Making Resistance Visible:  The Role of Nonverbal-Symbolic Opposition to Hate Speech Across Cultures

Making Resistance Visible: The Role of Nonverbal-Symbolic Opposition to Hate Speech Across Cultures | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Making Resistance Visible: The Role of Nonverbal-Symbolic Opposition to Hate Speech Across Cultures Jimena Zapata, Justin Sulik, Asya Evcil, Ophelia Deroy This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8554490/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 10 You are reading this latest preprint version Abstract Hate speech in public spaces is usually addressed and regulated as a primarily verbal phenomenon, as are efforts to counter it. Less attention has been paid to how citizens, within protected speech boundaries, oppose hate not only through words but also through visible, symbolic, and multimodal acts—such as wearing inclusive symbols, displaying specific colours, or aligning physically with marginalised groups—and how these responses shape perceptions of harm, tolerance, and democratic norms. This paper examines counterspeech not only as a verbal act but as a symbolic, visual, and multimodal form of civic expression. This experimental study (N = 827) investigates how citizens’ responses to homophobic speech are perceived by third-party observers and how these perceptions influence broader inferences about harm and societal tolerance. Using visual vignettes in Germany and the UK, the research compares three response types: verbal counterspeech, nonverbal-symbolic responses (like wearing inclusive symbols), and mixed responses that combine verbal objection with symbolic cues. By situating these within different socio-normative climates—from silence to majoritarian and unanimous opposition—the study conceptualises hate speech as a social phenomenon whose meaning emerges through interaction rather than isolated utterances. Findings reveal a differentiated pattern with key implications for free-speech debates. Symbolic responses alone did not reliably reduce perceived harm and sometimes heightened perceptions of unresolved tension. In contrast, mixed responses proved especially effective in Germany, reducing perceived harm and increasing perceptions of societal tolerance—particularly where opposition to hate speech was not yet widespread. These effects were weaker in the UK, highlighting how cultural legibility shapes the interpretation of symbolic expression. In both contexts, social consensus alone was insufficient; communicative clarity through multimodal expression was decisive. Theoretically, this study reframes counterspeech as a multimodal, embodied practice that operates within, rather than against, liberal speech democracies. It shows how citizens interpret and enact normative boundaries in daily interaction. Empirically, the findings suggest that visible, symbolic opposition—when combined with verbal engagement—can reinforce democratic resilience and social cohesion by clarifying shared values. The paper contributes to broader discussions on how institutional safeguards and citizen-led responses jointly shape inclusive and just societies. Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Introduction At a university cafeteria, a student mocks same-sex relationships. One listener wears a rainbow-coloured badge on their backpack, while another quietly says, “That’s not okay.” Others look at their phones or exchange uneasy glances. The room is full of reactions, but which of them actually challenge the comment? Does the badge signal resistance as clearly as words, or do people interpret it differently? And how are responses understood when verbal and nonverbal elements are combined? Situations like this illustrate that responses to hate speech are often multimodal, and that verbal, nonverbal, and mixed forms of action may carry distinct social meanings. Understanding how ordinary observers interpret and evaluate this range of response types is therefore crucial. To address these questions, this study compares verbal counterspeech, nonverbal-symbolic responses (such as wearing symbols of inclusivity), and mixed responses that combine both modalities, examining how each shapes perceptions of harm, opposition, and societal tolerance across cultural contexts. The study adopts a dignity-based approach to define hate speech (Waldron, 2012; Lepoutre, 2021; Bousquet, 2022). This approach accounts not only for hateful speech that degrades, stigmatizes, or threatens (Maitra & McGowan, 2012; Benesch, 2014; ECRI, 2016; Matsuda, 2018), but also for exclusionary discourses that present hateful narratives and expressions as socially accepted facts, implicitly or explicitly excluding individuals or groups based on their group belonging (Udupa, 2023; Becker et al., 2024; INACH, 2024). It acknowledges that hate speech is increasingly performed and amplified through multimodal cues that blend words, gestures, symbols, and collective presence (Bogerts et al., 2019; Yoon, 2016; Saul, 2021; Drainville & Saul, 2024; Aguilera-Carnerero et al., 2025) and that its force often derives not only from its propositional content but also from contextual resonance, social signals that mark belonging or exclusion, and shared background assumptions (Fricker, 2007; McGowan, 2009; Langton, 2018). Understanding resistance to such a speech requires examining not only what people say, but also what they do—or fail to do—visibly and collectively to counter it. Since harm is enacted through multiple modalities, resistance that engages only a single channel risks leaving critical social and perceptual dimensions unaddressed. Importantly, if hate speech derives its force from a shared social ecology, then the moral and civic challenge lies not only in the utterance itself but in how it is met, or not met, by others (Fumagalli, 2021; Wachs et al., 2024). Silent or indifferent reactions function as tacit assent, reinforcing the legitimacy of hate speech and enabling its normalisation in everyday interactions (Brauer & Chekroun, 2005; Langton, 2018; Fumagalli, 2021; Van Kleef, 2024). Research shows that leaving hate speech unopposed reshapes conversational norms by subtly redrawing what counts as acceptable public expression (Álvarez Benjumea & Winter, 2020; Zapata et al., 2024). Over time, exposure to unopposed hate speech weakens empathy and perspective-taking, suppressing neural activity associated with social understanding and eroding the collective capacity for moral engagement (Pluta et al., 2024). This normative pressure to respond does not mean that responding is straightforward (McGowan, 2018). It emerges in situations that are often socially constraining (Lepoutre, 2017). In public settings, whether offline or online, bystanders face power imbalances, uncertainty about others' support, and the risk of retaliation, social backlash, or escalation (Howard, 2021; Saul, 2021). For these reasons, counterspeech cannot be understood solely in terms of individual moral willingness or personal courage. It plays a crucial normative role by making opposition visible within the shared social space (Ayala-López, 2018; Langton, 2018; Popa-Wyatt & Wyatt, 2018). Empirical studies of real-world hate speech incidents demonstrate that observers distinguish sharply between passive bystanders and those who visibly disalign with perpetrators, showing that the social meaning of an incident is shaped by whether opposition, or its absence, is made perceptually evident (Wachs et al., 2024). Moreover, researchers indicate that social opposition’s effectiveness depends less on the content of responses and more on their collective nature. Resistance thus functions not merely as an individual moral act, but as a social coordination problem, gaining normative weight only when it becomes visible and shared (Blanchard et al, 1994; Álvarez Benjumea, 2023; Zapata et al., 2024). All these tensions between moral responsibility and practical constraints invite a broader understanding of resistance—one that situates responses to hate and exclusionary speech within perceptual, social, and cognitive frameworks, acknowledging that communicative meaning arises not only through semantic processing but also through perceptual and social simulation (Awad & Wagoner, 2020; Hardy, 2021; Schlenker, 2022; Gurung et al., 2025; Madella & Warthon, 2023). In search of alternative formats for counterspeech that complement, enhance, or substitute for purely verbal responses, our study explores a growing repertoire of democratic action: nonverbal-symbolic responses to hate and exclusion. We define such responses as communicative acts that employ visual and performative elements to convey messages of resistance or inclusivity without words. They often involve the use of symbols, colours, coordinated gestures, or shared insignia to challenge hate speech and demonstrate social solidarity. Research on social movements and protest shows these symbolic responses condense complex identities, histories, and moral stances into visible, salient cues that evoke solidarity and shape collective behaviour (Awad & Wagoner, 2020). Complementarily, visual and symbolic cues often evoke stronger emotional resonance, greater trust, and wider diffusion than textual or spoken rebuttals (Bünzli & Eppler, 2025), making them especially effective in public settings where verbal confrontation is impractical or easily ignored. Despite the potential of nonverbal-symbolic responses, empirical research on their efficacy and perception remains surprisingly limited. Most existing studies focus on verbal counterspeech (Cepollaro et al., 2023) and neglect the role of nonverbal strategies in countering hate speech. Our study aims to fill this gap. Nonverbal-symbolic responses and those that combine verbal and nonverbal elements are accessible and scalable, allowing individual expression and collective coordination of resistance in everyday spaces and teaching inclusivity to bystanders (and perpetrators) through visibility rather than persuasion (Esquith, 2011). They can function as social cues for norm activation by making inclusive norms perceptually salient (Castelli, Zogmaister & Tomelleri, 2012; Paluck & Green, 2009). Crucially, they avoid the binary between censorship and inaction, offering a democratic avenue for citizens to visibly express disalignment with exclusionary speech, signal solidarity with targets, and endorse values of tolerance while preserving open discourse. Symbolic opposition leaves space for dissent while clarifying the boundaries of acceptable expression. It is both democratic and preventive, a quiet affirmation that some forms of speech, though legally permitted, remain socially unacceptable. Philosophically, nonverbal-symbolic responses can be situated at the intersection of artistic and pre-emptive counterspeech. Symbols such as a rainbow badge may function simultaneously as artistic counterspeech, by aesthetically reframing and publicly contesting hateful meanings (Dixon, 2019; 2022), and as pre-emptive counterspeech, by establishing moral expectations of inclusion and tolerance before or alongside the occurrence of hate speech (Lepoutre, 2021). Together, these perspectives clarify how nonverbal interventions can render resistance perceptually salient and shape how observers interpret the social context in which hate speech circulates. Building on this framework, our study translates these philosophical insights into empirically testable hypotheses about the effects of nonverbal-symbolic responses on social perception and norm communication across cultural contexts. Cross-Cultural Legibility Importantly, our study acknowledges that the normative force of nonverbal-symbolic responses depends on cultural legibility, a property that is historically shaped and not guaranteed. A symbol becomes publicly recognisable as a marker of inclusion or resistance only through long-term diffusion, collective uptake, and cultural work; until that stabilises, misinterpretation and backlash remain possible. Research in formal semantics also defends the idea that communities assign specific “legibilities” to symbols or gestures that might be interpreted differently, or not at all, in another cultural context (Schlenker, 2022). The communicative traction of symbolic counterspeech is therefore highly mediated by its cultural legibility, shaped by historical trajectories, civic self-understandings, and local semiotic conventions. Considering this, our study investigates third-party perceptions of nonverbal-symbolic responses across two cultural contexts: Germany and the United Kingdom. We chose these countries due to their distinctive traditions of free speech, historical responsibility, and civic participation. In Germany, the history of post-war reconciliation and stringent anti-hate speech laws highlights a societal emphasis on historical responsibility. In contrast, the United Kingdom’s strong tradition of free speech coexists with contemporary challenges in regulating hate speech, shaped by colonial legacies and multicultural discourse. These differences make the meaning of symbolic counterspeech empirically non-obvious (Olabisi, 2018; Schmid et al., 2024). The chosen societies may differ in the extent to which visible signs of solidarity or resistance are recognised as public interventions rather than personal expressions, and in how such symbols are integrated into the moral expectations that govern everyday interactions. In some contexts, collective memory or deep-rooted civic narratives may heighten the salience of visible opposition; in others, traditions of expressive pluralism may render the same symbols less stable, requiring additional cues to anchor their meaning. These possibilities highlight the need for empirical and comparative investigation. Capturing the Social Ecology of Hate Speech To examine how nonverbal-symbolic responses shape the perception of hate and exclusionary speech, we used visual vignettes depicting group interactions involving homophobic speech in public settings. Unlike written descriptions, visual formats make nonverbal aspects of interaction directly observable, allowing participants to see how silence, collective presence, symbols, and words structure the encounter (Holm et al., 2018; Khanolainen & Semenova, 2020; Sahlström, 2025). This method captures features of everyday interactions that written vignettes struggle to convey, especially the immediate individual and social cues through which people assess the seriousness of an incident and the social norms it displays (Zapata et al., 2024). Building on this approach, our study considers how verbal, nonverbal-symbolic, and mixed responses operate within the broader context in which hate speech occurs and whether they influence evaluations of the harm created and the perceived tolerance of the society depicted in the scenes. Study description and tested hypotheses In this study, we examined how different types of citizens’ responses and the broader social norms guiding those responses shaped participants’ perceptions of homophobic speech incidents in Germany and the UK. Specifically, we focused on two central evaluative dimensions. First, participants judged the perceived level of harm created by the interaction, indicating how distressing they expected the incident to be for both the targeted individuals and the bystanders present (higher values reflecting greater perceived harm). Second, they evaluated the perceived societal tolerance in the environment where the incident occurred, with lower ratings indicating a more hostile or exclusionary climate. To examine these questions, we employed a mixed factorial design in which three types of citizens’ responses were manipulated within-subjects (verbal, nonverbal-symbolic, and mixed responses combining both elements), while two contextual factors were varied between-subjects. The first contextual factor concerned the socio-normative climate (Majoritarian Silence, Majoritarian Response, or Unanimous Response), and the second concerned the country in which participants were sampled (Germany or the UK). This design allowed each participant to directly compare different types of responses within a single normative environment and enabled us to examine cross-country variation in perceptions. Across vignettes, participants viewed a homophobic incident involving three bystanders whose reactions varied according to the assigned condition. This design allowed us to assess not only whether some forms of citizens' intervention were perceived as more effective or socially meaningful than others, but also whether their impact depended on the broader behavioural expectations of the group (i.e., prevailing social norm). In particular, we were interested in whether nonverbal-symbolic responses could serve as an adequate substitute for verbal intervention or, when added to verbal statements, operate as an enhancement, depending on the normative climate. Scenarios depicting the nine experimental combinations of Response Type and Social Norm (delivered separately by country) are summarised in Fig. 1 . Guided by previous research on bystander behaviour, symbolic communication, and normative influence, we formulated several predictions regarding harm perception. Firstly, we expected that the type of citizen response would shape perceived harm regardless of the prevailing social norm (H1), such that nonverbal-symbolic responses would be seen as reducing harm more effectively than verbal ones (H1a: NV < V), and mixed responses more effectively than purely verbal ones (H1b: M < V). Secondly, we hypothesised that the socio-normative context itself would modulate perceived harm (H2), with incidents presented in a context of Majoritarian Silence expected to elicit the highest harm assessments relative to those presented in contexts of Majoritarian Response and Unanimous Response. Beyond these main effects, we anticipated an interaction between response type and social norm (H3): under a Majoritarian Silence norm, mixed responses were expected to show an enhancement effect relative to purely verbal responses (H3a); under a Majoritarian Response, nonverbal-symbolic responses were expected to resemble verbal ones (a substitution effect; H3b); and under a Unanimous Response, we expected nonverbal-symbolic responses to reduce harm more than purely verbal ones (H3c). Finally, we expected these patterns to vary across national contexts (H4). For perceived societal tolerance, we formulated parallel expectations. We firstly predicted that response type would influence perceived societal tolerance across norms (H5), with nonverbal-symbolic responses generating higher tolerance ratings than verbal ones (H5a: NV > V), and mixed responses generating higher ratings than purely verbal responses (H5b: M > V). Secondly, we expected that norms promoting response, that is, Majoritarian Response and Unanimous Response, would yield higher perceived societal tolerance than Majoritarian Silence (H6). In addition, we anticipated an interaction between response type and social norms (H7), with mixed responses generating an enhancement effect under Majoritarian Silence (H7a), nonverbal-symbolic responses showing a substitution pattern under Majoritarian Response (H7b), and nonverbal-symbolic responses eliciting higher perceptions of societal tolerance than verbal ones under Unanimous Response (H7c). As with harm, we expected these patterns to differ between Germany and the UK (H8). Two sets of planned exploratory comparisons complemented our main hypotheses. First, we examined whether mixed responses differed systematically from purely nonverbal-symbolic ones across norms and countries (H9), focusing on whether the addition of verbal content strengthened or weakened perceived effectiveness. Second, we explored simple effects of socio-normative context within each response type (H10), enabling a more fine-grained examination of how the same behaviour is interpreted differently depending on the prevailing norm. Methods Participants We conducted a power analysis using R, running multiple simulations based on our pilot data. For each simulation, we generated datasets with various sample sizes and performed pairwise comparisons of Cumulative Link Mixed-Effects Models (CLMM; see the Analysis Strategy section below). After conducting 1,000 simulations per sample size, we calculated the proportion of simulations with statistically significant key effects (p < 0.05) to determine the sample size required to achieve 80% power. The results indicated that 363 participants were required to achieve such power. Considering the possibility that some participants would fail the attention checks, we increased the sample size by approximately 10%. As a result, we recruited 450 British and 450 German participants for the main study. Testing Materials We created nine colourful cartoons featuring a perpetrator shouting a homophobic remark at a gay couple (e.g., “You perverts make us sick!”, “You gays destroy our families!”, “You queers get out of here!”). The incident took place in front of three other citizens who either voiced opposition verbally (e.g., “Enough! Stop saying that!”, “Hey, not in my name!”), signalled opposition nonverbally-symbolically (e.g., carrying a heart-shaped, rainbow-coloured badge), responded verbally while carrying the badge (mixed response), or remained silent. While the number of citizens who witnessed the incident (three) remained constant across the cartoons, their responses were adjusted to illustrate different social norms, as shown in Fig. 2 . In the Majoritarian Silence norm, only one of the three opposes the hate speech. In the Majoritarian Response norm, two of the three express opposition. Finally, all three oppose the homophobic speech in the Unanimous Response norm. The visual vignettes depicted perpetrators as angry and disdainful, while the victims appeared intimidated or ashamed. The other citizens had neutral faces and a direct line of sight to the attack. The scenarios were gender-balanced, featuring both female and male perpetrators, victims, and bystanders. All characters had white skin to eliminate perception changes related to their