond that.

58  Foundations of Institutional Reality the rule wrong. Let me call these types of errors group-​wide errors. What I am claiming, then, is that group-​wide errors about the content of social rules by the population that follows them is not possible. Let us begin with what I called in Chapter 1 the negation problem. Take a population that believes that they follow the commands of God, and let us assume that they regularly conform in their behavior to that belief. In other words, suppose that Conditions-​H apply here: (a) Most people in S believe that God commands them to φ in C, and therefore most of them regularly φ in C, exert pressure on others in S to do the same, and criticize them when they don’t.

Clearly the truth of (a) does not depend on the existence of God; facts described by (a) can obtain whether God exists or not. Now compare this to (b) Most people in S believe that there is a social rule R requiring them to φ in C, and therefore most of them regularly φ in C, exert pressure on others in S to comply with R, and criticize them when they don’t. Now here, to say that the facts described by (b) do not depend on the existence of R would make little sense. And it would seem to make no sense because the very existence of R, as a social rule, is constituted by a set of attitudes members of S share and their corresponding behavior in light of those attitudes. This much, I hope, we established in the previous sections. However, the question of the possibility of group-​wide errors about the content of a social rule is not quite settled. Even if people cannot be mistaken about the fact that they follow a rule, they might be mistaken about what exactly the rule requires. Thus the question is this: Even if I am right that the content of a social rule is determined by the collective intention shared by the group (in the appropriate sense of sharing), can it be the case that most, perhaps even all, members of the group are mistaken about what it is that they collectively intend? One way to see why this is not possible is to look at the ways in which an individual member of a group can be mistaken about the collective intention of the group in question. The use of natural language would be an example: suppose, as I claim, that a semantic representation of the kind “ ‘X’ stands for/​signifies Y” (in English) is fully grounded in the collective

Grounding Social Rules  59 intentions of English speakers to signify Y by “X.” Now consider an individual A, learning to speak English, beginning to use the word “X.” Their aim or intention would be to use “X” as English speakers collectively intend the use of “X”; A’s intention would be to align with the collective one. But upon coming to learn what the collective intention of the use of “X” is, A might get it wrong—​she might form erroneous beliefs about what “X” is collectively meant to signify. And, presumably, if the mistake is not grave, if A is not totally off in her grasp of the general use of “X,” her mistake might persist for quite a while, unless or until she’s corrected by others. But now think about the question of what it is that renders A’s beliefs about the meaning of “X” wrong or mistaken. The answer has to be this: A’s beliefs are false because and insofar as they are not aligned with the relevant collective intention, the collective intention of English speakers with regard to what “X” signifies. But this type of explanation of an error is not available when you try to apply it to the collective itself, to all (or most) English speakers. It makes very little sense to say that the collective’s intentions are mistaken because they are not aligned with the collective’s intentions. In other words, if the content of a social rule just is the collective intention, then the collective in question cannot be mistaken about the content of the rule. But now one might worry that there is a hidden assumption here that might be false. One might suspect that I assume that people cannot be mistaken about their own intentions, and that may be a questionable assumption. Perhaps it is possible for an individual to believe that they intend to φ in C, without actually intending to φ in C. I am not sure about this, but for the sake of the argument, let us entertain the possibility that one can believe that they intend to do something without actually intending to do it. Could something like this apply to collective intentions? Could it be the case that there is a group of people sharing the collective intention to φ and yet they are somehow mistaken about what that intention is—​they have false beliefs about what they really intend? The condition of common awareness, which is essential for sharing a collective intention, precludes this possibility. Remember that a group of people can only share a collective intention if there is common awareness among members of the group of their collective intention. So take a simplified model of a group consisting of two agents, A and B. Suppose the following facts obtain: A We-​intends the group (A and B) to have the intention to φ in C. B We-​intends the group (A and B) to have the intention to φ in C.

60  Foundations of Institutional Reality A is aware of B’s We-​intention to φ in C and B is aware of A’s We-​intention to φ in C, and they are both aware of each other’s awareness. A and B assume that together they form the relevant group.

Under these conditions, whatever it is that A and B take their We-​intention to φ in C to be just is their collective intention. If A’s belief about what φ-​ing amounts to differs from B’s belief about what φ-​ing amounts to then they fail the condition of common awareness. Suppose, for example, that “φ-​ing in C” is ambiguous: it can either be φx or φy. And suppose that A believes that the We-​intention is to φx and B believes that the We-​intention is to φy. In this case, they would not be sharing the intention to φx and they would not be sharing the intention to φy. There is no collective intention here. Therefore, even if we allow for the possibility that individuals can have false beliefs about their intentions, a group of people cannot share a collective intention, in the appropriate sense of sharing it, without actually having that intention. If they are wrong about what that intention is, they cannot meet the common awareness condition. None of this is meant to entail that collective intentions, and therefore contents of social rules, cannot be vague. Vagueness about what is collectively intended is certainly possible, and not infrequent actually. But vagueness does not entail the possibility of group-​wide error. Take the example of the word “red”: in English this sound configuration is collectively intended to signify the color red. Surely “red” is vague, and there are bound to be irresolvable disagreements or differences in opinion about borderline cases of redness in applying the word “red.” But this would not show, on any plausible theory of vagueness, that a collective of English speakers can be mistaken about what they intend by the word “red.”25 The fact that the content of social rules is often vague is surely not surprising. But again, vagueness does not support the possibility of group-​wide errors. The implications of the impossibility of group-​wide errors about contents of social rules will turn out to be important in subsequent chapters; I will return to this topic in Chapters 5 and 6, elaborating with further details. For now, let me conclude the chapter by noting that it was only about the ontology 25 Note that, even according to so-​called epistemic theories of vagueness (e.g., Williamson, Vagueness), truths about presumed borderline cases are not knowable. Thus, even if there is some sense in which people can be mistaken about whether an x is F or not, when x is a borderline case of F, if the truth about x being or not being an F is not knowable, it is not the kind of error that should concern us here.

Grounding Social Rules  61 of rules—​there is much more to the nature of social rules than the conditions of their existence. People do not just come to have collective intentions, believing that they should conform with them and actually tending to comply for no reasons. All these things come about for various reasons, serving certain functions, that must have something to do with the collective nature of the attitudes and conduct that ground the rules. In the next chapter we will explore certain attempts to account for that. The task of this chapter was to show that the existence of social rules can be grounded in facts about individuals’ behavior and attitudes. What we have seen here is that it is a two-​step grounding, as it were, whereby the grounding of collective attitudes is one step and the grounding of social rules that take certain collective attitudes as their contents is the second step, but the building blocks of both steps are in line with the constraints imposed by methodological individualism. Social rules exist in virtue of individuals’ attitudes and dispositions, and regularities of behavior motivated or guided by those shared attitudes.

4 Constitution by Rules Institutional facts are the kind of facts that are, at least partly, structured or constituted by social rules. Obviously, then, the existence of institutional facts depends on the existence of social rules. Answering the question of what it is for a social rule to exist was the first step in working out the metaphysics of institutional facts. The second step is to unpack the relation of constitution between rules and the institutional facts they constitute. What is it for rules to constitute social practices and other kinds of institutional facts? Now here, to say that the relation is one of grounding, correct as it is, is not going to be helpful enough, for two reasons that will be articulated in some detail in this chapter. First, the grounding relation between rules and practices they constitute is essentially a functional one, and this functional relation needs some explanation. In particular, it needs to be shown that the type of function in play here is reduction friendly. We want an idea of social function that admits of grounding reduction, in line with the methodological individualism defended in the previous chapter. Second, I will argue that the constitutive relation between rules and their emergent practices is bound to be only partial grounding. It takes more than rules to constitute a social practice. John Searle gave us a very interesting account of the constitutive relation between rules and institutional facts, based on two main ideas: first, that the constitutive relation here is essentially functional; and second, that rules that constitute practices have certain structural features that make them different from ordinary rules of conduct.1 My purpose in this chapter is to show that Searle was correct about the functional aspect and wrong about the structural one. My main focus will be on the functional account of the constitutive relation between rules and practices, aiming to show that, with some adjustments, Searle’s theory is basically sound. On his view, however, the functional account is closely tied to the famous distinction he introduced, many years ago, between two types of rules, constitutive and regulative, 1 Searle, The Construction of Social Reality.

Foundations of Institutional Reality. Andrei Marmor, Oxford University Press. © Oxford University Press 2023. DOI: 10.1093/​oso/​9780197657348.003.0004

Constitution by Rules  63 and I will start by raising some doubts about this distinction. I will begin by explaining why the distinction between constitutive and regulative rules is one that we should lay to rest, and then in the next section focus on the idea of function and explain in some detail how it works to ground institutional facts in social rules. Once we have a functional explanation of the relevant type of grounding here, it will also become apparent that the distinction between regulative and constitutive rules was not needed to begin with. What is crucial for the constitutive functioning of rules in relation to social practices is their systematic nature. Finally, in the last section, we will see that it takes more than social rules to constitute practices; the latter also depend on some minimally shared understandings about the point of the practice or the values it is taken to instantiate.

4.1.  Constitutive Rules According to Searle, “regulative rules regulate antecedently or independently existing forms of behavior. . . . [C]‌onstitutive rules do not merely regulate, they create or define new modes of behavior.”2 The intuition here seems clear enough: some rules regulate behavior that can be fully described independently of the rule. A rule instructing you to drive on the right side of the road, for example, is regulative, because driving on the right side of the road “could be given the same description of specification . . . whether or not the rule existed.”3 On the other hand, a checkmate cannot be described without invoking the rules that constitute the game of chess. Furthermore, constitutive rules, according to Searle, typically have the form “X counts as Y in context C.”4 The rules of chess constitute what counts as permissible moves, what counts as winning the game, etc. Though the distinction is in many ways intuitively compelling, it is fraught with difficulties, and I doubt that as a distinction between two different kinds of rules it can be made to work. In particular, it is very doubtful that constitutive rules necessarily, or even typically, have the structure Searle claims that they do, complying with the “X counts as Y in C” formula. I think that 2 Searle, Speech Acts, 33. 3 Searle, Speech Acts, 35. 4 This is a point that Searle reiterated consistently (Speech Acts, 34–​35; The Construction of Social Reality, 43–​44; “Social Ontology and the Philosophy of Society”), but, as far as I can tell, he never said that the X counts as Y in C is a necessary feature of constitutive rules, only a typical one. And in fairness, the point is never quite defended. See also Guala and Hindriks, “A Unified Social Ontology.”

64  Foundations of Institutional Reality we need a much more permissive understanding of constitutive rules. I will argue that constitutive rules are any systematic set of rules that functions to constitute structured activities. What I take to be the distinguishing mark of constitutive rules, therefore, is entirely functional—​beyond that, there isn’t any structural or logical (or linguistic, for that matter) difference between rules that constitute some social practice and rules that merely regulate behavior.5 This is attested to by the fact that rules with identical or almost identical content can have a constitutive function in one context and lack it in another.6 To take one example for now, consider a rule requiring you to take off your hat upon entering X. Suppose that in one case, the X is a church, and in another case, it is airport security. The rule about no hat in the church stands in a constitutive relation to a certain practice of courtesy that has some symbolic function (together with some other rules in the vicinity), whereas a rule requiring you to take off your hat when passing through security at an airport would have no such constitutive function. Furthermore, the idea that constitutive rules can always be formulated by the “X counts as Y in C” formula seems to me an overgeneralization. No doubt many constitutive rules have this structure, such as rules determining what counts as a game of chess, or what counts as a limited liability corporation, but many other constitutive rules have no such structure; they look like ordinary rules regulating conduct.7 It is worth keeping in mind that many constitutive rules have a proscriptive content, prescribing avoidance of some sort, abstention, or even outright prohibition. Rules about sexual abstinence and proscription of marriage are constitutive of what it takes to be a nun or 5 The idea that the distinction between constitutive and regulative rules is essentially a linguistic difference has been advanced by several writers. See, for example Warnock, The Object of Morality; and Ruben, “John Searle’s The Construction of Social Reality.” 6 I must confess that in some of my earlier publications I have been more sympathetic to Searle’s distinction between constitutive and regulative rules, trying to show that by and large it can be made to work (e.g., Marmor, Social Conventions, ch. 2). But the truth is that I have never quite subscribed to the view that all constitutive rules share a certain structure or logic distinct from regulative rules. I have certainly not subscribed to the idea that all constitutive rules can be accounted for by the X counts as Y in C formula, which is the main point I’m doubtful about here. The idea that constitutive rules necessarily have the X counts as Y formula is widely shared in the literature. See, for example, Hindriks, “Constitutive Rules, Language, and Ontology.” For a view that is much more critical of Searle’s framework, see Epstein, The Ant Trap, ch. 6. 7 My point in the text should not be confused with a different critique of the distinction between constitutive and regulative rules, which is based on the idea that all social rules have always both constitutive and regulative functions, a point made by numerous writers, including Warnock, The Object of Morality, 37–​38; and Raz, Practical Reasons and Norms, 109. I have replied to this critique (Marmor, Social Conventions, ch. 2), showing that Searle actually anticipated the objection and responded to it convincingly. See also Hindriks, suggesting a different response (“Constitutive Rules, Language, and Ontology”).

Constitution by Rules  65 monk; rules of abstaining from the consumption of products from animal farming constitute veganism; similarly, the practice of keeping “kashrut” in Judaism is constituted by an elaborate set of rules telling Jews to avoid certain kinds of foodstuff, not to mix dairy products with meat, how to keep meat and dairy kitchenware separate, etc.—​a long list of “do’s and don’ts.” Even in games, though some rules take the form of X counts as Y in C, many rules tell players what they may or may not do, and often those prohibitive rules are essential to the game, making things deliberately more challenging or exciting. In sum, I do not think that there are structural differences between rules that aim to regulate conduct for this or that purpose and rules that function to constitute something that would not have been possible, as the kind of thing it is, without the rules.8 Could one reply, on behalf of Searle, that particular rules that do not comply with the X counts as Y in C formula are just fragments of a larger system of rules that together, as a system, comply with the X counts as Y in C formula? We could thus say, for example, that the rules of soccer, taken as a whole, as an entire system of rules, determine what counts as playing soccer: as a system, they comply with the X counts as Y in C formula. I doubt that this is what Searle meant and I doubt that this solves any problems. For one, it would be difficult, on this view, to draw a clear distinction between the Y item and the circumstances C. If the Y item is the practice as a whole, then in most cases it would be identical to the circumstances in which the rules play their constitutive role. More importantly, even though there is something quite right about the idea that a system of rules counts as the practice, the counts as relation would be just another way of labeling the constitutive relation between the rules and the practice. A note of caution before we proceed: if we want to be pedantic, we would have to concede that rules, in the abstract, do not constitute anything. It is only the practice of following the rules by a given population that instantiates the constitutive relation we are talking about. Just as scores do not make music without the orchestra actually playing it, rules constitute a practice only when they are actually followed. But since our sole focus here is on social rules—​and as we have seen in the previous chapter, the latter exist only if there is a population that exhibits a regularity of conduct in accordance with

8 It may be worth noting that not all rules formulated in terms of “X counts as Y in C” are constitutive. Many criminal offenses, for example, are formulated as X counts as criminal offense of type Y; their point is not to constitute anything, just to define the content of the criminal prohibition.

66  Foundations of Institutional Reality the rule—​this is not a worry for us. Therefore, I will keep talking about the constitutive relation between rules and practices without further mention of the fact that the rules have to be followed by a given population in order to instantiate this constitutive relation. One aspect of Searle’s characterization of constitutive rules, however, is quite right and crucially important. Rules that function to constitute some institutional facts or social practices are systematic; they form a web of interlocking rules that together constitute a certain type of activity. There are hardly any free-​floating, stand-​alone, constitutive rules.9 Rules of conduct, generally, are not in this way necessarily tied to a system. A stand-​alone rule, regulating a particular mode of conduct, is totally intelligible. Some rules can be fully understood, and complied with, regardless of any systematic relations to other rules. Consider, for example, a rule that prohibits parking a car near a fire hydrant. It does not depend, in terms of its intelligibility or practical point, on forming part of a larger system of rules. Even if there were no other rules in the vicinity, only cars and streets and fire hydrants, the rule prohibiting parking a car near a hydrant would make perfect sense. On the other hand, a rule of soccer, say, that a player is not allowed to pass the ball with their hands, cannot be seen as a stand-​alone rule that would have made sense outside the game, irrespective of its contribution to what soccer is and how to play it. The systematic nature of constitutive rules might also explain why constitutive rules do not lay down a set of necessary conditions for performing the constituted activity. As Williamson rightly noted, breaking a constitutive rule does not amount to ceasing engagement in the practice.10 When one breaks a rule of a game, one does not necessarily cease to be playing the game. One who speaks substandard or ungrammatical English is still speaking English. Similarly, changes that occur in the constitutive rules of a practice do not necessarily amount to a significant change in the practice itself. Rules of tennis, for example, have changed here and there over the years, but we are not compelled to reach the conclusion that the game is different from what it was. It might still be the same game, albeit with slightly different rules.11 Both of these, and similar features of the relations between constitutive rules 9 Some particular speech acts may look as if they are constituted by a stand-​alone constitutive rule; but I doubt it makes sense to isolate rules constituting a type of speech act from the constitutive rules of speech acts that are in its vicinity, similar to it, variations on it, etc. 10 Williamson, Knowledge and Its Limits, 240. 11 I elaborated on this in greater detail in Marmor, Social Conventions, 42–​44.

Constitution by Rules  67 and the practices they constitute, are explicable by the fact that constitutive rules have a systematic nature: there is always some cluster of interlocking rules that together, as a system, constitutes an activity type. The cluster of rules may be more or less tight, but it is never quite as tight as to require full compliance. In fact, most of the time most of us engage in practices constituted by rules exhibiting only partial compliance—​we follow most, but not necessarily all of the rules. You speak English even if your English is far from perfect. And you can play chess even if you do not follow all the International Chess Federation rules governing chess tournaments. The systematic nature of constitutive rules is closely tied to the fact that rules do not constitute particular acts; they constitute a type of activity. Actions are not constituted by rules, though activities or practices can be. It is only when we have a structured, rule-​governed activity that we can say that there is a practice constituted by rules, which is the flip side of the idea that there are no stand-​alone, free-​floating, constitutive rules. Consider the example I mentioned earlier, of a rule asking people to take off their hats upon entering a church. There is something that the rule constitutes here, for sure, but not by itself. The symbolic value of this outward gesture, its point or purpose, is tied with other rules or conventions that constitute a nonverbal language of symbolic gestures, a language of manifesting or symbolizing respect by certain types of outward behavior. In other words, the constitutive function of the “no hat in church” rule cannot be explicated in isolation from an activity type that this rule forms part of. The rule requiring you to take off your hat when you pass through airport security is not quite the same—​it’s not tied to any social practice, it doesn’t manifest any particular gesture, and it has no symbolic function.

4.2.  Constitutive Function and Function as Use What is it for a set of rules to constitute an activity type? Consider the example of chess: we say that playing the game is (partly, at least) constituted by the rules. The rules of chess constitute what it is to play that game. Is that relation of constitution one of grounding? In one sense, clearly yes. An activity is playing chess (when it is) in virtue of players following the rules of chess. So yes, playing the game is (partly) grounded in the rules that make it up. This is true, but not sufficiently illuminating. We still want to know what kind of grounding relation obtains here. On Searle’s account, constitutive rules

68  Foundations of Institutional Reality ground an institutional fact by way of assigning a function. His explanation of this functional assignment relies on the idea that constitutive rules have the typical structure of “X counts as Y in C.” Thus, when we have an “X that counts as Y in C,” the relation can be expressed by saying that the function of X is to count as, or stand for, Y in C. As I mentioned above, I don’t think we can accept this formula to apply to all cases. Nevertheless, the functional account of the relation of constitution can stand on its own, as long as we understand this functional aspect of rules in their systematic context. Let us work with two very simple examples: (1) The word [a type of sound configuration] “red” signifies or stands for red [the color] in English. (2) A silver coin [a type of physical object made of silver with a certain weight and shape and insignia] counts as money [means of commercial exchange] in C.

In both cases, the relation of constitution between the two relevant items, call them X and Y, assigns a function. In (1) X is a certain sound type that is assigned the function of signifying a color type; and in (2) X, a type of object, is assigned the function of serving as a means of commercial exchange, say Y. In both cases, however, the assignment of function to the X item, to count as or used as Y, does not stand on its own. The functional assignment makes a contribution to a practice that is already there, so to speak, at least in some rudimentary form. When people started using minted coins as means of exchange, thereby establishing the institution of money, they introduced a significant change to practices of commercial exchange that had existed before. The function of silver coins to serve as money is functional, for sure, but the function consists in a new contribution to a practice that was already constituted by other rules and conventions.12 Something similar applies to the use of color words, like “red.” Redness, the color itself, is not constituted by rules, of course. The Y item exists, as such, regardless of the X item that has the function of signifying it. But the function of the color word “red” cannot be 12 Means of commercial exchange had existed centuries, if not millennia, before the advent of money. The history of the invention of coined money is still not quite certain; according to some accounts it started in the Greek island of Aegina. According to other sources, Lydians were first to mint coins, and there are some other possibilities. In any case, it is widely assumed that the invention of minted coins as means of exchange dates to the seventh to fifth century BC in Greece. From there it quickly spread to Egypt and the Persian empire. The first coins were actually an alloy of silver and gold.

Constitution by Rules  69 isolated from its contribution to a language of systematic representations of colors and of lots of other things. The function of “red,” as of any other word, partly consists in its contribution to a set of institutional facts constituted by other rules. All of this just comes to express the idea that the function of a constitutive relation between rules and practices cannot be detached from their contribution to a system of rules that together, as a system, enable people to engage in certain types of activities. As I mentioned earlier, there are no free-​floating constitutive rules, constitution is systematic, and the function of particular rules usually consists in the contribution they make to the functioning of other rules or aspects of the relevant practice. Typically, as the examples above demonstrate, the Y item is constituted by a whole cluster of X items that together, in various combinations, function as Y and thereby constitute a practice that would not have existed without the rules and systematic relations between them. In other words, I think that Searle’s functional explanation of the constitutive relation between rules and institutional facts is basically correct, as long as we make two adjustments. First, we need to relax the structural requirement that constitutive rules typically abide by of the “X counts as Y in C” formula; that should not be seen as essential. Second, we need to expand the functional aspect of the constitutive relation to include the contribution of a rule to the functioning of other rules and institutional facts in their vicinity. Functions of rules tend to depend on functions of other rules that are systematically related to them.13 It may be worth mentioning that assignment of function does not have to be deliberate, and it often does not come about by some intentional design. Functional assignments tend to evolve by small steps that often take time to materialize. Historians sometimes tell us an interesting story about these evolutions, and we often see that the story is not about an invention, but about a gradual, almost inadvertent process that takes time to galvanize into a set of rules that constitute an established practice.14 In other words, functions are usually not deliberately assigned; they evolve gradually by 13 This is an old theme, well recognized at the time by the Oxbridge school of ordinary language analysis; philosophers like Wittgenstein and Ryle have emphasized that the functioning of words in language games is essentially systematic, depending on the functioning of other words in the vicinity and ways in which these functionings interact. See, for example, Ryle, Dilemmas, 33–​34. 14 The invention of paper money is a good example. Roughly, it’s a story about wealthy European merchants issuing individual letters of IOU, which gradually became transferable to third parties with increasing ease, endorsed by merchant banks, and later monopolized by central governments. There was no invention here, but only small steps evolving over time.

70  Foundations of Institutional Reality patterns of practical use. The practical use of an X to count as Y, or to serve Y purposes, etc., is what function actually is. But in order to see how this works, we need to say something about what functions are and how to understand the idea of a function in the context of our discussion here. There are countless things in our lives that serve various functions. Our heart functions to pump blood in our bodies. Tangible artifacts, such as chairs or screwdrivers, serve practical functions of allowing us to do this or that. Money serves the function of commercial means of exchanges. Stories may serve functions of enriching our imaginative cognition and emotional or social awareness. And so on. But what is a function, and are these different functional relations of one and the same kind? One aspect of functional terms that seems to apply to all functional relations is that they come with a whole vocabulary of success and failure, of serving a function better and worse, more or less efficiently or smoothly, and things like that.15 At least in this respect, function is very different from cause: causal relations do not come with a vocabulary of success and failure. One thing either causes another or it does not. There is no room for a vocabulary of better and worse about causal relations. And this lends some support to Searle’s view that, unlike causation, functions are not the kind of things that we can discover in the nature of things. Functions are not intrinsic features of things that exist independently of us humans. They are always observer relative, that is, assigned or imposed by us, relative to our values and interests. As Searle puts it, “As far as nature is concerned intrinsically, there are no functional facts beyond causal facts. The further assignment of function is observer relative.”16 Needless to say, not everybody agrees. The question of whether all functions are observer relative or extrinsic to nature is something that has been debated for a long time, in philosophy of mind and in philosophy of biology, but there is no need for us to try to resolve these issues here. First, in the social domain, when we talk about the relations between rules and practices they constitute, it seems quite plausible to assume that the functional terms we use are assigned, not discovered.17 In other words, even if Searle’s view about the nature of function doesn’t necessarily hold in all domains, it is clearly apt in the domain of social ontology. Second, we need 15 Searle, The Construction of Social Reality, 15. 16 Searle, The Construction of Social Reality, 16. 17 The term “assignment” of function is not meant to imply that the assignment is deliberate. Searle’s use of the word “assignment” is only meant to capture the idea that functions are not intrinsic to nature; they are extrinsic properties, depending on us, humans.

