Speech Repression and Threat Narratives in Politics: Social Goals and Cognitive Foundations

Abstract

Political movements are often bound together by mobilizing narratives about social threat. In devoted activists, this triggers moral motivations to protect the narrative from criticism and nuance. Speech repression phenomena include public shaming on social media, the “deplatforming” and “canceling” of controversial speakers, and the intimidation of dissidents. Speech repression phenomena are most puzzling when the narratives activists try to protect are simplistic and inaccurate, which is often the case in politics. Here, I argue that speech repression derives from at least three main sociocognitive motivations. First, hypersensitive dispositions to detect threat, from hostile outgroups in particular. Second, motivations to try to keep people committed to moral causes and mobilized against dangerous groups by controlling information flows and beliefs. Third, motivations to signal personal devotion to causes and ingroups to gain status. Members of most political groups engage in speech repression, even those ostensibly committed to freedom. Political activists and leaders only need to believe that speech restriction will bring about desired effects to engage in it. While speech repression can derive from sincere convictions, insincere self‐censorship and sanctioning are widespread.

Why do political activists repress speech, especially to protect simplistic ideological narratives? I argue that speech repression stems from at least three key motivations: hypersensitivity to social threats, desires to maintain mobilizations through information control, and status‐seeking through loyalty signaling. Even freedom‐loving groups engage in speech repression when believing it serves their goals. While sometimes based on sincere convictions, insincere self‐censorship and sanctioning are widespread across political movements.

1. Explanatory Target: Speech Repression on Sociopolitical Topics

Motivations to repress speech in political activists recur across cultures and time periods, especially in contexts of perceived social threat or conflict between political, ethnic, or religious groups [1]. Speech repression takes many forms from informal social control through public shaming or dogmatic replies to censorship of scientific knowledge to the intimidation of dissidents [2, 3, 4]. Historically, the medieval Catholic Church collaborated with governments to persecute those perceived as holding heretic views [1, 5]. In the 1950s, US elites were seized by an anticommunist panic and started suspecting intellectuals and public servants with left‐wing leanings of being Soviet spies [6, 7, 8]. Recently, the United States has seen an upsurge in attempts to silence views perceived as insufficiently supportive of racial and gender equality on the left, or a race‐blind and patriotic, Christian society on the right [9, 10, 11, 12, 13, 14].

While research on the social functions and cognitive foundations of political beliefs and communication is rich, there is little work on why people try to repress speech on sociopolitical topics and protect political orthodoxies. The notion that threat‐oriented narratives are central in public discourse is well‐known, with research on the “cultural success” of conspiracy theories, fake news, rumors, witchcraft, and religious beliefs [15, 16, 17, 18, 19, 20, 21, 22]. However, there is little scholarship on the strategic goals pursued by activists engaging in speech repression, especially in horizontal, nonstate‐supported forms of censorship. Normative philosophical defenses of free speech abound [23, 24] and histories of free speech have been written [1]. In social psychology, the study of individual differences in authoritarianism and cognitive rigidity has developed into an important research field since World War II [25, 26, 27, 28]. However, those research programs typically do not focus on analyzing the common underlying goals pursued by activists who repress speech touching on sociopolitical issues.

Yet, identifying those goals is crucial to our understanding of the dynamics of public opinion and political conflict. Speech repression is an amplifier of ideological polarization, as it contributes to create echo chambers, including among academic communities putatively committed to truth‐seeking [29, 30]. Speech repression is also a typical manifestation of affective polarization (i.e., between‐group hostility), to the extent that it promotes perceptions that outgroups are evil and dangerous [31, 32].

Moreover, at a cognitive level, the fact that politicized people often become motivated to protect simplistic ideological narratives from criticism calls for an explanation.¹ Classical cognitive perspectives would assume that beliefs and claims pursue the epistemic goal of faithfully portraying how things are to facilitate practical decisions [33, 34, 35, 36]. In this classical cognitive perspective, citizens should be motivated to ensure their political beliefs are accurate, and to update those beliefs in the face of compelling evidence and arguments. This is rarely what we observe, however, with numerous examples in politicized people of motivated cognition and resistance to update core political convictions, even in the face of strong counterarguments [37, 38, 39, 40, 41]. Social science shows that most social problems (e.g., poverty, crime, unemployment) have multiple causes, require complex solutions, and cannot seriously be blamed on only one political side. Social issues are thus almost never captured by the type of simplistic or dualistic accounts that thrive in times of political conflict.

Here, drawing on research in evolutionary psychology and anthropology, cultural evolution, and political communication, I argue that mischaracterizing social reality can yield social benefits [42, 43, 44, 45, 46, 47, 48, 49]. Our discourse and thought processes about sociopolitical topics do not only pursue epistemic goals of describing the social world accurately. They also pursue social or strategic goals of incentivizing other people to behave more prosocially, defending moral communities, and enhancing personal reputations for one's benefit [3, 4, 20, 29, 38, 41, 46, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69].

In line with this framework, I theorize speech repression on sociopolitical topics as being anchored in at least three key motivations: (1) to avoid social threats, from rival groups in particular; (2) to preserve the mobilization of other people in moral causes by manipulating their beliefs; and (3) to show devotion to moral causes and political ingroups to gain status. It is possible that other motivations than those three underpin speech sanctioning on sociopolitical topics, in some contexts. Together with other processes of social cognition, the motivations of speech repressors need not be fully conscious² [58, 70, 71, 72].

The primary focus of my analysis is to characterize the subjective motivations and inferences driving speech restriction measures taken by ordinary activists on politicized topics, especially in contexts of perceived social threat. By ordinary activists, I mean citizens endowed with strong moral convictions but who typically possess no government or administrative authority. My focus is not on the top‐down and state‐supported practices of authoritarian governments, which are the object of a literature in political science [2, 73], although I may occasionally touch upon them to illustrate the mechanisms. Similarly, less clearly politically motivated forms of speech regulation are not the primary focus of my analysis (such as community punishment of people who lied on practical matters, much of professional fact‐checking, or the retractation of bad scientific research not seen as having political repercussions).

Importantly, this theory is primarily descriptive, not normative. People with different belief systems, personality traits, and convictions will diverge about which speech restriction measures count as “legitimate protection” against dangerous ideas versus are “excessively authoritarian” [74, 75, 76, 77].

2. Detecting Social Threats, Mobilizing Allies, and Enhancing One's Reputation

Human cultural behaviors rely on the execution of underlying cognitive systems and motivations (or goals). Many of them were partly shaped by natural selection to deal with past recurrent problems faced by our species in small groups of hunter‐gatherers in ways which, on average, promoted biological fitness [48, 78, 79, 80]. Of course, cultural behaviors are always socially and historically situated, so they are influenced by local cultural, economic, and institutional variables as well. But a sociocognitive theory of speech repression should begin by identifying the evolved cognitive motivations likely causing attempts to repress speech, as they have been documented by evolutionary behavioral scientists.

