John R. Searle presented his first sketchy but influential analysis of institutions in the first book that he published in 1969, Speech Acts. Inspired by Elizabeth Anscombe's distinction between brute and institutional facts on the one hand and John Rawl's distinction between rules we use for convenience and rules that constitute certain practices, he maintained that institutions are systems of constitutive rules. Such rules, he proposed, facilitate the creation of institutional facts, for example the fact that a certain utterance counts as a promise. Searle developed his ideas about social relations and behaviours further in his 1995 book The Construction of Social Reality. There is a lot of continuity between this book and Searle's latest book. So the question arises why anyone who is familiar with the former would want to read the latter.

I can think of three reasons. (i) Searle presents his book as a more general version of the social ontology he offered in his 1995 work. He claims that this is because ‘at that time [he] did not see the centrality of Status Function Declarations in both creating and maintaining institutional facts’ (p. 19). (ii) Searle develops his social ontology in more detail. Over the years he has received many questions and criticisms. Many of the developments, adjustments and refinements are responses to those. (iii) They might, however, also have special relevance to social scientists. Although anthropologists, sociologists and economists have commented on his work and have regarded some of his ideas as useful for the development of their theories, some social scientists have also expressed doubts about the practical relevance of Searle's social philosophy.

The best way to tackle the third issue is by comparing social scientific theories of institutions with Searle's theory, and determining whether anything valuable can be learned along the way. In this review I provide some pointers as to how one might go about doing this. I shall start by asking whether Searle's theory of institutions is consistent with the idea that institutions are unintended consequences of individual actions. This idea seems to be challenged by Searle's claim that they depend on status function declarations. I argue that, if minor modifications are made, the idea can be salvaged (1).

Subsequently I discuss two kinds of social structures that Searle does not address: conventions and norms. In relation to conventions, I ask how institutions, which are inherently normative, might evolve from conventions, which need not be normative. As it turns out, conventions do tend to evolve into normative social structures, and can plausibly be regarded as precursors of institutions (2). Investigating the relation between institutions and social norms provides an opportunity to revisit Searle's distinction between regulative and constitutive rules. This distinction is by and large absent from social scientific theories, and it will be useful to determine whether anything of importance is lost. Although the distinction is important, a theory of institutions that appeals only to regulative rules need not be inadequate, or so I argue (3). Finally I turn to organizations, a kind of institution that has puzzled Searle to such an extent that he has come to subscribe to a fictionalist theory of it – this in spite of the fact that he is a staunch realist in other departments. I relate his view to the economic theory of the firm as a nexus of contracts, and argue that there is no need to go fictionalist, neither for Searle nor for those who support the nexus theory (4).

(1) The ‘main theoretical innovation’ in Searle's latest book is the notion of a status function declaration (11). A status function declaration is a declaration of something acquiring or possessing a certain status function – for instance, of two people becoming or being married. Searle maintains that all of institutional reality is created and maintained by status function declarations (or by representations that have the same logical form; 13). All status functions involve deontic powers such as rights and obligations. Marital status, for instance, comes with informal rights and obligations as well as formal powers (think of income tax deductions). At bottom, a status function declaration comes down to the declaration that the relevant people have the deontic powers involved in a particular status function (the notion of a function seems to do no work in the theory, in spite of Searle's claim to the contrary; 95n2).

Now consider Searle's claim that institutions depend on status function declarations. How might this conflict with the view that institutions are the unintended consequences of individual actions? One way in which a conflict could arise is this: status function declarations are actions that are intended to create or maintain institutional status functions. Another is that on Searle's view those status functions do not only depend on the agent who has performed the declaration, but also on the other individuals involved in the institution. Let me begin by considering this second claim. The idea is that a status function requires collective acceptance or, as he now also says, collective recognition. Collective acceptance or recognition is the collective intentional state involved in the creation and maintenance of institutions. In contrast, many social scientists and perhaps most economists hold that institutions depend only on individual attitudes such as individual preferences and expectations. This suggests there is a conflict here.

