Against acceptance theories of social norms

Two theories

On what I will call “acceptance theories,” a rule R counts as a social norm within community C if and only if it satisfies two conditions: first, people in C generally accept R; and second, people in C generally know that others generally accept R. ² These conditions constitute the shared core of all acceptance theories. On an acceptance theory, R_G in Groceries counts as a social norm because everyone accepts R_G, and everyone knows that everyone accepts R_G. In contrast, while rule R_A in Alcohol structures people's interactions with one another, it does not truly count as a social norm, because people do not generally accept it. Valentini claims that, in characterizing a situation like Alcohol, we ought not describe R_A as a social norm; rather, “the case is more aptly described as one in which individuals incorrectly believe that there exists a certain socially constructed norm… The norm is ‘illusory’, even if, from an external point of view, there are plenty of cues suggesting that a norm is there…” (Valentini, 2023: 32). Similarly, Brennan, Eriksson, Goodin, and Southwood (BEGS) argue that in this kind of scenario, we should say that people mistakenly think that a social norm prescribes certain behavior (Brennan et al., 2013: 35).

While acceptance theories conceptualize social norms as the rules that people generally and publicly accept, proponents of this approach disagree on many points. Most notably, they disagree about what it means to accept a rule. BEGS argue that people accept a rule if and only if they have “normative attitudes that appropriately reflect the content and normative force of” the rule (Brennan et al., 2013: 29). (For instance, if the rule prohibits ϕ-ing, then I accept the rule if I believe that ϕ-ing is wrong, am disposed to feel indignant when I witness other people ϕ-ing, am disposed to feel guilty if I ϕ, and so on.) In contrast, Valentini advances a bipartite theory of acceptance, according to which someone accepts a rule only if they both (a) believe the normative proposition that the rule expresses (so, for instance, if the rule requires that people ϕ, then one must believe that people must ϕ); and (b) commit to the rule as a standard for public behavior (Valentini, 2023: 23). But we can set these disagreements aside. All that matters for our purposes is that, on any acceptance theory, to accept a rule is not simply to decide to cooperate with it. ³ It must involve something else, like BEGS's normative attitudes or Valentini's commitments. Otherwise, students’ decisions to cooperate with R_A in Alcohol would count as acceptance, and the distinction between rules that people do generally accept and rules that people do not would lose all significance.

Unlike acceptance theories, practice theories do not require that people accept a rule in order for the rule to count as a social norm. They simply require that the rule structure a community's social practices. In a case like Alcohol, R_A clearly structures college students’ practices even though students do not generally accept R_A, so on a practice theory R_A is at the very least a plausible candidate for a social norm.

One of the main arguments against practice theories is that it is not clear that social practices can ground normative rules. For example, coordination conventions simply constitute solutions to coordination problems. In these cases, it is in each individual's interests to cooperate with the practice as long as other people generally cooperate (Mackie, 1996; Schelling, 1960; Ullmann-Margalit, 1977). This entails that it would be imprudent for individuals with typical interests not to cooperate with the practice, and that one party's non-cooperation would set back others’ interests and possibly disappoint their expectations. But it does not entail that defectors do anything wrong. It does not entail that people have any standing to demand others’ cooperation, or that there would be any grounds to criticize individuals with idiosyncratic interests who choose to defect. ⁴ In other words, coordination conventions lack what BEGS call “normative oomph” (Brennan et al., 2013: 38). This is not to say that coordination conventions are not an important part of our social world. It is just to say that coordination conventions ought not count as social norms. It is on these grounds that BEGS conclude that practice theories of social norms are inadequate (Brennan et al., 2013: 18).

But this shows only that coordination conventions on their own do not constitute social norms. It does not show that social practices generally cannot ground social norms. I propose that we focus our attention specifically on what I call representational practices, that is, on practices that express normative representations of the world. Generally, a rule R counts as a social norm within a community C if and only if the members of C generally cooperate with practices that represent R as a valid normative rule. ⁵ On this account, a rule like R_G counts as a social norm in Groceries because shoppers represent the act of putting an item in a cart or basket as a way of claiming the item, and so represent the act as imposing an obligation on others not to remove the item from the cart without justification. The representations in question are not subjective; that is, shoppers do not only (or even necessarily) believe that they claim the items by putting them into their carts. Rather, R_G counts as a social norm because people expressively represent R_G as a valid rule in their words and actions. Where R_G is deeply ingrained (as is the case in most contemporary grocery stores), shoppers will interpret mere conformity to R_G as expressively representing R_G as a valid rule: They will see other shoppers taking no interest in the items in one another's carts and will understand that this is no mere coincidence, but expresses a representation of the items as claimed. But it is easiest to see the expressive representation of a rule as a valid norm in people's responses to its violation. If someone were to remove an item from another shopper's cart, we can expect the victim to express indignation, to object to this kind of behavior, to complain to other shoppers and to store employees; and we can expect these bystanders to express support for the victim's claims, insisting that the perpetrator return the item, perhaps even asking them to leave the store if they will not respect other shoppers’ rights. Through all of these actions, victims and bystanders (and perhaps even perpetrators) represent R_G as a valid rule, so on a practice theory, R_G will count as a social norm. By the same token, a rule like R_A will count as a social norm on a practice account as long as students generally represent R_A as a valid norm. It will not matter that students generally do not accept R_A. All that matters is whether they demand that other students cooperate with R_A, blame students who violate R_A, express guilt and offer excuses when they violate R_A, and so on (again, not necessarily subjectively, but in their words and actions).