skin colour (See Supplementary Information section for the complete battery of testing materials). In addition, we created an extra vignette as an attention check. In this vignette, participants viewed a friendly dialogue between two people about a film. One character in the scene carried an unrelated symbol (Supplementary Fig. 1). For German participants, the vignettes were translated into German, ensuring that the remarks and responses were culturally and contextually equivalent while maintaining the integrity of the experimental design. Materials Validation To verify that participants interpreted the visual stimuli in line with our theoretical framework, we conducted a separate validation study focused on the badge, which served as a nonverbal symbolic response in the vignettes. We aimed to ensure the badge was perceived as a meaningful social cue (signalling values or a stance) rather than a neutral decorative element. We recruited 36 German participants and 36 British participants. Each participant was presented with one of six short scenarios that mirrored the main study’s structure: vignettes depicting nonverbal or mixed bystander responses (verbal plus badge) across three social-norm conditions (Majoritarian Silence, Majoritarian Response, and Unanimous Response). Participants were asked: “What meaning do you attribute to the badge worn by the character(s) in this scene?” They could select one of four response options: (1) “It represents his/their beliefs or values,” (2) “It is purely decorative with no specific meaning,” (3) “It signifies his/their affiliation or membership in a group,” or (4) “Other” (with an open field to specify). Frequency counts were calculated for each response category to assess whether the intended interpretation of the badge, as an expression of values or moral stance, was consistently recognised across both cultural samples and vignette conditions. Across both national samples, participants predominantly interpreted the badge as symbolising the holder’s beliefs or values. In total, 27 participants in Germany and 21 in the United Kingdom selected this option, suggesting strong alignment with the stimulus's intended meaning. The second most frequent interpretation was that the badge indicated affiliation or group membership, chosen by 8 German and 13 British participants. Only one participant in the UK perceived the badge as purely decorative, suggesting that participants rarely viewed it as a neutral or aesthetic element. Two participants provided alternative interpretations under “Other”: one German participant described the badge as “Pride,” while one UK participant specified “Represents love is love.” These findings confirm that the badge was interpreted mainly as a reliable and meaningful symbolic cue that communicated support for inclusivity and tolerance, consistent with our conceptualisation of nonverbal symbolic responses. The distribution of responses across Germany and the UK further indicates that this interpretation was stable across cultural contexts (See Fig. 3 ). Procedure and Analysis Strategy The study was conducted online using the Qualtrics platform ( www.qualtrics.com ). To start, participants were informed about the task. Upon their agreement, they provided informed consent before the experiment began. After the instruction, participants were randomly assigned to one of three social norms: Majoritarian Silence, Majoritarian Response, and Unanimous Response. Participants viewed four experimental scenarios within each, representing the three response types: Nonverbal, Verbal, and Mixed (see Fig. 1 ), and one attention check. After observing each scene, participants answered two questions on a 7-point Likert scale: Harm perception: "How distressing do you find the incident shown above? (For targeted people and bystanders).” Tolerance perception: "As a bystander, how would you rate the society where the incident occurred in terms of tolerance?" The Likert scale ranged from 1 (Not at all Distressing/Tolerant) to 7 (Extremely Distressing/Tolerant). All visual scenarios and their respective questions were presented in a randomised order. At the end of the experiment, participants answered demographic questions (age, education, gender, religion, socio-political attitude, and previous experience with discrimination). The study took approximately 5 minutes to complete. Participants received £0.90 for following the instructions correctly and completing the survey. For German participants, the same procedure was followed, but all instructions and demographic questions were adjusted to align with colloquial uses and expressions. Data Preprocessing Before the analysis, we excluded participants who provided incomplete responses, submitted duplicate entries, or failed the attention task by rating perceived harm at three or higher or tolerance at two or lower. In the German sample, 22 participants were excluded for failing the attention check, and one was excluded for submitting a duplicate response, leaving a final sample of 427 participants. Of these, 85 identified themselves as LGBTQ + and 342 as non-LGBTQ+. The gender distribution was 263 male, 155 female, seven non-binary or diverse, and two who preferred not to disclose their gender. In the British sample, 49 participants were excluded for failing the attention check, and 1 participant was excluded for submitting an incomplete response, resulting in a final sample of 400 participants. Among these, 51 identified themselves as LGBTQ + and 349 as non-LGBTQ+. The gender distribution was 224 females, 171 males, two non-binary or diverse, and three who preferred not to disclose their gender. Modeling Approach We implemented Cumulative Link Mixed-Effects Models (CLMM) using the ordinal package in R to analyse ordinal outcome variables (perceived harm and perceived societal tolerance ratings). These models accommodate the ordered nature of the dependent variables while accounting for the repeated-measures design through random effects. Each CLMM included fixed effects for three experimentally manipulated factors: Response Type: A within-subject factor with three levels (Verbal, Nonverbal-symbolic, and Mixed response). Social Norm: A between-subject factor with three levels (Majoritarian Silence, Majoritarian Response, and Unanimous Response). Country: A categorical variable with two levels (Germany and the UK). The complete factorial design incorporated all interaction terms (Country × Social Norm × Response Type). Participant-level variability was modelled using random intercepts to address within-individual correlation across repeated measures. Hypotheses for the Harm perception and Tolerance perception outcomes were examined with a two-step procedure. Two cumulative link mixed models (CLMMs) were fitted, one for Perceived Harm ratings and one for Perceived Societal Tolerance ratings. Both models included all fixed effects, including the three-way interaction, and a random intercept for participants to capture individual differences. Pairwise contrasts were then obtained with the emmeans package in R. Estimated marginal means were calculated for each combination of factor levels, with Tukey adjustments for multiple comparisons. Our analysis focused on seven sets of pairwise comparisons: 1. Response Type within each Country. 2. Response Type within each Country × Social Norm combination. 3. Social Norm within each Country. 4. Social Norm within each Country × Response Type combination. 5. Country within each Response Type. 6. Country within each Social Norm. 7. Country within each Response Type × Social Norm combination. For each contrast, the output includes the estimate, standard error (SE), z-ratio, and p-value. Although data collection took place in two stages, first with UK participants and later with German participants, we analysed all data in a single model. Country was included as an interactive factor, which improved model stability and accounted for more variance, allowing cross-country differences to be represented more clearly. Additional contrasts from sets (1) and (3) were used to examine the overall effect of Social Norm (without conditioning on Response Type) and Response Type (without conditioning on Social Norm). Sets (5), (6), and (7) were included to assess country differences in perceived harm and tolerance across scenarios. The seven sets of contrasts were applied to both the Harm and Tolerance models. Exploratory Analysis Reaction time differences between response types In addition to these planned analyses, we examined reaction-time differences to clarify the cognitive demands of interpreting each response type. Reaction time was defined as the time elapsed from stimulus onset (when the vignette appeared) to the participant’s first click. Because reaction time is a continuous measure of processing duration, we analysed it using a mixed-effects linear regression model. The fixed effects included Response Type, Country, and their interaction, while a random intercept by participants accounted for individual differences in baseline speed. After model estimation, we conducted pairwise contrasts within each country to compare reaction times across response types. For each contrast, we report the estimate, standard error (SE), t-value (t), and p-value (p). Reaction time is a widely used metric in cognitive and social psychology to assess processing speed, cognitive load, or relative task difficulty (Posner, 2005; Just & Carpenter, 1992). In tasks comparing visual and verbal stimuli, prior research shows that simple symbolic or pictorial cues are often processed more rapidly than words, especially in recognition or visual search tasks (Thorpe, Fize, & Marlot, 1996; Potter, Wyble, Hagmann, & McCourt, 2014; Wolfe, 2018). However, verbal cues can be processed efficiently when semantic meaning is directly relevant to the task or when the stimulus contains complex information that simple symbols alone cannot convey (Moreno & Mayer, 2007; Mayer, 2009). Measuring reaction time thus provides an indirect index of cognitive demand (Sweller, 2011), allowing us to explore whether nonverbal-symbolic responses were more rapidly processed than verbal ones and whether adding a nonverbal component to a verbal response (like in mixed responses that combined nonverbal and verbal elements) affected, positively or negatively, the temporal burden on participants’ processing. Covariate Analysis Finally, we conducted covariate analyses to assess whether demographic and experiential variables influenced harm and societal tolerance ratings across countries. We analysed covariates including gender, LGBTQ+ identity, age, education, religiosity, political attitudes, prior experience with discrimination, and experience living abroad. We adopted a Bayesian framework because Bayesian monotonic models are well-suited for ordered predictors. Bayesian approaches are also advantageous for handling ordinal data, as they offer flexibility for modelling complex hierarchical structures and incorporating prior information. This method enhances interpretability, especially when dealing with non-linear relationships, and provides a coherent way to quantify uncertainty, allowing for more robust conclusions across varying covariate effects. All covariate analyses were conducted within a Bayesian cumulative (ordinal) regression framework using the brms package in R, which differs from the frequentist framework used in the primary analysis. Accordingly, uncertainty for these models is reported using 95% credible intervals (CrI) rather than confidence intervals (CI). Participant-level random intercepts were included to account for individual variability, and weakly informative priors were used for all parameters. Four Markov chains with 2,000 iterations each (1,000 warm-up) were run, and convergence was verified (R̂ ≤ 1.01). All covariate models were estimated separately for harm and tolerance ratings. Results are reported as posterior means (β) with 95% credible intervals (CrI). For gender and LGBTQ+ identity, we fitted Bayesian cumulative models that included both predictors and their interactions with country. Average marginal contrasts were computed to estimate group differences (e.g., Female–Male and LGBTQ+–non-LGBTQ+) within each country. Experience living abroad was analysed as a separate nominal covariate with country and its interaction, following the same modelling and comparison procedure. For ordinal and continuous covariates such as age, education, religiosity, socio-political attitude, and discrimination experience, monotonic effect modelling was applied to represent ordered relationships with the outcomes. Each covariate was analysed in a separate model, including its interaction with the country. Results Perceived harm associated with the incident H1: Main Effect of Response Type on Harm Perception H1a, predicting that nonverbal-symbolic responses would reduce perceived harm more effectively than verbal ones, was not supported. In Germany, incidents accompanied by nonverbal-symbolic responses received higher harm ratings than verbal responses (estimate = 0.900, SE = 0.142, z = 6.340, p < 0.001), with a similar pattern in the UK (estimate = 0.944, SE = 0.145, z = 6.531, p < 0.001) (Fig. 4a). Mixed responses, however, differed by country. In Germany, they were associated with lower harm ratings than verbal responses (estimate = − 0.625, SE = 0.137, z = − 4.551, p < 0.001). In contrast, in the UK, the difference was nonsignificant (estimate = 0.311, SE = 0.141, z = 2.205, p = 0.070) (Fig. 4a). Therefore, H1b was supported in Germany but not in the UK. H2: Main Effect of Social Norm on Harm Perception No significant differences in perceived harm emerged across social-norm conditions in either country (all p > 0.1), indicating that the socio-normative context alone did not modulate harm ratings. H2 was not supported (Supplementary Fig. 2a). H3: Interaction Between Response Type and Social Norm on Harm Perception In Germany, under the Majoritarian Silence norm, mixed responses, combining nonverbal-symbolic and verbal elements, yielded lower harm perceptions than purely verbal responses (estimate = − 0.896, SE = 0.247, z = − 3.631, p < 0.001). In the UK, no significant difference was observed between mixed and verbal responses (estimate = 0.044, SE = 0.246, z = 0.179, p = 0.982). Accordingly, the enhancement effect predicted by H3a, was observed in Germany but not in the UK (Fig. 4b). Under the Majoritarian Response norm, nonverbal-symbolic responses were associated with higher perceived harm than verbal responses in both Germany (estimate = 0.771, SE = 0.245, z = 3.148, p = 0.005) and the UK (estimate = 1.016, SE = 0.245, z = 4.151, p < 0.001), showing no evidence for the substitution effect predicted by H3b (Fig. 4c). Lastly, under the Unanimous Response norm, nonverbal-symbolic responses again produced higher harm ratings than verbal ones (Germany: estimate = 1.320, SE = 0.237, z = 5.563, p < 0.001; UK: estimate = 0.918, SE = 0.253, z = 3.634, p < 0.001), contrary to predictions. H3c was, therefore, not supported (Fig. 4d). H4: Cross-Cultural Moderation Across socio-normative conditions, German participants rated homophobic incidents as more harmful than UK participants, particularly for verbal and nonverbal responses, while mixed responses showed similar ratings. Specifically: Verbal responses: Germany > UK (estimate = 0.943, SE = 0.255, z = 3.704, p UK (estimate = 0.900, SE = 0.258, z = 3.484, p UK (estimate = 1.039, SE = 0.399, z = 2.606, p = 0.009). Majoritarian Silence: Germany ≈ UK (estimate = 0.733, SE = 0.397, z = 1.847, p = 0.065). Unanimous Response: No difference (estimate = 0.080, SE = 0.389, z = 0.204, p = 0.838) (Supplementary Fig. 2a). Within norms: Verbal responses: Germany > UK under Majoritarian Silence (estimate = 1.142, SE = 0.444, z = 2.575, p = 0.010) and Majoritarian Response (estimate = 1.415, SE = 0.443, z = 3.190, p = 0.001), no difference under Unanimous Response (estimate = 0.274, SE = 0.434, z = 0.632, p = 0.528 (Supplementary Fig. 2b). Nonverbal-symbolic responses were higher in Germany under Majoritarian Response (estimate = 1.170, SE = 0.449, z = 2.604, p = 0.009) and marginally higher under Majoritarian Silence (estimate = 0.854, SE = 0.449, z = 1.902, p = 0.057), with no difference under Unanimous Response (estimate = 0.676, SE = 0.414, z = 1.532, p = 0.126) (Supplementary Fig. 2c). Lastly, for Mixed responses, no cross-country differences were found under any norm (Majoritarian Silence: estimate = 0.202, SE = 0.444, z = 0.455, p = 0.649; Majoritarian Response: estimate = 0.533, SE = 0.443, z = 1.203, p = 0.229; Unanimous Response: estimate = − 0.712, SE = 0.433, z = − 1.645, p = 0.099) (Supplementary Fig. 2d). These findings suggest a robust cross-cultural difference in harm perception for verbal and nonverbal-symbolic responses, with mixed responses showing consistent effects across countries, providing descriptive support for H4. Societal Tolerance Perception H5: Main Effect of Response Type on the societal tolerance perception H5a, predicting that nonverbal-symbolic responses would increase perceived societal tolerance more than verbal ones, was not supported in either country: In Germany, nonverbal-symbolic < verbal ones (estimate = − 0.411, SE = 0.136, z = − 3.029, p = 0.007). In the UK, nonverbal-symbolic < verbal (estimate = − 0.877, SE = 0.135, z = − 6.477, p Verbal (estimate = 1.133, SE = 0.134, z = 8.482, p < 0.001); but not in the UK: Mixed ≈ Verbal (estimate = 0.204, SE = 0.132, z = 1.552, p = 0.267). H6: Main Effect of Social Norm on Societal Tolerance Perception In Germany, perceived tolerance increased with stronger social consensus: Majoritarian Silence < Majoritarian Response (estimate = − 0.950, SE = 0.287, z = − 3.315, p = 0.003), Majoritarian Silence < Unanimous Response (estimate = − 2.486, SE = 0.285, z = − 8.711, p < 0.001) and Majoritarian Response < Unanimous Response (estimate = − 1.537, SE = 0.279, z = − 5.508, p 0.20). Therefore, H6 was supported in Germany but not in the UK (Fig. 6a). Figure 6: Perceived Tolerance Across Social Norms by Response Type and Country. (a) Mean perceived tolerance ratings (bars ± 95% CI) across the three social-norm conditions (Majoritarian Silence, Majoritarian Response, Unanimous Response) for all response types combined in Germany and the UK. (b) Mean perceived tolerance ratings for verbal responses. (c ) Mean perceived tolerance ratings for nonverbal responses. (d) Mean perceived tolerance ratings for mixed responses. H7: Interaction Between Response Type and Social Norm in Societal Tolerance Perception Within Majoritarian Silence: In Germany: Mixed > Verbal (estimate = 1.405, SE = 0.242, z = 5.810, p < 0.001), showing a strong enhancement effect. In the UK: Mixed ≈ Verbal (estimate = 0.169, SE = 0.221, z = 0.764, p = 0.725). Accordingly, H7a was supported in Germany but not in the UK (Fig. 5 b). Within Majoritarian Response: In Germany: Nonverbal-symbolic ≈ Verbal (estimate = − 0.491, SE = 0.233, z = − 2.104, p = 0.089) and in the UK: Nonverbal-symbolic < Verbal (estimate = − 0.677, SE = 0.234, z = − 2.888, p = 0.011), providing support to H7b in Germany but not in the UK (Fig. 5 c). Within Unanimous Response: Purely Nonverbal-symbolic responses were associated with lower tolerance perception than verbal responses in both countries, rejecting H7c (Germany: estimate = − 0.652, SE = 0.222, z = − 2.934, p = 0.009; UK: estimate = − 1.034, SE = 0.239, z = − 4.333, p < 0.001) (Fig. 5 d). H8: Cross-Cultural Moderation in Perceived Societal Tolerance When averaged across norms, German participants rated societies as more tolerant than UK participants for mixed responses (estimate = 0.410, SE = 0.195, z = 2.106, p = 0.035). In contrast, verbal responses were rated as less tolerant (estimate = − 0.518, SE = 0.197, z = − 2.634, p = 0.008). Nonverbal responses showed no cross-country difference (estimate = − 0.053, SE = 0.199, z = − 0.267, p = 0.780) (Fig. 5 a). Averaged across response types: Within Majoritarian Silence: Germany < UK (estimate = − 1.007, SE = 0.286, z = − 3.517, p UK (estimate = 1.053, SE = 0.281, z = 3.748, p < 0.001) and within Majoritarian Response: No difference (estimate = − 0.207, SE = 0.286, z = − 0.724, p = 0.469) (Fig. 6a). By response type within norms: Verbal responses: Germany < UK under Majoritarian Silence (estimate = − 1.695, SE = 0.347, z = − 4.891, p UK under Unanimous Response (estimate = 0.670, SE = 0.335, z = 2.000, p = 0.046) (Fig. 6b). Nonverbal-symbolic responses: Germany UK under Unanimous Response (estimate = 1.053, SE = 0.340, z = 3.092, p = 0.002) (Fig. 6c). Mixed responses: Germany > UK under Unanimous Response (estimate = 1.436, SE = 0.336, z = 4.277, p < 0.001), no difference under Majoritarian Silence (estimate = − 0.459, SE = 0.338, z = − 1.358, p = 0.174) or Majoritarian Response (estimate = 0.253, SE = 0.338, z = 0.748, p = 0.454) (Fig. 6d). Overall, German participants rated societies as more tolerant under consensus norms, whereas UK participants’ ratings were stable, providing descriptive support for H8. H9: Mixed vs. Nonverbal Responses Mixed responses consistently outperformed nonverbal-symbolic responses alone in perceived effectiveness. In Germany: Mixed < Nonverbal-symbolic for perceived harm (estimate = − 1.525, SE = 0.144, z = − 10.563, p Nonverbal for perceived societal tolerance (estimate = 1.544, SE = 0.136, z = 11.334, p < 0.001) (Figs. 4a, 5 a). In the UK: Mixed < Nonverbal for perceived harm (estimate = − 0.633, SE = 0.144, z = − 4.400, p Nonverbal for perceived tolerance (estimate = 1.081, SE = 0.136, z = 7.946, p < 0.001). Our results indicate that adding a nonverbal-symbolic component to verbal responses enhanced their perceived effectiveness, supporting H9. H10: Simple Effects of Social Norms Within Response Types This analysis examined how social-norm context affected perceived harm and tolerance across response types. Within Verbal responses: Germany: Perceived Harm is higher under Majoritarian Silence than Unanimous Response (estimate = 1.083, SE = 0.432, z = 2.504, p = 0.033) (Supplementary Fig. 2b). Perceived Tolerance is lower under Majoritarian Silence than Unanimous