I. Introduction
The scholarly discourse on social norms in the tradition of Thomas SchellingFootnote 1 often makes a sharp distinction between social conventions and social norms.Footnote 2 The distinction is drawn roughly as follows. Social conventions are based on prudential interest. The exemplary instance is joint choice of a coordination equilibrium in a recurrent coordination game. A common illustration is whether a country coordinates on driving on the right or on the left: if driving on the right is the inherited convention then one’s prudential interest in avoiding injury sufficiently motivates compliance with the convention. There is nothing “normative” about such conventions of common interest, according to the standard view.
Social norms too are based on beliefs about what other people in the relevant reference group do, but additionally are based on one’s anticipation of social approval or disapproval of the action by others in the reference group. The exemplary instance is a social dilemma where such anticipation motivates cooperation on a better joint choice than the inferior pre-normative equilibrium based only on prudential interest. A common illustration is the tragedy of the commons. Each farmer’s self-interest tempts him to graze his cattle on the common pasture, but if each does so the commons is ruined as are all of the farmers: a pattern among the farmers of social approval for each to sustainably graze and social disapproval of any who overgraze would change the structure of the game, create a stable equilibrium of cooperation by all, and make everyone better off. Regulation of the group by the social norm resolves the social dilemma.Footnote 3
Many in the Schelling tradition are rightly fascinated by the social dilemma. It seems like it describes many interdependent human situations, but at the same time we often observe circumstances where people cooperate together rather than pursue its apparent logic to the tragic destination of “continual fear, and danger of violent death: and the life of man, solitary, poor, nasty, brutish and short." There are many ways that resolution of the dilemma can come about in practice. One of the most obvious is informal regulation within the relevant group by a social norm. The understandable fascination with the problem of the social dilemma has driven much of the revival of the study of social norms, but at the same time has biased the content of that attention.
The distinction between social convention and social norm verges on standard, and is widely taught. In attempting to teach the distinction over ten years to undergraduates, graduates, and practitioners, I encountered frequent difficulties and confusions. Although students could grasp many difficult concepts, they frequently failed to grasp the distinction or how it would apply to real situations. A large part of the problem is that the illustrative examples we would offer of social convention—driving on the right, fashion, etiquette—were not merely matters of coordinated interest but also involved social disapproval of noncompliance that disappoints the expectations of others. In retrospect, students were not expressing confusions, but were making apt objections.
My own attempt to apply the distinction to empirical practices also foundered. My entry into the Schelling tradition had to do with my initial fascination with David Hume’s account of justice and government as conventions.Footnote 4 Next, I applied ideas about social convention to the harmful practice of female genital cutting in AfricaFootnote 5: people in an intramarrying group are stuck in the harmful coordination equilibrium of all cutting in order to maintain marriageability status of their girls, but would be better off to move to a beneficial coordination equilibrium of not cutting and also maintaining marriageability status. Consistent with what I had been taught, I construed this as a matter of mere prudential interest. Over many years I learned from field investigations of my own and from empirically expert scholars and practitioners that in some groups that practice is maintained by marriageability interest, in others by social approval and disapproval, and in many by both. Still in the grip of the distinction, I was unable to model how the practice could be both a social convention and a social norm. Perhaps because my original fascination was with coordination games rather than social dilemmas I am more apt to notice the erroneous tendency in the literature to think of social norms primarily as resolutions to social dilemmas. Thousands of social norms have nothing to do with social dilemmas; for instance, the subtle expectation among the English that one begin a conversation with an observation about the weather, and that the respondent agree at least somewhat with the initiator’s observation.Footnote 6
The purpose of this essay is to urge retirement of the standard distinction between social convention and social norm and a return to a version of Ullman-Margalit’s distinction between social norms of coordination and social norms of cooperation.Footnote 7 Her book was an influential early development of the game-theoretic approach to social norms. Ullman-Margalit had much to do with establishing the standard view that an effective social norm can resolve a social dilemma. The literature did not take up her idea that in a game of coordination, a social norm works to single out the coordination equilibrium followed in the particular group.
Southwood and Eriksson take the standard view to the extreme: they hold that social norms and social conventions are essentially different.Footnote 8 They say that social norms are necessarily normative, behavior-independent, and desire-independent, but that social conventions are not necessarily normative, and are behavior-dependent and desire-dependent. I dispute their views in the final section of the essay. I point out that both social conventions and social norms necessarily depend on what one believes about others, not on others’ actual behavior. I show by plausible counterexample that social norms need not be desire-independent in their sense of that term. Finally, accepting for the sake of argument their definition of social norm and of normative, I show by example that a social norm need not be normative.
Mary Burke and H. Peyton Young, in contrast, find the standard distinction between social conventions and social norms a distraction: “We would argue that there is a constellation of internal and external mechanisms that hold norms in place, and that the salience of these factors varies from one situation to another.”Footnote 9 For example, they continue, it can be a matter of honor to avenge an insult, and failure to do so could be negatively sanctioned in a group. Someone could comply with a social norm against littering even if they were not observed and sanctioned by others. Finally, to shake hands is a conventional way of greeting, but failing to do so expresses a social meaning that others would disapprove of.
I do not mean to proclaim the true definition of social norm. Definitions legitimately vary according to one’s theoretical and practical purposes. Alternative definitions can be compared, however, on how well each serves some particular set of purposes.
