Does J. S. Mill's doctrine of liberty permit society to coercively restrict the use of slurs or bigoted insults? In an important article in this journal, Melina Constantine Bell argues that it does. All that is required is to couple the harm principle with “a more expansive, modern, conception of harm” (Bell Reference Bell2021: 163). It is not necessary to supplement the harm principle with an offense principle to justify social prohibitions on bigoted insults.

I agree with Bell on many points. Most centrally, we agree that directing bigoted insults at someone ought to be considered harming them. Where we differ is in our explanations of precisely why this is true, which reflects a difference in how we understand harm. Bell and I are both willing to think broadly about what constitutes harm, perhaps more broadly than Mill himself. But Bell's approach, I will argue, has illiberal implications that mine avoids.

Bell takes Mill to understand harm in terms of damage to an individual's interests, where

She distinguishes this view from the account of harm that Piers Norris Turner attributes to Mill, according to which “any direct negative consequence for others,” including any unpleasant mental state, counts as a harm (Reference Bell2021: 166; cf. Turner Reference Turner2014). This account of harm is “too broad,” she maintains, commenting that “it is critically important to distinguish harm – a tangible, objectively verifiable setback to a person's interests – from offense, which is a subjective reaction to an experience that can vary greatly among persons.”

While it seems obvious that bigoted insults cause harm on the broad conception of harm that Turner outlines, Bell argues that once we appreciate the true effects of these insults the same holds true even on the “damage to interests” conception. Before explaining why, it will be helpful first to discuss what Bell means by “bigoted insults.” She opens her article with an account of the speech code that Stanford University attempted to adopt in the 1990s. (It was struck down by the courts.) This code classified expression as prohibited harassment if:

1. a) is intended to insult or stigmatize an individual or a small number of individuals on the basis of their sex, race, color, handicap, religion, sexual orientation, or national and ethnic origin;

2. b) is addressed directly to the individual or individuals whom it insults or stigmatizes; and

3. c) makes use of insulting or “fighting” words or non-verbal symbols (Byrne Reference Byrne1991: 414).

As I read her, Bell takes these criteria to serve as a working definition of “bigoted insult.” Her leading example of such an insult, which involves a group of men who yell “n______ whore” at a young woman of color, satisfies all three (Reference Bell2021: 175).

Bell acknowledges that there are forms of bigoted speech beyond bigoted insults. She says that while her article argues only that bigoted insults should be prohibited, she wants to leave open the question whether other forms of bigoted expression should also be restricted. She may also intend to argue that one other type of expression should be restricted along with bigoted insults, namely sexist and racist jokes. Such jokes – made, say, by a comedian on stage – would not have been restricted by the Stanford code. Nevertheless, Bell describes the harms of sexist jokes in considerable detail – more detail than we might expect if they fell outside the scope of her argument. Whether she means to leave the permissibility of restrictions on such jokes an open one or takes herself to have answered it is unclear.

Bell suggests that there is strong empirical evidence that bigoted insults harm the people they target. This stretch of her argument starts with her aforementioned discussion of sexist jokes. She cites recent empirical studies that show the detrimental effects that sexist humor can have on women that go beyond mere offense. Women exposed to sexist humor “had greater body surveillance, a kind of negative self-focus that can disrupt focused attention, usurping attentional resources and reducing performance on other cognitive tasks” (Reference Bell2021: 169). Moreover, misogynist jokes “can activate discrimination against women,” a claim that Bell takes to be established by research showing that “after watching sexist comedy skits, sexist men demonstrated significantly greater willingness to cut funding for a women's organization, compared to other types of organizations” (Reference Bell2021: 169–70).

Bell next turns her attention to the harms that African Americans suffer as a result of racist speech. She cites studies showing that racial discrimination has a negative effect on the health of African Americans even after researchers control for differences in socioeconomic status; this is measured through the analysis of markers like C-reactive protein, higher levels of which in the blood suggest inflammation. Bell's suggestion is that bigoted insults are at least partly responsible for the fact that members of the groups they target experience chronic stress, which in turn has detrimental effects on their health: “Harm to physical and mental health occurs when stress levels are perpetually elevated by living in a constant state of hyper-vigilance” (Reference Bell2021: 163). While the research that she adduces was not primarily concerned with the harmful effects of racist speech (as opposed to those of discriminatory policies and actions), Bell concludes that “it would be surprising if racist jokes and insults … failed to affect African Americans in similar ways, especially given the demonstrated connection between sexist humor and harm to women” (Reference Bell2021: 170–71).