Constitution by Rules  71 to distinguish between practical functions and functional explanations, and this might allow us to avoid some unnecessary complications.18 A practical function of X is the use we put X to.19 Practical functions are clearly observer relative in Searle’s sense. To say that the function of chairs (generally, as a type, not necessarily every token, of course) is to enable us to sit more comfortably, is to say that we make certain use of objects or things, practical use. And this applies not only to physical objects: practical use can be intellectual, aesthetic, or otherwise non-​tangible. Either way, practical function just is the use we make of things. And notice that the continual use of an object is what amounts to its function—​once the object is no longer used for whatever it had been used for, it no longer has that function. I can take a small stone and use it as a paperweight for a while, so the stone has the function of holding down papers on my desk. But once I throw the stone away and no longer use it as a paperweight, it no longer has that function. Of course the kind of functions we are interested in here—​concerning the function of constitutive rules—​are not about individual use. We are talking about general use by a certain population over time. Use, whether individual or collective, is typically a purposeful activity. But since we have already seen that collective attitudes are possible, and can be grounded in individuals’ attitudes, the collective-​attitude element would not introduce any new complications in our explanation of function. On the contrary, as long as functions can be analyzed in terms of purposeful use, we would have the metaphysical tools to explain practical function by way of reducing it to patterns of behavior and collective attitudes. Take the function of a screwdriver, for example. Its function consists in the use of it to turn screws in or out of a surface, or something like that. The use is followed as a pattern, by many individuals at different times, and in ways that are quite public—​we

18 The idea of function and functionalism in philosophy of mind is rather different from what I aim to explicate here. Function was understood, initially at least, in analogy with computational systems, and the functionalist argument was aimed to disprove the possibility of mind-​brain identity reduction. The thought was that the same mental states, understood as performing something like computational functions, can be realized by a whole range of biological organisms (the idea of multiple realizability) and that therefore it makes no sense to identify mental functions with brain states (Putnam, Mind, Language and Reality, ch. 21). Thus functionalism in philosophy of mind, at least in the old computational model, is actually an anti-​reductionist position (see, e.g., Polger and Shapiro, The Multiple Realization Book, ch. 2). As will become clear in this and the next chapters, I use the idea of functional explanation in the social domain in the exact opposite direction—​function as practical use is fully reducible to patterns of behavior and their accompanying attitudes. I do not claim, however, that this has any bearing on functionalism in philosophy of mind, however that notion is understood these days. 19 Searle calls this function as use “agentive function” (The Construction of Social Reality, 20).

72  Foundations of Institutional Reality get to learn from others what screwdrivers are for and how to use them. The function of screwdrivers is therefore determined by a pattern of purposeful use, that is, some regularity of conduct accompanied by collective, public, learnable attitudes. Thus (practical) function, generally, just is purposeful use. When I use something as a paperweight, it is thereby the function of the object to be a paperweight, and it is the object’s function only because and insofar as I use it that way.20 Practical use can change over time, of course. Something that had been used for F can come to be used for G, and then it would acquire a new function. Similarly, an object that is designed or mostly used for F can also be used, on occasion, for G, and then it would have multiple functions, as when a chair is used for standing on it to reach a high-​placed object, or such. Use, in other words, changes and adapts over time, which is consistent with the idea that functions are not necessarily unique to objects or things; one and the same item can have multiple functions or functions that change over time. Things can have a primary designated function and various derivative or occasional functions, standard and nonstandard uses, and so on.21 Not all talk about functions, however, is talk about use. To say that the function of the heart is to pump blood is not to suggest that we use our heart for this purpose, as if we could have used it for something else. Our body, as an organism, is not using the heart. Similarly, to say that the function of photosynthesis is to turn one type of energy into another is not a way of assigning purposeful use to plants. Our use of functional language in these cases is theoretical—​we give a functional explanation of something that happens irrespective of us.22 Hearts will keep pumping blood, or not, regardless of our practical purposes and interests. True enough, heart failure would be a failure for us, something bad that happens, but only because and insofar as we have an interest in the continued existence of the organism in question. 20 Houkes and Vermaas argue that a theory about the function of artifacts needs to be replaced by a theory of use (“Actions Versus Function”). Much of what they say about the use of artifacts seems right to me, but, strangely enough, they never consider the idea that function, at least in the practical domain, just is use. 21 See, for example, Preston, “Why Is a Wing Like a Spoon?,” 241. On the function of artifacts, however, I will have more to say in the next chapter. In her account of artifact functions, Preston borrows the notion of axaptation from biology, but I think that the account I offer makes this move somewhat unnecessary. On the general idea of functional changes and adaptations of artifacts I’m largely in agreement, however. 22 If I understand him correctly, this is what Searle calls “nonagentive” functions (The Construction of Social Reality, 20). He also hints at the idea that nonagentive functions are closely tied to explanations. But then he also suggests that there is no sharp dividing line between the two types of function, which I find a bit confusing.

Constitution by Rules  73 In other words, the vocabulary of success and failure in functional explanations is just as interest or value relative as in the case of function as use. But the function of “function” is different. When we use functional language as part of some theoretical explanation, the reference of “function” is not about practical use, as we would talk about the function of paperweights, chairs, stories, or myths. And this suggests that we may be talking about a rather different sense of function here. How to explain the concept of function in functional explanations of natural phenomena is a very tricky matter. Think about the heart example: What does it mean to say that the function of the heart is to pump blood? Our biological organism does not aim at anything, and cannot be said to use this or that part of it for some purpose or other. At the very least we can agree that the function of the heart does not depend on the use to which we put it to. So perhaps all we are saying here is that pumping blood in our body is simply what the heart does, like talking about what heat does to the velocity of molecules.23 But this won’t quite do. What the rain does, for example, is to water the soil. But we wouldn’t say that the function of rain is to water the soil. These are just things that happen, with no functional explanation that would seem apt here. I think that at least part of what makes the functional explanation of the heart appropriate has something to do with the fact that we are familiar with pumping mechanisms in other contexts. There are all sorts of contraptions, artificial and natural, that move liquid in tubes against gravitation in similar ways. Thus, by saying that the function of the heart is to pump blood, we implicitly invoke these similarities—​as if we say what the heart does is very similar to what a water pump (and oil pump, etc.) does to move liquid through tubes, or something like that. Functional ascription here is just a way of explaining something that happens by way of tying it to similar mechanisms and, in this case, to one that we have a concept for: the mechanism of a pump.

23 Analyzing natural functions in causal terms has been suggested in the literature by several philosophers. The leading paper is probably Wright, “Functions.” Ruth Millikan’s theory of function, which is also basically causal, strives to model functional explanations in biology, mostly, adding to the causal element a criterion of successful reproduction (Language, Thought, and Other Biological Categories). Her theory of function has been subject to considerable controversy (see, e.g., Preston, “Why Is a Wing Like a Spoon?”). However, I agree with Searle (The Construction of Social Reality, 17–​18), that even if Millikan’s technical definition of a “proper function” in terms of a history of reproduction is useful, it would not capture our ordinary notion of function, outside of biology, at least. Things can serve various functions without a history of reproduction, as my stone qua paperweight example would attest.

74  Foundations of Institutional Reality You might suspect, however, that this is not good enough. Some people’s heart does not pump blood well enough or as well as it should. When we speak of a function we can also speak of malfunction, and vice versa. But here I think that Searle is quite right to suggest that applying a language of success and failure to functional aspects of nature is always relative to human interests and concerns. We wouldn’t say that a virus malfunctions when it fails to infect a person with the flu. A star does not malfunction when it explodes—​supernova is not a malfunctioning of a star, it’s just what happens to it at a certain point. In other words, there is a clear sense in which functional explanations of natural phenomena are quasi-​metaphorical. We compare or measure things that happen in the world to our interests, concerns, and desires, as if aspects of nature were like artifacts we created, things we can use in this or that way. There is a good measure of anthropomorphism in the functional language we use when we talk about natural functions. Those who think that there are some inherent teloi in nature would not find any of this convincing. And that is fine. I did not intend to establish here a mechanistic conception of nature, even if it is the view that I do happen to hold. My point was to show that functional explanations are compatible with a mechanistic worldview, as long as we understand the idea of function as relative to our theoretical or explanatory interests, and thus functional explanations of natural phenomena as a form of explanation by similarities, analogy, or even metaphor.24 Function as use, however, is not metaphorical or incomplete in any sense. It is fully grounded in practical, observable aspects of our regularities of conduct and shared, learnable attitudes accompanying those regularities of use. And, of course, this is the sense in which I want to say that constitutive rules function to establish practices and institutional facts. The relation of constitution is a function of rules, exhibited in the rule-​following behavior of the relevant population, accompanied by some shared attitudes that render the rule following meaningful or purposive for the population in question. The idea that function is typically purposeful use might be objected to on grounds that there are cases in which functions are hidden, below the surface, as it were; things can function in ways that people are not necessarily aware 24 Many people think, and in my view, rightly so, that one of Darwin’s greatest achievements in his theory of evolution was precisely the ability to show that there is no need for genuine functional explanation of evolution; evolutionary processes are reducible to causal links, one thing brining about another, which looks as if there is function in nature. Functional ascriptions are ex post, explanatory perhaps, but not referring to genuine natural forces.

Constitution by Rules  75 of. In fact, sometimes it might even be the case that an X can fulfill the function of Y for S only if S is not aware of this functional relation. Marx’s theory about the function of ideology in masking class conflict and the exploitation of the proletariat is a case in point.25 The function of ideology, on Marx’s account, is not on the surface; it does not necessarily exhibit purposeful use of ideological tools by the bourgeoisie to deceive the exploited classes. On the contrary, ideology can function as a form of false consciousness precisely because its function is hidden, not widely conceived as such. Marx would not deny that ideology is used in certain ways, but he would have probably insisted that the use is not purposeful. And of course, Marx’s functional explanation of ideology is not the only example we have of explanations that rely on hidden functions. In Chapter 6 we will encounter other examples. Do we need to concede that even if function is practical use, the use need not be purposeful? Not quite. The idea of purposeful action needs to be understood in a rather relaxed manner. But this is not a problem. We would not be stretching the concept too far. Purposefully φ-​ing is not necessarily tantamount to φ-​ing with a particular purpose or aim in mind. Purposeful action is, generally, aim driven, but the particular aim of the conduct in question need not be at the front of the agent’s mind, as it were. There are many things we do purposefully without being aware of it, or without having the purpose of the act in the front of our mind. We will return to this issue, and elaborate on it further, in Chapter 6. For now, a few quick remarks might be in order. One, which we just noted above, is that not every functional explanation relies on its object actually having the function ascribed to it by the explanation. Functional explanations are often oblique, casual explanations, and that may also apply to the social domain, of course. Second, we need to be careful in dealing with the ascription of functionality on a massive scale, such as the functions we ascribe to ideology, or the state, or capitalism, and things like that. Ideology is comprised of countless social practices and collective endeavors, partly independent and partly interacting in various ways. Ascribing functions to things on such a massive scale is always a huge simplification—​perhaps theoretically warranted, but simplification nevertheless. Lots of discrete things can have particular functions, functions understood as purposeful use, while taken together, under some theoretical generalizations and abstractions, we can come to see that they have some 25 For a concise and elegant formulation of Marx’s view on this, see A Contribution to a Critique of Political Economy.

76  Foundations of Institutional Reality functions hitherto unnoticed or not actually aimed at by the relevant agents. Finally, it is worth keeping in mind that social functions involve collective attitudes, and those tend to be rather rough and vague sometimes, in ways that might obscure the distinction between what is clearly some people’s purpose and what is a bit more obscure or under the radar of social awareness. Let us return to our immediate concerns. So far, the idea is that the constitutive relation between rules and the practices they enable is functional, and this function consists in the practical use to which rules are put. But the explanation is not complete; we need one more ingredient, another important aspect of the idea of function as use, which is something we can call the added value of function. The idea is very simple but crucially important. Whenever we use X for Y, we add value to X (and sometimes to Y as well) compared to the value X had prior to its use as Y. Take the simplest of examples: I have a pile of papers on my desk, in danger of flying all over the place in case I open the window. I also happen to have a small stone nearby, so I take the stone, put in on the papers, using it now as a paperweight. The stone may have had some value prior to this use, or maybe none at all, but now, by using it in a certain way, value is added to it. It now serves a purpose, achieves some good (for me, at least), by functioning in a certain way. The function of X used for Y adds value to X—​it makes X valuable in ways that were not there prior to its acquired function. And then sometimes, quite often actually, when Y is complex, changing an aspect of it by using X may add some value to Y as well.26 Needless to say, the good or value added by function as use is not necessarily moral or aesthetic good, and it’s not necessarily good, objectively speaking. The good is always relative to the purposes or desires of the user. Furthermore, it is important to remember that the kind of values in play here are often symbolic or representational. When our ancient ancestors carved out a figurine to represent, let us assume, some deity or other they believed in, the material object gained symbolic value; its use was representational, standing in for something else, symbolizing it. Anthropologists tell us that human capacity to symbolize, that is, to use some material object to represent something beyond its physical properties, is the foundation of human culture, the mental capacity that made humans distinct from other species.

26 Once again, the introduction of coined money is a good example: the minted coins were used as means of exchange, thereby adding value to the objects by way of using them as means of exchange, but at the same time, the commercial means of exchange expanded in value, making them more efficient and easier to use.

Constitution by Rules  77 Symbolic functions (or “status” functions, as Searle calls them) remain, of course, central to our culture, and are pervasive aspects of our sociality.27 We are now in a position to pull some of the threads together. The constitutive relation between rules and practices is a functional matter: when a system of rules serves the function of enabling and structuring a type of activity that could not have existed without the rule-​following behavior, the rules constitute the practice. The added value of this function is the value of the practice, the value of the activity type that the rules constitute. I hope it is now evident that the constitutive relation, understood in terms of function as use, does not depend on the kind or the structure of the rules in play. It is the systematic nature of the rules, and their interconnectedness, that enables their constitutive function. Activities, by their nature, have a certain complexity that can only be structured by rules if the rules form an interlocking system. One rule hardly ever makes a practice, no matter how the rule is formulated. Practices have a structure that can only be constituted by a system of rules governing various aspects of the activity. The constitutive relation here, however, can be grounded in the ways in which the rules are used, that is, in patterns of conduct and the shared attitudes accompanying them. To put things simply: there is no distinction in form or logic or structure between constitutive and regulative rules. Rules of all kinds can have a constitutive function when they are used by a population to structure an activity that is made possible by some systematic rule-​following behavior. Rules are not constitutive; we can call them regulative, if you will, when they are used for other purposes. And those other purposes are multifarious: rules are sometimes used to coordinate behavior, to save deliberation time, to encourage morally desirable behavior, to educate, subjugate, or control a population, and whatnot. Rules have a constitutive function when they are used to structure an ongoing type of activity, a practice that would not have been possible but for the rule-​following aspect of it. What makes rules constitutive is their function in grounding institutional facts—​the relation between the rules and the activities they constitute is one of grounding. However, as we will see in the next section, this grounding relation is not complete; it takes more than rules to ground social practices.



27 Searle, The Construction of Social Reality, 41.

78  Foundations of Institutional Reality

4.3.  Constitutive Rules as Partial Grounds When rules constitute a social practice—​when rules constitute a certain type of activity that a given population engages in over some period of time, with some regularity, we have a grounding relation between the practice and the rules. The practice is what it is in virtue of the rules that make it up. All of this seems right, but now I would like to show that this grounding relation is not complete. It takes more than rules to ground a social practice. And this is nicely attested to by the fact that the grounding relation between rules and practices may fail transitivity. Let me use the example I gave in Chapter 1 about an imaginary practice of chess*, where the rules of the game are identical to chess, but the practice is partly religious in nature, aimed at revealing the divine grace of the gods. So now, think about the relation between the rules of the game and the practice in terms of a grounding relation. We assume that the rules of chess* in society Si are identical to the rules of chess in ours, S. Therefore, in both cases the rules constitute, that is, ground the game. But then in the case of chess*, the game is also constitutive of (viz., used for) some religious purposes with religious significance assigned to it in Si. Schematically, we can put it this way: In Si: Rc [rules of chess] grounds X and X grounds Y [Y =​religious significance of X]. Therefore, by transitivity: Rc grounds Y. In S: Rc ground X but not Y.

One would have to conclude that transitivity fails here. Otherwise it would be mysterious why Rc does not ground Y for us, in S, playing chess, not chess*. This seems to suggest that there are other factors, besides rules, that stand in a grounding relation to a social practice. The difference between chess and chess* points us in the direction of what is missing in the grounding relations between rules and practices—​namely, the collective understanding by the relevant population of what the rules are for, what the point of the practice is, for them. And this is closely related to the fact that when people engage in a certain type of activity, they do it for reasons that reflect the values and purposes they associate with the activity in question. Remember that constitution by rules is a functional grounding, and like any functional assignment, there is an added value to the objects assigned some function. In other words, it is never just rules that stand in a

Constitution by Rules  79 grounding relation to a social practice, but also some shared understandings of what the rules are for and what the point of the activity is. A constitutive relation between rules and practices is not complete without this further element of shared attitudes about the purpose of the activity or its point; participants must exhibit some shared understanding of what the practice is for, an understanding that would normally figure in their (motivating) reasons for following the rules that constitute the practice. A word of caution is in place here. We know from countless social practices we are familiar with that people who engage in the practice may have different, sometimes even conflicting, views about the values and aims they associate with the practice. In other words, a social practice can exist, and even thrive, despite considerable disagreement among people who engage in it (and sometimes those who refuse to engage in it) about the point of the practice and reasons to participate. Consider the example of kashrut mentioned earlier. Some people practice it because they believe that it has been decreed by God. Others practice kashrut not for religious reasons, but mostly because they value a tradition that they identify with, and wish to maintain it as such, as a collective tradition of the group they belong to. And this is true about most social practices we see—​there’s almost always some divergence of views about what the practice is good for and about the reasons for following its rules.28 So we must be careful not to assume that there has to be anything close to unanimous agreement among participants in a social practice about its aims and values. That would be patently false. And yet, some agreement in judgments there has to be. Suppose we have two groups of people following more or less the same rules that stand in a constitutive relation to their activities, but they have two very different views about the point of those rules and the values they associate with the activity—​ so much so that the two views have absolutely nothing in common. Suppose, for example, that one group regards the practice as a means of physical exercise and the other group regards it as a religious ritual. It would be strange to say that we have one social practice here, practiced for very different reasons by two groups of participants. We might as well say that we have two social practices here constituted by the same set of rules. The point, however, is not to work out the criteria for individuation of social practices. I doubt that it would be a fruitful line of inquiry. Practices can overlap and partly diverge in 28 Natural language might be sui generis in this respect, people do not tend to have divergent views about reasons to speak a language or the values of engaging in linguistic communication.

80  Foundations of Institutional Reality numerous ways, and thus trying to figure out criteria of individuation might be nothing more than a frustrating search for something that is not there. The point about some necessary agreement in judgments is different: it is to emphasize that practices must exhibit some minimal level of shared understanding among participants about the practice’s putative values. Remember that the constitutive relation between rules and activities is essentially functional—​rules have their constitutive function in virtue of the use they are put to by the population that follows them. Things that are put to a certain use must be presumed to serve a purpose, to enable or facilitate something that has some value for those who use it. I called it the added value of function. You cannot use something for nothing. Use necessarily reflects some evaluative dimension: if you use X for Y, then there is some sense in which you think that Y has some value for you. This value might be utterly mundane, as the value of nailing something to the wall, or it may be sublime, as the value of reading a great novel or using medicine to save lives. It cannot be overemphasized that talk of values here is meant in subjective terms, from the standpoint of those who engage in the practice. Remember that we confined all talk of reasons for action in this book to motivating reasons, not normative ones. In other words, from a metaphysical perspective, a neo-​Nazi club and a literary book club are on par even if the former is morally reprehensible and the latter is noble. We are not talking about values here objectively, from a normative standpoint. Values are to be understood here as the kind of purposes that rationalize some practical function, that explain why something is used for this or that purpose by those who use it. In this sense, practical function is necessarily reflective of some value judgment.29 The question we need to answer here, however, is why is it the case, and in what sense, that some minimal level of agreement in these evaluative judgments is necessary for a social practice to exist. The answer is basically practical, not metaphysical; it has something to do with the fact that conduct structured or regulated by rules is in need of interpretation, adjustment to new circumstances, and all sorts of modifications that are called for by the complexities of our social lives. To be sure, I am not suggesting that rules cannot be fully grasped and applied without some interpretation or other. Far from it.30 But practices are complex, and we 29 There are complex relations between practices and values. I elaborated on some possible relations in Marmor, Social Conventions, ch. 2. See also Guala and Hindriks, “A Unified Social Ontology.” 30 I have long argued that the application of rules is not necessarily mediated by interpretation. See, for example, Marmor, Interpretation and Legal Theory, ch. 7; and Philosophy of Law, ch. 6.

Constitution by Rules  81 have already seen that constitutive rules do not lay down a set of necessary conditions for performing the constituted activity. By breaking one of the rules of a game you do not necessarily cease to be playing it. There is, therefore, a great deal of flexibility in the actual compliance with rules that make up a social practice, and this flexibility is often needed in response to changing circumstances or new problems that might come up. We sometimes need to cut corners, or adjust rules to particular needs, and all sorts of things like that. The point is, however, that these practical challenges and adjustments we have to make in following constitutive rules require a minimal level of agreement in judgments about the point of the practice, about what makes it valuable for those who engage in it. The example of breaking a rule of a game can help us see the point here. When you see a group of kids playing soccer on the street, you can tell two things. First, they clearly do not follow all the rules of the game. A great deal of improvisation might be going on there. But you can also tell, with some confidence, that the kids are playing soccer (and not some other game). It is improvised soccer, for sure, but there is also a limit to how much improvisation is compatible with playing the game. If some of the kids start throwing the ball to each other by hand, they would no longer be playing soccer, not even an improvised version of it. Some rules, you might say, are essential to what the game is. But what makes it essential to the game that you follow, more or less, this rule and not necessarily that, is bound to depend on some shared understandings about what the point of the game is, what makes it the kind of game that it is. Kids who do not understand that the main challenge of soccer is to advance the ball by foot, not by hand, fail to grasp the nature of the game. They would be playing something, but not soccer. I think that this example generalizes to all cases. An improvised or modified version of a social practice is still the practice, but only up to a point. Without some shared understandings of what the practice is for, what its main points are, as it were, we do not engage in the same practice. In some cases, perhaps, we may have invented a new one, or split the practice into a new version of it, but that is a different matter. Practices evolve, and sometimes change character radically.31 That is something to be borne in mind, but it does not affect the present argument. And the argument is simple: we 31 The history of American football is a nice example: it was developed by college coaches in the late nineteenth and early twentieth century, introducing successive rule changes to UK football and rugby, up to the point where it has become a rather different game, only vaguely related to its predecessors.

82  Foundations of Institutional Reality can only come to understand why different versions of practices constituted by rules, and different interpretations of the rules, are still essentially the same practice, by realizing that the participants share some understandings of the main values that the practice is there to achieve or instantiate. That is why some minimal level of agreement in evaluative judgments is necessary for a social practice to exist. Without some agreement in judgments about what the practice is good for, participants would not be able to have any sense of what modifications, improvisations, or interpretations are acceptable while still engaging in the practice. One might object that there is too much essentialism presupposed by the argument, as if social practices must have an evaluative or purposeful essence, and then I am just assuming that participants must share an understanding of this essence, otherwise they do not engage in the practice. But this is not the argument. It is true, regardless of anything else, that in most social practices, some constitutive rules are more central or essential to what the practice is than other rules. It is difficult to deny this fact in face of abundant examples. The rule forbidding players to pass the ball by hand is clearly more central to soccer than the offense of offside. People can play soccer without observing the latter, but not without observing the former. Similarly, some institutional facts are much more central to law than others—​we cannot have a legal system without some institution making laws and some institution entrusted with applying them. But there are well-​functioning legal systems without all sorts of rules and practices we normally associate with a legal system, such as a written constitution or copyright protections.32 Nevertheless, the point of my argument about the need for a minimal level of agreement in judgments about what the practice is good for is not based on essentialism or, more generally, on metaphysics. The argument is meant to be understood in practical terms. It is simply a fact, practically speaking, that the way we go about making exceptions, adjustments, and modifications to the constitutive rules we follow requires some shared understandings about the aim of the practice, its main point or value. And that would be true even if, metaphysically speaking, practices have no constitutive essences. What matters, practically speaking, is the shared understandings of the participants about the point of their practice, not the question of whether those understandings reflect, or not, any truths that might exist independently of their attitudes. So yes, 32 Something similar is also the case about the role of some participants in comparison to others; in most practices some of the practitioners have a more central role to play than others. See Marmor, Social Conventions, 46–​47.

Constitution by Rules  83 perhaps some vague essentialism is hidden here, but not of a deep metaphysical kind, at least nothing that should strike one as esoteric or mysterious. Let me conclude by summarizing the main theses that I aimed to establish in this chapter. I purported to show that the constitutive relation between rules and practices is essentially functional: a system of rules constitutes a social practice when the system functions in a certain way. Function, I have argued, is a matter of practical use. When a system of rules is used to enable a certain type of activity that would not have been possible without the rule-​following behavior, the rules are in a grounding relation to the practice. But I have also argued that this grounding relation is not complete. Part of the facts that ground social practices are facts about shared understandings in the relevant population about the main values, purposes, or point of engaging in the activity type in question. Practices require some minimal level of agreement in judgments among participants about the things that make the practice valuable for them. In Chapter 6, I will have more to say about the methodological implications that follow from this account. In the next chapter, however, we will turn our attention to the nature of artifacts and ways in which an understanding of the metaphysics of artifacts contributes to an understanding of the metaphysics of our institutional reality.

5 Artifacts and the Limits of Error Artifacts are products of human agency. Many artifacts, particularly of the tangible kind, such as chairs or screwdrivers, do not depend on social rules, and thus facts about artifacts are not necessarily institutional facts. Institutional facts, however, do depend on the existence of artifacts. One of the main purposes of this chapter is to show that there is considerable overlap between the institutional aspects of reality and artifacts. Artifacts are building blocks of our institutional reality. Think about languages, laws, works of art, means of commercial exchange, political and social institutions, corporate entities—​in short, almost any kind of institutional facts—​and it is obvious that those facts depend on products of human agency. Not necessarily tangible products, of course. Many artifacts are intangible, and some are massive in scale, but artifacts they are. Therefore, a better understanding of the metaphysics of artifacts should be conducive to a better understanding of social ontology. My second main purpose of this foray into the nature of artifacts and their relation to institutional facts is to prove an epistemic point—​namely, that there are certain types of comprehensive, group-​wide errors about aspects of artifacts that are not metaphysically possible. A population that creates and uses an artifact has certain epistemic privileges about what the artifact is.1 Since part of what I aim to show here is the necessary dependence of institutional facts on artifacts, the epistemic considerations that follow from the fact that an object is a product of human agency are wide and significant. The chapter proceeds in two main steps. The first two sections are about the metaphysics of artifacts, quite generally, and their relation to institutional aspects of reality. The third section develops the epistemic argument while delineating its limits, and answers a number of objections. Some of the 1 The general idea of epistemic privilege about artifacts is not novel, of course. It has been suggested by, for example, Thomasson, “Artifacts and Human Concepts”; Baker, “The Ontology of Artifacts”; and actually, myself, though mostly in the context of the artifactual grounds of law and fiction: Marmor, Interpretation and Legal Theory, ch. 5, and “Law as Authoritative Fiction.” Part of what I hope to accomplish in the second part of this chapter is to draw distinctions between different types of comprehensive errors that are possible or not, and defend the view against some serious objections.