II. Social Norms are Distinguished by Social Approval and Disapproval
The literature on norms and practices distinguishes among ideal types. Comparison in a tabular matrix of the concepts of sixteen different theorists of norms and practices shows that, although their conceptual schemes lump and split in different ways, generally their schemes parse the same underlying entities.Footnote 10 Here is one typology:Footnote 11
• one’s personal attitude to a matter of prudence (following a paleo diet)
• happenstance social regularities (people standing in the shade on hot days)
• social proof (going to a restaurant because it has a long line)
• social convention (common prudential interest: driving on the right in France)
• social norm (upheld as well by a pattern of social dis/approval: sustainable grazing of the commons)
• legal norm (formal and coerced by the state: jailing a thief)
• moral norm (internally motivated by individual conscience: do not harm others).
What distinguishes ideal-typical social norms from other types of norms and practices? In Elizabeth Levy Paluck’s and Laurie Ball’s admirably simple formulation, social norms are people’s beliefs about which attitudes and behaviors are typical or desirable in their community.Footnote 12 Many definitions attempt to capture both those aspects of social norms. In my view, social norms, whether of coordination or cooperation, are marked by two features; first, one complies because one believes that enough others in the reference group comply, and second, in addition to any prudential valuation of the outcome, one’s compliance is motivated by the anticipation of approval for compliance and disapproval of noncompliance of enough others in the reference group. These features are what matter, not whether the underlying incentives of the situation constitute a coordination game or a social-dilemma game. Footnote 13
I cannot develop here the reasons why I favor this social-approval implementation of the desirability aspect of social norms. I think it better avoids some conceptual problems and better lends itself to empirical identification of social norms and measurement of their change over time. The most important point to state here is that the social-approval implementation makes clear that social norms cannot be understood or measured simply as physical behavior. Here’s why.
Facial expressions are the simplest form of social approval or disapproval. From there we can move inward or outward. Moving outward, approval and disapproval become more conspicuous, including multiple modalities of expression and action. The term sanction refers to both positive and negative expressions and actions, and includes admiration or contempt, verbal approval or disapproval, praise or rebuke, compliment or insult, prize or fine, promises of physical reward or threats of physical punishment, actual physical reward or punishment, and in the extreme, threats of death or actual death, among many other things.
Compliance follows not so much from application of sanctions but more from anticipation of them. For example, one is motivated to comply if one believes that others will negatively sanction noncompliance; notice that in this situation one would comply even if one’s beliefs were false. If what a social norm commands is clear, and if each believes that negative sanctions would be strong, then we would never observe application of negative sanctions in the group: the norm is maintained by what people believe would happen if one did not comply although everyone in fact complies. Thus a social norm can exist and have force even when not behaviorally indicated by the application of sanctions.
There are more reasons why norms cannot be identified by mere observation of physical behavior. Moving inward, one could believe, even in the absence of overt sanctions by others, that others form covert attitudes of approval or disapproval toward one, and these beliefs about others could motivate one’s compliance with a social norm.Footnote 14 Economists sometimes model norm compliance as instrumentally valued reputation: one values others’ approval (even covert) in the present because that approval will maintain or create benefit from cooperative transactions with them in the future.
However, it is not unusual for one also to value intrinsically the approval or disapproval of certain others— not just those one knows face to face, but, for example, and depending on context, strangers generally, people like oneself, those highly esteemed, or even one’s ancestors, deceased parents, or future generations. Intrinsic valuation of approval or disapproval can motivate one to comply even when there are no overt sanctions and no prospect of any relationship with referents in the future.
III. Social Norm of Coordination as Choice of Convention
A central identifying feature of social norms is that compliance is motivated in good part by one’s anticipation of approval and disapproval in the relevant reference group. I will now argue that conventions, one way or another, are almost always associated with that identifying feature. If the argument succeeds, then the standard distinction between social convention and social norm serves no purpose. I offer in its place an updated version of Ullman-Margalit’s distinction between social norms of cooperation and social norms of coordination.Footnote 15 Her social norm of cooperation is much the same as what the standard view now calls just a social norm. She called the followed convention in a recurrent coordination game a social norm of coordination. The norm singles out the coordination equilibrium actually followed within the relevant group.
Some background will help make sense of her idea. Hume developed an intriguing account of convention as “a sense of interest, suppos’d to be common to all . . . where every single act is perform’d in the expectation that others are to perform the like.”Footnote 16. Following Schelling,Footnote 17 David K. Lewis developed a detailed model of convention, illuminated by the application of game theory to the topic.Footnote 18 A canonical example of a social convention is whether people coordinate on driving on the left-hand side of the road or on the right-hand side of the road. One will drive on the left (right) if one believes that everyone else drives on the left (right) and believes that everyone else believes that one drives on the left (right). There is a drive-left equilibrium and a drive-right equilibrium, and in this example it doesn’t matter which is coordinated upon, all that matters is that one or the other is coordinated upon. In a recurrent coordination game, a convention can emerge out of people’s beliefs about what others will do in the situation. One believes that everyone has a strong prudential interest in coordinating on either one equilibrium or the other. Suppose driving left is the extant convention. One believes that in a drive-left country others will not deviate from driving on the left as they would risk injury or death by doing so.
