Review of “Understanding Institutions. The Science and Philosophy of Living Together”, Francesco Guala, Princeton University Press, 2016

The following is a (long) review of Francesco Guala’s recent book Understanding Institutions. The Science and Philosophy of Living Together (Princeton University Press, 2016).

Twenty years ago, John Searle published his influential account of the nature of institutions and institutional facts (Searle 1995). Searle’s book has been a focal point for philosophers and social scientists interested in social ontology and its claims and arguments continue to be hotly disputed today. Francesco Guala, a professor at the University of Milan and a philosopher with a strong interest in economics, has written a book that in many ways can be considered both as a legitimate successor but also a thoroughly-argued critique of Searle’s pioneering work. Understanding Institutions is a compact articulation of Guala’s thoughts about institutions and social ontology that he has developed in several publications in economic and philosophy journals. It is a legitimate successor to Searle’s book as all the central themes in social ontology that Searle discussed are also discussed by Guala. But it is also a strong critique of Searle’s general approach to social ontology: while the latter relies on an almost complete (and explicit) rejection of social sciences and their methods, Guala instead argues for a naturalistic approach to social ontology combining the insights of philosophers with the theoretical and empirical results of social sciences. Economics, and especially game theory, play a major role in this naturalistic endeavor.

The book is divided into two parts of six chapters each, with an “interlude” of two additional chapters. The first part presents and argues for an original “rules-in-equilibrium” account of institutions that Guala has recently developed in several articles, some of them co-authored with Frank Hindriks. Two classical accounts of institutions have indeed been traditionally endorsed in the literature. On the institutions-as-rules account, “institutions are the rules of the game in a society… the humanly devised constraints that shape human interactions” (North 1990, 3-4). Searle’s own account in terms of constitutive rules is a subspecies of the institutions-as-rules approach where institutional facts are regarded as being the products of the assignment of status function through performative utterances of the kind “this X counts as Y in circumstances C”. The institutions-as-equilibria account has been essentially endorsed by economists and game theorists. It identifies institutions to equilibria in games, especially in coordination games. In this perspective, institutions are best seen as devices solving the classical problem of multiple equilibria as they select one strategy profile over which the players’ beliefs and actions converge. Guala’s major claim in this part is that the relevant way to account for institutions calls for the merging of these two approaches. This is done through the key concept of correlated equilibrium: institutions are figured out as playing the role of “choreographers” coordinating the players’ choices on the basis of public (or semi-public) signals indicating to each player what she should do. Institutions then take the form of lists of indicative conditionals, i.e. statements of the form “if X, then Y”. Formally, institutions materialize as statistically correlated patterns of behavior with the equilibrium property that no one has an interest to unilaterally change her behavior.

The motivation for this new approach follows from the insufficiencies of the institutions-as-rules and institutions-as-equilibria accounts but also to answer fundamental issues regarding the nature of the social world. Regarding the former, it has been widely acknowledged that one the main defect of the institutions-as-rules is that it lacks a convincing account of the reason of why people are motivated in following rules. The institutions-as-equilibria approach for its part is unable to account for the specificity of human beings regarding their ability to reflect over the rules and the corresponding behavioral patterns that are implemented. Playing equilibria is far from being human specific, as evolutionary biologists have recognized long ago. However, being able to explain why one is following some rule or even to communicate through a language about the rules that are followed are capacities that only humans have. There are also strong reasons to think that the mental operations and intentional attitudes that sustain equilibrium play in human populations are far more complex than in any other animal population. Maybe the most striking result of this original account of institutions is that Searle’s well-known distinction between constitutive and regulative rules collapses. Indeed, building on a powerful argument made by Frank Hindriks (2009), Guala shows that Searle’s “this X counts as Y in C” formula reduces to a conjunction of “if X then Y” conditionals corresponding to regulative rules. “Money”, “property” or “marriage” are theoretical terms that are ultimately expressible through regulative rules.

The second part of the book explores the implications of the rules-in-equilibrium account of institutions for a set of related philosophical issues about reflexivity, realism and fallibilism in social ontology. This exploration is done after a useful two-chapter interlude where Guala successively discusses the topics of mindreading and collective intentionality. In these two chapters, Guala contends, following the pioneering work of David Lewis (1969), that the ability of institutions to solve coordination problems depends on the formation of iterated chains of mutual expectations of the kind “I believe that you believe that I believe…” and so on ad infinitum. It is suggested that the formation of such chains is generally the product of a simulation reasoning process where each player forms expectations about the behavior of others by simulating their reasoning, on the assumption that others are reasoning like her. In particular, following the work of Morton (2003), Guala suggests that coordination is often reached through “solution thinking”, i.e. a reasoning process where each player first asks which is the most obvious or natural way to tackle the problem and then assumes that others are reasoning toward the same conclusion than her. The second part provides a broad defense of realism and fallibilism in social ontology. Here, Guala’s target is no longer Searle as the latter also endorses realism (though Searle’s writings on this point are ambiguous and sometimes contradictory as Guala shows) but rather various forms of social constructionism. The latter hold that the social realm and the natural realm are intrinsically different because of a fundamental relation between how the social world works and how humans (and especially social scientists) reflect on how it works. Such a relationship is deemed to be unknown to the natural sciences and the natural world and therefore, the argument goes, “social kinds” considerably differ from natural kinds. The most extreme forms of social constructionism hold the view that we cannot be wrong about social kinds and objects as the latter are fully constituted by our mental attitudes about them.