2.1. Detecting Exploitation Risks and Social Threats

Over the course of their evolution, humans relied heavily on mutually beneficial cooperative exchanges with others for their survival and access to mating opportunities and resources (food, shelter, help, etc.) [81, 82, 83]. To ensure individuals do not frequently loose out in their social relationships, our minds evolved highly sensitive mechanisms to detect free‐riders and social threats³ [87, 88], in particular outgroup threats [16]. Threat‐avoidance mechanisms are designed to “overshoot” like smoke detectors: they favor mistakenly detecting a threat that is in fact absent over running the risk of failing to see a threat when it is present [89, 90]. With respect to political information, this is why our minds today are so susceptible to conspiracy theories, cues to free‐riders on community resources, and “moral panics” and “elite panics” about dangerous ideas spreading in society (e.g., communists, racism, enemies of the people), a point I return to below [75, 91, 92].

2.2. Mobilizing Followers Against Social Threats

Ancestrally, individuals without social support would have been at the mercy of those with multiple allies. Group cooperation can be costly for its participants and its long‐term benefits are uncertain, so there are often reasons to lack trust in others’ cooperative engagements [81, 83]. Early in childhood, humans intuitively see others as relatively selfish utility maximizers, who try to protect their own interests and avoid suffering [93, 94]. Moreover, engagement in any cooperative project mechanically reduces one's ability to engage in other groups or causes, making coalitional support a rival good [48]. Given those parameters, humans evolved to be “groupish” animals because of the reproductive benefits they get from the support of large coalitions of cooperators, and to possess motivations to create, join, and maintain such coalitions [20, 81, 95, 96, 97].

The benefits of creating and preserving collective mobilizations are all the greater when individuals are facing threats in their environment, as is the case with group‐based conflict. Group‐based conflict is particularly harmful because intelligent and coordinated humans can typically inflict much greater costs than animal predators and natural threats, causing fitness interdependence within groups to soar [98, 99, 100, 101, 102, 103]. In contemporary hunter‐gatherer and agricultural societies which likely resemble ancestral societies, violent raids sometimes lead to the killing of all male members of a group and to the abduction of women [102, 104, 105]. While the lethality and frequency of warfare in prehistoric forager societies is the matter of a current debate [102, 103, 104, 106, 107], there is strong evidence of at least occasional collective and lethal aggression in the human lineage, which likely constituted an important selection pressure on human minds.

Considered from a simplified computational perspective, efforts to mobilize allies require to persuade⁴ them that they will benefit more from engaging for the focal cause x than from doing nothing or engaging in other ventures y, z at a given time. This is frequently done by using a rhetoric of threat, that is, arguments that emphasize the perceived costs that individuals would avoid by engaging. Specifically, mobilization efforts aim to convince allies to increase their levels of investment or commitment based on arguments about how much they stand to lose if action is not taken. Mobilizations also require the coordination of motivated participants: establishing joint attention on a common problem and the identification of what to collectively do about it, who can serve as a leader, how many people are ready for action, and so on⁵ [48, 105, 110, 111, 112].

Reflecting those evolutionary pressures and strategic constraints, individuals across cultures often behave as intuitive activists. In traditional societies, moral entrepreneurs stimulate fears of enemy groups to mobilize ingroup members in ethnic conflicts, such as the 1994 Rwanda war [105]. They utilize fears of free‐riders on communal resources or beliefs in supernatural punishment to motivate more cooperative behaviors [51, 113]. In mass modern Western societies, ideological entrepreneurs try to mobilize others to engage against enemy states (France in 1914 Germany), or infiltrated enemies (communists in 1950s United States). They do so using state propaganda, historical narratives, religious teachings, traditional tales, and educational programs [95, 114, 115]. A similar logic underpins efforts to sensitize citizens to environmental protection or fight sexism and racism through door‐to‐door activism, student seminars, and social media posts today [116, 117].

2.3. Signaling Commitment to Ingroups and Their Moral Causes

Over evolutionary time, benefitting from mutualistic cooperation required being selected as a partner [81, 82, 118]. As a result, humans have been under selection pressures to develop abilities and motivations to secure a reputation for trustworthiness to keep their rank within any social group. Within communities in which there is a significant degree of partner choice, each member is incentivized to try to appear devoted personally to the values and projects the group cares about to be chosen as a friend, ally or mate. From a computational perspective, a group member seeking to persuade other members of her devotion to a focal cause x must manage to convey that she cares highly about x, that is, more than she cares about other causes y, z she could be pursuing.

Two key parameters increase the incentives to display devotion in contexts of group mobilizations. First, in contexts of greater perceived social threat (be it real or not), demands on side‐taking increase: individuals are expected to unambiguously show to potential allies that they truly belong to the ingroup and that they are not on the side of the enemy [53, 95, 119]. The second parameter is the average level of devotion to the cause already being displayed by allies (which, typically, is increased by threat perceptions). The greater the average level of investment in the ingroup and its cooperative projects, the more devoted a signaler should appear as being, following a “competitive signaling” logic [118].

Reflecting those evolutionary selection pressures, normally developing people in most cultures show strong motivations to acquire and maintain moral status among their communities [81, 120, 121, 122, 123, 124, 125]. Some of the well documented ways in which they try to do this is by appealing to moral principles their social circles value such as fairness, patriotism, a religion, and through policing and third‐party punishment⁶ [75, 113, 132, 133, 134, 135]. Importantly, motivations to acquire moral status from ingroup members seem harder to satiate in the highly interconnected societies of the social media era—with its millions of potential followers and voters—than at earlier time periods [136, 137, 138].

Figure 1 summarizes some of key the arguments presented in Section 2.

FIGURE 1.

The higher the perceived level of social threat, the higher the individual incentives to ensure the mobilization of ingroup members against the threat, and to signal devotion to allies and the prosocial mobilization.

3. Simplistic Threat Narratives as Perceived Tools for Prosocial Mobilizations

Ancestral, historical, and modern conflicts share basic properties. They all pit groups against each other in relationships seen as zero‐sum, and fuel rhetorics of social threat. Research suggests that in contemporary mass societies, contexts of mutual hostility and “culture wars” between groups activate the evolved mechanisms above of avoiding social threat, creating and protecting large mobilizations, and showing devotion to ingroups and their causes [95, 97, 110, 139, 140, 141].

