Mind the Economy/Justice 159

David Lewis and the unspoken world of conventions

by Vittorio Pelligra*

28 June 2026

11' min read

Translated by AI

Versione italiana

11' min read

Translated by AI

Versione italiana

The colour green has, in itself, no natural inclination to signify ‘go’. Red does not possess any metaphysical property that makes it suitable for signifying ‘stop’. We could have done things differently. In another possible world, red on a traffic light might have signalled ‘go’, whilst green might have signalled ‘stop’. And yet, once a community has settled on a particular correspondence, no one can change it on their own. If, out of a desire for philosophical originality, I were to decide that from today onwards red means ‘go’, I would not have invented a new freedom. I would simply cause an accident. The essence of David Lewis’s thought lies in this distinction between the arbitrariness of the sign and the social necessity of coordination. Conventions are like that. They might be different, but they cannot be different solely for the sake of individuals. They are not true because they are natural, nor are they binding because they are eternal. They are binding because they make practical understanding possible amongst those people who must act in concert due to their interdependence. This is why, when we enter a room, we lower our voices if others are speaking quietly. We queue up without there being any binding law on the matter. We cross the road by interpreting the gestures, lights and movements of those in traffic like ourselves. We greet one another in a certain way; we know when a meeting is about to begin, when a conversation has ended, and when a silence is awkward rather than respectful. Long before any rule comes into play, long before any penalty is threatened, long before anyone explicitly says ‘this is how it must be done’, social life is already governed by patterns that allow us to anticipate each other’s behaviour. These are not mere habits, because a habit can be solitary. I might drink coffee every morning, go for a walk at the same time every day, or read before going to sleep. A convention, on the other hand, is something else entirely. It exists only within the realm of interdependence. It is not just about what I do, but about what I do because I expect others to do something, and because they, too, expect me to behave accordingly.

The philosopher David Lewis was the undisputed pioneer in the study of conventions. In his seminal work *Convention. A Philosophical Study (Harvard University Press 1969), he was the first to attempt to clarify what a convention is without reducing it to an explicit agreement, a moral norm or a rule imposed from above. ‘Language,’ the philosopher states, ‘is just one of many activities governed by conventions that we have not created by mutual agreement and that we are unable to describe’ (p. 3). Language is governed by conventions, but language is merely a specific example. So too are many forms of economic, political and institutional life. The real question is to understand how we can be bound by regularities that no one has formally established and to which no one has formally attributed the power to do so.

The fragile power of precedent

There are situations in which it is in everyone’s interest to act in a way that is compatible with others, but there are several possible ways of achieving such compatibility. If everyone drives on the right, it is best to drive on the right. If everyone drives on the left, it is best to drive on the left. The point is not that one of the two solutions is morally superior to the other. The point is that each becomes rational only if others adopt it too. If there were only one sensible course of action, we would not need a convention. We would simply have a necessary solution. Lewis captures a profound aspect of communal life here. Many human institutions are not based on the intrinsic evidence of what they prescribe, but on the stability of the expectations they make possible. The side of the road on which we drive, the value attributed to money, the meaning of words, forms of greeting, protocols of courtesy, etc.—all of this could be different. But it cannot be so by private initiative. A convention is arbitrary in content, but not in function. It serves to make mutual action predictable.

In its original formulation, Lewis defines a convention as ‘a regularity R in the behaviour of individuals in a population P’ in a recurring situation, where everyone conforms to it, everyone expects others to conform to it, and everyone prefers to conform to it provided that others do so too (p. 42). The definition may seem technical, but it highlights an essential point. A convention is not simply what happens frequently. It is what happens frequently because everyone takes it as a basis for anticipating the behaviour of others. This is why Lewis places such emphasis on the role of precedent. If a certain solution has worked in the past, it becomes salient. “Precedent is simply the source of an important kind of salience” (p. 36), he continues. Precedent makes one solution more visible than others, not because it is necessarily better, but because it provides a foothold for mutual expectations. That’s what we did yesterday. We’ll probably do the same today. And precisely because everyone thinks that others think this too, the pattern tends to repeat itself. Lewis describes this process as a metastable system of preferences, expectations and actions. His formula is almost chant-like: ‘we are here because we are here because we are here because we are here’ (p. 42). We conform because we expect conformity. Observed conformity reinforces that expectation. The reinforced expectation underpins further conformity. In this sense, convention does not require a solemn founding moment. It may arise from an agreement, but it is not the same as the agreement. It may be reinforced by law, but it is not the same as the law. It may take on normative value, but it does not entirely coincide with a moral norm.

