Cultural Conventions as Group-Makers

In: Journal of Cognition and Culture
Marc Slors Professor of Philosophy of Mind and Cognition, Faculty of Philosophy, Theology and Religious Studies, Radboud University Nijmegen The Netherlands

Search for other papers by Marc Slors in
Current site
Google Scholar
Open Access


In most literature on human cultural evolution and the emergence of large-scale cooperation, the main function of cultural conventions is described as providing group-markers. This paper argues that cultural conventions serve another purpose as well that is at least as important. Large-scale cooperation is characterized by complex division of labour and by a diversity of social roles associated with cultural institutions. This requires ubiquitous ‘role-interaction coordination’ – as it will be labelled. It is argued that without cultural conventions this type of coordination would be cognitively intractable. Thus, apart from functioning as group markers, they are first and foremost important group-makers.


In most literature on human cultural evolution and the emergence of large-scale cooperation, the main function of cultural conventions is described as providing group-markers. This paper argues that cultural conventions serve another purpose as well that is at least as important. Large-scale cooperation is characterized by complex division of labour and by a diversity of social roles associated with cultural institutions. This requires ubiquitous ‘role-interaction coordination’ – as it will be labelled. It is argued that without cultural conventions this type of coordination would be cognitively intractable. Thus, apart from functioning as group markers, they are first and foremost important group-makers.

Humans are ‘groupish’ (Ridley, 1996: 39–40). We defend the integrity of our group against other groups and suppress selfishness in its interest. We identify with our group and are prone to serious in-group bias: we prefer the company of in-group members over out-group members, trust in-group members much more than out-group members, and are considerably more affected by the judgments of in-group members than those of out-group members. Some argue that groupishness is a feature of human sociality that sets us apart from other animals (Tomasello, 2014: 84). There are various explanations for this phenomenon. For example, group identification is held to mitigate epistemic uncertainty (Greenberg and Arndt, 2011; Hogg, 2011) and boost self-esteem (Reynolds et al., 2017; Sibley and Duckitt, 2008). Evolutionary explanations of groupishness focus on the need for intra-group collaboration and the reality of inter-group competition (Tomasello, 2014, 2020; Richerson and Boyd, 2005) or on the idea of an evolved ‘coalitional psychology’ (Tooby and Cosmides, 2010; Boyer, 2018). There are also those who argue that such explanations may be redundant, since minimal identity – ‘mere membership’ of artificial groups through meaningless group-tags – turns out to be sufficient for many effects associated with in-group bias (Dunham, 2018). What all these approaches have in common is the idea that the role of cultural conventions – social etiquette, customs, dress codes, etcetera – is that of group-markers. With respect to our groupishness, cultural conventions are mainly supposed to serve the purpose of identifying in-group and out-group members as such. The aim of this paper is to argue that cultural conventions play a much more substantial role as well. They are not just group-markers, but indispensable group-makers.

In order to argue for this, I will start with a discussion of Michael Tomasello’s views on the nature and origins of human groupishness. For the view I will put forward can be seen as a proposal to extend Tomasello’s outlook. According to Tomasello, “in-group favoritism (and out-group disfavoritism) is basically a scaled-up version of children’s tendency to favor collaborative partners over non-collaborators (i.e., free riders)” (Tomasello, 2020: 8). Group identification, on this view, is based on a large-scale parallel of the joint intention to collaborate that we find in dyads: a collective intention to collaborate for the survival and thriving of our group. In Section 2, I argue that the parallel between dyads and large groups is only partial because they differ quite considerably in the amount and kinds of coordination that collaboration requires. In Section 3, I argue that cultural conventions are an indispensable tool for a specific type of coordination that is ubiquitous in groups whose collaboration is based on complex division of labour. This coordination-enabling feature of cultural conventions makes us cognitively dependent on them. And this dependency explains our immediate preferences for the company of those who share our conventions – our in-group bias. In the final section of this paper, I will compare the idea of conventions as group-makers with Tomasello’s idea that collective intentionality is the main force behind group formation. I will argue that the idea of conventions as group-makers shows that the mechanisms that drive group formation need not rely on our wanting or intending to form a group.

1 Collaboration and Collective Intentionality as the Basis for Groupishness

According to Tomasello, group identification and in-group bias is grounded in collective intentional attitudes that are basically scaled-up versions of the joint intentionality that we find in collaborating dyads. This scaling up applies both to ontogenetic development of humans and to phylogenetic processes of cultural evolution.

