Discussion: Making Sense of Public Sensemaking Relative to the COVID-19 Crisis

Abstract

Disease outbreaks motivate human groups to engage in sensemaking efforts to give meaning to the event. These sensemaking processes often involve narratives framing where a disease comes from, how it spreads, and how to prevent and cure infections. At least four generic narratives are typically used as symbolic resources make sense of disease outbreaks: A medical science narrative and three lay narratives, i.e., (1) infectious disease as divine punishment, (2) infectious disease as caused by actions of outgroups (3) infectious disease as caused by evil elites. The contributions to this Special Issue are discussed in relation to this narrative sensemaking perspective.

Human societies have always been threatened by infectious diseases. Even today, infectious diseases constitute one of the main causes of death worldwide (Morens et al., 2004). Human societies have developed strategies of coping with disease threat, from instinctive behaviors evolved in ancestral environments (Schaller & Park, 2011) to traditional ethnobotany to the pharmacological and institutional arsenals of modern medical science. Part of this array is constituted by culturally shared representations of where diseases come from, how they spread, and how to prevent and cure infection (Eicher & Bangerter, 2015). If these representations are accurate, they may protect individuals’ physical integrity, but they also serve to protect the symbolic integrity of groups that support and propagate them. Cultural (or social) representations of diseases are produced, sustained, modified and shared as part of collective sensemaking (Weick, 1995; Wagner & Hayes, 2005) or collective symbolic coping (Wagner et al., 2002), by which groups interpret novel, unexpected or traumatizing events. Sensemaking is a motivated process where data is organized into a simpler, coherent form (Chater & Loewenstein, 2016). One important tool for sensemaking is creating or using a narrative, i.e., storytelling activities which give meaning to experience (Bietti et al., 2019). Narratives organize events into logical, causal or temporal sequences, appealing to pre-existing knowledge structures (Moscovici, 1984) and facilitating collective engagement of audiences (Grall et al., 2021).

The COVID-19 crisis is no exception to these sensemaking processes (Christianson & Barton, 2021) ¹ . But it presents novel features, not least its massively disruptive scope, and its arrival in a “post-truth” era dominated by social media, information proliferation (especially misinformation), opinion polarization and fracturing of the public sphere into self-insulated communities, “echo chambers” or “filter bubbles” (Lewandowsky et al., 2017). Information proliferation creates an economy of attention, where information competes for attention from individual minds. In such an environment, belief-consistent information tends to get preferentially selected for consumption (Hills, 2019), such as the increasingly prevalent alternative worldviews - narratives - that motivate rejection of the science behind health communication messages. The motivation for sensemaking in uncertain situations can direct cognitive processes like confirmation bias or information avoidance to make incoming data “fit” with these narratives (Chater & Loewenstein, 2016).

The current Special Issue on language and communication relative to COVID-19 showcases investigations of cognition, perception and intentions relative to health-related and social behaviors. This is potentially of immense practical importance. In the first waves of COVID-19, containing spread and limiting infections was largely contingent on adoption of non-pharmacological interventions like social distancing, hand-washing and wearing masks. Even immunological measures like vaccination rely on individuals’ willingness to get vaccinated. Accordingly, several articles adopt a health communication approach (Glanz et al., 2008), investigating linguistic features of public health messages to optimize their uptake. But in the current social and media landscape, such messages, however well-crafted, are not the only voices at play. As such, the articles may benefit from being related to a sensemaking approach. Indeed, such an approach is often implicit in the studies’ design. All the studies except Malik et al. (2021) ask participants to make sense of experimental stimuli in more or less narrative form; Malik et al., analyze news, which is per se a form of sensemaking (Wagner et al., 2002). I thus adopt a sensemaking perspective in my comments. Of course, each article reflects the authors’ own theoretical and methodological commitments and makes their own contributions; I thus also comment on them on their own terms. I describe sensemaking narratives of infectious disease as a backdrop before commenting on the articles.