Here, Lewis moves away as much from the Hobbesian view of order as a product of fear as from the contractualist view of order as the outcome of an original pact. Of course, pacts, laws and sanctions matter. But a society that had to resort to them for every minor form of coordination would be unliveable. No state could prescribe in detail everything that makes orderly coexistence possible. No contract could anticipate all the circumstances of interaction. No surveillance could replace the tacit expertise with which individuals interpret situations and regulate one another.

Before the vote of confidence

Following on from Roemer and Hollis, whose ideas we have analysed in the ‘Mind the Economy’ columns of recent weeks, David Lewis enables us to take a further step. Roemer, with his concept of Kantian optimisation, describes an agent capable of asking not only which choice maximises their own benefit, but also what behaviour would be justifiable if everyone were to adopt it. Hollis has shown that a purely prudential rationale struggles to explain promises, trust and collective action. Lewis addresses the issue at an even more fundamental level. Before full trust, before promises, before deliberation in the first-person plural, there is the problem of coordination. This point is important because it prevents an overly moralised interpretation of social life. Not every order arises from virtue, just as cooperation does not always arise from benevolence. Similarly, not every shared rule arises from a conscious collective decision. Very often we live within structures that function because they make our behaviour mutually intelligible. Convention is, in this sense, a grammar of social action. It does not necessarily tell us what is right in the ultimate sense. It tells us what one can reasonably expect in a given situation. Lewis’s more refined definition adds a decisive element: all this must be ‘common knowledge’. It is not enough that many people do the same thing. It is necessary for everyone to know – and to know that others know – that this pattern is followed and expected. Lewis writes that, for a pattern to be a convention, it must be ‘common knowledge in P’ that the conditions of the convention are satisfied (pp. 58–59). I must not merely expect you to do a certain thing. I must also expect you to expect something of me. And you must expect me to expect that you expect something of me. The formula may seem abstract, but it actually describes an ordinary structure of experience. When I walk into a classroom and see that everyone is sitting in silence, I am not merely registering a fact. I am grasping a situation. I know that the others know that a lesson is about to begin. It is this interweaving of expectations that transforms a regularity into a convention. And precisely for this reason, Lewis observes, ‘there is no convention that is the only one possible’ (p. 70). A convention is such precisely because another pattern could have served the same function, had collective expectations stabilised differently.

The line between coordinating and excluding

This apparent analytical modesty has profound political consequences. A society is not made up solely of stated values; it is made up of practical expectations. It is not enough to write that everyone is equal before the law if common experience suggests that some can circumvent it without consequence. It is not enough to proclaim merit if the informal conventions governing access, knowledge, social codes, language and self-presentation make selections even before the competition begins. It is not enough to invoke responsibility if everyday practices teach us that those who follow the rules are naive and those who bend them are clever. Here, Lewis’s reflection touches on the theme of justice. Conventions are not neutral simply because they are unspoken. They can coordinate, but they can also exclude. They can reduce uncertainty, but they can also crystallise hierarchies. They can make communal life run smoothly, but they can also render the inequalities running through it invisible. A linguistic, professional, bureaucratic or cultural convention may work perfectly for those who know it and remain opaque to those on the outside. Those who belong to the community in the know understand ‘how things work’. Those who do not belong remain exposed to embarrassment, to making mistakes, to dependence on intermediaries, and to the feeling of not being entitled to participate.

Let’s think about school. A significant part of academic success does not depend on intelligence or hard work, but on familiarity with a set of conventions. How to interact with a teacher, how to phrase a question, how to interpret an assignment, how to organise one’s time, how to voice a doubt without feeling ashamed, how to envisage one’s future within an educational pathway. For some pupils, these conventions come almost naturally, as they are absorbed within the family and social environment. For others, they are a foreign language. The right school is not one that pretends these differences do not exist. It is one that makes them explicit, teaches them and makes them accessible to all.