3.1. Threatening and Moralistic Information Is Cognitively Attractive, and Ubiquitous in Political Discourse

One way of conveying mobilization benefits to followers is through a narrative depicting the utopian end state of a revolution [75, 142] or the glorious achievements of heroes with whom one shares a national filiation [115]. But the most efficient way of making people believe they should pay attention to an issue and take action is through a narrative about threat. Experimental studies reveal that negatively framed information is attended to, remembered, and judged more credible than neutral information [16, 143, 144]. Threatening information is more likely to be judged interesting‐if‐true than neutrally valanced information [145]. Threat‐oriented messages are particularly successful in online communication [146, 147] and tend to be amplified in social networks, whereas many other aspects of communicated information rapidly decay [148]. Expressing threatening information also makes speakers appear competent and prosocial, since it can benefit others [17], which enables speakers to gain moral status. Group‐based threats in particular, entailing that individuals are coordinating to hurt oneself and kin, are given high priority by the mind given the costs of group conflict [149]. Political rhetoric often portrays groups as being trapped in relationships which are inescapably zero‐sum (“if they win, we lose; if we win, they lose”) and thus as leaving no choice but struggle.⁷

To make threat perceptions more compelling, ideological narratives often make heavy use of moralizing language (“wrong,” “immoral,” “evil”) to communicate that enemy ideas or groups are threatening the interests of everyone, regardless of their position [65, 75, 150, 151]. Moralistic language tends to attract more engagement and to be amplified by social media algorithms [137, 152, 153, 154]. In US contexts, perceptions of a message as touching upon a societal issue of “absolute moral importance” (e.g., racial equality, gun rights) amplify intentions to share it on social media [44]. People are also more receptive to potentially flawed reasoning if they see its conclusion as morally good [155], which can increase the “cultural success” of moralizing political messages.

These cognitive mechanisms contribute to explain why political movements recurrently craft and bind around moralistic narratives emphasizing how oppressive, existential, ubiquitous, intentional, long‐lasting or imminent a social threat is [75, 115, 156, 157]. Such narratives can help political activists and leaders solve the fundamental problem of leadership: finding “how to make my problem, our problem?” [158].

3.2. Ideological Narratives Are Abstract and Difficult to Falsify

Threat narratives vary widely in the extent to which they are grounded in facts. Claims unconstrained by reality can be made as “juicy” and attention‐grabbing as one likes, so activists who manufacture them are often tempted to embed in them malevolent intentions, which are frequently disconnected from reality [19, 105, 110, 159]. At one end of the spectrum, claims such as “the United States is structurally racist,” under certain interpretations, are based on robust historical evidence (e.g., measures of “redlining” by insurance companies and banks against Black majority neighborhoods⁸). By contrast, many political narratives postulate evil (and sometimes cartoonish) intentions like conspiracy theories about Jews controlling the world, claims that “America is invaded by a communist plot” (in 1954) or that “Democratic politicians are running a satanic child sex‐traffic” (QAnon, 2020s).

Regardless of how empirically grounded they are, however, it is important to realize that sociopolitical narratives portray events that are outside individual perception. This makes them difficult to falsify. You intuitively use your perceptual systems to assess the position of walls and furniture items in your immediate environment to avoid bumping into them. In contrast, statements like “America [or Russia] is invaded by secret foreign agents” or “There is a great replacement going on with immigration” are macro social statements about entire nations. Most of the events they point to take place in distant cities or organizations, or are unobservable in principle (e.g., harmful private thoughts; agents conspiring in hidden places). Like other culturally transmitted beliefs (tales, village gossip, etc.), we learn much of the content of political narratives through the word of others, that is, through informal conversations, education, and the media, rather than inferring them entirely from personal experience⁹ [95, 160, 161, 162]. Assessing the narratives’ accuracy would require systematic observation and induction, which almost nobody undertakes apart from a few professional social scientists. As a result, the referents of ideological narratives can only be metarepresented in the “mind's eye,” as reflective beliefs, rather than being derived from perceptual inference [15, 163, 164]. Importantly, the fact that ideological narratives and their threat components are metarepresentations implies that they can be weakly correlated with actual levels of social threat. They emerge largely from the interplay between media and politicians’ communications and citizens’ moral priorities at a given time.¹⁰

Let us also note that the concepts which compose political narratives refer to collective or aggregate dynamics, which make them semantically fuzzy [170]. What distinguishes a society in which there is a communist conspiracy from one in which there isn't? At what point does a racial mental bias or hiring practice become so frequent that it counts as “structural”? There is no clear criterion for determining what events in the world would make such abstract claims true or false.

Last but not least, those who endorse and affirm the narratives are statistically unlikely to be personally affected by the narrative being wrong. In mass modern politics, policies have low probabilities of affecting the average person [171, 172]. This is one of the mechanisms leading median citizens to remain little motivated to test and correct their beliefs about social and economic processes [172, 173, 174, 175, 176, 177, 178, 179, 180].

3.3. Stronger Perceptions of Social Threat Should Push Toward More Extreme and Dualistic Narratives

The abstract nature of social threat narratives entails that it is difficult to get feedback from experience unambiguously demonstrating the narratives miss important aspects of reality and should be updated or abandoned. Combined with their ability to tap hypersensitive systems of social threat avoidance, this enables them to appear as plausible and relevant, and thus as effective‐seeming tools for mobilizations in informal conversations, social media posts, partisan media, state propaganda, education campaigns, and university curricula.

How people reason and communicate about political topics is shaped by epistemic goals of describing the world accurately, and by social goals of getting others to behave more prosocially, fitting in communities, and enhancing personal reputations [38, 41, 57, 59, 63]. As explained in Section 2, perceptions of social threat (group‐based threat in particular) are among the evolutionary challenges that made it most useful to achieve mobilizations, as well as to signal devotion to one's community. In mass contemporary societies, one would thus expect that (H1) as fears of social threats develop, individuals become more motivated to craft, assert, and endorse simplistic (or dualistic) threat narratives as tools of mobilization (H1a) and reputational enhancement (H1b). Those hypotheses are illustrated in Figure 2. See Table 1 for a summary of the main directional hypotheses formulated in the paper.

FIGURE 2.

H1: As political activists perceive (rightfully or not) increased levels of social threat in their environment, this should increase their social motivations to exaggerate threat in their speech to persuade others to take collective action (H1a) and signal devotion to the mobilization for reputational benefit (H1b) (red arrows). Those social motivations push in opposite directions from the motivation to describe the world accurately. This gives rise to less nuanced and more simplistic, dualistic, threat‐oriented accounts (green arrow).

TABLE 1.

The main directional hypotheses formulated in this article.