Response (estimate = − 2.813, SE = 0.346, z = − 8.132, p < 0.001) and lower under Majoritarian Response than Unanimous Response (estimate =-1.598, SE = 0.335, z = − 4.776, p 0.20). Within Nonverbal-symbolic responses: Germany: Tolerance lower under Majoritarian Silence than Unanimous Response (estimate = − 2.253, SE = 0.345, z = − 6.532, p < 0.001) and lower under Majoritarian Response than Unanimous Response (estimate = − 1.437, SE = 0.337, z = − 4.260, p 0.10) UK: No differences for tolerance or harm (all p > 0.20). Within Mixed responses: Germany: Tolerance substantially lower under Majoritarian Silence (estimate = − 2.393, SE = 0.337, z = − 7.106, p < 0.001) and Majoritarian Response (estimate = − 1.574, SE = 0.329, z = − 4.779, p < 0.001) compared to Unanimous Response; harm showed a nonsignificant trend toward reduction (estimate = 0.777, SE = 0.430, z = 1.804, p = 0.168). UK: Neither perceived tolerance (estimate = − 0.497, SE = 0.341, z = − 1.461, p = 0.310) nor perceived harm varied across norms. Overall, greater normative consensus was associated with higher perceived tolerance in Germany, whereas UK participants’ ratings remained largely stable, supporting H10 in Germany but not in the UK. Exploratory Results: Reaction Time Differences Between Response Types In Germany: Mixed slower than nonverbal-symbolic (estimate = 2.060, SE = 0.773, t = 2.665, p = 0.021), Mixed vs. verbal: nonsignificant (estimate = − 1.354, SE = 0.773, t = − 1.752, p = 0.186) and, Nonverbal faster than verbal (estimate = − 3.414, SE = 0.773, t = − 4.417, p < 0.001). In the UK: Mixed slower than nonverbal (estimate = 3.699, SE = 0.799, t = 4.631, p < 0.001), Mixed vs. verbal: nonsignificant (estimate = 0.083, SE = 0.799, t = 0.104, p = 0.994) and, Nonverbal faster than verbal (estimate = − 3.616, SE = 0.799, t = − 4.527, p < 0.001). These results suggest verbal responses required more processing time, while nonverbal-symbolic elements were processed more fluently (Fig. 7 ). Covariate Results Nominal Predictors Figure 8. Estimated effects of gender and LGBTQ+ identity on perceived harm and tolerance . Posterior mean estimates (β) with 95% credible intervals for gender (male vs. female) and LGBTQ+ identity (yes vs. no) in Germany and the UK, derived from Bayesian cumulative regression models for harm and tolerance ratings. Specifically, within Gender: Male participants rated harm lower than female participants (Germany: β = − 1.110, 95% CrI [–1.688, − 0.537]; UK: β = − 1.599, 95% CrI [–2.194, − 1.019]). Regarding tolerance perception: Germany uncertain (β = 0.084, 95% CrI [–0.362, 0.522]); UK Male participants gave higher tolerance ratings (β = 0.598, 95% CrI [0.139, 1.028]) (Fig. 8). LGBTQ+ identity: No apparent differences in harm perception (Germany: β = 0.505, 95% CrI [–0.209, 1.177]; UK: β = 0.302, 95% CrI [–0.549, 1.162]) or tolerance (Germany: β = − 0.389, 95% CrI [–0.969, 0.157]; UK: β = − 0.211, 95% CrI [–0.894, 0.438]). Experience living abroad: No meaningful effects on harm or tolerance (all β ≈ 0) (Fig. 8). Ordinal and Continuous Covariates Religion: No association with harm (Germany: β = − 0.267, 95% CrI [–0.791, 0.234]; UK: β = − 0.185, 95% CrI [–0.895, 0.460]); for tolerance, UK positive (β = 0.716, 95% CrI [0.135, 1.287]), Germany uncertain (β = − 0.148, 95% CrI [–0.550, 0.288]). Education: No meaningful effects on harm or tolerance (all β ≈ 0). Socio-political attitude: Germany positively associated with harm (β = 0.617, 95% CrI [0.383, 0.928]); UK: negative/uncertain for perceived harm (β = − 0.169, 95% CrI [–0.427, 0.085]); no effect on tolerance. Age: No effect on harm (all β ≈ 0); tolerance slightly higher in older UK participants (β = 0.304, 95% CrI [0.021, 0.679]). Experience with discrimination: No clear effect on harm; tolerance lower in UK participants with more exposure compared to Germany (β = − 0.487, 95% CrI [–0.963, − 0.071]) (See Supplementary Figs. 3 and 4). Discussion Our findings lend empirical support to the view that responses to hate speech operate beyond mere words, and that visible, symbolic actions constitute a distinct and consequential dimension of civic resistance. By examining how third-party observers interpret verbal, nonverbal-symbolic, and mixed responses to homophobic speech across Germany and the UK, this study shows that citizens’ responses shape perceptions of harm and societal tolerance in differentiated ways. Rather than identifying a single form of counterspeech as uniformly superior, the results point to a more complex pattern in which response modalities fulfil distinct social functions, depending on their combination, the visibility of social norms, and the cultural legibility of the cues involved. Regarding perceived harm, the results do not support a straightforward substitution account. Across both Germany and the UK, purely nonverbal-symbolic responses were associated with higher perceived harm than verbal responses (H1a; H3). This pattern suggests that, for ordinary observers, symbolic opposition alone may not provide sufficient interpretive clarity to mitigate the perceived severity of a hate incident. In the absence of explicit verbal rejection, visible symbols may even heighten the salience of the incident by drawing attention to unresolved tension or ambiguity. By contrast, multimodal responses—combining verbal and symbolic cues—were associated with lower harm in Germany and under conditions of majoritarian silence (H1b; H3), indicating that symbolic elements can strengthen verbal opposition when they operate in concert rather than isolation. Significantly, social norms alone did not reliably reduce harm perceptions (H2). Even when most citizens in the scene visibly opposed the hateful remark, harm ratings remained largely stable unless response modality was taken into account. This suggests that harm evaluations are relatively resistant to socio-normative modulation and depend more on the clarity and explicitness of opposition than on mere consensus. Cross-cultural differences further reinforce this point: German participants consistently rated incidents as more harmful than UK participants in verbal and nonverbal conditions (H4), indicating baseline differences in sensitivity to homophobic speech that cannot be explained solely by response patterns. A different dynamic emerged for perceived societal tolerance. Here, social norms played a central role, particularly in Germany, where tolerance ratings increased in a complex and patterned manner as opposition became more widely shared (H6; H10). For citizens observing these scenes, a visible consensus appeared to recalibrate expectations about what is socially acceptable, even when the harmful act itself remained salient. Mixed responses, combining symbolic and verbal elements, were especially effective under conditions of majoritarian silence (H7a), suggesting that symbolic cues can compensate for the absence of verbal unanimity by making normative opposition perceptually salient. In contrast, tolerance judgements in the UK were comparatively stable across response types and norm conditions, suggesting weaker norm updating in response to bystander behaviour. Notably, the German results point to a decisive advantage of mixed responses over purely verbal ones, indicating that the combination of spoken objection with visible symbols of inclusion carries distinctive normative force in this context. This finding lends empirical support to civic and institutional initiatives in Germany that encourage the public display of inclusive symbols—such as flags, colours, or badges—not as substitutes for verbal opposition, but as amplifiers that make dissent socially legible and collectively anchored. These findings highlight a significant dissociation between harm and tolerance. While harm judgements reflect the perceived severity of the incident itself, tolerance judgements capture broader inferences about the society in which the incident unfolds. Mixed responses appear particularly well suited to shaping the latter, not by directly repairing harm, but by signalling collective values and expectations. This interpretation aligns with research on metanorms and sanctioning, which shows that individuals’ willingness to intervene depends on their expectations about how others will evaluate such intervention (Fehr & Fischbacher, 2004; Willer, Kuwabara, & Macy, 2009; Horne et al., 2009; Horne, 2014). By making opposition visible, mixed responses reduce uncertainty about social approval and clarify that intervening against hate is likely to be supported rather than sanctioned. From a theoretical standpoint, this supports accounts in communication science and cognitive theory that treat symbolic and bodily cues not as peripheral to meaning, but as constitutive elements through which social understanding and normative expectations are established before, and alongside, explicit verbal interpretation. Reaction-time results further support this interpretation. Nonverbal-symbolic responses were processed more rapidly than verbal responses across both countries, whereas multimodal responses did not impose additional processing demands beyond verbal responses. Symbolic cues function as perceptually efficient signals that can be integrated quickly into social judgements. Longer reaction times for verbal responses likely reflect deeper semantic engagement rather than inefficiency. Together, these findings suggest that mixed responses combine perceptual immediacy with interpretive clarity, offering a form of civic intervention that is both cognitively accessible and normatively informative. Overall, our results suggest that responses to hate speech operate within a shared social ecology in which meaning is inferred not only from what is said, but from what is made visible. Multimodal responses do not replace verbal opposition, but augment it by clarifying social alignment, especially for citizens navigating uncertainty about others’ support. Limitations Several limitations should be considered when interpreting these findings. First, symbolic cues are inherently dependent on cultural legibility. While mixed and nonverbal-symbolic responses shaped tolerance perceptions in Germany, their effects were weaker and less systematic in the UK. This asymmetry suggests that symbols do not carry stable meanings across contexts and that their normative force depends on historically sedimented conventions and shared interpretive frameworks. For citizens witnessing hate incidents, a symbol may signal collective resistance in one context and personal expression in another. Future research should therefore vary the types of symbols and their degrees of conventionalisation to assess how legibility conditions norm inference. Second, the limited impact of symbolic responses on perceived harm highlights a boundary condition. Multimodal responses appear better suited to signalling collective values than to alleviating the immediate perceived severity of harm. This distinction is normatively important: visible opposition may shape future expectations and behaviour without undoing the emotional or dignitary damage already inflicted. Symbolic responses should therefore be understood as complementary to, rather than substitutes for, verbal or institutional mechanisms that directly address harm and accountability. Third, the effects observed here rely on collective visibility. The strong role of unanimous and majoritarian response conditions indicates that isolated acts of opposition may be insufficient to shift social expectations. For third-party observers, it is the perception of shared alignment, rather than individual moral expression, that stabilises norms. This raises questions about threshold effects, coordination, and durability that cannot be fully captured in vignette-based designs. Fourth, reaction-time measures offer only indirect insight into cognitive processing. Faster responses to symbolic cues indicate perceptual fluency, but do not capture longer-term persuasion, learning, or behavioural change (Sweller, 2011; Paas & Sweller, 2014). Combining experimental methods with longitudinal or qualitative approaches would help clarify how momentary norm inferences translate into sustained civic engagement. Finally, the study does not address potential backlash or strategic misuse of symbolic opposition. While visible resistance can clarify norms, it may also provoke counter-mobilisation or devolve into tokenistic signalling. Understanding when multimodal responses empower democratic participation and when they risk trivialisation remains an important task for future work. Conclusions and Implications Taken together, these findings support a view of counterspeech as a coordination problem rather than a purely individual moral act. Citizens do not respond to hate speech in a vacuum; they act under uncertainty about others’ values, likely reactions, and the social costs of intervention. Multimodal responses appear particularly valuable in this respect because they reduce uncertainty about collective alignment. By making opposition perceptually salient, they help witnessing citizens infer not only that a comment is objectionable, but that opposing it is socially supported. The dissociation between harm and tolerance outcomes is normatively significant. While multimodal responses do not reliably repair harm at the level of immediate perception, they play a crucial role in shaping the social meaning of the incident and the normative environment in which future speech will occur. This suggests that resistance to hate speech should not be evaluated solely in terms of immediate harm reduction, but also in terms of its capacity to stabilise inclusive norms and discourage normalisation over time. From a theoretical perspective, the study contributes to research on hate speech, social norms, and civic response by demonstrating that resistance operates through multiple communicative channels. It empirically supports accounts that emphasise the importance of visibility, collective uptake, and perceptual cues in norm transmission, extending them beyond verbal interaction. By focusing on third-party perceptions, the study highlights how meaning is co-constructed by those who witness hate incidents and interpret what responses, if any, signal about shared values. Practically, the findings have implications for how democratic societies think about responding to hate speech in everyday settings. Mixed responses, combining verbal and nonverbal-symbolic elements, offer citizens ways to express disalignment that do not rely solely on confrontation or institutional sanction. They provide a means of making values visible in public spaces, particularly where power dynamics, risk, or uncertainty constrain verbal engagement. In contexts such as Germany, where symbolic cues appear to be widely legible, our findings suggest that campaigns promoting visible signs of inclusion are not merely expressive gestures, but can meaningfully enhance the perceived social impact of verbal opposition when both are deployed together. At the same time, the results caution against treating symbolic opposition as a standalone solution. Its effectiveness depends on cultural legibility, collective visibility, and integration with verbal and institutional responses. More broadly, this study suggests that the civic challenge posed by hate speech is not only a matter of regulating speech or persuading speakers, but of sustaining shared expectations about what a society tolerates. Multimodal responses contribute to this task by marking the boundaries of acceptable expression in ways that are publicly observable, socially coordinated, and cognitively accessible to those who witness them. In conclusion, responding to hate speech is not only about speaking back, but about shaping the social space in which speech is heard. By examining how citizens interpret verbal, nonverbal-symbolic, and multimodal responses across cultural contexts, this study shows that visible opposition matters—not because it always lessens harm, but because it helps sustain the normative conditions under which harm is recognised, contested, and less likely to be normalised. Multimodal responses thus emerge as a meaningful form of democratic action, complementing verbal engagement and institutional measures by making resistance visible where it might otherwise remain silent. Declarations Data & code availability All anonymised data and analysis scripts supporting the findings of this study are openly available in the OSF at [https://osf.io/uyhce/overview?view_only=8bbf4526080f4406babe7b2bc9cd73b5]. The repository includes the raw and processed datasets, the R scripts used for analyses, and the code for figure generation. Acknowledgements JZ and AE were supported by the Mentoring Program from [redacted for peer review]. The funders had no role in study design, data collection and analysis, the decision to publish or the preparation of the manuscript. Author contributions JZ: Conceptualization, methodology, investigation, writing original draft, project administration, funding acquisition; JS: Methodology, formal analysis, visualisation, data curation; AE: Formal analysis, investigation, resources, visualisation, data curation; OD: Methodology, supervision and funding acquisition. Competing Interests The authors declare no competing interests as defined by Nature Research or other interests that might be perceived to influence the results and/or discussion reported in this paper. Ethical Approval The Ethics Committee of the Faculty of Philosophy, Philosophy of Science and Religious Studies at the Ludwig Maximilian University of Munich (Germany), as the competent approval body for the study design and procedures, approved the main protocol for this study on February 10, 2022 (ID-Number 131874). The ethical approval covered online recruitment and data collection involving adult participants residing in Germany and the United Kingdom. The approval remained valid at the time of data collection. All research involving human participants was conducted in accordance with relevant international and institutional ethical guidelines and regulations, including the principles of the Declaration of Helsinki. Informed Consent Informed consent was obtained from all participants prior to their participation in the study. The study was conducted entirely online, and participants were recruited via the Prolific (https://www.prolific.com/) platform, with data collected using the Qualtrics (https://www.qualtrics.com/) platform. Consent was obtained electronically by the research team before any study procedures commenced, through an online consent form presented on the Qualtrics platform. Participants were required to actively indicate their consent by clicking an “I wish to participate” button before proceeding to the study. Informed consent was obtained between July 2024 and September 2024 from all adult participants residing in the United Kingdom and Germany. Before providing consent, participants were fully informed about the purpose of the research, the non-interventional nature of the study, the procedures involved (completion of a short online questionnaire following the presentation of four visual scenarios), the expected duration of participation (approximately five minutes), and the compensation provided (£0.90 via Prolific upon successful completion). Participants were informed that some visual or verbal stimuli might contain homophobic speech and could be potentially distressing. They were informed of their right to withdraw from the study at any time without penalty and without loss of compensation. Participants were explicitly informed that their participation was voluntary, that their responses would remain anonymous, and that no personally identifiable information would be collected. They were informed that data would be analysed only in aggregated form, used solely for research purposes, and reported in scientific publications at the group level. The scope of consent covered participation in the study, the anonymous collection and analysis of data, and the use of anonymised data for academic publication. Participants were also informed about the reasons for conducting the research, how their data would be utilised, and that no risks beyond those ordinarily associated with viewing online content were anticipated. Information about appropriate support services was provided to participants, including resources specific to the United Kingdom and Germany, should they experience distress related to the study content. The study involved only adult participants and did not include minors or other vulnerable populations. No oral consent procedures were used References Aguilera-Carnerero C, Becker MJ, Scheiber M (2025) Same tools, different target: countering hate speech with memes. Álvarez-Benjumea A (2023) Uncovering hidden opinions: social norms and the expression of xenophobic attitudes. Eur Sociol Rev 39(3):449–463 Álvarez-Benjumea A, Winter F (2020) The breakdown of antiracist norms: a natural experiment on hate speech after terrorist attacks. Proc Natl Acad Sci USA 117(37):22800–22804 Ayala S, Vasilyeva N (2016) Responsibility for silence. J Soc Philos 47(3) Ayala-López S (2018) A structural explanation of injustice in conversations: it ’ s about norms. Pac Philos Q 99(4):726–748 Becker MJ, Ascone L, Bolton M, Bundzíková V, Chapelan A, Haupeltshofer P, et al (2024) Decoding antisemitism: an AI-driven study on hate speech and imagery online (Discourse Report No. 6) Benesch S (2014) Defining and diminishing hate speech. In: State of the world ’ s minorities and indigenous peoples 2014, pp 18–25 Blanchard FA, Crandall CS, Brigham JC, Vaughn LA (1994) Condemning and condoning racism: a social context approach to interracial settings. J Appl Psychol 79(6):993–997. https://doi.org/10.1037/0021-9010.79.6.993 Bogerts L, Fielitz M (2019) “ Do you want meme war?”Understanding the visual memes of the German far right. Brauer M, Chekroun P (2005) The relationship between perceived violation of social norms and social control: situational factors influencing the reaction to deviance. J Appl Soc Psychol 35(7):1519–1539 Bünzli F, Eppler MJ (2025) Calls to