The general problem tackled by Guala in this part is what he characterizes as the dependence between mental representations of social kinds and social kinds. The dependence can be both causal and constitutive. As Guala shows, the former is indeed a feature of the social world but is unproblematic in the rules-in-equilibrium account. Causal dependency merely reflects the fact that equilibrium selection is belief-dependent, i.e. when there are several equilibria, which one is selected depends on the players’ beliefs about which equilibrium will be selected. Constitutive dependency is a trickier issue. It assumes that an ontological dependence holds between a statement “Necessarily (X is K)” and a statement “We collectively accept that (X is K)”. For instance, on this view, a specific piece of paper (X) is money (K) if and only if it is collectively accepted that this is the case. It is then easy to see why we cannot be wrong about social kinds. Guala claims that constitutive dependence is false on the basis of a strong form of non-cognitivism that makes a radical distinction between folk classifications of social objects and what these objects are really doing in the social world: “Folk classificatory practices are in principle quite irrelevant. What matters is not what type of beliefs people have about a certain class of entities (the conditions they think the entities ought to satisfy to belong to that class) but what they do with them in the course of social interactions” (p. 170). Guala strengthens his point in the penultimate chapter building on semantic externalism, i.e. the view that meaning is not intrinsic but depends on how the world actually is. Externalism implies that the meaning of institutional terms is determined by people’s practices, not by their folk theories. An illustration of the implication of this view is given in the last two chapters through the case of the institution of marriage. Guala argues for a distinction between scientific considerations about what marriage is and normative considerations regarding what marriage should be.

Guala’s book is entertaining, stimulating and thought-provoking. Moreover, as it is targeted to a wide audience of social scientists and philosophers, it is written in plain language and devoid of unnecessary technicalities. Without doubt, it will quickly become a reference work for anyone believing that naturalism is the right way to approach social ontology. Given the span of the book (and is relatively short length – 222 pages in total), there are however many claims that would call for more extensive arguments to be completely convincing. Each chapter contains a useful “further readings” section that helps the interested reader to go further. Still, there are several points where I consider that Guala’s discussion should be qualified. I will briefly mention three of them. The first one concerns the very core of Guala’s “rules-in-equilibrium” account of institutions. As the author notes himself, the idea is not wholly new as it has been suggested several times in the literature. Guala’s contribution however resides in his handling of the conceptual view that institutions are both rules and equilibria with an underlying game-theoretic framework that has been explored and formalized by Herbert Gintis (2009) and even before by Peter Vanderschraaf (1995). Vanderschraaf has been the first to suggest that Lewis’ conventions should be formalized as correlated equilibria and Gintis has expanded this view to social norms. By departing from the institutions-as-equilibria account, Guala endorses a view of institutions that eschews the behaviorism that characterizes most of the game-theoretic literature on institutions, where the latter are simply conceived as behavioral patterns. The concept of correlated equilibrium indeed allows for a “thicker” view of institutions as sets of (regulative) rules having the form of indicative conditionals. I think however that this departure from behaviorism is insufficient as it fails to acknowledge the fact that institutions also rely on subjunctive (and not merely indicative) conditionals. Subjunctive conditionals are of the from “Were X, then Y” or “Had X, then Y” (in the latter case, they correspond to counterfactuals). The use of subjunctive conditionals to characterize institutions is not needed if rules are what Guala calls “observer-rules”, i.e. devices used by social scientists to describe an institutional practice. The reason is that if the institution is working properly, we will never observe behavior off-the-equilibrium path. But this is no longer true if rules are “agent-rules”, i.e. devices used by the players themselves to coordinate. In this case, the players must use (if only tacitly) counterfactual reasoning to form beliefs about what would happen in events that cannot happen at the equilibrium. This point is obscured by the twofold fact that Guala only considers simple normal-form games and does not explicitly formalize the epistemic models that underlie the correlated equilibria in the coordination games he discusses. However, as several game theorists have pointed out, we cannot avoid dealing with counterfactuals when we want to account for the way rational players are reasoning to achieve equilibrium outcomes, especially in dynamic games. Avner Greif’s (2006) discussion of the role of “cultural beliefs” in his influential work about the economic institutions of the Maghribi traders emphasizes the importance of counterfactuals in the working of institutions. Indeed, Greif shows that differences regarding the players’ beliefs at nodes that are off-the-equilibrium path may result in significantly different behavioral patterns.