Foundations of Institutional Reality. Andrei Marmor, Oxford University Press. © Oxford University Press 2023. DOI: 10.1093/​oso/​9780197657348.003.0005

Artifacts and the Limits of Error  85 theoretical implications of the epistemic argument will be further developed in Chapter 6.

5.1. Artifacts Aristotle famously divided things that exist into those that “exist by nature,” that is, irrespective of us, humans, and those we make or produce, which would be “artificial products” or, in short, artifacts.2 Over the centuries since, most scientific and metaphysical interest focused on things existing by nature; artifacts did not seem all that exciting or foundational. But in recent decades this imbalance has started to shift, with a growing interest in the nature of artifacts.3 Understandably so. If you think about it, most of the things we encounter in our everyday lives do not “exist by nature.” Right now, for example, I am working on a laptop computer, positioned on my desk, with a bunch of books and articles scattered around it, writing something in English, sitting on a chair, looking out the window at tall buildings on Broadway Avenue, observing the cars passing by on the street, and listening to music in the background. Every noun item I mentioned in the last sentence is an artifact. The Aristotelian distinction, however, between things that exist by nature and artifacts is overly simplistic. There are things in nature we manipulate in various ways, without thereby making things that we would normally call artifacts. Damming a river and thereby creating a lake where none had existed before creates the lake, in some sense, but the lake is not quite an artifact; it is still a natural lake. Similarly, cherry tomatoes, seedless grapes, and countless other agricultural produce are not artifacts, even if they are created by some human manipulation of natural things. Some people call this intermediary category “naturefacts” or “biological artifacts,” and it makes sense to have a separate category for them, because these things occupy an intermediary position between things that exist by nature and ordinary artifacts, created by us.4 And then we can also come to see that these distinctions are vague, partly depending on how much we manipulate nature, thereby creating new things. 2 Physica, Book II, 192 b 9–​18. 3 See, for example, Dipert, Artifacts, Art Works, and Agency; Hilpinen, “Authors and Artifacts” and “Artifact”; Baker, “The Ontology of Artifacts”; Margolis and Laurence, Creations of the Mind; and Burazin, Himma, and Roversi, Law as an Artifact. 4 Sperber, “Seedless Grapes.”

86  Foundations of Institutional Reality Moving a small stone a few feet away does not create anything; moving it on top of a pile of papers makes it a paperweight. But these categories are somewhat rough and vague, and they are bound to have many borderline cases. Having said this, I will focus on clear cases of artifacts, those things humans build, make, or invent. Countless artifacts are tangible, material objects with physical properties that make them the kind of artifacts they are, like chairs, cars, buildings, and appliances. But there are also countless intangible artifacts that are not, or not primarily, material objects. Languages are artifacts, and so are many things we create with language, such as novels, poems, and narratives.5 For the atheists among us, deities are clearly artifacts, as are myths and legends. Theories are artifacts. In short, an artifact is anything that is a product of human agency, whether tangible or intangible.6 But then, you might wonder, what makes things as diverse as a building or a car, a treatise in philosophy, a symphony, a natural language, and almost everything surrounding us in an urban environment belong to the same metaphysical kind or category? The answer is that all these things, that is, artifacts, are products of human agency. I use the word “product” advisedly, because it captures two aspects of artifacts that are essential. Product is something that has to be produced, that is, made or created somehow by human agency. But it is also something that is identifiable as a product, an entity of a kind, recognizable as such. Not every consequent of human agency is a product. The discordant tune I hum in the shower is the result of something that I do, deliberately, but trust me, a product it is not. Products are the kind of things that have some stability over time, are replicable or reproducible, and publicly recognizable as things of a kind.7 I am not suggesting, of course, that artifacts must be tokens of an already known type; many of the greatest inventions in human history created new types, hitherto nonexistent as a type. The invention of carriages, trains, money, airplanes, computers, novels, cinema, etc., created new types 5 The artifact nature of various aspects of natural languages is controversial, of course. Frege (The Thought) held a Platonist view about sense, and various forms of Platonism about aspects of language have been advocated since. My remarks about artifact aspects of language do not require disputing these views. I do not have to assume here that language is artifact through and through. Even if some aspects of it, like the sense in Frege’s view, is not artifactual, surely some aspects of language are, at least those that are undisputedly conventional, like surface grammar or sound-​sense relations. 6 Even between the tangible and the nontangible artifacts, there are borderline cases. For example, many inventions involve the creation of new modes of production, like Ford’s invention of the assembly line that transformed the entire manufacturing industry. Is the assembly line a tangible or an intangible artifact? Maybe both, somehow. 7 The idea of recognizability is mentioned by several philosophers; see, for example, Dipert, Artifacts, Art Works, and Agency, 109; Baker, “The Ontology of Artifacts.”

Artifacts and the Limits of Error  87 of things. My only point here is that products of human agency must have some durability or sustainability over time, have to be generally reproducible, and their tokens publicly recognizable as tokens of a type. Artifacts are the kind of things that need to be made; they result from human agency, either individual or collective. Many artifacts are made by individual agents: a carpenter can make a chair, or a graduate student can write a paper. But most artifacts in our world (including most chairs, actually) are made by a group of individuals, collectively somehow. It takes a lot of people, in many different roles, to build an airplane. And in many cases, these groups are not just considerable in size, but also with very diffuse membership. Natural languages, cities, commercial means of exchange, and myth and legends are examples of artifacts created by very large and diffuse groups of people, often spanning considerable tracts of time. I will call these massive artifacts. Either way, human agency must be involved, individually or collectively. The idea of agency is important; it implies that the creation of artifacts is typically a purposeful, deliberate action that has been intentionally carried out. But we need to be careful here. It is not uncommon for philosophers to define an artifact as an object that has been intentionally made for some purpose, and then they are drawn to the idea that the purposes of the creator (individual or group) are partly constitutive of what the artifact is. But the truth is that none of this is necessary. Think about a natural language, like English or French. A language is an artifact. It is—​or, at least, significant aspects of it are—​the product of human agency. And, of course, it is true that the agency in question is grounded in intentional actions. But it would be absurd to assume that English was created with someone having the purpose of creating it in the front of their mind, as it were, even if we allow for group agency here. Languages are not created in the way that airplanes or novels are created. The same is true about countless other artifacts, like towns and cities, gods, myths, or urban legends. These are products of human agency, for sure, but there is something misleading about the suggestion that they come about as something that was intentionally made for a purpose.8 Some philosophers resist the idea that something like a natural language or a myth could be an artifact, because such things do not result from

8 Religion is an interesting example where we see all sorts of cases; some religions have a founder, an inventor that gets others to follow him (usually) in a new creed. Other religions evolved much more gradually with mythological founding figures invented later, as part of the religious myth.

88  Foundations of Institutional Reality authorship. These are not the kind of things intentionally created with a particular purpose in the creators’ minds. According to Hilpenin’s influential account, artifacts must have an author; authorship is defined as the intentional creation of something for a purpose.9 Therefore, Hilpenin claims, “An object o is an artifact authored by A only if some properties of o depend on the content of A’s intentions.”10 But these restrictions are somewhat questionable. First, they stipulate away massive artifacts, things that are clearly products of human agency, but not in any way that we can see as a result of authorship. As I just mentioned, natural languages, myth and legends, gods, and even tangible artifacts like a town or a city would not be artifacts on this account. Hilpenin admits as much, and calls them “cultural artifacts” to designate the fact that the condition of authorship does not apply to them.11 There is nothing wrong with labeling a subcategory of X as Y, but this does not entail that Y is not an X type. Cultural or, as I call them here massive artifacts, are still products of human agency, and if they do not meet the condition of authorship, so much the worse for the latter. It just means that the condition of authorship does not apply to all artifacts, though it no doubt applies to many. The problem here is not taxonomical. Forging too tight a connection between intentional creation for a purpose and the nature of artifacts leads to all sorts of distortions. In particular, it creates the misconception that the constitutive properties of an artifact, the properties that determine what kind of artifact something is, depend on the intentions of their creator. But as we’ve just seen, in many cases this is not plausible, and with respect to massive artifacts is not even an option. The constitutive properties of an artifact depend on its prevalent use. If something is generally used for X purposes, if it is used as an X type, then it is an X, regardless of the intention of its creator. Imagine a carpenter with artistic aspirations, creating a chair as a work of art, not intending it to be sat on. Unfortunately for the carpenter, however, the chair gets into market circulation, people buy and sell it as an ordinary chair, and they sit on it without any awareness that it was meant to be a work of art. Unlucky carpenter, perhaps, but if the artifact is used as a chair, a chair it is.12 The phenomenon of artifacts created for one purpose and later used for another is pervasive. Functions change over time, and thereby endow 9 Hilpenin, “Authors and Artifacts” and “Artifact.” 10 Hilpenin, “Artifact,” 159. Baker holds a similar view, though somewhat less rigidly so (“The Ontology of Artifacts,” 7). 11 Hilpenin, “Authors and Artifacts,” 167. 12 Baker, “The Ontology of Artifacts,” 3.

Artifacts and the Limits of Error  89 artifacts with new features, not intended by their creators, sometimes creating new artifacts. Now, one might be tempted to say that the change in the function is itself a new authorship. Well, yes, sometimes it is, as when a boat is turned into a boathouse, or when a church is converted into a mosque or into a cafe. But this is not a necessary or even a typical process. Many times changes in function are very gradual, almost invisible for a while, and not attributable to any particular agents who can be said to have the intention of modifying the use in this or that way. Villages and towns become huge cities without authorship; one thing evolves gradually into something else. One’s god becomes somebody else’s myth; tools become toys; fashion accessories created for one purpose get used for a different purpose over time; and so on.13 In short, appropriations, modifications, and changes in function are sometimes introduced by intentional authorship, and sometimes, quite often, happen gradually and almost inadvertently. You might now suspect that I am relying on a purely functional conception of artifacts, and that in itself may be problematic. Objects don’t necessarily become an artifact in virtue of being used in a certain way. People can use a fallen tree trunk to sit on, like on a bench, but that does not make the tree trunk an artifact. True, but remember that I have defined artifacts as products of human agency. In other words, the function of an artifact is not what makes it an artifact to begin with. Something does not become an artifact solely in virtue of its use or function; it becomes an artifact by being made or created by human beings. The prevalent function of an artifact, its standard or common use by the population who uses it, determines what kind of artifact it is. A pipe cleaner is an artifact because it has been manufactured. If it comes to be used mostly as a toy or craft, and no longer used to clean pipes, then a craft or toy it becomes. Continuous and prevalent use determines what kind of artifact it is. Furthermore, I think that we have to allow for the possibility that an act of converting the use of a certain type of object into a different type may sometimes constitute the making of a new artifact. When people started using pipe cleaners to make crafts, this 13 Here are a few more examples: holy sites tend to be appropriated by other religions by way of associating a different myth with the physical location, as anyone who visits Jerusalem comes to learn very quickly. In fashion things change in strange ways, for example, high heels were used in men’s shoes until the eighteenth to nineteenth centuries. Explosive devices were invented and used in China for show, and their conversion to military use has been very gradual, spanning centuries. In the Middle Ages, countless crafts and ornamental artifacts were transported from the Islamic world into Christian Europe, changing functions, often without any changes to the physical properties of the objects.

90  Foundations of Institutional Reality new use may have created a new artifact, even if the physical properties of pipe cleaners remained the same. Obviously, there is room for a great deal of vagueness here. What counts as invention, creation, making of a new type of object is sometimes disputable, and we cannot expect to have very clear criteria about it. Most of the artifacts we encounter in everyday life are composites of other artifacts. Cars, airplanes, buildings, etc., are made of tens of thousands of parts that are artifacts. Almost every artifact we use is made of lots and lots of artifacts as parts.14 This reminds us that most artifacts are created by a great multiplicity of human agents, often working in concert to create a complex artifact. A great many of these projects, perhaps most when it comes to tangible artifacts, are products of shared agency, analyzable along the lines of Bratman’s model.15 People work in concert to achieve a certain goal, following a plan divided into many sub-​plans, with a great deal of division of labor, sometimes with enormous predetermined complexity. However, as we have seen earlier, not all massive artifacts, created by a large and often diffuse group of people, come about by way of shared agency. Collective actions and collective attitudes can form without plans or anything like an ex ante design. It’s how languages, myths, narratives, cities, borders, and countless other artifacts come about. Group or collective agency is not necessarily “shared agency” (in Bratman’s sense). The properties of a tangible artifact can be grounded in a combination of the object’s physical (or biological, if you want to include “naturefacts”) properties, and its practical function(s), its use(s). Intangible artifacts are a bit more complex. First, because they are almost inevitably mixed with the tangible, some material aspects that make it possible to create and maintain their intangible being. A novel or poem needs to be written or told. Even a myth or a legend needs to have been told by people at some point and maintained in collective imagination by stories written or spoken. A work of art, as intangible content that is conveyed artistically, as it were, needs to have some tangible aspect; art, quite generally, expresses some content by way of manipulating material objects, including, sometimes, human behavior, as in the 14 An easy Google search reveals that an ordinary car is made of about 30,000 manufactured parts. 15 Not very easily, however. As Bratman himself admits, his model might be confined to relatively small nonhierarchical groups of agents who can communicate with each other (Shared Agency, 7). Whether the same or very similar model can be applied to shared agency on very large scale, involving thousands of agents in different locations, is not entirely clear. When numbers are big, a hierarchical-​authoritative structure of agency is almost inevitable. List and Pettit’s account of group agency is not confined to small groups (Group Agency).

Artifacts and the Limits of Error  91 case of performing arts. In short, it is difficult to think of any intangible artifact that would have no material aspect, some physical means of conveying the relevant content. What allows us to talk about these contents as intangible artifacts is the fact that they are contents or products that can be grasped as tokens of a type, talked about, described, appraised, and seen as something separate from (though not necessarily independent of) their means of conveyance or expression. The idea of content is extremely difficult to define. We know that propositional content is the kind of thing that can be true or false. But not all content we convey to each other is propositional, of course. Content can be evocative, associative, affective, etc. I will not attempt to define what content is, hoping to rely on rough-​and-​ready intuitions we have about it. Contents are to be understood here as thoughts, propositions, emotions, ideas, desires, etc., that we can convey to each other by expressive means, and identify them as such, as particular contents expressed. Now, there is a serious worry that if we understand intangible artifacts as any kind of content that is a product of human agency, then we would get an extremely broad notion of artifacts, so broad that it becomes useless. Surely we do not want to end up with the idea that any thought expressed by someone is an artifact. But here again I need to revert to the idea of a product. Artifacts have to be products that can be identified as products of a kind, with some durability or sustainability over time, reproducible or replicable, identifiable as tokens of a type. A poem is an intangible artifact; it is a product of human agency that can be identified as such, as an artistic expression of a certain type. Zeus (the Olympian god) is an artifact, constituted by many stories and legends told about him, identifying him as a product of a kind, a deity with features attributed to him by Greek mythology over the centuries. There are no sharp boundaries here, not even about poems or novels or anything of the sort. There are countless clear, paradigmatic cases of intangible artifacts of various kinds, and then there are many borderline cases, things that have some of the features we talked about but not others. Remember, however, that artifacts, like functions generally, come with a vocabulary of success and failure. Since artifacts are products made by human beings, created or manufactured, there is always the possibility of failure. One can fail to make the product in question, or partially fail, or fail to make a good one. And vice versa, of course. Creations can be ingenious, admirable, useful, etc. And this applies, generally speaking, to intangible artifacts as well. If I try to write a poem, I am likely to fail miserably.

92  Foundations of Institutional Reality But again, we must admit that the idea of a product is vague and bound to have countless borderline cases.

5.2.  Situational Properties As mentioned above, most of the ordinary artifacts we encounter and use in our daily lives are composites, made of many parts that are themselves artifacts. The relation of parts to the whole here is one of construction: a car is constructed from its parts, and in a somewhat more complex sense, we can say that a text is constructed of its words and sentences, or rather, the meaning of those. But apart from construction, there is another sense in which properties of artifacts may depend on their relation to other artifacts, and in some contexts this is quite important. The background here is the familiar distinction between intrinsic and extrinsic properties of objects. Take, for example, a natural object like a stone: we would say that its mass and molecular composition are intrinsic properties; they do not depend on the stone’s relation to anything else. But its relational properties, say, that it is bigger than a pebble nearby, are extrinsic. The fact that the stone looks kind of grayish to us in daylight is also an extrinsic property; it depends on the reflection of light and on our perception. In the case of artifacts, however, some extrinsic properties are such that they depend on the artifact’s position within or its relation to other artifacts. The positioning, spatial or social, tends to make a difference to the kind of artifact it is. In other words, artifacts sometimes have situational properties, affecting ways in which we use them, see them, or think about them. A situational property is a subset of extrinsic properties; it is the kind of property of an object that depends on the object’s position or location within a certain artificial environment, material or social. Artifacts that have this feature, whereby some of their properties are situational, I will label “situational artifacts.” The world of contemporary art provides some of the clearest examples, so let me start there. Take a pile of sand with a mirror stuck in its middle. If you put this up in your backyard, all you get is a mirror stuck in a pile of sand. But if you go to the Dia Beacon gallery in New York, you will find a mirror stuck in a pile of sand that is displayed as a work of art: “Leaning Mirror,” by Robert Smithson.16 I don’t want to get into a dispute here over whether objects of

16 Truth is that the mirror in this work is double-​sided.

Artifacts and the Limits of Error  93 this kind become a work of art only because they are exhibited in a museum, or whether they are exhibited in museums because they are art.17 But even if you hold the latter view, and think that Leaning Mirror is clearly a work of art, you would have to agree that its presentation in a museum is quite significant. If the exact same object happened to be found near a construction site, it might not be a work of art. Its location in an art museum, with a title and the artist’s name and all, makes a difference to the kind of artifact it is. Museums themselves are artifacts. A museum serves various functions, and different functions according to the kind of museum it is, the period and culture in which it operates, and things like that. And again, this is often quite significant. One and the same object can be an exotic ornament if displayed in an anthropological museum, or a work of art in an art museum.18 In other words, objects displayed in a museum are often (not necessarily, perhaps) situational artifacts: their features and associated functions are partly determined by the nature or functions of the artifact in which they are located or otherwise related to. Situational artifacts are not confined to the domain of art. We find them everywhere, both with respect to material objects and, perhaps even more so, in the realm of intangible artifacts. A Vespa motor scooter in the middle of a trendy clothing store is not there to be bought as a means of transportation—​ it’s a fashion statement. Now think about intangible artifacts and you will quickly realize that many of them are situational in this way. A poem has to be structured by following certain rules, and which rules the poet follows determines, in part, what kind of poem it is. Similarly, novels are structured by certain conventions, and those vary with different genres and subgenres. The generic context with its genre-​defining conventions makes a difference. Literary works are inevitably situational artifacts. The law provides another paradigmatic example. Laws or legal directives are products of authoritative pronouncements. But there have to be rules or conventions in the background, determining who counts as a legal authority, the scope of the authority’s power, how it is to be exercised, and things like that. Laws result, by and large, from speech acts performed by people or 17 See Levinson, “Artworks as Artifacts.” 18 Current critiques of Eurocentricism in the art world often lament the fact that Western museums downgraded and marginalized Indigenous forms of art by exhibiting those objects in anthropological museums instead of art museums. I don’t want to imply, necessarily, that everything that looks like art to us is, indeed, art, but the debates about exhibition forms and locations attests to the fact that situational properties of artifacts are familiar and their cultural impact widely recognized, at least by experts.

94  Foundations of Institutional Reality institutions in authority, which means that their legal character and content depends on the rules or conventions that determine what counts as law around here, who gets to make it or modify it, and how. Bear in mind that the exact same normative or prescriptive content can be law in one legal order, but not elsewhere, in a different jurisdiction. In short, legal directives are clearly situational artifacts.19 You might suspect that there is a missing link in the chain of the argument here: Are social rules or conventions artifacts? We have not yet established that. But I think that we have all the tools in our toolkit now to show that social rules and conventions are intangible artifacts. As we saw in Chapter 3, social rules have contents constituted by the collective attitude that is shared by the population whose rule it is. Facts about social rules, we argued, are grounded in facts about patterns of behavior exhibited by individuals in the relevant population, coupled with a set of attitudes and dispositions accompanying that behavior. In short, social rules exist in virtue of individuals’ attitudes, dispositions, and regularities of behavior. Two things are crucially important for our purposes here. First, social rules have contents (prescriptive in nature) fully determined by the collective attitude that is shared, in the appropriate way, by the relevant population. Second, social rules emerge by way of human agency; for a social rule to exist there has to be a pattern of conduct, intentional actions performed by individuals, in light of the collective attitudes they share. And these two points, taken together, warrant the conclusion that social rules are artifacts, their contents intangible products of human agency. I don’t think we have any reason to doubt that rules are products, in the sense we characterized here. Social rules have a certain durability, they persist over time, can be recognized as such, as things of a kind, and are clearly replicable. What we do need to admit, of course, is that most social rules and conventions do not result from any particular authorship; they are not the kind of things deliberately created for a purpose. But I hope I have said enough here to prove that countless artifacts, tangible and intangible, fail the authorship condition. The fact that social rules and conventions are not created by anyone (individual or group) with the purpose of creating something 19 For the purpose of the argument here, there is no need to assume that law is only a product of authoritative directives; even anti-​positivist legal philosophers who would insist that there is more to law than authoritative pronouncements and directives would readily concede that much of the law is the result of deliberate enactments by legislatures and judicial decisions. And that is all we need to assume in order to show that law is a situational artifact, even if, perhaps, not all of it is.

Artifacts and the Limits of Error  95 of its kind—​they tend to emerge organically, often slowly in an almost invisible manner, evolving in all sorts of ways that are often difficult to retrace with great accuracy—​is not something that counts against their nature as artifacts. Many artifacts are created in this way, without authorship. Where does all this lead? It shows that the institutional aspects of reality are partly constituted by artifacts. Which is not surprising, actually—​we tend to think about institutional facts as the kind of facts that are socially constructed. The metaphysics of artifacts, particularly of the massive kind, helps us provide some content and shape to the idea of social construction. Things are socially constructed when they are products of human agency. Seeing that massive artifacts, as products of human agency, can be created by a large and diffuse population, often spanning a considerable amount of time, enables us to see that social construction is a form of artifact creation, as it were. Languages, cities, religions, works of art, countries and borders, and countless other major aspects of our institutional world are socially constructed, of course. Which is also just a way of saying that they are all massive artifacts. And if I am right about the artifact nature of social rules, then the point generalizes to all cases of institutional facts. Our institutional world is the world we have made. In the next section I argue that there are some epistemic conclusions about institutional facts that follow from this.

5.3.  Artifacts and Group-​Wide Errors If the history of science teaches us anything, surely it is the fact that our theories about the true features of natural things, even widely accepted theories, can turn out to be mistaken. But perhaps not so much about artifacts: since artifacts are, essentially, products of human agency, it would seem reasonable to maintain that there are certain types of group-​wide errors we cannot make about their nature.20 Recall (from Chapter 3) that by the idea of an error that is group-​wide (or comprehensive) I mean one that is shared, over some period of time, by the entire community of those whose artifact it is, those who have created the artifact and use it in certain ways. An error would not be group-​wide or comprehensive, in this sense, if it is a mistake of some



20 Thomasson, “Artifacts and Human Concepts.”

96  Foundations of Institutional Reality individuals within that community, even if their mistake is consistent or systematic.21 Consider the spelling of words, for example: many people can misspell a word in their writing, but if a misspelling becomes pervasive, at some point it is not a misspelling anymore. An entire community of language users cannot have group-​wide errors about proper spelling. And the same goes for meaning and surface grammar. Some people, for example, systematically err in their use of a certain word in their natural language; they may systematically misuse the word or misuse it on occasion. That would not be a case of group-​wide error. If the misuse of a word in a natural language becomes pervasive over time, and we reach a point where most people who use the word use it in a way that had been erroneous in the past, then the word acquires a new meaning, at some point is no longer an error.22 In other words, an entire community of language users cannot be systematically mistaken about the meaning of the words they use (or some grammatical structure in their language).23 Meanings are artifacts, constituted by consistent and systematic use.24 But there are also some mistakes about the nature of artifacts, group-​wide mistakes, that are quite possible. In particular, though this is much more common with intangible artifacts, group-​wide errors are possible with respect to the artifact nature of the objects in question. People can comprehensively mistake artifacts for natural kinds. How exactly to specify these epistemic modalities, however, and where to draw the lines, is a tricky business. But given how prominent artifacts are as building blocks of our sociality, it is important to try to be clear about these issues. Let me start with an example, and then show that its lesson generalizes to all cases. Consider the character of Zeus, the chief Olympian god in Greek mythology. Let us assume that the ancient Greeks firmly believed that Zeus exists, as a real entity. We think that they were comprehensively mistaken about this; we would regard Zeus now as a fictional character, something 21 A note on usage: by the term “comprehensive error” or “comprehensively mistaken,” I will mean error that is group-​wide, involving an entire population of users. In other words, group-​wide and comprehensive will be used interchangeably. 22 The word “meat” in English is a good example: some centuries ago it used to mean any kind of foodstuff, not just the flesh of animals, as “meat” is used these days. 23 Of course, that does not apply to the nature of the reference of words. Group-​wide errors about that are possible, as famously shown by Putnam, Mind, Language and Reality, ch. 12. 24 Needless to say, if you are a Platonist about these things, you would disagree; but then we can just focus on other aspects of language. Surely nobody is a Platonist about spelling of words, or some rules of surface grammar. Generally speaking, Platonism about a type of entity would be directly opposed to its nature as an artifact. If you are a Platonist about X you would deny that X is socially constructed. That is not surprising or in need of refutation here.