action and user engagement: the role of visual and verbal requests in nonprofit advocacy on Facebook. Comput Hum Behav Rep 18:100611 Castelli L, Carraro L, Pavan G, Murelli E, Carraro A (2012) The power of the unsaid: the influence of nonverbal cues on implicit attitudes. J Appl Soc Psychol 42(6):1376–1393 Cepollaro B, Lepoutre M, Simpson RM (2023) Counterspeech. Philos Compass 18(1):e12890 de Melo CM, Terada K, Santos FC (2021) Emotion expressions shape human social norms and reputations. iScience 24(3) Dixon D (2019) Alterpieces: artworks as shifting speech acts. Dissertation Dixon D (2022) Artistic (counter) speech. J Aesthet Art Crit 80(4):409–419 Drainville R, Saul J (2024) Visual and linguistic dogwhistles. ECRI (European Commission Against Racism and Intolerance) (2016) General Policy Recommendation No. 15 on Combating Hate Speech. https://www.coe.int/en/web/european-commission-against-racism-and-intolerance/recommendation-no.15 Esquith SL (2011) The political responsibilities of everyday bystanders. Penn State Press Fehr E, Fischbacher U (2004) Social norms and human cooperation. Trends Cogn Sci 8(4):185–190. https://doi.org/10.1016/j.tics.2004.02.007 Fricker M (2007) Epistemic injustice: power and the ethics of knowing Fumagalli C (2021) Counterspeech and ordinary citizens: how? When? Polit Theory 49(6):1021–1047 Gennaro G, Derksen L, Abdelrahman A, Broggini E, Green MA, Haerter VA, et al (2025) Counterspeech encouraging users to adopt the perspective of minority groups reduces hate speech and its amplification on social media. Sci Rep15(1):22018 Gurung MI, Agarwal N, Bhuiyan MMI, Poudel D (2025) Symbolic signals on Instagram: how visual media shapes engagement, emotion, trust, and diffusion. Soc Netw Anal Min 15(1):57 Hangartner D, Gennaro G, Alasiri S, Bahrich N, Bornhoft A, Boucher J et al (2021) Empathy-based counterspeech can reduce racist hate speech in a social media field experiment. Proc Natl Acad Sci USA 118(50):e2116310118 Hardy BW (2021) Embodied cognition in communication science. Commun Theory 31(4):633–653. https://doi.org/10.1093/ct/qtab021 Hess U, Fischer A (2024) Emotional mimicry. In: Emotional communication and social interaction. Springer Horne C, Chen CF, Berg J, Evermann-Druffel K (2009) Metanorm expectations: determining what to sanction. In: Altruism and prosocial behavior in groups, pp 199–223. Emerald Group Publishing Horne C (2014) The relational foundation of norm enforcement. In: The complexity of social norms, pp 105–120. Springer, Cham Holm G, Sahlström F, Zilliacus H (2018) Arts-based visual research. In: Leavy P (ed) Handbook of arts-based research, pp 311–335. Guilford Press Howard JW (2021) Terror, hate and the demands of counter-speech. Br J Polit Sci 51(3):924–939. https://doi.org/10.1017/S000712341900053X INACH (International Network Against Cyber Hate) (2024) What is cyber hate and why do we want to counter it. https://www.inach.net/cyber-hate-definitions Just MA, Carpenter PA (1992) A capacity theory of comprehension: individual differences in working memory. Psychol Rev 99(1):122–149 Khanolainen D, Semenova E (2020) School bullying through graphic vignettes: developing a new arts-based method to study a sensitive topic. Int J Qual Methods 19. https://doi.org/10.1177/1609406920922765 Langton R (2018a) Blocking as counter-speech. Philos Public Aff 46(4):363–393. https://doi.org/10.1111/papa.12138 Langton R (2018b) The authority of hate speech. Oxford Stud Philos Law 3:123–152 Langton R (2018c) Speech acts and pornography. In: Speech and harm Lepoutre M (2017) Hate speech in public discourse: A pessimistic defense of counterspeech. Soc Theory Pract:851–883 Lepoutre M (2021) Democratic speech in divided times. Oxford University Press, Oxford Madella P, Wharton T (2023) Nonverbal communication and context: multimodality in interaction. In: The Cambridge handbook of language in context. Cambridge University Press Maitra I, McGowan MK (2012) Speech and harm: controversies over free speech Matsuda MJ (2018) Words that wound: critical race theory, assaultive speech, and the First Amendment. Routledge Mayer RE (2009) Multimedia learning, 2nd edn. Cambridge University Press McGowan MK (2018) Responding to harmful speech: the more speech response, counter speech, and the complexity of language use. In: Voicing dissent, pp 182–199. Routledge Moreno R, Mayer RE (2007) Interactive multimodal learning environments. Educ Psychol Rev 19:309–326 Olabisi OO (2018) A comparative perspective on the First Amendment: free speech, militant democracy, and the primacy of dignity as a preferred constitutional value in Germany. Alabama Civ Rights Civ Liberties Law Rev 10(1):169–204 Paas F, Sweller J (2014) Implications of cognitive load theory for multimedia learning. In: Cognitive load theory. Springer Paluck EL, Green DP (2009) Deference, dissent, and dispute resolution: an experimental intervention using mass media to change norms and behavior in Rwanda. Am Polit Sci Rev 103(4):622–644. https://doi.org/10.1017/S0003055409990128 Pluta A, Zeller F, Koglin U (2024) Hate speech exposure and empathic desensitization: neural and behavioral correlates. J Exp Psychol Gen 153(2):295–310 Popa-Wyatt M, Wyatt JL (2018) Slurs, roles and power. Philos Stud 175(11):2879–2906 Posner MI (2005) Timing the brain: mental chronometry as a tool in neuroscience. Oxford University Press Potter MC, Wyble B, Hagmann CE, McCourt ES (2014) Detecting meaning in RSVP at 13 ms per picture. Atten Percept Psychophys 76(2):270–279 Sahlström F (2025) Arts-based visual research. In: Handbook of arts-based research, p 309 Saul J (2021) Someone is wrong on the internet: is there an obligation to correct false and oppressive speech on social media? In: MacKenzie A, Rose J, Bhatt I (eds) The epistemology of deceit in a postdigital era: dupery by design, pp 139–157. Springer, New York Schlenker P (2022) What it all means: semantics for (almost) everything. MIT Press Schmid U, Kümpel AS, Rieger D (2024) Social media users ’ motives for (not) engaging with hate: a survey study in Germany. Soc Media Soc 10(1) Sweller J (2011) Cognitive load theory. Psychol Learn Motiv 55:37–76 Thorpe S, Fize D, Marlot C (1996) Speed of processing in the human visual system. Nature 381(6582):520–522 Udupa S (2023) Extreme speech. In: Strippel C, Paasch-Colberg S, Emmer M, Trebbe J (eds) Challenges and perspectives of hate speech research, pp 233–248. Berlin. https://doi.org/10.48541/dcr.v12.14 Van Kleef GA (2024) The social life of emotions. Cambridge University Press Wachs S, Wettstein A, Bilz L, Espelage DL, Wright MF, Gámez-Guadix M (2024) Individual and contextual correlates of latent bystander profiles toward racist hate speech: a multilevel person-centred approach. J Youth Adolesc 53(6):1271–1286. https://doi.org/10.1007/s10964-024-01968-x Waldron J (2012) The harm in hate speech. Harvard University Press Willer R, Kuwabara K, Macy MW (2009) The false enforcement of unpopular norms. Am J Sociol 115(2):451–490 Wolfe JM (2018) Visual search: how do we find what we are looking for? Annu Rev Vis Sci 4:539–562 Zapata J, Sulik J, von Wulffen C et al (2024) Bystanders ’ collective responses set the norm against hate speech. Humanit Soc Sci Commun 11:335. https://doi.org/10.1057/s41599-024-02761-8 Additional Declarations No competing interests reported. Supplementary Files SupplementaryFigure1.png SupplementaryFigure2.png SupplementaryFigure3.png SupplementaryFigure4.png Cite Share Download PDF Status: Under Review Version 1 posted Reviews received at journal 11 May, 2026 Reviews received at journal 07 May, 2026 Reviewers agreed at journal 20 Apr, 2026 Reviewers agreed at journal 20 Apr, 2026 Reviewers agreed at journal 10 Apr, 2026 Reviewers invited by journal 08 Apr, 2026 Editor assigned by journal 01 Apr, 2026 Editor invited by journal 30 Mar, 2026 Submission checks completed at journal 25 Jan, 2026 First submitted to journal 25 Jan, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8554490","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":603184792,"identity":"1337ddd4-6471-4607-ad39-84e6b3e5cd6b","order_by":0,"name":"Jimena Zapata","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAuElEQVRIiWNgGAWjYBACxgYog58dRLIRr8WAQbKZWC1QYMBgcJhYLcwN7A8/F9T8kTc+zJ0mwVBmQ5TDkqVnHDMw3HaYd5sEw7k0orQckOZhM2AEa2FsO0yMFsbm3zz/DOw3N4O1/CdGCzObNG+bQeIGZrCWA0RoaWZjs57ZZ5w84zDvZouEc8mEtRi2tz++XfBNzra/vXfjjQ9ldkRoAcYgM5yXQFgDA4M8A7KWUTAKRsEoGAXYAACUxDIUuxtfHwAAAABJRU5ErkJggg==","orcid":"","institution":"Ludwig-Maximilians-Universität München","correspondingAuthor":true,"prefix":"","firstName":"Jimena","middleName":"","lastName":"Zapata","suffix":""},{"id":603184793,"identity":"a48ae33d-0f73-4694-9475-88924aa35899","order_by":1,"name":"Justin Sulik","email":"","orcid":"","institution":"Ludwig-Maximilians-Universität München","correspondingAuthor":false,"prefix":"","firstName":"Justin","middleName":"","lastName":"Sulik","suffix":""},{"id":603184794,"identity":"441d5b39-f5ea-40a4-8b11-b4b9e4f7cdd5","order_by":2,"name":"Asya Evcil","email":"","orcid":"","institution":"Ludwig-Maximilians-Universität München","correspondingAuthor":false,"prefix":"","firstName":"Asya","middleName":"","lastName":"Evcil","suffix":""},{"id":603184795,"identity":"f34436a1-c755-4b42-91f1-9b0e584b5100","order_by":3,"name":"Ophelia Deroy","email":"","orcid":"","institution":"Ludwig-Maximilians-Universität München","correspondingAuthor":false,"prefix":"","firstName":"Ophelia","middleName":"","lastName":"Deroy","suffix":""}],"badges":[],"createdAt":"2026-01-08 18:53:11","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8554490/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8554490/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":104545462,"identity":"e813d55c-c913-4543-8d36-bfb6daa0ee2d","added_by":"auto","created_at":"2026-03-13 07:13:30","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":212660,"visible":true,"origin":"","legend":"\u003cp\u003eExperimental conditions. Overview of the three Social Norm conditions and Response types.\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8554490/v1/7813505c92d919a7be9fcae1.png"},{"id":104780886,"identity":"ec41e6fb-ad43-4da3-b651-82228a6425f4","added_by":"auto","created_at":"2026-03-17 07:54:12","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":595705,"visible":true,"origin":"","legend":"\u003cp\u003eSample materials illustrating mixed-response conditions: Examples depict bystanders exhibiting a combination of verbal and nonverbal symbolic opposition to homophobic speech across the three social-norm types: Majoritarian Silence, Majoritarian Response, and Unanimous Response.\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8554490/v1/c314bb1f3b786889ca5789ec.png"},{"id":104545465,"identity":"800827ed-c714-4621-83be-ea1caef44d83","added_by":"auto","created_at":"2026-03-13 07:13:30","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":124465,"visible":true,"origin":"","legend":"\u003cp\u003eBadge interpretation by country. Pie charts show how respondents in Germany (n = 36) and the UK (n = 36) interpreted the badge.\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8554490/v1/f914699bfb978136794d2c3d.png"},{"id":104545470,"identity":"4d37d72e-57ec-47e8-8031-cc6c98033269","added_by":"auto","created_at":"2026-03-13 07:13:30","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":237610,"visible":true,"origin":"","legend":"\u003cp\u003ePerceived harm ratings across response types and social norm conditions by country. (a) Mean ratings of perceived harm (bars ± 95 % CI) for different response types (Nonverbal, Verbal, Mixed) across Germany and the UK, combining all social norms. (b) Mean ratings for the Majoritarian Silence norm. (c ) Mean ratings for the Majoritarian Response norm. (d) Mean ratings for the Unanimous Response norm. Asterisks indicate significance levels: * p \u0026lt; 0.05, ** p \u0026lt; 0.01, *** p \u0026lt; 0.001.\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8554490/v1/ab57b5c8002c768564b43255.png"},{"id":104781457,"identity":"dda3259a-75df-4438-8f8e-371de7e65d1c","added_by":"auto","created_at":"2026-03-17 07:55:42","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":238402,"visible":true,"origin":"","legend":"\u003cp\u003ePerceived tolerance across response types by social norm and country. (a) Mean ratings of perceived tolerance (bars ± 95 % CI) for different response types (Nonverbal, Verbal, Mixed) across Germany and the UK, combining all social norms. (b) Mean ratings for the Majoritarian Silence norm. (c ) Mean ratings for the Majoritarian Response norm. (d) Mean ratings for the Unanimous Response norm. Asterisks indicate significance levels: * p \u0026lt; 0.05, ** p \u0026lt; 0.01, *** p \u0026lt; 0.001.\u003c/p\u003e","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-8554490/v1/ab4db6bffa77ce880d2d2cc8.png"},{"id":104545467,"identity":"d5361598-3754-4f15-bd8f-4504a183f029","added_by":"auto","created_at":"2026-03-13 07:13:30","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":167970,"visible":true,"origin":"","legend":"\u003cp\u003ePerceived Tolerance Across Social Norms by Response Type and Country. (a) Mean perceived tolerance ratings (bars ± 95 % CI) across the three social-norm conditions (Majoritarian Silence, Majoritarian Response, Unanimous Response) for all response types combined in Germany and the UK. (b) Mean perceived tolerance ratings for verbal responses. \u0026nbsp;(c ) Mean perceived tolerance ratings for nonverbal responses. (d) Mean perceived tolerance ratings for mixed responses.\u003c/p\u003e","description":"","filename":"floatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-8554490/v1/dd01acd0add5eabbf9b36184.png"},{"id":104545468,"identity":"df9dbf74-4a36-42b5-84d5-84bcba3a2a3c","added_by":"auto","created_at":"2026-03-13 07:13:30","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":263970,"visible":true,"origin":"","legend":"\u003cp\u003eReaction time distribution across response types and countries. Distribution of reaction times (in seconds) for Nonverbal, Verbal, and Mixed response types in Germany and the UK. Shaded density curves represent the distribution of reaction times for each response type, and dashed vertical lines indicate the corresponding mean reaction times.\u003c/p\u003e","description":"","filename":"floatimage7.png","url":"https://assets-eu.researchsquare.com/files/rs-8554490/v1/3fd41c393edeeeee31b9721e.png"},{"id":104545472,"identity":"4964effe-84b6-4eb9-b40f-ea127d076b65","added_by":"auto","created_at":"2026-03-13 07:13:30","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":115668,"visible":true,"origin":"","legend":"\u003cp\u003eEstimated effects of gender and LGBTQ+ identity on perceived harm and tolerance. Posterior mean estimates (β) with 95% credible intervals for gender (male vs. female) and LGBTQ+ identity (yes vs. no) in Germany and the UK, derived from Bayesian cumulative regression models for harm and tolerance ratings.\u003c/p\u003e","description":"","filename":"floatimage8.png","url":"https://assets-eu.researchsquare.com/files/rs-8554490/v1/26755aa5b9d4f347455d60b0.png"},{"id":104784633,"identity":"28b3e177-812c-416c-85d2-6f051da2e012","added_by":"auto","created_at":"2026-03-17 08:08:23","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2531983,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8554490/v1/b39c375e-9ec7-40b3-8c60-bbe042c9d946.pdf"},{"id":104781748,"identity":"25c94a89-b439-4c35-839f-038ae2a775e6","added_by":"auto","created_at":"2026-03-17 07:56:16","extension":"png","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":379319,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryFigure1.png","url":"https://assets-eu.researchsquare.com/files/rs-8554490/v1/9651c60f64c87e769b0f40a1.png"},{"id":104781661,"identity":"fe63e3ba-7084-403f-9cbc-07dcb0ce9798","added_by":"auto","created_at":"2026-03-17 07:56:06","extension":"png","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":167675,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryFigure2.png","url":"https://assets-eu.researchsquare.com/files/rs-8554490/v1/884747ea651b5a686922d9ae.png"},{"id":104780835,"identity":"ba512cf7-8687-4e85-aa89-701ed65cf790","added_by":"auto","created_at":"2026-03-17 07:54:05","extension":"png","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":783411,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryFigure3.png","url":"https://assets-eu.researchsquare.com/files/rs-8554490/v1/874f4e66468da0051bf9cfa5.png"},{"id":104545473,"identity":"d677c9df-2752-4a63-8b8b-bbee4fbdcb2e","added_by":"auto","created_at":"2026-03-13 07:13:31","extension":"png","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":785147,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryFigure4.png","url":"https://assets-eu.researchsquare.com/files/rs-8554490/v1/c14a03dec301848709a7f0bc.png"}],"financialInterests":"No competing interests reported.","formattedTitle":"Making Resistance Visible: The Role of Nonverbal-Symbolic Opposition to Hate Speech Across Cultures","fulltext":[{"header":"Introduction","content":"\u003cp\u003eAt a university cafeteria, a student mocks same-sex relationships. One listener wears a rainbow-coloured badge on their backpack, while another quietly says, “That’s not okay.” Others look at their phones or exchange uneasy glances. The room is full of reactions, but which of them actually challenge the comment? Does the badge signal resistance as clearly as words, or do people interpret it differently? And how are responses understood when verbal and nonverbal elements are combined? Situations like this illustrate that responses to hate speech are often multimodal, and that verbal, nonverbal, and mixed forms of action may carry distinct social meanings. Understanding how ordinary observers interpret and evaluate this range of response types is therefore crucial. To address these questions, this study compares verbal counterspeech, nonverbal-symbolic responses (such as wearing symbols of inclusivity), and mixed responses that combine both modalities, examining how each shapes perceptions of harm, opposition, and societal tolerance across cultural contexts.\u003c/p\u003e \u003cp\u003eThe study adopts a dignity-based approach to define hate speech (Waldron, 2012; Lepoutre, 2021; Bousquet, 2022). This approach accounts not only for hateful speech that degrades, stigmatizes, or threatens (Maitra \u0026amp; McGowan, 2012; Benesch, 2014; ECRI, 2016; Matsuda, 2018), but also for exclusionary discourses that present hateful narratives and expressions as socially accepted facts, implicitly or explicitly excluding individuals or groups based on their group belonging (Udupa, 2023; Becker et al., 2024; INACH, 2024). It acknowledges that hate speech is increasingly performed and amplified through multimodal cues that blend words, gestures, symbols, and collective presence (Bogerts et al., 2019; Yoon, 2016; Saul, 2021; Drainville \u0026amp; Saul, 2024; Aguilera-Carnerero et al., 2025) and that its force often derives not only from its propositional content but also from contextual resonance, social signals that mark belonging or exclusion, and shared background assumptions (Fricker, 2007; McGowan, 2009; Langton, 2018). Understanding resistance to such a speech requires examining not only what people say, but also what they do—or fail to do—visibly and collectively to counter it. Since harm is enacted through multiple modalities, resistance that engages only a single channel risks leaving critical social and perceptual dimensions unaddressed.\u003c/p\u003e \u003cp\u003eImportantly, if hate speech derives its force from a shared social ecology, then the moral and civic challenge lies not only in the utterance itself but in how it is met, or not met, by others (Fumagalli, 2021; Wachs et al., 2024). Silent or indifferent reactions function as tacit assent, reinforcing the legitimacy of hate speech and enabling its normalisation in everyday interactions (Brauer \u0026amp; Chekroun, 2005; Langton, 2018; Fumagalli, 2021; Van Kleef, 2024). Research shows that leaving hate speech unopposed reshapes conversational norms by subtly redrawing what counts as acceptable public expression (Álvarez Benjumea \u0026amp; Winter, 2020; Zapata et al., 2024). Over time, exposure to unopposed hate speech weakens empathy and perspective-taking, suppressing neural activity associated with social understanding and eroding the collective capacity for moral engagement (Pluta et al., 2024).\u003c/p\u003e \u003cp\u003eThis normative pressure to respond does not mean that responding is straightforward (McGowan, 2018). It emerges in situations that are often socially constraining (Lepoutre, 2017). In public settings, whether offline or online, bystanders face power imbalances, uncertainty about others' support, and the risk of retaliation, social backlash, or escalation (Howard, 2021; Saul, 2021). For these reasons, counterspeech cannot be understood solely in terms of individual moral willingness or personal courage. It plays a crucial normative role by making opposition visible within the shared social space (Ayala-López, 2018; Langton, 2018; Popa-Wyatt \u0026amp; Wyatt, 2018). Empirical studies of real-world hate speech incidents demonstrate that observers distinguish sharply between passive bystanders and those who visibly disalign with perpetrators, showing that the social meaning of an incident is shaped by whether opposition, or its absence, is made perceptually evident (Wachs et al., 2024). Moreover, researchers indicate that social opposition’s effectiveness depends less on the content of responses and more on their collective nature. Resistance thus functions not merely as an individual moral act, but as a social coordination problem, gaining normative weight only when it becomes visible and shared (Blanchard et al, 1994; Álvarez Benjumea, 2023; Zapata et al., 2024).\u003c/p\u003e \u003cp\u003eAll these tensions between moral responsibility and practical constraints invite a broader understanding of resistance—one that situates responses to hate and exclusionary speech within perceptual, social, and cognitive frameworks, acknowledging that communicative meaning arises not only through semantic processing but also through perceptual and social simulation (Awad \u0026amp; Wagoner, 2020; Hardy, 2021; Schlenker, 2022; Gurung et al., 2025; Madella \u0026amp; Warthon, 2023).