Stratified social norms

Abstract

This article explains how social norms can help to distinguish and understand a range of different kinds of social inequality and social hierarchy. My aim is to show how the literature on social norms can provide crucial resources to relational egalitarianism, which has made social equality and inequality into a central topic of contemporary normative political theorizing. The hope is that a more discriminating and detailed picture of different kinds of social inequality will help relational egalitarians move beyond a discussion of the justice or injustice of social equality as a single general category.

1. Introduction

One of the main reasons for moral and political philosophers to take an interest in social norms has been as an explanation of cooperative behaviour. As much as social norms can motivate people to act in their common interest, however, they can also lead to the marginalization of individuals or groups, put people in positions of influence and control over others, and function as systems of oppression and domination. This article explains how social norms (and Bicchieri’s theory of social norms in particular) can help to distinguish and understand a range of different kinds of social inequality and social hierarchy. My aim is to show how the large and well-developed literature on social norms can provide crucial resources to relational egalitarianism, which has made social equality and inequality into a central topic of contemporary normative political theorizing. The hope is that a more discriminating and detailed picture of different kinds of social inequality will help relational egalitarians move beyond a discussion of the justice or injustice of social equality as a single general category.

2. Social Norms

When I take the tube in London I am subject to a norm requiring me not to look at other travellers. One looks at advertisements, shoes, one’s phone or book, but it is unseemly to look directly at people. A key feature of social norms is their conditionality: in a sense to be explained, social norms exist, apply, or are in force because they are accepted or endorsed in the relevant population or group. Tube travellers generally don’t directly look at others, or at least not for too long, they generally expect others not to look at them, and they generally believe that people shouldn’t look at other people. If it weren’t for the prevalence of such behaviours and attitudes, nothing would be amiss in looking at one’s fellow passengers. Of course, one can do so in intrusive or threatening ways and there may not need to be a social norm to make such behaviour unacceptable, but one can easily imagine, or experience for that matter, more socially interactive yet unproblematic public transit systems than the London tube.

One of the main differences between different philosophical accounts of the nature of social norms is how the conditionality of social norms is understood. In this paper, I will rely on Bicchieri’s account of social norms. Bicchieri provides the following conditions for the existence of social norms:

Let R be a behavioural rule. We say that R is a social norm in a population P if and only if there exists a sufficiently large subset P _cf of P such that, for each individual i in P _cf ,

Conditional preference: i prefers to conform to R on the condition that:

1. (a) Empirical expectations: i believes that a sufficiently large subset of P conforms to R;

and

1. (b) Normative expectations: i believes that a sufficiently large subset of P believes that i ought to conform to R (and may sanction non-compliance).

Bicchieri then adds that a social norm R does not merely exist but is followed by population P if there exists a sufficiently large subset P _f of P _cf such that, for each individual i in P _f conditions (a) and (b) are met and, as a result, i prefers to conform to R. ¹ Since I’m not comfortable with the implicit behaviourism in labelling rules that people prefer to conform to as “followed” I will refer to such norms as operative social norms instead.

Ignoring all manner of subtleties and exceptions, the rule in my tube example would say ‘do not look at other travellers’. Sufficiently many members of the relevant population (London tube travellers) prefer to conform to this rule conditional on their belief that sufficiently many others will also follow the rule, and on their belief that sufficiently many others believe that they ought to follow the rule (and may impose various sanctions). It is hard to say whether the preferences of most tube travellers really are conditional in this way, but I for one would on many occasions (too tired or carsick to read, for example) rather be people-watching than staring at shoes or ads if many others would do the same, or if I thought that hardly anyone would disapprove. If sufficiently many people are like me, then the ‘do not look at other travellers’ rule would be a social norm on Bicchieri’s account.

The empirical expectations and normative expectations conditions are Bicchieri’s way of capturing the conditionality of social norms. Competing accounts do this in different ways. Brennan e t al., for example, hold that social norms are patterns of normative attitudes that are grounded in presumed social practices for the individuals who hold those attitudes (2013: 72). This article does not adjudicate between Bicchieri’s and other accounts of social norms. Starting from Bicchieri’s work, I aim to show how we can account for a range of hierarchy and inequality related social phenomena. I take Bicchieri’s work as my starting point because it provides the most well-developed, influential and empirically tested account of social norms available in the current literature. In addition, my discussion in section 3 makes use of some distinctive features of Bicchieri’s view, such as the distinction between normative and empirical expectations, and the way in which individual preferences are conditional on beliefs about the members of a specific reference class of individuals. Nonetheless, for all I say in this article, other theories of the nature of social norms may provide equally good bases for thinking about social hierarchy and inequality.