A theory that grounds social norms in representational practices is not vulnerable to BEGS's objection that social practices cannot ground normative rules. After all, since the relevant practices expressively represent certain rules as valid rules within a community, they explicitly have normative content. In contrast, coordination conventions need not count as social norms because they do not necessarily involve the relevant representational practices. For example, participants in a coordination convention need not demand one another's cooperation, offer excuses when they fail to cooperate, or blame one another for failures to cooperate. On the practice theories of social norms, then, a rule cannot count as a social norm if people do not generally represent it as a valid rule for their community, but a rule can count as a social norm even if people do not generally accept it.

In the remainder of this paper, I will say that any rule that people generally and publicly accept within a particular community counts as an acceptance norm within that community, and that any rule that people generally represent as a valid norm counts as a practice norm. Moreover, I will call practice norms that people do not accept mere practice norms.

Methodology

Why should we think that only acceptance norms count as genuine social norms? BEGS do not provide a clear argument. Commenting on cases (like Alcohol) in which people falsely believe that the members of their community widely accept a rule, BEGS simply assert that:

It seems… that one should say that there appears to be a norm within the community but that appearances are deceiving; that the members of the community mistakenly think there is a norm. Cases of merely apparent or mistaken norms may, of course, have an important bearing on how people behave. But they must not be confused with norms proper. (Brennan et al., 2013: 35)

But why should one say this? On what grounds should we think that it would be “confused” to construe rules like R_A as “norms proper”? BEGS provide no explicit answer. In fact, we face a choice between two terminologies, both of which seem viable at first glance. On the one hand, if we embrace an acceptance theory of social norms, we will adopt terminological rule T1: a rule counts as a social norm only if people generally and publicly accept it; rules like R_A at most count as “illusory” or “perceived” norms. On the other hand, if we embrace a practice theory of social norms, we will adopt an alternative terminological rule T2: a rule counts as a social norm as long as people generally represent it as a valid rule, whether they accept the rule or not; some social norms enjoy widespread acceptance, while others do not. If we are going to choose between T1 and T2 (and if we are not going to view the choice as purely arbitrary), we will need some kind of argument. BEGS provide none.

We might say that we should embrace T1 simply in order to highlight the descriptive difference between acceptance norms and mere practice norms: namely, that people accept the former, and do not accept the latter. But we do not need our conceptual vocabularies to capture every descriptive difference between distinct phenomena. On the contrary, there are infinitely many descriptive differences between any two phenomena, so we should build our conceptual vocabularies around the differences that matter. The question is: does the difference between acceptance norms and mere practice norms actually matter?

In order to answer this question, we will need some way of distinguishing the differences that matter from the differences that do not. Here, Valentini provides a principled path forward. First, she advances a methodological proposal: we should define social norms primarily with an eye toward our theoretical goals. As Valentini observes, however, we might find multiple conceptions of social norms theoretically fruitful, especially if we call on social norms to do multiple kinds of explanatory or moral work. So, Valentini concludes, we should not insist that there is a single univocal concept of social norms (Valentini, 2023: 18). Instead, we should specify the phenomena that we hope to explain, and we should build conceptions of social norms that provide us with the tools we need to explain those phenomena. Working within this methodological framework, Valentini argues that we should embrace acceptance theories of social norms because they equip us to explain some important phenomena. But she explicitly allows that alternative theories might better equip us to explain other phenomena, and so might be worth our attention in other theoretical contexts (Valentini, 2023: 33). In principle, then, Valentini's approach does not treat acceptance theories and practice theories as competitors, but as complements that do distinct kinds of philosophical and social scientific work.

Nevertheless, I will argue that acceptance theories are not viable approaches to social norms. To do so, I will take up Valentini's methodology, and I will argue that acceptance norms lack explanatory power with respect to the phenomena that we typically call on social norms to explain. In the empirical domain, the fact that people generally and publicly accept some rule is not necessarily explanatorily significant. When it is not, the distinction between social norms (understood as acceptance norms) and allegedly “illusory” norms obscures more than it illuminates. In the moral domain, the claim that acceptance norms are morally significant seems implicitly to depend on democratic commitments. But acceptance norms stand outside of the institutional and political processes that underlie democratic legitimacy. In both empirical and moral inquiry, then, distinguishing acceptance norms from mere practice norms does not advance our theoretical goals and may actually confound them. We are better served by the practice theories of social norms.

Extractive and identitarian practices

First, there are contexts in which widespread acceptance of a norm does not explain widespread cooperation with the norm; instead, self-interest explains both widespread cooperation and widespread acceptance. ⁶ I will describe two related kinds of social practice that often fit this description: extractive and identitarian practices.

Extractive practices enable one group to extract labor or resources from another group on unjust terms. By way of illustration, consider a new case:

Labor: Waged laborers have legal rights to regular breaks. However, managers frequently express contempt for any laborers who take full advantage of their legally mandated breaks, blaming these laborers for any delays in service or production. Some laborers—call them acquiescent workers—support management in this practice, while others—call them dissident workers—actively resist it. Managers and acquiescent workers accept R_L: Workers should forego their legally mandated breaks in order to speed service and production.