Sensemaking Narratives of Infectious Disease

In the modern era, at least four generic narratives function as symbolic resources (Zittoun, 2007) to make sense of disease outbreaks (Eicher & Bangerter, 2015). The most widespread is the institutionalized narrative of medical science, elaborated by scientists (based on the germ theory of infectious disease) and propagated by educational, governmental and political institutions and the media. It is recent, having attained its dominant position in the last century or so (Washer, 2010). Three other narratives are lay theories, elaborated and propagated by non-scientists (“the public”). These are (1) infectious disease as divine punishment, (2) infectious disease as caused by actions of outgroups (Joffe, 1999), and (3) infectious disease as caused by evil elites. Infectious disease framed as a divine punishment harks back to biblical stories of plagues and recurs regularly, for example during the Black Death or when AIDS (“the gay plague”) emerged in the 1980s ² . It is sometimes fused with infectious disease framed as the result of the actions of outgroups. Many outgroups and minorities are associated with or blamed as the cause of disease outbreaks due to their purportedly unhygienic or immoral practices. This is accomplished via othering (Joffe, 1999) a projection of risk posed by disease onto those groups. Thus, Jews were accused of poisoning wells during the Black Death, AIDS was initially attributed to minorities like intravenous drug users, gay men, prostitutes or Haitians (Fouron, 2013; Washer, 2010), and avian flu has been attributed to purportedly unhygienic Chinese practices (Joffe & Lee, 2004; Washer, 2004). Infectious diseases are also framed as being intentionally engineered by “evil elites”, typically in conspiracy theories. This narrative is perhaps more recent (Campion-Vincent, 1989); in any case it has become a standard feature of recent disease outbreaks (e.g., COVID-19 as a “man-made virus” or a “hoax”).

Narratives feature a cast of characters (Propp, 1968) whose intentions and actions give meaning to the narrative. Narratives about disease outbreaks stage collectives as actors: nations, political or public health institutions, corporations like pharmaceutical companies, social categories based on gender, age or sexual orientation, and mass media (Wagner-Egger et al., 2011). They can be grouped into three categories: heroes, villains and victims. In the context of swine flu, poor countries were often depicted by laypersons as victims, who elicited pity, but were sometimes described as deserving of their plight. Experts (scientists) and authorities were depicted as heroes. Villains were constituted by the media and private corporations like pharmaceutical companies (Wagner-Egger et al., 2011). Analogous figures may be emerging in COVID-19 narratives, albeit in different roles.

Sensemaking is dynamic. Wagner et al. (2002) describe four “stages” of collective symbolic coping: Awareness, divergence, convergence and normalization. Awareness corresponds to the emergence of an issue in the public sphere. Divergence is characterized by the coexistence of multiple, often incompatible narratives emerging from sensemaking attempts. Convergence is when a single dominant narrative emerges, suppressing others and reducing epistemic uncertainty. Normalization is when the issue has become a part of everyday life. These stages can be illustrated by the case of AIDS. When it first emerged in the early 1980s, it initially attracted sporadic reporting in medical journals and the media (awareness) (Washer, 2010). As the disease became more prevalent, multiple explanations coexisted (divergence). Gradually, a dominant medical science narrative of AIDS as caused by the HIV virus emerged (convergence). Normalization occurred after several decades (Rosenbrock et al., 2000), when AIDS became perceived less as a fatal disease and more as a chronic condition. In further research on collective symbolic coping, Mayor et al. (2013) showed how Swiss laypersons’ and media depictions of blame (focusing on villains) shifted during the swine flu pandemic. When the outbreak was a remote disease, sensemaking focused more on collectives like foreign nations. When it arrived in Switzerland, however, sensemaking (and blame) shifted to local collectives like authorities, risk groups, pharmaceutical companies and the media. In another study, in the context of avian flu (Gilles et al., 2013), attributing the disease to Asian countries occurred in the divergence stage of collective symbolic coping. While the collective symbolic coping model may not be applicable to all cases of sensemaking, it highlights how sensemaking adapts to the changing conditions of disease outbreaks.