I make one amendment to Bicchieri’s account: in my view, social norms can demand not only that individuals behave in certain ways, but also that individuals display various dispositions, emotions, feelings, or other attitudes. Social norms surrounding the death of loved ones, for example, may demand that we grieve them, where grief is not simply a matter of displaying certain kinds of behaviour. Bicchieri’s view can straightforwardly account for norms that demand attitudes if we simply allow ‘R’ to be rules that demand attitudes and not just behaviour. ² This will matter for some of the claims I make below, although much could be reformulated using social norms limited to behaviour alone.

It will be important for my discussion later to briefly discuss the sense in which social norms are normative. One way to consider this issue is in terms of the reasons social norms give individuals to comply with the rule. Enoch (2011: 4–6) distinguishes between three ways in which people may be given reasons. For illustration, imagine you are a patient suffering from gout. First, suppose your doctor tells you that you should refrain from eating organ meats. Here, the doctor gives you a reason to refrain from eating organ meats in an epistemic sense: you have a reason not to eat organ meats because you suffer from gout, and your doctor’s statement makes you aware of this reason. Second, suppose that your doctor tells you instead that she will prescribe treatment for your gout, but only if you pay her 50 guilders under the table. You are given a reason to pay the doctor. In this case, you have a standing conditional reason to pay-the-doctor-if-treatment-is-conditional-on-payment. The doctor gives you a reason to pay by triggering this conditional reason. Third, imagine you promise the doctor that you will refrain from eating organ meats. On at least some views, you thereby exercise a normative power, giving yourself a reason to do as promised over and above any promise-independent reasons you may have. Enoch calls this robust reason-giving. If the doctor were to order you to refrain from eating organ meats, this would not give you a reason in the robust sense because the doctor has no authority to order you to do this, but her order would arguably purport to be robustly reason-giving.

In which of these senses are social norms reason-giving? On Bicchieri’s view, social norms are patterns of (conditional) motivational states. The presence of such patterns can give people reason to comply with the rule in the triggering sense. This may be because people have standing reasons to be well-regarded by other people or to avoid social sanctions (Bicchieri 2006: 23). Social norms make desirable outcomes of this kind contingent on following the rules, and so trigger reasons to comply with those rules. Further, the presence of a social norm may give individuals evidence that many people believe that they ought to follow the rule, and that in turn may provide individuals with evidence that they ought to follow the rule (Bicchieri 2006: 23). In this way, social norms can give people reasons in the epistemic sense. Nothing in Bicchieri’s account, however, requires that existing or operative social norms give people reason in a further, robust, sense. This does not preclude that some social norms are robustly reason-giving, but I will proceed on the assumption that the social norms I discuss below do not have any kind of normative force or authority beyond giving people reasons in the triggering and epistemic senses. This fits well with the primary purpose of Bicchieri’s theory of social norms, which is to help explain, predict and modify behaviour, not to justify behaviour.

3. Social Norms and Inequality

Relational egalitarians hold that we should relate to one another as equals. Relational egalitarianism has been very influential in political philosophy over the past two decades; it has reoriented egalitarianism away from narrowly distributive concerns, and it has done much to bring questions about the politics of race, gender, and class back to the centre of normative political thought. ³ Of course, this very generally stated view usually comes with various qualifications. Maybe justice requires that we relate as equals, maybe individuals have a valid complaint against unequal relations, maybe all persons should relate as equals, maybe the members of a given political society, and so on. Whatever the specifics, the view needs an account of what relating to one another as equals amounts to. It is relatively uncontroversial, I think, to observe that relational egalitarians have not found it easy to do so. ⁴

Two main types of view are pursued in the literature. The first is to moralize the idea of relating as equals. All human beings, or all persons, are one another’s moral equals – we all have the same basic moral standing, or are equal in moral worth. Relational egalitarianism is the claim that our social and political arrangements ought to reflect this moral equality. ⁵ One problem with this type of view is that it threatens to trivialize relational egalitarianism. The moral equality of persons is very broadly accepted, and those who accept it generally take it to place constraints on how we should relate to others in social and political contexts (see Steinhoff 2014). Trivially, our social and political relationships should conform to those constraints. Of course, there is plenty of room for disagreement about what these constraints are, but ‘relational egalitarianism’ would not be a distinctive position.