A second, related point on which I would slightly amend Guala’s discussion concerns his argument about the unnecessity of public (i.e. self-evident) events in the generation of common beliefs (see his chapter 7 about mindreading). Here, Guala follows claims made by game theorists like Ken Binmore (2008) regarding the scarcity of such events and therefore that institutions cannot depend on their existence. Guala indeed argues that neither Morton’s “solution thinking” nor Lewis’ “symmetric reasoning” rely on the existence of this kind of event. I would qualify this claim for three reasons. First, if public events are defined as publicly observable events, then their role in the social world is an empirical issue that is far from being settled. Chwe (2001) has for instance argued for their importance in many societies, including modern ones. Arguably, modern technologies of communication make such events more common, if anything. Second, Guala rightly notes in his discussion of Lewis’ account of the generation of common beliefs (or common reason to believe) that common belief of some state of affairs or event R (where R is for instance any behavioral pattern) depends on a state of affairs or event P and on the fact that people are symmetric reasoners with respect to P. Guala suggests however that in Lewis’ account, P should be a public event. This is not quite right as it is merely sufficient for P to be two-order mutual belief (i.e. everyone believes P and everyone believes that everyone believes P). However, the fact that everyone is a symmetric reasoner with respect to P has to be commonly believed (Sillari 2008). The issue is thus what grounds this common belief. Finally, if knowledge and belief are set-theoretically defined, then for any common knowledge event R there must be a public event P. I would argue in this case that rather than characterizing public events in terms of observability, it is better to characterize them in terms of mutual accessibility, i.e. in a given society, there are events that everyone comes to know or believe even if she cannot directly observe them simply because they are assumed to be self-evident.

My last remark concerns Guala’s defense of realism and fallibilism about social kinds. I think that Guala is fundamentally right regarding the falsehood of constitutive dependence. However, his argument ultimately relies on a functionalist account of institutions: institutions are not what people take them to be but rather are defined by the functions they fulfill in general in human societies. To make sense of this claim, one should be able to distinguish between “type-institutions” and “token-institutions” and claim that the functions associated to the former can be fulfilled in several ways by the latter. Crucially, for any type-institution I, the historical forms taken by the various token-institutions I cannot serve as a basis to characterize what I is or should be. To argue for the contrary would condemn one to some form of traditionalism forbidding the evolution of an institution (think of same-sex marriage). The problem with this argument is that while it may be true that the way people represent a type-institution I at a given time and location through a token-institution I cannot define what I is, it remains to determine how the functions of I are to be established. Another way to state the problem is the following: while one (especially the social scientist) may legitimately identify I with a class of games it solves, thus determining its functions, it is not clear why we could not identify I with another (not necessarily mutually exclusive) class of games. Fallibilism about social kinds supposes that we can identify the functions of an institution but is this very identification not grounded on collective representations and acceptance? If this is the case, then some work remains to be done to fully establish realism and fallibilism about social kinds.


Binmore, Ken. 2008. “Do Conventions Need to Be Common Knowledge?” Topoi 27 (1–2): 17.

Chwe, Michael Suk-Young. 2013. Rational Ritual: Culture, Coordination, and Common Knowledge. Princeton University Press.

Gintis, Herbert. 2009. The Bounds of Reason: Game Theory and the Unification of the Behavioral Sciences. Princeton University Press.

Greif, Avner. 2006. Institutions and the Path to the Modern Economy: Lessons from Medieval Trade. Cambridge University Press.

Hindriks, Frank. 2009. “Constitutive Rules, Language, and Ontology.” Erkenntnis 71 (2): 253–75.

Lewis, David. 1969. Convention: A Philosophical Study. John Wiley & Sons.

Morton, Adam. 2003. The Importance of Being Understood: Folk Psychology as Ethics. Routledge.

North, Douglass C. 1990. Institutions, Institutional Change and Economic Performance. Cambridge University Press.

Searle, John R. 1995. The Construction of Social Reality. Simon and Schuster.

Sillari, Giacomo. 2008. “Common Knowledge and Convention.” Topoi 27 (1–2): 29–39.

Vanderschraaf, Peter. 1995. “Convention as Correlated Equilibrium.” Erkenntnis 42 (1): 65–87.

Rational Expectations and the Standard Model of Social Ontology

Noah Smith has an interesting post where he refers to an article of Charles Manski about the rational expectations hypothesis (REH). Manski points out that in a stochastic environment it is highly unlikely that expectations are rational in the sense of the REH. However, he ultimately concludes that there are no better alternative. In this post, I want to point out that the REH is actually well in line with what the philosopher Francesco Guala calls in an article the “Standard Model of Social Ontology” (SMOSO), including the fact that it lacks empirical support. This somehow echoes Noah Smith’s conclusion that “rational Expectations can’t be challenged on data grounds”.

Guala characterizes the SMOSO by the following  three elements:

1) Reflexivity: Guala defines this as the fact that “social entities are constituted by beliefs about beliefs” (p. 961). A more general way to characterize reflexivity is that individuals form attitudes (mainly, beliefs) about the systems they are part of and thus attitudes about others’ attitudes. If it is assumed that these attitudes determine people’s actions and in turn, these actions determine the state of the system, then people’s attitudes determine the system. This may lead to the widely discussed phenomenon of self-fulfilling prophecies where the agents’ beliefs about others’ beliefs about the (future) state of the system bring the system to that state.

2) Performativity: it can be defined as the fact that the social reality is literally made by the agents’ attitudes and actions. The classical example is language: performative utterances like “I promise that Y” or “I make you man and wife” not only describe the social reality, they (in the appropriate circumstances) make it by creating a state of affairs that makes the utterance true. Other cases are for instance the fact that some pieces of paper are collectively regarded as money or the fact that raising one’s hand is regarded as a vote in favor of some proposition or candidate.