Artifacts and the Limits of Error  97 invented by the collective imagination of the ancient Greeks. In short, Zeus is an artifact, an intangible product of human agency. So there is clearly one type of group-​wide error that is possible about such artifacts—​namely, errors about the ontology of the object in question. People can think of an object, o, that o is something they have discovered in the nature of things, irrespective of human invention, whereas the truth is that o is an artifact, a product of human creation. If you think about the number of deities thought up by human cultures over millennia, you would see that this is a fairly common mistake. However, precisely because Zeus is an artifact, certain group-​ wide mistakes about his essential attributes are not possible. We cannot come to discover that Zeus was not, after all, the chief Olympian god for the ancient Greeks, or that Zeus was not a god at all, just an ordinary fellow with a big beard. The constitutive features or attributes of Zeus are those, and only those, that emerge from the stories and legends told about him. There is nothing more to it that we can come to discover. Of course we do not need to rule out the possibility that historians or archaeologists would discover that there was an alternative narrative in ancient Greece, depicting Zeus very differently from what we presently know. But what such a discovery would amount to is that there was more than one “Zeus” artifact in Greek mythology. That’s quite possible. There is something very similar, and not unrelated, about possibilities of errors concerning social conventions. Roughly, a social rule is conventional only if there is an alternative rule that the relevant population could have followed instead, achieving the same purposes or solving the same problem. But a community that follows certain conventions can be mistaken about the conventional nature of the norms they follow. They may think that the norms are not conventional—​perhaps because they are not aware of any alternative rules that they could have followed instead, achieving the same purposes—​ where in fact they are conventional. But they cannot be mistaken about the content of their conventions. As we saw in Chapter 3, social rules’ contents are fully grounded in the collective attitudes shared by the population whose rule it is.25 25 The idea that people (that is, an entire community of rule followers) can be mistaken about the conventional nature of rules they follow was first argued by Burge (“On Knowledge and Convention”), who argued that the condition of common knowledge about the conventional nature of a rule that a given population follows, assumed by Lewis (Convention), is wrong. I have explained this, supporting Burge’s position, in greater detail in Marmor, Social Conventions, 5–​6.

98  Foundations of Institutional Reality In other words, the main features of products of our culture and imagination cannot be discovered to have features previously hidden from us, as it were, things that we can come to discover. Of course we can come to discover things about ourselves, our psychology, dispositions, imagination, etc., by learning more about our cultural creations and ways in which we come to believe this or that. But then we are talking about natural objects, the human psyche, not about artifacts. It is also true that group-​wide errors about the ontology of artifacts are usually in one direction: people fail to realize that a product of human creation is what it is, an artifact, produced by our collective attitudes or imagination, while errors in the other direction are much more rare. It is difficult to think of examples where natural kinds are comprehensively mistaken for artifacts. But some cases do come to mind. Realists about domains like mathematics, or ethics, or aesthetics, can be understood to claim that if we think about these domains as humanly constructed, in essence, as artifacts, we are comprehensively mistaken. Either way, the lesson from the story of Zeus is that we need to draw a distinction between two kinds of comprehensive or group-​wide errors about intangible artifacts. Group-​wide errors about the contents of intangible artifacts is not possible; errors about the ontological status of artifacts are possible, and not uncommon. Would something like this apply to tangible artifacts as well? With some modifications, clearly yes. Tangible artifacts have no contents, they are material objects used in this or that way; their nature, as artifacts, is determined by their practical functions, the ways in which they are used by a population. So there is some parallel between the practical functions of tangible artifacts and the contents of intangible ones. Just as we cannot be comprehensively mistaken about the contents of intangible artifacts, we cannot be comprehensively mistaken about the main functions of tangible ones. We cannot come to discover that chairs are not really there to sit on, or that red chairs are not really chairs. Group-​wide errors about the ontological status of tangible artifacts, though rather uncommon, are possible. Many ancient mythologies contain beliefs about all sorts of artifacts as gifts of gods. In the 1970s stories abounded about the alien origins of the Egyptian pyramids—​many people were convinced that such magnificent structures could not have been constructed by humans five thousand years ago.26 So there are possibilities for group-​wide errors about the ontology of 26 Functions of ancient artifacts are often not very well known; we often do not have enough information about practical uses of artifacts in the past. But that, of course, does not show the possibility of

Artifacts and the Limits of Error  99 tangible artifacts. But group-​wide errors about their constitutive functions and features are not possible. Artifacts are what we make of them, grounded in our practical uses and collective attitudes. The idea that group-​wide errors about contents or functions of artifacts is not possible has been resisted by some philosophers, so before we proceed, I would like to answer three objections. One objection to the idea that there is some epistemic privilege about artifacts relies on the idea of division of labor.27 As we noted earlier, most artifacts, tangible and intangible, are created and sustained by large groups of people with considerable division of labor among them. As Putnam showed a long time ago, division of linguistic labor is what enables people to refer to natural kinds even if they have only a partial knowledge of the true nature of the word’s reference.28 I can talk about gold as a distinct element in nature even if I know very little about Au 79; I can rely on others who know better. Similarly, Kornblith claims that we participate in the creation of artifacts by taking part in some large collective endeavor, partly a linguistic endeavor that endows these products with linguistic/​conceptual meanings, and therefore, just as we can be mistaken about the true nature of the reference of “gold,” we can be mistaken about the true nature of the reference of artifact words.29 Thomasson’s response to this argument strikes me as correct: the objection misses the distinction between errors made by individuals about the nature of the artifact they participate in creating, which is always possible, with the idea of group-​wide errors.30 The kind of epistemic privilege we claim to exist with respect to artifacts only prevents the possibility of a whole community of users from being mistaken about the nature or content of the artifacts they make and use. Each and every member within that community can be mistaken about these things, whether the individual is a user or a creator. It is quite possible that a worker on an assembly line would know very little about the final product assembled.31 But if it is an artifact product, then it is just not possible that we can be comprehensively mistaken about the nature of that product. If the product of the assembly line is pickup trucks, and people group-​wide error. Even if we do not quite understand the functions of pyramids, the Egyptians who made them certainly did.

27 Kornblith, “How to Refer to Artifacts.”

28 Putnam, Mind, Language and Reality, ch. 12.

29 Kornblith, “How to Refer to Artifacts,” 145–​146.

30 Thomasson, “Artifacts and Human Concepts,” 65. 31 This is Kornblith’s example, actually.

100  Foundations of Institutional Reality use them as such, the ignorance or potential error of any individual worker is neither here nor there. As I mentioned earlier, people can systematically misuse a word or an expression in their natural language, but that does not make it a group-​wide error. Here is a second possible objection to consider: think about an individual creator, say, a novelist, writing about a character she invented. So we have a fictional character, say P, which is clearly an artifact, created, let us assume, by A, the novelist. Admittedly, some philosophers would want to allow for the possibility that A may come to have some mistaken beliefs about the character of P. A may even come to recognize her mistake. She may come to think that she had thought of P as the smartest fellow in her story, but it turns out that P was not that smart after all—​Q (another character in the story) was actually smarter.32 So it seems that we must allow for the possibility that even the creator of an artifact can be mistaken about certain aspects of the artifact she created. And if this holds about an individual creator of an artifact, there is no reason to believe that group-​wide errors can be ruled out. Just as a novelist can be mistaken about the fictional character she created, the Greeks could have been mistaken about the attributes of Zeus. Maybe they thought that Zeus was the mightiest God in their pantheon, but in fact Poseidon was mightier. The crucial question here is, of course, what is this “in fact” so and so? A thinks that the character P that she created is X, but in fact P is Y? Fact about what? The true nature of P? There is a whole minefield here we should tread very carefully. Let’s take a concrete example of a fictional character, say, Agatha Christie’s Hercule Poirot. There are a lot of things we learn about Poirot in Christie’s mysteries. For example, we are told that he is Belgian, a private detective working in London, and very smart. Can it turn out somehow that Poirot is not Belgian but in fact, French? Or that he is not a detective but a fishmonger? Not really, that is, not unless there is some indication in the text that allows for such twists in the narrative.33 It does seem possible, however, that upon reflection, reading everything Christie wrote about Poirot, he turns out to be not that smart after all. (And perhaps even Christie 32 I am indebted to David Enoch for raising this issue for me. For the idea that fictional characters are artifacts see, for example, Thomasson, Fiction and Metaphysics, 21–​23. For my own stab at this issue, see Marmor, “Law as Authoritative Fiction.” 33 There are some complicated issues lurking in the background here about truth in fiction, though they are not directly relevant to the point I make in the text. I have elaborated on these issues in Marmor, “Law as Authoritative Fiction”; building on Lewis, “Truth in Fiction”; and Thomasson, Fiction and Metaphysics.

Artifacts and the Limits of Error  101 herself could have come to wonder about that at some point, not being so sure herself.) But notice that the expression “turns out to be not that smart after all,” has a critical evaluative dimension. People who create an artifact can fail to make the artifact they had intended to create. It is surely possible that Christie wanted to create a character with feature X but has not quite succeeded in that; perhaps the resultant character is more Y than X. In creating artifacts of all kinds, success in the creation is not guaranteed. Success and failure are evaluative terms—​they can pertain to the relation between the intention of the creator and the final product, or they can relate to the evaluation of the product relative to some critical standards. As I mentioned earlier, artifacts come with a rich vocabulary of success and failure. A product of human agency is not always the product intended, and it is certainly not always a good product. And, then, when we apply critical standards to the evaluation of an object, disagreements and errors, even group-​wide errors, cannot be ruled out. The conclusion here is that the epistemic privilege a population has with respect to the nature of the artifact created or sustained by the population in question does not entail immunity from critical or normative errors. Far from it. When I suggested that there is nothing more to discover about the attributes of Zeus than what is told about him in Greek mythology, I did not mean to include any critical perspectives on the character of Zeus—​whatever the Greeks may have thought about Zeus in this respect is, of course, subject to critical or normative challenge. It is certainly possible that Zeus, as the fictional character that he is, is actually evil, morally speaking, even if the Greeks would not have agreed. Epistemic privilege is not immunity from critical evaluative mistakes. The third objection I want to consider is in the same vicinity but it draws on the possibility of reasonable disagreements about the classification of artifacts. We relied on the idea that artifacts have to be products of human agency, and mentioned the fact that products are not just any results of human agency—​they need to have some durability or sustainability over time, be replicable or reproducible, in short, be recognizable as things of a certain kind. It was also assumed that what counts as or makes things a kind of product is grounded in actual patterns of use, in the functions and contents of the artifacts in question. But now it can be argued that the idea of a product of a kind in this context is often very unstable, open to competing interpretations and all sorts of misconceptions. And that is particularly so if we detach, as I argued that we should, the constitutive features of artifacts

102  Foundations of Institutional Reality from their original authorship. Thus, the objection goes, if we can have reasonable disagreements about what kind of artifact an object is, then there is no epistemic privilege here. Just as one interpretation can get it wrong, in principle, all of them could get it wrong. Or perhaps all of them, inconsistent as they potentially are, can be equally right, which would amount to the same thing: no epistemic privilege. Let us take an example of a work of art, which would be particularly conducive to the objection under consideration here. Recall Leaning Mirror, exhibited as a work of art in the Dia Beacon gallery. We can certainly have a reasonable disagreement about the question of whether this object is a work of art or not. Surely, the fact that the artist, Robert Smithson, made it with the intention of creating a work of art is not enough to make it so; people can fail to make the kind of product they wish to make, whether in art or in any other context. I take it as not disputable that there is room for sensible disagreement here. It is not obvious or self-​evident that Leaning Mirror is a work of art.34 Would any of this prove that we have no epistemic privilege about what works of art are? Can we come to discover that art is not what we had thought it is? In some rather loose sense, yes, perhaps, but this not the sense in which the objection under consideration can be made to work. Art is one of those cultural types that changes all the time. Conventions structuring artistic genres are constantly under pressure, open to be challenged by artists striving to create new conventions and new kinds of work of art. To take a familiar example, for centuries in European art, at least, people thought that paintings and sculptures needed to depict figurative images. Abstract art, introduced in the early decades of the twentieth century, changed this convention, quite radically. So perhaps there is a sense in which we can say that art was not what we had thought that it was; we came to learn and appreciate that it doesn’t have to be figurative.35 In some loose sense this is true. There is a sense here in which we could say that the invention of abstract art allowed us to see art in some new light, perhaps hitherto unappreciated. But there is something potentially very misleading here as well. Abstract art was not a discovery about the true nature of art, as it were, but an invention of a new kind of art, 34 As we will see in Chapter 6, these kinds of disagreements are bound to depend, in part, on evaluative conceptions about what makes works or art valuable and worthy of our appreciation. Different conceptions about the values we associate with art are bound to result in different views about this particular object, whether it is art or not. 35 This statement is overly Eurocentric; nonfigurative art was very widespread in the Islamic world for centuries.

Artifacts and the Limits of Error  103 hitherto nonexistent. And the same goes for the inception of readymades art, conceptual art, more recently video art, and so on. Art is a very loose and vague type of product, with new kinds and genres emerging all the time.36 Consider a much more mundane example. Some decades ago we had telephones, then came mobile phones, and then smartphones. All of these devices share some features—​they enable people to converse from a physical distance. But they are different kind of devices. The mobile phone, and then the smartphone, were inventions of new kinds of artifacts. It would be very strange to say that by the invention of smartphones we came to discover something new about the true nature of telephones.37 Not at all; people invented a new kind of artifact, obviously related in their functions to artifacts we had had before, but this does not mean that the invention is a form of discovery, telling us something we had not known about the nature of telephones, something that was there all along but we failed to notice. The same is true about new genres of art. The invention of abstract art was an invention, and a very creative one at that, but not a discovery. New kinds of artifacts, tangible or intangible, are still inventions, even if they are closely related to earlier types. Such innovations certainly do not come to show that we had been comprehensively wrong about previous types, only that new valuable or functional types can be brought into existence. Let us return to the potential disagreement about Leaning Mirror. Art is one of the most elusive kinds of artifact human cultures have created. Little wonder that many borderline cases of art are to be found. But the crucial point to bear in mind here is that the idea of epistemic privilege is not meant to rule out the possibility of genuine vagueness. The essential features that make an artifact what it is are just as vague as any other concepts. As I mentioned several times earlier, borderline cases of mundane artifacts, like chairs or houses, are abundant. That certainly would not show that we can be comprehensively mistaken about the nature of chairs or houses, that we can somehow come to discover something about their constitutive features we had not known before.38 So the real issue here is to explain why disagreements about borderline 36 A number of art critics and philosophers have argued that art itself, and not just new genres of it, may have been reinvented or, at least, radically reshaped, in the middle of the twentieth century. See, for example, Levinson, “Artworks as Artifacts,” Danto, After the End of Art. 37 Of course we did come to discover something about the boundaries and possibilities of technology, and all sorts of scientific aspects of telecommunication. But that is beside the point; scientific discoveries are not in any doubt here. 38 This is not really challenged by so-​ called epistemic theories of vagueness (Williamson, Vagueness). Even if, metaphysically speaking, there is always a truth about borderline cases, those truths are not knowable.

104  Foundations of Institutional Reality cases are often rather silly or pointless, while in others, such disagreements can be serious and interesting. It may not be very interesting to argue about whether a broken chair is a chair, but it is interesting and potentially fruitful to argue about whether Leaning Mirror is a work of art. There is an explanatory challenge here, and I will turn to it in the next chapter. To conclude this chapter, what I have tried to demonstrate here is that we cannot understand the socially constructed aspects of reality without a meticulous metaphysical account of the nature of artifacts. Since social rules and conventions are artifacts, institutional facts are partly constituted by artifacts. And then many familiar types of institutional aspects of reality are themselves massive artifacts, tangible and intangible, like languages, legal systems, art, money, cities, borders, corporations, and so on. None of these massive artifacts got created by purposeful design—​they came about gradually and organically, so to speak, by way of collective behavior and collective attitudes developing over time. But artifacts they are, and as such, we cannot be comprehensively mistaken about their constitutive features, though we can have group-​wide errors about their ontology and about their values relative to various critical perspectives.

6 Rationalizing Practices The content of social rules, as we saw in Chapter 3, is fully grounded in the collective attitudes of the population whose rule it is, which means that a population that follows a social rule has a certain epistemic privilege with respect to the content of the rule they follow: comprehensive, group-​wide errors about it are not possible. Similarly, we saw in Chapter 5 that the artifactual aspects of institutional reality render certain types of comprehensive, group-​wide errors about contents and main functions of the things we construct impossible. Finally, as we saw in Chapter 4, for a social practice or institution to exist, there has to be a minimal level of agreement in judgments among its participants about the reasons for having the practice and the values it instantiates for them. These three independent conclusions sit comfortably together. The concern I want to address in this chapter, however, is that the direction of these conclusions might lead us to another idea, but one that we have good reasons to resist. The thought might be that any philosophical account of a social practice would need to offer the kind of explanation that would rationalize the practice for its participants, make it intelligible to them why they engage in it, and what the point of it is, for them. I call this the internal rationalization constraint and will explain it shortly. My main argument in this chapter is that a reductive metaphysical explanation of a social practice is not subject to the internal rationalization constraint, and that this is one of the main differences between a metaphysical theory of social practices and a hermeneutical one. I will use some of the key debates about the nature of law as my main examples in this chapter. I will try to show that Hart’s conception of what to expect from a theory about the nature of law is metaphysical and reductive. The interpretative critique leveled against it, predominantly by Dworkin, presupposes a very different methodology, and one that is clearly guided by the internal rationalization constraint. My point is going to be that metaphysical theories about the nature of law, or any other social practice for that matter, do not need to rise to the challenge. A grounding-​reductive account of social practices does not need to meet the internal rationalization Foundations of Institutional Reality. Andrei Marmor, Oxford University Press. © Oxford University Press 2023. DOI: 10.1093/​oso/​9780197657348.003.0006

106  Foundations of Institutional Reality constraint. Hermeneutic theories of practices do take the rationalization of the practice for its participants as an important objective, but then they must also rely on a different, essentially normative methodological framework. In the first part of the chapter, I will try to explain the distinction between internal and external rationalizations of social practices. In the second part of the chapter, we will see how the distinction, and some of the arguments deployed in previous chapters, play out in one of the most prominent debates in philosophy of law. I focus on legal philosophy here for two main reasons. The first is a matter of theoretical convenience—​we have a long history of philosophical debates about the ontology of law, nicely exemplifying the kind of concerns we are interested in here. The second reason has to do with the nature of law itself. The law is one of the most structured and complex practices human cultures ever created. If we can shed light on some of the traditional philosophical debates about the ontology of law, we can safely assume that other social practices would not require very different tools.

6.1.  The Internal Rationalization Constraint Social practices are structured by rules. Rules have constitutive functions in making the practice what it is, functions that are grounded in purposeful use, which itself is grounded in patterns of conduct and collective, learnable attitudes and dispositions. As we noted in Chapter 4, social practices require, if for no other than practical reasons, a certain level of agreement among its participants about its main functions and values, at least for those who willingly engage in the practice.1 One pervasive feature of social practices (as well as of most massive artifacts) is their multiple instantiations. Practices and massive artifacts can differ from each other as different tokens or subtypes of the same general type. There are quite a few different schools of yoga, but despite significant differences, yogis will tell you that they are just different practices of yoga. Cornell University is quite different from Oxford University, in its teaching practices, governance structure, admission procedures, and countless other aspects, yet they are both research 1 We should not lose sight of the fact that there are many social institutions that are imposed by force and maintained by violence or at least violent threats. In most cases, I think, this only means that the relevant population whose practice we need to focus on is the population that imposes the institution by force and manages to maintain it as such. The relevant participants in our institutions of incarceration, for instance, are not the inmates, but those who put them there and manage to keep them incarcerated.

Rationalizing Practices  107 universities. Practices and massive artifacts allow for different versions of the same general type and they are fully grasped as such by their respective participants. This is a pervasive aspect of social practices—​they often exist in different, sometimes significantly different, versions or instantiations. But they are seen by the relevant agents as different instances of the same general practice, the same kind of social institution. It would be impossible to explain this variability if we tried to account for social practices and institutions without taking into account their overall purposes, social functions, and, generally, some explanation of what the practice is there for. And this, of course, is closely tied to the fact (mentioned in Chapter 4) that some of the rules or subparts structuring a social practice are more essential or central to it than others. What makes certain rules or parts of a practice more central to it than others must have something to do with the general purposes or aims of the practice—​its main point, as it were. When you think about a university, for example, you would think of the pursuit of knowledge, the value of research and teaching, the freedom of thought and innovation, and things like that, as essential aspects of what a university is for and what the point of being part of it is for you. Giving us a list of the rules and conventions that constitute an institution like a university, with their corresponding functions, is just not going to give us the full story of what universities are, what motivates people to take part in the practice or, as often the case, seek to change aspects of the practice to better serve its overall purposes. But precisely because we understand these things, we can easily see how different versions of a practice or institution are just different versions of essentially the same thing. In other words, the idea here is that combining the functions of rules structuring a social practice is not going to generate the overall functions and purposes of the institution or practice as a whole. We need to think about features of social practices that rationalize their existence, aspects of the practice that would explain what the practice is all about, and how it is tied to things people value. Thus, social practices and institutions must have rationalizing features, which would be aspects of the practice that explain why it exists and why it is valued by those who practice it and maintain its existence as the kind of practice it is. But here we need to draw a crucially important distinction: rationalizing features of a social practice can be external or internal. A rationalization is internal if it is the kind of explanation that the participants in the practice, those whose practice it is, can see as their own, so to speak, explaining the point of the practice, the reasons for engaging in it, for them. Thus, an

108  Foundations of Institutional Reality aspect or feature, X, of a practice, P, would be internally rationalizing for participants of P if they can come to see X as a reason to engage in P, or something that ties the practice to things they value or care about. A rationalizing feature of a practice would be external, therefore, if it explains why people have the practice the way they do, or why they engage in it, without necessarily offering an explanation that participants can see as a reason to engage in the practice or things they value about it. Thus, Y would be an externally rationalizing feature of a social practice P if Y contributes to an explanation of P as the kind of practice it is, even if Y is not something that participants in P would be willing to see as a reason to engage in P. Now this, of course, does not mean that externally rationalizing features of a practice cannot be internal as well; it only means that they do not have to be. Let me emphasize, however, that the distinction between internal and external rationalizing features is not the distinction between motivating and normative reasons. First, the internally rationalizing features of a practice need not be on the surface—​they include factors or features of a practice that participants may not be aware of, but would be willing to recognize as their own once presented to them. More importantly, external rationalizing features do not have to be reasons in the normative sense. Many social practices and institutions are such that there are no reasons—​in the normative sense—​to have them, or have them in the ways they are actually practiced. That does not mean that they have no rationalizing features. There are no reasons to have racist or sexist practices and institutions, but their racism, sexism, or similar features are such that they can contribute to an explanation of why people have the practice and what functions and values it serves for them.2 And, crucially, such features can be rationalizing even if the participants in the practice would not tend to recognize those features as normatively significant for them. Let me give an example or two. Consider the practice of the so-​called traditional family. It is a social practice, partly but significantly structured 2 The distinction between internal and external rationalizing features has nothing to do with Hart’s famous distinction between internal and external points of view about social rules (The Concept of Law, 86–​88). The internal point of view is already incorporated in the Conditions-​H of what social rules are, as explained in Chapters 1 and 3. As I have also explained in some detail (Marmor, Philosophy of Law, 53–​55), Hart’s distinction between the two (or rather, three) points of view is meant to reinforce his reductionist explanation of law, not to call it into doubt. Hart’s main point here is that we can account for the internal perspective reductively, by accounting for people’s shared attitudes, without having to endorse those attitudes or presuppose their soundness or anything else about them. The distinction is a critique of Kelsen, not mainly Austin, as most commentators tend to think.

Rationalizing Practices  109 by social and legal rules—​ with some considerable local and temporal variations—​collective attitudes, and familiar, widely shared patterns of behavior. Clearly it is the kind of social institution that has a great deal of meaning and personal significance for those who practice it. For many people it is the most significant institution in their lives. There are, no doubt, rationalizing features of family practices that are internal—​they would be the kind of features that participants would recognize as their own, something that rationalizes for them why they value the institution and their reasons for maintaining it. Needless to say, there is plenty of room for disagreement here. Some people value the idea of a family partly for religious reasons, while others care nothing about those. But the fact is that we can easily discern the internally rationalizing features of the institution of the family despite possible disagreements about them and different interpretations of them. There are also some rationalizing features of the institution of family that are clearly external, such as a causal-​evolutionary explanation. Perhaps these practices got reproduced over millennia because they turned out to be biologically successful (in whatever evolutionary terms we want to explicate this). But there are some possibilities in the middle, neither internally rationalizing nor merely causal. One example I have in mind, quite obviously, is the feminist critique of the family as an institution that serves to entrench patriarchal dominance of women and children by men.3 Allow me to assume here that there is a great deal of truth in this feminist critique. And yet it highlights a rationalizing feature of the institution that most participants would tend to find alien; it is not something that rationalizes the practice for them. Most people, men and women, wouldn’t think that male domination is a reason for their adherence to traditional family structures, something that counts in favor of getting married and raising their biological children together. In this respect, it is very similar to the Marxist account of religion as ideology, as a form of false consciousness. It is just not the kind of perspective on religion, an explanation of its rationale, as it were, that any religious person can coherently endorse. Perhaps the feminist perspective on the family is not quite as alienating as a Marxist account of religion. If you are sincerely religious, you just cannot be a Marxist about it, but you can appreciate your family even if you recognize the truths and insights of the feminist critique. It’s not easy, and 3 The literature on the essential patriarchal nature of the family is vast; see, for example, MacKinnon, Feminism Unmodified, ch. 3.

