Immunization effect of inoculation messages: interventions against consensus and misinformation to improve public acceptance of recycled water

Abstract

The global freshwater crisis is accelerating, posing an urgent need for sustainable water solutions. Despite its potential as a viable alternative, recycled water has yet to gain widespread public acceptance. A key barrier is the public’ low intention to use it. Grounded in inoculation theory, this study explores the mechanism through which different types of information influence public acceptance of recycled water. Using eye-tracking technology, it examines the mediating role of information reception effects. The cognitive processing involved in the interaction between inoculation messages, scientific consensus, and misinformation is examined through the lens of dual-processing theory. The results demonstrate a public preference for misinformation about recycled water, leading to lower acceptance when both scientific consensus and misinformation are presented simultaneously. However, inoculation messages effectively mitigate the impact of misinformation while enhancing the reception effects of scientific consensus, resulting in increased acceptance among participants exposed to inoculation. From a micro-level perspective of eye movement behavior, this study illustrates the underlying cognitive mechanism underlying inoculation theory - specifically, how inoculation messages exert immunization effects by either suppressing cognitive processing of misinformation or reinforcing motivated reasoning of scientific consensus. By integrating inoculation theory and dual-processing theory, this study also provides empirical support for negative preference theory within a cognitive science framework. The findings provide practical insights for public communication campaigns, highlighting the value of inoculation strategies in shaping public attitudes toward recycled water. We recommend that marketing efforts incorporate inoculation strategies to foster public attitudinal resilience and prioritize the integration of scientific consensus and inoculation messages in public outreach materials.

Introduction

Water shortages and pollution are crucial challenges that constrain global economic growth, social progress, and even human survival. In response, the broad application of unconventional water sources has been put on the agenda (Tortajada and van Rensburg, 2020). Among these, recycled water–municipal wastewater that undergoes rigorous treatment to meet specific water quality standards for safe reuse (Fu et al. 2020)–is increasingly recognized as a key solution to the global water crisis (Contzen et al. 2023). However, strong psychological resistance from the public presents a significant barrier to the implementation of recycled water projects (Bass et al. 2022). For example, in the United States, some individuals refuse to purchase recycled water (Geipel et al. 2018), or consume products produced with recycled water, such as food (Li et al. 2018) and agricultural products (Savchenko et al. 2019). This resistance stems from negative preferences, wherein recycled water is perceived as unclean or potentially harmful to health (Adapa 2018; Garcia-Cuerva et al. 2016).

To enhance public acceptance of recycled water, many scholars have investigated the key factors influencing recycled water acceptance (Dolnicar et al. 2011; Ross et al. 2014), focusing particularly on the influence of information and strategies to shift public perception (Wester et al. 2016). Scientific information, such as on water quality data (Mainali et al. 2013), treatment processes (Dolnicar et al. 2010), and health implications (Gerrity et al. 2018) has been found helpful to enhance acceptance. However, such technical explanations often require a high level of public scientific literacy (Fu et al. 2022). In other public communication domains, scholars have adopted a more concise and effective communication approach - scientific consensus (Lewandowsky et al. 2012b). Unlike detailed technical descriptions, scientific consensus conveys the collective judgment of experts on complex issues (e.g., “97% of climate scientists have concluded that human-caused global warming is happening” (van der Linden et al., 2017)). Research shows that individuals’ beliefs are significantly influenced by scientific consensus (Goldberg et al. 2019).

At the same time, advances in generative artificial intelligence (AI) has dramatically transformed information environments. While AI has accelerated the dissemination of scientific consensus, it has also made it easier to generate and spread misinformation (Feuerriegel et al. 2023). AI tools can easily distort legitimate scientific consensus into misleading but professional-looking misinformation (Monteith et al. 2023). Research has shown that misinformation can significantly harm public attitudes, particularly in areas, such as public health (Shin, 2024). Although fact-checking is widely used to counter misinformation (Shin and Akhtar, 2024), it is often ineffective due to the continued influence effect, whereby individuals continue to believe misinformation even after it has been corrected (Walter and Tukachinsky, 2019). As a more proactive alternative, researchers have proposed inoculation intervention to combating misinformation (Ecker et al. 2022; Pennycook and Rand, 2021; Feuerriegel et al. 2023). Inoculation theory suggests that exposures to weakened persuasive information (inoculation messages) can stimulate cognitive resistance, enabling individuals to develop attitudinal immunity against misinformation in the future (McGuire, 1964). Previous research on inoculation intervention has primarily examined their effectiveness (Basol et al. 2021; Banas and Rains, 2010; Amazeen et al. 2022), and found that different inoculation messages yield varying intervention effects across diverse contexts (Amazeen et al. 2022).