3) Collective intentionality: attitudes (in particular beliefs) constitutive of the social reality are in some way or another “collective”. Depending on the specific model, collective intentionality can refer to a set of individual attitudes (intentions, beliefs) generally augmented by an epistemic condition (usually, mutual or common knowledge of these attitudes) or a distinct collective attitude of the form “we intend to” or “we believe that”.

The three elements constitutive of the SMOSO are common to almost all the theories and models developed in social ontology and the philosophy of social science for the last thirty years. That does not mean that they fully determine the content of these theories and models: there are several and mutually exclusive accounts of collective intentionality, as well as there are different ways to account for performativity and reflexivity. Now, I want to suggest that many economic models using the REH fall within this general model of social ontology. The REH states that economic agents do not make systematic errors in the prediction of the future value of relevant economic variables. In other words, they make correct predictions on average. Denote X(t) the value of any economic variable you want (price, inflation, …) at time t and X(t+1)^ei the expected value of X at time t+1 according to agent i. Formally, an expectation corresponds to X(t+1)^ei = E[X(t)ΙI(t)^i] with I(t)^i the information available at t for i and E the expectation operator. The REH is the assumption that X(t+1) = X(t+1)^ei + Eu where u is an error term of mean 0. The proximity of the REH with the three elements of the SMOSO is more or less difficult to see but is nevertheless real.

The relationship between the REH and reflexivity is the easiest to state because discussions on rational expectations in the 1950s find their roots in the treatment of the reflexivity issue which itself originates in Oskar Morgenstern’s discussion of the “Holmes-Moriarty paradox”. Morgenstern was concerned with the fact that if the state of affairs that realizes depends on one’s beliefs about others’ beliefs about which state of affairs will realize, then it may be impossible to predict states of affairs. In 1950s, papers by Simon and by Modigliani and Grunberg tackle this problem. Using fixed-point techniques, they show that under some conditions, there is at least one solution x* = F(x*) such that the prediction x* regarding the value of some variable is self-confirmed by the actual value F(x*). In his article on rational expectations, Muth mentions as one of the characteristic of the REH the fact that a public prediction in the sense of Grunberg and Modigliani “will have no substantial effect on the operation of the economic system (unless it is based on inside information)”. So, the point is that a “rational prediction” should not change the state of the system.

The relationship of the REH with performativity and collective intentionality is more difficult to characterize. Things are somewhat clearer however once we realize that the REH implies mutual consistency of the agents’ beliefs and actions (see this old post by economist Rajiv Sethi which makes this point clearly). This is due to the fact that in an economic system, the value X(t+1) of some economic variable at time t+1 will depend on the decisions si made by thousands of agents at t, i.e. X(t+1) = f(s1(t), s2(t), …, sn(t)). Assuming that these agents are rational (i.e. they maximize expected utility), the agent’s decisions depend on their conjectures X(t+1)^ei about the future value of the variable. But then this implies that one’s conjecture X(t+1)^ei is a conjecture about others’ decisions (s1(t), …, si-1(t), si+1(t), …, sn(t)) for any given functional relation f, and thus (assuming that rationality is common knowledge) a conjecture about others’ conjectures (X(t+1)^e1, …, X(t+1)^ei-1, X(t+1)^ei+1, …, X(t+1)^en). Since others’ conjectures are also conjectures about conjectures, we have an infinite chain of iterated conjectures about conjectures. Mutual consistency implies that everyone maximizes his utility given others’ behavior. In general, this will also imply that everyone forms the same, correct conjecture, which is identical to the REH in the special case where all agents have the same information since we have X(t+1) = X(t+1)^ei for all agent i. As Sethi indicates in his post, this is equivalent to what Robert Aumann called the “Harsanyi doctrine” or more simply, the common prior assumption: any disagreement between agents must come from difference in information.

In itself, the relationship between the REH and the common prior assumption is interesting. Notably, if we consider that the common prior assumption is difficult to defend on empirical grounds, this should lead us to consider the REH with suspicion. But it also helps to make the link with the SMOSO. Regarding performativity, we have to give up the assumption (standard in macroeconomics) that the equilibrium is unique, i.e. there are at least two values X(t+1)* and X(t+1)** for which the agents’ plans and conjectures are mutually consistent. Now, any public announcement of the kind “the variable will take value X(t+1)* (resp. X(t+1)**)” is self-confirming. Moreover, this is common knowledge.[1] The public announcement play the role of a “choreographer” (Herbert Gintis’ term) that coordinates the agents’ plans. This makes the link with collective intentionality. It is tempting to interpret the common prior assumption as some kind of “common mind hypothesis”, as if the economic agents were collectively sharing a worldview. Of course, as indicated above, it is also possible to adopt a less controversial interpretation by seeing this assumption as some kind of tacit agreement involving nothing but a set of individual attitudes. The way some macroeconomists defend the REH suggests a third interpretation: economic agents are able to learn about the economic world and this learning generates a common background. In game-theoretic terms, we could also say that agents are learning to play a Nash equilibrium (or a correlated equilibrium).