110  Foundations of Institutional Reality tensions are bound to come to the surface here and there, but it is (I hope) possible. And of course, it is also true that the more publicly recognized and appreciated the feminist critiques of the traditional family become, the more social pressure builds up for change, and we have witnessed significant changes unfolding in the last few decades. External critiques can become internalized over time and push for social and legal changes. But what are these external rationalizing features—​ metaphysically speaking, that is? When we say that the traditional family structure serves to entrench male domination and patriarchal rule, we seem to imply that there is some functional explanation in play here. The question is, however, what kind of functional explanation? As we saw in Chapter 4, not every functional explanation necessarily attributes practical function, function as use, to the explanandum. Functional explanations are often about causal chains. Therefore, one option here as well is that such factors are simply elements of a causal explanation. To say that traditional family structures serve to entrench male domination can be just a way of saying that it is what the institution does; it’s what the effects of it are, socially speaking. Perhaps the causal chain starts earlier. Perhaps the desire and ability of men to dominate women and children is what brought about the family practice, pushed for it, as it were, in a complex causal chain over centuries and millennia. Surely there is some truth in this too. Social practices and institutions come into being by a complex causal chain and they have a myriad of causal effects, whether intended or not. But the problem with this causal account is that it leaves out some of the explanatory power of what a rationalizing feature is. There are countless kinds of items or elements that figure in a casual explanation of a social practice, and many causal effects of it, but only some of them would seem to bear on the practice’s intelligibility, its purpose, its point, and people’s (motivating) reasons for engaging in it. A second and more promising option is to think about externally rationalizing features in terms of practical functions, function as purposeful use, even if some of the purposes are not on the surface, as it were. In other words, invoking the idea of hidden functions, mentioned in Chapter 4, seems particularly attractive here. It is very tempting to suggest that something like entrenchment of male domination is a hidden practical function of the traditional family institution. In a way, we can say that it is what the institution is used for, even if this use is not explicitly endorsed by the participants or recognized by them. This seems right to me, but there are some difficulties to overcome. One difficulty stems from the fact that when a hidden function

Rationalizing Practices  111 is exposed, as it were, when you present it to the relevant participants, it is still possible that participants might find it alienating—​they would not necessarily exhibit a kind of “aha” moment, “so yes, now we see what reasons we have for doing this!” If we want to retain the idea of function as purposeful practical use, the relevant question here is whether people can actually use an X for F while sincerely rejecting the idea that X serves F for them. If the answer to this question is yes, then there should be no objection to seeing externally rationalizing features of a social practice in terms of practical function, even if, as sometimes is the case, those functions are “hidden.” So let us explore the possibility. Before we proceed, however, let me emphasize that I am not claiming here that all externally rationalizing features consist of hidden functions. Far from it. Externally rationalizing features would include all those putative values, purposes, and functions that would contribute to an explanation of what a practice or institution is for, what explains its existence, and ways in which it is practiced. What I am trying to show here is that externally rationalizing features can accommodate hidden functions as well, or any other rationalizing feature that would fail the internal rationalization criterion. So now the question is this: Is it possible for A to use X for F, even if A would sincerely reject the idea that she is using X for F? As I mentioned in Chapter 4, the idea of a purposeful action is not confined to those cases in which the aim or purpose of the action is at the forefront of the agent’s mind, which comes to show that there is a sense in which the purpose of one’s conduct is occasionally not fully avowable by an agent. Those who research and write about racism would surely agree. Many people act in ways that reflect their racist biases and prejudices while sincerely rejecting any suggestion of a racist element in their motives and purposes. Racism (and sexism, and the like) would have been much less invidious and prevalent had it been the case that people cannot act on racist prejudices unless they are fully aware of them.4 So it seems that the idea of a hidden function is not so suspect after all, even if we retain the idea of practical function as purposeful use. In other words, racism, sexism, xenophobia, and similar aspects of social practices can be rationalizing features even if they clearly fail the internal rationalizing constraint. They are features we can employ in an explanation of why

4 The literature on so-​called implicit biases is considerable and somewhat controversial. See, for example, Banaji and Greenwald, Blindspot; Payne, Niemi, and Doris, “How to Think about Implicit Bias.”

112  Foundations of Institutional Reality a practice exists, what the point of it is, why people engage in it in the ways they do, how they adapt the practice to changing circumstances, and all sorts of explanatory tasks of this nature, without assuming that the participants themselves would avow those features as rationalizing the practice for them. Now one might worry that the possibility of including hidden functions in the rationalizing features of a social practice is in tension with the kind of epistemic privileges I argued for in previous chapters. If it is not possible for an entire community of participants in a social practice to misconceive the contents of the rules they follow, or to be mistaken about the main purposes and functions of the artifacts they create, how can those things have practical functions that are not fully avowable by the participants? In response, two points to bear in mind. First, even if there is a product of human agency that has functions that are not fully avowable by the users, nothing can have a function that is entirely hidden. Hidden functions are parasitic on functions as purposeful use in the ordinary, self-​avowable sense. We do not have artifacts as products of human agency without considerable epistemic transparency about the practical use of the product, its designated or widely used function. (Remember that practical use is something that we normally learn from others.) Hidden functions, if and when they are present, are always parasitic on ordinary, epistemically transparent functions. First, we need to have an object generally used for something, and then we can also come to realize that it serves something else as well, perhaps not widely recognized as such. In other words, the idea of hidden functions does not undermine the idea of epistemic privilege, it only limits its scope. Second, this limit in scope is itself limited to a certain type of possible error. It is not really the case that by discovering the hidden function of an artifact or institutional practice we come to see that the population in question is comprehensively mistaken about what it is that they collectively created. They may be mistaken, perhaps gravely so, about the real purpose of their creation, but not about what it is that they created. In other words, the kind of epistemic privilege we argued for is privilege about content, or constitutive features, not about purposes. When a Marxist claims that religion is ideology, the Marxist need not deny that what people take their religion to be is what it is. The Marxist does not claim that Christianity, for example, is not what Christians believe it to be in terms of its contents, practices, proscriptions, dogmas, etc. She only denies that a complete account of the real purposes or functions of religion, what it is actually used for, is what religious people assume. Similarly, by claiming that traditional family structures serve purposes

Rationalizing Practices  113 of male domination the feminist critique does not claim that people can be comprehensively mistaken about what a family is. We can acknowledge that people sometimes act for reasons they are not fully aware of, or even with purposes they would not be willing to avow. This is compatible with the view articulated in previous chapters that there are certain types of errors about the contents and functions of socially constructed reality that are not possible on a group-​wide scale. The question that remains open, however, is the one we started with: Would we have good reasons to think that a metaphysical account of a social practice must meet the internally rationalizing constraint? In particular, would it be a valid argument against a grounding-​reductive explanation of a social practice that it fails the internal rationalization constraint? In the next section I will explore this issue using the example of some of the main controversies about the nature of law in legal philosophy.

6.2.  Legal Positivism, Reduction, and Hermeneutics There are at least three main debates about the nature of law in analytical jurisprudence, aiming to answer different questions. One is a debate about the ontology of law; the other is about determinates of legal content; and the third is about law’s normative or moral, reason-​giving aspects. Naturally, our focus here is going to be on the ontological debate, which is not to deny that there are all sorts of connections between these three debates and in certain ways they may be intertwined.5 Legal positivism, however, is a theory, if theory is not too big a word here, focused on the ontological aspect of law, aiming to answer the question of what makes things legal, part of the legal domain, as it were, when they are. It is a question of the general form What makes an o an F? where o stands for some normative or prescriptive content, and F stands for legality, in some relevant sense. This is a fairly narrow question, and an answer to it, even if fully successful, would hardly amount to a theory about the nature of law. Nevertheless, this is the question we will focus on here.

5 I explored some of the ways in which these debates are intertwined in Marmor, “Law as Authoritative Fiction.” For my take on the normativity of law, where I also explain why different views on the normativity of law are not necessarily aligned with the traditional positivism versus anti-​positivism positions; see Marmor, “Norms, Reasons and the Law.”

114  Foundations of Institutional Reality So what makes it the case that a norm or requirement or such is part of the law (somewhere at some point in time)? As mentioned in Chapter 2, the background against which this ontological question arises is very familiar. We often say things like, “It is the law (here and now) that . . . such and such”; or “Such and such is legally required”; or “It had been a legal requirement to φ until a year ago but it no longer is.” We normally talk about laws as things that exist, in some sense. Propositions about what the law is in a given jurisdiction can be true or false. To be sure, we do not need to assume complete determinacy here—​sometimes there is genuine indeterminacy about whether something is or is not part of the law. We probably owe this focus on the ontological question about the nature of law to the influential work of John Austin.6 There are two main ambitions Austin envisaged about the nature of his project. First, he sought to provide a reductive explanation as the answer to the ontological question. Austin argued that the answer to the question of what makes norms legal, when they are, can be fully articulated in terms of social facts about political sovereignty, people’s conduct and attitudes, and particular actions and events that take place in the world.7 Second, Austin expected the answer to be such that it is capable of explaining how law differs from similar normative domains, such as morality, religion, and social customs. Thus, Austin aimed to meet two criteria for the adequacy of an answer to the ontological question: reducibility and uniqueness. An account of legality has to be reductive, and it has to show how law is different from other, similar normative domains. In the next chapter I will argue that Austin’s second desideratum, uniqueness, does not need to be met, at least not as strictly as Austin assumed. In this chapter I will focus on the reducibility ambition. Thus, following Austin, let us assume that the question we ask is: What makes an object, o, an F? What is o, the object, in the legal case? We tend to talk about norms—​norms are the kind of things, it is generally assumed, that can be legally valid or not. And that is largely true, but not entirely accurate. Particular instructions, such as “You [a person] do this . . . [a particular act or act type],” can be legal instructions. Laws do not have to be norms, though they usually are. And then perhaps somewhat looser prescriptive contents, 6 Austin, The Province of Jurisprudence Determined. 7 I use the expression “social facts” advisedly; one of Hart’s convincing critiques of Austin (though not in these exact terms, of course) is precisely the idea that Austin failed to see the distinction between social facts and institutional facts. Austin failed to see that facts about sovereignty are institutional, depending on rules. I explain this in some detail in Marmor, Philosophy of Law, 44–​59.

Rationalizing Practices  115 such as general doctrines, even ideologies, may also form part of the law, even if they are difficult to formulate in canonical normative terms.8 Bearing these reservations in mind, at least for the sake of convenience, we can assume that the question is this: What makes norms legal norms? Now the F in question here is a bit more difficult to formulate precisely. For one thing, F is always relative to a particular jurisdiction at a particular time. Norms are legally valid in some legal order or other, at a given time, not in the abstract. Thus the question of what makes an o an F, fully articulated, would need to be spelled out in a way that is relative to particular jurisdictions and time.9 But, of course, the philosophical question is not about the specific determinates of legal validity in particular jurisdictions. What we want to know is this: What makes it the case, generally speaking, that any given norm would form part of the law where and when it does? Which is to say that we can formulate the question conditionally: If a given norm is part of the law in Si at t, what has to be true to ground this fact? What are the grounding facts of legality? I will not dwell on the details of Austin’s answer. What matters is the philosophical ambition, which is to provide the answer reductively, in terms of people’s actual modes of conduct, the attitudes they share, and particular actions and events that take place. In other words, Austin clearly attempted to reduce legality to patterns of conduct and attitudes exhibited by individuals.10 As I have explained elsewhere in some detail, Hart shared the same theoretical ambition.11 His dispute with Austin was not about the reductive nature of the explanation sought, but about the particular building blocks that Austin had offered. Hart argued that Austin’s building blocks are too crude and inadequate to the task. Nevertheless, Hart clearly shared Austin’s theoretical ambition to explain the elements that ground legality by way of reducing them to more foundational facts about people’s actual modes of conduct, their shared attitudes and dispositions, and various actions and events that take place in the world. As we saw in Chapter 1, the hallmark of Hart’s theory of law is the idea that social rules are at the foundation of law, and that social rules, in turn, can be explained by 8 For example, I think that it would be difficult to deny that a commitment to capitalism forms part of US law. 9 International law included; I am not doubting its legal status here. 10 Austin also sought to provide a reductive explanation of law’s normative aspects (The Province of Jurisprudence Determined). In that respect, his project is widely regarded as a spectacular failure. But the truth is that there are some reductive ambitions in Hart’s account of law’s normativity as well, though clearly the latter is much more sophisticated. This is not an issue I will take up here at all. 11 Marmor, Philosophy of Law, ch. 2.

116  Foundations of Institutional Reality way of a grounding reduction. With the exception of Kelsen (and perhaps to some extent Raz), this project of trying to articulate a grounding-​reductive explanation of legality is characteristic of the legal positivist tradition.12 Hart’s main critique of Kelsen’s theory of law is actually very instructive here because Kelsen explicitly rejected any attempt to provide a reductive account of legality. Kelsen thought that legality cannot be reduced to other types of facts; it must be theoretically presupposed.13 Hart and Kelsen shared the view that there has to be some normative framework structuring ways in which law is made in any given legal order, and they both thought that this normative framework is hierarchical, eventually deriving from one fundamental norm in every legal system from which others are derived in various ways. Kelsen called it the Basic Norm, Hart called it the Rule of Recognition. But for Kelsen, the Basic Norm was a postulate, a presupposition we need to make in order to make sense of the idea of legal validity. And this is precisely the main point that formed the target of Hart’s critique. In direct response to Kelsen, he argued that there is no need for presuppositions about the Basic Norm; the norms in play are social rules, reducible, as we have seen, to patterns of conduct and shared attitudes of the population in question. As Hart put it, facts about the rules of recognition “can be established by reference to actual practice: to the way in which courts identify what is to count as law, and to the general acceptance of or acquiescence in these identifications.”14 In other words, Hart’s theory of legality, his answer to the ontological question about the nature of law, is explicitly reductive. What Hart sought to achieve is a full grounding reduction of legal validity by way of a two-​ stage grounding: legality is grounded in social rules and events in the world that gain their particular legal meanings by those rules, and social rules are grounded in patterns of conduct and their accompanying attitudes, as we saw in detail in Chapter 3. As I mentioned in Chapter 2, reductive projects in philosophy are inherently ambitious because they fail if the reduction is not complete. Partial or incomplete grounding does not satisfy the conditions of a reductive 12 Kelsen’s theory of law spans many books and articles; most of the main ideas are articulated in General Theory of Law and State and Pure Theory of Law, 1st ed., translated to English as Introduction to the Problems of Legal Theory. See also Marmor Philosophy of Law, ch. 1, where I explain Kelsen’s anti-​reductionism and why it fails as such. 13 Hence he called his theory a pure theory of law (Kelsen, Introduction to the Problems of Legal Theory). 14 Hart, The Concept of Law, 105.

Rationalizing Practices  117 explanation. It is, therefore, not surprising that objections to legal positivism came up by way of showing that an attempt to reduce legality to institutional facts is bound to be incomplete. In particular, as many have argued, truths about morality are also in play, always and necessarily so, according to Dworkin, or at least sometimes, as others claimed. I want to focus here on Dworkin’s critique of Hart’s reductionist project because there are two stages in which Dworkin developed his critical arguments, and they nicely capture two different critiques of Hart’s grounding reductionism. Let me call them the argument from judicial reasoning, and the argument from interpretation, respectively. Both aim to show that legality cannot be reduced to institutional facts because legality depends, at least partly, on moral truths. My aim here, however, is not to defend Hart’s particular theory of legal ontology, but to show the methodological differences between the two approaches and how they bear on the nature of a metaphysical account of a social practice like law.

6.2.1  The Argument from Judicial Reasoning It is common ground to critics of legal positivism to argue that, at least in some cases, truths about morality, what is morally right or wrong, determine the legality of norms. The standard move here, especially among legal philosophers in the common law tradition, is to draw this conclusion from observing judicial behavior and taking seriously ways in which judges actually reason to their legal opinions. We are reminded that judges, sometimes at least, take principles of justice, fairness, and morality to bear on the legal validity of norms they rely on. The metaphysical import of this critique, however, stems from the widely shared assumption that even if facts about morality are social facts (ontologically dependent on human interactions), they are not institutional facts. More precisely, the metaphysical interest in the anti-​positivist stance is closely tied to the assumption that facts about morality are not fully grounded in facts about people’s patterns of behavior and shared attitudes.15 Why is that? If you hold the view that whatever people believe to be morally right—​individually or collectively—​is, ipso facto, morally right, then the fact that legality is partly a matter of moral truths is not going 15 Dworkin clearly recognized this, and made an effort to substantiate his metaethical views in ways that align with his legal argument. See his “Objectivity and Truth.” Finnis also made it very clear that his views on the nature of law are closely tied with his cognitivism about morality (Natural Law and Natural Rights and “Natural Law Theory”).

118  Foundations of Institutional Reality to change the building blocks that ground legality; it would just add another set of attitudes to the inventory of the building blocks that ground legality. In order to show that a grounding reduction of facts about legality is bound to fail because legality depends on moral truths, one needs to assume that facts about morality are not fully constituted by whatever it is that people believe to be morally right. This does not seem to be problematic because very few philosophers would adhere to a simple-​minded version of moral subjectivism, whereby attitudes about moral issues are fully constitutive of moral truths. And yet this plausible assumption in the background entails that anti-​ positivism must concede that comprehensive errors about moral judgments are possible. After all, if what is morally right or wrong is not fully grounded in what people believe to be morally right or wrong, then it is possible that we might be comprehensively mistaken about moral truths. But, then, if truths about morality, sometimes at least, ground legal validity of norms, it follows that group-​wide, comprehensive errors about legality are also possible. And this, in my mind, is the Achilles’ heel of the anti-​reductionist argument under consideration. Dworkin’s theory of legal principles, published in the 1970s, is the most famous and perhaps still best-​developed argument purporting to show that a grounding reduction of legality is bound to fail because the kind of facts that determine what counts as a legally relevant consideration is often a matter of moral truths.16 Ordinary legal rules may well result from deliberate enactment by various authorities, Dworkin argued, but the law also relies on principles deemed legally binding, and those principles are derived, partly but necessarily, by way of moral arguments. Judges reason to the legality of a principle by striving to articulate the best moral justification of the relevant body of law. Therefore, what is morally right may determine what is legally valid. Now the details of Dworkin’s distinction between rules and principles, and the details of the argument itself, do not matter for our concerns here. I have expressed various doubts about it elsewhere.17 What matters is the plausibility of the conclusion, and the conclusion is shared by many critics of positivism who do not necessarily adhere to Dworkin’s theory of legal principles. Generally, then, the anti-​positivist critique of Hart’s reductionism is committed to the following theses:



16 Dworkin, Taking Rights Seriously, ch. 2. 17 See Marmor, Philosophy of Law, ch. 4.

Rationalizing Practices  119 (1) Facts about legal validity are sometimes determined by facts about what is morally right or wrong. (2) Facts about what is morally right or wrong, of the kind that (partially) grounds legality, are not fully grounded in what people believe to be morally right or wrong. (3) If there are facts about morality that are not fully grounded in what people believe to be morally right or wrong, then comprehensive mistakes about what is morally right or wrong are possible. (4) Therefore, comprehensive errors about what is legally valid are possible.

Since we are largely inclined to think that (2) and (3) are correct, anti-​ positivism clearly entails a commitment to (4). But the problem is that (4) is rather implausible. It makes little sense to suggest that an entire community of legal experts and legal practitioners would get their laws wrong. What would it mean to get the laws wrong? There are two main possibilities here. People can get their laws wrong either thinking that a norm is legally valid, whereas in fact, it is not (because it is immoral), or else people can fail to see that a norm is legally valid (because it has a great deal of moral support) even if the norm is not recognized as legally valid by the relevant legal actors. But both of these kinds of comprehensive mistakes are implausible. If all those who practice law in the United States, for example, believe that decisions of the US Supreme Court amount to the final settlement of what the law is, and the US Supreme Court holds that the law is X, then it’s very difficult to see how the law in the United States would not be X because the court had made a moral mistake. Furthermore, suppose that the US Supreme Court makes a systematic moral error over a period of time—​though not necessarily a grave one—​reiterated in numerous holdings. The conclusion would be that the entire legal community follows putative laws that are not really laws, norms that lack legal validity. This makes very little sense.18 Now of course it’s a bit more complicated than this. Laws may conflict or they may be indeterminate in all sorts of ways. It’s possible, for example, that one holding of the Supreme Court is in some conflict with another holding, and therefore the law on that issue might remain in some doubt. But legal conflicts and indeterminacies aside, the idea that an entire community of legal experts and practitioners would misidentify what is legally valid in their jurisdiction, in either sense of misidentification mentioned above, seems rather far-​fetched.

18 For a more detailed version of this argument, see Marmor, Philosophy of Law, ch. 4.

120  Foundations of Institutional Reality You might find this a bit too rough and vague, so let me try to be more precise about this idea that comprehensive errors about what is law in our system are not possible. Remember that, according to Hart, the grounding of legality is a two-​stage affair: legality, in any given system, is grounded in the social rules that determine what makes law in the system, what counts as a legally relevant act, and how. (As I will explain in the next chapter, these are the basic power-​structuring rules.) This is one grounding stage, whereby we ground the legality of particular norms or directives in social rules and particular actions or events taking place in the world (including, possibly, customarily so). And then, as we saw in detail in Chapter 3, social rules, as such, are grounded in regularities of conduct coupled with the appropriate collective attitudes. So let me start with the more foundational stage of grounding, the basic power-​structuring rules that determine what counts as legal in the relevant society, or the “rules of recognition” in Hart’s terminology. Since these are social rules, comprehensive errors about their basic contents are not possible. But remember that the impossibility of comprehensive errors is confined to the content of the social rules, not other aspects of it. People can be mistaken about the nature of the rules, their origins, and whatnot. If you remember the history of Europe in centuries past, the ideology of the Divine Right of Kings was widely shared, which means that people actually believed, or so it seems, that legal-​political authority is sanctioned by God. That was a mistake, and a fairly comprehensive one at that. But even under this misconceived conception of the ontological status of the rules of recognition, the content of the rules of recognition could not have been comprehensively misconceived. There is room for some vagueness, with some borderline cases in the margins, as I will argue in the next chapter, but that is not the point of contention here.19 None of this, however, would seem to rule out the possibility that the social rules of recognition would be such that their contents render legality, either necessarily or in some cases, depend on moral truths or moral right and wrong. Perhaps in our society, the Leader’s directives are law only if they

19 In later work Dworkin went to considerable lengths to show that lawyers and judges actually disagree about the criteria of legal validity in their jurisdictions, emphasizing that the disagreements are not about borderline cases (Law’s Empire, ch. 1). He took this to show that there are no rules of recognition because rule following requires agreement on the content of the rules. Few found Dworkin’s argument convincing on these points; the disagreements about legality that Dworkin pointed out seem to be very much in the margins, and it is not clear that they are disagreements about rules, as much as they are about interpretative methods, concerning ways of changing the law by judges and adapting laws to novel circumstances.

Rationalizing Practices  121 are just! Let us assume that this is possible.20 But then another kind of comprehensive error is made possible. Suppose the Leader directs that the law is X, and suppose that for a while we are fine with it; we do not regard X as unjust. So for a while, we regard X as legally valid, part of our laws. Could we have made a mistake about that? Can we come to discover, perhaps decades later, that X is actually unjust? As I mentioned earlier, on any plausible view of morality and justice that anti-​positivism presupposes, the answer has to be yes—​we could have thought that something is just when in fact it is not. But, then, if anti-​positivism is correct, we would also need to conclude that X had failed to be law—​it has not been part of our legal order for all this time when we thought that it was. So it may turn out, on this view, that an entire community of lawyers and legal experts, including judges, were wrong, and perhaps have been wrong for a long time, to have thought that X is law. And this is what I find very doubtful. Now you might think that I find this type of comprehensive error impossible because I have assumed, without much of an argument, that laws are necessarily artifacts, products of human agency. But some, like Dworkin perhaps, may disagree. They may think that even if some laws are clearly artifacts, or some aspects of legality share features with massive artifacts of other kinds, laws are also another type of entity, the moral type, which is not entirely a product of human agency.21 I do not have a metaphysical argument to disprove this possibility. Pointing to the fact, and a fact it is, that countless laws are morally wrong, even iniquitous, is not a good argument. A Dworkin type of anti-​positivism can allow for the view whereby legality is a mix of the moral with the product of human agency, they may come into some conflict, and sometimes we end up with laws that are morally far from good. My argument here is not an attempt to refute the anti-​positivist stance with a direct argument; it is an objection based on the price tag. And the price tag, I contend, is very high: the anti-​positivist argument that allows for legal validity to depend on moral truths makes it possible for a whole community of legal experts and legal practitioners to be mistaken about the laws they have, about what is legal and what is not in their community. To me this makes very little sense. Notice, however, that even if Dworkin and other critics are right and I am wrong about this, none of it 20 The assumption is arguendo here; I have raised doubts about its plausibility elsewhere for reasons that should not detain us. See, for example, Marmor, Philosophy of Law, 92–​95. 21 I think that Atiq’s interesting new version of anti-​positivism relies on such an assumption. It can be construed as suggesting that law is partly an artifact but partly a moral issue, combining some elements of both kinds (“There Are No Easy Counterexamples to Legal Anti-​positivism”). I find this rather mysterious, but I also acknowledge that mystery is often in the eyes of the beholder.

122  Foundations of Institutional Reality would amount to a critique of the metaphysical arguments deployed thus far. The disagreement pertains to the nature of law, not the nature of artifacts or social rules, their grounding and ontology. In other words, the anti-​positivist argument under consideration does not contest the idea that if and to the extent that law is an artifact, it could not be subject to comprehensive errors; it contests the assumption that law is only a product of human agency. And this may be the point in the argument where we hit rock bottom. All I can show is what follows from the fact that law is a massive artifact, essentially a product of human agency. I’m not sure how one can show that law is nothing more than that.

6.2.2  The Argument from Interpretation The second stage in Dworkin’s critique of reductionism in legal philosophy challenges Hart’s reductive ambitions from a methodological perspective, and in a way that might have much wider implications for the metaphysics of sociality generally. It is based on Dworkin’s general theory of interpretation, arguing that explanations of social practices are inevitably interpretative and, therefore, necessarily intertwined with evaluative judgments.22 To put this in terms more closely aligned with our previous discussion, Dworkin’s idea here is that reductionism cannot be reconciled with the internal rationalization constraint. The latter calls for a hermeneutical approach that requires any philosophical explanation of a social practice to engage, argumentatively, with the evaluative perspective of the participants.23 Let me state from the outset that I think that Dworkin is right about one thing and wrong about another here. He is right to argue that an account of internally rationalizing features of social practices is going to be inevitably interpretative, and therefore inescapably evaluative. I think that he is wrong to assume that this amounts to an argument against grounding reductionism; the latter is not subject to the internal rationalization constraint. Dworkin’s theory of interpretation is complex and insightful, and parts of it are controversial. Since I have written on it extensively elsewhere, I will

22 See, mainly, Dworkin, Law’s Empire and Justice in Robes. 23 Historically this theme is not new; hermeneutical methods were seen as critiques of positivism in social sciences from the start, so this is just the latest iteration of a debate that has been going on since the mid-​nineteenth century.

Rationalizing Practices  123 not belabor it again here.24 Suffice it to acknowledge that I am not going to dispute Dworkin’s claim that interpretation is partly but necessarily a matter of evaluative judgments. Generally speaking, I think that Dworkin is quite right to argue that we cannot begin to interpret anything, be it a work of art or a piece of legislation, without having a sense of what kind of thing it is and, crucially, what makes things of that kind better or worse, worthy of our appreciation, and similar views about values and merit.25 In other words, I will just assume here that Dworkin is quite right to maintain that, once we are in the business of interpretation, we cannot avoid relying on some evaluative views about the merits of objects of the kind we interpret. Interpretation is not mere description, for sure. But, now, in order to see why Dworkin and others would come to think that interpretation is inevitable in the context that is relevant to our discussion, let me revert to the example of works of art, and the question of whether something like the Leaning Mirror is a work of art or not. When I first encountered this work in the museum, I was accompanied by a friend who happens to be a renowned art historian, and we got into an argument about it. I was somewhat dubious that what we were looking at (just a mirror stuck in a pile of sand, remember) is a work of art. But my friend objected, saying that it certainly is, though perhaps not a great one at that. Thus we ended up having an argument about the question of whether an o is an F or not, whether this object is art or not. You can easily guess the arguments we deployed here. I pushed the point that a work of art needs to demonstrate some special talent or skills in creating it, while my friend pointed out that the talent is sometimes just the ability to see something that had eluded others, not necessarily to do something that others cannot readily accomplish. I argued that an object that can be easily mistaken for a pile of rubbish if found on the street doesn’t quite get to be art. My friend reminded me of the important contribution of Dada and its counter-​art revolution, and the ways in which it changed our conception of the social role of art and how to appreciate it. And so on and so forth, we kept arguing for a while. But, of course, our premises and arguments were closely entangled with our views about the values of art, in general, about what makes art worthy of our appreciation, and things like that. In other words, it seems to be clearly the case that different views about values we associate with art entail different answers

24 See, for example, Marmor, Interpretation and Legal Theory and Philosophy of Law, ch. 4. 25 Dworkin, Law’s Empire, ch. 2.