Although managers and acquiescent workers accept R_L, that does not entail that they cooperate with R_L because they accept it. On the contrary, we might expect that managers cooperate with R_L because cooperation with R_L advantages them in their ongoing conflict with workers, and that acquiescent workers cooperate with R_L in the hope that this cooperation will keep them in management's good graces (and might even earn them a chance to join the managerial class). If managers and acquiescent workers cooperate with R_L because it is in their interests, they also might accept R_L because they cooperate with R_L, and because most people find it psychologically difficult to cooperate with a rule that they know to be unjust. To be clear, my claim is not that managers and acquiescent workers might pretend insincerely to accept R_L. It is that they might come quite sincerely to accept R_L in virtue of their interests in cooperating with R_L and the psychological pressures that cooperation generates. In this kind of case, then, widespread acceptance of R_L does not explain widespread cooperation with R_L; instead, self-interest explains widespread cooperation with R_L, and widespread cooperation explains widespread acceptance.

Identitarian practices enable one group to define itself in opposition to other groups. (These can be related to extractive practices, since they can draw the boundaries between the groups who benefit from extractive practices, and the groups from whom they extract labor and resources.) Identitarian practices can define groups along various dimensions: Where did your ancestors come from? What does your body look like? What language do you speak? How do you dress? Most relevantly to us, identitarian practices often define groups in part in terms of members’ attitudes, thereby subjecting them to what Han van Wietmarschen calls attitudinal social norms (van Wietmarschen, 2021). For example, sexist practices define two genders, men and women, primarily in terms of their putatively natural orientation toward patriarchal family roles: men are providers, protectors, and sexual agents; women are caregivers, homemakers, and sexual objects. On the terms that sexist practices prescribe, those who do not clearly fit into these roles are improperly gendered. They are deviant, corrupt, or monstrous, the apt objects of medical intervention, management, exclusion, or even extermination. Moreover, sexist practices define the “real” man and woman in part in terms of their commitment to the rules that define the patriarchal family's gendered roles. These rules include the following:

R_S1: Men should provide for their families.

If enough people come to accept these rules, and if their acceptance is common knowledge, then these rules will count as acceptance norms. At first glance, then, we might think that widespread acceptance of R_S1 through R_S5 explains the persistence of sexist practices. But as long as there is general cooperation with sexist practices, a failure to present oneself as though one accepts these rules constitutes a failure to present oneself as a “real” man or woman, and this can expose one to significant risks, including the risks of medical and punitive violence. So people might present themselves as though they do accept these rules primarily in order to mitigate these risks. And the fact that people maintain this self-presentation over the course of their lives may itself explain why many of them do accept R_S1 through R_S5. As a result, general cooperation with sexist practices (which distinguishes “real” men and women from the deviant, corrupt, and monstrous) creates the very conditions that encourage people to accept the rules that themselves define sexist practices. If people were to cease cooperating with these practices, then people would have no social incentive to present themselves as though they accept these rules, which would alleviate much psychological pressure on them to actually accept these rules. In this case, as in the case of R_L, widespread acceptance does not drive general cooperation with R_S1 through R_S5. Rather, widespread cooperation with sexist practices in general explains why people accept the rules defining these practices (including R_S1 through R_S5). On this interpretation, sexism does not survive because people accept its rules; rather, people accept sexist rules because sexism survives.

The upshot is that the norms internal to extractive and identitarian practices often satisfy the criteria for acceptance norms, even though people ultimately cooperate with these practices on the basis of “prudential considerations and conformism” (to use Valentini's phrase). If so-called “illusory norms” like R_A should not count as “genuine” norms because people only cooperate with them on the basis of self-interest, then what should we make of rules like R_L and R_S1 through R_S5?

Moreover, it is not hard to conceptualize cases in which people do not accept R_L or R_S1 through R_S5, and yet pretend that they do, and invoke these rules in their evaluation of one another's activities. Consider a modified version of Labor:

Labor*: Managers and acquiescent workers do not, in fact, accept R_L. Instead, they act in bad faith when they cite R_L to blame workers who take full advantage of their legal breaks. They do so only because they believe that cooperation with R_L is in their interests. ⁷

The difference between Labor and Labor* is that, in Labor*, people cooperate with R_L in order to advance their own interests; and in the original Labor, people cooperate with and accept R_L in order to advance their interests. As a result, R_L counts as an acceptance norm in Labor but only as a mere practice norm in Labor*. But is this a meaningful difference?

Following Valentini's lead, we might think that this would be a meaningful difference if different strategies would enable us to change or eliminate R_L in the original and modified scenarios. But in both cases, people ultimately cooperate with R_L on the basis of self-interest. We certainly could not expect people to stop cooperating with R_L in Labor* simply by revealing that nobody actually accepts the rule. The people who cooperate with R_L in Labor* already do so in bad faith and will continue to do so as long as this cooperation advances their interests. So it is likely that, if we want people to stop cooperating with R_L in Labor*, we will have to alter the schedule of costs and benefits that cooperation and resistance entail. Similarly, we should not expect rational argumentation to change people's minds or behavior in the original Labor. Management's and acquiescent labor's beliefs are fixed by their interests, and so are not likely to respond to reason. In both cases, we should instead undermine R_L by empowering workers or disempowering managers. If we could create conditions in which workers can effectively ignore managers’ efforts to represent them as lazy when they take full advantage of legally mandated breaks, or even to sanction managers who engage in this representational practice, then we can hope that the practice will wither away. Neither workers nor managers will have any reason to cooperate with or to accept R_L because it no longer will be in their interests to do so. In both Labor and Labor*, then, what is most salient is the distribution of power. The fact that people generally accept R_L in the original case turns out to be a red herring.