Comments on the Articles

Shulman et al., describe an experimental investigation of participants’ motivation to engage with complex (jargon-laden) messages as a function of topic urgency (comparing COVID-19 to less urgent topics). They additionally analyze how motivation fluctuates over three waves of data collection, making clever use of the ongoing pandemic. The findings that the relationship between jargon and processing fluency depend on topic urgency and time have interesting theoretical implications. Fluency and central vs. peripheral processing can be viewed as specific processes in the service of sensemaking. The fact that motivation to process jargon over time decreases fits with an economy-of-attention perspective. One could speculate that, concurrently, the relative influence of motivated reasoning processes (perhaps based on participants’ worldviews) increases. If that is the case, then one would expect that ideological variables like perhaps social dominance orientation or conspiracy mentality may emerge as predictors in latter phases of the pandemic.

Tu et al. (2021) describe a carefully controlled experiment showing that you-framing is more effective than we-framing in increasing intentions to adhere to a stay-at-home recommendation from the World Health Organization (WHO). This effect is moderated by self-control. The authors emphasize potential applications for health messaging of their findings. I agree, although the small size of the effect (the difference in stay-at-home likelihood is 6.88 versus 6.74 on a nine-point scale for you vs. we) does limit their applicability. In terms of the sensemaking narratives described above and the characters they stage, I also wondered how we is interpreted, in terms of a sense of imagined community with the WHO. Depending on the position the WHO occupies in participants’ narrative worldviews (e.g., hero or villain), you or we may acquire somewhat different meanings. In this vein, I noticed that trust in WHO affects adherence intentions, and wondered whether it might also potentially affect the potency of the manipulation (that is, a trust-by-pronoun type interaction). Further, the potency of the manipulation may also depend on the organization producing the message (e.g., the WHO versus a local politician or public health agency).

In another carefully controlled experiment, Tian et al. (2021) investigated the effects of you-framing vs. we-framing and hope appeals in supportive messages from a neighbor on coping. This study is a study of sensemaking. Indeed, in a sense, coping is sensemaking (Wagner et al., 2002), especially as the messages constitute invitations to construe the crisis in specific ways. As with Tu et al., here I am skeptical about effects of generic personal pronouns independently of evidence about who those pronouns refer to. This skepticism is borne out by the weak effects of the manipulated variables. In the regression models, message realism, trait hope, hope appeals and communal coping orientations are the strongest predictors. The effect of message realism might reflect a methodological issue (participants who perceived the message as low in realism may have been less willing to engage with the hypothetical scenarios). But it also points to the importance of the content of the message (a sensemaking phenomenon). Perhaps the most interesting theoretical issue relates to hope. As a future-related emotion, hope plays an important motivational role in sensemaking (Steigenberger, 2015). Further, there is a connection to be made between leadership and hope (Helland & Winston, 2005). In terms of sensemaking narratives, leaders (local community leaders, workplace leaders or national leaders) played (and continue to play) an important role during COVID-19 (Crayne & Medeiros, 2020). Part of this role involves communicating hope-based messages. That suggests potential future studies to pursue the authors’ agenda: investigating how hope-based message content in naturalistic content affects message outcomes, or manipulating the leadership status of a message's source.

In another carefully controlled experiment, Janssen et al. (2021) manipulated expressions of epistemic uncertainty via lexical hedges in a message about mask-wearing and the source of the message (a scientist versus a politician), investigating effects on perceived trustworthiness. Hedges had little effect, but scientists were perceived as more trustworthy than politicians. Uncertainty in science is an intricate issue (Kienhues et al., 2020) and its relation to trustworthiness is complex. Janssen et al., surmise that whether expressions of uncertainty increase perceptions of trustworthiness (Hypothesis 1) may depend on the period in the pandemic when the message is produced. I agree, and further speculate that expressions of uncertainty can also serve strategic purposes like undermining scientific consensus, as in the case of climate change (Lewandowsky et al., 2013), in which case it would serve to decrease perceived trustworthiness. Another interesting dimension of this study is, again, related to leadership: Scientists and politicians are leaders who help the public make sense of disease outbreaks (Crayne & Medeiros, 2020). Perceived trustworthiness is an important component of this role. Whether expressing uncertainty increases or decreases their trustworthiness may depend on their charisma, but also on follower characteristics like need for closure or preferences for effortful reasoning (Hendriks & Jucks, 2020).