Hypotheses	Section
H1: As fears of social threats develop, individuals will develop stronger motivations to craft, assert, and endorse simplistic threat narratives as tools of mobilization (H1a) and reputational enhancement (H1b).	3
H2: The more perceptions of threat and group conflict develop, the more activists should try to repress claims and sources seen as reducing people's engagement for the ingroup's causes or against the social threat.	4
H3: Speech repression should be greater on messages, persons, and information sources believed to have large (vs. small) undermining effects on mobilization efforts for a moral cause or against a threat.	4
H4: Repressing views and sources that question the importance of the cause, or hurt the reputation of the ingroup, can be done to signal commitment to those.	5
H5: Refraining from airing opinions seen as taboo or demobilizing with respect to a prosocial cause can be motivated by intentions to safeguard impressions of commitment to the cause, to avoid reputational costs.	5

3.4. Ideological Narratives Do Not Need to Have Strong Behavioral Effects to be Regarded as Mobilizing

Importantly, I do not want to assume that threat narratives are always potent at effectively getting people to take actions. It is enough that ideological activists and leaders believe they have behavioral effects on others for the activists to be motivated to spread them.

Granted, repeated exposure to ideological narratives may sometimes succeed in increasing people's willingness to engage more in the cause. In line with this persuasion hypothesis, some studies find that some costly behaviors are associated with strong threat perceptions and conspiratorial beliefs. This is the case of terrorist attacks by extreme left [181] and extreme right groups in 1970s Europe [182], jihadism [99, 183], violent street protests by the French Yellow Vests [184], and the invasion of the Washington Capitol by Trump supporters on January sixth 2021 [185, 186]. In those cases, deeply held fears of group‐based threat combined with strong moral grievances undoubtedly played a causal role in motivating individuals to take personal risks.

On the other hand, it would be naïve to assume that just because they are judged relevant and propagated, ideological narratives have strong effects on behavior. First, exposure to an ideological narrative does not entail its being endorsed by the audiences. The success of mass persuasion, political rhetoric, and propaganda is strongly constrained by the existence of favorable prior beliefs¹¹ [161, 191, 192].

Second, talk is cheap and explicit cultural beliefs must not be taken at face value. As emphasized by Hugo Mercier, Dan Sperber et al. on epistemic vigilance—our intuitive mechanisms of defense against misinformation and deception—there should have been selection pressures against engaging in personally costly behaviors on the basis of ambiguous sensory evidence [191, 193, 194, 195]. Abstract ideological narratives always have this ambiguity. In fact, it is a routine feature of political attitudes that people claim to endorse them—sometimes with high confidence—without the beliefs having strong motivational effects,¹² especially when their behavioral effects are not reputationally rewarded. This belief‐action gap is clear in the case of moral commitments to protect the environment, where it is such an effort for us to stop taking the plane and accept paying train tickets at 3× the price [198].

Regardless of their actual motivational effects, it is enough that activists and leaders believe that ideological narratives will get people to mobilize for moralized causes and against threats for them to be motivated to spread the narratives in society [113, 115, 199].

4. Speech Repression as a Perceived Strategy for Preserving Mobilizations and Commitments

Let's hope it's not true; but if it is true,

let's hope that it does not become widely known.¹³

4.1. Repressing Speech to Preserve Prosocial Mobilizations and Commitments

Ensuring that mobilizing narratives are widely disseminated in society and protected from criticism becomes the stake of fierce political battles of influence from leaders and activists. A typical ideological narrative is a description of social threat from which followers’ minds infer a need for collective action. From the claim that “American corporate values are threatened by communist influence,” implications such as we must detect socialist sympathies wherever they are in the administration can be drawn. From the statement that “racism is endemic in society,” people can draw conclusions like it is probably around us now and we must be vigilant about racially offensive remarks on campus and in elites’ discourse. In this sense, a statement about threat operates like a cognitive “vector” that pushes with more or less force toward more cooperative investment in moralized cooperative efforts. This entails that the vector's force is constantly undermined by contrarian arguments, including true scientific facts, that if they spread from mind to mind, might lead followers to lower their investment in the cause. This could be the case of statements like “communism/racism/illegal immigration is not that widespread, you're dramatizing.” Or, hypothetically, of nuanced and accurate scientific papers finding that the police does not proportionately kill more Black than White defendants (for liberals denoucing police violence), or demonstrating that White Americans lack knowledge about economic inequality along racial lines (for White conservatives who deny structural racial disparities).

From the perspective of an ideological entrepreneur concerned with maintaining widespread support for their cherished cause, free speech therefore opens a risk. It creates the danger that views nuancing the saliency of a threat—or the benefits of a mobilization—travel through social networks and reach an increasing number of people. It is the dissemination of communicated information susceptible to “dilute” accounts of social threat, or undermine beneficial commitments, that speech repressors viscerally want to prevent. This fear, I suspect, takes the form of a (potentially unconscious) mental representation of a slippery slope or snowball effect. That is, of a representation of contrarian claims spreading in information cascades, thereby leading individuals to care less about important moral causes and tolerating bad behaviors more¹⁴ (e.g., racism, sexism, communism, criticism of national grandeur).

Specifically, the demobilizing effects of contrarian information traveling may unfold through two anticipated mechanisms. First, directly, through shifts in individuals’ personal beliefs toward less perceived benefits of the mobilization. Second, the effects may happen indirectly, through individuals getting to believe that contrarian views will cause other people to see less benefits in the mobilization and decrease their investments. The latter is the belief that individuals may feel “socially authorized” to care less about the cause because they believe other people do not care about it. It is about a change in common knowledge in society [62, 112]. Many speech repression measures on sociopolitical topics likely aim to avoid shifts in personal beliefs and in common knowledge. As Pinker puts it, “people hold many factual beliefs not because the beliefs have been shown to be true but because they are felt to uphold a moral order” [62]. As a result, activists try to restrict contrarian speech, even on scientific facts, that is seen as threatening “moral orders.” Figure 3 provides a graphic illustration of this logic of repressing speech to preserve perceptions of social threat and of the benefits of mobilizations.

FIGURE 3.

Bell curves are prior beliefs among the ingroup of a leader or activist, or among neutral audiences. If free speech: Free speech opens the risk that contrarian views create a shift in prior beliefs among audiences that may undermine the perceived truth of a threat or mobilizing narrative. If speech is repressed: Active repression of contrarian views and sources by ideological activists reduces the likelihood of such prior beliefs shift. Systematic repression of contrarian views can create an echo chamber and further extremism of the narrative (dark blue arrow, bottom left).

In support of this account, people with stronger beliefs that information will negatively impact others’ behaviors—who are more subject to the “third‐person” effect—are more motivated to censor the media [202] and more afraid of misinformation on TV and social media [203]. Similarly, a study of exposure to ISIS propaganda found that being more subject to the third‐person effect predicted greater support for government censorship and restrictions of pro‐ISIS content on social media platforms [204]. Finally, a study of censorship of scientific findings by Clark et al. found that across “six sets of scientific findings, [lay US] people consistently overestimated others’ harmful reactions (medium to large average effect sizes)” (abstract), even when financially incentivized to guess accurately [205]. Moreover, “harm overestimations were also consistently associated with stronger desires to censor science” (p. 844).