Let’s look at dyads first. From a large number of experiments on both human infants and chimpanzees, Tomasello concludes that in contrast with chimps, 3 year-old humans “begin to feel the normative force of joint commitments to collaborate.” (Tomasello, 2020: 4) In human dyads, “the joint agent ‘we,’ (..) not only conducts but also self-regulates the collaboration. (…) [P]articipation in joint intentional collaboration, especially as initiated by a joint commitment, is the earliest source of children’s (and early humans’) feeling of obligation to their partner.” (Tomasello, 2020: 6) The normative1 bonds in collaborating dyads are neither based on personal sympathy, nor on pre-existing societal norms, Tomasello argues; the crucial force, rather, is joint intentionality and the way in which this makes collaborators view their respective roles in the collaboration. Tomasello:

Key in all of this is the notion of role that joint intentional collaboration creates. In collaborative activities performed by a “we,” partners each have their role to play. As both play their roles, they come to see one another as equally deserving individuals, equally worthy of respect. This recognition is based on a dawning understanding of self-other equivalence (…). Most basically, as children participate in joint intentional collaboration, they see that: (1) both participants are equal causal forces in producing the mutually intended outcome; (2) both partners could switch roles as needed; and, most crucially, (3) the standards of performance for each role (…) are impartial in the sense that they apply to anyone in that role.

Tomasello, 2020: 6

Thus, joint intentionality lies at the basis of a normative relation between collaborators that expresses itself as a sense of obligation toward each other; a sense that is not felt towards non-collaborators. A collaborator cares about her partner’s behaviour, and about the way her partner views her own contribution to their shared project. Collaboration makes partners directed to each other in ways that differ from their relation to anyone outside of the collaboration. In a sense, they are a tiny in-group.

Now to the scaling up. The central assumption here (one that I shall question, to some extent, below) is this:

A cultural group (…) is nothing more or less than one big collaborative activity in which “we” as a people operate with a collective commitment to the group’s surviving and thriving. Each individual has her role(s) to play in this collective commitment (…) and this generates, in a scaled-up manner, more universal normative expectations.

Tomasello, 2020: 7

From an evolutionary point of view, the idea is that

with increasing population sizes and competition among humans, the members of human groups began to think of themselves and their groupmates (known and unknown, present and past) as participants in one big, interdependent, collaborative activity aimed at surviving and thriving in competition with other human groups.

Tomasello, 2014: 84

In-group formation is thus the result of collective intentionality. When I identify an in-group stranger as “one of us”, this means I am assuming, implicitly, that this person is part of the overall collaborative effort of the group I belong to.

Cultural norms, which are defining features of a given in-group, can also be viewed in terms of the parallel between the joint intentionality of collaborating dyads and the collective intentionality of whole groups. Tomasello sees them as scaled up versions of the principles of partner choice and of partner control (the forms of protest we find in individuals who think their collaboration partner is not living up to their expectations): “The enforcement of social norms is (…) a kind of scaled-up, third-party version of the second-personal protest characteristic of joint intentional collaboration: It is a new, group-level form of partner control comprising group-level protest backed by a threat of exclusion (partner choice).” (Tomasello, 2020: 7) Even when such norms are formalized into cultural institutions, they are still ‘descendants’ from normative relations in collaborating dyads.

One crucial difference between dyads and large groups, though, is the fact that large groups do, but dyads do not, use elaborate cultural conventions. By cultural conventions’ I mean contingent customs, collective habits, rules, and regularities pertaining to the ordering and aesthetics of our socio-cultural environment. These include social etiquette, dress codes, social scripts, and even the design and use of artefacts, public spaces and interiors. Think of such things as queueing, knowing the proper connections between accent and class, using and grasping understatement and humor, shaking hands or bowing as a means of greeting, table manners, tacit expectations about bodily posture, eye contact, gestures, and accent and vocabulary in different social groups and different situations. Think of the rules we employ in board meetings, or the typical interaction scripts that we use while ordering wine in a restaurant, check out at the supermarket, address a superior, or have a casual conversation with friends. Or think of the way we understand which clothes to wear when and what messages they send in which situation, our understanding of how to greet whom and how to take leave.