Bell depicts bigoted insults as one part of a larger framework of oppression, and she insists that we cannot appreciate their harmfulness without seeing them in this way.

Bell does not claim that bigoted insults are the only harmful forms of speech. However, she claims that they are unusual, if not unique, in terms of how little they contribute to the “marketplace of ideas.” This makes it comparatively easy for a Millian to justify restrictions on them. Bell presupposes an orthodox interpretation of the harm principle, one according to which decisions about whether society ought to interfere with someone's conduct should be made in two stages. Ignoring certain complications, we might say that the first stage is a determination of whether the conduct is harmful and so falls within society's jurisdiction. The second stage is a utilitarian calculation to ascertain whether the benefits of interference would outweigh the costs. Interference is warranted just if society has jurisdiction over the conduct and the proposed restriction would be optimal. Society is justified in prohibiting the use of bigoted insults, Bell asserts, not only because they are harmful but also because they “fail to communicate a viewpoint that serves as a basis for consideration or discussion” (Reference Bell2021: 173). As a result, Mill's arguments for the liberty of discussion in chapter 2 of On Liberty are not applicable. The utilitarian calculation accordingly favors restriction: “From the standpoint of the Greatest Happiness Principle, because [bigoted insults] do not benefit humans as progressive beings by contributing to discussion, they have no public value and produce no benefit that can justify tolerating the harms they cause.”Footnote ²

Questions can always be raised about back-of-the-envelope utilitarian calculations, and Bell's is no exception. Henry Louis Gates, Jr., for instance, challenges the notion that bigoted insults do not communicate a viewpoint (Reference Gates and Menand1996: 127–28). He also expresses doubt about whether prohibitions on these insults would accomplish very much, given that laws or policies resembling the Stanford speech code would still permit so many kinds of expression that are even more damaging. Gates observes that telling a student of color at an elite university that if they find themselves struggling in courses it is not their fault, since they were probably only admitted due to affirmative action, will likely be more “‘wounding’ and alienating to its intended audience” than attacking them with a slur (Reference Gates and Menand1996: 146).

I will not pursue these questions, however, for two reasons. First, I am chiefly concerned with whether and why bigoted insults are harmful, and hence whether the harm principle would assign society jurisdiction over them, not with the costs and benefits of restricting them. Second, while Bell and Gates are both mainly focused on restrictions on bigoted insults that take the form of formal coercive rules, e.g., speech codes, it is important to remember that for Mill the harm principle is equally applicable to ad hoc informal social punishments like angry rebukes.Footnote ³ Whether it would be optimal for a state to pass civil or criminal laws against bigoted insults or for universities to adopt the Stanford code is one question. Whether it would maximize utility to “tell off” a person who used a bigoted insult, calling them a jerk if not something stronger, is another. Even if the answer to the former question is no, the answer to the latter is surely sometimes yes. The critical question is whether society has jurisdiction over their expression in the first place, i.e., whether it is harmful.

It may not be immediately apparent that Bell is offering a more expansive conception of harm than Mill's own. She (rightly) takes him to understand harm in terms of damage to interests, and she asserts that the empirical evidence establishes that bigoted insults damage the interests of the groups they target.Footnote ⁴ It may therefore look as if she's simply taking the conception of harm that she finds in Mill and straightforwardly applying it. Yet Mill leaves the contours of his account of harm vague. Bell makes choices about how to understand aspects of harm that Mill does not explicitly discuss – choices about how to resolve this ambiguity – and those choices are in the direction of a broader conception of the concept.

What is distinctive about the harms produced by bigoted insults that concern Bell is that they involve cases in which setbacks to people's interests are a secondary consequence of a primary effect that Mill would not count as other-regarding harm. In this case, the primary effects of bigoted insults are the subjective reactions of their targets. Bell contends that these subjective reactions in turn cause the effects on people's objective states that Bell considers harms:

So, for instance, being the subject of bigoted insults contributes to long-term experiences of mental distress, and this chronic stress then results in a worsening of a person's objective states, e.g., their health.