Similarly, in our original characterization of sexist norms, we stipulated that people generally accept R_S1 through R_S5. But we should not hope to defeat patriarchy by presenting people with moral arguments against these beliefs. We might hope to win over some hearts and minds through these practices, but we should also expect some people to dig in their heels, either because they cling to patriarchy's benefits or because they fear the consequences of losing their status as “real” men and women. If we hope to reach these people, we ought to begin by changing prevailing practices, undermining cooperation with sexist rules like R_S1 through R_S5. For example, we ought to cultivate feminist and queer spaces in which people can safely try on new personae, and we ought to pass and enforce legislation that protects people from the kinds of violence that sexist practices impose on those who do not present themselves as though they naturally fit into the patriarchal family mold. Acceptance theories focus our attention on epiphenomena; practice theories, in contrast, enable us to identify the forces that really explain the development, persistence, and change of the norms internal to extractive and identitarian practices.

Expressive norms

So far, I have argued that there are cases in which prudential considerations explain both widespread acceptance of and widespread cooperation with a normative practice, and that in these cases, practice theories of social norms are more fruitful than acceptance theories. Now we turn our attention to a set of cases in which people cooperate with a normative practice, not because they accept its constitutive rules, but because they expect cooperation to facilitate meaningful interaction with others. Consider one of Valentini's examples, in which a Japanese tourist dines at a restaurant in Germany (Valentini, 2023: 97). Germany and Japan have distinct tipping practices. In both countries (and unlike in the United States), restaurants pay servers a reasonable wage, so that servers do not depend substantially on gratuities from patrons. Nonetheless, in Germany, it is expected that patrons will tip 5% to 10% for good service, while in Japan, it is the norm not to tip one's servers. In Germany, then, we seem at first glance to have a simple rule R_T mandating that restaurant patrons tip their servers between 5% and 10%, which counts as a social norm if and only if Germans generally accept it. But this mischaracterizes the case. It ignores the fact that tipping is an essentially expressive practice. Tipping (and, by the same token, a failure to tip) says one thing in Germany, and another in Japan. By attending to this expressive significance, we will see that tipping is normatively somewhat complicated in ways that exhaust acceptance theories’ conceptual resources.

It is easiest to see the normative complexity of tipping by taking up the perspective of the Japanese tourist. We might expect a conscientious Japanese tourist to tip in Germany in order to express appreciation for good service, even while knowing that the same act might express contempt or arrogance in Japan. The tourist accepts the principle that one should express respectful appreciation for good service at a restaurant, and they recognize that different actions express this appreciation in different places. They adjust their actions to their context because they wish for others to understand them, and they know that others will not understand them if they do not speak the local language. In order to clarify this case, we ought to separate R_T into two distinct rules: a primary rule, R_T1, which mandates that restaurant patrons express appreciation for the service; and a secondary semantic rule R_T2, which specifies that tipping 5% to 10% has the relevant expressive content. Once we analyze R_T into R_T1 and R_T2, we open up the possibility of ambivalence toward R_T. Individuals might neither straightforwardly accept nor reject R_T; instead, they might accept the primary rule R_T1 while regarding the semantic rule R_T2 as arbitrary. They cooperate with R_T2 only because, when they attempt to cooperate with R_T1, they want others to recognize the expressive content of their actions, and this entails that they must defer to public interpretive standards. In other words, people (like Valentini's Japanese tourist) who are themselves indifferent to R_T2 nonetheless have reason to cooperate with R_T2 because that is the local semantic rule. They would gladly cooperate with a different rule in other contexts (as the tourist does when in Japan).

To be clear, the point is not that proponents of acceptance theories cannot make sense of the fact that travelers can and do accommodate themselves to local norms. It is that they rely on oversimplified interpretations of social rules in ways that distort our interpretations of these accommodations. The conscientious tourist in Germany need not simply tip between 5% and 10% because R_T is a local social norm; rather, they might tip because they accept R_T1, and because the semantic rule R_T2 specifies what cooperation with R_T1 looks like in Germany. This is important because it is not only tourists who might take this attitude toward local semantic rules. On the contrary, people within a community might generally cooperate with semantic rules not because they are committed to them or believe them to be correct, but only because these are the rules that they expect others to deploy in the interpretation of their words and actions, and because they wish for others to understand them. When describing a community like this, it would distort the facts to claim that people generally accept the original rule R_T. It may be that they do accept R_T1, but they have no particular commitments to or attitudes about tipping in itself. It is even possible for people generally to have reason to cooperate with R_T2 even though they generally oppose it. A case like this need not involve the kind of “pluralistic ignorance” that sustains R_A in Alcohol. On the contrary, even if people were aware of general opposition to R_T2, they might not abandon the semantic rule if (a) it is important to them that they cooperate with R_T1, (b) they need to cooperate with some semantic rule in order to do so, and (c) they would face a coordination problem were they to attempt to move from R_T2 to a new semantic rule. At the very least, there may be multiple alternatives to R_T2, and if people see no reason to converge on any one alternative, they may reasonably fear that abandoning R_T2 would lead to expressive chaos, undermining people's ability to cooperate with R_T1 in a way that they can expect others to understand. To make matters worse, people might disagree about which alternatives to R_T2 are best. In these conditions, a status quo could persist in which prevailing norms obligate people to tip between 5% and 10% even though people generally believe they ought not have to do so. ⁸