Activists may try to restrict speech to protect prosocial mobilizations and commitments with slightly different goals in mind. Many cases of speech repression, especially in Western societies, are justified by fears of offending and hurting the feelings or status of racial and sexual minorities (cf. uses of “hate speech” accusations) [1, 9, 12, 206, 207, 208]. This is presumably because, in Western societies that widely condemn physical and verbal violence, care motivations are increasingly high among citizens and seen as being least objectionable.¹⁵ It is important to appreciate that many speech repression measures invoking concerns for victims, however, may be policing‐oriented in the sense that they are (i) primarily concerned with how neutral third parties in society would start misbehaving if antisocial information were to enter their cognitive systems.

Other speech repression motivations may have more to do with mobilizations for perceived intergroup conflict strictly speaking. In line with arguments about the high need for ingroup motivation and coordination in group conflict (Section 2), some cases of speech repression may be (ii) primarily aimed at controlling beliefs within the political ingroup to ensure they remain favorable to the mobilization (e.g., preserving strong convictions that the United States is structurally racist today/threatened by communism in 1950s). Here, the goal would primarily be to preserve ingroup orthodoxy and consensus to preserve ingroup coordination and cohesion [5, 28, 210]. The spirit of this would be similar to that of an army general trying to maintain coordination among its troops as they are marching to battle. Such concerns may be key when preparing costly, real life collective actions, like a political protest.

There are other possibilities. Pinsof et al. make the point that informal political discussions and public communications are often about embellishing the reputation of one's political ingroup and its allies and derogating the image of rival groups. According to Pinsof et al., the goal of such expressions is to increase support for one's side and decrease support for rival groups, in particular among neutral or undecided audiences [211]. Similarly, some speech repression measures may aim to influence undecided audiences’ beliefs by (iii) restricting views perceived as damaging the image of the ingroup and its causes, and by (iv) blocking speech giving reasons to see rival outgroups and causes favorably (e.g., arguments and economic facts favorable to capitalism or CEOs for committed liberals; arguments highlighting the usefulness of illegal immigrants to the economy for strong conservatives).

It is often difficult to know whether speech repression measures primarily aim to influence the beliefs and conducts of ingroup members, neutral audiences, or outgroup members, or a mix of those. It is also often unclear when speech repression primarily aims to protect ingroup unity and boundaries for their own sake versus to protect more goal‐driven or instrumental mobilizations (e.g., a protest or national social movement demanding a new political right). Disentangling those nuanced mobilization‐related motivations (i–iv) requires controlled experiments, which to my knowledge do not exist. And of course, different activists within a given movement may have different motivations to restrict speech they fuzzily see as harmful or demobilizing.

4.2. Stronger Perceptions of Group Conflict and Source Authority Should Push Toward More Mobilization‐Oriented Speech Repression

Elevated perceptions of social threat, such as in times of perceived group competition, may be expected to incentivize activists to affirm more extreme and dualistic threat narratives as tools of mobilization (H1a). A related hypothesis can be drawn. Namely, that (H2) the more perceptions of social threat and group conflict develop, the more activists should try to repress claims and sources seen as reducing citizens’ engagement for the ingroup's causes or against the social threat¹⁶ (see Table 1). In support of this hypothesis, historical evidence suggests that speech repression measures often covary with beliefs about core moral values or ingroups’ cohesion and safety being under threat, whether by external enemies or betrayers inside the ingroup. Throughout Western history, councils of the medieval Catholic Church issued dogmas (i.e., policies) which severity could be predicted by the degree of perceived threat to the Church's unity, which ranged from minor controversies over rituals to schisms sparked by Luther and his rejection of core Church teachings [1, 5]. In the 1950s Unites States, the anticommunist witch‐hunt (dubbed “Second Red Scare” and “McCarthyism”) took place in a context of geopolitical rivalry with the USSR and moral panic about communist influence in US cultural institutions and governmental agencies [6, 7, 8, 214]. McCarthy justified his action by the view that the Unites States was “engaged in a final, all‐out battle between communist atheism and Christianity” [1, 6]. Today, dictatorial regimes such as China and Russia are engaged in a global struggle of influence with Western democracies. They use their administrations to surveil, censor, and imprison critics of the regimes lest they might endanger narratives about exemplarity of the nation (e.g., role in WW2) or their leaders, and lead to citizen mobilizations demanding more freedoms and democracy [215, 216, 217].

American “culture wars” took place in the highly affectively polarized context of the 2010s–2020s Unites States. Activists on the left engaged in public displays of outrage on social media (“dunking”). Some demanded the “cancelling” and “deplatforming” of thinkers, celebrities, and scholars whose statements or research conclusions were seen as insufficiently supportive of racial and gender equality [9, 12, 29, 206, 218]. Within academia, liberal social scientists, and students in recent decades have engaged in attempts to block the social dissemination of empirical conclusions—on the genetics of IQ for instance—seen as creating slippery slopes toward biological racism. They have actively tried to censor talks, theories, and empirical reports suggesting innate psychological differences between individuals and the sexes, out of fears anchored in the fallacy that these might undermine moral equality between groups [30, 61, 205, 219, 220]. On the US political right, Republican leaders hostile to liberals and Democrats have attempted to censor teachings of critical race theory and the social influences on gender in school manuals. Those leaders were often driven by the belief that those teachings might threaten widespread support for a race‐blind and socially conservative vision of US society, in a context of intense criticism of traditional values by secular progressives [220, 221, 222, 223].

The notion that prosocial mobilizations depend on controlling information flows, private beliefs, and common knowledge leads to another key prediction. Namely, that when speech repression is motivated by keeping people committed to causes seen as morally desirable, (H3) speech repression should be greater on messages, persons, and information sources believed to have large (vs. small) undermining effects on mobilization efforts. For instance, the greater the authority of a social media influencer or politician, the larger the number of people reading its public posts, and so forth, the more one would expect activists to want to “cancel” them if they disseminate discourse seen as harmful or demobilizing. Attempts to “cancel” Harry Potter’s author J. K. Rowling after she made statements judged by liberal activists as derogatory toward transgender women on Twitter probably illustrate this logic [224]. So does the suspension of US President Donald Trump's Facebook and Twitter accounts after his incitement of the January sixth 2021 Capitol invasion [225]. Still in support of H3, in a qualitative study of 23 professional fact‐checkers from 15 countries from the West and Global South, Sehat et al. [226] found that fact‐checkers typically focus their efforts on messages and sources they see as spreading highly societally “harmful” misinformation and which they believe a lot of people are attending to.

4.3. Speech Repression Is Based on “Folk” Beliefs That It Will Preserve Mobilizations, Which Can be Wrong

I argue that many speech restriction tactics are anchored in motivations for strategically keeping people mobilized around prosocial causes through the manipulation of their beliefs. Those motivations, which likely evolved in small forager societies to facilitate ingroup cooperation, activate in mass modern environments via perceptions of social threat and group competition. This gives rise to folk beliefs in political activists that doing speech repression will help them preserve prosocial mobilizations and commitments.