Searle (Reference Searle, Cohen, Morgan and Pollack1990, Reference Searle1995) has argued that collective intentionality cannot be reduced to individual intentionality. The content of a collective intention to perform action A is ‘We intend to A’. A social fact exists when two individuals share a collective intention with the same content. Due to the fact that its content refers to a collective, a collective intention cannot be reduced to an individual intention. This view clearly conflicts with what is perhaps the dominant view in economics. However, Searle has changed his view in a way that resolves at least most of that conflict. He now allows for two kinds of collective intentionality, one of which is reducible to individual intentionality plus mutual beliefs (58). This is the kind of collective intentionality that is needed for institutions. Searle does not discuss common knowledge. Many economists regard mutual belief as too weak for institutions and require common knowledge instead. However, if this is the residual disagreement between them, the distance between what Searle claims about the attitudes involved in institutions and the views economists typically have on this is substantially smaller as compared with the position Searle defended in The Construction of Social Reality.

Even if Searle's talk of collective acceptance and recognition does not conflict with individualist theories of institutions, the role that status function declarations play in his theory might stand in the way of embracing the claim that institutions are unintended consequences of individual actions. Status function declarations will typically, if not always, be intentional actions. These declarations pertain to status functions, which means that they concern deontic powers. So someone who performs a status function declaration intentionally institutes deontic powers. Does this leave any room for institutions as unintended consequences?

Although Searle (Reference Searle1983) denies this, an act can be intentional without it being intended. Imagine yourself walking through a park. You follow others in taking a shortcut through the grass. Suppose you realize that you thereby contribute to the damage inflicted on the grass, and think to yourself that you should not be doing this. However, all you really care about is getting to your destination as soon as possible. It is plausible to say in this situation that you damage the grass intentionally even though you do not intend to damage it (Harman Reference Harman1976, Knobe Reference Knobe2003, Reference Knobe2010, Hindriks Reference Hindriks2008a). The shortcut can develop into a collectively accepted path and people may acquire the informal or even formal right to walk there. Searle's use of the term ‘status function declaration’ appears to be very permissive. The core issue he tries to get at is that people can have deontic powers due to the fact that this is collectively accepted. Against the background of this idea, it may be so permissive as to accommodate your contributing to the evolution of the path by damaging the grass (which makes me wonder why Searle puts so much emphasis on declarations). Given what was said above, you intentionally contribute to the creation of the path without intending to do so (Hindriks Reference Hindriks2009b). The path, then, is the unintended consequence of your and other people's actions. The consequence is unintended even if all people were like you and were fully aware of it. So if we let go of Searle's claim that intentional action is always intended action, the idea that a particular institutional fact can be the unintended consequence of individual actions can be salvaged at least for some cases.

Suppose we grant that a particular institutional fact is created on purpose. Does that mean that the relevant institution is also intended? Not necessarily. When someone participates in an institution, her concerns will usually be with the institutional fact that is at issue and not with the institution as such. A bride and groom need not care about the institution of marriage as such. All they need to care about are their wedding vows. Although it may well be that some participants, for example the priest or the civil servant, cares about the marriage as such, this still does not entail that any of the people involved intends to contribute to the existence of marriage as an institution. In this example, the institutional fact is an intended consequence of the actions of the agents involved, but the institution need not be. The upshot is that Searle's framework allows for institutions as unintended consequences after all.

(2) On Searle's current view, any institution involves deontic powers. Institutions are in effect created and maintained by declarations according to which certain people have particular deontic powers. This suggests that institutions are inherently normative. Many economic models of institutions do not recognize the (alleged) normative nature of institutions. Those that do, usually cash it out in terms of normative expectations. Perhaps Searle's current conception of what is involved in the collective recognition of an obligation comes close to a collection of individuals having certain normative expectations, i.e. having beliefs that some people in certain situations ought to perform particular actions.

It is not obvious that the normativity of institutions can be captured in this way. Margaret Gilbert (Reference Gilbert1989) and Raimo Tuomela (Reference Tuomela1995) have, for instance, argued that behavioural structures such as institutions require joint or collective commitment. The idea is that deontic powers need a stronger normative basis that is grounded not in a loosely connected collection of individuals but in the collective as a whole. The collective commitment is what binds the individuals together. Accepting the idea that an institution depends on collective commitment of that institution, however, does not sit comfortably with the idea that institutions are the unintended consequences of individual actions.

Perhaps institutions proper can indeed not be the unintended consequences of individual actions. It could still be, however, that behavioural structures closely analogous to institutions are unintended. Unfortunately, Searle only discusses institutions. Furthermore, he focuses on how institutions are made and hardly pays any attention to how they might evolve. But institutions can originate with non-normative behavioural structures that are plausibly conceived of as the unintended consequences of individual actions. Conventions might fit the bill. Conventions are usually taken to be relatively arbitrary patterns in behaviour that are not inherently normative (Lewis Reference Lewis1969). As such, there is no reason to question the idea that they (or many of them) are the unintended consequences of individual actions. And conventions might be the precursors of institutions.