The second is to say that relating as equals refers to a distinctive kind of social phenomenon. ⁶ There are social structures, or sets of social relationships, and some are egalitarian and others are not. Relational egalitarianism’s central normative claim is that our social structures or relationships should be of the egalitarian kind. The issue with this view is that it is implausibly broad. Egalitarian and inegalitarian social relationships exist in many forms, in many contexts, and serving many different purposes. Unless inegalitarian social relationships are defined as unjust or wrongful (which would again trivialize the central normative claim), it would be astounding if such relationships were objectionable across the board. This invites a disaggregative approach: the relational egalitarian should distinguish different types of social inequality, and argue that certain types of social inequality are objectionable while others are not, and that when social inequalities are objectionable they can be so for different reasons. My aim is to show that the idea of a social norm is a valuable resource for this disaggregative project. The concept of a social norm, and Bicchieri’s theory of social norms in particular, can be used to distinguish a range of different types of social phenomena, all of which fit naturally under the heading of social inequality.

Social norms may not seem natural candidates to play this role, because a lot of the literature on social norms is concerned with the explanation of cooperative behaviour. In this context, social norms are argued to support cooperation in collective action problems (e.g. Ullmann-Margalit 1977; Den Hartogh 2002; Verbeek 2002; Skyrms 2004; Sugden 2004), to facilitate coordination (Lewis 1969), or to make us accountable to one another (Brennan et al. 2013: 35–39). Characteristic examples used to illustrate the phenomenon fit this picture. Bicchieri, for example, writes of norms against blowing one’s nose in the tablecloth (2006: 43), norms of reciprocity (2006: 12) and norms against premarital sex (2006: 13). Brennan et al. discuss norms to wear black at funerals (2013: 67), to pass the port on the left (2013: 64), and against urinating in public swimming pools (2013: 13). Pettit (1990: 725) mentions norms of truth-telling, promise-keeping and abstinence from theft, fraud and violence. The social norms in such examples (a) apply to all the individual members of the relevant population; (b) make the same demands of each individual; and (c) are upheld by all the individuals in the population, in the sense that the preferences of individuals are conditional on their expectations about the behaviour and beliefs of all other individuals.

Of course, not all examples are like this – Bicchieri mentions norms requiring Muslim women to wear a veil (2006: 14), and norms of revenge that apply to Corsican men (2006: 42). Brennan et al. write that norms can be used to create repressive, hierarchical, exclusive, and unjust relationships (2013: 39). More recently, there has been a more explicit and sustained focus on social norms that are bad (Bicchieri 2017; Thrasher 2018) or unfair (O’Connor 2019), and on the role social norms play in the explanation of disadvantage (Bruner 2017) including intersectional disadvantage (O’Connor et al., 2019). My discussion in this section aligns with this more recent work. However, much of this literature is concerned with the explanation of certain outcomes – greatly simplifying, bad norms produce suboptimal outcomes, unfair norms leave some groups worse off than others, and disadvantage is understood in relation to how bargaining problems are resolved. My discussion will instead largely focus on ways in which (sets of) social norms can themselves constitute different forms of social inequality, which can be understood independently of resulting distributions of resources or other goods. ⁷

3.1. Compliance and non-compliance

The first way in which social norms are tied to social inequality is by creating a distinction between compliant and non-compliant individuals. For every operative social norm, the members of the relevant population can conform to the rule or fail to conform to the rule, and they can be seen or believed to violate the rule, against a background of general compliance. As condition (b) indicates, such individuals may be subjected to various kinds of sanction. These sanctions in turn can establish a variety of inequalities between individuals. In some cases, social sanctions produce disadvantage in straightforwardly material terms. The shopkeeper may lose customers after being seen to violate a social norm of truth-telling. In other cases, social sanctions may leave people disadvantaged in what Cordelli (2015) calls relational resources, and Bourdieu (1986) might call social capital. Norm violators might be excluded from lunch tables, after-work drinks, birthday parties and from the social networks associated with such social gatherings, and they may have reduced opportunities for establishing friendships or marriages. Further, the sheer fact of being regarded as having conducted oneself contrary to how one ought to conduct oneself may for some people be a burdensome social stigma, even absent any further social sanctions.