This last point is interesting when it is put in perspective with Guala’s critique of the SMOSO. Guala criticizes the SMOSO for its lack of empirical grounding. For instance, discussions about collective intentionality are typically conceptual, but almost never build on empirical evidence. Most critics of the REH in economics make a similar point: the REH is made for several reasons (essentially conceptual and theoretical) but has no empirical foundations. The case of learning is particularly interesting: since the 1970s, one of the “empirical” defenses of the REH has been the casual claim that “you can’t fool people systematically”. This is the same as to say that on a more or less short term, people learn how the economy works. This is a pretty weak defense, to say the least. Economists actually do not know how economic agents are learning, what is the rate of the learning process, and so on. Recently, a literature on learning and expectations has been developing, establishing for instance the conditions of convergence to rational expectations. As far as I can tell, this literature is essentially theoretical but is a first step to provide more solid foundations to REH… or to dismiss it. The problem of the empirical foundations for any assumption regarding how agents form expectations is likely to remain though.


[1] Going a little further, it can be shown that if the public announcement is made on the basis of a probabilistic distribution p where each equilibrium is announced with probability p(X(t+1)*), then p also defines a correlated equilibrium in the underlying game, i.e. agents behave as if they were playing a mixed strategy defined by p.

Frame Principles and the Grounding of Social Facts

ant trapI am currently reading Brian Epstein’s book The Ant Trap (Oxford University Press). Epstein is Assistant Professor of Philosophy at Tufts University and is a specialist of social ontology and philosophy of social science more generally. Though I do not like the subtitle at all (“Rebuilding the Foundations of the Social Sciences”), the book provides an interesting and stimulating attempt to build a metaphysical framework for studying the social world. Epstein is mainly interested in working out the metaphysical reasons that ground social facts, i.e. what is it that makes facts like “I have a 20$ bill in my pocket” or “Barack Obama is the President of the United States of America” possible. The book has two parts: the first one develops the metaphysical framework on the basis of a critique of the “standard model of social ontology”. The second part applies the framework to the specific topic of groups and what ground facts about groups. A recurring theme throughout the book is the critique of ontological individualism, i.e. the claim that only facts about individuals ground social facts, including facts about groups.

In this post, I will only discuss Epstein’s key concept of frame principles. Epstein offers this concept as an alternative to Searle’s constitutive rules and it is instructive to see if and how it avoids the problems I discuss in my preceding post. Epstein’s framework builds on a key distinction between anchoring and grounding. This distinction is not essential here but helps to better understand both the critique of ontological individualism and Epstein’s points about the nature of social facts. Grounding is a relation between two facts through what the author calls a frame principle: it states the conditions (the “metaphysical reasons”) for a fact to generate another (social) fact. For instance, through a given frame principle, the (physical) fact that I raise my hand at some time and some place grounds the (social) fact that I have voted for some candidate in an election. Anchoring is different: it is “a relation between a set of facts and a frame principle” (p. 82). An anchor is what is making a given frame principle to hold in some population. The nature of the anchor may vary depending one’s favorite model of social ontology. For instance, in Searle’s account of institutional facts the anchor is the collective acceptance or recognition of some constitutive rules. Epstein does not much discuss anchoring but argues convincingly against the “conjunctivist” (scholars who conflate grounding and anchoring) that the distinction is important because it is the only way to avoid falling in an infinite regress.

For the rest of this post, I will ignore issues related to anchoring. The grounding relation is more significant because Epstein suggests that it is an alternative to Searle’s account of constitutive rules (which, according to the author, are “neither constitutive nor are they rules” (p. 77)). As said above, the grounding relation is a relation between two facts or set of facts. More exactly, it is established through a frame principle that articulates a link between a grounding (set of) condition(s) X and a grounded fact of type Y. This gives the following formula for a frame principle (p. 76):

For any z, the fact “z is X” grounds the fact “z is Y”.

Consider for instance the following frame principle:

For all z, the fact “z is a bill printed by the Bureau of Printing and Engraving” grounds the fact “z is a dollar.

Given this frame principle, any fact of the type “this particular bill z* is printed by the bureau of Printing and Engraving” grounds the social fact “this particular bill z* is a dollar”. A frame principle can also be alternatively formulated in a semantic model of possible worlds: a frame is simply a set of possible worlds P where the grounding conditions for social facts are fixed. Denote w(z-X) and w(z-Y) as the propositions “In world w, the fact “z is X” holds” and “In world w, the fact “z is Y” holds” respectively. Then, if zX is the event “z is X” (i.e. the subset of possible worlds where the proposition w(z-X) is true), then zX ⊆ zY for all w ∈ P, with zY the event “z is Y”. In words, for any possible worlds where the frame holds, whenever z has property X, it also has property Y.