It is in fact often maintained that conventions tend to evolve into social norms (ibid.). Francesco Guala (Reference Guala2010) has recently put this claim to the test. He has people play a coordination game for nine rounds. During these rounds they tend to converge on a common strategy and thereby on a solution of the game that can plausibly be seen as a convention. In the tenth round a temptation is introduced for one of the players. As it turns out, few people give in to this temptation and most cooperate also in this tenth round of the game. By comparing the results with those of a one-shot game in which no convention is present, Guala is in a position to conclude that ‘conventions acquire normative power’ (Reference Guala2010, 755). All this suggests that conventions start out as non-normative social structures that subsequently acquire a normative dimension. They evolve into social norms. And it is a small step from norms to institutions, or so it seems.

(3) But how do institutions relate to social norms? This is where Searle's distinction between regulative and constitutive rules comes in. Searle maintains that institutions are systems of constitutive rules. This means that their logical structure is: ‘X counts as Y in context C’. This schema is the counts-as locution, and is closely related to the notion of a status function declaration. Status function declarations often pertain to particular instantiations of status functions (e.g. the marriage of John and Mary). Constitutive rules, however, are general and always pertain to institutions as such (e.g. any two people who have exchanged wedding vows in an appropriate ceremony count as being married). Searle (now) characterizes constitutive rules as ‘standing declaratives’ (97). A constitutive rule pertains to actions that cannot be performed independently of that rule (think of marriage licence laws).

Regulative rules concern actions that can be performed independently from those rules. Traffic rules and the rules of etiquette are examples, as making a right turn and using a napkin are possible independently of these rules. In light of this characterization and these examples, social norms are plausibly conceived of as regulative rules (e.g. it is permissible to walk across the lawn here). This creates a problem for the idea that social norms can evolve into institutions. After all, it implies that these two kinds of behavioural structures are based on different kinds of rules. And Searle maintains that the distinction between regulative rules and constitutive rules is a strict one. This suggests that, just as it is impossible to transform a regulative rule into a constitutive rule, it is impossible for a social norm to be transformed into an institution.

I think, however, that such a transformation can be achieved. The crucial step that has to be made is that a term for a status function needs to be introduced. But it is not so difficult to come up with such a term. In a way Searle discusses the process himself when he imagines a leader being appointed on an ad hoc basis. This might be a matter of accepting the rule ‘If Rex tells you to do something, you do what he says’. Now suppose the practice of leadership in the relevant community develops into a monarchy. Then Rex turns into a king with the power to issue orders to all his servants. And some procedure such as hereditary succession to determine the conditions under which someone is a king. At this point, the community has accepted a constitutive rule that specifies the conditions under which someone has the status function of a monarch (anyone who stands in a suitable hereditary relation to Rex is our monarch). And being a monarch comes with certain deontic powers (which on my view are to be explicated in terms of status rules, which explicate among others that you are supposed to obey the monarch; Hindriks Reference Hindriks, Schmid, Schulte-Ostermann and Psarros2008b, Reference Hindriks2009a). The upshot is that a regulative rule has been transformed into a constitutive rule. This line of argument suggests that Searle's analysis of institutions might not be so far removed from theories of institutions in the social sciences as one might have thought. The main points of contention are the formulation of the rules and the alleged constitutive role of language (Hindriks Reference Hindriks2009a, Reference Hindriks, Jarvie and Zamora-Bonilla2011).

(4) Searle encounters a problem when he applies his social ontology to organizations. At first, the case of organizations seems to be fairly straightforward. As we have seen, Searle maintains that institutions consist of status functions that involve deontic powers. If organizations are status functions, they involve deontic powers. It is not obvious, however, who possesses these deontic powers. In other words, the question is: What is the X to which status functions and deontic powers are ascribed?

Searle discusses the creation of a limited liability corporation. He maintains that there is no pre-existing object that is turned into a corporation. As evidence, Searle refers to the fact that Californian law states that forming a corporation is a matter of executing and filing articles of incorporation. This all too brief description of how limited liability corporations are created does indeed not involve an object on which a status is imposed. However, the law he cites reads as follows: ‘One or more natural persons, partnerships, associations or corporations, domestic or foreign, may form a corporation under this division by executing and filing articles of incorporation.’ (97) So prior to the existence of a corporation that is instituted are one or more natural persons, partnerships, associations or other corporations. On the natural assumption that partnerships and associations consist of people, this suggests that corporations are statuses imposed on one or more people. Searle accepts that president is a status function that is imposed on a person. This raises the question why he would not simply maintain that this holds for corporations as well. Instead, he claims that corporations are created ‘so to speak out of thin air’ (98).