Two special cases of the interaction between the violation of social norms and social inequality are worth highlighting. The first is the removal of individuals from a social group as a sanction for the violation of the social norms of that group. This can involve the physical relocation of the individual, as would be the case with classical practices of ostracism (see Forsdyke 2000) and modern punishments of banishment and deportation. One is placed outside the community, where the community has a specific geographic location. In other cases, however, the revocation of group membership may consist primarily, if not entirely, in the termination of a set of social relationships. One can be excommunicated from a church, defrocked as a member of the clergy, or cashiered as a military officer, without significant changes in one’s access to physical spaces. Through the application of these kinds of sanctions, social norms and their enforcement can contribute to the marginalization of individuals or groups.

Removal from a social group is typically not just a matter of how the individual relates to a set of social norms. This means that social norms cannot fully account for this form of social sanction. Still, Bicchieri’s account provides the resources to capture certain key features of it. First, on Bicchieri’s account, social norms are constituted by patterns of preferences conditional on people’s beliefs about the behaviour and normative beliefs of the members of a certain population. Removal from a social group can involve the removal of the individual from this population: people’s readiness to conform to the rule will no longer be conditional on the behaviour and normative beliefs of the removed individual. In other words, a key aspect of the power of the individual to influence the compliance and non-compliance of others is removed. Second, membership in this or that social group may feature in the content of relevant rules (more on this in section 3.3). A norm of reciprocity, for example, may demand of the members of a given population that they reciprocate the cooperative behaviour of other members of that population. Removal from a social group may partly consist in no longer being counted as a member of various groups that figure into the content of the social norms operative in that group.

The second special case is the imposition of degradation as a sanction for non-compliance. What I have in mind here is the lowering of an individual’s position in a social hierarchy as a sanction for the (perceived) violation of a social norm operative in the hierarchical social group (Garfinkel 1956). This social hierarchy may itself not just be a matter of the presence of certain social norms, so just as I was previously assuming the existence of a relevant social group from which individuals may be removed, I am here assuming a social hierarchy in which individuals can be repositioned. For example, one may be a member of a hierarchically organized military unit. As a penalty for violating a social norm internal to the military organization – say, a norm against malingering – one may be reduced from the rank of sergeant to corporal. Perhaps the rules that might lead to such a demotion in military rank are too formalized to count as social norms, but we can similarly imagine a professor being lowered in a prestige hierarchy upon violating certain informal rules of professional conduct.

3.2. Different groups, different rules

A second main way in which social norms interact with various kinds of social inequality and social hierarchy is by imposing different rules on different individuals. This in turn can happen through two main kinds of mechanism. The first is that there may be a number of distinct social norms with different contents, where different individuals are subject to different norms. The second is that a single social norm may have a complex content such that it makes different demands of different people. Imagine a school class in which it is the norm that boys wear their hair short and girls long. We might say that there are two social norms. One that applies to boys, the content of which is ‘wear your hair short’, and one that applies to girls that says ‘wear your hair long’. We might also say, however, that there is one social norm, the content of which is ‘wear your hair short if you are a boy, wear your hair long if you are a girl’. Unless further details are filled in, the distinction between these two ways of describing the situation may be without a difference. Bicchieri’s view of social norms gives us resources to make this distinction have force in explaining different kinds of social inequality.

In this subsection, I focus on the idea that there are distinct social norms for different (sub)groups. We say that P is the population of all the pupils in the class, PG refers to all and only the girls, and PB refers to all and only the boys. Of course, pupils in school classes may not be so neatly categorized into two gendered groups, but I will ignore such complications here. Let RG be the rule ‘wear your hair long’ and RB be the rule ‘wear your hair short’. We can now say that there is a norm of wearing one’s hair long among the girls in the following way:

There is a sufficiently large subset PG _cf of PG such that, for all members i of PG _cf :

i prefers to conform to RG conditional on (a) i believes that a sufficiently large subset of PG conforms to RG; and (b) i believes that a sufficiently large subset of PG believes that i should conform to RG (and may sanction non-compliance).

We get a social norm of wearing one’s hair short among boys by substituting PB and RB for PG and RG.