However, the driving rule can also be understood as a social norm. One who deviated would elicit social disapproval possibly accompanied by overt negative sanctions, not just from those endangered by the deviation but by third parties as well. Anticipation of social disapproval could be a reason for compliance with the rule. It is also a legal norm in that the state formally prohibits deviation and threatens to punish it. Additionally, individuals could be motivated to comply by the moral norm not to risk the injury and death of others. Generally, it’s possible that for any individual, the prudential, social, legal, or moral motivation, or some combination of them, would suffice for compliance. And the set of motivations could vary across individuals in the group. In this example, one might speculate about compliance with the rule in the absence of one or more of the motivations. Social regulation is likely not that strong, since cars don’t exhibit the same cues as humans. The moral motivation could be stronger than the skeptic would expect: one can walk on the side of the road expecting that nearly all drivers (there are malicious exceptions) will take care not to hit you. Nevertheless, all drivers have a strong prudential motivation, and if there were neither moral nor legal regulation, it seems likely that prudential motivation would suffice for compliance.
Ullman-Margalit would say that a social norm of coordination prescribes one of the coordination equilibria (and proscribes any others): Hereabouts, we all drive to the left. If we are already under a convention where past precedent leads one to believe that people will select the same equilibrium indefinitely, it may seem redundant to state that there is a social rule to drive on the left. But for those new to the context, children, or strangers, it would be useful to let them know that the rule here is that we keep to the left. Social norms of cooperation as well are often obvious to insiders but it’s useful to be able to state the rule to new arrivals.
IV. Four More Ways Social Norms of Coordination are Influenced by Social Approval and Disapproval
Beyond Ullman-Margalit’s idea of social norm of coordination as choice of convention, there are at least four more ways that conventions stabilized by mutual prudential interest are also maintained or shifted by social approval and disapproval, supporting my proposal that what are standardly called social conventions would better be called social norms of coordination. One way that happens was pointed out originally by Lewis: all else equal and as a presumptive reason, others legitimately expect one’s compliance with a social convention. Next, social approval and disapproval can also assist in the transition from an inferior coordination equilibrium to a superior one, or from an unfair coordination equilibrium to a fair one. Of wider scope, when an action has a conventional meaning in a reference group, standing for something valued such as politeness, virginity, marriageability, equal respect, or courage, not to comply with the convention invites covert attitudes of social disapproval and overt sanctions.
A. Legitimate expectation of compliance
The first comes from David Lewis, who argued that conventions are a species of norm that one ought to comply with. If there is a convention in population P, the actions and preferences of most members of P probably imply that I have reason to conform to the convention and most others have reason to expect me to conform, according to Lewis.Footnote 19 Moreover, (all else equal) if one is expected by others to conform, then failure to conform would evoke unfavorable responses from others. Anticipation of such responses would strengthen one’s conditional preference for complying with the convention.
Some conventions are of great importance, such as which side of the road to drive on or the runway lights that guide the night landing of passenger aircraft. Some are of small importance, such as remembering in conversation with an Italian that the word for cat is gatto rather than cano.
I propose that members of the group are more entitled to rely on the compliance by others with an extant convention
1) the more uniform is the convention. If everyone in the population complies in every instance, then the convention is perfectly uniform. The convention is less uniform if some of the population complies, or the population complies in some instances, or both.
2) the greater the expected value of loss from miscoordination.
The two constraints tend to be correlated, but they need not be so.
Thus, one can comply with a convention not just from prudential interest but also because one believes that others are entitled to rely upon such compliance. If one does not comply such that the legitimate expectation of one or more others is disappointed, then members of the group can disapprove of that noncompliance and possibly manifest disapproval of it with negative sanctions. The rebuke would be extreme for failing to get the airport landing lights right; using the wrong word that confused another in conversation would elicit shallow and transient disapproval from the interlocutor.
As above, the approval or disapproval of others is one more reason to comply for those who already have a prudential reason to comply. Next, we shall see, there are ways that social approval and disapproval can be a reason to comply for those who do not have a prudential reason to comply.
According to Francesco Guala, some scholars dismiss the Lewis account of the normativity of conventions, because the normativity is external to the convention, deriving, all else equal, from both the oughtness of instrumental rationality (I should do what I prefer) and the oughtness of not making others worse off (I should do what others prefer me to do).Footnote 20 Social norms proper are intrinsically normative, but Lewis’s conventions are extrinsically normative, and that is too weak, according to these scholars. In a controlled laboratory experiment Guala had two human subjects play a coordination game together for nine rounds; on the tenth round he changed the payoffs for one player such that she would gain from departing from what had been the conventional choice. If convention is only a matter of prudential interest then that player should depart from the established convention; however, only 29 percent of such players did so. Guala suggests that this supports the view that convention—recurrent choice in a coordination game creating expectations in others—can be intrinsically normative.
Next, formal definitions of social convention tend to obscure the reality that a portion of the population is not sufficiently motivated to act in their own interest, the interests of the general population, or both. In part, a social convention exists in a population P, if everyone,Footnote 21 almost everyone,Footnote 22 most,Footnote 23 or a sufficiently large number of peopleFootnote 24 believes that others comply with it. In P everyone, almost everyone, or most have approximately the same preferences regarding all possible combinations of actions (Lewis), or for a sufficiently large subset the situation is a coordination game without non-strict Nash equilibria (Bicchieri). What about the deviants who don’t follow the rule? For Lewis, if most conform, then they prefer that even more conform. Presumably, for Bicchieri, the noncompliant must not consider the situation a coordination game without non-strict equilibria; the noncompliant nevertheless are in the same reference group as the compliant and thus I infer that the compliant interact with the noncompliant.