Searle (now) comes close to accepting the claim mentioned. He maintains that the whole point of declaring the existence of a corporation is ‘to create a rather elaborate set of power relationships between actual people; indeed, the corporation consists of such relationships’ (98). Corporations ‘always bottom out in actual human beings’ (108). He adds to this, however, that a corporation ‘is just a placeholder for a set of actual power relationships among actual people’ (22). Furthermore, in contrast to ‘president’ Searle regards ‘corporation’ as a so-called ‘freestanding Y term’, that is a Y term that is instantiated without there being an independently existing X. In light of this he claims that corporations are fictitious entities (100).

Perhaps the key to this puzzle lies in the fact that Searle regards the president as in some sense identical to the person who occupies the office (102), whereas it makes little sense to identify a corporation with the people who constitute it. As Searle recognizes, a corporation remains in existence when new people come to occupy its key positions (98). If it were identical to one or more persons, it would change whenever there would be a change in the people involved in it.

The idea that Searle regards constitution as identity is supported by the fact that he objects to identifying corporations to their members (20). If so, the problem lies in Searle's conception of constitution as identity. As something is always identical to itself, identity is a symmetric relation. The salient alternative is to regard constitution as an asymmetric relation instead. This is very natural for the following reason: it makes a lot of sense that a person that meets condition X constitutes an entity with status function Y, whereas the reverse is nonsense. An advantage of such an asymmetric conception of constitution is that it can easily accommodate the fact that a status function can remain operative when the object that constitutes it changes in some way. Applied to corporations, this means that the following two claims can be true at the same time: (1) Corporations can survive changes in their constituency; (2) Corporations are constituted by collections of individuals.

How is this relevant to economists? The problem at issue is intimately related to the theory of the firm as a nexus of contracts, which plays an important role in the economics of organization. On this theory, firms are fictions. Firms do not exist. All there is to a firm is a collection of contracts (Jensen and Meckling Reference Jensen and Meckling1976). Those contracts define certain relations between people, and thereby they provide boundary conditions for the transactions that can or should take place between them.

But economists could turn this theory on its head. Rather than denying their existence, they could just say that firms are real because they are nexuses of contracts. This problem is similar to the problem I identified in relation to what Searle says about corporations. I argued against Searle that, as long as one adopts a suitable account of constitution, corporations turn out to be legitimate social entities that are as real as money and presidents. Economists who subscribe to the nexus theory can maintain that firms come into existence when certain statuses such as that of a limited liability corporation are imposed on particular people.

Accepting Searle's social ontology means that they can hold on to their conception of firms as nexuses of contracts. It does not, however, compel them to say that firms are fictions. Instead, they could say that firms are real and owe their existence to the constitutive rules people recognize. Those rules specify the conditions that have to be met in order for the deontic powers involved in firms to be instantiated. In the case of limited liability corporations these are legal rules, which means that the deontic powers are clearly specified in the law. These powers are in one way or another executed by the persons who are constitutive of these corporations (or by people who they have authorized to do so). This implies that the boundaries of such firms are clearly delineated. Things may be different for other kinds of firms. The principle, however, is the same.

To sum up, I have drawn four conclusions all of which bear on the relevance of Searle's social ontology for the social sciences.(1) In spite of appearances, Searle's social ontology is consistent with the idea that institutions are the unintended consequences of individual actions. (2) On Searle's view, institutions are inherently normative. It is plausible that (some) institutions originate with non-normative conventions that evolve into social norms. (3) Searle, however, cannot accept this, because he takes institutions to be systems of constitutive rules and social norms are to be explicated in terms of regulative rules. This problem dissolves once it is appreciated that, pace Searle, the distinction between regulative and constitutive rules is not categorical but (primarily) linguistic. (4) Searle regards organizations as fictitious entities. They are, however, better thought of as status functions constituted by collections of individuals. On this conception, they are as real as money and presidents. I found this exploration of the most recent version of Searle's social ontology rewarding. To the extent that the reader shares this experience, Searle's latest book is more than recommendable.