Consider a religious order, the members of which are vowed to chastity and poverty. It may be that in many or most cases, members of religious orders are committed to such rules unconditionally, but it is not hard to imagine a particular group in which such rules operate as social norms. These norms can be fully egalitarian within the order, in the sense that all members of the order are committed to these rules, the rules make the same demands of poverty and chastity of each member, and the rules are upheld by all the members. Nonetheless members may face significantly greater constraints on their behaviour and attitudes than others simply because non-members are not constrained by these social norms at all. This has a good claim to being a kind of social inequality: in virtue of the kind of social relationships individuals stand in to others, individuals are subject to more or less stringent demands on their conduct and attitudes. Social inequalities of this kind can in turn produce familiar differences in relative advantage. Norms of poverty and chastity may leave members significantly worse off in terms of their material resources and valuable relationships than non-members.

In some cases, social norms that operate in a particular (sub)population only regulate attitudes and behaviour ‘within’ that (sub)population. For example, a norm of reciprocity may require of group members that they help when fellow group members are in need, but may say nothing about what one is to do when non-members are in need. Similarly, the norms regulating one’s conduct as a member of the military, or as a worker on a factory floor, may be silent about one’s conduct and attitudes when off-duty or clocked out. In other cases social norms that operate in a particular subgroup nonetheless have much wider scope. Social norms may ask for various kinds of modesty, abstinence, dress, or diet, for example, wherever the individual goes and with whomever they interact. No special explanation of this distinction is needed: norms with wider and narrower scope are simply norms with different contents. It is worth pointing out the distinction, however, because social norms that operate in a particular subgroup but that nonetheless make requirements with a very wide scope produce especially consequential forms of social inequality. In these cases, the rules that apply to the individual because of their membership in a particular group do not just restrict their dealings within that group, instead these restrictions follow the individual across social contexts.

3.3. Complex content

I now turn to social norms with complex content, such that a single norm imposes different constraints on different members of the group. In our example, we say that R is the rule: ‘wear your hair short if you are a boy, wear your hair long if you are a girl’. Such a rule is a followed social norm in a given population P under the usual conditions. A key difference between this scenario and the separate norms for boys and for girls scenario of section 3.2 is the conditionalization of individuals’ preferences. In the section 3.2 scenario, girls’ preferences to conform to RG are conditional on their beliefs about the compliance and normative beliefs of other girls. If girls start wearing their hair short or if they stop believing that girls should wear their hair short, and if these changes are noticed, then the social norm erodes. But if boys start wearing their hair long or don’t care whether girls wear their hair long or short, we expect this to have no effect on the RG norm. In the complex content scenario, the preferences of all members of P _cf , girls and boys alike, are conditional on the believed compliance and normative beliefs of boys as well as girls. ⁸ Given these differences, we can capture additional kinds of social inequality and hierarchy.

First, norms with complex content allow for further ways in which social norms can constrain or disadvantage some individuals more than others. In my stylized example of classroom hairstyle norms, a single norm requiring long hair of girls and short hair of boys does not, by itself, establish any kind of social inequality – what is asked of boys and girls is merely different. Examples of norms that do establish inequalities along gendered lines easily come to mind. Social norms might say that girls and women are to be modest, passive, pliant, and should focus on domestic tasks, whereas boys and men are to be aggressive, active, assertive, and should pursue an education and a career. What makes it the case that these kinds of differential norms create or constitute inequalities and not mere differences is not always straightforward. In some cases, it might be enough to point to the fact that norms more tightly constrain the behaviour and attitudes of some individuals compared with others. A simple example would be a social norm in a religious community demanding that women wear a certain type of head covering. This may be a social norm that operates in the community as a whole, not just to the subpopulation of women, but that leaves half the population entirely unconstrained. In other cases, inequality-establishing norms may place constraints on all parties in ways that make it non-obvious that anyone is constrained more or less than another. A norm requiring boys to go to school and girls to stay in the home may leave as few options for both. Here it is tempting to say that the norm creates an inequality because it puts girls at a disadvantage compared with boys. Being excluded from education is bad for people, not just in terms of access to material resources, but also in terms of well-being, the development of capabilities, or various other metrics we use to measure advantage and disadvantage.

As we saw in section 3.2, social norms with simple content can also lead to inequalities in constraint and relative advantage. The added dimension here is that given a complex content, a single norm that applies to all the members of a given population can create such inequalities between members of that population. It is of course also possible that a simple norm imposes different burdens on different individuals. A social norm requiring chastity of each member of a religious order may affect the well-being or flourishing of some more than others. Still, this is a different kind of phenomenon than that of social norms making different demands on the conduct and attitudes of different individuals based on their gender or on other distinguishing features.