Let’s say that, for a convention to be stable, enough of the population recurrently engaging in the interaction must conform with the convention. Even if everyone believes that the convention is quite stable, those who do conform to the convention sometimes interact with those who do not comply. In those interactions, their expectations are disappointed, they are inconvenienced or worse, and those thus harmed or third parties in the group could disapprove of and negatively sanction the noncompliant. For another convention, just barely enough of the population complies for it to be stable, and those who do comply accumulate losses from miscoordination. Those who comply would strongly approve of and might positively sanction others who comply, and would strongly disapprove and might negatively sanction others who do not comply. They fear that otherwise the useful convention would collapse.
Deviants might fail to comply with a social convention because of weakness of will, incompetence, carelessness, perversity, not sharing in the common goal, or youthful or outsider ignorance of the convention. Here are a few examples. For Hume, rowing a boat together was a good illustration of a social convention. Indeed, both rowers have a strong interest in coordinating on pulls, otherwise much effort is wasted. One may be paired with a lackadaisical younger cousin, however, who is clumsy and easily distracted by the sights. It’s in his interest to coordinate, but he is both incompetent and incontinent, or weak of will. Both rowers would be better off if the more competent and continent rower were to chastise, train, and focus her partner. Elsewhere, some might be ignorant of the convention. In rural Senegal I broke a rule about how to walk by the side of the road in the midst of heavy pedestrian and vehicle traffic, and was angrily rebuked by a local. I learned that I did break a rule, but my translator could not understand what the content of the rule was. I wanted to comply but I couldn’t.
B. Failure to coordinate on the best convention
Often discussions of social convention assume that choice of coordination equilibrium is a matter of indifference, for example that for everyone to drive on the right is as good as for everyone to drive on the left, and that either is better than the disequilibrium of some to drive on the right and some to drive on the left. Sometimes this observation leads to the conclusion that evaluation is arbitrary, always relative to a particular group of people. Some of the Sophists who roused Plato’s derision held this view, but Aristotle understood both that conventions implement an underlying value and also that one convention can implement the value better than another. He writes, in the Nicomachean Ethics (Book V, Section 7),Footnote 25
Of political justice part is natural, part legal, natural, that which everywhere has the same force and does not exist by people’s thinking this or that; legal, that which is originally indifferent, but when it has been laid down is not indifferent, e.g. that a prisoner’s ransom shall be a mina, or that a goat and not two sheep shall be sacrificed . . . .
First, if the choice of one convention or another is a matter of indifference, still the convention adopted has to do with some valuable purpose.
Now some think that all justice is of this sort, because that which is by nature is unchangeable and has everywhere the same force (as fire burns both here and in Persia), while they see change in the things recognized as just. This, however, is not true in this unqualified way, but is true in a sense . . . . The things which are just by virtue of convention and expediency are like measures; for wine and corn measures are not everywhere equal, but larger in wholesale and smaller in retail markets. Similarly, the things which are just not by nature but by human enactment are not everywhere the same, since constitutions also are not the same, though there is but one which is everywhere by nature the best.
Second, even with drab conventions of measure, larger measures are more valuable in wholesale markets and smaller measures in retail markets. And although constitutions conventionally differ, they also differ in the justice each provides, and there is one constitutional convention that is best by nature.
One convention can be better for all than another convention. And one can’t assume that people are coordinated on the best convention. Varieties of coordination failure, including failing to coordinate on the best convention, are surprisingly frequent in controlled human-subject experiments,Footnote 26 as well as in everyday life (English orthography).
Sweden followed the convention of driving on the left. After World War II, better road and ferry connections to right-driving Europe made driving on the right a better coordination equilibrium than driving on the left. Sweden switched from driving on the left to driving on the right; a law proclaiming the plan to switch and a four-year education and road engineering program culminated in a switch from left to right as of 5 a.m., September 3, 1967. More recently Samoa switched from right to left, because its neighboring countries drive on the left.
A coordination game in which one equilibrium is Pareto-superior, better for all, than a second equilibrium we call a Hi-Lo coordination game. A group can discover that an alternative convention is better than the one it is currently following, but that does not entail that the better convention is adopted. The population is stuck in an inferior convention. No one or few people can make the move to the superior convention on their own; in the extreme case of the driving rule any who tried would be injured or killed.
People most like if everyone drives on the right, next like if everyone drives on the left, and least like miscoordination where they crash into one another. Preference orders over four states of the world are as follows.
The two-by-two matrix represents those preference rankings of two actors over four states of the world.
$$
The upper left quadrant is the state of the world where Self drives on the right and stylized Others also drive on the right. Self’s payoff in a world where all drive on the right is shown in italics in the lower left hand corner of the upper-left box. Other’s payoff in a right-driving world are shown in bold in the upper-right hand corner of the upper-left box. The lower right-hand box shows the state of the world where Self drives on the left and Other drives on the left. Each values the all-left equilibrium less than the all-right equilibrium, but values the all-left equilibrium more than the miscoordination represented in the lower-left or upper-right boxes. An equilibrium state is one where no single player has an incentive to deviate. To check, if everyone is driving on the left, would Self make herself worse-off by driving on the right? No. Would Other make herself worse-off by driving on the left? No. There are two coordination equilibria in this game, all drive right and all drive left. The equilibrium states are marked in the matrix by an asterisk.