Tomasello acknowledges the fact that large groups do and dyads do not use cultural conventions, emphasizing the prominence of cultural conventions and practices in large groups at many occasions. But this marked difference between dyads and large groups (societies, cultures) does not undermine the parallel that is invoked to explain groupishness. This is because cultural conventions solve a problem that does not occur in dyads: the identification of those we should count (or discount) as co-group members. Groups that are larger than hunter-gatherer bands involve in-group strangers who must be distinguished from out-group strangers. This is where cultural conventions are of obvious use. Speaking of our evolutionary past, Tomasello writes that “Group members were identified most readily by specific cultural practices, and so teaching and conformity to the group’s lifeways became a critical part of the process. (…) Group identification means that human groups each have their own set of conventional cultural practices.” (Tomasello, 2014: 84–85) Thus, the possibility to identify group members hinges on a conventional group-identity: “cultural practices and social norms in large part identify our group as who we are: ‘We’ are those people who talk, think, dress, and eat in these particular ways.” (Tomasello, 2020: 7).

In what follows I will argue that there is room for a completely different but equally substantial function of cultural conventions: not as group identifiers but as group-constitutors.

2 Dyads, Groups and the Relevance of Different Types of Coordination

The parallel between dyads and societies drawn by Tomasello, i.e. between joint intentions to collaborate and collective intentions to collaborate, hides an important difference between the kinds of collaboration at issue. In this section I will discuss the fact that collaboration in groups involves significantly different kinds of coordination than collaboration in dyads. In the next section I will argue that a ubiquitous type of coordination found in groups (but not in dyads) is facilitated by cultural conventions, and that that turns conventions into a kind of social glue that is as important as collective intentionality.

To appreciate the difference between the kinds of collaboration in dyads and in societies, it is helpful to look at two features of the latter that are absent in the former. Societies are characterized by synchronic social organization and by the diachronic transmission of knowledge and skills. Interestingly, when writing about the stages in human evolution where these societal features emerged, Tomasello emphasizes the important connection between coordination and conventions in a way that is very close to what I want to argue later on:

The combination of these changes in the two dimensions of human sociality created some totally new cultural realities. The transformative process was conventionalization, which has both a coordinative component, as individuals implicitly “agree” to do something in a consistent way (everyone wants to do it this way as long as everyone else does, too), and a transmitive component, as this way of doing things sets a precedent to be copied by others who want to coordinate as well. The result is what we may call cultural practices, in which individuals, in effect, coordinate with the entire cultural group via collectively known cultural conventions, norms, and institutions.

Tomasello, 2014: 81

I believe Tomasello is entirely right here. But I also think that he concentrates on one type of coordination only – one that is dominant in the literature that connects conventions to coordination – while ignoring other types and that this prevents him from appreciating the full use we make of cultural conventions.2

The kind of coordination Tomasello speaks of is well known from the debate on conventions initiated by David Lewis, to which Tomasello refers. For Lewis, a coordination problem is a situation “of interdependent decision by two or more agents in which coincidence of interest predominates and in which there are two or more proper coordination equilibria. (…) [T]hey are situations in which, relative to some classification of actions, the agents have a common interest in all doing the same one of several alternative actions.” (Lewis, 1969: 24) Coordination, in this sense, is about doing things the same way. We coordinate when we manage to be at the same time and place for Friday drinks after work. We coordinate when we keep to one side of a narrow alley so as to not bump into each other. We coordinate when we all wear hats for someone’s party. Conventions serve the purpose of maintaining such coordination in recurring situations. Being at a certain pub around 5 o clock every Friday for drinks might become a convention. Always keeping to the right side of the alley – like driving on one side of the road – might become a convention when you and I navigate the same alley every day. And wearing hats for parties might become a convention like wearing neck ties to the office and T-shirts at neighbourhood barbecues.

This type of coordination involves doing the same things in the same situations, that is, it involves following certain rules. Here it is important to distinguish rule-selection from rule-use. Rule-selection, that is, settling on the same convention, is solving a coordination problem in Lewis’s sense. We may not have selected a time and place for our Friday afternoon drinks. It is in all our interests to be at the same place at the same time, so selecting a rule for this is solving our coordination problem. Salience and precedent are the main motors behind rule selection in such cases, Lewis argues (based on Schelling, 1960; see also Sudgen, 2011). If there is only one really good pub that is close to our office, that will probably be where most people go when meeting for the first time. And once most people have ended up there the first time, everyone will probably go there the next. This is how a convention may emerge.

So, selecting a conventional rule serves the purpose of solving a coordination problem. In some cases, the use of such a rule also serves a coordinative purpose. For example, after settling which side of the road all cars should keep to, which is a rule-selection coordination problem, we can put that rule to use to solve the further coordination problem of avoiding collision and not being in each other’s way (we can see that this is indeed a further coordination problem by considering the fact that there is an alternative way of solving it: not through adopting a rule, but through driving slowly and carefully). But in other cases, rule-use serves non-coordinative purposes. Thus, for example, after settling the rule selection problem of determining which words to use for which things, we can then use these rules for non-coordinative purposes such as influencing each-others thoughts (see Schiffer, 1972: 151; Lewis, 1983: 166). The difference between rule selection and rule use will become important at the end of this section.