Second, social norms with complex content can explain some ways in which subgroups can be oppressed by larger groups. As we have seen, norms with simple content can impose significant constraints and burdens on individuals. But no matter how demanding such social norms may be, each individual member of the relevant population faces the same requirements. We may want to say that such norms can be oppressive, but this would be a situation in which each individual is similarly oppressed by the group as a whole – those who are not similarly constrained and burdened are not the agents of oppression. The way this can be different once we allow for norms with complex content is especially vivid in cases in which there is a burdened and constrained minority. Take the following description of some of the requirements of caste as an example:

the Untouchable was not allowed to use the public streets if a Hindu was coming along, lest he should pollute the Hindu by his shadow. The Untouchable was required to have a black thread either on his wrist or around his neck, as a sign or a mark to prevent the Hindus from getting themselves polluted by his touch by mistake. In Poona, the capital of Peshwa, the Untouchable was required to carry, strung from his waist, a broom to sweep away from behind himself the dust he trod on, lest a Hindu walking on the same dust should be polluted. (Ambedkar 2014 [1936]: 5)

I won’t claim that the described phenomenon is in fact best understood as a set of operative social norms with complex content. Nonetheless, we can imagine that at issue here is a number of social norms that apply to, and are followed and upheld by, a society as a whole, including both ‘Untouchables’ and Hindus, but that place specific behavioural and attitudinal requirements on ‘Untouchables’. We can call these the norms of caste. It is true that all members of the group, ‘Untouchables’ and Hindus alike, are subject to the norms of caste, but, given their content, it is also fair to say that through these social norms ‘Untouchables’ in particular are oppressed by the group as a whole. Moreover, ‘Untouchables’ may be a sufficiently small minority that the believed compliance and the believed normative beliefs of Hindus alone may suffice for the empirical expectations and normative expectations conditions of most people in the group to be met. Consequently, norms of this kind may be operative even if many ‘Untouchables’ are known to resist and are known to reject the idea that they ought to comply. In such cases it seems fair to say not just that ‘Untouchables’ are oppressed by the group as a whole, but that they are oppressed by the Hindu population. ⁹

Third, social norms with complex content can account for social hierarchies in which individuals occupy ranked positions. For illustration, consider a simplified picture of a command hierarchy. Suppose that the members of the relevant population P are numbered i ₁ , i ₂ , … i _n . Suppose that i ₁ has the authority to command all other members of P; i ₂ has authority to command all other members of P except i ₁ ; etc., all the way down to i _n who has no authority to command anyone. Real command hierarchies are inevitably more complex than this. They may have multiple individuals at the same rank, they need ways to resolve conflicts between orders of differently and similarly ranked individuals, and they need to specify a domain of authority (the sergeant can command his privates to charge to their near certain death, but cannot command them to write him into their will). Bracketing all of this, could such a structure of asymmetrical authority relationships be a matter of social norms among these people?

To capture this kind of command hierarchy, we can give a rule R the following kind of content: i _n obeys i _n-1 −i ₁ ; i _n-1 obeys i _n-2 −i ₁ ; and so on. About this rule, we then say the usual:

There is a sufficiently large subset P _cf of P such that, for all members i of P _cf : i prefers to conform to R conditional on (a) i believes that a sufficiently large subset of P conforms to R; and (b) i believes that a sufficiently large subset of P believes that i should conform to R (and may sanction non-compliance).

If this rule is followed in P, then for most individuals it is true that they prefer to obey the commands of higher-placed individuals. On this picture, such willingness to obey is explained by reference to the group as a whole upholding and enforcing a hierarchical pattern of authority and obedience. ¹⁰ In similar ways, we might understand dominance hierarchies in terms of rules specifying that different (sets of) individuals are to defer to different (sets of) individuals, and we might understand prestige hierarchies in terms of rules asking individuals to esteem, admire, and emulate different (sets of) individuals to greater and lesser degrees.

My claim here is very modest. All I say is that we can specify complex contents for R such that if R is a followed social norm, then the resulting social relationships resemble those we find in command, dominance, and prestige hierarchies. Maybe some such hierarchies are in fact social norms with complex content, but there may also be very different social structures that count as social hierarchies of these kinds. In addition, the explanatory power of my proposal here is limited by the fact that the relevant rules contain substantial hierarchy-related notions such as obedience, deference, or esteem. My proposal does not explain what these terms refer to.