The question whether these secondary consequences count as other-regarding harms that give society jurisdiction over bigoted insults is not explicitly answered in the text of On Liberty, so Bell's approach to thinking about harm does not outright contradict anything Mill says. It is, however, in tension with some of his remarks. Mill distinguishes between actions' direct and indirect consequences, and he observes that an action is taken out of the self-regarding sphere only if it negatively affects someone else's interests “directly, and in the first instance” (Reference Mill and Robson1977: 225).

Much could be said about how the notions of directness and indirectness should be understood here, but Mill does not say any of it; as with so many other technical points in On Liberty, he leaves this distinction frustratingly vague, offering only some examples of actions that are self-regarding even though they indirectly result in setbacks to the interests of others. These involve cases in which these effects stem from harm that the agent is inflicting on themselves, e.g., when the agent's dissolute lifestyle sets a bad example that others follow (Reference Mill and Robson1977: 281–82).

The setbacks to people's interests that Bell takes to result from bigoted insults resemble those in Mill's examples in that they are produced indirectly. Where they differ is that they are mediated not by harms that agents inflict on themselves but rather by effects on people's subjective feelings. Since Mill gives us no general theory about when harms are too indirect to take actions out of the self-regarding sphere, it is possible that he would agree with Bell that bigoted insults are harmful in virtue of their setting people's interests back. However, whatever Mill himself might have said, there is reason for Millians to worry that Bell's approach may count too many effects as harms.

Bell and I fully agree, contra Turner, that Mill would not say that the fact that a homophobe is distraught about the fact that her new gay neighbors are having sex in the privacy of their bedroom means that they are harming her. Even if the Victorian Mill would never have included it in On Liberty, this seems like a paradigmatic example of self-regarding conduct. But suppose that her stress reaches the level where her blood pressure rises dangerously high or she suffers other physical consequences. I am not sure how Bell could avoid saying that now she is being harmed, given that her objective states are being detrimentally affected. Of course, even if she is being harmed it is improbable that on balance any kind of social intervention against her neighbors would be optimal. Yet it seems that Millians should want to deny that she is being harmed at all.

Someone might object here that the disquiet that the homophobe in my example experiences should not be equated with the sort of chronic stress that Bell takes bigoted insults to produce. But we can imagine that in her mind the new neighbors are just the nearest manifestation of systemic shifts in sexual norms with which she is confronted whenever she is in public or consuming media. I do not see anything unrealistic about the possibility that people could be deeply attached enough to norms of any sort that living in a society in which those norms are regularly transgressed could induce them to experience chronic stress.Footnote ⁵ What is at issue here is how some individuals might perceive things, since it is their perceptions that drive their subjective reactions, not whether we believe that their perceptions are warranted or reasonable. Every day's news seems to feature stories that illustrate that certain individuals intensely dislike any manner of things that they ought not to dislike and experience being confronted by them on an ongoing basis as being under assault or siege – as akin, in their minds, to an “oppressive system.” There is nothing surprising in the suggestion that some of these individuals could suffer objective setbacks to their interests, e.g., to their physical or mental health, as a result. If Bell's approach to thinking about harm does not weaken the harm principle as much as Turner's, it still weakens it enough that anyone who counts themselves a Millian liberal ought to view it with suspicion. For a liberal to suggest that one is harmed by anything that one dislikes intensely enough to raise one's blood pressure or C-reactive protein levels is thus to give a hostage to fortune.

If we do not follow Bell in counting setbacks to interests that are secondary to unpleasant subjective reactions of persons as harms, then on what basis could we say that bigoted insults are harmful? There is an important difference between the mental distress experienced by the young woman in Bell's example and the distress experienced by the homophobe in mine. In Bell's example the young men intend to cause distress – that is the point of their yelling – where in my example the gay couple presumably does not. In one case, the action that causes distress is aimed at someone else; in the other, it is not. If we need to enlarge Mill's conception of harm to account for the harmfulness of bigoted insults, then doing so in a way that reflects this key difference may help us arrive at a conception that captures our intuitions about what makes certain actions harmful yet that does not include too much.