124  Foundations of Institutional Reality to the question of whether this object is a work of art. More generally, it is one of those cases in which the answer to the question of the type Is o an F? depends on values we associate with F. I hope you can see that if a similar dialectic applies to law as well, a reductive explanation of legality is in trouble. After all, it is precisely the point of legal positivism’s critics that views about what makes law or particular legal domains valuable, and reasons, in particular moral reasons, for appreciating various aspects of law partly determine what counts as legal. And if that is the case, then the kind of grounding reduction of legality we explored above is not likely to succeed. One line of response to this challenge is to question the analogy between law and art in this context. As I argued elsewhere in some detail, there are, indeed, reasons to be very careful with analogies between law and art or, at least, with taking those analogies too far.26 But this is not the main line of argument I want to pursue here. Instead, I want to show that, even in the context of art, a hermeneutical approach is called for only if we seek the kind of explanation that aims to articulate internally rationalizing features of the practice. A grounding reduction of social practices, including art, need not be constrained by that. Let us return to the argument I had with my friend about Leaning Mirror, whether it is a work of art or not. It was, basically, an argument about internally rationalizing features of art. What we disagreed about concerned questions about what makes art valuable for us, what makes us appreciate it as a distinct form of creative achievement, what kind of creativity we value, and things like that. It is therefore easy to agree with Dworkin that such arguments are inevitably interpretative. They combine elements of art history—​that is, facts about what people regarded as art in the past at different places and in different cultures—​how these things developed over time, and evaluative views about the kind of things that make us appreciate works of art, what it is that we value about it. We were trying to make sense of what the practice is and what it means for us, those who value it and participate in it in different

26 See, for example, Marmor, “Law as Authoritative Fiction.” In a nutshell, the two main differences between law and art that are relevant here are these: art is typically created (or, at least, generally perceived to be created) with the intention of becoming an object of artistic appreciation, inviting different interpretations of it; law is something that we have for practical purposes, not for purposes of appreciation or admiration. Second, there is some sense in which ascribing artistic qualities to an object implicates a positive appreciation of it, something that is valuable or worth our appreciation. It is far from clear that this implication necessarily applies to law. By saying that it is the law in Si that X, I do not necessarily implicate anything positive about that.

Rationalizing Practices  125 roles. But this is not the only kind of theorizing about art one can envisage. A sociologist, historian, or philosopher might ask other kinds of questions. In particular, one might be interested in what counts as art in different times and places and why, on what grounds. What makes it the case that people treat certain artifacts or activities as distinctly artistic? In other words, a metaphysical inquiry into the nature of art might take a very different shape: we can be interested in the kind of facts that ground a complex (and admittedly rather vague) set of practices, and the kind of facts that would explain what makes it the case that certain activities count as activities of an artistic type or products of it as works of art, if and where they do. Can we have a metaphysical grounding reduction of art, fully explaining it in terms of people’s conduct and shared attitudes, complying with the constraints of methodological individualism? I see no principled reason why we could not accomplish such a task.27 Many facts about art are institutional facts, partly dependent on social conventions that structure different artistic genres and various general expectations about what counts as an artistic accomplishment. These conventions and collective attitudes tend to vary with location, culture, and period, undergoing changes over time.28 In this respect, the analogy between law and art actually holds nicely—​in both cases, we have products of human agency that are significantly structured by rules and conventions. Conventions and collective attitudes play a significant role in making it the case that we identify certain types of products of human agency as products of that type, laws in one case and works of art in the other. In other words, both law and art are clear examples of situational artifacts. The main difference between them in this respect is that the constitutive conventions of law are power-​structuring rules, because the law is an authoritative institution, extensively deploying and exercising normative powers. (More on this in the next chapter.) Art is not a practical authority, and normative powers play no role in its instantiations. The constitutive conventions of various forms of art are obviously not power-​structuring rules; they are rules, conventions, and general social understandings that define what counts as a product with certain qualities or as a work of art or art of a certain type.

27 Dickie’s institutional theory of art certainly comes very close. (Art and the Aesthetic). See also Danto, “The Artworld.” 28 In previous work, I have argued that these types of conventions tend to come in two layers: we may have deep conventions determining what counts, generally, as art, instantiated by various surface conventions about particular genres that determine ways of creating particular types of art (Marmor, Social Conventions, chs. 3 and 6). And I argued that something similar applies to law as well.

126  Foundations of Institutional Reality Be this as it may, I would not venture to try to work out the details of such a grounding-​reductive account of art, or anything similar for that matter. The relevant question for us is whether an attempt to carry out such a project would need to comply with the internal rationalization constraint. I don’t see why it should. The task of a grounding account of art, or law for that matter, is not to explain the participants’ normative reasons to engage in the practice. As I mentioned earlier, there are practices and institutions we have no reasons to have; they may serve functions better not served at all, such as entrenching sexism, xenophobia, or racism. But this does not mean that they are practices we cannot fully explain. Whatever grounds their existence and functions would not be categorically different, metaphysically speaking as it were, from whatever grounds noble and admirable practices such as the arts or sports or law for that matter. To take an extreme, but not irrelevant, example: I think that, in principle, we can give a full account of the belief systems and collective attitudes prevalent in the ancient Mayan culture to explain their extensive use of human sacrifice in their religious rituals, even if there is almost nothing about that belief system that we can come to share or normatively appreciate. Which also means that it is quite possible, more than likely, actually, that our explanation of human sacrifice in the ancient Mayan culture would not be the kind of explanation that meets the internal rationalization constraint. The Mayan people would not have been able to see our explanation as one that rationalizes human-​sacrifice rituals to them. That, in itself, would not make our explanation deficient in any metaphysical sense. One might immediately object here, saying that even if we do not share, normatively speaking, the values and purposes that rationalize a practice for its participants, any philosophical account of the practice must articulate its rationalizing features, what makes the practice intelligible and desirable for those who engage in it. After all, we have already seen that no social practice can exist for long without its participants sharing, at least to some minimal extent, views about the practice’s point or value, the kind of things the practice is there to promote or accomplish. Now that is true, of course. We must be able to articulate the rationalizing features of a social practice, but it does not entail that the rationalizing features of the practice must be of the internal kind. For the purposes of a grounding-​reduction account, externally rationalizing features are quite sufficient. We must explain the rationalizing functions of the practice, what makes participants engage in it, but not necessarily in terms that the participants would come to recognize as their own, as their own reasons to engage in the practice or value various aspects of it. What

Rationalizing Practices  127 we need for the purposes of a metaphysical explanation of a social practice is a full account of the attitudes and dispositions shared by the relevant population, the rule-​following behavior exhibited in light of those shared attitudes, and the participants’ beliefs about the point or the value of the practice for them. A grounding reduction of such elements in the explanation does not have to make sense, normatively speaking, to the participants themselves. Only an interpretative account of a social practice, one that initially seeks to rationalize the practice for the participants, striving to illuminate their own self-​understandings of the point of the practice and their reasons to engage in it, would have to engage with those self-​understandings argumentatively, on the basis of normative reasons and values. I think that this is precisely how Dworkin saw his own theory of law, and where he found the failures of Hart’s legal positivism. Dworkin argued that the latter cannot give us an adequate account of the internally rationalizing features of law, at least not in the common law tradition. In particular, Dworkin argued that legal positivism does not have the tools to explain the moral-​political significance for judges and other legal actors to adhere to past legal decisions, why they should care about the constraints of the past, and to what extent they should.29 And perhaps Dworkin was right about that.30 But if I am correct to argue that a grounding-​reduction theory of law should not be expected to meet the internal rationalization constraint, then Dworkin’s argument misses the target; it just answers a different kind of question. From the perspective of internally rationalizing features, it may well be the case that whether an o is an F or not depends on values participants associate with F; that is true in many cases, and it may well be true in the law as well. However, I hope it is also clear that a metaphysical reductive account of law, or art for that matter, strives to account for a different kind of question. It purports to establish why something counts as an F in a given society, S, and to the extent that we need, as we usually do, to include in the explanation various aspects of F that rationalize it for members of S, those rationalizing features need not be internal; they do not necessarily have to make sense, normatively speaking, to the participants in S. 29 Dworkin, Law’s Empire. 30 Interestingly, Raz probably sides with Dworkin on this point; he also held the view that a theory about the nature of law should make it fully intelligible to those whose practice it is, and in ways they can come to recognize as rationalizing the practice for them. His disagreement with Dworkin is mostly about what it is that rationalizes the practice. In Raz’s view, it is law’s essential authoritative nature and conclusions about legal validity that follow from it. See Raz, Ethics in the Public Domain, chs. 8–​10; and “Can There Be a Theory of Law?”

128  Foundations of Institutional Reality In other words, the argument of this chapter shows that a suspicion many philosophers have had over the years, that Hart and Dworkin were arguing past each other, may well be quite right.31 A metaphysical reductive account of law aims to achieve one thing, while an interpretative account of law aims to achieve something else. This does not mean, however, that the projects are entirely unrelated. Far from it. Any theory about aspects of social practices and institutions has to make some metaphysical sense. However, within the bounds of metaphysical foundations and epistemic constraints that follow from them, there is ample room for interpretative disagreements among the participants themselves, about various aspects of any social practice or institution. Metaphysics, by itself, cannot settle those disagreements. They are, as Dworkin rightly claims, partly evaluative in nature. Finally, let me conclude by reminding us that law was meant to be only an example here. The differences between metaphysical and hermeneutical theories of social practices apply much more broadly, and pertain to any theoretical explanation of social practices.



31 Including Hart himself in the postscript to The Concept of Law.

7 Power-​Structuring Rules Writing about social issues in the analytical philosophy tradition, we tend to forget that perhaps the most important aspect of our social lives consists in brute power some people exercise over others. Individuals and groups can exploit others, sometimes many, many others, subjugate them to their needs and desires, and control the behavior of people in myriad ways. I will assume here that brute power is (or at least is measured by) the ability to overcome resistance. If I can make you do something that you would not have done otherwise, or make you refrain from doing something that you would have wanted to do, then I have power over you. Brute power often tracks physical power, the actual ability to physically harm the other, but it is not limited to that. People can exercise brute power by intimidation, by deceit and manipulation, by exploiting the needs of others, perhaps even by promises and all sorts of bribes, real or imagined, and whatnot. Those who have power can subordinate others to their desires. In this respect we are not all that different from other animal species. The remarkable aspect of our sociality is that human societies managed to transform a great deal of brute power into an elaborate normative framework as well. For now, suffice it to say that normative power is the ability to unilaterally introduce a change in the normative situation of another person or persons.1 It is the ability to change what another person is obligated to do, or has a right to do, or has power to do. In some fundamental way, this is what law is all about. The law structures the power relations in society by normative means, determining who gets to rule over others in myriad ways, big and small, and how power relations are supported, or not, by collective, organized force. The law is not unique, however, in its extensive use of normative powers. Countless social practices and institutions deploy and exercise normative powers. If you work 1 One might worry that the qualifier “unilaterally” is too strong; after all, sometimes it takes two to tango, for example, to get married, or to reach an agreement. That is true, of course; some normative changes require the coordinated exercise of power by two or more agents. But each agent has the unilateral ability to introduce, or not, the normative change that is relevant. A contract or agreement requires an offer and an acceptance of that offer; each exemplifies a unilateral ability to change the normative situation of the other.

Foundations of Institutional Reality. Andrei Marmor, Oxford University Press. © Oxford University Press 2023. DOI: 10.1093/​oso/​9780197657348.003.0007

130  Foundations of Institutional Reality in a university, you would not have to look far for examples—​I doubt that I can list all the committees, deans and associate deans, chairs, and other administrators, who have some normative powers over various aspects of my university work. Generally, employers have many powers directing their employees or granting them various powers over other employees. Religious institutions tend to deploy normative powers within their institutional settings. Referees and umpires in sports competitions would normally have the power to declare scores, remove players from the game, and things like that. There is no social practice or institution with some hierarchical structure without a range of normative powers that are recognized and practiced within it.2 In this chapter, I want to analyze what normative powers are and show that they play a foundational role in an account of our institutional reality. There are two main conclusions I want to establish in this chapter. First, normative powers are basic and foundational to our social reality despite the fact that they seem to operate on preexisting normative relations. Second, there are basic, power-​structuring social rules, and they are at the foundations of hierarchical social practices, such as law. Along the way, I also hope to show that basic power-​structuring rules are inevitably complex, coming in sets of intertwined social rules, and yet they allow for a considerable amount of vagueness and indeterminacy in their applications.

7.1.  Normative Powers and Institutional Facts A close cousin to normative powers is the ability to change people’s situation by saying things, by expressing some words or sentences (or sometimes by some other nonverbal expressions). Speech acts of various kinds can affect the condition of others in myriad ways, ranging from consequences that are trivial or mild, to changes in people’s lives that are significant, sometimes profound, or even tragic. In these days of the coronavirus pandemic, governments instruct us to stay at home for weeks on end. We are instructed where can we go, whom we can socialize with, what kinds of work we are allowed to do, and countless other directives that affect our lives in profound and consequential ways. At least on the surface, all of these consequences are achieved by officials expressing directives, orders, and recommendations. 2 The distinction between hierarchical and flat social practices was introduced in Chapter 1, and I will return to it toward the end of this chapter.

Power-Structuring Rules  131 But, of course, this is just an extreme version of what happens in normal times—​our lives are significantly shaped and governed by countless directives and regulations, legal and other. How can words do so much? How is it possible that by expressing a few words or sentences, others’ lives can be changed so profoundly and in so many ways? One answer that covers many but not all the cases consists in the fact that there is an institutional framework in the background that enables people to have this impact on others’ normative condition by saying things in certain ways. And these abilities are, typically, normative powers. There is a substantial overlap, but not identity, between the idea of normative powers, institutional facts, and people’s ability to change others’ situation by expressive means. Normative powers do not necessarily depend on institutional facts; they may be moral powers, supported by reasons and values that may or may not depend on institutional facts. For example, we have the power to wave others’ obligations by consent, or the power to grant the addressee the right to one’s performance by making a promise. Arguably, these kinds of normative powers do not depend on institutional facts—​they may be moral facts, entirely supported by moral reasons.3 Furthermore, though normative powers are typically exercised by expressive means, the expression does not have to be verbal. And then, of course, not every expression or speech act that affects the condition of another is an exercise of normative power, though sometimes it is an exercise of brute power. My expressions of love to my children may affect them in important ways, sometimes even morally relevant ways, but it’s not an exercise of normative power. A humiliating or offensive speech act may be an exercise of brute power, but not, normally, a normative one. In this section, I will try to unpack these distinctions and some lessons to be learned from them. In the second section, I will explicate the notion of basic power-​structuring rules and show how they lie at the foundation of significant aspects of our social reality. It is common ground to all understandings of normative power that it is the ability or capacity to introduce, unilaterally, some change in the normative condition of another.4 Roughly, then, we can say that A has normative 3 I say “arguably” because the case of promising is controversial; according to the Humean conception, promises are essentially institutional, depending on a social practice. See, for example, Kolodny and Wallace, “Promises and Practices Revisited.” According to Valentini, all exercises of power are practice dependent (“Normative Powers”). I’m not so sure about that, but as will become clear later, my argument does not depend on accepting or rejecting such views. 4 The idea of normative power probably originates with the influential work of Hohfeld on the nature of legal rights (Fundamental Legal Conceptions), although Hart attributes the main ideas to Bentham (Essays on Bentham, ch. 8). Hohfeld identified power as one of the four basic forms of a legal

132  Foundations of Institutional Reality power over B if A can make it the case that by φ-​ing B acquires a right, permission, duty, or power that B would not have had but for A’s φ-​ing. This would seem to entail that normative powers are not the foundational building blocks of normative relations or normative concepts. First there have to be duties, rights, prohibitions, permissions, and things like that for an ability to change them to come into the picture. And indeed, analytically, this sounds right.5 But it doesn’t mean that powers are dispensable or somehow not quite basic. Some rights and obligations make no sense without the power to modify or waive them—​they would be very awkward and hardly usable if people did not have the power to waive or condition them at will, such as the right not to be touched (physically), or the right to own property. Without the accompanying power to waive the right, to modify it or condition it in various contexts, such rights would not make much sense, and there would be no point in having them. So perhaps it is true that power, as the ability to introduce a change, operates on existing rights and obligations and other normative relations, but powers are not less basic or foundational for that. We could not eliminate normative powers from our normative landscape without altering that landscape beyond recognition. The rough definition of normative power, as the ability to change rights or obligations, etc., of others needs some refinement. Consider this example: A walks too close to the edge of a cliff, in immediate danger of losing his balance and falling down the cliff. B happens to be very close by and can easily lend a hand, preventing A’s disastrous fall. One would think that this puts B under an obligation to lend a hand and prevent A’s demise. Has A exercised normative power over B in this situation? Suppose A had created this situation purposefully, precisely in order to force B into this obligation to help—​would that make a difference? In these days of the pandemic, if A is infected with the coronavirus and gets in close physical proximity to B, B would immediately come under an obligation to self-​quarantine for a few days. Whether A’s actions in these examples is deliberate or not, the cases do not seem, intuitively, to be exercises of normative power. But it is very difficult to specify what is missing here. In both examples, A’s action changes the normative condition of B, putting B under an obligation that B would not have had but for A’s action. And yet it does seem a bit strange to suggest that right. Hart and many other scholars since have adopted Hohfeld’s definition, extending it beyond the confines of legal rights, and used the idea of normative power as I use it here as well. 5 I will come to qualify this slightly later, in section 7.2.

Power-Structuring Rules  133 these are examples of an exercise of normative power. Raz suggests that normative powers are vague and borderline cases are to be expected.6 Perhaps he is right about it; perhaps we should not feel surprised that some cases are neither here nor there. In some respect, they seem like ordinary examples of normative power, but in others, not quite. But this still leaves us wondering why putting someone in a position to be obliged to help another from falling off the cliff, or be obliged to help others by self-​quarantining, are not quite exercises of normative powers. What is it, intuitively, that is missing here, or at least make us think of them as borderline cases? Raz may have a very good answer. On his view, all normative powers are normative in a robust, moral, or evaluative sense: “a person’s act is an exercise of normative power if it brings about or prevents a normative change because it is, all things considered, desirable that that person should be able to bring the change about or prevent it by performing that act.”7 In other words, Raz explicitly identifies normative powers by “the considerations that establish their justification, taking their justification to be sufficient for their existence.”8 If this understanding of normative power is correct, then we can come to see why the examples I gave above are borderline cases. It is doubtful that, all things considered, people should have the powers exemplified by the two cases. It is somewhat doubtful that people should be able to impose an obligation to help by putting themselves (or others) in danger. So if we think about normative powers as grounded in their justification, all things considered, then we can certainly see how we get these borderline cases. Is it plausible to think that normative powers are only those that it is justified, all things considered, for people to have? Can we have an account of normative powers that renders them entirely moral in nature, entailing that one has a normative power if and only if one should have that power all things considered? There is nothing amiss here, as far as I can tell, as long as we are clear about the phenomenon we strive to account for. If, but only if, Raz’s account is confined to an explication of normative powers we have from a moral or, perhaps more generally, evaluative point of view, then the idea that the powers we have are those that serve the relevant values, and are all things considered justified, is certainly plausible. Not uncontroversial, but quite plausible.9 The problem is that we often talk about normative 6 Raz, “Normative Powers.” 7 Raz, “Normative Powers,” 163. 8 Raz, “Normative Powers,” 163. 9 Raz might object, and perhaps rightly so, to the idea that the moral domain is special or distinct from other domains of value. My formulation should not be taken to suggest that morality has clear

134  Foundations of Institutional Reality powers that exist, in some sense, even if the power holder should not have that power. People can have normative powers that, all things considered, they ought not to have, morally or otherwise. This suggests that, regardless of any moral or other evaluative considerations that may apply, there is a social sense of normative powers, distinct from brute power, that people have, and we need to explicate what this sense is and what makes those powers exist. The obvious examples of normative powers that do not necessarily depend on their justification is in the legal domain. Governments, and agents acting on their behalf, as well as private agents acting in various specified capacities, sometimes have normative powers, granted by law, that they ought not to have, all things considered. And I do take the “all things considered” seriously here. It is, of course, open for Raz to argue that many legal powers we may have preferred not to exist are nevertheless justified, all things considered, because it is justified, all things considered, for the law to have the power to create those powers.10 This reply works, up to a point, but not all the way down. Some legal regimes or legal institutions are such that all things considered there is no good reason to have them, no justification for their existence. And yet, there is a clear sense in which people exercise normative powers even if those powers are such that they should not have had them, all things—​including the power to create those powers—​considered. The law, however, is not the only domain in which normative powers exist even if they are not, all things considered, justified. Many social practices, some conventionally grounded, deploy normative powers within the practice, and some of those powers may not be justified, all things considered.11 Generally speaking, we should bear in mind that the main function of some social practices and institutions is morally questionable, sometime profoundly so, entrenching as they do questionable hierarchies, subordination, exploitation, and things like that. There is no justification for having these boundaries or even that it is all that different from other domains of value. Either way, Raz’s understanding of normative powers is explicitly derived from values, whether moral or any other. That’s essentially what I mean in the text. Raz is not alone, of course, in arguing that normative powers are derived from moral considerations. See, e.g., Tadros, “Appropriate Normative Powers”; but cf. Valentini, “Normative Powers.” 10 Raz, “Normative Powers.” 11 The institution of slavery, in its various forms, is the obvious example of normative powers that there is absolutely no justification for having. But we don’t need to rely on such extreme cases. Existing labor relations in many countries sanction normative powers of employers over their employees in ways that are morally deplorable; some educational institutions may have normative powers over their students beyond what is necessary and warranted; many organized religions exercise normative powers over their believers in ways that are morally questionable; and so on.

Power-Structuring Rules  135 practices, and all things considered, it might be better not to have them, but there is no denying that such practices include normative powers. It is part of our social reality that those powers exist, even if they should not have existed, all things considered. One could argue, however, that the exercise of unjustified power does not really create the changes in the condition of others that it purports to create. If A purports to impose an obligation on B by exercising a normative power that A should not have had to begin with, all things considered, then B does not come under a genuine obligation in virtue of A’s action. Now of course this is plausible from a moral point of view. If A’s normative power is not morally justified, then it makes sense to conclude that A’s purported use of the power does not impose, or at least may not impose, a moral obligation. But obligations, rights, and powers are not confined to the moral domain. A legal power may confer a legal right or legal obligation whether the right or the obligation is justified or not. The world would be a better place if all our legal obligations perfectly matched our moral ones, but sadly that is not the world we live in. All this comes to suggest that there are two senses in which we can talk about the existence of normative powers: we can talk about normative powers as moral facts or as institutional facts. I take Raz’s account to be about the former, and my intention here is to articulate an account of the latter. Therefore, from now on, when I talk about normative powers, I will focus on institutional facts, of the kind grounded in social rules, whether those rules and the powers that emerge from them are morally justified or not. To be sure, I am not assuming here that Raz and others are wrong to maintain that many normative powers people have do not depend on institutional facts. Some normative powers we have are such that we have them solely in virtue of the moral considerations that apply. My focus in this chapter, however, is on the idea of normative powers as institutional facts, facts that form part of our social reality, regardless of their moral merit. Note, however, that one and the same normative power can be both moral and institutional. The law, for example, can recognize and sanction normative powers we morally have, or stipulate new powers that are morally justified. Overdetermination is not ruled out, and does not pose any particular problems here. Normative powers are usually exercised by communication. If you want to change somebody’s normative condition, you would normally need to communicate that content to the person whose normative condition you intend to change. Different types of speech acts might be in play, depending

136  Foundations of Institutional Reality on the kind of normative power exercised and other circumstances. In order to impose an obligation on someone, it normally takes issuing a directive or an order.12 To give someone a reason, weaker than obligation, it usually suffices to express a request or to give a kind of authoritative advice.13 To grant someone a new right, it might be enough to make a promise, or to agree to some terms in a contractual or quasi-​contractual way. Either way, normative powers are generally exercised by the performance of speech acts. Sometimes, however, power is exercised expressively, that is, by some communicative act that does not involve speech. Expressions of consent, for example, vary considerably, and are often given by nonverbal communication. Upon entering a familiar establishment, like a restaurant or concert hall, one would normally be held to have expressed consent to abide by the rules and conventions that regulate behavior in such establishments. Expressions of consent to sex can take many subtle forms, often nonverbal, and sadly we know this all too well because these subtleties often give rise to serious moral concerns. Consent is not the only example, however. Even promises do not necessarily require an explicit speech act to undertake, and other normative powers can be expressed nonverbally.14 It is much more difficult, however, to think of examples where normative power can be exercised without communication. In order to bring about a change in somebody else’s normative condition, one needs to express that intended change by some means of communication. Two kinds of cases might be thought to be counterexamples. First, if you are inclined to think that my earlier pandemic example—​whereby simply by being in physical proximity to someone when you are infected, you oblige them to self-​quarantine—​is an exercise of normative power, then that might be a case of a noncommunicative exercise of normative power. As I mentioned earlier, however, I am doubtful that these kinds of cases are examples of exercising normative power. Another kind of case that might be thought to be relevant here concerns the idea of consent. There is a controversy in the literature on the morality of consent over the question of whether consent needs to be expressed at all;

12 Bach and Harnish, Linguistic Communication and Speech Acts, 47. 13 I have argued that there is a certain type of reason giving—​I called it presumptive reasons—​ which consists in an expectation to do as requested or at least to explain why not (Marmor, “Soft Law, Authoritative Advice and Non-​binding Agreements”). 14 A nice example is the legal power to dispense of property by “abandonment”; for example, by leaving your old couch on the street at some designated place, you confer the power on others to take it if they wish, that is, you legally abandon your property rights in the couch. I take it that the placement of the couch in the street is a nonverbal expression of your intention to abandon it.