Nonetheless, I think there are some advantages to conceiving of these kinds of social hierarchies in terms of social norms with complex content. The first is that this seems to get the normativity of social hierarchies right. As I said in section 2, on Bicchieri’s account social norms are not generally robustly reason-giving. Perhaps in some cases we do want to say that commands robustly provide reasons to obey. This might be the case in well-organized militaries which fight just wars justly. But there are also authority hierarchies in criminal organizations and dictatorial regimes. In such cases, issued commands give individuals reason to obey in the triggering and possibly the epistemic sense – such commands can trigger standing reasons to avoid disapproval and sanctions, and can provide (misleading) evidence of pre-existing reasons to do as commanded. We should, however, resist the idea that such commands are robustly reason-giving. Thinking of command hierarchies as social norms with complex content accurately reflects this picture of the reasons with which the commands of superiors provide lower-ranked individuals. Similar remarks can be made about prestige hierarchies in which the qualities for highly esteemed individuals are not genuinely estimable – hierarchies in which people acquire prestige proportional to the severity of the crimes they commit, for example. Thinking of such hierarchies as social norms with complex content leads to the correct verdict that participants in such hierarchies are given reason to esteem prestigious individuals in the triggering and epistemic senses, but not in a robust sense.

Second, the proposal explains the involvement of the wider social context in the constitution of two-person hierarchical relationships. In certain dominance hierarchies among non-human animals, the rank of individual animals is directly determined by their size or their capacity to inflict physical violence. In human authority hierarchies, however, the rank of individuals is not straightforwardly determined by the intrinsic properties of individuals (Martin 2009: Ch. 7). Why would a captain in the mafia obey the commands of a mob boss he could easily beat in a fight? The answer is obvious: his obedience is supported by the broader social structure in which these two individuals operate. My proposal is to think, at least in some cases, of this broader social structure as an operative social norm. Obedience is motivated by the empirical and normative expectations of compliance with a rule that demands obedience to particular commands.

3.4. Influence

I now turn to one final way in which social norms can give rise to social inequality: social norms can place individuals in positions of greater and lesser influence. Bicchieri’s account of social norms refers to a ‘sufficiently large subset’ in three different places: a sufficiently large subset of the population has the requisite conditional preferences, and those preferences are conditional on the belief that a sufficiently large subset will comply and the belief that a sufficiently large subset believes that one ought to comply. This introduces a degree of vagueness into the account, but this seems appropriate in order to capture a sufficiently broad range of cases in which we would want to say that a social norm is operative. Without removing this element of vagueness, we can nonetheless introduce some additional structure to capture an additional type of social inequality. In particular, I will look at some further specification of the conditional preferences of individuals.

If R is an operative social norm in P, then a sufficiently large subset of P _cf of P have the requisite conditional preferences. These members of P _cf prefer to conform to R conditional on the believed compliance and the believed normative beliefs of sufficiently many others. One straightforward way of specifying this conditionalization is as follows: each individual in P _cf prefers to conform provided that they believe a threshold number or percentage of the members of P conforms and believes one ought to conform. For example, if we imagine P to have exactly 100 members, each individual in P _cf may prefer to conform to R provided they believe that 70 or more members of P conform to R, and provided they believe that 70 or more members of P believe they ought to conform to R. Once we have this simple picture in view, we can introduce further complications. ¹¹

I will focus on the possibility that individuals attach different ‘weights’ to the believed compliance and normative expectations of different group members. Here is a stylized example. Imagine a group of 21 total diners at the royal court of Henry the Eighth. Henry is seated at the head of a long table, with all the other diners seated in descending order of royal favour corresponding to their distance to the king. Suppose there is a social norm within the group as a whole with rule R being that one does not pick up food from a plate with one’s hands; instead, one uses a fork. We say the usual to capture this fact, with P _cf being the subset of individuals with the requisite conditional preferences. Now suppose, as seems reasonable, that members of P _cf care more about the believed compliance and believed normative beliefs of the king than of other participants. Further, we can imagine that each individual cares about the believed compliance and normative beliefs of others in correspondence with their distance from the king.