Importantly, however, speech repression tactics may very well fail to produce their intended effects of preserving mobilizations or have side effects that overcome their intended effects. It is difficult to control information flows on large scales in mass modern societies, especially when one does not have power over the administration or the rules of moderation on social media platforms. Moreover, speech sanctioning seen as excessively authoritarian may turn opponents into martyrs and cause the restricted speech to become more popular, undermining the speech repressors’ goals [1, 13]. Whether speech restriction brings about its desired effects is an empirical question, not analyzed in this article.

More generally, many cultural practices can remain widespread simply because agents believe they help them reach their goals¹⁷ [199]. In small‐scale societies, examples of effective‐seeming (but superstitious) cultural practices are beliefs and rituals around rain magic [227], shamanism [21], divination [228], and supernatural punishment [51, 113]. In mass modern societies, policy makers sometimes deploy economic or health policies that do not bring about their desired effects, or which evidence has shown to be ineffective or counterproductive¹⁸ [170, 229, 230, 231].

5. Speech Repression to Signal Devotion and Gain Status

Let us now analyze another key goal of speech restriction: the signaling of moral and political allegiances. Research on costly helping and moralistic punishment has shown that those behaviors can be motivated, including unconsciously, by reputational benefits. The evidence shows that those who engage in third‐party punishment¹⁹ tend to be seen as more trustworthy, and potentially also that they behave in more prosocial ways, and are thus more likely to be chosen as partners in Trust games and in real life [81, 133, 234]. This creates incentives for people to engage in more punishment when they know they are being watched, to signal commitment to norms of prosocial behavior (e.g., fairness, group loyalty) and enhance their reputation. Similarly, in psychological studies, expressions of moral outrage are predicted by guilt about one's own past moral failings, expressions presumably motivated (unconsciously) by attempts at reputational embellishment [235]. Motivations to share hostile news targeting the political outgroups are amplified when people believe their political friends are reading them [236], and among people with status‐seeking orientations [45, 47, 124, 237, 238], highlighting the importance of reputational concerns in political expressions.

5.1. Repressing Speech to Signal Moral and Political Commitments

In contexts where there are ongoing efforts to keep people motivated around prosocial causes, cracking down on views and sources that question the importance of the cause, or hurt the reputation of the ingroup, can be done to signal commitment to those (H4). What would be the mechanism here? Audiences have a folk model of human nature, such that they expect a normal agent to value several things simultaneously: to be committed to the prosocial cause or ingroup x, but also to value the truth and to value letting people sincerely express themselves and to value getting along with other people, and so on. As a result, an agent's attempt to restrict the speech of a target group to protect a narrative promoting x indicates to observers that the agent prioritizes advancement of x above the other values of getting along with the people being censored, promoting free expression and truth‐seeking, saving her own energy, and so forth.²⁰ In other words, the extent to which an activist cares about a cause is measured by all the things she is willing to sacrifice for that cause.

A value that speech repressors will often choose (or accept) to sever, beyond the energetic costs of censoring, is good relationships with other groups. That is, good relationships not just with those whose views are being repressed, but also with neutral audiences who would be expected to value free expression and truth‐seeking. Precisely because damaging those connections entails loosing cooperative opportunities, it can credibly signal commitment to a focal group and its causes. It can also advertise what side of a political divide one wants to be seen as siding with. Both effects entail “burning bridges” and can confer reputational benefits [67, 95, 161]. Note that for the speech repression measures to reveal political convictions and group allegiances, observers must possess implicit “mental maps” of the different groups composing society, and of who typically believes and opposes which opinion at the time [128, 211].

Historically and today, activists do seem to repress views perceived as undermining core mobilizing narratives to signal personal devotion to cherished causes and groups (H4). There is also evidence that this strategy sometimes translates into reputational capital. One of the main reasons people gave for supporting McCarthy's witch‐hunt was that “They admire greatly his ‘courage’ and ‘sincerity,’ feeling that he is not afraid to ‘get tough’” [7] (p. 252). Expressed intentions to crack down on individuals suspected of socialist sympathies in the State Department and believed to weaken US security turned McCarthy from an unknown senator into a national figure almost overnight. A study of the Anti‐Rightist Campaign (1957–1959) in Maoist China found that “bureaucrats whose backgrounds are not trusted by the leader tend to implement repressive tasks more fervently in order to signal their loyalty to the autocrat” [239]. Other authoritarian leaders seem to have succeeded in benefitting electorally from expressing unwillingness to compromise with critics of idealized nationalistic narratives, for Vladimir Putin [240, 241] and the status privileges of declining white demographic groups, for Donald Trump [242, 243, 244]. Those observations dovetail with some lab experiments showing that lay respondents see authoritarian leaders as more appropriate and competent in group conflict like wars. Citizens seem to infer that in conflict, strong leaders’ antisocial tendencies become useful to bargain for better treatment by rival groups and suppress free‐riding within the ingroup [109, 243, 245, 246].

Among ordinary activists on social media, sharing vitriolic posts and making grandstanding claims attacking political opponents are often correlated with status‐seeking inclinations. Those behaviors can be amplified by recognition from fellow partisans [45, 47, 124, 136, 237]. On social media like Twitter/X, users who “dunk” or “pile on” publicly on those who said something they strongly disagree with can do so as a form of moral grandstanding for reputational embellishment [47, 124, 235]. Online experiments on speech restriction confirm those real‐world observations. Studies by Ahn et al. indicate that sanctioning speech seen as offensive to, for example, transgender rights, is seen by partisans as sending as reliable a signal of Democratic partisanship than explicitly campaigning for a Democratic candidate²¹ [9]. Symmetrical effects were found for sanctioning of proliberal speech. Furthermore, although most study participants were reluctant to partake in speech sanctioning, they were more likely to express willingness to “report” its author when being primed about the existence of restrictive speech norms among their ingroup [9]. Witnessing that other copartisans had sanctioned the political Tweet (vs. being a first mover in critiquing it) similarly increased speech sanctioning. Those effects suggest that speech repression happens more when its reputational benefit among the ingroup increases, or its reputational cost from neutral and uncongenial audiences decreases [9].