In these kinds of cases, people do not unambiguously accept R_T, so R_T is not an acceptance norm. It is worth noting that we should not say that people in Germany generally accept R_T2 in the sense that they choose to cooperate with it. As noted in the section “Acceptance and practice theories of social norms,” if we adopt this minimalist conception of acceptance, then the distinction between acceptance norms and mere practice norms disappears entirely. At the same time, it would be misleading to label R_T an “illusory norm,” or to say that people “mistakenly” believe R_T is a local norm. As we saw earlier, Valentini claims that people only cooperate with an “illusory norm” because they falsely believe that others accept it; as a result, “illusory norms” vanish when people learn the truth. But even if people were aware of general indifference or opposition to R_T2, R_T may remain stable if people generally accept R_T1 and if people were to face a coordination problem were they to abandon R_T2. So while it would distort the facts to label R_T an accepted norm, we would do no better to label it an “illusory” norm.

Could we say that, properly speaking, R_T1 (rather than R_T) is the true social norm here, since that is the rule people generally accept? This move would push our theory of social norms to such a high level of generality that it would compromise our ability to speak with any specificity about many of the social norms that operate within distinct communities. It may be that many people accept general normative requirements like “be respectful” or “express gratitude in appropriate circumstances,” while recognizing that social practices provide concrete interpretations of these requirements within specific historical contexts. But if we wish to understand social norms, then surely what we want to know is how these specific interpretations acquire authority within these contexts. Even if many communities share an expectation that restaurant patrons express appreciation for the service, part of our goal in advancing a theory of social norms is to develop the vocabulary with which to identify the distinct ways in which patrons do so in different places and at different times. And it should be clear that acceptance theories cannot explain this.

A messy world

If acceptance theories aspire to focus our attention on explanatorily significant phenomena, they face a third problem, as well: even when people do generally accept some rule R, and even when this plays some role in explaining why people generally cooperate with R, the total explanation often will be complex and unstable. That is, the fact that people generally accept some rule R will often provide only a partial explanation for general cooperation with R, and its significance compared with other partial explanations will likely to change over time. Certainly individuals will have multiple shifting motivations for cooperating with some rule. For example, people in sexist communities often cooperate with sexist conceptions of masculinity and femininity for various reasons:

they individually have internalized sexist propaganda since childhood;

The relative significance of these explanations will vary from individual to individual, and for each individual, it will vary from one time to another. Within the wider community, the role that general acceptance plays in explaining widespread cooperation with sexist norms likely will shift over time, due to changes in the wider social structure (like the rise of particular social movements, like feminism or the Civil Rights Movement), changing economic circumstances (like the mass entry of women into the workforce and the disappearance of the Fordist family wage), or the rise of new technologies (like the Internet and social media).

I do not mean to claim that proponents of acceptance theories cannot acknowledge that various factors might contribute to a total explanation of any community's cooperation with a rule. But once we acknowledge that general acceptance might play only a partial role in this explanation, we find that acceptance theories draw arbitrary distinctions between similar cases and obscure important contrasts between different cases. A condition in which general acceptance plays a significant, but non-exclusive, role will be similar in some respects to a condition in which people do not accept the relevant rules, and similar in some respects to a condition in which general acceptance plays an exclusive role; but it will be exactly like neither of these conditions, so it seems arbitrary to lump it in with the latter while distinguishing it from the former. Indeed, a condition in which general acceptance plays only a small explanatory role may be more similar to a condition in which people do not accept the relevant rules than a condition in which acceptance plays an outsized or exclusive role. To be sure, there is good reason to investigate the role that different factors play in explaining people's cooperation with social rules. But if we insist that these rules count as social norms only if people generally and publicly generally them, we arbitrarily highlight one factor over all others. It will be more productive to deploy a practice theory of social norms, which is indifferent between the various possible reasons why people might cooperate with a rule and thus allows us to investigate these reasons with an eye toward their complexity and dynamism.

Beyond behavior

If we wish to explain patterns of human behavior, and we want a conceptual vocabulary that enables us to draw distinctions that matter, then acceptance theories of social norms come up short. These theories generate a compromised map of our social world. They draw arbitrary boundaries between similar cases (like Labor and Labor*), they leave expressive rules off the map entirely, and they arbitrarily privilege one (frequently partial) explanatory factor over others.

But these arguments do not only show that acceptance theories leave us ill-equipped to explain patterns of human behavior. We can redeploy these same arguments to show that, for any class of empirical phenomena we wish to explain, acceptance theories obscure more than they illuminate. For example, we might wish to invoke social norms to explain why certain attitudes prevail within a community. Say, for instance, that people tend to expect one another to ϕ, to feel guilty when they fail to ϕ, and to feel indignation when others fail to ϕ. Why do these patterns persist? It could be that these attitudes circulate within a community because people generally accept rule R which obligates people to ϕ. (To be clear, this is no circular explanation. Rather, the proposal now on the table is that, because of their social natures, individuals are more likely to sustain certain attitudes if those around them sustain similar attitudes.)