\u003c/p\u003e \u003cp\u003e In search of alternative formats for counterspeech that complement, enhance, or substitute for purely verbal responses, our study explores a growing repertoire of democratic action: nonverbal-symbolic responses to hate and exclusion. We define such responses as communicative acts that employ visual and performative elements to convey messages of resistance or inclusivity without words. They often involve the use of symbols, colours, coordinated gestures, or shared insignia to challenge hate speech and demonstrate social solidarity. Research on social movements and protest shows these symbolic responses condense complex identities, histories, and moral stances into visible, salient cues that evoke solidarity and shape collective behaviour (Awad \u0026amp; Wagoner, 2020). Complementarily, visual and symbolic cues often evoke stronger emotional resonance, greater trust, and wider diffusion than textual or spoken rebuttals (Bünzli \u0026amp; Eppler, 2025), making them especially effective in public settings where verbal confrontation is impractical or easily ignored.\u003c/p\u003e \u003cp\u003eDespite the potential of nonverbal-symbolic responses, empirical research on their efficacy and perception remains surprisingly limited. Most existing studies focus on verbal counterspeech (Cepollaro et al., 2023) and neglect the role of nonverbal strategies in countering hate speech. Our study aims to fill this gap. Nonverbal-symbolic responses and those that combine verbal and nonverbal elements are accessible and scalable, allowing individual expression and collective coordination of resistance in everyday spaces and teaching inclusivity to bystanders (and perpetrators) through visibility rather than persuasion (Esquith, 2011). They can function as social cues for norm activation by making inclusive norms perceptually salient (Castelli, Zogmaister \u0026amp; Tomelleri, 2012; Paluck \u0026amp; Green, 2009). Crucially, they avoid the binary between censorship and inaction, offering a democratic avenue for citizens to visibly express disalignment with exclusionary speech, signal solidarity with targets, and endorse values of tolerance while preserving open discourse. Symbolic opposition leaves space for dissent while clarifying the boundaries of acceptable expression. It is both democratic and preventive, a quiet affirmation that some forms of speech, though legally permitted, remain socially unacceptable.\u003c/p\u003e \u003cp\u003ePhilosophically, nonverbal-symbolic responses can be situated at the intersection of artistic and pre-emptive counterspeech. Symbols such as a rainbow badge may function simultaneously as artistic counterspeech, by aesthetically reframing and publicly contesting hateful meanings (Dixon, 2019; 2022), and as pre-emptive counterspeech, by establishing moral expectations of inclusion and tolerance before or alongside the occurrence of hate speech (Lepoutre, 2021). Together, these perspectives clarify how nonverbal interventions can render resistance perceptually salient and shape how observers interpret the social context in which hate speech circulates. Building on this framework, our study translates these philosophical insights into empirically testable hypotheses about the effects of nonverbal-symbolic responses on social perception and norm communication across cultural contexts.\u003c/p\u003e \u003cp\u003eCross-Cultural Legibility\u003c/p\u003e \u003cp\u003eImportantly, our study acknowledges that the normative force of nonverbal-symbolic responses depends on cultural legibility, a property that is historically shaped and not guaranteed. A symbol becomes publicly recognisable as a marker of inclusion or resistance only through long-term diffusion, collective uptake, and cultural work; until that stabilises, misinterpretation and backlash remain possible. Research in formal semantics also defends the idea that communities assign specific “legibilities” to symbols or gestures that might be interpreted differently, or not at all, in another cultural context (Schlenker, 2022).\u003c/p\u003e \u003cp\u003eThe communicative traction of symbolic counterspeech is therefore highly mediated by its cultural legibility, shaped by historical trajectories, civic self-understandings, and local semiotic conventions. Considering this, our study investigates third-party perceptions of nonverbal-symbolic responses across two cultural contexts: Germany and the United Kingdom.\u003c/p\u003e \u003cp\u003eWe chose these countries due to their distinctive traditions of free speech, historical responsibility, and civic participation. In Germany, the history of post-war reconciliation and stringent anti-hate speech laws highlights a societal emphasis on historical responsibility. In contrast, the United Kingdom’s strong tradition of free speech coexists with contemporary challenges in regulating hate speech, shaped by colonial legacies and multicultural discourse. These differences make the meaning of symbolic counterspeech empirically non-obvious (Olabisi, 2018; Schmid et al., 2024). The chosen societies may differ in the extent to which visible signs of solidarity or resistance are recognised as public interventions rather than personal expressions, and in how such symbols are integrated into the moral expectations that govern everyday interactions. In some contexts, collective memory or deep-rooted civic narratives may heighten the salience of visible opposition; in others, traditions of expressive pluralism may render the same symbols less stable, requiring additional cues to anchor their meaning. These possibilities highlight the need for empirical and comparative investigation.\u003c/p\u003e \u003cp\u003eCapturing the Social Ecology of Hate Speech\u003c/p\u003e \u003cp\u003eTo examine how nonverbal-symbolic responses shape the perception of hate and exclusionary speech, we used visual vignettes depicting group interactions involving homophobic speech in public settings. Unlike written descriptions, visual formats make nonverbal aspects of interaction directly observable, allowing participants to see how silence, collective presence, symbols, and words structure the encounter (Holm et al., 2018; Khanolainen \u0026amp; Semenova, 2020; Sahlström, 2025). This method captures features of everyday interactions that written vignettes struggle to convey, especially the immediate individual and social cues through which people assess the seriousness of an incident and the social norms it displays (Zapata et al., 2024). Building on this approach, our study considers how verbal, nonverbal-symbolic, and mixed responses operate within the broader context in which hate speech occurs and whether they influence evaluations of the harm created and the perceived tolerance of the society depicted in the scenes.\u003c/p\u003e \u003cp\u003eStudy description and tested hypotheses\u003c/p\u003e \u003cp\u003e In this study, we examined how different types of citizens’ responses and the broader social norms guiding those responses shaped participants’ perceptions of homophobic speech incidents in Germany and the UK. Specifically, we focused on two central evaluative dimensions. First, participants judged the perceived level of harm created by the interaction, indicating how distressing they expected the incident to be for both the targeted individuals and the bystanders present (higher values reflecting greater perceived harm). Second, they evaluated the perceived societal tolerance in the environment where the incident occurred, with lower ratings indicating a more hostile or exclusionary climate.\u003c/p\u003e \u003cp\u003e To examine these questions, we employed a mixed factorial design in which three types of citizens’ responses were manipulated within-subjects (verbal, nonverbal-symbolic, and mixed responses combining both elements), while two contextual factors were varied between-subjects. The first contextual factor concerned the socio-normative climate (Majoritarian Silence, Majoritarian Response, or Unanimous Response), and the second concerned the country in which participants were sampled (Germany or the UK). This design allowed each participant to directly compare different types of responses within a single normative environment and enabled us to examine cross-country variation in perceptions.\u003c/p\u003e \u003cp\u003eAcross vignettes, participants viewed a homophobic incident involving three bystanders whose reactions varied according to the assigned condition. This design allowed us to assess not only whether some forms of citizens' intervention were perceived as more effective or socially meaningful than others, but also whether their impact depended on the broader behavioural expectations of the group (i.e., prevailing social norm). In particular, we were interested in whether nonverbal-symbolic responses could serve as an adequate substitute for verbal intervention or, when added to verbal statements, operate as an enhancement, depending on the normative climate. Scenarios depicting the nine experimental combinations of Response Type and Social Norm (delivered separately by country) are summarised in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eGuided by previous research on bystander behaviour, symbolic communication, and normative influence, we formulated several predictions regarding harm perception. Firstly, we expected that the type of citizen response would shape perceived harm regardless of the prevailing social norm (H1), such that nonverbal-symbolic responses would be seen as reducing harm more effectively than verbal ones (H1a: NV \u0026lt; V), and mixed responses more effectively than purely verbal ones (H1b: M \u0026lt; V). Secondly, we hypothesised that the socio-normative context itself would modulate perceived harm (H2), with incidents presented in a context of Majoritarian Silence expected to elicit the highest harm assessments relative to those presented in contexts of Majoritarian Response and Unanimous Response. Beyond these main effects, we anticipated an interaction between response type and social norm (H3): under a Majoritarian Silence norm, mixed responses were expected to show an enhancement effect relative to purely verbal responses (H3a); under a Majoritarian Response, nonverbal-symbolic responses were expected to resemble verbal ones (a substitution effect; H3b); and under a Unanimous Response, we expected nonverbal-symbolic responses to reduce harm more than purely verbal ones (H3c). Finally, we expected these patterns to vary across national contexts (H4).\u003c/p\u003e \u003cp\u003eFor perceived societal tolerance, we formulated parallel expectations. We firstly predicted that response type would influence perceived societal tolerance across norms (H5), with nonverbal-symbolic responses generating higher tolerance ratings than verbal ones (H5a: NV \u0026gt; V), and mixed responses generating higher ratings than purely verbal responses (H5b: M \u0026gt; V). Secondly, we expected that norms promoting response, that is, Majoritarian Response and Unanimous Response, would yield higher perceived societal tolerance than Majoritarian Silence (H6). In addition, we anticipated an interaction between response type and social norms (H7), with mixed responses generating an enhancement effect under Majoritarian Silence (H7a), nonverbal-symbolic responses showing a substitution pattern under Majoritarian Response (H7b), and nonverbal-symbolic responses eliciting higher perceptions of societal tolerance than verbal ones under Unanimous Response (H7c). As with harm, we expected these patterns to differ between Germany and the UK (H8).\u003c/p\u003e \u003cp\u003eTwo sets of planned exploratory comparisons complemented our main hypotheses. First, we examined whether mixed responses differed systematically from purely nonverbal-symbolic ones across norms and countries (H9), focusing on whether the addition of verbal content strengthened or weakened perceived effectiveness. Second, we explored simple effects of socio-normative context within each response type (H10), enabling a more fine-grained examination of how the same behaviour is interpreted differently depending on the prevailing norm.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003eParticipants\u003c/p\u003e\u003cp\u003eWe conducted a power analysis using R, running multiple simulations based on our pilot data. For each simulation, we generated datasets with various sample sizes and performed pairwise comparisons of Cumulative Link Mixed-Effects Models (CLMM; see the Analysis Strategy section below). After conducting 1,000 simulations per sample size, we calculated the proportion of simulations with statistically significant key effects (p \u0026lt; 0.05) to determine the sample size required to achieve 80% power. The results indicated that 363 participants were required to achieve such power. Considering the possibility that some participants would fail the attention checks, we increased the sample size by approximately 10%. As a result, we recruited 450 British and 450 German participants for the main study.\u003c/p\u003e\u003cp\u003eTesting Materials\u003c/p\u003e\u003cp\u003eWe created nine colourful cartoons featuring a perpetrator shouting a homophobic remark at a gay couple (e.g., “You perverts make us sick!”, “You gays destroy our families!”, “You queers get out of here!”). The incident took place in front of three other citizens who either voiced opposition verbally (e.g., “Enough! Stop saying that!”, “Hey, not in my name!”), signalled opposition nonverbally-symbolically (e.g., carrying a heart-shaped, rainbow-coloured badge), responded verbally while carrying the badge (mixed response), or remained silent.\u003c/p\u003e\u003cp\u003eWhile the number of citizens who witnessed the incident (three) remained constant across the cartoons, their responses were adjusted to illustrate different social norms, as shown in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e. In the Majoritarian Silence norm, only one of the three opposes the hate speech. In the Majoritarian Response norm, two of the three express opposition. Finally, all three oppose the homophobic speech in the Unanimous Response norm. The visual vignettes depicted perpetrators as angry and disdainful, while the victims appeared intimidated or ashamed. The other citizens had neutral faces and a direct line of sight to the attack. The scenarios were gender-balanced, featuring both female and male perpetrators, victims, and bystanders. All characters had white skin to eliminate perception changes related to their skin colour (See Supplementary Information section for the complete battery of testing materials).\u003c/p\u003e\u003cp\u003eIn addition, we created an extra vignette as an attention check. In this vignette, participants viewed a friendly dialogue between two people about a film. One character in the scene carried an unrelated symbol (Supplementary Fig.\u0026nbsp;1).\u003c/p\u003e\u003cp\u003eFor German participants, the vignettes were translated into German, ensuring that the remarks and responses were culturally and contextually equivalent while maintaining the integrity of the experimental design.\u003c/p\u003e\u003cp\u003eMaterials Validation\u003c/p\u003e\u003cp\u003e To verify that participants interpreted the visual stimuli in line with our theoretical framework, we conducted a separate validation study focused on the badge, which served as a nonverbal symbolic response in the vignettes. We aimed to ensure the badge was perceived as a meaningful social cue (signalling values or a stance) rather than a neutral decorative element. We recruited 36 German participants and 36 British participants. Each participant was presented with one of six short scenarios that mirrored the main study’s structure: vignettes depicting nonverbal or mixed bystander responses (verbal plus badge) across three social-norm conditions (Majoritarian Silence, Majoritarian Response, and Unanimous Response).\u003c/p\u003e\u003cp\u003eParticipants were asked: “What meaning do you attribute to the badge worn by the character(s) in this scene?” They could select one of four response options: (1) “It represents his/their beliefs or values,” (2) “It is purely decorative with no specific meaning,” (3) “It signifies his/their affiliation or membership in a group,” or (4) “Other” (with an open field to specify). Frequency counts were calculated for each response category to assess whether the intended interpretation of the badge, as an expression of values or moral stance, was consistently recognised across both cultural samples and vignette conditions.\u003c/p\u003e\u003cp\u003eAcross both national samples, participants predominantly interpreted the badge as symbolising the holder’s beliefs or values. In total, 27 participants in Germany and 21 in the United Kingdom selected this option, suggesting strong alignment with the stimulus's intended meaning. The second most frequent interpretation was that the badge indicated affiliation or group membership, chosen by 8 German and 13 British participants. Only one participant in the UK perceived the badge as purely decorative, suggesting that participants rarely viewed it as a neutral or aesthetic element. Two participants provided alternative interpretations under “Other”: one German participant described the badge as “Pride,” while one UK participant specified “Represents love is love.”\u003c/p\u003e\u003cp\u003eThese findings confirm that the badge was interpreted mainly as a reliable and meaningful symbolic cue that communicated support for inclusivity and tolerance, consistent with our conceptualisation of nonverbal symbolic responses. The distribution of responses across Germany and the UK further indicates that this interpretation was stable across cultural contexts (See Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eProcedure and Analysis Strategy\u003c/p\u003e\u003cp\u003eThe study was conducted online using the Qualtrics platform (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ewww.qualtrics.com\u003c/span\u003e\u003cspan class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). To start, participants were informed about the task. Upon their agreement, they provided informed consent before the experiment began. After the instruction, participants were randomly assigned to one of three social norms: Majoritarian Silence, Majoritarian Response, and Unanimous Response. Participants viewed four experimental scenarios within each, representing the three response types: Nonverbal, Verbal, and Mixed (see Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e), and one attention check.\u003c/p\u003e\u003cp\u003eAfter observing each scene, participants answered two questions on a 7-point Likert scale:\u003c/p\u003e\u003cul\u003e \u003cli\u003e \u003cp\u003eHarm perception: \"How distressing do you find the incident shown above? (For targeted people and bystanders).”\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eTolerance perception: \"As a bystander, how would you rate the society where the incident occurred in terms of tolerance?\"\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e\u003cp\u003eThe Likert scale ranged from 1 (Not at all Distressing/Tolerant) to 7 (Extremely Distressing/Tolerant). All visual scenarios and their respective questions were presented in a randomised order. At the end of the experiment, participants answered demographic questions (age, education, gender, religion, socio-political attitude, and previous experience with discrimination). The study took approximately 5 minutes to complete. Participants received £0.90 for following the instructions correctly and completing the survey.\u003c/p\u003e\u003cp\u003eFor German participants, the same procedure was followed, but all instructions and demographic questions were adjusted to align with colloquial uses and expressions.\u003c/p\u003e\u003cp\u003eData Preprocessing\u003c/p\u003e\u003cp\u003eBefore the analysis, we excluded participants who provided incomplete responses, submitted duplicate entries, or failed the attention task by rating perceived harm at three or higher or tolerance at two or lower. In the German sample, 22 participants were excluded for failing the attention check, and one was excluded for submitting a duplicate response, leaving a final sample of 427 participants. Of these, 85 identified themselves as LGBTQ + and 342 as non-LGBTQ+. The gender distribution was 263 male, 155 female, seven non-binary or diverse, and two who preferred not to disclose their gender. In the British sample, 49 participants were excluded for failing the attention check, and 1 participant was excluded for submitting an incomplete response, resulting in a final sample of 400 participants. Among these, 51 identified themselves as LGBTQ + and 349 as non-LGBTQ+. The gender distribution was 224 females, 171 males, two non-binary or diverse, and three who preferred not to disclose their gender.\u003c/p\u003e\u003cp\u003eModeling Approach\u003c/p\u003e\u003cp\u003eWe implemented Cumulative Link Mixed-Effects Models (CLMM) using the ordinal package in R to analyse ordinal outcome variables (perceived harm and perceived societal tolerance ratings). These models accommodate the ordered nature of the dependent variables while accounting for the repeated-measures design through random effects. Each CLMM included fixed effects for three experimentally manipulated factors:\u003c/p\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003e Response Type: A within-subject factor with three levels (Verbal, Nonverbal-symbolic, and Mixed response).\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eSocial Norm: A between-subject factor with three levels (Majoritarian Silence, Majoritarian Response, and Unanimous Response).\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eCountry: A categorical variable with two levels (Germany and the UK).