Epstein does not state clearly why his formulation his superior to Searle’s. One advantage is that it may help to make the distinction between grounding and anchoring more salient. In particular, it appears clearly that grounding is captured by a “possible worlds/unique frame” model, while anchoring corresponds a “unique world/possible frames” model. Another advantage is that it seems to rule out all the debates over the nature of rules. Epstein’s frame principles are not (necessarily) rules, so that to ask whether they are regulative or constitutive seems meaningless. Still, as far as I can see, the “linguistic argument” presented in the preceding post is still valid. The issues of the nature of (regulative) rules and of their differences with frame principles remain in Epstein’s framework. Does a regulative rule of the kind “In Britain, drive at the left side of the road” also counts as a frame principle? At first sight, it seems not. But the problem is that it is not difficult to reformulate any frame principle as a regulative rule along exactly the same lines that the one discussed in the preceding post. This is not surprising: Epstein’s frame principles are semantically identical to Searle’s constitutive rules (i.e. the underlying semantic model is the same). And any regulative rule can be captured by a similar semantic model[1]. So the question of the nature of the grounding relation is not completely answered. Epstein’s account suggests nevertheless a possible direction to look at: it is possible that the more or less “constitutive” nature of frame principles depends on the conditions of their anchoring. That is a possibility that could be worth to be explored.


[1] Consider a regulative rule “if z is X, then z is Y” (e.g. if Bill is under 18, then Bill cannot vote”). Now, using the same notation than above, in all possible worlds w ∈ P where the rule holds, zX ⊆ zY.

Are There Constitutive Rules?

The philosopher John Searle is well-known for his work in the philosophy of language and in the philosophy of mind (see in particular the “Chinese room” thought experiment). He has also made an important contribution to social ontology with his books The Construction of Social Reality (1995) and Making the Social World (2010). An important feature of Searle’s account of the nature of social reality is his distinction between constitutive and regulative rules. Actually, he already made this distinction in 1969 in his work on speech acts. Searle’s point is that some rules are straightforward statements of the kind “Do X” or “If Y, then X”. Other rules however are of the form “This X count as Y in (circumstances) C”. The former are regulative rules, the latter are constitutive rules. The key difference is that constitutive rules make some kinds of actions or facts possible while regulative rules only regulate a practice that is not logically tied to the rule.

Consider the following facts: “I have been checkmated”, “Bill hits a home run”, “I have a 20$ bill in my wallet”. All these facts depend on constitutive rules that define what counts as a checkmate, a home run or a 20$ bill. Without these rules, the above facts cannot exist. Now, contrast with the fact “In Britain, people drive at the left side of the road”. This fact only depends on a regulative rule (“if you’re in Britain, then drive at the left side of the road”); the very practice of driving does not seem to depend on the peculiar content of the rule.

The distinction has some intuitive appeal. It also seems significant because constitutive rules, unlike regulative rules, have the ability to create the institutional reality. This is reflected in the fact that the “count-as” locution generates what Searle calls status functions: a constitutive rule attributes to some entity X (which can be a person, an object or anything else) a status defined in terms of deontic powers. For instance, the rule “such and such pieces of papers count as dollars in the United States of America” gives these pieces of paper the power to buy things. Once the rule is collectively accepted in some community, pieces of paper with the appropriate characteristics actually have this property. So, constitutive rules generate institutional facts.

However, Searle’s account of constitutive rules has been widely criticized. The most significant critique has been that the distinction is a false one: all rules are both constitutive and regulative. Whether or not all regulative rules are also constitutive is a complex debate. In some ways, the rule “if in Britain, then drive at the left side of the road” is constitutive of the practice “to drive in Britain”: assume some possible world identical to the actual world except for the fact that people in Britain are driving at the right because the rule says so. Then, the practice “to drive in Britain” would not be the same. This involves a problem of identity on which I will briefly return at the end of this post. The reverse is easier to analyze: all constitutive rules can be reformulated as regulative rules. This point is forcefully made by Frank Hindriks in this paper (see also here). He shows that all constitutive rules actually consist in the conjunction of two propositions corresponding respectively to a “base rule” and a “status rule”. Consider for instance the case of a proto-institution we call property* which is defined by the constitutive rule “X[this piece of land l] counts as Y[property* of person p] in C[p was the first person to claim so and such and such other conditions obtain]”. The base rule states the conditions for the piece of land to be owned by person p:

Base Rule: if the set of conditions c obtain, then l is property* of p.

Suppose that the proto-institution of property* grants the right of exclusive use and nothing else. Then, the status rule is:

Status Rule: if l is property* of p, then p has the right of exclusive use of l.

It should be noted that both the base rule and the status rule are of the form “if… then”, i.e. they are regulative rules. Now, it is easy to demonstrate that Searle’s distinction between constitutive and regulative rules seems merely to be a linguistic one, rather than logical or ontological. Indeed, we can define the rule for property* by combining the base rule and the status rule:

Rule for Property*: if the set of conditions c obtain, then p has the right of exclusive use of l.

Once again, the statement of the rule is of the form “if… then”. Moreover, the rule is stated without any reference to the institution of property* itself. It seems that we have reduced a constitutive rule to a regulative rule and thus that constitutive rules have nothing specific. They are merely a linguistic artifact.

One may think that this is only due to Searle’s specific account of the distinction and that it may be possible to defend it in some other way. For instance, in his paper “Two Concepts of Rules”, John Rawls seems to offer an alternative way to account for constitutive rules. Rawls distinguishes between what he calls the “summary conception” and the “practice conception” of rules. The former defines a rule as a mere behavioral pattern generated by acts persons have made because of their efficiency. According to the latter, a rule defines a practice in the sense that the practice consists in following the rule. For instance, “hitting a home run” or more generally the practice of “playing baseball” consist precisely in the fact of following some set of rules.