I assume that this kind of interaction between individuals’ attitudes and their social environment can explain why we see certain cognitive and affective patterns within particular communities. The question is whether this gives us good reason to distinguish those rules that people generally accept from those rules that people do not generally accept. And the arguments that we have encountered already show that it does not. People might have the relevant attitudes, not because people generally accept rule R obligating people to ϕ, but because

people generally recognize that cooperation with R is in their interests, and (for various psychological reasons) ongoing cooperation encourages people to develop the relevant attitudes; or because

Moreover, when general acceptance does help to explain the circulation of the relevant attitudes within a community, it may constitute only a partial explanation, and the relative significance of this explanation may shift over time. Therefore, acceptance norms prove no more apt for our efforts to explain attitudinal patterns within a community than they were for our efforts to explain patterns of human behavior. We can reformulate these same arguments to challenge any effort to show that acceptance theories are not fruitful resources for empirical explanation. In the empirical domain, we ought to prefer practice theories of social norms.

Success conditions for a defense of normative acceptance theory

If proponents of normative acceptance theory aim to demonstrate that acceptance norms constitute a morally significant category, then they owe us an account of the moral significance of the rules that people generally accept. Notice, though, that this is not the same as an account of the moral significance of the rules that individuals accept. Even if we assume that individuals’ normative attitudes or commitments are morally significant, that does not entail that we have any reason to distinguish those rules that people generally accept from those rules that a minority accepts, or even from those rules that a handful of individuals idiosyncratically accept. By way of illustration, consider Valentini's defense of the claim that generally and publicly accepted norms are morally significant. Valentini begins with a parable. Imagine that you are building a sandcastle at the beach. “This is a project that matters to you,” she writes, “and one that you have taken up authentically: without fear, pressure, or brainwashing” (Valentini, 2023: 86). The fact that you have taken up this project—that you have committed to it—imposes obligations on other beach-goers (Valentini claims). They have obligations to walk around your sandcastle, rather than straight through it; if you ask them to pass you a handful of pebbles to decorate its walls, they have obligations to do so (provided that doing so is of no great inconvenience to them). These obligations, Valentini claims, derive from our more general obligations to respect people's commitments, which are expressions of their agency (Valentini, 2023: 87). And, Valentini continues, “The reason why it is wrong to violate permissible socially constructed norms… is analogous to the reason why it is wrong to interfere with your sandcastle. The norms, just like the castle, are the product of people's commitments… To respect the agency of the members of our community, those who find themselves in the context of that community must not undermine, or interfere with, the normative world they have created” (Valentini, 2023: 87).

There are two problems with Valentini's argument. First, Valentini claims that her parable about building a sandcastle demonstrates that we have obligations to respect people's authentic commitments. But this story illustrates at most the claim that we should honor people's authority within certain domains. It tells us nothing about the scope of those domains. It certainly does not vindicate the claim that people have the authority to subject one another to morally arbitrary rules simply by committing to them. But the second problem with Valentini's argument is more important. The argument begins with a parable that purports at most to concern the moral significance of an individual's normative commitments, and it moves without explanation to the conclusion that the rules that people generally accept are morally significant. But even if we have some kind of obligation to respect other individuals’ normative commitments, that does not entail that the rules that people generally accept within a community are any more morally significant than the rules that minorities or individuals accept. The claim that generally accepted rules, specifically, are morally significant seems to depend on an implicit commitment to majoritarianism or supermajoritarianism—a commitment that Valentini does not defend. And this commitment stands in need of defense, since it comes at an apparent cost. As Valentini acknowledges, strict adherence to majoritarianism or supermajoritarianism would impose permanent burdens on the members of minority communities who do not share the majority's commitments. But Valentini claims that the minority simply must accede to the majority's commitments. She writes that there is something “unsatisfactory” about this. But, she continues, “This… is an unavoidable fact of life, especially within pluralistic societies…. There is no world… which doesn’t involve frustrating valid agency-respect claims” (Valentini, 2023: 172). But this simply ignores the central question: How should we distribute the relevant frustrations? A majoritarian principle, according to which we have moral obligations to cooperate with norms that are widely accepted, necessarily privileges the commitments of the many over the commitments of the few, and Valentini gives us no reason to think this is appropriate. So those (like Valentini) who argue that acceptance norms are morally significant need to bridge the gap between, on the one hand, the claim that individuals’ attitudes and commitments are morally significant; and on the other hand, the claim that the rules people generally accept are distinctively significant.