\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003cp\u003eThe complete factorial design incorporated all interaction terms (Country × Social Norm × Response Type). Participant-level variability was modelled using random intercepts to address within-individual correlation across repeated measures.\u003c/p\u003e\u003cp\u003eHypotheses for the Harm perception and Tolerance perception outcomes were examined with a two-step procedure. Two cumulative link mixed models (CLMMs) were fitted, one for Perceived Harm ratings and one for Perceived Societal Tolerance ratings. Both models included all fixed effects, including the three-way interaction, and a random intercept for participants to capture individual differences. Pairwise contrasts were then obtained with the emmeans package in R. Estimated marginal means were calculated for each combination of factor levels, with Tukey adjustments for multiple comparisons.\u003c/p\u003e\u003cp\u003eOur analysis focused on seven sets of pairwise comparisons:\u003c/p\u003e\u003cp\u003e1. Response Type within each Country.\u003c/p\u003e\u003cp\u003e2. Response Type within each Country × Social Norm combination.\u003c/p\u003e\u003cp\u003e3. Social Norm within each Country.\u003c/p\u003e\u003cp\u003e4. Social Norm within each Country × Response Type combination.\u003c/p\u003e\u003cp\u003e5. Country within each Response Type.\u003c/p\u003e\u003cp\u003e6. Country within each Social Norm.\u003c/p\u003e\u003cp\u003e7. Country within each Response Type × Social Norm combination.\u003c/p\u003e\u003cp\u003eFor each contrast, the output includes the estimate, standard error (SE), z-ratio, and p-value.\u003c/p\u003e\u003cp\u003eAlthough data collection took place in two stages, first with UK participants and later with German participants, we analysed all data in a single model. Country was included as an interactive factor, which improved model stability and accounted for more variance, allowing cross-country differences to be represented more clearly. Additional contrasts from sets (1) and (3) were used to examine the overall effect of Social Norm (without conditioning on Response Type) and Response Type (without conditioning on Social Norm). Sets (5), (6), and (7) were included to assess country differences in perceived harm and tolerance across scenarios. The seven sets of contrasts were applied to both the Harm and Tolerance models.\u003c/p\u003e\u003cp\u003eExploratory Analysis\u003c/p\u003e\u003cp\u003eReaction time differences between response types\u003c/p\u003e\u003cp\u003eIn addition to these planned analyses, we examined reaction-time differences to clarify the cognitive demands of interpreting each response type. Reaction time was defined as the time elapsed from stimulus onset (when the vignette appeared) to the participant’s first click. Because reaction time is a continuous measure of processing duration, we analysed it using a mixed-effects linear regression model. The fixed effects included Response Type, Country, and their interaction, while a random intercept by participants accounted for individual differences in baseline speed. After model estimation, we conducted pairwise contrasts within each country to compare reaction times across response types. For each contrast, we report the estimate, standard error (SE), t-value (t), and p-value (p).\u003c/p\u003e\u003cp\u003eReaction time is a widely used metric in cognitive and social psychology to assess processing speed, cognitive load, or relative task difficulty (Posner, 2005; Just \u0026amp; Carpenter, 1992). In tasks comparing visual and verbal stimuli, prior research shows that simple symbolic or pictorial cues are often processed more rapidly than words, especially in recognition or visual search tasks (Thorpe, Fize, \u0026amp; Marlot, 1996; Potter, Wyble, Hagmann, \u0026amp; McCourt, 2014; Wolfe, 2018). However, verbal cues can be processed efficiently when semantic meaning is directly relevant to the task or when the stimulus contains complex information that simple symbols alone cannot convey (Moreno \u0026amp; Mayer, 2007; Mayer, 2009). Measuring reaction time thus provides an indirect index of cognitive demand (Sweller, 2011), allowing us to explore whether nonverbal-symbolic responses were more rapidly processed than verbal ones and whether adding a nonverbal component to a verbal response (like in mixed responses that combined nonverbal and verbal elements) affected, positively or negatively, the temporal burden on participants’ processing.\u003c/p\u003e\u003cp\u003eCovariate Analysis\u003c/p\u003e\u003cp\u003eFinally, we conducted covariate analyses to assess whether demographic and experiential variables influenced harm and societal tolerance ratings across countries. We analysed covariates including gender, LGBTQ+ identity, age, education, religiosity, political attitudes, prior experience with discrimination, and experience living abroad.\u003c/p\u003e\u003cp\u003eWe adopted a Bayesian framework because Bayesian monotonic models are well-suited for ordered predictors. Bayesian approaches are also advantageous for handling ordinal data, as they offer flexibility for modelling complex hierarchical structures and incorporating prior information. This method enhances interpretability, especially when dealing with non-linear relationships, and provides a coherent way to quantify uncertainty, allowing for more robust conclusions across varying covariate effects.\u003c/p\u003e\u003cp\u003eAll covariate analyses were conducted within a Bayesian cumulative (ordinal) regression framework using the brms package in R, which differs from the frequentist framework used in the primary analysis. Accordingly, uncertainty for these models is reported using 95% credible intervals (CrI) rather than confidence intervals (CI). Participant-level random intercepts were included to account for individual variability, and weakly informative priors were used for all parameters. Four Markov chains with 2,000 iterations each (1,000 warm-up) were run, and convergence was verified (R̂ ≤ 1.01). All covariate models were estimated separately for harm and tolerance ratings. Results are reported as posterior means (β) with 95% credible intervals (CrI).\u003c/p\u003e\u003cp\u003eFor gender and LGBTQ+ identity, we fitted Bayesian cumulative models that included both predictors and their interactions with country. Average marginal contrasts were computed to estimate group differences (e.g., Female–Male and LGBTQ+–non-LGBTQ+) within each country. Experience living abroad was analysed as a separate nominal covariate with country and its interaction, following the same modelling and comparison procedure.\u003c/p\u003e\u003cp\u003eFor ordinal and continuous covariates such as age, education, religiosity, socio-political attitude, and discrimination experience, monotonic effect modelling was applied to represent ordered relationships with the outcomes. Each covariate was analysed in a separate model, including its interaction with the country.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003ePerceived harm associated with the incident\u003c/p\u003e \u003cp\u003eH1: Main Effect of Response Type on Harm Perception\u003c/p\u003e \u003cp\u003e H1a, predicting that nonverbal-symbolic responses would reduce perceived harm more effectively than verbal ones, was not supported. In Germany, incidents accompanied by nonverbal-symbolic responses received higher harm ratings than verbal responses (estimate\u0026thinsp;=\u0026thinsp;0.900, SE\u0026thinsp;=\u0026thinsp;0.142, z\u0026thinsp;=\u0026thinsp;6.340, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), with a similar pattern in the UK (estimate\u0026thinsp;=\u0026thinsp;0.944, SE\u0026thinsp;=\u0026thinsp;0.145, z\u0026thinsp;=\u0026thinsp;6.531, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) (Fig.\u0026nbsp;4a). Mixed responses, however, differed by country. In Germany, they were associated with lower harm ratings than verbal responses (estimate = \u0026minus;\u0026thinsp;0.625, SE\u0026thinsp;=\u0026thinsp;0.137, z = \u0026minus;\u0026thinsp;4.551, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). In contrast, in the UK, the difference was nonsignificant (estimate\u0026thinsp;=\u0026thinsp;0.311, SE\u0026thinsp;=\u0026thinsp;0.141, z\u0026thinsp;=\u0026thinsp;2.205, p\u0026thinsp;=\u0026thinsp;0.070) (Fig.\u0026nbsp;4a). Therefore, H1b was supported in Germany but not in the UK.\u003c/p\u003e \u003cp\u003eH2: Main Effect of Social Norm on Harm Perception\u003c/p\u003e \u003cp\u003eNo significant differences in perceived harm emerged across social-norm conditions in either country (all p\u0026thinsp;\u0026gt;\u0026thinsp;0.1), indicating that the socio-normative context alone did not modulate harm ratings. H2 was not supported (Supplementary Fig.\u0026nbsp;2a).\u003c/p\u003e \u003cp\u003eH3: Interaction Between Response Type and Social Norm on Harm Perception\u003c/p\u003e \u003cp\u003eIn Germany, under the Majoritarian Silence norm, mixed responses, combining nonverbal-symbolic and verbal elements, yielded lower harm perceptions than purely verbal responses (estimate = \u0026minus;\u0026thinsp;0.896, SE\u0026thinsp;=\u0026thinsp;0.247, z = \u0026minus;\u0026thinsp;3.631, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). In the UK, no significant difference was observed between mixed and verbal responses (estimate\u0026thinsp;=\u0026thinsp;0.044, SE\u0026thinsp;=\u0026thinsp;0.246, z\u0026thinsp;=\u0026thinsp;0.179, p\u0026thinsp;=\u0026thinsp;0.982). Accordingly, the enhancement effect predicted by H3a, was observed in Germany but not in the UK (Fig.\u0026nbsp;4b). Under the Majoritarian Response norm, nonverbal-symbolic responses were associated with higher perceived harm than verbal responses in both Germany (estimate\u0026thinsp;=\u0026thinsp;0.771, SE\u0026thinsp;=\u0026thinsp;0.245, z\u0026thinsp;=\u0026thinsp;3.148, p\u0026thinsp;=\u0026thinsp;0.005) and the UK (estimate\u0026thinsp;=\u0026thinsp;1.016, SE\u0026thinsp;=\u0026thinsp;0.245, z\u0026thinsp;=\u0026thinsp;4.151, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), showing no evidence for the substitution effect predicted by H3b (Fig.\u0026nbsp;4c). Lastly, under the Unanimous Response norm, nonverbal-symbolic responses again produced higher harm ratings than verbal ones (Germany: estimate\u0026thinsp;=\u0026thinsp;1.320, SE\u0026thinsp;=\u0026thinsp;0.237, z\u0026thinsp;=\u0026thinsp;5.563, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001; UK: estimate\u0026thinsp;=\u0026thinsp;0.918, SE\u0026thinsp;=\u0026thinsp;0.253, z\u0026thinsp;=\u0026thinsp;3.634, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), contrary to predictions. H3c was, therefore, not supported (Fig.\u0026nbsp;4d).\u003c/p\u003e \u003cp\u003eH4: Cross-Cultural Moderation\u003c/p\u003e \u003cp\u003eAcross socio-normative conditions, German participants rated homophobic incidents as more harmful than UK participants, particularly for verbal and nonverbal responses, while mixed responses showed similar ratings. Specifically: Verbal responses: Germany\u0026thinsp;\u0026gt;\u0026thinsp;UK (estimate\u0026thinsp;=\u0026thinsp;0.943, SE\u0026thinsp;=\u0026thinsp;0.255, z\u0026thinsp;=\u0026thinsp;3.704, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Nonverbal-symbolic responses: Germany\u0026thinsp;\u0026gt;\u0026thinsp;UK (estimate\u0026thinsp;=\u0026thinsp;0.900, SE\u0026thinsp;=\u0026thinsp;0.258, z\u0026thinsp;=\u0026thinsp;3.484, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Mixed responses: No difference (estimate\u0026thinsp;=\u0026thinsp;0.007, SE\u0026thinsp;=\u0026thinsp;0.254, z\u0026thinsp;=\u0026thinsp;0.030, p\u0026thinsp;=\u0026thinsp;0.976) (Fig.\u0026nbsp;4a).\u003c/p\u003e \u003cp\u003eAveraged across response types: Majoritarian Response: Germany\u0026thinsp;\u0026gt;\u0026thinsp;UK (estimate\u0026thinsp;=\u0026thinsp;1.039, SE\u0026thinsp;=\u0026thinsp;0.399, z\u0026thinsp;=\u0026thinsp;2.606, p\u0026thinsp;=\u0026thinsp;0.009). Majoritarian Silence: Germany\u0026thinsp;\u0026asymp;\u0026thinsp;UK (estimate\u0026thinsp;=\u0026thinsp;0.733, SE\u0026thinsp;=\u0026thinsp;0.397, z\u0026thinsp;=\u0026thinsp;1.847, p\u0026thinsp;=\u0026thinsp;0.065). Unanimous Response: No difference (estimate\u0026thinsp;=\u0026thinsp;0.080, SE\u0026thinsp;=\u0026thinsp;0.389, z\u0026thinsp;=\u0026thinsp;0.204, p\u0026thinsp;=\u0026thinsp;0.838) (Supplementary Fig.\u0026nbsp;2a).\u003c/p\u003e \u003cp\u003eWithin norms: Verbal responses: Germany\u0026thinsp;\u0026gt;\u0026thinsp;UK under Majoritarian Silence (estimate\u0026thinsp;=\u0026thinsp;1.142, SE\u0026thinsp;=\u0026thinsp;0.444, z\u0026thinsp;=\u0026thinsp;2.575, p\u0026thinsp;=\u0026thinsp;0.010) and Majoritarian Response (estimate\u0026thinsp;=\u0026thinsp;1.415, SE\u0026thinsp;=\u0026thinsp;0.443, z\u0026thinsp;=\u0026thinsp;3.190, p\u0026thinsp;=\u0026thinsp;0.001), no difference under Unanimous Response (estimate\u0026thinsp;=\u0026thinsp;0.274, SE\u0026thinsp;=\u0026thinsp;0.434, z\u0026thinsp;=\u0026thinsp;0.632, p\u0026thinsp;=\u0026thinsp;0.528 (Supplementary Fig.\u0026nbsp;2b). Nonverbal-symbolic responses were higher in Germany under Majoritarian Response (estimate\u0026thinsp;=\u0026thinsp;1.170, SE\u0026thinsp;=\u0026thinsp;0.449, z\u0026thinsp;=\u0026thinsp;2.604, p\u0026thinsp;=\u0026thinsp;0.009) and marginally higher under Majoritarian Silence (estimate\u0026thinsp;=\u0026thinsp;0.854, SE\u0026thinsp;=\u0026thinsp;0.449, z\u0026thinsp;=\u0026thinsp;1.902, p\u0026thinsp;=\u0026thinsp;0.057), with no difference under Unanimous Response (estimate\u0026thinsp;=\u0026thinsp;0.676, SE\u0026thinsp;=\u0026thinsp;0.414, z\u0026thinsp;=\u0026thinsp;1.532, p\u0026thinsp;=\u0026thinsp;0.126) (Supplementary Fig.\u0026nbsp;2c). Lastly, for Mixed responses, no cross-country differences were found under any norm (Majoritarian Silence: estimate\u0026thinsp;=\u0026thinsp;0.202, SE\u0026thinsp;=\u0026thinsp;0.444, z\u0026thinsp;=\u0026thinsp;0.455, p\u0026thinsp;=\u0026thinsp;0.649; Majoritarian Response: estimate\u0026thinsp;=\u0026thinsp;0.533, SE\u0026thinsp;=\u0026thinsp;0.443, z\u0026thinsp;=\u0026thinsp;1.203, p\u0026thinsp;=\u0026thinsp;0.229; Unanimous Response: estimate = \u0026minus;\u0026thinsp;0.712, SE\u0026thinsp;=\u0026thinsp;0.433, z = \u0026minus;\u0026thinsp;1.645, p\u0026thinsp;=\u0026thinsp;0.099) (Supplementary Fig.\u0026nbsp;2d). These findings suggest a robust cross-cultural difference in harm perception for verbal and nonverbal-symbolic responses, with mixed responses showing consistent effects across countries, providing descriptive support for H4.\u003c/p\u003e \u003cp\u003eSocietal Tolerance Perception\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eH5: Main Effect of Response Type on the societal tolerance perception\u003c/p\u003e \u003cp\u003eH5a, predicting that nonverbal-symbolic responses would increase perceived societal tolerance more than verbal ones, was not supported in either country: In Germany, nonverbal-symbolic\u0026thinsp;\u0026lt;\u0026thinsp;verbal ones (estimate = \u0026minus;\u0026thinsp;0.411, SE\u0026thinsp;=\u0026thinsp;0.136, z = \u0026minus;\u0026thinsp;3.029, p\u0026thinsp;=\u0026thinsp;0.007). In the UK, nonverbal-symbolic\u0026thinsp;\u0026lt;\u0026thinsp;verbal (estimate = \u0026minus;\u0026thinsp;0.877, SE\u0026thinsp;=\u0026thinsp;0.135, z = \u0026minus;\u0026thinsp;6.477, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e5\u003c/span\u003ea). In addition, H5b, predicting that mixed responses would increase perceived societal tolerance more than verbal ones was supported in Germany: Mixed\u0026thinsp;\u0026gt;\u0026thinsp;Verbal (estimate\u0026thinsp;=\u0026thinsp;1.133, SE\u0026thinsp;=\u0026thinsp;0.134, z\u0026thinsp;=\u0026thinsp;8.482, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001); but not in the UK: Mixed\u0026thinsp;\u0026asymp;\u0026thinsp;Verbal (estimate\u0026thinsp;=\u0026thinsp;0.204, SE\u0026thinsp;=\u0026thinsp;0.132, z\u0026thinsp;=\u0026thinsp;1.552, p\u0026thinsp;=\u0026thinsp;0.267).\u003c/p\u003e \u003cp\u003eH6: Main Effect of Social Norm on Societal Tolerance Perception\u003c/p\u003e \u003cp\u003eIn Germany, perceived tolerance increased with stronger social consensus: Majoritarian Silence\u0026thinsp;\u0026lt;\u0026thinsp;Majoritarian Response (estimate = \u0026minus;\u0026thinsp;0.950, SE\u0026thinsp;=\u0026thinsp;0.287, z = \u0026minus;\u0026thinsp;3.315, p\u0026thinsp;=\u0026thinsp;0.003), Majoritarian Silence\u0026thinsp;\u0026lt;\u0026thinsp;Unanimous Response (estimate = \u0026minus;\u0026thinsp;2.486, SE\u0026thinsp;=\u0026thinsp;0.285, z = \u0026minus;\u0026thinsp;8.711, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and Majoritarian Response\u0026thinsp;\u0026lt;\u0026thinsp;Unanimous Response (estimate = \u0026minus;\u0026thinsp;1.537, SE\u0026thinsp;=\u0026thinsp;0.279, z = \u0026minus;\u0026thinsp;5.508, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). In the UK, tolerance ratings did not differ across norms (all p\u0026thinsp;\u0026gt;\u0026thinsp;0.20). Therefore, H6 was supported in Germany but not in the UK (Fig.\u0026nbsp;6a).\u003c/p\u003e \u003cp\u003eFigure 6: \u003cem\u003ePerceived Tolerance Across Social Norms by Response Type and Country.\u003c/em\u003e (a) Mean perceived tolerance ratings (bars\u0026thinsp;\u0026plusmn;\u0026thinsp;95% CI) across the three social-norm conditions (Majoritarian Silence, Majoritarian Response, Unanimous Response) for all response types combined in Germany and the UK. (b) Mean perceived tolerance ratings for verbal responses. (c ) Mean perceived tolerance ratings for nonverbal responses. (d) Mean perceived tolerance ratings for mixed responses.\u003c/p\u003e \u003cp\u003eH7: Interaction Between Response Type and Social Norm in Societal Tolerance Perception\u003c/p\u003e \u003cp\u003eWithin Majoritarian Silence: In Germany: Mixed\u0026thinsp;\u0026gt;\u0026thinsp;Verbal (estimate\u0026thinsp;=\u0026thinsp;1.405, SE\u0026thinsp;=\u0026thinsp;0.242, z\u0026thinsp;=\u0026thinsp;5.810, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), showing a strong enhancement effect. In the UK: Mixed\u0026thinsp;\u0026asymp;\u0026thinsp;Verbal (estimate\u0026thinsp;=\u0026thinsp;0.169, SE\u0026thinsp;=\u0026thinsp;0.221, z\u0026thinsp;=\u0026thinsp;0.764, p\u0026thinsp;=\u0026thinsp;0.725). Accordingly, H7a was supported in Germany but not in the UK (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e5\u003c/span\u003eb). Within Majoritarian Response: In Germany: Nonverbal-symbolic\u0026thinsp;\u0026asymp;\u0026thinsp;Verbal (estimate = \u0026minus;\u0026thinsp;0.491, SE\u0026thinsp;=\u0026thinsp;0.233, z = \u0026minus;\u0026thinsp;2.104, p\u0026thinsp;=\u0026thinsp;0.089) and in the UK: Nonverbal-symbolic\u0026thinsp;\u0026lt;\u0026thinsp;Verbal (estimate = \u0026minus;\u0026thinsp;0.677, SE\u0026thinsp;=\u0026thinsp;0.234, z = \u0026minus;\u0026thinsp;2.888, p\u0026thinsp;=\u0026thinsp;0.011), providing support to H7b in Germany but not in the UK (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e5\u003c/span\u003ec). Within Unanimous Response: Purely Nonverbal-symbolic responses were associated with lower tolerance perception than verbal responses in both countries, rejecting H7c (Germany: estimate = \u0026minus;\u0026thinsp;0.652, SE\u0026thinsp;=\u0026thinsp;0.222, z = \u0026minus;\u0026thinsp;2.934, p\u0026thinsp;=\u0026thinsp;0.009; UK: estimate = \u0026minus;\u0026thinsp;1.034, SE\u0026thinsp;=\u0026thinsp;0.239, z = \u0026minus;\u0026thinsp;4.333, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e5\u003c/span\u003ed).\u003c/p\u003e \u003cp\u003eH8: Cross-Cultural Moderation in Perceived Societal Tolerance\u003c/p\u003e \u003cp\u003eWhen averaged across norms, German participants rated societies as more tolerant than UK participants for mixed responses (estimate\u0026thinsp;=\u0026thinsp;0.410, SE\u0026thinsp;=\u0026thinsp;0.195, z\u0026thinsp;=\u0026thinsp;2.106, p\u0026thinsp;=\u0026thinsp;0.035). In contrast, verbal responses were rated as less tolerant (estimate = \u0026minus;\u0026thinsp;0.518, SE\u0026thinsp;=\u0026thinsp;0.197, z = \u0026minus;\u0026thinsp;2.634, p\u0026thinsp;=\u0026thinsp;0.008). Nonverbal responses showed no cross-country difference (estimate = \u0026minus;\u0026thinsp;0.053, SE\u0026thinsp;=\u0026thinsp;0.199, z = \u0026minus;\u0026thinsp;0.267, p\u0026thinsp;=\u0026thinsp;0.780) (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e5\u003c/span\u003ea).\u003c/p\u003e \u003cp\u003eAveraged across response types: Within Majoritarian Silence: Germany\u0026thinsp;\u0026lt;\u0026thinsp;UK (estimate = \u0026minus;\u0026thinsp;1.007, SE\u0026thinsp;=\u0026thinsp;0.286, z = \u0026minus;\u0026thinsp;3.517, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), within Unanimous Response: Germany\u0026thinsp;\u0026gt;\u0026thinsp;UK (estimate\u0026thinsp;=\u0026thinsp;1.053, SE\u0026thinsp;=\u0026thinsp;0.281, z\u0026thinsp;=\u0026thinsp;3.748, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and within Majoritarian Response: No difference (estimate = \u0026minus;\u0026thinsp;0.207, SE\u0026thinsp;=\u0026thinsp;0.286, z = \u0026minus;\u0026thinsp;0.724, p\u0026thinsp;=\u0026thinsp;0.469) (Fig.\u0026nbsp;6a). By response type within norms: Verbal responses: Germany\u0026thinsp;\u0026lt;\u0026thinsp;UK under Majoritarian Silence (estimate = \u0026minus;\u0026thinsp;1.695, SE\u0026thinsp;=\u0026thinsp;0.347, z = \u0026minus;\u0026thinsp;4.891, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), Germany\u0026thinsp;\u0026gt;\u0026thinsp;UK under Unanimous Response (estimate\u0026thinsp;=\u0026thinsp;0.670, SE\u0026thinsp;=\u0026thinsp;0.335, z\u0026thinsp;=\u0026thinsp;2.000, p\u0026thinsp;=\u0026thinsp;0.046) (Fig.\u0026nbsp;6b). Nonverbal-symbolic responses: Germany\u0026thinsp;\u0026lt;\u0026thinsp;UK under Majoritarian Silence (estimate = \u0026minus;\u0026thinsp;0.867, SE\u0026thinsp;=\u0026thinsp;0.348, z = \u0026minus;\u0026thinsp;2.492, p\u0026thinsp;=\u0026thinsp;0.013), Germany\u0026thinsp;\u0026gt;\u0026thinsp;UK under Unanimous Response (estimate\u0026thinsp;=\u0026thinsp;1.053, SE\u0026thinsp;=\u0026thinsp;0.340, z\u0026thinsp;=\u0026thinsp;3.092, p\u0026thinsp;=\u0026thinsp;0.002) (Fig.