1. GROUP AGENTS

People who interact do not necessarily do something together. You might get up in a movie theatre to let me pass. We end up watching the same movie. But watching the movie together requires more than this. We need to have watching the movie as our joint goal. Suppose this is our joint goal. Each of us might be committed to this goal for private reasons. If so, we do something together in what Tuomela (Reference Tuomela and Schmitt2002) calls the ‘I-mode’. In contrast, those who act in the ‘we-mode’ are collectively committed to the relevant goal (70). Part of being collectively committed to a goal is to think as a group member. This means that an individual member derives the reasons from which she acts from the intentions that the group agent forms. The intention to paint the clubhouse, for instance, may be based on the fact that the old paint has worn out. In such a situation, the members may intend to do their part simply for the reason that the group agent has formed this intention (37–40).

Tuomela argues that we-mode collective intentionality cannot be reduced to I-mode collective intentionality. This thesis of conceptual collectivism can be illustrated in terms of the notion of a group reason. Given that members derive their participatory intentions from the intention of the group agent, the level of the group agent is in an important sense prior to that of an individual member (101–115). This idea is strengthened by the fact that, as Tuomela has it, the level of the group agent also has normative authority in that each member ought to accept its goals or its ethos (27, 115–120). We-mode reasons can come apart from I-mode reasons, as the former do not even supervene on the latter (93).

According to Tuomela, the we-mode is causally efficacious. Strategic or I-mode reasoning does not always suffice to settle on the best outcome, whereas intuitively people converge on it without much effort. Tuomela argues that this is because in such situations people engage in we-mode or group reasoning that serves to restrict the range of feasible action alternatives. In the we-mode people do not ask: What should I do? Instead, they ask: What should we do? The least beneficial of two mutually beneficial outcomes can then be excluded from the start, even though it is an equilibrium for I-mode groups. Thus, there is a ‘functional difference’ between the modes: the we-mode can explain how people coordinate in collective action dilemmas such as the Hi-Lo game whereas the I-mode cannot. Tuomela concludes that group agents cannot be reduced to I-mode social groups because, in contrast to I-mode reasoning, we-mode reasoning successfully reduces the number of equilbria in a range of strategic interactions.

Tuomela defends a number of other claims which suggest that, in addition to conceptual collectivism, he embraces ontological collectivism. He argues that group members entify the groups to which they belong (3) and thereby construct them as group agents (22). Furthermore, he maintains that their psychology is autonomous from the private psychologies of their members and in this sense they enjoy autonomy (22; see also List and Pettit Reference List and Pettit2011). Not only do group agents exhibit a mental unity, they can also be reflectively self-conscious (4). Furthermore, he recognizes joint causal powers (26), maintains that groups are systems that as such have the power to act (51), and embraces the idea that group agency involves emergent features (22, 48, 52, 91). Tuomela concludes that group agents have what he calls ‘functional existence’ (2–3, 237).

In spite of this, Tuomela combines his conceptual collectivism with ontological individualism. The main reason for this is that he believes that ‘the only causally initiating agentive motors are the individual agents, and hence the agency of group agents must ontologically bottom out in the behavior of its members’ (22; see also 13, 15, and 93). He provides further support for his ontological individualism by commenting on the intentionality of group agents. First, he points out that people project intentional properties onto group agents, which can thereby seem real. Such attributions, however, give rise only to ‘intentional existence’ (2–3). As a consequence, it is ‘not literally true’ that group agents have intentional states (47). Instead, group agents are mind-dependent, ‘do not exist as fully intentional agents’ and are ‘partly fictitious’ (47, 46).

Tuomela goes on to argue that the intentionality that group agents exhibit is extrinsic rather than intrinsic (47). Intrinsic intentionality is internal to someone's mind – think of the intentionality of beliefs and desires. By contrast, intentionality that someone attributes to something – think of the intentionality of words and signs – is extrinsic intentionality (Searle Reference Searle1983). Only biological organisms exhibit intrinsic intentionality. According to Tuomela, the intentionality of group agents is extrinsic, because it depends on attributions people make, in particular those made by group members (3, 23). Hence, group agents only exhibit extrinsic intentionality and do not possess intrinsic intentionality.

At this point the question arises whether Tuomela's views about group agents are consistent. Many collectivists accept the claim that groups ‘can act only through their members’ activities’ (13). They deny, however, that this entails that groups are not proper agents (French Reference French1984; Pettit Reference Pettit2007). It is unclear to me why Tuomela does not acknowledge this. After all, he grants them functional existence, which means that they have irreducible causal consequences. In light of this, it appears that Tuomela should embrace the existence of group agents as well.