5.2. Reputationally Motivated Self‐Censorship

The logic of signaling for reputational benefit also explains motivations to self‐censor and even lie about one's true beliefs in contexts of established political orthodoxies. Political subcultures often demand of their members to endorse (or at least seemingly respect) core moral beliefs to preserve internal cohesion and harmony [3, 4, 28, 60, 65, 151, 210, 248]. This is especially true in contexts of perceived threat because of higher demands on commitment and coordination among the ingroup [5, 75, 249]. In those contexts, (H5) refraining from airing opinions seen as morally taboo or demobilizing with respect to a cause x can be motivated by intentions to safeguard impressions of commitment to x, to avoid losing standing in the eyes of people committed to the cause²² [3, 4, 9, 250]. Self‐censorship skills are deeply ingrained in nonclinical human psychology, and their expression comes in degrees. “Spirals of silence” in Noelle‐Neumann's sense arise when people collectively refrain from expressing a socially disapproved opinion [60]. “Preference falsification” and “knowledge falsification,” as theorized by Kuran, are more active processes in which people figure they had better broadcast beliefs they do not hold to avoid reputational costs [3]. When people do not realize the extent of self‐censorship in society, these two processes can generate pluralistic ignorance: situations in which publicly expressed speech (e.g., about a social group's traits or policy's popularity) becomes misaligned with privately held beliefs without people being aware of it [3, 4, 62, 251, 252].

There is ample experimental and historical evidence of self‐censorship and preference falsification. In the lab, Asch's experiments famously showed that even on simple perceptual tasks, people will lie about what they see to align with the group [253]. During the anticommunist witch‐hunt of McCarthyism, many US citizens strategically chose to silence their politically left‐wing convictions. The “American tolerance for anticonformists waned” and “the fear of being branded a communist put the fight for racial equality through civil liberties on pause” [1, 254] (Mchangama, 2022, p. 297). Through list experiments run in the 1990s–2000s meant to reduce desirability bias in responses, political scientists have shown that up to 40% of US liberals under‐report their level of disapproval toward affirmative action [255] and immigration [256] to avoid backlash from other liberals.

5.3. Collective Self‐Censorship Can Create False Illusions of Consensus

In the United States and other Western countries, recent cycles of speech sanctioning have given rise to claims that left‐wing and right‐wing forms of “cancel culture” would have set in [13, 206, 257, 258]. It is important to stress, however, that impressions of widespread support for speech sanctioning measures can be largely explained by the costs of speaking out against what are believed to be political orthodoxies [3, 49, 140, 259]. The more people believe that an active minority of activists around them might punish them for airing views that challenge a sacred political narrative, the higher the perceived social and material costs of publicly challenging the narrative (e.g., lost jobs, contracts, paying subscribers, damaged academic reputations, etc.). One should therefore be wary of inferring consensus or the existence of an ingrained “culture” from the rarity of contrarian speech. In the presence of a minority of righteous activists willing to call out and punish contrarians, collective spirals of silence [60] and preference falsification [3] may be more parsimonious explanations. Recent studies of the US “culture wars” suggest that most liberals did not privately support practices of sanctioning speech skeptical of left‐wing causes such as trans rights or Black Lives Matter on social media [9]. However, many avoided expressing their disapproval of restrictive speech codes out of a fear of being sanctioned themselves.²³

5.4. Speech Repression for Reputational Benefit: To Stand Out Versus to Protect Oneself

Some types of reputation‐driven speech repression aim at standing out from the crowd by signaling more devotion to a group or mobilization than the average group member (the “stand out” or “show off” type). By contrast, other forms of speech repression are meant to avoid appearing insufficiently committed to the cause (the “self‐protection” type). Political movements that developed into repressive moral panics, such as the Second Red Scare of McCarthyism [7, 214], the French Revolution's Reign of Terror [260, 261], and the Cultural Revolution in China [262], typically displayed both types. To illustrate, in the early days of the Second Red Scare, Senator McCarthy gained popularity as a leader by organizing inquisitive congressional hearings in response to fears among the electorate about communist influence threatening US security. By undertaking an uncompromising search for leftists in government and entertainment, McCarthy made a name for himself as a dedicated, patriotic politician. Being among the first ones to take the initiative to censor contrarian speech—versus the hundredth person to do so—enables ideological entrepreneurs to score status points among congenial audiences [7, 239]. This is because by virtue of being isolated, the initiative they take is more costly reputationally (and energetically), given likely backlash from uncongenial audiences, and without the insurance of being backed up by coordinated allies.

Then, more and more people opportunistically participate in the censoring and denunciation, as competitive signaling sets in [118, 131, 134]. But claiming to denounce immoral views or traitors gradually appears as banal and even socially demanded of ingroup members. Speech repression's ability to operate as a signal of devotion progressively wanes. It is those who do not participate who now risk appearing as insufficiently devoted and suspicious. The movement turns into a witch‐hunt when an atmosphere of paranoia spreads [1, 7]. In this new context, publicly denouncing others as witches or “enemies of the people” and so forth can become a defensive tool for deflecting accusations of insufficient commitment to the community's values. Existence of a veil of ignorance about who might become the target of the witch‐hunt next—that is, an atmosphere of random accusations—may be a key trigger of self‐protective speech repression.²⁴ Future research would benefit from examining whether the motivations that initiate moral panics and witch‐hunts differ in kind from those that perpetuate them (Karl Frost, correspondence).

5.5. Speech Repression to Eliminate Rivals

Section 5 focuses on the reputational and status benefits of censoring. It should not omit, however, that taking the initiative of denouncing heretics or disseminators of harmful or demobilizing speech can also be a way to accrue power and influence for oneself or one's clique within a movement. This has frequently been done by ambitious party leaders to eliminate competitors. Stalin's purges of “Trotskyism” in the Soviet Union of the 1920–1930s, Mao's tactics during the Cultural Revolution in China, and McCarthyism provide examples of this [2, 214, 262]. Allegiance signaling primarily benefits the censor indirectly by improving their standing with ingroup members and those already in power. By contrast, the elimination of rivals directly benefits the censor's influence by removing obstacles to their advancement. Because the main focus of my article is on speech repression measures by citizens who are not professional politicians, I will not expand further on this motivation.

6. Well‐Intended Speech Repression Has Negative Effects

6.1. Speech Repression Can Amplify Hostility

Speech repression is a frequent amplifier of affective polarization or between‐group hostility. This is because witnessing speech restriction acts by a political side give their opponents and neutral observers reasons to believe they are illiberal, refuse compromise, and to reciprocate the censorship [31, 32, 263]. Witnessing speech sanctioning can also reinforce the belief that average outgroup members are more radical than they really are, a misperception which decreases willingness to compromise with and tolerate ideological opponents [264, 265, 266].

6.2. Speech Repression Impairs Public Opinion Accuracy and Scientific Inquiry

Moreover, speech suppression is a contributor of ideological polarization and widespread ignorance. If sufficiently large scale—in particular when ideological leaders can control entire administrative and educational systems—speech repression and self‐censorship can generate echo chambers and cause the extinction of contrarian views. Reduction of competition on the marketplace of ideas can lead to increasingly extreme ideological narratives among activists, widespread ignorance of the status quo's disadvantages, and less pragmatic policy propositions. It can also undermine political movements’ and institutions’ missions, and prevent them from solving the issues listed above [3] (see bottom of Figure 3). As J. S. Mill emphasized in the XIXth century, free speech is the only correcting mechanism that humans have for becoming aware of the blind spots of their beliefs and of suboptimal decisions [267].