Rule-selection (but, as I will argue below, not rule use) is a form of what is known as correlative coordination. Correlative coordination – how do we manage to do things the same way in our common interest – is sometimes contrasted with what is known as complementary coordination. Here coordination requires precisely not doing the same things. Consider the example of shared intentionality proposed in an earlier BBS target article by Tomasello and colleagues (Tomasello et al., 2005). Two people have to open a box which is too big to hold steady and cut open at the same time by one person. So, one person must hold the box and the other must cut it open. Coordination is achieved by complementing each-others actions here, not by doing the same thing. Solving a coordination problem of this type is not determining how we shall behave (for this is more or less given), but settling the question “who does what?”3

Solving a complementary coordination problem is done, not via rule- selection, but through what we may call ‘role-selection’. Role-selection may, but need not, involve agreements and decisions. Here’s a toy example of emergent role-selection (taken from Abramova and Slors, 2015). Suppose a group of parents have to clean up a classroom after a school party. One person enters the room and decides to start vacuum cleaning, thinking that when that is done she will pick another chore. The next person enters and sees that vacuum cleaning is already ‘taken’. She decides to start wiping the desks. And so on, until all parents are busy with their respective, complementary chores. When one chore is finished the relevant person will look for a new one, or starts helping another parent. What this example shows is that there is a tight connection between role-selection or complementary coordination and division of labour. Deciding how to divide labour is solving a role-selection problem.

Human societies are characterized by an often massively complex division of labour, usually but not always subserved by emergent role-selection as in the classroom case, and by division of social roles, which are connected with cultural institutions. The diversity of roles in societies leads to a class of coordination problems that I believe is overlooked by Tomasello and many others. These are neither the correlative coordination problems analysed by Lewis, nor the complementary coordination problems solved by role-selection. This is a class of coordination problems that is connected not with the selection but with the execution of different roles in societies with complex division of labour. Once we have a wide range of different roles in a given society, these roles have to interact every now and then for the exchange of goods, services and/or information. And if they do not interact, they must not hinder each other. This leads to three types of problem: (1) how do we recognize who plays which role? (2) How do we standardize the interaction between roles so that the exchange of goods, services and information proceeds as smoothly as possible? And (3) how do we make sure that when people carry out various roles in a small space, such as a market hall, an office or a factory hall, they are not in each-others way when they are not interacting? These are problems of what I will call role-interaction coordination.

In the next section, I will argue that many instances of role-interaction coordination are facilitated by conventions about dress, etiquette, and behavioural templates. Before discussing this, however, it is important to stress two things. First, role-interaction coordination problems can be found in abundance in societies, but not usually in dyads. This is not to say that there is no role-interaction coordination in dyads. There is. Think of the coordination between the person holding the box and the person cutting it open in Tomasello et al.’s example. After it has been decided who does what (role-selection coordination), the person holding the box needs to hold it in such a way that the other person is best able to cut it open without unnecessary effort or damage – both collaborators need to attune their behaviour to each-other. That is role-interaction coordination too. But such role-interaction coordination does not usually require rules or conventions. It requires tracking each other’s behaviour carefully. Role-interaction coordination in dyads is usually sufficiently served by embodied or sometimes verbal negotiation. In societies, I will argue below, it requires more than that.

Secondly, cultural conventions can be conceived of as rules about how to behave and how to dress in specific situations. And since settling on which conventional rules we adopt is the type of coordination problem Lewis is concerned with, it may seem as if role-interaction coordination is covered by Lewis and thus taken into account by Tomasello. That is a mistake, however. Role-interaction coordination is an instance of rule-use, not rule selection (in terms of Lewis’ definition of a coordination problem: smoothly coordinated role interaction is the “common interest” that is served by settling on a convention). Just like the process of deciding on which side of the road cars should drive (rule-selection) is logically distinct from the coordination problem that the use of the rule ‘everyone should drive on the left-hand side of the road’ is meant to solve (avoiding collision and not being in each-others way – which is our “common interest”), the process in which cultural conventions emerge or are decided on (rule-selection) is logically distinct from the coordination problems associated with interacting cultural roles. It is the latter type of coordination problem I will be concerned with.