Power-Structuring Rules  137 according to some views, what matters, morally speaking, is the attitude or mental state of the agent, not her expression of that mental state.15 Since my concern here is not moral philosophy, I wish to take no sides in this debate. I would only say that, on the purely mental state view of consent, if what really matters is only the attitude of the consenter regardless of her expression of the attitude, it would be rather difficult to see consent as an exercise of normative power. On this view, consent matters morally only because and insofar as it is wrong to treat people in ways they would not welcome, at least in certain contexts.16 That well may be so, but it would not necessarily show that normative powers can be exercised without communication. Consent matters on this view not as an exercise of normative power but as an indication, as imperfect evidence, of subjective preferences, which is what really matters. But, of course, these are controversial issues. Nevertheless, it is safe to assume that, by and large, barring some very unusual cases, the exercise of normative power is an expressive act, typically a speech act of one type or another. Why is this significant? In a way, it comes to show that a complex normative framework needs to be in the background in order to constitute the normative effect of speech acts exercising the relevant powers. Think about orders or commands, for example. When A orders B to φ, the linguistic expression conveys the idea that A intends to impose an obligation on B to φ, and intends B to recognize this intention, and to take it as a reason for being under an obligation to φ, and A presumes herself to have the relevant normative standing for expecting all of this. But this does not guarantee, of course, that the expression was a felicitous order. I can order the president of the United States to resign, but my “order” would be nothing but a farce. An employer can fire an employee by saying, “You’re fired,” but the employee cannot fire the employer by saying something to the same effect, just as I cannot order the president to resign. Our words have the powers they do, when they are exercises of power, only due to a complex normative background that grants those speech acts or expressions, uttered by the appropriate agents in the appropriate circumstances, the normative effect they purport to have.17 15 E.g., Hurd, “The Moral Magic of Consent.” 16 Consent to sex is usually the context in which these debates play out. I have great sympathy for the view that sex should be consensual in a truly subjective sense, aligned with the actual wishes and intentions of the parties, not overly depending on expressions of consent or lack of it. But surely sex is one of those contexts where things ought not be done to people in any way that they would not genuinely welcome. 17 The point I am making in the text here should not be confused with a larger issue that has been debated in the literature on so-​called performative speech acts. J. L. Austin (How to Do Things with

138  Foundations of Institutional Reality As Raz rightly notes, normative powers can be either derivative or basic.18 A derivative power is one that has been granted, as such, by the exercise of an existing power. To give an obvious example, legislatures have the power to grant a great variety of powers to other agencies, to the executive branches, to private citizens acting in various capacities, etc. Thus the power these agents have would be derivative—​they are powers granted by the exercise of the legislature’s power to grant them. And this is a very pervasive phenomenon, including in much more mundane contexts. An employer would normally have the power to grant normative powers to some of their employees, making them in charge of exercising certain normative powers relative to other employees. Most of the normative powers attached to official roles are granted by the normative powers of those who can create those offices and regulate them. A trustee may have the power to delegate some of her powers to another trustee, and so on. The simple idea is that most normative powers we encounter in our everyday lives are powers that were created by exercising normative powers to create them. All these we can call derivative powers. But not all powers can be derivative, of course, otherwise we would get an infinite regress, which makes no sense. Some powers are basic, not created or granted by the exercise of any other power. And then, of course, the question is what kind of things can ground basic normative powers? In particular, what grounds basic powers of the institutional kind? If it is not the justification for having them, all things considered, then what kind of social facts make it the case that basic normative powers exist? There are two main options here: basic normative powers can be created by agreement between the relevant agents, or they can be constituted by social rules. The former option is very limited in scope; agreement between a number of people to grant some power to an agent, constituted by the agreement itself, would be limited to groups consisting of a fairly small number of people, who can communicate with each other and reach an agreement or an understanding, and would be limited to the parties’ willingness to

Words) and Searle (Speech Acts and “How Performatives Work”) have maintained that performative speech acts are only made possible by conventions in the background that grant those speech acts their performative effect. Bach and Harnish (“How Performatives Really Work”), following an early lead by Strawson, argued against this conventionalist account, showing that most performatives are accomplished, as the kind of speech act they are, without any conventional background. I have mostly sided with Bach and Harnish in this debate (Marmor, Social Conventions, ch. 5).

18 Raz, “Normative Powers.”

Power-Structuring Rules  139 feel bound by their agreement for the duration of the power they grant.19 As David Lewis rightly observed, social rules tend to emerge precisely when agreement between the relevant agents is practically impossible to obtain, mostly due to the large number of agents involved.20 Thus, given the context of our discussion, I will ignore the option of creating powers by agreement, and assume that the kind of basic normative powers that structure our social practices and institutions are constituted by social rules.

7.2.  Basic Power-​Structuring Social Rules Jean-​Jacques Rousseau famously claimed that “the first person who, having enclosed a plot of land, took it into his head to say this is mine and found people simple enough to believe him, was the true founder of civil society.”21 A tad sarcastic, perhaps, but I think there is a deep and important insight in Rousseau’s words, and I hope that the idea of basic power-​structuring rules will help us see the point of it. What Rousseau has in mind, I think, is something I alluded to in the beginning of this chapter concerning the transformation of brute power to normative power. Imagine that you ask yourself at what moment in our history brute power got transformed into a normative idea. There was no such moment, of course, but we can try to imagine what would have to be the case for such a moment to have existed, as it were. Rousseau’s answer is that it takes two elements: claims and beliefs. Somebody has to claim normative power over others and the others need to believe him, or at least exhibit acquiesce in the claim. In one sense, however, Rousseau’s idea might seem a bit circular—​to claim that something “is mine” assumes a great deal of normative complexity. For one thing, it is not a physical relation to an object, but a normative one. And it is a rather complex one at that. Ownership, as well recognized, is a complex web of norms structuring relations between persons in relation to an 19 One might worry that agreement is itself an exercise of normative powers. Often it is, of course, but I doubt that always. An agreement to do something together, like taking a walk together, involves the undertaking of some commitments, perhaps, but I am far from sure that these undertakings necessarily amount to exercises of normative powers. Either way, since I’m going to bracket powers constituted by agreements, I need not prove the point here that basic normative powers can be constituted by agreements. 20 Lewis, Convention. Lewis was interested in a particular type of social rules, those that emerge to solve large-​scale, recurrent coordination problems. However, this particular point about social rules can be generalized to most cases. 21 Rosseau, Discourse on the Origin of Inequality, 44.

140  Foundations of Institutional Reality object, determining such things as what one may or may not do with the object relative to others, and how one can transfer rights in the object to others, abandon the object, and extract rent from its use.22 But this might only show that starting with property rights is not advisable—​they are too complex and they already incorporate a web of normative powers. Perhaps, but the complexity, in and of itself, is of the essence here. As we will see shortly, power-​ structuring rules are inevitably complex, though complexities often evolve over time, as need arises. In order to get a better sense of how foundational basic power-​structuring rules are, allow me a rather clichéd hypothetical. Imagine that a few thousand people find themselves on a desert island, totally isolated, with no feasibility (or intention) of ever leaving it. Perhaps they were shipwrecked, or perhaps voluntarily settled there to escape the next pandemic; it doesn’t matter. What matters is that they are ordinary people, by and large morally decent human beings, and that they need to organize their new life on this island for generations to come, best they can. Let us also stipulate, however, that their starting point on the island is one where they lack any social practices or institutions. No practices, customs, or institutional aspects of their previous lives are carried over to the island, as it were. Obviously, some people would naturally have potential for exercising brute power over others, and those potentials will operate at the background of their new lives. And now suppose that these people begin to organize their new lives, structuring a new society, with forms of governance, some property regime, division of labor in various fields, and so on. The point here is that it would not be difficult for us to imagine what kind of institutional settings we will find on this island after a while, maybe after a few generations. We can easily identify certain social practices that are bound to emerge sooner or later. The story of the hypothetical is not important. If you are weary of such stories, just imagine it as a hypothetical exercise in social engineering. Suppose you are given the task to design the institutional setting for a large community of people who are supposed to live together for generations. Ask yourself what kind of social practices and institutions you would absolutely have to include in this design. Let us focus on some fundamental aspects of necessities of societal life. First and foremost, it is difficult to imagine that when we revisit the island generations after its settlement, we will not find many rules that the islanders

22 Certainly, if not much before, by Hohfeld, Fundamental Legal Conceptions.

Power-Structuring Rules  141 follow. I don’t think we can even begin to imagine societal life without social rules. Why is that? Probably for the prosaic reason that many of our interpersonal problems and practical challenges tend to recur with considerable regularity, and it would be tremendously inefficient if we tried to resolve each problem ab initio, as if we had not encountered it before. Social amnesia is not a recipe for success; organizing our lives by following some rules is much more efficient. Furthermore, rules can be taught, and people can become habituated in using them, thus further reducing social costs and smoothing interpersonal relations. Therefore, it is fair to assume that our imagined island colony could not possibly thrive without social rules emerging there. Second, the island is bound to have some governing structure. It is difficult to imagine a society without some authoritative roles. Somebody would need to make new rules, and adapt rules to changing circumstances. Somebody would sooner or later need to make sure that rules are followed, respond to cases of noncompliance, adapt the rules to novel cases, and things like that.23 All of these authoritative roles would be structured by rules conferring specific normative powers. Since, ex hypothesis, the islanders had not had practices and institutions upon arriving to the island, whatever normative powers they confer and authoritative institutions they create would need to start with what I call basic power-​structuring rules. That is, some rules conferring power ab initio, without the power to create these powers granted by other norms. Furthermore, the basic power-​structuring rules would exhibit some inevitable complexity. Suppose, for the sake of simplicity, that the islanders start off with a simple dictatorship model. They choose a person to be the island’s Supreme Leader, and confer all practical authority to that person. Nothing can be as simple as that, however. The basic power-​conferring norms establishing the office of the Supreme Leader

23 In their provocative and controversial book, The Dawn of Everything, Graeber and Wengrow take us on a long historical and intellectual journey to disprove the naive assumption I make here that complex societies must have some authoritative roles. Their main point in the book is to prove that large and complicated societies existed and thrived at prehistoric and ancient times without any political authoritative structures. The historical and archeological arguments deployed in the book are controversial at best. Mostly they seem to point to lack of evidence of political authorities in some ancient societies despite their considerable population size. But, of course, from the fact that there is no evidence for P it does not follow that we have evidence for not-​P. The scholarly controversies over the rich and thought-​provoking content of this book will no doubt continue for a while. But even if all the evidence and arguments the authors present in The Dawn of Everything is true and accurate, it would only show that some societies, rather large and complex perhaps, managed to thrive without coercive political authorities. It would still not follow that we can run our lives in complex societies without some authoritative roles. As long as people have quarrels and disagreements, somebody would have to make some decisions taken to be binding on others. That is basically all we need to assume here.

142  Foundations of Institutional Reality would soon end up being more complex: there would have to be rules about succession (e.g., what happens when the Supreme Leader dies, who gets to replace her, and how?).24 Sooner or later questions about the limits of the Supreme Leader’s power will come up and need to be settled, questions about the Leader’s power to delegate power to others will arise, and so on. In other words, even the most rudimentary power-​structuring rules would have to embody some complexity, consisting of a set of intertwined norms defining normative powers of an office or institution.25 Roughly the same holds true about a property regime the islanders are bound to have. Given the fact that humans cannot live without material resources, and our relation to objects and material resources is essential to our survival, determining what is mine and what is yours, and what it practically means that something is mine and not yours, is one of the most fundamental aspects of interpersonal relations between humans. Some norms need to determine what belongs to whom and how those things are going to be decided, implemented, and modified as things unfold. Any set of basic power-​structuring rules would need to determine, among other things, some type of property regime. Finally, we are likely to find a range of specializations. People would have acquired certain roles in society, as farmers, teachers, technicians, healers, judges, political leaders, artisans, perhaps even artists, historians, and philosophers, and so on. Notice, however, that some of these social roles come with normative powers attached to them, while others do not. Therefore, we can draw a distinction between social roles and offices. People who occupy an office, like judges and legislators, teachers, or referees in sports competitions, play a role in society partly defined by the normative powers they get to exercise. Other roles, like those of a farmer, artisan, or medical healer (and, alas, philosopher), involve no normative powers—​people in those occupations perform certain roles in society, economic, social, or other, but they are not offices (in the sense I use the term here); their role is not fulfilled by exercising normative powers. Thus offices, unlike other social roles, are institutional in nature—​they need to be structured by rules, and many of those rules will be power-​conferring rules, and would constitute the normative powers attached to the office. 24 On the necessity of rules of continuity, see Hart, The Concept of Law, 61. 25 I have argued for an institutional conception of practical authorities, showing that what counts as practical authority is inevitably determined by social norms conferring normative powers (Marmor, “An Institutional Conception of Authority”).

Power-Structuring Rules  143 Is there a lesson in all of this? The main lesson is that we cannot imagine the evolution of society without the emergence of basic power-​structuring rules. Our islanders could not have any form of governance, a property regime, and many forms of social specializations without them. Whatever regime they end up having, including a property regime, it is going to be one that is structured by some basic power-​conferring rules, rules that determine who gets to decide on what, and how those decisions are going to be binding on others. I don’t think that as philosophers we would have much to say about ways in which social rules actually emerge; we better leave those explanations to sociologists and historians.26 But we know what social rules are—​at least I have tried to offer an account of them in previous chapters. The question now is whether there is anything special about basic power-​structuring social rules. And there might be an appearance of a puzzle here, which I alluded to earlier: normative powers enable people to introduce a change in the normative condition of others. Therefore, normative powers seem to operate on preexisting normative relations, enabling the power holder to change them at will. People must be assumed to have some rights and obligations for the possibility of changing them to arise. Presumably, it does not mean that we cannot have a basic power that grants a specific right or obligation ab novo; a new kind of right or obligation can be created by basic power-​conferring rules. Even so, some normative relations and some basic normative concepts, if you will, need to be in place before normative powers come into the picture. And this might give the impression that social rules structuring normative powers cannot be quite basic—​they are inevitably parasitic on some normative framework that must already be there, existing, as it were, before powers can change them. To some extent, this must be correct. The function of normative powers is to grant people or institutions the ability to introduce a change in the normative conditions of others. Normative powers have the function of structuring normative relations, modifying them, or creating new ones.

26 David Lewis might have begged to differ; his theory about the emergence of social conventions contains an explanation of how people tend to converge on salient solutions to coordination problems, drawing on game-​theory tools (Convention). In the last few decades, a thriving game-​ theory literature on these issues has emerged, modeling the emergence of various types of rules. Getting into the complexities of game-​theory modeling and what such modeling can show is far beyond the scope of this work (and my competence, for that matter). For a more comprehensive account of how various types of social rules emerge, see Bicchieri, The Grammar of Society.

144  Foundations of Institutional Reality All this seems right, but a real puzzle is not created by it. For basic power-​ structuring rules to have a function, it is enough that people take themselves to have rights and obligations and powers, even if there is nothing more to it than claims, demands, and attitudes. It is at this point that I find Rousseau’s insight relevant—​as long as people claim to have some rights or entitlements, or obligations and duties, or things like that, and their claims gain some traction in their social vicinity, power-​structuring rules would have a function to serve, regulating, as it were, those claims and demands. And this remains the case whether the presumed normative claims are justified or not. In other words, the basic function of power-​structuring rules is to allocate decision-​ making powers, establishing practical authority, as it were, determining who gets to decide certain things for others. Basic power-​structuring rules are the means by which brute power is transformed or molded into a normative framework. In order to get a better sense of what is at stake here, I want to contrast the view I strive to defend here with Hart’s account of secondary rules at the foundations of law. My point is that basic power-​structuring rules do not have to be “secondary rules,” in Hart’s terminology.27 They do not have to be rules about rules—​they can be rules about a wide range of claims or demands people make, and attitudes about normative relations they exhibit, whether those are embedded in social rules or not. Hart’s theory of the rules of recognition at the foundations of law certainly gave many readers the impression that he had assumed that there is necessarily a sequential development, taking place over time, from primary rules of conduct, that is, social rules directly regulating conduct by way of imposing obligations or granting rights, to the emergence of secondary rules, rules that are about the rules of conduct, determining ways of identifying them conclusively, granting powers to create new rules or modifying existing one, powers to apply rules in various changing circumstances, etc.28 And this would be in line with the thought that power-​structuring rules must operate on preexisting primary rules of conduct, sequentially, as it were. First we need to have rules determining rights and obligations, and then we can introduce powers to change them. I am not entirely sure that this is what Hart really meant, but if he did, he was wrong about it.



27 Hart, The Concept of Law, 79.

28 Hart, The Concept of Law, ch. 5.

Power-Structuring Rules  145 Before we proceed, let me clarify the relation, as I see it, between what Hart called secondary rules and power-​structuring rules—​they are not identical. Secondary rules are rules about rules. They are the kind of rules that take other rules as their object of regulation. Most of them, however, are directly or indirectly power structuring in the sense we used the idea of power here. Secondary rules tend to be rules about ways of enacting new rules, modifying existing ones, creating offices for adjudicating conflicts and disagreements about the application of rules, and establishing ways of monitoring compliance and reactions to noncompliance. In other words, most secondary rules would also be power-​structuring rules—​they would determine ways in which agents can introduce changes to the normative situation of others. Part of what I want to show here, however, is that basic power-​structuring rules do not have to be secondary rules—​they do not have to take other rules as objects of regulation. In a famous passage in The Concept of Law, Hart asks us to envision a society where there are only primary rules of conduct—​namely, rules imposing obligations or conferring claim rights, and things like that. Such a society, Hart claims, would sooner or later face serious problems of uncertainty: “if doubts arise as to what the rules are or as to the precise scope of some given rule, there will be no procedure for settling this doubt . . . This defect in the simple structure of primary rules we may call its uncertainty.”29 And then, a few pages later, Hart introduces the rule of recognition as the remedy: “The simplest form of remedy for the uncertainty of the regime of primary rules is the introduction of what we shall call a ‘rule of recognition,’ ” which is “a very simple form of secondary rule: a rule for conclusive identification of the primary rules of obligation.”30 Hart then further identifies other secondary rules, granting powers of changing rules and granting powers of adjudicating disputes and conflicts about the rules’ application.31 It is, of course, Hart’s celebrated contribution to legal philosophy that the rules of recognition are at the foundations of legality. As he put it, “in the simple operation of identifying a given rule as possessing the required feature of being an item on an authoritative list of rules we have the germ of the idea of legal validity.”32 But the question of what constitutes legality is not what I want to focus on here; my interest is in Hart’s diagnosis, seeing

29 Hart, The Concept of Law, 90.

30 Hart, The Concept of Law, 92.

31 Hart, The Concept of Law, 93–​94.

32 Hart, The Concept of Law, 93.

146  Foundations of Institutional Reality power-​structuring rules as a remedy for uncertainty. I think that Hart was wrong about that, and in ways that are instructive. To begin with, Hart himself never quite explains why uncertainty about primary rules of conduct is such a major concern. At points he nods toward accepting the view, widely held by other legal theorists, that the use of organized coercion makes it important for people to have some certainty about the rules that are going to be forcefully enforced.33 At other times, he rejects views that forge too tight a connection between law and coercion or brute power. Either way, I think we have already noted the reasons, in previous chapters, to be skeptical about this preoccupation with certainty and clear boundaries of social rules and institutional facts. Generally speaking, it is not practically necessary to have precise boundaries; rough and vague boundaries will do just as well. We have already noted, in Chapter 4, that full compliance with constitutive rules of a practice is not required for the practice to exist, even to thrive. Breaking a constitutive rule does not necessarily amount to ceasing engagement in the practice. In fact, most of us most of the time follow constitutive rules of social practices somewhat partially, even improvising at times as we go along. It is only a small step from here to recognize that a certain amount of vagueness or uncertainty about the rules that constitute a practice is easily tolerated. There is no reason to assume that the boundaries of legality are different: some vagueness about them would not undermine the main functions of law and legal institutions in society.34 Be this as it may, it is time to recall the distinction between two types of social practices I mentioned in Chapter 1: hierarchical practices, which include secondary rules, and practices that are flat, lacking rule-​governed mechanisms for enacting new rules, changing existing ones, or monitoring and responding to noncompliance. Examples of hierarchical practice would include the law, of course, many organized religions with elaborate institutional settings, corporate entities, universities, and sports organizations and

33 E.g., Austin, The Province of Jurisprudence Determined; and Kelsen, General Theory of Law and State. 34 The disagreements that prevail in the United States—​for example, about various aspects of the Supreme Court’s power of determining constitutional matters—​is a good case in point. The extent of these powers and how the court is to exercise them is controversial. And similar controversies in the margins of basic power-​structuring rules are ubiquitous. Dworkin (Law’s Empire, ch. 1) argued that these disagreements and indeterminacies about the rules of recognition are very pervasive, so much so that it is doubtful that rules of recognition actually exist. Elsewhere I have argued that Dworkin took this way too far, confusing, as it were, borderline cases with definite extension (Marmor, Philosophy of Law). But I think that few jurists would doubt that there are some borderline cases and indeterminacies about the boundaries of legality.

Power-Structuring Rules  147 leagues. Examples of flat practices, lacking secondary rules, include natural language, practices of etiquette, and perhaps leisurely games played by amateurs. Now take a flat practice, like the use of natural language. Our use of language is governed by rules of all kinds, but there are no power-​structuring rules included. There is no rule-​governed mechanism for changing the rules of language, enacting new rules, or responding to failures of compliance with the existing rules.35 The use of language, as a social practice, is of the flat type. And yet, uncertainty about the rules is not a major concern. There is no defect in our use of language due to uncertainty in need of a remedy by way of introducing secondary rules. And the same goes for practices of etiquette—​ the rules may be uncertain at times, but secondary rules have not emerged. Etiquette is a flat social practice. In short, Hart’s focus on the need for certainty is in the wrong direction. We need a fresh start here. The distinction between flat and hierarchical practices helps us see why basic power-​structuring rules are foundational, not parasitic on primary rules of conduct. In other words, certain types of social practices and institutions are hierarchical in their nature, ab initio, so to speak, which means that they are partly constituted by power-​structuring rules; without them there is no practice to speak of. In order to see this, we can ask why, generally, some practices are essentially hierarchical. After all, the existence of flat social practices, like practices of etiquette and countless other conventional practices, shows that hierarchical structure is not necessary. Some practices can thrive without any normative powers embedded in the practice. So what makes it the case that, in some practices but not in others, basic power-​structuring rules are essential? The answer is almost embarrassingly simple: given the complex nature of interactions between humans living in a community, and their mutual interdependence, many situations emerge whereby some people need to make decisions that are binding on others. When these circumstances are such that they tend to recur with some regularity, the need for a decision-​making rule becomes evident. Let me reiterate that I do not pretend to know how social rules actually emerge, in reality, but we can easily discern the basic human 35 One might wonder about various national institutes or academies, as they are often called, entrusted with official modifications of a natural language, as the famous Académie française or its pale counterpart in Israel, the Academy of the Hebrew Language. These institutions are usually marginal in their actual effects, and more symbolic than anything else; national sovereignty over language use can only go so far.

148  Foundations of Institutional Reality needs that different types of rules respond to. We can observe their essential functions. And the simple point here is that power-​structuring rules respond to the need for authority. They are needed whenever some agents need to make decisions for others. Think about Rousseau’s example of property. There is no idea of property, no property regime that we can envisage, without the idea that unilateral decisions by an agent bind others. If something is mine, I can decide to give it to you, or I can give it to you on some conditions. And even if every asset is held in common by a group, some decisions need to be made about the asset’s allowed uses. Even if nothing is owned by anyone in particular (if that is at all possible), one should be able to decide what to do with their body and mind, or the food they prepared for dinner. The nature of property is such that it requires authority; it requires some decision-​making powers. When those powers go beyond brute power, and get transformed somehow into a normative idea, typically embedded in social rules, then we get the idea of basic power-​structuring rules. Now, of course, property is just one of many examples; the complexity and interdependence of human interactions create many kinds of circumstances in which recurrent needs for practical authority, the need to have some agents decide for others, gets molded into social rules, and those rules will be of the basic power-​structuring kind. But then, of course, not every type of human interaction is such that it concerns the need for authority. In using language, for example, as in many other contexts of solving recurrent coordination problems, once a solution is salient or conventionally entrenched, there is no need for decision-​making powers. And the emergent social practices, in such cases, tend to remain flat—​they do not include power-​conferring rules.36 I hope that these considerations are sufficient to show that basic power-​ structuring rules are indeed basic, that they do not have to emerge as secondary rules regulating, as it were, an already existing set of rules of conduct. When the emergence of a social practice is closely tied to the need for authority, practically speaking, when the situation requires some people to make decisions for others, the emergent practice is bound to include

36 It is an interesting question I think why rules of etiquette, like table manners for instance, remained flat over the centuries. I venture to guess that the explanation would have something to do with their function in enabling and reinforcing social stratification; since their main function is symbolic, signaling that if you don’t follow the rules, you don’t belong here; there is no need for decision-​ making powers. The rules are not there to solve practical problems; they are there to create solidarity and distance (Elias, The Civilizing Process).

Power-Structuring Rules  149 power-​structuring rules, and thus the practice is essentially hierarchical. And now, circling back to Hart’s famous rules of recognition, we can see that the better understanding of such rules is to see them as the basic power-​ structuring rules of a legal order. These are the social rules that determine who gets to make the law around here, and how, who gets to apply it to particular cases and interpret it when need arises, who implements those decisions and how, and so on.37 But again, the law is not unique in this respect. Basic power-​structuring rules do not have to concern the law—​they are at the foundations of all hierarchical social practices. There are many practices and institutions created by basic power-​structuring social rules that have little to do with law, at least initially, when those practices and institutions emerge. The law in modern times tends to incorporate those institutions into a legal framework, taking over the regulation of institutions by setting some legal boundaries to the powers they deploy and by providing, or withholding, legal sanctions from various aspects of such institutions. Thus, looking at the institutional world around us these days, almost everything is legal, and nearly all institutions operate within some legal regulatory regime. But this is a matter of law’s expansive reach, not necessarily an indication of how practices and institutions emerge and come to be practiced.38 The main conclusion I hope to have established here is that basic power-​ structuring social rules are foundational building blocks of significant aspects of our social reality. Such rules ground the existence of hierarchical social practices. Practices and institutions are hierarchical in virtue of the power-​structuring rules that constitute their existence. Given the ubiquitous practical need for some authoritative relations in society, the recurrent needs to have some people make decisions binding on others, hierarchical practices are all around us, shaping a very significant portion of our everyday lives. I do not mean to suggest that this a good thing (or bad)—​surely some hierarchical practices and institutions are morally justified, while others are morally questionable, sometimes even deplorable. The transformation of brute power into a normative idea is a remarkable achievement of our species, but it is not necessarily moral progress, not in all cases, for sure. Oppression 37 This is the view about the rules of recognition I have supported in previous work. See Marmor, Social Conventions, ch. 7; and Philosophy of Law, ch. 2. 38 There are many social institutions that have evolved independently of any legal regime, and that later, sometimes much later, came to be incorporated into a legal-​regulatory scheme. Universities, established in the late Middle Ages, would be one such example; religious institutions would be another obvious example. Some institutions, however, such as limited liability corporations, which emerged in the early seventeenth century, are arguably legal creations from their start.