The negative epistemic side effects of (self‐)censorship easily arise on political topics—such as economic policy or technological risk—on which beliefs are acquired mostly through the word of politicians, journalists, and pundits, rather than through direct perception [3, 161, 162, 170]. The negative epistemic side effects of (self‐)censorship are perhaps best illustrated in the case of dictators, who lose any ability to receive useful feedback on the efficiency of their policies or military campaigns once they have managed to deter any expression of dissent [2, 3]. Thus, the restriction of free expression is likely to come at an epistemic cost even for those who commandeered it in the first place.

Free expression is thus an obvious prerequisite to scientific inquiry. Researchers need to be able to pinpoint the limits of existing empirical findings and theories, introduce new data and research questions, and mutually correct their individual biases [58, 268, 269]. Politically motivated (self‐)censorship among scientists severely impairs policy makers’ ability to quantify and explain social issues accurately, and to design effective corrective interventions²⁵ [29, 30, 91, 205, 272].

7. Speech Repression Is Politically Recurrent and Sometimes Based on Sincere Convictions

7.1. Most, If Not All, Political Groups Hypocritically Engage in Speech Restriction

Looking at the historical record, most if not all political and religious groups are tempted to engage in speech restriction when it favors their interests. This is true even of groups putatively committed to free speech and in favor of “freedom” and “democracy.” This point is explicitly made by J. Mchangama,²⁶ author of a history of free speech in 2022 [1]. Psychologically, the default state of political cognition with respect to free expression may be a state of moral hypocrisy [273]. That is, intuitive inclinations to herald commitments to free speech to adorn personal and ingroups’ reputations, all the while applying free speech rights with double standards when this can help preserve mobilizations and express devotion—as long as the hypocrisy is not called out [56, 274]. If this is true, political and legal efforts to ensure free speech, even in democratic societies, imply a continuous fight against deep‐seated tendencies of human nature [1].

7.2. Speech Repression and Sincere Convictions

Average humans are skilled at identifying their audiences’ fears and interests, and at invoking the moral principles that make them appear as trustworthy in their social circles [135, 275]. Political leaders, press secretaries, and spin doctors have dedicated years to honing such skills as professionals [57, 71, 276]. Many political activists may thus reap mobilization and reputational benefits from speech restriction measures without genuinely endorsing the moral views being protected.

That being said, an important principle of strategy is that people who sincerely endorse the values they preach are more likely to persuade addressees to support them [57, 63, 66, 277]. After all, audiences evolved cognitive mechanisms of epistemic vigilance, which incline them to distrust individuals who appear as pursuing their own interests to the detriment of others [191, 195]. In response, speakers must seem committed to causes other people care about to attract trust [48, 132, 135, 278].

Second, people who are genuinely taken over by the narratives they preach are more creative at finding arguments in support for them, and at elaborating convincing post hoc rationalizations. They are less likely to be caught making inconsistent claims or lying [66]. Third, given that speakers benefit from appearing devoted to impartial principles [132, 135, 275, 279], speakers may be most persuasive when the links between the political views they promote and their self‐interest are inaccessible to consciousness [66, 135]. For all these reasons, sincere endorsement of political beliefs may facilitate persuasiveness.²⁷

In line with those notions, ethnographic and experimental studies of devoted actors such as jihadists, environmental activists, conspiracy theorists, or investigative journalists find that they are driven by sincere moral convictions [40, 99, 281, 282, 283, 284]. Devoted political activists do not see themselves as protecting a contingent ideological narrative or set of hypotheses about how the social world works. Only the narratives of others are perceived as stories.²⁸

In particular, doubts about the ideological narrative's truth are all the more likely to vanish as activists—building on initial convictions—have invested resources and time in promoting the narrative and bonded with other faithful members. Illustrating the power of self‐persuasion, experimental studies find that when being requested to advocate for a particular moral or political position—including a position they have no prior stake in defending—people's beliefs shift. They move slightly toward the position they have been asked to pitch [287, 288], or which are in their economic interest of the moment [42].

Granted, it is always difficult to empirically establish moral sincerity (and people often have mixed motives). Still, the above arguments and evidence make one expect that in political movements, at least some devoted activists will want to restrict speech based on deeply sincere moral convictions.²⁹ (Of course, many other individuals may self‐censor for reputational reasons, as stressed in Sections 5 and 6.) Recent experimental studies by Ashokkumar et al. of censoring behavior in US contexts support this hypothesis about sincerity. Motivations to censor counterattitudinal social media comments touching on abortion and gun rights were stronger among respondents whose subjective identity was “fused” with those causes than among those who were not fused [290]. Similarly, in recent empirical studies of Western academics, genuine moral concerns about true scientific findings having dangerous social repercussions (e.g., “could harm vulnerable groups,” “reinforce negative stereotypes”) came out as a prominent motivation to prevent their publication and dissemination [29, 205, 207].

8. Conclusion

Moral motivations to protect mobilizing, often threat‐oriented, narratives from dilution by criticism through speech repression recur in times of political conflict. They are puzzling given that motivations to form true, nuanced beliefs about the social world and to be open to flexibly update them might appear most adaptive at first sight. However, joining a growing number of social cognition scholars, I emphasized that there can be social benefits to mischaracterizing social reality. I proposed that speech repression on sociopolitical topics may emerge from the activation of at least three evolved motivations (others might exist). Namely, to detect and avoid social threats; to control information flows to preserve other people's commitment to moral causes; and to signal devotion to causes one's allies value to gain status.

Activists and leaders from most political groups opportunistically take speech restriction measures when it advances their interests, worldviews and reputations, above and beyond claimed adhesion to free speech. Speech sanctioning can be anchored in sincere convictions, but it can also arise from deliberate strategies, or from collective spirals of self‐censorship and preference falsification. Importantly, speech restriction measures may or may not bring about the effects the activists aimed at and potentially generate side effects (such as escalating between‐group hostility). In my account, it is enough that activists and leaders believe that restricting speech will produce its desired effects for them to be motivated to do it.

Avenues for future research include testing more systematically the hypotheses laid out above on the motivations behind speech repression using laboratory experiments, quantitative historical studies and cross‐cultural ethnographic data. Future work would also benefit from investigating the actual effects of speech repression on the dynamics of public opinion and political conflict.

Conflicts of Interest

The author declares no conflicts of interest.

Marie A., “Speech Repression and Threat Narratives in Politics: Social Goals and Cognitive Foundations.” Annals of the New York Academy of Sciences 1553, no. 1 (2025): 75–95. 10.1111/nyas.70089