And there are several reasons to doubt that we can bridge this gap. First, however we articulate the moral significance of individuals’ attitudes and commitments, it is unlikely that this significance will be additive. That is, it is unlikely that the mere fact that multiple people share attitudes and commitments will make those attitudes and commitments more significant than they would be if fewer people shared them. By way of illustration, consider a consequentialist account on which it is bad to violate a rule that an individual accepts, say, because this violation will cause them psychic distress (assuming they are aware of it), and because psychic distress is bad. It is not true on this account that it would be twice as bad to violate a rule that two people accept as it would be to violate a rule that one person accepts, because the two parties involved might not care equally about the rule. Nor is it true that the violation of a rule that many people accept will be worse than the violation of a rule that a few people accept, since the few might care more deeply about their rule than the many do about theirs. Moreover, psychic distress might not be the only consequence of the violation of a rule that someone accepts. On the contrary, the violation might constitute an invitation to recognize the moral arbitrariness of the rule and to consider alternative normative possibilities. Indeed, the violation of a rule that many people accept may constitute a valuable challenge to a stagnant or straitened culture. Attempting to infer from the claim that (a) the violation of a rule that someone accepts will cause them psychic distress to the conclusion that (b) the rules that people generally accept are distinctively morally significant simply ignores these complications.

Or consider a Kantian account on which we owe it to people as moral agents to respect their attitudes and commitments. (This is the kind of defense that Valentini develops in her discussion of the sandcastle.) It is an essential characteristic of Kantian moral frameworks that we must respect persons as individuals. That is, on a Kantian approach, if we must respect the fact that one person S₁ accepts some rule R, then the fact that two other people S₂ and S₃ reject R does not diminish our obligations to S₁. Rather, we ought to understand the moral significance of S₁'s acceptance of R in a way that is compatible with other people's competing normative perspectives—and by the same token, we ought to understand the moral significance of S₂'s and S₃'s rejection of R in a way that is compatible with S₁'s commitments. In order to vindicate the claim that the rules people generally accept are morally significant, we would need to show that majoritarianism (or something like it) enables us to respect everyone's moral agency equally.

At first glance, it might seem obvious that a majoritarian principle will respect everyone's moral agency equally. After all (the thought goes), the principle of majority rule is a cornerstone of democratic governance for a reason: it affords each individual exactly one “vote,” equal to everyone else's. The fact that S₁ accepts rule R is worth exactly as much as is the fact that S₂ rejects R, and it is only after other people cast their “votes” that we can determine whose normative perspectives will win the day. Normative acceptance theories, then, seem to depend only on the claim that democratic socio-political structures respect democratic citizens as equals. Must we reject this commonsense claim in order to deny that acceptance norms are morally significant?

On the contrary, there are reasons to doubt that democratic commitments will vindicate the moral significance of acceptance norms. That is because democracy is in large part about processes, not states. A rule does not count as a valid law in a democracy because all people currently accept it, but because it survives a democratically structured process—that is, a process (a) that aspires to preserve the coherence of law and policy and (b) in which all subjects enjoy equal standing to participate. Acceptance norms do not emerge from these kinds of processes, and so cannot enjoy democratic legitimacy. (I should emphasize that I am not arguing that democratic legitimacy is not an important political ideal, nor will I argue that acceptance norms lack democratic legitimacy simply because some acceptance norms are unjust. Instead, my arguments show that acceptance norms generally are not apt objects of democratic legitimacy.)

Institutional processes

First, consider the institutional processes that aim to preserve the coherence of law and policy. Voters within democratic communities often belong to multiple overlapping groups, and any two groups might agree about some rules while disagreeing about others. As a result, majorities often do not coalesce around a single homogeneous political perspective; instead, majorities sometimes arise when multiple groups successfully organize an electoral coalition around the issues on which they agree. But the members of these coalitions often continue to disagree about other issues, which may in principle form the basis of new coalitions. In light of this heterogeneity, if we simply were to take a snapshot of the rules and policies that democratic subjects endorse at any given moment, it is possible that the results would be normatively incoherent. Consider a community comprised of three equal blocs M₁, M₂, and M₃. The members of blocs M₁ and M₂ all accept rule R₁: People must ϕ. The members of blocs M₂ and M₃ accept rule R₂: If you ϕ, you must ψ. And finally, the members of blocs M₁ and M₃ accept rule R₃: You must not ψ. These rules do not cohere; they jointly entail that people must both ϕ and not-ϕ, and must both ψ and not-ψ. Critics of democracy sometimes appeal to these kinds of considerations to criticize the idea that democratic institutions express the will of the people. If individuals’ commitments do not add up to a coherent whole, what is the will of “the people” supposed to be? ⁹ But as Henry Richardson has argued, democratic institutions ought not aspire to identify the normative framework that “the people” endorse ex ante, but to generate a coherent normative framework which can count as “the will of the people” precisely because it emerges from a set of egalitarian political, bureaucratic, and legal processes (Richardson, 2002: 65). Legislatures have an implicit obligation to maintain coherent legal rules; as a result, members ought not simply write laws that capture majority opinion, but instead should craft coherent laws through debate, negotiation, and compromise. And when bureaucrats and judges do encounter apparent incoherence in the law, they have an obligation to interpret the law in the most coherent way possible. When these processes break down (as they sometimes do), incoherence within the legal and political order undermines democratic legitimacy. Democracy does not survive when the law commands and prohibits the same actions.