3 Conventions and Role-Interaction Coordination

Let’s go back to the parents who cleaned up the classroom after a party at some elementary school. They may have been surprised at the smoothness of their (role-selection) coordination and the speed at which they collectively managed to clean up, without anyone being in charge. As a result, they decide to gather every Friday afternoon and clean up the classroom so that the school can save money on their cleaning service. Perhaps fixed roles will start to emerge – one parent being ‘the vacuum cleaner’ and another being ‘the desk wiper’. But especially when some chores are more coveted than others, it may also be the case that roles switch or that role-selection is emergent every time. Independently of this, however, it is likely that some conventions start to emerge that pertain to role-interaction.

For example, cloths are used for wiping desks, cleaning the sink and cleaning the blackboard. But it may not be a good idea to use the same cloth in each of these instances (e.g. to avoid coating blackboard and the desks with a smelly film of organic waste). Especially when roles change frequently, it may therefore be useful to use simple reminders: red cloths are for cleaning the sink, yellow ones for desks and blue ones for the blackboard. Also, given that parents are frequently in each other’s way, and given that it is easier for a desk wiper to step aside than it is for someone handling a vacuum cleaner (on top of the fact that vacuum cleaning needs to be done first, so that the floors can be mopped later on), it may become standard practice that people step aside for the vacuum cleaner. And since many roles involve the use of buckets every now and then, which are scarce, the custom may emerge that all buckets are to be returned to their place under the sink, right after use. We may also fantasise that the example of the parents of this one class is followed by parents of other classes. Perhaps some parents start to specialize; there may be one expert vacuum cleaner who vacuums all the classes, or someone who waters all the plants. In this situation parents are likely to interact with unknown others whose role they must infer (“who’s the desk-wiper in this class? We’re out of yellow cloths”). Here again, standard signposting practices may emerge over time (say, desk-wipers are recognizable by their aprons).

Of course, this is a toy example and perhaps not even a very realistic one (although the initial example was taken from real life). The point here is to illustrate how emergent conventions are used for facilitating role-interaction coordination. In what follows, I will argue that regardless of how they emerged, many or most cultural conventions are put to similar use. Let me briefly discuss the three kinds of role-interaction coordination identified in the previous section and indicate how they are facilitated and made cognitively tractable through cultural conventions.

Let me start with the signalling and signposting of roles. When we look at economic division of labour and at the division of roles in social institutions, one obvious example of the way in which conventions are used for signposting is clothing. The clearest example is probably the use of uniforms that allow us to immediately recognize policemen, doctors, shop personnel, etcetera. But we can also think of the typical styles of dressing that are associated with certain professions. These are usually not as standardized as uniforms, but we certainly have no problem telling a rock guitarist from a corporate lawyer or a butcher. And when we pay a little more attention, we can discern dress codes that obtain within specific sub-groups in society. Think of the subtle cues that are used by teenagers to signal membership (or desired membership) of various sub-cultures. And within sub-cultures, even more subtle cues are used to signal hierarchy or dominance positions. It is very hard to think of any style of clothing that does not signal roles within groups in any way. Even such everyday garments as jeans and T-shirts will send messages in different situations: at a street barbecue they will signal blending in and being part of the neighbourhood, but at formal occasions they will signal playing the role of a rebel or someone who is deliberately aiming to offend.

Apart from clothing there are other conventional signposts and signals. For the trained perceiver, the use of specific gestures, table manners, and accents can reveal a host of information about social status, upbringing, and cultural roles. Usually such signals are combined. Unlike the use of uniforms, it is not often the case that one specific type of convention reveals one specific social role. Just like we use letters in an alphabet to form words and sentences, most conventions that are used for the signposting of informal social roles are combined to send the intended message.

But just like in the case of more unambiguous signals such as uniforms, it would be hard to play most of our social roles without the use of relevant signposts. That is almost certainly an understatement. It is not a priori impossible that a combination of verbal communication, theory of mind-use and inference, for example, would allow a hypothetical population that does not use conventional signposts for social and economic roles (and that is large enough to not know everyone’s roles by heart) to discern these roles nevertheless. But there is no doubt that this way of identifying roles will be cognitively taxing, compared to the use of more or less standardized conventional cues. The fact that all societies that are large enough to contain in-group strangers use conventional signposts to mark social and economic roles, suggests that the extra cognitive effort that would be required for the identification of roles in our hypothetical population is substantial indeed. In all likelihood, it will go at the expense of these roles themselves, or the complexity of their division and interrelation.