The same applies to work. Every organisation thrives on conventions. These include ways of speaking, response times, levels of informality, codes of reliability and rituals of belonging. Those who master them have an edge. Those who do not master them risk appearing inadequate even when they possess substantial skills. In this sense, conventions are tools for coordination, but also mechanisms for selection. They do not merely produce order; they produce recognisability. And recognisability is a powerful form of social capital.

The issue then becomes more complex. A theory of justice cannot be limited to distributing resources or formal opportunities. It must scrutinise the conventions through which a society defines who appears credible, who appears competent, who appears deserving, and who appears out of place. Many instances of exclusion do not occur through explicit prohibitions. They occur through unspoken codes that some learn early on and others too late, or never at all. Inequality, even before it manifests as a gap in outcomes, is often a gap in mastery of these conventions.

This is why Lewis’s notion of convention can enrich the line of thought initiated by Roemer. Equality of opportunity is not merely about compensating for initial circumstances. It also concerns the possibility of accessing the social frameworks that make those opportunities a reality. If a procedure is public, but its language is comprehensible only to those with the appropriate cultural resources, that publicity is incomplete. If a right exists, but the means of claiming it is socially costly, that right remains fragile.

When conventions are broken

Conventions are both robust and fragile. Robust, because once they have become established, they tend to perpetuate themselves. Fragile, because they depend on shared expectations. When we begin to doubt that others will continue to conform, the convention loses its hold. And when this loss of strength becomes visible in turn, the process can accelerate. This applies to small, everyday conventions, but also to those that underpin public life. Paying taxes, queuing in order, not abusing one’s position, not turning every procedure into an opportunity for personal gain, accepting the result of a vote, recognising the legitimacy of one’s opponent, not confusing dissent with hostility. All these practices have a normative dimension, but also a conventional one. They function as long as many people expect that many others will continue to respect them. When, however, the belief spreads that ‘everyone does it anyway’, the alternative norm begins to take centre stage. Deviance no longer appears as an exception, but as the new norm. In this sense, corruption is not merely a sum of unlawful acts. It is the destruction of a shared understanding. It does not merely cause economic damage. It brings about a shift in expectations: those who previously assumed reliability begin to assume opportunism; those who previously followed procedure begin to feel foolish; those who previously thought the rules applied to everyone begin to seek protection, shortcuts and connections. Corruption, like any profound institutional erosion, does not merely alter incentives. It changes the interpretative framework within which those incentives are understood. The same happens when public discourse loses all connection with the truth. Truth, too, has a conventional dimension – not in the sense that it is arbitrarily invented, but in the sense that public communication presupposes shared rules.

Justice as the upholding of the common world

Lewis does not offer a moral theory of justice. His aim is analytical. He seeks to understand what a convention is, and how it differs from an agreement, a norm, a rule or a mere regularity. But it is precisely for this reason that his contribution is invaluable. He reminds us that justice does not exist solely in principles, constitutions, rights and redistributive policies. It also exists in the minute expectations that make a social order liveable. It lives in the practices through which a person understands whether they can trust a procedure, whether they can speak without being humiliated, whether they can enter a space without feeling like an intruder, whether they can claim a right without appearing a nuisance. Some move through the social world as if they were at home. Others must constantly decipher it. Some can afford to be spontaneous. Others must watch every move they make. Some break a convention and are seen as original. Others break it and are judged unsuitable. This, too, is inequality: not the most visible, but one of the most pervasive. The task of institutions, then, is not merely to impose fair rules. It is also to nurture the conventions that make those rules workable. To make hidden codes explicit. To lower the cost of entry into the realms of citizenship. To prevent professional conventions from becoming castes. To prevent public language from becoming a barrier. To ensure that procedures are not only formally correct, but socially intelligible. Justice requires principles, of course. But it also requires the maintenance of the shared world.

Perhaps this is David Lewis’s most relevant lesson today. We do not live solely by laws. We live within a cloud of expectations. We are able to act in concert within the shared space because we inherit patterns that allow us to anticipate, interpret and recognise one another. But what we inherit may be fair or unfair, just as it may be inclusive or exclusionary. A democratic society must therefore prevent conventions from becoming rigid and turning into destiny. It must be able to distinguish between the patterns that coordinate and those that subordinate. Because democracy truly begins when even its unwritten rules cease to speak only the language of those who have always understood them.

(*) Professor of Economics, Department of Economics and Business, University of Cagliari