However, the same problem remains, as shown by David Lewis in his article “Scorekeeping in a Language Game” (note that Lewis does not make reference to Rawls’ article). Consider any well-run baseball game G (either a professional game or an informal game between friends). At any stage t of G, there is a score S which is defined by the septuple of numbers < rv, rh, h, i, s, b, o > with rv and rh the number of runs of the visiting team and the home team respectively, h the half (first or second) in the inning, i the inning, s the number of strikes, b the number of balls and o the number of outs. According to Lewis, a codification of the rules of baseball would consist in the conjunction of four kinds of rules:

1) Rules specifying the evolution of S: if S(t) is the score at stage t, and if between t and t’ the players behave in a manner m, then the score S(t’) is determined in a certain way by both S(t) and m.

2) Specifications of correct play: for any score S(t) and any other stage t’, there is a set M of manners to behave which corresponds to correct play.

3) Directives concerning correct play: throughout G (i.e. for all tt’ sequences), players ought to adopt manners to behave belonging to M.

4) Directives concerning scores: players have behave such as their teams score the maximum runs and the opposing teams the minimum runs.

Lewis notes that sets of rules 1) and 2) correspond to constitutive rules, while sets of rules 3) and 4) rather correspond to regulative rules. Consider in particulars rules about the evolution of score S. That these rules cannot be seen as a mere summary of past behaviors is reflected by the fact that the evolution of score is both a function of how the players behave m and of the current score S(t), i.e. S(t’) = f(S(t), m). The function f encompasses a set of constitutive rules regarding, for instance, what counts as a strike. The way the players behave seems not sufficient as such to make the score evolve.

But this is clearly a linguistic artifact, again. As Lewis states, “[o]nce score and correct play are defined in terms of the players’ behavior, then we may eliminate the define terms in the directive concerning requiring play and the directing concerning scores”. In other words, constitutive rules can be reformulated in regulative rules, which themselves can be stated as summary of (past) behaviors. The implication seems to be that there is no social reality beyond the actions of persons: institutions are reducible to individual behavior, and there is nothing more to the social reality.

There may be several ways to avoid this conclusion however. A first possibility is suggested by Hindriks in the article I have linked to: the fact that the distinction between constitutive and regulative rules is linguistic does not mean that it is ontologically and scientifically irrelevant. The debates over reductionism in science give a great illustration. In principle, all facts about the economy (“interest rates are rising”, “growth is slowing”, “Amazon is losing money”, “Bill buys a car”) can be described without any economic terms and concepts. It could be possible to describe all these facts as facts about the movement of atoms and molecules. But not only this would be extremely complicated, this would also be unhelpful to explain and to predict economic phenomena. The distinction between constitutive and regulative rules can thus be grounded not on ontology, but rather on a pragmatic account of scientific explanation and prediction.

Another alternative consists in acknowledging that the distinction is not clear-cut. It may well be that all rules are both constitutive and regulative. But our attitudes toward rules may vary. In a given community and at a given time, it may be a fact that a rule is regarded as being constitutive of some practice or institution, while others are not. Compare for instance a proposal to create a four-points basket in basketball with another one making tackles permissible. Both rules would change the nature of the game, but while the former would probably not be rejected on the ground that “this is not basketball”, the latter probably would. The point is that rules are constitutive not per se, but through (or because of) our practices and attitudes toward them. Behind these attitudes and practices, lies the difficult issue of identity: what is it to an institution to be this institution and nothing else according to some community? This is a subject for another post.

Ullmann-Margalit’s Game-Theoretic Account of Social Norms


Game theory is now a fairly standard tool in the study of social norms and institutions. Among the pioneering game-theoretic accounts of norms figures Edna Ullmann-Margalit’s book The Emergence of Norms. Originally published in 1977, it has been recently reedited by Oxford University Press. This offers an opportunity to rediscover an interesting study which anticipates several developments in the analysis of social norms.

The title of the book seems to indicate that Ullmann-Margalit is interested in the way norms have appeared and evolved, i.e. what today we would rather call the evolution of norms. Her approach is “structural” rather than “historical”: types of norms and their emergence are related to their functions in different kinds of strategic interactions which differ in their properties. This gives Ulmann-Margalit’s account a strong functionalist stance which she explicitly recognizes: the existence of norms is explained by their functions. I will return at the end of this post on this point but it can be already noted that this makes the title of the book slightly misleading. Ullmann-Margalit’s account is far more convincing if viewed as an account of the way norms work (i.e. determine individual’s behaviors) than an account of the mechanisms by which norms evolve.