Something similar applies to a second way in which cultural conventions are used for the purpose of coordinating role-interaction. Role-interactions that are aimed at the exchange of goods, services and/or information are often facilitated by conventional interaction templates, i.e. more or less standardized ways in which such exchanges are supposed to proceed. Some exchanges, like sampling the wine one has ordered before the waiter pours the glasses of the other persons at the table in an expensive restaurant, showing one’s passport at the airport customs, or checking out at the supermarket are more or less scripted (Shank and Abelson, 1977; Meng, 2008). Other exchanges are not scripted, but aided by conventional cues for specific transactions. The simplest example of these are greetings, such as shaking hands or bowing, as openings and closings of transactions. But we can also think of hammering as a means of conveying that a decision has been taken in a meeting, waiting for the host to start eating as a sign that the rest of the table can follow, raising one’s glass in a toast, etcetera. Like with signposting roles, interaction templates facilitate exchanges in such a way that any non-conventional alternative will require considerably more cognitive effort.

There is a third way in which conventions facilitate role-interaction. Role- interactions do not just consist in exchanges of goods, services and information. They also consist of carrying out roles simultaneously in small spaces in which one can be in each-others way. These are situations in which role-interaction coordination consists in securing ‘non-interference’. Think of manoeuvring through a crowded shopping mall, of letting people get off the bus before getting on one, of not speaking in a library, of queuing, or of holding the door for each other. We have a set of cultural conventions that function as non-strict versions of traffic laws and that serve the common interest of each pursuing our own goals – playing our own roles when not connecting with others – as undisturbed as possible in situations where we are likely to be in each other’s way.

So, the idea is that conventions facilitate role-interactions by making them cognitively low-cost. Thus, conventions are a necessary cognitive scaffolding for scaling up collaboration in a way that involves complex division of labour and the formation of complex social institutions, both formal and informal. This means that conventions are not merely group markers, which is the standard position with respect to the function of conventions in human groupishness, but true group makers as well.

4 Group Formation beyond Collective Intentionality

On Tomasello’s account, large groups survive and thrive because everyone, or nearly everyone in the group has their minds set on collaborating with everyone else in the group, for the sake of the group’s thriving and surviving. I do not intend to question the necessity of something like such collective intentionality for large group formation or collaboration within large groups. But collective intentionality isn’t sufficient for large-scale collaboration. Large scale collaboration involves complex division of labour and complexly interrelated socio-cultural roles. The kind of coordination involved in the interaction of these roles requires more than intentions. In order to become cognitively tractable, role-interaction coordination cannot do without a range of conventional signals, templates, scripts and rules that function as a socio-cultural infrastructure – a medium that does not dictate a specific division of roles and tasks, but that enables many such divisions to function smoothly.4

The function of cultural conventions as group-makers is not implied by and thus independent from group-making collective intentions to further the surviving and thriving of the group through collaboration. Once groups have a complex division of tasks and roles, the use of cultural conventions as a shared infrastructure for role-interaction coordination becomes imperative. Shared conventions – many of which consist of embodied, automated behavioural patterns (Voestermans and Verheggen, 2013) – thus play a similar but distinct function as shared intentions on Tomasello’s view, namely that of social glue. Large groups survive and thrive not merely because group members intend this, but because group members can simply focus on their respective complementary tasks and roles, without having to spend a great deal of cognitive effort on coordinating these with the tasks and roles of others. Cultural conventions enable a situation in which a focus of individuals on their own tasks and roles, rather than on the collective, still allows a large group to flourish. Parallel to and partly5 overlapping with Adam Smith’s notion of the ‘invisible hand’ (Smith 1776), the idea here is that a focus on one’s individual situation – within a massively complex network of interrelated roles and tasks – benefits not just the individual but the group as well.

As Joseph Henrich points out, cultural evolution makes sure that adaptive cultural practices are selected and transmitted, regardless of whether their adaptive use is transparent to group members (Henrich, 2016). This is directly applicable to the selection and transmission of cultural conventions that facilitate role-interaction coordination. The fact that we need not be aware of the coordinative function of cultural conventions underscores the fact that no intention – collective or individual – is required for cultural conventions to function as social glue.

From the viewpoint of the group member, when cultural conventions are used, they simply open up a landscape of social affordances to which she can respond (Rietveld and Kiverstein, 2014). In this sense conventions are transparent: when someone stretches out her hand to me, I do not infer an intention to greet through recognizing the use of a conventional sign for greeting – shaking hands. Rather, I directly perceive such an intention in the conventional behaviour. The transparency of cultural conventions can be compared to the transparency of language. When reading a novel, say, we do not perceive shapes of ink on paper first and then letters, and then words and then meanings. Rather we read a story ‘though’ the shapes of ink on paper. The connection between meanings and words, once learned, is very strong and cannot be undone by will. The famous Stroop task is a clear indication of this: when colour words are printed in the ‘wrong’ kind of ink (ink of a colour that is different than the one indicated by the word), we are much slower to name the colour of the ink, because the meaning of the word gets in the way of selecting the ink-colour word. Similarly, we do not first perceive an outstretched hand and only then interpret this as a gesture of polite greeting: we immediately see an intention to greet. Thus when, in times of social distancing during the Covid pandemic, I accidentally offer a hand to someone who then refuses to shake it, for good reasons that I recognize, I might nevertheless involuntarily experience a pang of insult.