150  Foundations of Institutional Reality is sometimes worse, not better, if it is normatively structured. In any case, this book was not about the morality of social practices. On the contrary, I hope to have demonstrated, even if in a very limited sphere, that we can do metaphysics of sociality without moral judgments necessarily implicated or presupposed.

Bibliography Armstrong, D. M. A World of States of Affairs (Cambridge: Cambridge University Press, 1997). Atiq, Emad H. “There Are No Easy Counterexamples to Legal Anti-​positivism.” Journal of Ethics and Social Philosophy 17, no. 1 (2020): 1–​26. Audi, Paul. “Grounding: Toward a Theory of the In-​Virtue-​of Relation.” Journal of Philosophy 109, no. 12 (2012): 685–​711. Audi, Paul. “A Clarification and Defense of the Notion of Grounding.” In Correia and Schnieder, Metaphysical Grounding: Understanding the Structure of Reality. Austin, John. The Province of Jurisprudence Determined. 1832 (New York: Prometheus Books, 2000). Austin, J. L. How to Do Things with Words (Cambridge: Cambridge University Press, 1962). Bach, Kent, and Robert M. Harnish. “How Performatives Really Work.” Linguistics and Philosophy 15, no. 1 (1992): 93–​110. Bach, Kent, and Robert M. Harnish. Linguistic Communication and Speech Acts (Cambridge, MA: MIT Press, 1979). Baker, Gordon. “The Private Language Argument.” Language & Communication 18, no. 4 (1998): 325–​356. Baker Lynne Rudder. “The Ontology of Artifacts.” Philosophical Explorations 7, no. 2 (2004): 99–​111. Banaji, Mahzarin, and Anthony G. Greenwald. Blindspot: Hidden Biases of Good People (New York: Delacorte Press, 2013). Bennett, Karen. “Construction Area (No Hard Hat Required).” Philosophical Studies 154, no. 1 (2011): 79–​104. Bennett, Karen. Making Things Up (Oxford: Oxford University Press, 2017). Bicchieri, Cristina. The Grammar of Society: The Nature and Dynamics of Social Norms (Cambridge: Cambridge University Press, 2006). Bratman Michael E. Faces of Intention: Selected Essays on Intention and Agency (Cambridge: Cambridge University Press, 1999.) Bratman Michael E. Shared Agency: A Planning Theory of Acting Together (Oxford: Oxford University Press, 2014). Burazin Luka, Kenneth Einar Himma, and Corrado Roversi, eds. Law as an Artifact (Oxford: Oxford University Press, 2018). Burge, Tyler. “On Knowledge and Convention.” Philosophical Review 84, no. 2 (1975): 249–​255. Correia, Fabrice, and Benjamin Schnieder. Metaphysical Grounding: Understanding the Structure of Reality, edited by Correia and Schnieder (Cambridge: Cambridge University Press, 2012). Daly, Chris. “Scepticism about Grounding.” In Correia and Schnieder, Metaphysical Grounding, 81100. Danto, Arthur. After the End of Art (Princeton, NJ: Princeton University Press, 1997).

152 Bibliography Danto, Arthur. “The Artworld.” The Journal of Philosophy 61 (1964): 571–​584. Dickie, George. Art and the Aesthetic, An Institutional Analysis (Ithaca, NY: Cornell University Press, 1974). Dipert, Randall R. Artifacts, Art Works, and Agency (Philadelphia: Temple University Press, 1993). Dorr, Cian. “To Be F Is to Be G.” Philosophical Perspectives 30, no. 1 (2016): 39–​134. Dworkin, Ronald M. Justice in Robes (Cambridge, MA: Harvard University Press, 2006). Dworkin, Ronald M. Law’s Empire (Cambridge: Harvard University Press, 1986). Dworkin, Ronald M. “Objectivity and Truth: You’d Better Believe It.” Philosophy & Public Affairs 25, no. 2 (1996): 87–​139. Dworkin, Ronald M. Taking Rights Seriously (London: Duckworth, 1977). Dummett, Michael A. E. The Interpretation of Frege’s Philosophy (London: Duckworth, 1981). Elias, Norbert. The Civilizing Process. 1939 (Hoboken, NJ: Blackwell, 1978). Enoch, David. Taking Morality Seriously: A Defense of Robust Realism (Oxford: Oxford University Press, 2011). Epstein Brian. The Ant Trap: Rebuilding the Foundations of the Social Sciences (Oxford: Oxford University Press, 2015). Fine, Kit. “Guide to Ground.” In Correia and Schnieder, Metaphysical Grounding, 37–​80. Fine, Kit. “Ontological Dependence.” Proceedings of the Aristotelian Society 95 (1995): 269–​290. Fine, Kit. “The Question of Realism.” Philosopher’s Imprint 1, no 1 (2001): 1–​30. Finnis, John. Natural Law and Natural Rights (Oxford: Oxford University Press, 1980). Finnis, John. “Natural Law Theory: Its Past and Its Present.” In The Routledge Companion to Philosophy of Law, edited by Andrei Marmor, 16–​30 (New York: Routledge, 2012). Frege, Gottlob. “The Thought: A Logical Inquiry.” Mind 65, no. 259 (1956): 289–​311. Originally published in Beiträge zur Philosophie des Deutschen Idealismus, 1918–​1919. Gilbert, Margaret. On Social Facts (Princeton, NJ: Princeton University Press, 1989). Graeber David, and David Wengrow. The Dawn of Everything: A New History of Humanity (New York: Farrar, Straus and Giroux, 2021). Guala, Francesco, and Frank Hindriks. “A Unified Social Ontology.” Philosophical Quarterly 65 (2015): 177. Hart, H. L. A. The Concept of Law (Oxford: Oxford University Press, 1961). Hart, H. L. A. The Concept of Law. 2nd ed. Edited by Penelope A. Bulloch and Joseph Raz (Oxford: Oxford University Press, 1994). Hart, H. L. A. Essays on Bentham: Jurisprudence and Political Theory (Oxford University Press, 1982). Hilpinen, Risto. “Artifact.” Stanford Encyclopedia of Philosophy. Edited by Edward N. Zalta (Spring 1999). Hilpinen, Risto. “Authors and Artifacts.” Proceedings of the Aristotelian Society 93 (1993): 155–​178. Hindriks, Frank. “Constitutive Rules, Language, and Ontology.” Erkenntnis 71, no. 2 (2009): 253–​275. Hindriks Frank, and F. Guala. “A Unified Social Ontology.” Philosophical Quarterly 65 (2015): 177. Hindriks Frank, and F. Guala. “The Functions of Institutions: Etiology and Teleology.” Synthese 198, no. 3 (2021): 2027–​2043.

Bibliography  153 Hofweber, Thomas. “Ambitious, Yet Modest, Metaphysics.” In Metaphysics: New Essays on the Foundations of Ontology, edited by David Chalmers, David Manley, and Ryan Wasserman, 260–​289 (Oxford: Oxford University Press, 2009). Hohfeld, Wesley Newcomb. Fundamental Legal Conceptions. Edited by W. W. Cook (New Haven, CT: Yale University Press, 1919). Houkes, Wybo, and Pieter Vermaas. “Actions Versus Functions: A Plea for an Alternative Metaphysics of Artifacts.” Monist 87, no. 1 (2004): 52–​71. Hurd, Heidi M. “The Moral Magic of Consent.” Legal Theory 2, no. 2 (1996): 121–​146. Kelsen, Hans. General Theory of Law and State. 1945. Translated by Anders Wedberg (New York: Russell and Russell, 1961). Kelsen, Hans. Introduction to the Problems of Legal Theory. 1934. Translated by Bonnie Litschewski Paulson and Stanley L. Paulson (Oxford: Oxford University Press, 2002). Kolodny, Niko, and R. Jay Wallace. “Promises and Practices Revisited.” Philosophy & Public Affairs 31, no. 2 (2003): 119–​154. Kornblith Hilary. “How to Refer to Artifacts.” In Margolis and Laurence, Creations of the Mind: Theories of Artifacts and Their Representation, 138–​149. Koslicki, Kathrin. “Varieties of Ontological Dependence.” In Correia and Schnieder, Metaphysical Grounding, 186–​213. Kripke, Saul. Naming and Necessity (Oxford: Blackwell, 1980). Kripke, Saul. Wittgenstein on Rules and Private Language (Oxford: Blackwell, 1982). Levinson, Jerrold. “Artworks as Artifacts.” In Margolis and Laurence, Creations of the Mind: Theories of Artifacts and Their Representation, 74–​84. Lewis, David. Convention: A Philosophical Study (Oxford: Blackwell, 1969). Lewis, David. “Truth in Fiction.” In Philosophical Papers, vol. 1, 261 (Oxford: Oxford University Press, 1983). List, Christian, and Philip Pettit. Group Agency: The Possibility, Design, and Status of Corporate Agents (Oxford: Oxford University Press, 2011). MacKinnon, Catherine A. Feminism Unmodified (Cambridge, MA: Harvard University Press, 1987). Margolis, Eric, and Stephen Laurence, eds. Creations of the Mind: Theories of Artifacts and Their Representation (Oxford: Oxford University Press, 2007). Marmor, Andrei. “An Institutional Conception of Authority.” Philosophy & Public Affairs 39, no. 33 (2011): 238–​261. Marmor, Andrei. Interpretation and Legal Theory. 2nd ed. (Oxford: Hart Publishing, 2005). Marmor, Andrei. The Language of Law (Oxford: Oxford University Press, 2014). Marmor, Andrei. “Law as Authoritative Fiction.” Law & Philosophy 37, no. 5 (2018): 473–​497. Marmor, Andrei. “Norms, Reasons and the Law.” In Unpacking Normativity: Conceptual, Normative, and Descriptive Issues, edited by Kenneth Einar Himma, Miodrag Jovanovic, and Bojan Spaic, 95–​118 (Oxford: Hart Publishing, 2018). Marmor, Andrei. Philosophy of Law (Princeton, NJ: Princeton University Press, 2011). Marmor, Andrei. Social Conventions: From Language to Law (Princeton, NJ: Princeton University Press, 2009). Marmor, Andrei. “Soft Law, Authoritative Advice and Non-​binding Agreements.” Oxford Journal of Legal Studies 39, no. 3 (2019): 507–​525. Marx, Karl. A Contribution to a Critique of Political Economy. Berlin: Verlag von Franz Dunder, 1859. McDaniel, Kris. The Fragmentation of Being (Oxford: Oxford University Press, 2017).

154 Bibliography Millikan, Ruth Garrett. “Language Conventions Made Simple.” Journal of Philosophy 95, no. 4 (1998): 161–​180. Millikan, Ruth Garrett. Language, Thought, and Other Biological Categories (Cambridge, MA: MIT Press, 1984). Payne, Keith, Laura Niemi, and John M. Doris. “How to Think about Implicit Bias.” Scientific American. March 27, 2018. https://​www.sci​enti​fica​meri​can.com/​arti​cle/​how-​ to-​think-​about-​impli​cit-​bias. Polger, Thomas W., and Lawrence Shapiro. The Multiple Realization Book (Oxford: Oxford University Press, 2016). Preston, Beth. “Why Is a Wing Like a Spoon? A Pluralist Theory of Function.” Journal of Philosophy 95, no. 5 (1998): 215–​254. Putnam, Hilary. “The Meaning of Meaning.” In Mind, Language and Reality, vol. 2 (Cambridge, 1975), chap. 12: 215–​271. Putnam, Hilary. “The Nature of Mental States.” In Mind, Language and Reality, 2 (Cambridge, 1975), chap. 21. Raz, Joseph. “Can There Be a Theory of Law?” In Between Authority and Interpretation, 17–​46 (Oxford: Oxford University Press, 2009). Raz, Joseph. Ethics in the Public Domain (Oxford: Oxford University Press, 1994). Raz, Joseph. “Normative Powers.” In The Roots of Normativity, edited by Ulrike Heuer, 162–​178 (Oxford: Oxford University Press, 2022). Raz, Joseph. Practical Reasons and Norms. 1975 (Princeton, NJ: Princeton University Press, 1990). Rosen, Gideon. “Metaphysical Dependence: Grounding and Reduction.” In Modality: Metaphysics, Logic, and Epistemology, edited by Bob Hale and Aviv Hoffman, 109–​136 (Oxford: Oxford University Press, 2010). Rosen, Gideon. “Real Definition.” Analytic Philosophy 56, no. 3 (2015): 189–​209. Rousseau, Jean-​Jacques. Discourse on the Origin of Inequality. 1775. Translated by Donald A. Cress (Cambridge, MA: Hackett Publishing, 1992). Ruben, David-​Hillel. “John Searle’s The Construction of Social Reality.” Philosophy and Phenomenological Research 57, no. 2 (1997): 443–​447. Ruben, David-​Hillel. The Metaphysics of the Social World (Oxford: Routledge & Kegan Paul, 1985). Ryle, Gilbert. Dilemmas (Cambridge: Cambridge University Press, 1954). Salmon, Nathan. “Two Conceptions of Semantics.” In Semantics versus Pragmatics, edited by Zoltan Gendler Szabo, 317–​328 (Oxford: Oxford University Press, 2005). Schaffer, Jonathan. “Grounding, Transitivity, and Contrastivity.” In Correia and Schnieder, Metaphysical Grounding, 122–​138. Schaffer, Jonathan. “On What Grounds What.” In Metametaphysics: New Essays on the Foundations of Ontology, edited by David Chalmers, David Manley, and Ryan Wasserman, 347–​383 (Oxford: Oxford University Press, 2009). Searle, John R. The Construction of Social Reality (New York: The Free Press, 1995). Searle, John R. “How Performatives Work.” Linguistics and Philosophy 12, no. 5 (1989): 535–​558. Searle, John R. “Social Ontology and the Philosophy of Society.” In Margolis and Laurence, Creations of the Mind: Theories of Artifacts and Their Representation, 3–​17. Searle, John R. Speech Acts (Cambridge: Cambridge University Press, 1969). Sider, Theodore. “Ground Grounded.” Philosophical Studies (forthcoming). Sider, Theodore. Writing the Book of the World (Oxford: Oxford University Press, 2011).

Bibliography  155 Soames, Scott. “Drawing the Line between Meaning and Implicature—​and Relating Both to Assertion.” Noûs 42, no. 3 (2008): 440–​465. Sperber, Dan. “Seedless Grapes: Nature and Culture.” In Margolis and Laurence, Creations of the Mind: Theories of Artifacts and Their Representation, 124–​137. Tadros, Victor. “Appropriate Normative Powers.” Aristotelian Society Supplementary 94, no. 1 (2020): 301–​326. Tahko, Tuomas E., and E. Jonathan Lowe. “Ontological Dependence.” Stanford Encyclopedia of Philosophy (Fall 2020). https://​plato.stanf​ord.edu/​archi​ves/​fall2​020/​entr​ies/​dep​ende​ nce-​onto​logi​cal. Thomasson, Amie. “Artifacts and Human Concepts.” In Margolis and Laurence, Creations of the Mind: Theories of Artifacts and Their Representation, 52–​73. Thomasson, Amie. Fiction and Metaphysics (Cambridge: Cambridge University Press, 1999). Tuomela, Raimo. The Philosophy of Sociality: The Shared Point of View (Oxford: Oxford University Press, 2007). Valentini, Laura. “Normative Powers.” Unpublished manuscript. Warnock, Geoffrey J. The Object of Morality (London: Methuen, 1971). Williamson, Timothy. Knowledge and Its Limits (Oxford: Oxford University Press, 2000). Williamson, Timothy. Vagueness (London: Routledge, 1994). Wittgenstein, Ludwig. Philosophical Investigations. 1953. Translated by G. E. M. Anscombe (Oxford: Blackwell, 1958). Wright, Larry. “Functions.” Philosophical Review 82, no. 2 (1973): 139–​168.

Index For the benefit of digital users, indexed terms that span two pages (e.g., 52–​53) may, on occasion, appear on only one of those pages. Aristotle on artifacts, 85–​86 on metaphysical inquiry, 23–​25 metaphysics and grounding, 30–​31 artifacts authorship, 87–​88 changes in function, 88–​90, 89n.13 as composites of other artifacts, 90 constitutive properties of, 88–​89 content of, 91–​92 intangible artifacts, 90–​91 laws as, 93–​94 massive artifacts, 87 ontology of objects, 96–​97 as products, 86–​87, 86n.6 social construction and, 95 social rules as, 94–​95 tangible vs. intangible, 98–​99 artifacts and limits of error, 21–​22, 84–​ 85, 104 artifacts, characteristics of, 85–​92 epistemic privilege, 84–​85, 84n.1 group-​wide errors, 95–​104 situational properties, 92–​95 ascription, functional, 73–​75 asymmetry, and grounding, 26 Audi, Paul grounding, 25n.3 grounding relations, 29, 33n.18, 35n.22, 36n.26 reduction, 37n.30 worldly conception of facts, 25, 28 Austin, John, 114–​16 authorship, and artifacts, 87–​88 behavior, regularities of, 9–​10 biological artifacts, 85–​86

Bratman, Michael E., 5–​6, 45–​46, 45n.7, 45n.10, 90, 90n.15 circularity, and the conduct element, 52–​53 collective intentionality, 13–​14, 15–​ 18, 41–​51 Hart’s Conditions-​H, 41 individual and group attitudes, 41–​ 43, 46 individual and group identities, 43–​ 45, 46–​47 individual shared intentions, 47–​49, 51 supervenience, 49 communication, and normative powers, 135–​37 conduct, and pattern of social rules, 11–​13 conduct element, and grounding social rules, 52–​57 vagueness, 54–​57, 60 conformity, and the conduct element, 53–​55 constitution by rules, 21–​22, 62–​63 constitutive function, 67–​77 constitutive rules, 63–​67 constitutive rules as partial grounds, 78–​83 proscriptive content of constitutive rules, 64–​65 social practices, differing perspectives on, 79–​83 constitutive function, and function as use, 67–​77 added value of function, 76–​77 functional assignment, 68–​71, 73–​75 functions as observer relative or extrinsic to nature, 70–​76, 71n.18

158 Index constitutive function, and function as use (cont.) and grounding, 67–​68 practical function, 71–​73 purposeful function, 74–​76 constitutive rules, vs. regulative rules, 62–​67 putting rules into practice, 65–​66 stand-​alone constitutive rules, 66 systematic nature of constitutive rules, 66–​67 constitutive vs. regulative rules, 64n.6 definition, real, 34–​36 disposition, and pattern of social rules, 11–​13 Dorr, Cian, 33n.18, 35n.23, 36n.26 Dworkin, Ronald M., 118, 120–​23, 127–​28 eliminativism, 50 epistemic privilege, and artifacts, 84–​ 85, 84n.1 epistemic privilege, and grounding social rules, 57–​61 group-​wide errors, 57–​60 Epstein, Brian, 4n.4, 14n.23, 44n.4, 64n.6 Fine, Kit, 4n.5, 25n.3, 28n.7, 30n.10 flat social practices, 7–​8, 146–​47 function added value of, 76–​77 functional ascription, 73–​75 functional assignment, 68–​71 observer relative or extrinsic to nature, 70–​76 purposeful function, 74–​76 function as use, 67–​77 added value of function, 76–​77 functional assignment, 68–​71, 73–​75 functions as observer relative or extrinsic to nature, 70–​76, 71n.18 and grounding, 67–​68 practical function, 71–​73 purposeful function, 74–​76 Gilbert, Margaret, 5–​6, 47n.14, 53n.19 Graeber, David, 141n.23

grounding complete vs. partial grounds, 27–​ 29, 29n.9 and constitutive function, 67–​68 constitutive rules as partial grounds, 78–​83 grounded facts, 29–​33 and hierarchical structure, 30–​ 31, 30n.10 and reduction, 20–​21 vs. identity relations, 33 grounding, social rules, 21, 40–​41 circularity, and the conduct element, 52–​53 and collective intentions, 41–​51 conduct element, 52–​57 conformity, and the conduct element, 53–​55 epistemic privilege, 57–​61 vagueness, and epistemic privilege, 60 vagueness, and the conduct element, 54–​57 group-​wide errors, and epistemic privilege, 57–​60 group-​wide errors, artifacts and, 95–​104 Hart, H. L. A., x–​xi, 4–​5, 8 grounding facts of legality, 115–​16 on institutional facts, 10 practice theory of rules, 11–​20 secondary rules and foundations of law, 144–​46 Hart’s Conditions-​H, 41 hierarchical social practices, 7–​8, 146–​47 hierarchical structure, grounding and, 30–​31, 30n.10 Hilpenin, Risto, 85n.3, 87–​88 Hindriks, Frank, 63n.4, 64n.6, 64n.7, 80n.29 human agency, artifacts as products of, 86–​87 identity relations and reduction, 36–​37 vs. grounding, 33 institutional facts, and normative powers, 130–​39

Index  159 institutional facts, and social facts, 1–​11 natural rules vs. social rules, 8–​10 shared agency, 5–​6 social practices, 6–​8 social solidarity, 6 institutional facts, metaphysics of, 62–​63 institutions, examples of, ix irreflexivity, and grounding, 26 judicial reasoning, and rationalization of social practices, 117–​22 Kelsen, Hans, 108n.2, 115–​16, 116n.13, 146n.33 Kornblith, Hilary, 99 languages, as artifacts, 86, 86n.5 law, as social institution, xi law, philosophy of, xi and metaphysical inquiry, 23–​25 laws, as artifacts, 93–​94 legality, reductive account of, 115–​17 legal positivism, 113–​17 Lewis, David conventions, 56n.21, 97n.25 rules and agreement, 10n.18 social rules, emergence of, 138–​39, 139n.20, 143n.26 social rules and agreement, 56 truth in fiction, 100n.33 List, Christian, 49–​50, 49n.15, 50n.17 methodological individualism, 11–​12 Millikan, Ruth Garrett, 53n.19, 73n.23 moral facts, and normative powers, 135 museums, as artifacts, 93, 93n.18 natural language semantics of, 18 as social practice, 7–​8 natural rules vs. social rules, 8–​10 naturefacts, 85–​86 normative powers, and institutional facts, 130–​39 objects intrinsic vs. extrinsic properties of, 92

ontology of, 96–​97 ontological dependence, 1–​4 Pettit, Philip, 49–​50, 49n.15, 50n.17 philosophy of law, xi and metaphysical inquiry, 23–​25 positivism, and philosophy of law, xii power-​structuring rules, 22, 129–​30 basic social rules, 139–​50, 141n.23 and normative powers, 130–​39 secondary rules and foundations of law, 144–​45 practices, rationalizing, 22 Preston, Beth, 72n.21, 73n.23 products, artifacts as, 86–​87, 86n.6 purposeful function, 74–​76 Putnam, Hilary, 49n.15, 71n.18, 96n.23, 99 Quine, Willard Van Orman on metaphysical inquiry, 23–​25 rationalization, of social practices, 105–​6 argument from interpretation, 122–​28 argument from judicial reasoning, 117–​22 internal rationalization constraint, 106–​13 legality, reductive account of, 115–​17 legal positivism, 113–​17 Raz, Joseph, 133, 135 real definition, and reduction, 34–​36 reduction and contemporary metaphysics, 33–​39 forms of, 37–​39 grounding and, 20–​21 and identity relations, 36–​37 and real definition, 34–​36 and social ontology, 38–​39 regularities of behavior, 9–​10 regulative rules, vs. constitutive rules, 62–​ 67, 64n.6 proscriptive content of constitutive rules, 64–​65 stand-​alone constitutive rules, 66 systematic nature of constitutive rules, 66–​67 Rosen, Gideon, 34, 35–​36, 35n.22 Rousseau, Jean-​Jacques, 139–​40

160 Index rules, power-​structuring, 22 rules, practice theory of, 11–​20 collective intentionality, 13–​14, 15–​18 and methodological individualism, 11–​12 social rules and rules of conduct, 19 structuralism, 14–​15 Schaffer, Jonathan, 23n.1, 25n.3, 27n.6, 30–​32, 45n.9 Searle, John, x–​xi on collective intentionality, 43–​45 constitutive vs. regulative rules, 62–​ 67, 64n.6 on institutional facts, 10 shared agency, and social facts, 5–​6 situational properties, of artifacts, 92–​ 95, 125 social construction, and artifacts, 95 social conventions, errors concerning, 97 social facts, and institutional facts, 1–​11 shared agency, 5–​6 social solidarity, 6 social hierarchy, and social practices, 7–​8 sociality of animals, 3–​4 social practices definition and examples of, 6–​7 determining purpose of, 107–​8 differing perspectives on, 79–​83 examples of, ix external vs. internal rationalizing features, 107–​11 hidden functions of, 111–​13 hierarchical vs. flat, 7–​8, 146–​47 and normative powers, 134–​35 practical functions of, 110–​11 social practices, rationalization of, 105–​6 argument from interpretation, 122–​28

argument from judicial reasoning, 117–​22 internal rationalization constraint, 106–​13 legality, reductive account of, 115–​17 legal positivism, 113–​17 social rules as artifacts, 94–​95 ontology of, 10 patterns of, 11–​13 and power-​structuring rules, 139–​50 and rules of conduct, 19 vs. natural rules, 8–​10 social rules, grounding, 21, 40–​41 circularity, and the conduct element, 52–​53 and collective intentions, 41–​51 conduct element, 52–​57 conformity, and the conduct element, 53–​55 epistemic privilege, 57–​61 vagueness, and epistemic privilege, 60 vagueness, and the conduct element, 54–​57 social solidarity, 6 structuralism, 14–​15 supervenience, and individual vs. collective intention, 49 Thomasson, Amie, 84n.1, 99–​100, 100n.32, 100n.33 transitivity, and grounding, 26 vagueness, and the conduct element, 54–​ 57, 60 Wengrow, David, 141n.23 Williamson, Timothy, 60n.25, 66–​67, 103n.38