\u0026nbsp;6c). Mixed responses: Germany\u0026thinsp;\u0026gt;\u0026thinsp;UK under Unanimous Response (estimate\u0026thinsp;=\u0026thinsp;1.436, SE\u0026thinsp;=\u0026thinsp;0.336, z\u0026thinsp;=\u0026thinsp;4.277, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), no difference under Majoritarian Silence (estimate = \u0026minus;\u0026thinsp;0.459, SE\u0026thinsp;=\u0026thinsp;0.338, z = \u0026minus;\u0026thinsp;1.358, p\u0026thinsp;=\u0026thinsp;0.174) or Majoritarian Response (estimate\u0026thinsp;=\u0026thinsp;0.253, SE\u0026thinsp;=\u0026thinsp;0.338, z\u0026thinsp;=\u0026thinsp;0.748, p\u0026thinsp;=\u0026thinsp;0.454) (Fig.\u0026nbsp;6d). Overall, German participants rated societies as more tolerant under consensus norms, whereas UK participants\u0026rsquo; ratings were stable, providing descriptive support for H8.\u003c/p\u003e \u003cp\u003eH9: Mixed vs. Nonverbal Responses\u003c/p\u003e \u003cp\u003eMixed responses consistently outperformed nonverbal-symbolic responses alone in perceived effectiveness. In Germany: Mixed\u0026thinsp;\u0026lt;\u0026thinsp;Nonverbal-symbolic for perceived harm (estimate = \u0026minus;\u0026thinsp;1.525, SE\u0026thinsp;=\u0026thinsp;0.144, z = \u0026minus;\u0026thinsp;10.563, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and Mixed\u0026thinsp;\u0026gt;\u0026thinsp;Nonverbal for perceived societal tolerance (estimate\u0026thinsp;=\u0026thinsp;1.544, SE\u0026thinsp;=\u0026thinsp;0.136, z\u0026thinsp;=\u0026thinsp;11.334, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) (Figs.\u0026nbsp;4a, \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e5\u003c/span\u003ea). In the UK: Mixed\u0026thinsp;\u0026lt;\u0026thinsp;Nonverbal for perceived harm (estimate = \u0026minus;\u0026thinsp;0.633, SE\u0026thinsp;=\u0026thinsp;0.144, z = \u0026minus;\u0026thinsp;4.400, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and Mixed\u0026thinsp;\u0026gt;\u0026thinsp;Nonverbal for perceived tolerance (estimate\u0026thinsp;=\u0026thinsp;1.081, SE\u0026thinsp;=\u0026thinsp;0.136, z\u0026thinsp;=\u0026thinsp;7.946, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Our results indicate that adding a nonverbal-symbolic component to verbal responses enhanced their perceived effectiveness, supporting H9.\u003c/p\u003e \u003cp\u003eH10: Simple Effects of Social Norms Within Response Types\u003c/p\u003e \u003cp\u003eThis analysis examined how social-norm context affected perceived harm and tolerance across response types. Within Verbal responses: Germany: Perceived Harm is higher under Majoritarian Silence than Unanimous Response (estimate\u0026thinsp;=\u0026thinsp;1.083, SE\u0026thinsp;=\u0026thinsp;0.432, z\u0026thinsp;=\u0026thinsp;2.504, p\u0026thinsp;=\u0026thinsp;0.033) (Supplementary Fig.\u0026nbsp;2b). Perceived Tolerance is lower under Majoritarian Silence than Unanimous Response (estimate = \u0026minus;\u0026thinsp;2.813, SE\u0026thinsp;=\u0026thinsp;0.346, z = \u0026minus;\u0026thinsp;8.132, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and lower under Majoritarian Response than Unanimous Response (estimate =-1.598, SE\u0026thinsp;=\u0026thinsp;0.335, z = \u0026minus;\u0026thinsp;4.776, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). UK: No significant differences for harm or tolerance (all p\u0026thinsp;\u0026gt;\u0026thinsp;0.20). Within Nonverbal-symbolic responses: Germany: Tolerance lower under Majoritarian Silence than Unanimous Response (estimate = \u0026minus;\u0026thinsp;2.253, SE\u0026thinsp;=\u0026thinsp;0.345, z = \u0026minus;\u0026thinsp;6.532, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and lower under Majoritarian Response than Unanimous Response (estimate = \u0026minus;\u0026thinsp;1.437, SE\u0026thinsp;=\u0026thinsp;0.337, z = \u0026minus;\u0026thinsp;4.260, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001); harm did not vary (all p\u0026thinsp;\u0026gt;\u0026thinsp;0.10) UK: No differences for tolerance or harm (all p\u0026thinsp;\u0026gt;\u0026thinsp;0.20). Within Mixed responses: Germany: Tolerance substantially lower under Majoritarian Silence (estimate = \u0026minus;\u0026thinsp;2.393, SE\u0026thinsp;=\u0026thinsp;0.337, z = \u0026minus;\u0026thinsp;7.106, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and Majoritarian Response (estimate = \u0026minus;\u0026thinsp;1.574, SE\u0026thinsp;=\u0026thinsp;0.329, z = \u0026minus;\u0026thinsp;4.779, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) compared to Unanimous Response; harm showed a nonsignificant trend toward reduction (estimate\u0026thinsp;=\u0026thinsp;0.777, SE\u0026thinsp;=\u0026thinsp;0.430, z\u0026thinsp;=\u0026thinsp;1.804, p\u0026thinsp;=\u0026thinsp;0.168). UK: Neither perceived tolerance (estimate = \u0026minus;\u0026thinsp;0.497, SE\u0026thinsp;=\u0026thinsp;0.341, z = \u0026minus;\u0026thinsp;1.461, p\u0026thinsp;=\u0026thinsp;0.310) nor perceived harm varied across norms. Overall, greater normative consensus was associated with higher perceived tolerance in Germany, whereas UK participants\u0026rsquo; ratings remained largely stable, supporting H10 in Germany but not in the UK.\u003c/p\u003e \u003cp\u003eExploratory Results: Reaction Time Differences Between Response Types\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eIn Germany: Mixed slower than nonverbal-symbolic (estimate\u0026thinsp;=\u0026thinsp;2.060, SE\u0026thinsp;=\u0026thinsp;0.773, t\u0026thinsp;=\u0026thinsp;2.665, p\u0026thinsp;=\u0026thinsp;0.021), Mixed vs. verbal: nonsignificant (estimate = \u0026minus;\u0026thinsp;1.354, SE\u0026thinsp;=\u0026thinsp;0.773, t = \u0026minus;\u0026thinsp;1.752, p\u0026thinsp;=\u0026thinsp;0.186) and, Nonverbal faster than verbal (estimate = \u0026minus;\u0026thinsp;3.414, SE\u0026thinsp;=\u0026thinsp;0.773, t = \u0026minus;\u0026thinsp;4.417, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). In the UK: Mixed slower than nonverbal (estimate\u0026thinsp;=\u0026thinsp;3.699, SE\u0026thinsp;=\u0026thinsp;0.799, t\u0026thinsp;=\u0026thinsp;4.631, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), Mixed vs. verbal: nonsignificant (estimate\u0026thinsp;=\u0026thinsp;0.083, SE\u0026thinsp;=\u0026thinsp;0.799, t\u0026thinsp;=\u0026thinsp;0.104, p\u0026thinsp;=\u0026thinsp;0.994) and, Nonverbal faster than verbal (estimate = \u0026minus;\u0026thinsp;3.616, SE\u0026thinsp;=\u0026thinsp;0.799, t = \u0026minus;\u0026thinsp;4.527, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). These results suggest verbal responses required more processing time, while nonverbal-symbolic elements were processed more fluently (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e7\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eCovariate Results\u003c/p\u003e \u003cp\u003eNominal Predictors\u003c/p\u003e \u003cp\u003eFigure 8. \u003cem\u003eEstimated effects of gender and LGBTQ+ identity on perceived harm and tolerance\u003c/em\u003e. Posterior mean estimates (β) with 95% credible intervals for gender (male vs. female) and LGBTQ+ identity (yes vs. no) in Germany and the UK, derived from Bayesian cumulative regression models for harm and tolerance ratings.\u003c/p\u003e \u003cp\u003eSpecifically, within Gender: Male participants rated harm lower than female participants (Germany: β = \u0026minus;\u0026thinsp;1.110, 95% CrI [\u0026ndash;1.688, \u0026minus;\u0026thinsp;0.537]; UK: β = \u0026minus;\u0026thinsp;1.599, 95% CrI [\u0026ndash;2.194, \u0026minus;\u0026thinsp;1.019]). Regarding tolerance perception: Germany uncertain (β\u0026thinsp;=\u0026thinsp;0.084, 95% CrI [\u0026ndash;0.362, 0.522]); UK Male participants gave higher tolerance ratings (β\u0026thinsp;=\u0026thinsp;0.598, 95% CrI [0.139, 1.028]) (Fig.\u0026nbsp;8). LGBTQ+ identity: No apparent differences in harm perception (Germany: β\u0026thinsp;=\u0026thinsp;0.505, 95% CrI [\u0026ndash;0.209, 1.177]; UK: β\u0026thinsp;=\u0026thinsp;0.302, 95% CrI [\u0026ndash;0.549, 1.162]) or tolerance (Germany: β = \u0026minus;\u0026thinsp;0.389, 95% CrI [\u0026ndash;0.969, 0.157]; UK: β = \u0026minus;\u0026thinsp;0.211, 95% CrI [\u0026ndash;0.894, 0.438]). Experience living abroad: No meaningful effects on harm or tolerance (all β\u0026thinsp;\u0026asymp;\u0026thinsp;0) (Fig.\u0026nbsp;8).\u003c/p\u003e \u003cp\u003eOrdinal and Continuous Covariates\u003c/p\u003e \u003cp\u003eReligion: No association with harm (Germany: β = \u0026minus;\u0026thinsp;0.267, 95% CrI [\u0026ndash;0.791, 0.234]; UK: β = \u0026minus;\u0026thinsp;0.185, 95% CrI [\u0026ndash;0.895, 0.460]); for tolerance, UK positive (β\u0026thinsp;=\u0026thinsp;0.716, 95% CrI [0.135, 1.287]), Germany uncertain (β = \u0026minus;\u0026thinsp;0.148, 95% CrI [\u0026ndash;0.550, 0.288]). Education: No meaningful effects on harm or tolerance (all β\u0026thinsp;\u0026asymp;\u0026thinsp;0). Socio-political attitude: Germany positively associated with harm (β\u0026thinsp;=\u0026thinsp;0.617, 95% CrI [0.383, 0.928]); UK: negative/uncertain for perceived harm (β = \u0026minus;\u0026thinsp;0.169, 95% CrI [\u0026ndash;0.427, 0.085]); no effect on tolerance. Age: No effect on harm (all β\u0026thinsp;\u0026asymp;\u0026thinsp;0); tolerance slightly higher in older UK participants (β\u0026thinsp;=\u0026thinsp;0.304, 95% CrI [0.021, 0.679]). Experience with discrimination: No clear effect on harm; tolerance lower in UK participants with more exposure compared to Germany (β = \u0026minus;\u0026thinsp;0.487, 95% CrI [\u0026ndash;0.963, \u0026minus;\u0026thinsp;0.071]) (See Supplementary Figs.\u0026nbsp;3 and 4).\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eOur findings lend empirical support to the view that responses to hate speech operate beyond mere words, and that visible, symbolic actions constitute a distinct and consequential dimension of civic resistance. By examining how third-party observers interpret verbal, nonverbal-symbolic, and mixed responses to homophobic speech across Germany and the UK, this study shows that citizens’ responses shape perceptions of harm and societal tolerance in differentiated ways. Rather than identifying a single form of counterspeech as uniformly superior, the results point to a more complex pattern in which response modalities fulfil distinct social functions, depending on their combination, the visibility of social norms, and the cultural legibility of the cues involved.\u003c/p\u003e \u003cp\u003eRegarding perceived harm, the results do not support a straightforward substitution account. Across both Germany and the UK, purely nonverbal-symbolic responses were associated with higher perceived harm than verbal responses (H1a; H3). This pattern suggests that, for ordinary observers, symbolic opposition alone may not provide sufficient interpretive clarity to mitigate the perceived severity of a hate incident. In the absence of explicit verbal rejection, visible symbols may even heighten the salience of the incident by drawing attention to unresolved tension or ambiguity. By contrast, multimodal responses—combining verbal and symbolic cues—were associated with lower harm in Germany and under conditions of majoritarian silence (H1b; H3), indicating that symbolic elements can strengthen verbal opposition when they operate in concert rather than isolation.\u003c/p\u003e \u003cp\u003eSignificantly, social norms alone did not reliably reduce harm perceptions (H2). Even when most citizens in the scene visibly opposed the hateful remark, harm ratings remained largely stable unless response modality was taken into account. This suggests that harm evaluations are relatively resistant to socio-normative modulation and depend more on the clarity and explicitness of opposition than on mere consensus. Cross-cultural differences further reinforce this point: German participants consistently rated incidents as more harmful than UK participants in verbal and nonverbal conditions (H4), indicating baseline differences in sensitivity to homophobic speech that cannot be explained solely by response patterns.\u003c/p\u003e \u003cp\u003eA different dynamic emerged for perceived societal tolerance. Here, social norms played a central role, particularly in Germany, where tolerance ratings increased in a complex and patterned manner as opposition became more widely shared (H6; H10). For citizens observing these scenes, a visible consensus appeared to recalibrate expectations about what is socially acceptable, even when the harmful act itself remained salient. Mixed responses, combining symbolic and verbal elements, were especially effective under conditions of majoritarian silence (H7a), suggesting that symbolic cues can compensate for the absence of verbal unanimity by making normative opposition perceptually salient. In contrast, tolerance judgements in the UK were comparatively stable across response types and norm conditions, suggesting weaker norm updating in response to bystander behaviour. Notably, the German results point to a decisive advantage of mixed responses over purely verbal ones, indicating that the combination of spoken objection with visible symbols of inclusion carries distinctive normative force in this context. This finding lends empirical support to civic and institutional initiatives in Germany that encourage the public display of inclusive symbols—such as flags, colours, or badges—not as substitutes for verbal opposition, but as amplifiers that make dissent socially legible and collectively anchored.\u003c/p\u003e \u003cp\u003eThese findings highlight a significant dissociation between harm and tolerance. While harm judgements reflect the perceived severity of the incident itself, tolerance judgements capture broader inferences about the society in which the incident unfolds. Mixed responses appear particularly well suited to shaping the latter, not by directly repairing harm, but by signalling collective values and expectations. This interpretation aligns with research on metanorms and sanctioning, which shows that individuals’ willingness to intervene depends on their expectations about how others will evaluate such intervention (Fehr \u0026amp; Fischbacher, 2004; Willer, Kuwabara, \u0026amp; Macy, 2009; Horne et al., 2009; Horne, 2014). By making opposition visible, mixed responses reduce uncertainty about social approval and clarify that intervening against hate is likely to be supported rather than sanctioned. From a theoretical standpoint, this supports accounts in communication science and cognitive theory that treat symbolic and bodily cues not as peripheral to meaning, but as constitutive elements through which social understanding and normative expectations are established before, and alongside, explicit verbal interpretation.\u003c/p\u003e \u003cp\u003eReaction-time results further support this interpretation. Nonverbal-symbolic responses were processed more rapidly than verbal responses across both countries, whereas multimodal responses did not impose additional processing demands beyond verbal responses. Symbolic cues function as perceptually efficient signals that can be integrated quickly into social judgements. Longer reaction times for verbal responses likely reflect deeper semantic engagement rather than inefficiency. Together, these findings suggest that mixed responses combine perceptual immediacy with interpretive clarity, offering a form of civic intervention that is both cognitively accessible and normatively informative.\u003c/p\u003e \u003cp\u003eOverall, our results suggest that responses to hate speech operate within a shared social ecology in which meaning is inferred not only from what is said, but from what is made visible. Multimodal responses do not replace verbal opposition, but augment it by clarifying social alignment, especially for citizens navigating uncertainty about others’ support.\u003c/p\u003e \u003cp\u003eLimitations\u003c/p\u003e \u003cp\u003eSeveral limitations should be considered when interpreting these findings.\u003c/p\u003e \u003cp\u003eFirst, symbolic cues are inherently dependent on cultural legibility. While mixed and nonverbal-symbolic responses shaped tolerance perceptions in Germany, their effects were weaker and less systematic in the UK. This asymmetry suggests that symbols do not carry stable meanings across contexts and that their normative force depends on historically sedimented conventions and shared interpretive frameworks. For citizens witnessing hate incidents, a symbol may signal collective resistance in one context and personal expression in another. Future research should therefore vary the types of symbols and their degrees of conventionalisation to assess how legibility conditions norm inference.\u003c/p\u003e \u003cp\u003eSecond, the limited impact of symbolic responses on perceived harm highlights a boundary condition. Multimodal responses appear better suited to signalling collective values than to alleviating the immediate perceived severity of harm. This distinction is normatively important: visible opposition may shape future expectations and behaviour without undoing the emotional or dignitary damage already inflicted. Symbolic responses should therefore be understood as complementary to, rather than substitutes for, verbal or institutional mechanisms that directly address harm and accountability.\u003c/p\u003e \u003cp\u003eThird, the effects observed here rely on collective visibility. The strong role of unanimous and majoritarian response conditions indicates that isolated acts of opposition may be insufficient to shift social expectations. For third-party observers, it is the perception of shared alignment, rather than individual moral expression, that stabilises norms. This raises questions about threshold effects, coordination, and durability that cannot be fully captured in vignette-based designs.\u003c/p\u003e \u003cp\u003eFourth, reaction-time measures offer only indirect insight into cognitive processing. Faster responses to symbolic cues indicate perceptual fluency, but do not capture longer-term persuasion, learning, or behavioural change (Sweller, 2011; Paas \u0026amp; Sweller, 2014). Combining experimental methods with longitudinal or qualitative approaches would help clarify how momentary norm inferences translate into sustained civic engagement.\u003c/p\u003e \u003cp\u003eFinally, the study does not address potential backlash or strategic misuse of symbolic opposition. While visible resistance can clarify norms, it may also provoke counter-mobilisation or devolve into tokenistic signalling. Understanding when multimodal responses empower democratic participation and when they risk trivialisation remains an important task for future work.\u003c/p\u003e "},{"header":"Conclusions and Implications","content":"\u003cp\u003eTaken together, these findings support a view of counterspeech as a coordination problem rather than a purely individual moral act. Citizens do not respond to hate speech in a vacuum; they act under uncertainty about others’ values, likely reactions, and the social costs of intervention. Multimodal responses appear particularly valuable in this respect because they reduce uncertainty about collective alignment. By making opposition perceptually salient, they help witnessing citizens infer not only that a comment is objectionable, but that opposing it is socially supported.\u003c/p\u003e\u003cp\u003eThe dissociation between harm and tolerance outcomes is normatively significant. While multimodal responses do not reliably repair harm at the level of immediate perception, they play a crucial role in shaping the social meaning of the incident and the normative environment in which future speech will occur. This suggests that resistance to hate speech should not be evaluated solely in terms of immediate harm reduction, but also in terms of its capacity to stabilise inclusive norms and discourage normalisation over time.\u003c/p\u003e\u003cp\u003eFrom a theoretical perspective, the study contributes to research on hate speech, social norms, and civic response by demonstrating that resistance operates through multiple communicative channels. It empirically supports accounts that emphasise the importance of visibility, collective uptake, and perceptual cues in norm transmission, extending them beyond verbal interaction. By focusing on third-party perceptions, the study highlights how meaning is co-constructed by those who witness hate incidents and interpret what responses, if any, signal about shared values.\u003c/p\u003e\u003cp\u003ePractically, the findings have implications for how democratic societies think about responding to hate speech in everyday settings. Mixed responses, combining verbal and nonverbal-symbolic elements, offer citizens ways to express disalignment that do not rely solely on confrontation or institutional sanction. They provide a means of making values visible in public spaces, particularly where power dynamics, risk, or uncertainty constrain verbal engagement. In contexts such as Germany, where symbolic cues appear to be widely legible, our findings suggest that campaigns promoting visible signs of inclusion are not merely expressive gestures, but can meaningfully enhance the perceived social impact of verbal opposition when both are deployed together. At the same time, the results caution against treating symbolic opposition as a standalone solution. Its effectiveness depends on cultural legibility, collective visibility, and integration with verbal and institutional responses.\u003c/p\u003e\u003cp\u003eMore broadly, this study suggests that the civic challenge posed by hate speech is not only a matter of regulating speech or persuading speakers, but of sustaining shared expectations about what a society tolerates. Multimodal responses contribute to this task by marking the boundaries of acceptable expression in ways that are publicly observable, socially coordinated, and cognitively accessible to those who witness them.\u003c/p\u003e\u003cp\u003eIn conclusion, responding to hate speech is not only about speaking back, but about shaping the social space in which speech is heard. By examining how citizens interpret verbal, nonverbal-symbolic, and multimodal responses across cultural contexts, this study shows that visible opposition matters—not because it always lessens harm, but because it helps sustain the normative conditions under which harm is recognised, contested, and less likely to be normalised. Multimodal responses thus emerge as a meaningful form of democratic action, complementing verbal engagement and institutional measures by making resistance visible where it might otherwise remain silent.