The idea of cultural conventions as a transparent medium that opens up a landscape of affordances for smooth role-interaction coordination explains what happens when I identify a given stranger as an in-group member by the conventions she uses (the language she speaks, her accent, the gestures she makes, the clothes she wears, etcetera). I do not only experience these conventions as group markers that identify someone as sharing in the collective intentionality of my group, I also experience them as group-makers through which I immediately recognize an array of possibilities for role-interaction. This recognition is recognizing a stranger as a member of one’s group.

Thus, the idea of conventions as group makers complements Tomasello’s idea that groups are held together primarily by collective intentions to collaborate.


I would like to thank an anonymous reviewer for this journal for insightful comments.


  • Abramova, E. and Slors, M. (2015). Social Cognition in Simple Action coordination: A Case for Direct Perception. Consciousness and Cognition, Vol. 36, 519531.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Boyer, P. (2018). Minds Make Societies. How Cognition Explains the World Humans Create. New Haven: Yale University Press.

  • Bratman, M. E. (2014). Shared agency: A planning theory of acting together. New York: Oxford University Press.

  • Dunham, Y. (2018). Mere membership. Trends in Cognitive Science Vol. 22 (9), 780793.

  • Gilbert, M. (2014). Joint Commitment: How we make the social world. New York: Oxford University Press.

  • Greenberg, J. and Arndt, J. (2011). Terror management theory. In Handbook of Theories of Social Psychology (Vol. 2) (P. A. M. Van Lange, A. W. Kruglanski, & E. T. Higgins, Eds.) pp. 398415, Sage Publications.

    • Search Google Scholar
    • Export Citation
  • Henrich, J. (2016). The Secret of Our Success. How Culture is Driving Human Evolution, Domesticating Our Species, and Making Us Smarter. Princeton: Princeton University Press.

    • Search Google Scholar
    • Export Citation
  • Hogg, M. A. (2011). Uncertainty-identity theory. In Handbook of Theories of Social Psychology (Vol. 2) (P. A. M. Van Lange, A. W. Kruglanski, & E. T. Higgins, Eds.) pp. 6280, Sage Publications.

    • Search Google Scholar
    • Export Citation
  • Lewis, D. (1969). Convention. A Philosophical Study. Oxford: Blackwell.

  • Lewis, D. (1983). Philosophical Papers Volume 1. New York: Oxford University Press.

  • Meng, H. (2008). Social Script Theory and Cross-Cultural Communication. Intercultural Communication Studies Vol. 17 (1), 132138.

  • O’ Connor, C. (2019). The Origins of Unfairness: Social Categories and Cultural Evolution. New York: Oxford University Press.

  • Pettit, P. (2020). A Hard Choice for Tomasello. Behavioral and Brain Sciences Vol. 43, 3738.

  • Reynolds, K. J. et al. (2017). Self-categorization theory. In The Wiley Blackwell Encyclopedia of Social Theory (2nd edn) (Turner, B. S., ed.), 14, John Wiley & Sons.

    • Search Google Scholar
    • Export Citation
  • Richerson, P. J. and Boyd, R. (2005). Not by Genes Alone: How Culture Transformed Human Evolution. Chicago: University of Chicago Press.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ridley, M. (1996). The Origins of Virtue. Human Instincts and the Evolution of Cooperation. New York: Penguin.

  • Rietveld, E., and Kiverstein, J. (2014). A rich landscape of affordances. Ecological Psychology Vol. 26 (4), 325352.

  • Schiffer, S. R. (1972). Meaning. Oxford: Clarendon Press.

  • Schelling, Thomas (1960). The Strategy of Conflict. Cambridge MA: Harvard University Press.

  • Shank, R. C. and Abelson, R. P. (1977). Scripts, Plans, Goals and Understanding. An Inquiry into Human Knowledge Structures. Hillsdale NJ: Erlbaum.