But acceptance norms are not the rules that emerge from an institutionalized process whose structure aims to maintain normative coherence. They simply are the rules that people generally and publicly accept within their communities. There generally are no processes that aspire to ensure that the rules that people generally accept are coherent; there are no bureaucrats or judges to interpret apparent inconsistencies in the most coherent way possible. In the case we described above, rules R₁, R₂, and R₃ all count as acceptance rules, even though they jointly put people into a normatively impossible situation. In the formalized domains of law and politics, this would represent some kind of institutional failure. But outside of these domains, it simply represents ordinary normative disagreement among the individual members of a community. There is no failure here precisely because the mere facts that people generally accept R1, R2, and R3 are not morally significant in themselves. To be sure, someone living within this community will find it impossible to act in a way to which none of their neighbors will object. But it simply is a fact of life in pluralist communities that our actions may violate rules that our neighbors accept, and this fact has no moral consequences in itself.

Political processes

Second, consider the political processes that generate a democratic community's law and policy. These processes enable people to discuss their political judgments with one another, to educate, to debate, to negotiate, to mobilize, to build (or to undermine) coalitions. To be sure, democratic theorists do not agree about the specific activities that are essential to democratic legitimacy. (Deliberative democrats, for instance, sometimes cast a skeptical eye on less obviously deliberative activities, like negotiation or certain forms of coalition building; and instead associate democracy with the more clearly deliberative activities, like discussion, education, and debate.) But whatever activities we associate with democratic governance, they only preserve democratic legitimacy if they have the right internal structure. At a minimum, individuals must have opportunities to participate in the relevant activities. A political system that locks some people out of the spaces in which political discussion and debate occur hardly counts as a democracy, even if its elections operate on principles of majority rule. More robustly, individuals must have equitable opportunities for political participation. A political system that affords some people significantly greater influence over the political process than others surely does not count as a democracy. In order to determine whether a political system enjoys democratic legitimacy, it is not enough to ask whether its people generally support its laws. We need to investigate the structure of the processes through which people arrive at and express that support.

In contrast, normative acceptance theories only define social norms as those norms that people generally and publicly support. But it is not hard to conceive of contexts in which prevailing norms might deny minorities the opportunity to express dissent. Say that a minority group M rejects a rule R₁ that the majority within their community accepts. And consider the following exclusionary rules:

R_E1: Do not discuss R₁ if members of M are present.

Although I have framed R_E1 through R_E5 in abstract terms, it should not be hard to think of relevantly similar rules that undermine various groups’ standing to criticize prevailing rules in actual communities. Consider the familiar forms of testimonial injustice, in which people reliably dismiss testimony from the members of subordinate groups (women, non-white people, queer people, unemployed people, and so on) (Fricker, 2007). Or consider sociologists Picca and Feagin's finding that white people in the United States tend to discuss race only in all-white spaces, and to insist on a norm of “colorblindness” in multiracial contexts (Picca and Feagin, 2007). This practice undermines non-white people's standing to challenge white people's conceptions of race and the role that it plays within their communities (Lawless, 2022). Proponents of normative acceptance theories thus face a normative and an empirical question. Normatively, how many exclusionary rules like R_E1 through R_E5 can prevail in a society before its acceptance norms lose democratic legitimacy? Empirically, how exclusionary are actual societies? Depending on our answers to these questions, we may find that acceptance norms enjoy democratic legitimacy in very few (if any) actual communities. (To be clear, this point does not depend on the assumption that R₁ is itself immoral. Rather, the point is that inegalitarian social processes might sustain R₁, and that this would deprive R₁ of democratic legitimacy even if people generally accept R₁.)

But the problem is not just that acceptance norms will not typically enjoy democratic legitimacy in the real world. It is that normative acceptance theories themselves conflict with democratic principles. If we think that social norms can and should enjoy some kind of democratic legitimacy, then we need to think about the ways in which people can enjoy equal standing to participate in the construction and revision of these norms. In the domain of formal democratic politics, people will have opportunities to cast votes for or against specific laws, policies, parties, or representatives. But there are no elections in the informal domain. Rather, people change the informal norms that prevail in their communities primarily by building competing normative orders. And this practice generally involves the refusal to cooperate with hegemonic norms. In other words, people can vote to repeal a law that criminalizes ϕ-ing without thereby ϕ-ing (and so engaging in criminal conduct), but people typically cannot effectively challenge an informal norm that prohibits ϕ-ing unless some of them start to ϕ while the norm is still in place. And on normative acceptance theories, this would be morally objectionable. For her part, Valentini agrees that “norm-entrepreneurs” might reasonably violate prevailing norms in order to change them (Valentini, 2023: 118). But she explicitly insists that their actions would be “mildly wrong,” and they would have a responsibility to apologize for their actions and to justify their violations to others. This is akin to claiming that people have a responsibility to apologize for voting against existing laws. These conclusions would limit our moral standing to participate actively in the construction of our relationships with others. They are, if anything, antidemocratic.

None of this is to say that social norms cannot be morally significant. As we observed at the beginning of this section, social norms can be morally significant in virtue of their content. People can have responsibilities to apologize for the violation of social norms when this violation recklessly endangers others, fails to respect others’ moral equality, or otherwise violates people's moral rights. Nor is it to say that people need not have equal or equitable standing to participate in the construction of the “normative worlds” that define their concrete relations with the members of their communities. Exclusionary rules like R_E1 through R_E5 are objectionable in part because they compromise this standing for some members of our communities. Nor, indeed, is it to say that individuals cannot subject themselves to a rule by committing to it. My claim is only that the fact that people generally accept a rule is not morally significant. We have no content-independent reason to cooperate with acceptance norms.