The book distinguishes between three kinds of strategic interactions with specific features which give rise to three kinds of social norms: prisoner’s dilemma (PD) norms, coordination norms and norms of partiality. Ullmann-Margalit’s analysis of coordination norms partially builds on Schelling’s and Lewis’ game-theoretic accounts. Coordination norms are defined as solutions to coordination problems, i.e. interactions where the players’ interests are perfectly aligned and where at least two profiles of actions fully satisfy them. Like Schelling and Lewis, she points out that salience is the most general way through which a recurrent coordination problem is solved. Once a solution has been singled out by the participants, the repetition of the interactions gives rise to a social norm. For novel coordination problems however, or at least some of them, she argues that it is most likely that the solution comes from a norm dictated and enforced by some external authority. Moreover, contrary to Schelling and Lewis, Ullmann-Margalit suggests that coordination norms are “real” norms and not simply conventions. That means that coordination norms have a normative force: social pressure and/or moral obligation rather than simply convergent expectations contribute to explain why people conform to some specific norm.

As their name suggests, PD norms are solutions to prisoner’s dilemma type of interactions. Basically, PD norms help to foster cooperation in strategic interactions where defection is the dominant strategy. Ullmann-Margalit’s account can be seen as one of the numerous attempts made by social scientists and philosophers to show that it can be rational to cooperate in a PD. However, contrary to other philosophers (for instance, David Gauthier’s theory of constrained maximization), Ullmann-Margalit does not attempt to argue that playing dominated strategies is rational. Rather, she suggests that PD norms foster cooperation on the basis of several “payoff-transforming” mechanisms depending on external sanctions and/or moral commitments. The payoff-matrix then no longer corresponds to a PD but to a game where mutual cooperation is (possibly the only) an equilibrium. On this point, there is some similarity between Ullmann-Margalit’s account and the theory of social norms developed recently by Christina Bicchieri. Bicchieri’s suggests that social norms rely on a conditional preference for conformity which in some cases a PD into a coordination game.

Ullmann-Margalit’s study of the third kind of norms – norms of partiality, is the most original and intriguing. Norms of partiality stabilize situations where some parties are favored and other disfavored. More exactly, they legitimize a status quo of inequality. The analysis builds on an interesting (though not totally convincing) distinction between equilibrium and strategic stability. The former corresponds to the standard Nash equilibrium solution concept and follows from the fact that each player is rationally searching for improving his absolute position. An equilibrium is simply a state of affairs where no player can improve his absolute position by changing his behavior. Strategic stability matters as soon as we assume that the players are also concerned by their relative position (assuming of course that this concern is not already incorporated into the payoff matrix). In a situation of inequality, a disfavored party may seek to reduce the inequality level even if this leads to a worsening of her absolute position. The threat of such a move becomes credible once it is realized that in some cases, the favored party’s rational response to this move leads to an improvement in both the disfavored party’s relative and absolute position. In this case, a state of affairs may be an equilibrium but still be strategically unstable: one may want to change his behavior even though it worsens temporarily his absolute position. According to Ullmann-Margalit, norms of partiality’s function is to stabilize state of affairs which are strategically unstable through they correspond to a game-theoretic equilibrium. The matrix below corresponds to the paradigmatic illustration:


Assume that R1-C1 is the status quo. Though it is an equilibrium, it is not strategically stable since the column player may try to convince the row player that it will play C2 to improve his relative position. Then, row player’s best response will be to switch to R2, thus leading to R2-C2 with a reversal of fortune for the two parties. Norms of partiality prevent such kinds of strategic move by generating some kind of normatively binding constraint.

Though ingenious, this functional account of the role of norms in stabilizing situations of inequality is not totally convincing because of a lack of concrete examples (which the author herself points out). As noted by Cass Sunstein in his review of Ullmann Margalit’s book, the examples cited in the latter (property rights, rights inheritance) are not enforced by norms but rather by law (at least in developed countries). In other cases, obvious situations of inequality (for instance between men and women regarding wages) seem to lack any broad normative or moral support but still endure thanks to other social mechanisms. It is not clear therefore whether norms of partiality really have an empirical counterpart.

This leads me to the last point of this post. As I briefly note above, Ullmann-Margalit’s study is not really an account of the emergence of norms. Such an account would propose one or several causal mechanisms for the creation and the evolution of norms. Since the 1980’s, evolutionary game-theoretic accounts of norms have been developed. They remain largely unconvincing however because the emergence problem is fundamentally an empirical one. At the very first line of her book, Ullmann-Margalit states clearly that her essay belongs to “speculative sociology” (nowadays, we would rather call it “social ontology”). She claims that she is intending to propose a “rational reconstruction” of norms rather than an historical account. By this, she means that her goal is to provide a list of reasons or features that may explain why norms exist. As noted above, her approach is functionalist because she tries to relate the existence of norms to the functions they fulfill. But as argued by Jon Elster, functional explanations are no explanation at all; only causal explanations are. The title thus wrongly suggests that the book provides a causal explanation (either theoretical or historical) of the emergence of norms, which is not the case. However, Ullmann-Margalit’s work is highly valuable if it is taken to give an account of the way norms are working, i.e. how they actually affect people’s behavior. Because she insists on the functional stance of her approach, Ullmann-Margalit does not make this point clearly enough. However, for each kind of norms, one or several mechanisms are suggested to explain why people cooperate or succeed in coordinating: commitment (strategic or moral), framing effects, social pressure, and so on are all hinted as explanations for the working of norms. From this point of view, The Emergence of Norms anticipates many contemporary developments in social ontology, game theory and experimental economics and for this reason remains a valuable read.