The mind-dependence of group agents does not provide a good basis for denying their reality. To be sure, the idea that mind-dependent entities do not exist is deeply entrenched in philosophy. However, not all kinds of mind-dependence are on a par in this respect. When something depends only on conceptual schemes, or if those schemes constitute it, it is rather plausible that it is fictitious. However, when people rely on a concept in their deliberations about what to do, those schemes can have causal consequences. And when they do, it does not follow that the entity to which the concept refers is fictitious. Richard Boyd (Reference Boyd1991) and Uskali Mäki (Reference Mäki2011) argue that the fact that something has causal consequences provides a suitable basis for concluding that it is real. In the case at issue, this is plausible because group agents do not depend on the mind of a single individual, but on the minds of a number of individuals. Furthermore, there are semantic, epistemic and causal relations between the minds of these individuals. Because of these irreducibly collective causal features, the line of reasoning from mind-dependent to fictitious can and should be resisted (Baker Reference Baker2000).

Finally, the distinction between intrinsic and derived intentionality is not a solid ground for resisting ontological collectivism either. The reason for this is that it supports scepticism with respect to collective intentional states in general, and thereby threatens Tuomela's own views about collective intentional states in the I-mode. Tuomela argues that the reality of group agents should be denied on the basis of the fact that the intentionality they exhibit is extrinsic rather than intrinsic. This seems to commit him to believing that we-mode attitudes only have intentional existence. There is ample reason to believe that this also holds for collective attitudes in the I-mode. After all, both are a matter of extrinsic intentionality: rather than being states of a mind that supervenes on a brain, they are attributed by individuals who have states of a mind that supervenes on a brain. Two people who carry a piano upstairs do so due to a joint intention. They do not, however, have a joint brain.

Tuomela comes close to recognizing the problem when he points out that ‘the ontological gap between the non-fictitious joint states and actions and the fictitious intentional attitudes and actions attributed to the group agent figuratively speaking is rather “small”’ (49). This reveals how dangerous his reductionism with respect to group agents is to his overall project in social ontology. The gap that Tuomela mentions is too small to ground a difference in reality. The upshot is that, because group agents have irreducible causal powers, ontological collectivism about group agents is more attractive than Tuomela's individualism.

2. SOCIAL INSTITUTIONS

According to Tuomela, institutions are norm-governed social practices. Institution terms such as ‘money’, ‘marriage’, and ‘property’ can be used to refer to social practices, as well as to the norms governing those practices (214). Social practices are repeated, collective actions that are performed for a social reason or a shared we-attitude. A pottery practice, for instance, might involve ingrained skills including some hand movement. Those who participate in the practice possess these skills to some degree and make pots using these skills. Furthermore, they believe that others are manufacturing pots in the same way, and this is mutually believed. A social practice such as this one can be turned into a social institution by introducing social norms that govern the practice (216–218).

The main function of social institutions is to establish a stable and persistent order that is conducive to satisfying the needs of those involved (223). Social institutions serve this function by solving coordination problems and dissolving collective action dilemmas. Tuomela argues that they serve this function better if they involve we-mode collective acceptance rather than I-mode collective acceptance (229; see 16, 44, 175).

It is a platitude that social institutions constrain and enable behaviour. Institutions constrain by prohibiting certain kinds of behaviour and possibly by sanctioning violations. The natural next thought is that they enable by permitting certain kinds of behaviour. The problem with this claim, however, is that, for all we know, the permitted behaviour is possible independently of the permission. In the absence of prohibitions, the question of whether one can perform the action does not depend on whether it is permitted. This raises the question whether the sense in which institutions enable behaviour is rather trivial.

Tuomela explicates the enabling role of institutions in terms of the notion of a constitutive rule, which contrasts to that of a regulative rule. Regulative rules guide behaviour that is possible independently of those rules – think of the rules of etiquette. Constitutive rules make possible the very behaviour they guide – think of the rules of chess (Searle Reference Searle1969). Tuomela uses educational institutions as an example. Without those institutions, professors would not be in a position to decide whether a student is qualified for a university degree (in fact, there would not even be any professors; 224). This example serves to bring across the intuition that institutional behaviour is novel as compared with the behaviour people can display in a more basic social order. But it remains to be seen whether the notion of a constitutive rule or norm captures this.