Even though Sweden’s shift from left to right was coordinated by legal decree, especially as to the exact moment of shift, it took years of publicity and education to prepare for the change in reciprocal expectations. Even with a law in place, there was likely also social and moral regulation of the change. A short time before the shift, those driving on the right would not only endanger others, but also be subject to social and moral regulation. A short time after the shift there was likely high social and moral regulation supporting the change given the inertia in individuals’ daily habits.
In another coordination situation the shift from inferior equilibrium to a superior equilibrium need not be so instantaneous.. I applied a simple dynamic model of convention shift to the end of footbinding in China and proposed that the similar methods would help end female genital cutting (FGC) in Africa.Footnote 27 People found themselves in the inferior equilibrium where all families bind their daughters’ feet. Everyone would be better-off with natural feet. The details of the model correspond to the history of rapid abandonment of footbinding and to instances since its publication of the abandonment of FGC.
C. Failure to coordinate on the fairest convention
The battle of the sexes is a commonly discussed game of coordination and conflict. Both male and female want to go to an event together, but the male favors one event and the female the other. The female prefers going to the event favored by the male rather than not going together to an event. Unfortunately, this game describes many of the conventions of marriage, the extant conventions tending to be ones that favor the male.
$$
If enough males come to value fair marriage arrangements, through acceptance of public justification, a new pattern of social approval and disapproval, or both, then women and men would become better-off by shifting to the fairer convention. After valuations have changed, a further effort, probably involving social regulation, is required to move from the now inferior convention of male-advantaged marriage to the superior new convention of fair marriage.
$$
D. Conventions of social meaning
There are two usages of the term social convention that I think are sometimes confused. In its ordinary usage the term often refers to recommended but not mandatory social norms. The implicit distinction is that there are weakly enjoined social conventions and strongly enjoined social norms. The other usage is game-theoretic, in the tradition of Schelling, Lewis, and beyond.
Authors sometimes describe fashion or etiquette as social conventions in the ordinary sense (for example, “we also abide by conventions of etiquette, dress, eating, and so on”Footnote 28) and then struggle to explain them in the game-theoretic sense. What is the joint prudential interest in a convention of fashion? One complies with dress requirements because, for some reason, it is in her prudential interest to dress like others do, it has been suggested. Not in my experience. I have almost never wanted to dress like my peers, but I frequently do because of my beliefs about others’ approval for complying and disapproval for not complying with the dress code. Any significant deviation from the fashion norm elicits remarks, teases, and insults from some of one’s peers on a matter that is of no prudential interest to them whatsoever. There are fashion norms, socially enforced. Another prudential interest explanation is that one conforms to a social convention of etiquette in order to realize the joint prudential interest of demarcating one’s group from inferiors. But when you violate a rule of etiquette, no one accuses you of blurring the boundaries of the group. They say that you have offended someone and they might rebuke you for it. And the lowest group of inferiors follow their own rules of etiquette even though they have no lower group to demarcate themselves from.
If we think of the social norm primarily as a device that works to resolve social dilemmas and similar situations we leave out the thousands of little social rules that are maintained in part by mutual coordination interest and in part by social approval and disapproval related to their social meaning. For example, a popular book by Oxford anthropologist Kate Fox relates hundreds of mundane social rules in English life, having to do with: talk about the weather, ordinary conversation, rules of humor, linguistic class codes, mobile phone use, pub talk, home rules, rules of the road, rules at work, rules at play, dress codes, food rules, rules of sex, and rites of passage.Footnote 29 Many of these rules do not regulate situations of the social-dilemma type. Nor do many of them concern mutual prudential interest in coordination with no further purpose. Rather, I suggest, most of them are conventions of social meaning. I suspect that the bulk of social norms are of this type.
When an action has a conventional meaning to others in the reference group, standing for something valued, such as politeness, virginity, marriageability, equal respect, courage, and so on, not to perform the action means (regardless of intention or truth) impoliteness, harlotry, unmarriageability, disrespect, or cowardice. In a sound-sense game we want to coordinate on a label for the creature, either the cat or il gatto. Here, the label is not an utterance but an action. The action has a social meaning.
To wear a red dress to ball conventionally means to others harlotry, lack of chastity and fidelity, unmarriageability, regardless of the intention of the wearer. This is unforgettably portrayed in the 1938 movie Jezebel. The antebellum southern belle played by Bette Davis rashly wears a red dress to the ball, which upon her entry provokes searing contempt and scorn, ruining her and driving away her fiancé. In New Orleans, maidens must wear white to the ball; to wear a red dress means that one is a prostitute. To wear a white dress to a wedding means to others virginity, chastity, fidelity, regardless of the intention of wearer, and regardless of the truth. With honor killing the issue is social meaning, not truth. The girl will be killed even if there is false gossip about her virtue, gossip that most know to be false (this is the result of extreme positional competition as to chastity and fidelity). To talk with a boy in the square means harlotry, dishonor to her group and unmarriageability of her, her sisters and cousins, until cleansed by death. The girl knows that it means that, and insiders believe that her knowing that makes her culpable; she must do absolutely everything to avoid inspiring even false gossip.