A simple way to represent this is as follows. Let each member of P _cf have a threshold score for their empirical expectations and a threshold score for their normative expectations, expressed as a real number. The individual prefers to follow the rule provided both of these scores are reached. Each individual in P _cf scores the level of compliance in P by determining which members in P they believe do and which do not comply with R. In doing so, however, they assign different weights to different individuals, also expressed as a real numbers. An individual i’s compliance score for R in P is the sum total of the scores for each individual in P, where individuals i believes not to comply are scored 0 and all the others are scored their weights. Similarly, each individual in P _cf scores the level of normative belief in P. The individual prefers to comply with R if both scores are at or above their threshold score.

For example, let both threshold scores be 50 for each of our 21 diners. Each member of P _cf scores the king at 11, his two neighbours at 10, their non-king neighbours at 9, and so on. ¹² Each diner in P _cf prefers to use a fork to pick up food from a plate provided that the people they expect to comply with the rule add up to 50 or more, and likewise for the people they expect to believe that they ought to comply with the rule. When different weights are attached to different individuals in this way, and especially when those weightings align across individuals (everyone weights the king at 11, his neighbours at 10, and so on), some people can have much greater influence over the preferences and behaviour of group members than others. This kind of influence is not captured by the previous subsections: no distinction between compliant and non-compliant individuals is being made, all members are subject to the same norm, and the norm specifies the same requirement for all participants. We might see differences in this kind of influence be instantiated by popular kids in school, or by ‘influencers’.

With this basic picture in view, we can see that many other kinds of complications can be built into the conditionalization of individual preferences. My normative and empirical expectations about one person could, for example, pre-empt the weight of my normative and empirical expectations about another. I might give weight to the (believed) conformity and normative beliefs of the bishop, for example, but if I believe that the pope complies and expects me to comply, then my beliefs about the bishop may have 0 weight beyond those about the pope. Further, there doesn’t seem to be a principled reason why the weighting in a given individual’s normative and empirical expectations with regard to another person would have to be the same. That is, it seems in principle possible that my conditional preference is highly sensitive to whether i in fact conforms to R, but not to whether i believes that I should conform to R, or vice versa. If the king, for example, picks up a bunch of food with his hands, this might not change my readiness to conform to the rule one bit; but if I believe that the king no longer thinks that I should conform to the rule, that might more or less suffice for me to use my hands.

If inequalities of influence are pushed to extremes, then the resulting social situation will run up against the limits of the notion of a social norm. Consider groups in which there is a single individual of overwhelming influence. Gurus may play this kind of role in certain cults and sects (see Zablocki 1980). Say that sufficiently large subset P _cf of population P prefer to conform to a set of rules – rules dictating certain modes of dress, exercise and prayer, for example – conditional on sufficiently many others also conforming to those rules and sufficiently many others believing that one ought to conform to those rules. So far they fit Bicchieri’s definition. But imagine that there is one person, the guru, such that for almost all of the members of P _cf their belief that the guru follows the rules and their belief that the guru believes that one ought to follow the rules counts as ‘sufficiently many’ regardless of the believed behaviour and normative beliefs of other members of P. We may hesitate to say that these rules of dress, exercise, and prayer are social norms in population P. Nothing much hangs on whether we would apply the label, I think, but we can see that if we allow the conditionalization of individual preferences to include differences in weighting, then extreme concentration of influence of our guru would be the limit case.

4. Conclusion

Taking Bicchieri’s theory of social norms as my starting point I have, sometimes by simply using the theory and sometimes by expanding upon the theory, shown how social norms can create, support and perpetuate a number of different ways in which individuals can be unequal to one another. In each case I have given examples intended to show that these inegalitarian social structures are not mere theoretical possibilities. Of course, I have not shown that these social structures do in fact exist, but it is at least plausible that social relationships with which we are familiar from sociology, anthropology, history, and simply everyday life are captured by the variations on Bicchieri’s theory described in section 3.

When relational egalitarians object to social inequality, examples of each of the four types of phenomena I have described fit their target. But these different kinds of inegalitarian social norms may well be objectionable (when they are) under different conditions and for different reasons. My hope is, then, that this paper can help move social egalitarianism away from discussions of the normative status of ‘social inequality’ understood as one broad phenomenon. Furthermore, if some of the unjust social arrangements with which relational egalitarianism is concerned are indeed closely related to social norms, then the large literature on the dynamics of social norms can help us understand how these injustices may be remedied.