Tuomela argues that the constitutive norm confers an institutional status on the activity that is characteristic of the social practice it governs, or to items that play a central role in that practice (226). The status is social because it depends on collective acceptance, and it is normative because it concerns a rule that stipulates that the action at issue is permitted, prohibited or required. The status is also symbolic, as the norm is a conceptual presupposition of the action (227). In this respect, it contrasts with norms that merely regulate behaviour. Behaviour guided by a regulative norm does not presuppose that norm.

I do not see how this analysis of norms that are constitutive of institutions supports the claim that institutional actions are novel in an ontologically significant sense. Just like constitutive norms, regulative norms are social and normative. Both kinds of norms are in force exactly if they are collectively accepted, and both prohibit, permit, or require some kind of action. This means that the only respect in which these two kinds of norms differ from one another is their symbolic status. Only constitutive norms confer such a status. The symbolic status is conceptual or linguistic. Focusing on its linguistic manifestation, it is a matter of us having certain terms such as ‘professor’. Having such terms is of great practical significance as it helps people to economize on cognitive resources. However, economy of thought is as such ontologically neutral.

The thing to see is that regulative rules can enable new forms of behaviour as well. Collective acceptance of a regulative rule introduces the normative features that are pivotal for institutions. Consider a community in which men and women tend to live together in twos. Over time, people become conscious of this and they realize that ‘this is how we do things here’. At some point, a ritual is invented to mark the moment at which a particular man and woman start living together. Perhaps the leader of the community kisses the man and the woman on the forehead. People in this community start to expect men and women to focus their attention on the person they live with and they start to frown on those who engage a lot with others. Over time, the frowns transform into explicit norms concerning sexual activity and the use of goods.

This example describes how a social practice of co-habitation – a kind of proto-marriage – turns into a social institution. All the norms that play a role in it are regulative norms. Co-habitation, sexual activity and using goods are possible independently of them. One might say that they are constitutive because the practices depend on the norms in that without the norms they would not exist. Given how I laid out the example, however, this is simply false. The social practice preceded the norms. One might say instead that they are constitutive because the practices would not be norm-governed if it were not for the norms. Although true, this claim is trivial.

Imagine next that an outsider enters the community – perhaps an anthropologist. She invents the terms ‘husband’ and ‘wife’ for individuals who underwent the ritual together, as well as the term ‘married’ that serves to mark their common status. This does not affect the institution, which still owes its existence to a regulative rule. It could happen, however, that the members of the community come to adopt the terms that the outsider introduced. They use the terms ‘husband’ and ‘wife’ for individuals who underwent the ritual together, and the term ‘married’ to mark their common status. This enables them to formulate constitutive norms such as ‘any man that has undergone the kissing ritual with a woman is her husband, and any woman that has undergone the ritual with a man is his wife’. In light of this, it is necessarily true that any two people who have undergone the kissing ritual are husband and wife. Due to the norms that are characteristic of the classificatory or symbolic practice that has been adopted, husbands and wives have certain rights and obligations when it comes to sexual activity and to the use of goods. The very same persons, however, had those rights and obligations already before people started using the terms.

This example reveals that there is no deep distinction between regulative and constitutive rules (Hindriks Reference Hindriks2009). The key difference is that constitutive rules deploy certain concepts or terms that regulative norms do not. However, the example reveals that little would have been lost if the outsider had never come to the community and the terms she came up with had never been adopted. The upshot is that collectively accepted norms enable new forms of behaviour, irrespective of whether they are constitutive or regulative. Thus, Tuomela mistakenly links the enabling role of institutions to the notion of a constitutive rule.

If regulative rules can enable new forms of behaviour, what is it exactly that they enable? At this point, Tuomela's claim that we-mode collective acceptance of rules or norms serve to solve coordination problems and collective action dilemmas becomes relevant. People can achieve more efficient equilibria by collectively accepting rules or norms, which can also issue in a more stable and robust social order. In this way, institutions have real consequences insofar as preference-satisfaction is concerned. I would like to suggest that this provides for an alternative way of giving substance to the claim that institutions enable new forms of behaviour: institutions enable certain forms of coordination and cooperation. As a consequence, the extent to which the preferences of those who participate in those institutions are satisfied increases. Thus, I venture, the enabling role of institutions is intimately bound up with the value they create.

I warmly recommend Tuomela's new book. It is perceptive, thorough, and broad. Among the many books he has written, it is the most accessible one. And it presents a social ontology that is rich both in theories and arguments that deserve to be welcomed as well as scrutinized.