    • Search Google Scholar
    • Export Citation
  • Sibley, C. G. and Duckitt, J. (2008). Personality and prejudice: a meta-analysis and theoretical review. Personality and Social Psychology Review Vol. 12, 248279.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Slors, M. (2021). A cognitive explanation of the perceived normativity of cultural conventions. Mind & Language Vol. 36, 6280.

  • Smith, A. (1776/1970). The Wealth of Nations, Harmondsworth: Penguin.

  • Sudgen, R. (2011). Salience, Inductive Reasoning, and the Emergence of Conventions. Journal of Economic Behavior and Organization Vol. 79, 3547.

    • Search Google Scholar
    • Export Citation
  • Tomasello, M. (2020). The Moral Psychology of Obligation. Behavioral and Brain Sciences 43, 112.

  • Tomasello, M. (2014). A Natural History of Human Thinking. Cambridge MA: Harvard University Press.

  • Tomasello, M., Carpenter, M., Call, J., Behne, T. and Moll, H. (2005). Understanding and sharing intentions: The origin of cultural cognition. Behavioral and Brain Sciences Vol. 28, 675735.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Tooby, J. and Cosmides, L. (2010). Groups in mind: the coalitional roots of war and morality. In Høgh-Olesen (ed.). Human Morality and Sociality Evolutionary and Comparative Perspectives. London: Palgrave Macmillan, 191234.

    • Search Google Scholar
    • Export Citation
  • Tuomela, R. (2005). We-Intentions revisited. Philosophical Studies Vol. 125, 327369.

  • Voestermans, P. and Verheggen, T. (2013). Culture as Embodiment. The Social Tuning of Behavior. Oxford: Wiley Blackwell.


Philip Pettit argues that Tomasello does not succeed in deriving normative bonds from joint intentionality (Pettit, 2020). Tomasello refers to Bratman’s (2014) and Gilbert’s (2014) accounts of joint intentionality. Pettit argues that Bratman’s account does not involve normativity while Gilbert’s account requires agents to master the concept of moral obligation before they are able to have joint intentions – in which case joint intentionality cannot underlie normative bonds. It remains to be seen whether for instance Tuomela’s (2005) account of joint intentionality would be able to save Tomasello here (see also Tomasello et al., 2005). For my purposes, however, normativity and moral obligation do not play a leading role – ‘normative bonds’ may be read as ‘social bonds’ without this affecting the argument of this paper.


The argument in this section and the next is based in part on Slors 2021. However, I have made a number of improvements to the taxonomy and labelling of types of coordination proposed in that paper.


Lewis was sceptical about the difference between correlative and complementary coordination. This is connected with one of his main examples of a coordination problem. In this example, a phone call is disconnected. Who calls back? If both interlocutors call back or if both wait, the connection will not be restored. So, one has to call back and the other has to wait. This looks like complementary coordination. But, Lewis argued, this is a matter of description. In his definition of a coordination problem, people have a common interest in doing the same thing “under some classification of actions.” Settling this coordination problem through establishing a convention means settling on a rule – either the caller calls back, or the receiver. Hence, what looks like complementary coordination can be re-described as correlative coordination where both interlocutors use the same rule. However, and this is crucial, such re-description is not possible in all cases of complementary coordination. Let’s look at the box opening case. Coordination might be re-described as following the rule ‘one person holds the box steady and the other person cuts it open’. This is basically an action plan, but so is the rule ‘the caller calls back and the receiver waits’. What, then, is wrong with this re-description? First of all, there is no alternative rule that we can choose. In the telephone case, there are two rules we can pick and picking one is establishing coordination. Secondly, after we have selected this one rule – one person holds the box steady and the other cuts it open – the main question that needs answering in order to achieve coordination is still left unanswered: who does what? Who holds the box, and wo cuts it open? We may want to re-describe that problem as determining whether we follow the rule ‘Sue holds the box and Bob cuts it open’ or the rule ‘Bob holds the box and Sue cuts it open’? But these aren’t rules in the relevant sense: unlike in the telephone case, for each different pair of people, the same coordination problem would have to be solved by a different rule. As Cailin O Connor argues, the equivocation of both kinds of coordination only works when we can tag the people involved in the problem in impersonal ways – e.g. as ‘the caller’ and ‘the receiver’ (O Connor, 2019).


See Slors 2021, Section 5 for a more elaborate explanation of the connection between the cognitive tractability of role-interaction coordination and the use of conventions.


The overlap is only partly because Smith’s focus is on division of labour in the free market, whereas I intend to include other forms of divided labour and the division of roles in socio-cultural institutions as well.

Content Metrics

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 110 110 22
PDF Views & Downloads 165 165 39