A usual observation about etiquette is that there is a common goal but there are alternative conventions that signify it. A wonderful example is how to show respect to those more highly ranked in society. In one culture, you walk behind the chief in order to give him temporal precedence; to walk in front would show disrespect by showing him your backside. In another culture, you show respect by walking in front of the chief, in order to intercept any magic poison that could kill him.Footnote 30 One or the other convention will do, but not both! Following the right convention in the right context means to show respect. If one fails to show respect one can anticipate disapproval from second and third parties. Being a foreigner is a good excuse, but not always. If social norms have to do with social approval and disapproval, then conventions of social meaning are social norms, even though they have nothing to do with resolving a situation like a social dilemma.
Goffman says of etiquette that it is “a conventionalized means of communication, of secondary or no significance in its own right, which expresses one’s character or conveys his appreciation of other parties in the situation.”Footnote 31 Felicia Ackerman says of politeness that, “for its intended beneficiaries [reference group], the goal is to make social life orderly, predictable, comfortable, and pleasant,” doing so by making social life aesthetically appealing; minimizing embarrassment, hurt feelings, and unpleasant surprises; showing consideration for others, respecting their autonomy and privacy; providing security of conventional forms; reflecting distinctions of rank and privilege—all conventionally signified. Finally, “it is socially sanctioned to take a violation of the rule by other people in situations involving oneself as an affront to oneself.”Footnote 32 These conventions of social meaning are clearly social norms, I suggest.
Impoliteness, even if it is unintended, means that one does not share the common goal of making the situation orderly, predictable, comfortable, and pleasant. The way politeness is shown is conventional and differs from one group to another, but not to show politeness is to violate a social norm. Complying members would disapprove and negatively sanction a noncomplying member, including exclusion from the group or even death. One way to be deliberately rude is to violate the local social norms of politeness in order to assert subordination of members of the group or otherwise to express disrespect for them. Elsewhere, there can be a social norm of rudeness, such as in sports competition training where insults, name-calling, and obscenities are considered appropriate and believed to be ultimately beneficial for those involved.Footnote 33
Lessig too offers a concept of social meaning. For example, some time ago, to wear a seatbelt in a Budapest taxicab had the meaning that one did not trust the driver and thus was an insult. That was a harmful convention of social meaning. Around 1900 in the United States the new practice of tipping was thought to be corrupting of all involved and insulting to the recipient. Now, its social meaning is an obligation to support otherwise underpaid hospitality workers. To fly a Confederate flag in the American South unmistakably means to many black southerners, and is intended to mean by some whites, slavery, Jim Crow, and caste subordination.Footnote 34
E. Why social norms of coordination and of cooperation?
I have argued, contrary to the standard distinction between social convention and social norm, that social norms of coordination are influenced by social approval and disapproval, just as are social norms of cooperation. Why not collapse norms of coordination and cooperation then? We could, but there is a good reason to maintain the new distinction. If we are working to shift from harmful social norms to more beneficial ones, then for science and policy purposes it is useful to know whether the situation is one of coordination or of cooperation. Footbinding in China was a convention that people abandoned for the alternative of natural feet, indicated by the fact that once the practice ended in an area, no one was tempted to revert to its use. The change was stable because it was self-enforcing. The city and people of Bogotá, Colombia in the 1990s overcame severe disorder and violence with a program of harmonized change in moral, social, and legal norms. Civil peace is not entirely self-enforcing, however. It is more like a game of cooperation that requires ongoing moral , social, and legal regulation for its stable continuation.
V. Are Social Conventions and Social Norms Essentially Different?
So far I have argued that the standard ideal types of social convention and social norm are more usefully distinguished in theory and practice as social norms of coordination and social norms of cooperation. Southwood and Eriksson argue that social conventions and social norms are essentially different (hereafter, I shall refer to theirs as the ANU view because it was developed at the Australian National University). Social conventions are not normative, and are behavior-dependent and desire-dependent, they say. Social norms are normative, and are behavior-independent and desire-independent.Footnote 35
A. Counterexample to Lewis’ claim that conventions are a species of norm
Southwood and Eriksson offer a counterexample to Lewis’ arguments that social conventions are a species of social norm. A number of people are eager to meet one another on a weekly basis at one coordinated time and place. Each believes that the others will be at one coordination point, each desires that everyone including himself be there, and this is common knowledge among them. However, each of them “has a peculiar loathing for normatively binding social arrangements . . . . [T]hey only want to meet in this way conditional on its being clear and explicit that there is no requirement to turn up and no criticism or censure that flows from their not doing so . . . . [T]he meeting must be, so to speak, optional on all sides,” and this is common knowledge among them.
There are two ways to challenge this example, each showing that the participants’ goal to meet together requires social regulation to attain. First, the example presented by the ANU view concludes that participants do manage to coordinate on meeting together. I argue that, given the details of the example, to coordinate on a meeting would under quite ordinary assumptions require social regulation. Can the vignette be described as a coordination game with more than one strict equilibrium? Yes, they could meet Mondays at 12, or Wednesdays at 1, and so on. In the recurrent coordination game would people be able to establish a convention with one another? Yes, but likely the one inferior for all, defeating the mutual desire to meet together. Suppose that all (or, alternatively, enough) would be better-off meeting at A, and next better-off meeting at B. If all but one (or, alternatively, a few) are meeting at A, and one goes to B, then all are worse-off through miscoordination than if everyone had met at A. If all but one are meeting at B, and one goes to A, then all are worse-off through miscoordination than if everyone had met at A. Next, I believe it’s unobjectionable to assume that it takes more effort to go to either A or B and less effort simply to go your own way. That leaves a third choice: for each to go her own way, which is better for everyone than the miscoordination that results whenever only one person fails to meet at A or only one person fails to meet at B. There are three equilibria: all meet at A, all meet at B, and all go their own way.