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003eData \u0026amp; code availability\u003c/p\u003e\n\u003cp\u003eAll anonymised data and analysis scripts supporting the findings of this study are openly available in the OSF at [https://osf.io/uyhce/overview?view_only=8bbf4526080f4406babe7b2bc9cd73b5]. The repository includes the raw and processed datasets, the R scripts used for analyses, and the code for figure generation.\u003c/p\u003e\n\u003cp\u003eAcknowledgements\u003c/p\u003e\n\u003cp\u003eJZ and AE were supported by the Mentoring Program from [redacted for peer review]. The funders had no role in study design, data collection and analysis, the decision to publish or the preparation of the manuscript.\u003c/p\u003e\n\u003cp\u003eAuthor contributions\u003c/p\u003e\n\u003cp\u003eJZ: Conceptualization, methodology, investigation, writing original draft, project administration, funding acquisition; JS: Methodology, formal analysis, visualisation, data curation; AE: Formal analysis, investigation, resources, visualisation, data curation; OD: Methodology, supervision and funding acquisition.\u003c/p\u003e\n\u003cp\u003eCompeting Interests\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests as defined by Nature Research or other interests that might be perceived to influence the results and/or discussion reported in this paper.\u003c/p\u003e\n\u003cp\u003eEthical Approval\u003c/p\u003e\n\u003cp\u003eThe Ethics Committee of the Faculty of Philosophy, Philosophy of Science and Religious Studies at the Ludwig Maximilian University of Munich (Germany), as the competent approval body for the study design and procedures, approved the main protocol for this study on February 10, 2022 (ID-Number 131874). The ethical approval covered online recruitment and data collection involving adult participants residing in Germany and the United Kingdom. The approval remained valid at the time of data collection. All research involving human participants was conducted in accordance with relevant international and institutional ethical guidelines and regulations, including the principles of the Declaration of Helsinki.\u003c/p\u003e\n\u003cp\u003eInformed Consent\u003c/p\u003e\n\u003cp\u003eInformed consent was obtained from all participants prior to their participation in the study. The study was conducted entirely online, and participants were recruited via the Prolific (https://www.prolific.com/) platform, with data collected using the Qualtrics (https://www.qualtrics.com/) platform. Consent was obtained electronically by the research team before any study procedures commenced, through an online consent form presented on the Qualtrics platform. Participants were required to actively indicate their consent by clicking an \u0026ldquo;I wish to participate\u0026rdquo; button before proceeding to the study. Informed consent was obtained between July 2024 and September 2024 from all adult participants residing in the United Kingdom and Germany.\u003c/p\u003e\n\u003cp\u003eBefore providing consent, participants were fully informed about the purpose of the research, the non-interventional nature of the study, the procedures involved (completion of a short online questionnaire following the presentation of four visual scenarios), the expected duration of participation (approximately five minutes), and the compensation provided (\u0026pound;0.90 via Prolific upon successful completion). Participants were informed that some visual or verbal stimuli might contain homophobic speech and could be potentially distressing. They were informed of their right to withdraw from the study at any time without penalty and without loss of compensation.\u003c/p\u003e\n\u003cp\u003eParticipants were explicitly informed that their participation was voluntary, that their responses would remain anonymous, and that no personally identifiable information would be collected. They were informed that data would be analysed only in aggregated form, used solely for research purposes, and reported in scientific publications at the group level. The scope of consent covered participation in the study, the anonymous collection and analysis of data, and the use of anonymised data for academic publication. Participants were also informed about the reasons for conducting the research, how their data would be utilised, and that no risks beyond those ordinarily associated with viewing online content were anticipated. Information about appropriate support services was provided to participants, including resources specific to the United Kingdom and Germany, should they experience distress related to the study content. The study involved only adult participants and did not include minors or other vulnerable populations. No oral consent procedures were used\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eAguilera-Carnerero C, Becker MJ, Scheiber M (2025) Same tools, different target: countering hate speech with memes.\u003c/li\u003e\n\u003cli\u003e\u0026Aacute;lvarez-Benjumea A (2023) Uncovering hidden opinions: social norms and the expression of xenophobic attitudes. Eur Sociol Rev 39(3):449\u0026ndash;463\u003c/li\u003e\n\u003cli\u003e\u0026Aacute;lvarez-Benjumea A, Winter F (2020) The breakdown of antiracist norms: a natural experiment on hate speech after terrorist attacks. Proc Natl Acad Sci USA 117(37):22800\u0026ndash;22804\u003c/li\u003e\n\u003cli\u003eAyala S, Vasilyeva N (2016) Responsibility for silence. J Soc Philos 47(3)\u003c/li\u003e\n\u003cli\u003eAyala-L\u0026oacute;pez S (2018) A structural explanation of injustice in conversations: it\u003cspan dir=\"RTL\"\u003e\u0026rsquo;\u003c/span\u003es about norms. Pac Philos Q 99(4):726\u0026ndash;748\u003c/li\u003e\n\u003cli\u003eBecker MJ, Ascone L, Bolton M, Bundz\u0026iacute;kov\u0026aacute; V, Chapelan A, Haupeltshofer P, et al (2024) Decoding antisemitism: an AI-driven study on hate speech and imagery online (Discourse Report No. 6)\u003c/li\u003e\n\u003cli\u003eBenesch S (2014) Defining and diminishing hate speech. In: State of the world\u003cspan dir=\"RTL\"\u003e\u0026rsquo;\u003c/span\u003es minorities and indigenous peoples 2014, pp 18\u0026ndash;25\u003c/li\u003e\n\u003cli\u003eBlanchard FA, Crandall CS, Brigham JC, Vaughn LA (1994) Condemning and condoning racism: a social context approach to interracial settings. J Appl Psychol 79(6):993\u0026ndash;997. https://doi.org/10.1037/0021-9010.79.6.993\u003c/li\u003e\n\u003cli\u003eBogerts L, Fielitz M (2019) \u003cspan dir=\"RTL\"\u003e\u0026ldquo;\u003c/span\u003eDo you want meme war?\u0026rdquo;Understanding the visual memes of the German far right.\u003c/li\u003e\n\u003cli\u003eBrauer M, Chekroun P (2005) The relationship between perceived violation of social norms and social control: situational factors influencing the reaction to deviance. J Appl Soc Psychol 35(7):1519\u0026ndash;1539\u003c/li\u003e\n\u003cli\u003eB\u0026uuml;nzli F, Eppler MJ (2025) Calls to action and user engagement: the role of visual and verbal requests in nonprofit advocacy on Facebook. Comput Hum Behav Rep 18:100611\u003c/li\u003e\n\u003cli\u003eCastelli L, Carraro L, Pavan G, Murelli E, Carraro A (2012) The power of the unsaid: the influence of nonverbal cues on implicit attitudes. J Appl Soc Psychol 42(6):1376\u0026ndash;1393\u003c/li\u003e\n\u003cli\u003eCepollaro B, Lepoutre M, Simpson RM (2023) Counterspeech. Philos Compass 18(1):e12890\u003c/li\u003e\n\u003cli\u003ede Melo CM, Terada K, Santos FC (2021) Emotion expressions shape human social norms and reputations. iScience 24(3)\u003c/li\u003e\n\u003cli\u003eDixon D (2019) Alterpieces: artworks as shifting speech acts. Dissertation\u003c/li\u003e\n\u003cli\u003eDixon D (2022) Artistic (counter) speech. J Aesthet Art Crit 80(4):409\u0026ndash;419\u003c/li\u003e\n\u003cli\u003eDrainville R, Saul J (2024) Visual and linguistic dogwhistles.\u003c/li\u003e\n\u003cli\u003eECRI (European Commission Against Racism and Intolerance) (2016) General Policy Recommendation No. 15 on Combating Hate Speech. https://www.coe.int/en/web/european-commission-against-racism-and-intolerance/recommendation-no.15\u003c/li\u003e\n\u003cli\u003eEsquith SL (2011) The political responsibilities of everyday bystanders. Penn State Press\u003c/li\u003e\n\u003cli\u003eFehr E, Fischbacher U (2004) Social norms and human cooperation. Trends Cogn Sci 8(4):185\u0026ndash;190. https://doi.org/10.1016/j.tics.2004.02.007\u003c/li\u003e\n\u003cli\u003eFricker M (2007) Epistemic injustice: power and the ethics of knowing\u003c/li\u003e\n\u003cli\u003eFumagalli C (2021) Counterspeech and ordinary citizens: how? When? Polit Theory 49(6):1021\u0026ndash;1047\u003c/li\u003e\n\u003cli\u003eGennaro G, Derksen L, Abdelrahman A, Broggini E, Green MA, Haerter VA, et al (2025) Counterspeech encouraging users to adopt the perspective of minority groups reduces hate speech and its amplification on social media. Sci Rep15(1):22018\u003c/li\u003e\n\u003cli\u003eGurung MI, Agarwal N, Bhuiyan MMI, Poudel D (2025) Symbolic signals on Instagram: how visual media shapes engagement, emotion, trust, and diffusion. Soc Netw Anal Min 15(1):57\u003c/li\u003e\n\u003cli\u003eHangartner D, Gennaro G, Alasiri S, Bahrich N, Bornhoft A, Boucher J et al (2021) Empathy-based counterspeech can reduce racist hate speech in a social media field experiment. Proc Natl Acad Sci USA 118(50):e2116310118\u003c/li\u003e\n\u003cli\u003eHardy BW (2021) Embodied cognition in communication science. Commun Theory 31(4):633\u0026ndash;653. https://doi.org/10.1093/ct/qtab021\u003c/li\u003e\n\u003cli\u003eHess U, Fischer A (2024) Emotional mimicry. In: Emotional communication and social interaction. Springer\u003c/li\u003e\n\u003cli\u003eHorne C, Chen CF, Berg J, Evermann-Druffel K (2009) Metanorm expectations: determining what to sanction. In: Altruism and prosocial behavior in groups, pp 199\u0026ndash;223. Emerald Group Publishing\u003c/li\u003e\n\u003cli\u003eHorne C (2014) The relational foundation of norm enforcement. In: The complexity of social norms, pp 105\u0026ndash;120. Springer, Cham\u003c/li\u003e\n\u003cli\u003eHolm G, Sahlstr\u0026ouml;m F, Zilliacus H (2018) Arts-based visual research. In: Leavy P (ed) Handbook of arts-based research, pp 311\u0026ndash;335. Guilford Press\u003c/li\u003e\n\u003cli\u003eHoward JW (2021) Terror, hate and the demands of counter-speech. Br J Polit Sci 51(3):924\u0026ndash;939. https://doi.org/10.1017/S000712341900053X\u003c/li\u003e\n\u003cli\u003eINACH (International Network Against Cyber Hate) (2024) What is cyber hate and why do we want to counter it. https://www.inach.net/cyber-hate-definitions\u003c/li\u003e\n\u003cli\u003eJust MA, Carpenter PA (1992) A capacity theory of comprehension: individual differences in working memory. Psychol Rev 99(1):122\u0026ndash;149\u003c/li\u003e\n\u003cli\u003eKhanolainen D, Semenova E (2020) School bullying through graphic vignettes: developing a new arts-based method to study a sensitive topic. Int J Qual Methods 19. https://doi.org/10.1177/1609406920922765\u003c/li\u003e\n\u003cli\u003eLangton R (2018a) Blocking as counter-speech. Philos Public Aff 46(4):363\u0026ndash;393. https://doi.org/10.1111/papa.12138\u003c/li\u003e\n\u003cli\u003eLangton R (2018b) The authority of hate speech. Oxford Stud Philos Law 3:123\u0026ndash;152\u003c/li\u003e\n\u003cli\u003eLangton R (2018c) Speech acts and pornography. In: Speech and harm\u003c/li\u003e\n\u003cli\u003eLepoutre M (2017) Hate speech in public discourse: A pessimistic defense of counterspeech. Soc Theory Pract:851\u0026ndash;883\u003c/li\u003e\n\u003cli\u003eLepoutre M (2021) Democratic speech in divided times. Oxford University Press, Oxford\u003c/li\u003e\n\u003cli\u003eMadella P, Wharton T (2023) Nonverbal communication and context: multimodality in interaction. In: The Cambridge handbook of language in context. Cambridge University Press\u003c/li\u003e\n\u003cli\u003eMaitra I, McGowan MK (2012) Speech and harm: controversies over free speech\u003c/li\u003e\n\u003cli\u003eMatsuda MJ (2018) Words that wound: critical race theory, assaultive speech, and the First Amendment. Routledge\u003c/li\u003e\n\u003cli\u003eMayer RE (2009) Multimedia learning, 2nd edn. Cambridge University Press\u003c/li\u003e\n\u003cli\u003eMcGowan MK (2018) Responding to harmful speech: the more speech response, counter speech, and the complexity of language use. In: Voicing dissent, pp 182\u0026ndash;199. Routledge\u003c/li\u003e\n\u003cli\u003eMoreno R, Mayer RE (2007) Interactive multimodal learning environments. Educ Psychol Rev 19:309\u0026ndash;326\u003c/li\u003e\n\u003cli\u003eOlabisi OO (2018) A comparative perspective on the First Amendment: free speech, militant democracy, and the primacy of dignity as a preferred constitutional value in Germany. Alabama Civ Rights Civ Liberties Law Rev 10(1):169\u0026ndash;204\u003c/li\u003e\n\u003cli\u003ePaas F, Sweller J (2014) Implications of cognitive load theory for multimedia learning. In: Cognitive load theory. Springer\u003c/li\u003e\n\u003cli\u003ePaluck EL, Green DP (2009) Deference, dissent, and dispute resolution: an experimental intervention using mass media to change norms and behavior in Rwanda. Am Polit Sci Rev 103(4):622\u0026ndash;644. https://doi.org/10.1017/S0003055409990128\u003c/li\u003e\n\u003cli\u003ePluta A, Zeller F, Koglin U (2024) Hate speech exposure and empathic desensitization: neural and behavioral correlates. J Exp Psychol Gen 153(2):295\u0026ndash;310\u003c/li\u003e\n\u003cli\u003ePopa-Wyatt M, Wyatt JL (2018) Slurs, roles and power. Philos Stud 175(11):2879\u0026ndash;2906\u003c/li\u003e\n\u003cli\u003ePosner MI (2005) Timing the brain: mental chronometry as a tool in neuroscience. Oxford University Press\u003c/li\u003e\n\u003cli\u003ePotter MC, Wyble B, Hagmann CE, McCourt ES (2014) Detecting meaning in RSVP at 13 ms per picture. Atten Percept Psychophys 76(2):270\u0026ndash;279\u003c/li\u003e\n\u003cli\u003eSahlstr\u0026ouml;m F (2025) Arts-based visual research. In: Handbook of arts-based research, p 309\u003c/li\u003e\n\u003cli\u003eSaul J (2021) Someone is wrong on the internet: is there an obligation to correct false and oppressive speech on social media? In: MacKenzie A, Rose J, Bhatt I (eds) The epistemology of deceit in a postdigital era: dupery by design, pp 139\u0026ndash;157. Springer, New York\u003c/li\u003e\n\u003cli\u003eSchlenker P (2022) What it all means: semantics for (almost) everything. MIT Press\u003c/li\u003e\n\u003cli\u003eSchmid U, K\u0026uuml;mpel AS, Rieger D (2024) Social media users\u003cspan dir=\"RTL\"\u003e\u0026rsquo; \u003c/span\u003emotives for (not) engaging with hate: a survey study in Germany. Soc Media Soc 10(1)\u003c/li\u003e\n\u003cli\u003eSweller J (2011) Cognitive load theory. Psychol Learn Motiv 55:37\u0026ndash;76\u003c/li\u003e\n\u003cli\u003eThorpe S, Fize D, Marlot C (1996) Speed of processing in the human visual system. Nature 381(6582):520\u0026ndash;522\u003c/li\u003e\n\u003cli\u003eUdupa S (2023) Extreme speech. In: Strippel C, Paasch-Colberg S, Emmer M, Trebbe J (eds) Challenges and perspectives of hate speech research, pp 233\u0026ndash;248. Berlin. https://doi.org/10.48541/dcr.v12.14\u003c/li\u003e\n\u003cli\u003eVan Kleef GA (2024) The social life of emotions. Cambridge University Press\u003c/li\u003e\n\u003cli\u003eWachs S, Wettstein A, Bilz L, Espelage DL, Wright MF, G\u0026aacute;mez-Guadix M (2024) Individual and contextual correlates of latent bystander profiles toward racist hate speech: a multilevel person-centred approach. J Youth Adolesc 53(6):1271\u0026ndash;1286. https://doi.org/10.1007/s10964-024-01968-x\u003c/li\u003e\n\u003cli\u003eWaldron J (2012) The harm in hate speech. Harvard University Press\u003c/li\u003e\n\u003cli\u003eWiller R, Kuwabara K, Macy MW (2009) The false enforcement of unpopular norms. Am J Sociol 115(2):451\u0026ndash;490\u003c/li\u003e\n\u003cli\u003eWolfe JM (2018) Visual search: how do we find what we are looking for? Annu Rev Vis Sci 4:539\u0026ndash;562\u003c/li\u003e\n\u003cli\u003eZapata J, Sulik J, von Wulffen C et al (2024) Bystanders\u003cspan dir=\"RTL\"\u003e\u0026rsquo; \u003c/span\u003ecollective responses set the norm against hate speech. Humanit Soc Sci Commun 11:335. https://doi.org/10.1057/s41599-024-02761-8\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"humanities-and-social-sciences-communications","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"palcomms","sideBox":"Learn more about [Humanities \u0026 Social Sciences Communications](http://www.nature.com/palcomms/)","snPcode":"41599","submissionUrl":"https://submission.springernature.com/new-submission/41599/3","title":"Humanities and Social Sciences Communications","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Nature AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-8554490/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8554490/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eHate speech in public spaces is usually addressed and regulated as a primarily verbal phenomenon, as are efforts to counter it. Less attention has been paid to how citizens, within protected speech boundaries, oppose hate not only through words but also through visible, symbolic, and multimodal acts\u0026mdash;such as wearing inclusive symbols, displaying specific colours, or aligning physically with marginalised groups\u0026mdash;and how these responses shape perceptions of harm, tolerance, and democratic norms. This paper examines counterspeech not only as a verbal act but as a symbolic, visual, and multimodal form of civic expression. This experimental study (N\u0026thinsp;=\u0026thinsp;827) investigates how citizens\u0026rsquo; responses to homophobic speech are perceived by third-party observers and how these perceptions influence broader inferences about harm and societal tolerance. Using visual vignettes in Germany and the UK, the research compares three response types: verbal counterspeech, nonverbal-symbolic responses (like wearing inclusive symbols), and mixed responses that combine verbal objection with symbolic cues. By situating these within different socio-normative climates\u0026mdash;from silence to majoritarian and unanimous opposition\u0026mdash;the study conceptualises hate speech as a social phenomenon whose meaning emerges through interaction rather than isolated utterances. Findings reveal a differentiated pattern with key implications for free-speech debates. Symbolic responses alone did not reliably reduce perceived harm and sometimes heightened perceptions of unresolved tension. In contrast, mixed responses proved especially effective in Germany, reducing perceived harm and increasing perceptions of societal tolerance\u0026mdash;particularly where opposition to hate speech was not yet widespread. These effects were weaker in the UK, highlighting how cultural legibility shapes the interpretation of symbolic expression. In both contexts, social consensus alone was insufficient; communicative clarity through multimodal expression was decisive. Theoretically, this study reframes counterspeech as a multimodal, embodied practice that operates within, rather than against, liberal speech democracies. It shows how citizens interpret and enact normative boundaries in daily interaction. Empirically, the findings suggest that visible, symbolic opposition\u0026mdash;when combined with verbal engagement\u0026mdash;can reinforce democratic resilience and social cohesion by clarifying shared values. The paper contributes to broader discussions on how institutional safeguards and citizen-led responses jointly shape inclusive and just societies.\u003c/p\u003e","manuscriptTitle":"Making Resistance Visible: The Role of Nonverbal-Symbolic Opposition to Hate Speech Across Cultures","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-03-13 07:13:25","doi":"10.21203/rs.3.rs-8554490/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"editorInvitedReview","content":"","date":"2026-05-11T07:40:24+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-05-07T09:41:42+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"171734012375209721849959991653289536301","date":"2026-04-20T15:07:41+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"22029549892450077238019735398615299788","date":"2026-04-20T07:24:44+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"279283770447466040254417187456689287818","date":"2026-04-10T07:30:02+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-04-08T14:33:26+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-04-01T12:05:54+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2026-03-30T08:28:24+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-01-25T21:14:16+00:00","index":"","fulltext":""},{"type":"submitted","content":"Humanities and Social Sciences Communications","date":"2026-01-25T21:05:00+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"humanities-and-social-sciences-communications","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"palcomms","sideBox":"Learn more about [Humanities \u0026 Social Sciences Communications](http://www.nature.com/palcomms/)","snPcode":"41599","submissionUrl":"https://submission.springernature.com/new-submission/41599/3","title":"Humanities and Social Sciences Communications","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Nature AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"dc180f80-1a76-42fe-aff0-42cb88dbdafc","owner":[],"postedDate":"March 13th, 2026","published":true,"recentEditorialEvents":[{"type":"editorInvitedReview","content":"","date":"2026-05-11T07:40:24+00:00","index":43,"fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-05-07T09:41:42+00:00","index":42,"fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-04-08T14:39:19+00:00","versionOfRecord":[],"versionCreatedAt":"2026-03-13 07:13:25","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8554490","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8554490","identity":"rs-8554490","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}