In another carefully controlled experiment, Lux et al. (2021) investigated the effects of high vs. low message complexity on trust in science and trust in a message advocating mask-wearing. This effect was mediated by social exclusion, but only for participants high in conspiracy mentality. Presumably, these participants feel excluded by a more complex scientific text, which in turn decreases their trust. The possibility that the exclusion effect is of an intergroup nature is intriguing; conspiracy theories and social exclusion are bidirectionally related: Members of socially excluded groups may turn to conspiracy theories to make sense of their disadvantaged situation (Crocker et al., 1999), but being asked to endorse conspiracy theories also leads people to fear social exclusion (Lantian et al., 2018). Indeed, in Lux et al.'s data, conspiracy mentality correlates with exclusion, suggesting a sensitivity to this issue that may be cued by the text complexity (mechanisms by which complexity cues exclusion would be important to understand). In any case, here conspiracy-minded participants may also be engaging in sensemaking using the narrative of disease as caused by evil elites, which frames mask-wearing as a “hoax”; this is consistent with the strong negative correlations between conspiracy mentality and trust.

Malik et al., analyzed news content in world regions varying in historical pathogen prevalence, showing that it predicts COVID-19 news attention, and that moralizing language in news is predicted by the interaction of historical pathogen prevalence and acute infection risk. This work is based on research on the behavioral immune system, which proposes that regions with high pathogen stress in ancestral environments have developed behavioral and cultural adaptations to decrease the risk of infection. Moralizing language is interesting in relation to sensemaking because it may potentially constitute an indicator for the presence of outgroup blame sensemaking narratives or divine punishment narratives. This article is noteworthy relative to the others in the Special Issue because it uses naturalistic textual data to test hypotheses – some of the other research questions could be amenable to such analyses.

Vacchini et al. (2021) are inspired by an actual episode reported in the media during COVID-19 where a taxi driver refused to let a customer of a COVID-19-stigmatized nationality get into his taxi because of fear of contagion (the experimental condition; in the control condition, no reason for the refusal or information about the victim's nationality was given). Effects on perceptions of the protagonists are investigated. The driver is perceived as more moral in the COVID-19 condition and the customer as feeling more social pain. The study materials are an example of the outgroup-blame narrative (Joffe, 1999; Wagner-Egger et al., 2011), featuring potential villain and victim characters. Participants’ adhesion to this narrative may lead them to blame the driver (or not) for his actions, and to help the victim (or not). The way participants relate to this issue illustrates the ambiguous status of victims. The authors also measured social dominance orientation (SDO) in their study but did not relate it to the main measures. This would have been interesting because othering is predicted by SDO, especially during the divergence phase of collective symbolic coping (Gilles et al., 2013).

Conclusion

The articles in this Special Issue test theoretically-grounded propositions, several about effects of linguistic features of health-related or supportive messages. Methodologically, the studies are of high internal validity. Rigorous experimental designs use controlled stimuli (control often equals decontextualization: scientists, politicians, and neighbors are abstract labels). These stimuli constitute the message context for testing subtle linguistic manipulations. The authors are to be commended for their careful work, especially as many studies were probably performed in urgent conditions as the pandemic unfolded.

Part of the strength of the studies is that they attempt to test theories in the context of COVID-19, and yet they often do so in a decontextualized manner, which limits our understanding of how participants make sense of the scenarios. Accordingly, the effects of experimental manipulations are small, while larger effects accrue to individual-difference variables or ideological orientations. Some studies also rely on single messages to instantiate concepts, which limits the possibility of generalizing to the population of messages (Clark, 1973; Reeves et al., 2016). The tradeoff between contextualization (using multiple messages with realistic protagonists to increase message realism, participant engagement, and ultimately, validity) and de-contextualization (which increases control and internal validity) is a fine line to walk. I see four avenues for increasing contextualization while maintaining control. First, scenarios can be fleshed out (completing narratives, using more concrete or real-life protagonists, like Vacchini et al.,). Second, message receiver characteristics (individual differences in cognition, perception and ideology) could be measured more systematically. Third, more use of the wealth of available naturalistic data (Malik et al., are an inspiring example) could be made to complement experiments. Fourth, as demonstrated by Shulman et al. (2021) the current studies could be contextualized in the time course of the pandemic – running some of them again (with potential adaptations to the changing situation) would allow insights into potentially shifting sensemaking patterns.