$$
If at time 1 all coordinate on meeting at A, would a convention emerge to meet at A? That’s hard to say. At time 2, Nemo would choose A if he believed that all others would choose A. But would they? If only one were not to show, then Nemo would face the penalty of miscoordination. He would be better-off by going his own way than he would be under circumstances of miscoordination. Everyone else reasons in the same way. Any consideration of an initial convention is fragile to anticipation of weakness of will or mistake on the part of only one of the other members. If Nemo is contemplating meeting at A (or at B if there were reason to believe others would be assembling at B) he would reason as follows: if all others are going to A I will go to A, but if any other person goes her own way I will be worse-off by going to A than if I go my own way. In other words, the option to all go their own way—the safest choice for each if there is too much uncertainty about what others would do—would be selected rather than to all meet at A—the better choice for each if there is enough certainty about what others would do. If people in the group knew that a regime of social approval and disapproval were in the background, then that would remove the temptation for each to go her own way and thereby stabilize the best convention.
This is a form of the stag hunt game. Ullman-Margalit remarked that as in the social dilemma game, “the jointly satisfactory outcome produced by joint cooperation stands in need of some assistance . . . . by their feeling themselves under some kind of obligation to keep cooperating . . . or, alternatively, by effective external sanctions. So in situations of the stag-hunt type, too, it might be expected that norms will be generated.”Footnote 36
Second, the vignette is baffling. Suppose I ask you to wake me at 6 a.m., and you say, “Sure, but if I fail to do so, don’t react negatively: because of my peculiar upbringing I can’t agree with you to do something unless you agree not to rebuke me for not keeping my agreement. I’m just not into normative practices.” I respond, “What a coincidence, I had the same upbringing, and I’ll agree not to rebuke you only if you agree not to rebuke me if I do happen to rebuke you.” Whatever you and I are talking about, it becomes clear that neither of us are talking about the practice of reaching an agreement. Each of the friends who would meet for lunch will do so only if each is required not to criticize or censure someone who doesn’t show up. However, what would happen to one who broke that rule and censured others for not showing up at the lunch? Nothing. Because of their peculiar upbringing no one would be able to censure someone who censures. A requirement not to require people to comply with requirements cannot be expected to be effective. The hovering threat of abhorrent rebuke would prevent the anti-normatives from ever attempting to coordinate on meeting together for lunch, even though all want to so coordinate.
B. Are social norms behavior-independent but social conventions behavior-dependent?
The ANU view holds that a social norm is behavior-independent, that is, a social norm can exist even if no one complies with it. Their (regrettable) example is a norm in a certain country not to urinate in the swimming pool. The example trades on the undetectability of transgression: everyone could believe that others should comply and could believe that most others are complying but be factually mistaken about whether others are compliant. The norm could exist even though its targets do not actually comply. But a social norm is not constituted by whether enough others actually comply, but by whether one believes that enough others comply and believes that enough others would approve or disapprove. Social norms are distinct from social conventions, they say, because norms are behavior-independent and conventions are not. But it seems to me that conventions are behavior-independent in just the same way that norms are behavior-independent. Each is dependent on one’s beliefs about others. A convention as a solution to a recurrent coordination problem is based on participants’ beliefs that enough others will select the same coordination equilibrium on the next iteration. And as a conceptual possibility there could be conventions deviation from which is undetectable. Everyone could believe that the convention is followed but be mistaken. That is plausible. Where I went to elementary school in America there was a convention to wear green on St. Patrick’s Day. It was a conventional way to respect the Irish heritage. Failure to wear green meant that one showed disrespect. Since no one could be made to show their underwear, an effective way to show support was to claim that there was green in one’s underwear, and many did not wear any detectable green.
C. Are social norms always normative and social conventions not?
The ANU view also holds that social norms are essentially normative, and that social conventions can be normative but need not be. To challenge this claim we have to show that there is a social norm that is not normative in the sense they intend. What do they mean by normative? For them, social norms are rules or normative principles accepted in a particular group. They do not include objectively valid normative principles that apply to all agents (moral norms). One essential feature of a social norm is normative principles, which require that the target must do what is enjoined. A second essential feature is that the principle applies among those in a particular group. The principle is accepted among enough members of the group. Acceptance has to do with “certain kinds of normative attitudes: being disposed to regard certain responses as appropriate and inappropriate, the ‘right ones’ and the ‘wrong ones’; being disposed to criticise those who do wrong and to regard such criticism as legitimate, and so on.”Footnote 37
This account neglects the existence of harmful social norms, which are not a tortuous logical possibility but rather are quite real. I am reliably told that in a certain city it was the social norm for a teenage boy to beat up his girlfriend if she was suspected of flirting with another boy. On further investigation it turned out that many of the boys did not want to beat up their girlfriends for suspected flirtations, but felt strong social pressure to do so. They experienced and anticipated attitudes of disapproval and negative sanctions from some of their peers for failure to comply. Is this social norm accepted, that is, considered valid or correct by the boys and girls in this group? Would each member of the group consider the social norm to be appropriate or right? Would each consider criticism of those who failed to beat their girlfriends as legitimate?