Accordingly, I propose that when applying the harm principle we might say that A has harmed B if A has directly damaged B's interests and/or A has intentionally caused B subjective pain or distress – even if that mental state does not rise to the level of threatening B's interests. This disjunctive way of conceiving harm allows for the existence of unintentional harm, since direct damage to another's objective interests harms them regardless of intent. But it also captures the plausible notion that no one can assert that society at large has no interest in what they are doing when their very intention is to harass, annoy, or in some other way cause pain to someone else – even if these subjective reactions are not so intense or sustained as to result in damaging objective states.Footnote ⁶ This way of conceiving harm therefore takes a step in the direction of Turner's, since in some cases it does count subjective displeasure as a harm, but only a small, measured step. If these subjective effects do rise to the level of affecting the individual's objective states, then this may suggest that the amount of harm being caused is greater and the case for social intervention is stronger. Perhaps the law should generally get involved only when people's interests are at least indirectly being affected. But as I noted earlier, the harm principle governs when we can tell people off as much as it does when we can criminalize their behavior, and the bar for when we are allowed to tell someone that they are being a jerk should not be set too high. If one person is going out of their way to annoy another then the harm principle should not shield them from angry rebukes, whether by the target of their harassment or a third party.

I should be clear that while I am proposing a way for Millians to think about harm, I am not making the interpretative claim that this is how Mill thought about it. I cannot point to a passage in which he even hints that subjective pain is harmful just if it is intentionally caused. At most I can say that I am also not aware of any passage in which he considers and rejects the proposal and that my proposal seems to be more in the spirit of his liberalism than is Bell's. One might see my proposal, like Bell's, as making choices about how to fill in details that Mill leaves unspecified in his account, choices that are in the direction of a more expansive conception of the concept. Or one might see it as an amendment to his account.

I have noted that there are points at which Bell's article seems to be arguing for social prohibitions not only on bigoted insults but also on biased or misogynist jokes. It might appear that my proposal would not count such jokes as harmful, at least not as a general rule. Sometimes these jokes might be told with the intention of annoying or humiliating members of the groups who serve as the jokes' butt. Often, though, the worst of these jokes are not told in their hearing. And sometimes it might be hoped that they will enjoy the humor, even if it is at their own expense. This might be seen as a deficiency of my proposal. However, Mill seems to believe that when an audience is unable to rationally consider whether to act on a speaker's words then the speech and the audience's subsequent behavior can constitute one compound action. If the behavior is harmful, then so too is the expression that sparked it. Mill illustrates this with a speaker who shouts that property is theft to an agitated mob (Reference Mill and Robson1977: 260). Bell suggests that the same reasoning applies to sexist and misogynist jokes (although not to bigoted insults), since humor bypasses “rational consideration” and these jokes damage people's interests (Reference Bell2021: 170). If she is right, then telling these jokes is harmful on my account, since if an action damages someone's interests then whether it is harmful does not depend on the agent's intent.Footnote ⁷

Some people might not experience any mental pain when a given bigoted insult is directed at them; they may, perhaps, have been inured. One implication of my proposed way of thinking about harm is that those particular people are not harmed by those particular insults. Note, however, that rebuking the person who used the insult could still be justified in terms of preventing harmful conduct. The harm principle does not prevent us from restraining or punishing someone who threw an unprovoked punch just because their blow happened to miss its intended target.

A further implication of my proposal is that it is possible to harm people by exploiting the fact that they strongly dislike things that they should not. Suppose, for instance, that the gay couple in my earlier example actually does go out of their way to hold hands or kiss in front of their homophobic neighbor, just because they know that this will disturb her. On my account, they are harming her. While this means that society has jurisdiction over their actions, however, it does not mean that intervention is warranted, e.g., that someone ought to tell them that they are being jerks. Mill believes that many harmful actions should be tolerated and that some are even commendable. Whether the couple should be criticized for aggravating their neighbor, applauded, or just ignored will depend on a utilitarian calculation that takes account of the specific circumstances of the case.

My intent in the foregoing discussion has been to position myself not as Bell's critic but as a fellow traveler, pointing out a safer and more direct route to the destination that we both want to reach. No doubt the way of thinking about harm that I have sketched here stands in need of further refinement; my ambition for this brief note has only been to draw attention to a possibility that seems worthy of fuller discussion. Note that strictly speaking one is not forced to choose between Bell's strategy for explaining why bigoted insults are harmful and mine. One might agree with Bell that the setbacks to interests that these insults cause are harms, the fact that they are mediated by subjective reactions notwithstanding, yet also agree with me that the painful subjective reactions are also harms in virtue of being intentionally caused. Note also that my proposal has implications that extend far beyond bigoted insults, since it entails that we harm others when we do anything that is intended to cause them distress or even merely to annoy them.