I think not. Sometimes the behavior enjoined by a social norm is accepted by some of those whom it targets, but the behavior enjoined can also be contrary to the personal attitudes or even the moral commitments of those targeted. Social norms are not uniformly moral in content. In Germany long ago, the Nazis changed not only the laws, but also deliberately engineered the adoption of new social norms, for example, pertaining to one’s association with Jews. Shunning Jews could be contrary to one’s moral commitments, but still mandated given overwhelming social pressure that would interfere with the performance of other moral obligations (just this happened to Haffner, to his surprise).Footnote 38
Social norms can be accepted, but more generally they are complied with. The member of a group can believe a social norm is appropriate, but more generally she believes that enough others consider it appropriate. A member can think compliance with a social norm is right, but more generally the typical member believes it is approved or disapproved of by others in the group.
But at least some of those governed by the norm must have such normative attitudes? Not necessarily. Pluralistic ignorance Footnote 39 is the idea that many could comply with a social norm because they believe most others accept it even though most in fact privately reject it. It is not a fanciful concept; the social norm I mentioned of beating up one’s girlfriend for flirtatious behavior could have been maintained by pluralistic ignorance. And it is conceptually possible that everyone in a reference group could comply with a norm because they believe others approve of it and would overtly sanction noncompliance even though in fact none of the others do. Hence, the typical individual’s normative attitudes are not essential to a social norm; a norm rather has to do with the typical individual’s beliefs about the approval and disapproval of others in the reference group, and those beliefs can be mistaken.
James Coleman taught that not all social norms are conjoint, where the group that expects the behavior is identical to those expected to comply with the behavior; some norms are disjoint.Footnote 40 An example of a disjoint norm would be men enforcing a social norm prohibiting women from speaking in public. The men approve, disapprove, and sanction; the women are approved of, disapproved of, and sanctioned. The women might consider their subordination acceptable, appropriate, and right, but even if they did not they would comply with the norm because of social pressure from the men. Or, sanctioners and sanctioned can overlap. Consider the Jim Crow code of racial etiquette, for example, that a black can’t offer a handshake to a white; that whites must address blacks by first names and blacks address whites as Mr., Mrs., Sir, Ma’am; that a black may never suggest that a white is lying; that a black can never laugh at a white; and so on. A norm like this can be enforced both by whites and, evidence suggests, by blacks amongst themselves, not out of internal acceptance but in order to spare oneself and one’s kind from suffering more brutal punishments by whites. Are normative principle, normative attitude, appropriateness, rightness, general acceptance, and general legitimacy of punishing noncompliance the first ideas that come to mind in struggling to describe and explain such norms?
ANU overmoralizes social norms. In their scheme, for both the moral and the social norm, the normative attitude is within each individual, who applies it to self and others. I have argued that a social norm is in part distinguished by the feature that compliance with it can be motivated only by one’s beliefs about the approval or disapproval of others; personal motivation of any kind to comply can be insufficient or even absent. If one happens to endorse the content of a reigning social norm, and has sufficient personal motivation to comply, such social regulation might go unnoticed; but were one to retract that endorsement and lose personal motivation one would suddenly become aware of the surrounding social pressure. Suppose that the social norm in a group is to drive home drunk from parties: that’s just what people do. When through new information from elsewhere one comes to believe that drunk driving is too likely to harm others, one then sadly realizes that to state, follow, and invoke that view in this group would occasion unbearable ridicule and slighting by the others.
D. Are social norms always desire-independent and social conventions not?
Finally, ANU argues that social conventions are desire-dependent, but that all norms, including social norms, are desire-independent. The idea of desire-independent reasons to act seems to be borrowed from Searle. The issues are difficult and controversial; also, I find the ANU reasoning difficult to understand. In this space I am unable to review and respond to their specific arguments and must offer instead a brief response to their central proposition.
I have argued that many social conventions are supported by beliefs about social approval and disapproval, and have proposed to call them social norms of coordination. Perhaps one could concoct, at the expense of triviality, social conventions the violation of which would fail to elicit any social approval or disapproval in the relevant group. I would be happy not to consider those as social norms of coordination.
Next, consider a social dilemma and a corresponding social norm that resolves the dilemma by motivating enough people to coordinate on cooperation. Given the stipulated set of preferences, an actor could also “desire” to cooperate if (she believes) enough others would cooperate. One way for the typical actor to be assured that enough would cooperate is if she had reason to believe that enough people are disposed to approve of those who cooperate and disapprove of those who defect, and reason to believe that enough people would be motivated to cooperate by their “desire” for social approval and aversion to social disapproval. ANU would not consider the desire to gain approval or avoid disapproval a normative motive. I have provided a desire-based account of a social norm of cooperation.
Searle rejects the idea that every intentional action is an expression of a desire to perform that action. Rather, he says, desire as mere inclination is one reason to act, and there are reasons other than mere inclination that agents can create, for example, obligations, that provide a secondary desire to act that may outweigh or displace a mere inclination (primary desire) to the contrary. A way for a social norm of cooperation to exist in the ANU sense is if each of the parties is an unconditional cooperator and believes that enough others are. One could call unconditional cooperation a desire-independent motivation. But the desire-independent way is not the only way for a social norm of cooperation to exist.
VI. Conclusion
Hence, I conclude that the ANU attempt to establish that social conventions and social norms are strictly different does not withstand challenge. I have also argued that the more relaxed view of social convention and social norm as ideal types is of little use. I think that the idea of social norms of coordination and of cooperation is more useful for theorization, teaching, research, and practical application.