Despite the growing interest in scientific consensus and misinformation, it remains unclear how these two influential factors interact to influence public acceptance of recycled water. Furthermore, the mechanisms by which inoculation messages interact with both scientific consensus and misinformation in this context remain underexplored. Significant research gap exists concerning the effectiveness and comprehensiveness of information provision in this field. Notably, no prior studies have simultaneously examined the positive effects of scientific consensus and the negative impacts of misinformation in shaping attitudes toward recycled water. Much of existing literature relies on idealized information environments that overlook the complexities of public perception. To address this gap, it is crucial to consider the combined effects of scientific consensus and misinformation on public acceptance of recycled water. Understanding how these elements interact will provide deeper insights into public attitudes and effective communication strategies. In addition, the generalizability of inoculation theory and the intrinsic mechanisms of inoculation intervention are still poorly understood. No existing research has evaluated the effectiveness of inoculation intervention in countering misinformation about recycled water and how it interacts with scientific consensus and misinformation within complex information environments. Investigating these interactions may uncover cognitive mechanisms that explain how inoculation achieves its immunization effects.

This study addresses these research gaps by examining the combined effects of scientific consensus and misinformation on public acceptance of recycled water through information dissemination strategies. Specifically, it aims to explore whether inoculation intervention can effectively protect public attitudes from misinformation and to identify the mechanisms through which inoculation intervention exert immunization effects.

To achieve this, this study employs eye-tracking technology - a cognitive neuroscience tool that captures participants’ real-time intuitive responses to information stimuli (Burlingham et al. 2024). This approach advances traditional experimental techniques for assessing information effectiveness. Through the completion of two eye-tracking experiments, the study investigates how different types of information influence public acceptance of recycled water: Study 1 explores the underlying mechanisms through which scientific consensus and misinformation influence recycled water acceptance by analyzing public information reception effects. Study 2 introduces inoculation messages and compares their impacts with other message types (scientific consensus and misinformation), evaluating their immunization impact and potential interactive mechanisms.

By examining how inoculation messages interact with scientific consensus and misinformation, this research addresses the limitation in prior work that focused exclusively on positive messaging, which often overlooked the complexity of real-world information environment. It advances the internal validity and theoretical mechanisms of inoculation theory and provides insights into how different message types interact to shape public attitudes. Understanding the effectiveness of inoculation theory in the context of recycled water provides valuable guidance for combatting the spread of misinformation and improving the implementation of water reuse initiatives. By revealing the cognitive processed underlying attitudinal immunization, this study informs the optimal design of communication strategies that foster greater public acceptance of recycled water. Ultimately, the findings can provide actionable references for public policy, enhance communication campaigns, and contribute to environmental protection and sustainable socio-economic development.

Literature review

Scientific consensus

Scientific consensus, which simplifies complex scientific knowledge into a concise descriptive fact (van der Linden et al. 2017), represents the consistent judgment of domain experts regarding a phenomenon (Kahan et al. 2011). Unlike direct factual information (e.g., “water treatment technology has achieved to reduce the level of colony count to less than 500 CFU/mL” (Chen et al. 2022)), consensus messaging does not provide detailed technical explanations of the issue itself. Instead, it signals expert agreement, which plays a crucial role in fostering public support and enhancing acceptance - particularly in contexts, such as recycled water. Communicating perceived scientific consensus helps bridge attitudinal divides by positively shaping public beliefs and reinforcing support for actions on controversial issues (Linden et al. 2017). Research has shown that when individuals do not engage in detailed cognitive elaboration, they tend to rely on consensus messages to form judgments about complex social issues (Lewandowsky et al. 2012b). For instance, a study on climate change showed that scientific consensus messaging significantly improved public perceptions of scientific agreements (van der Linden et al., 2017). Similarly, in the domain of recycled water, there is broad academic consensus on several key issues. Many studies have identified wastewater reclamation as an essential solution to both current and future water scarcity (Tortajada and van Rensburg, 2020; Maier et al. 2022). Public concerns about recycled water safety often center on the effectiveness of treatment technology and the adequacy of regulation. Scientific evidence shows that modern water treatment technologies have already succeeded in producing water that meets drinking standards (Fielding et al. 2019; Yao et al. 2021). Therefore, scientific consensus in recycled water research represents a strategic and evidence-based communication approach aimed at increasing public acceptance by emphasizing the expert agreement on its safety and necessity.

Misinformation

Misinformation refers to information that contradicts scientific consensus or objective factual reality (Ecker et al. 2022), encompassing false, erroneous, or scientifically unsubstantiated claims (van der Linden, 2022). A more specific subset, disinformation, refers to fabricated or manipulated content designed to mislead or cause harm (Ecker et al. 2022). In this study, we adopt “misinformation” as a broader umbrella term encompassing any ultimately false information, while reserving the term “disinformation” for instances of intentional deception.

The inherent uncertainty and provisional nature of science, along with the limitations of scientific consensus (Kreps and Kriner, 2020), can facilitate the fabrication of seemingly creditable misinformation. Such misinformation can create erroneous beliefs about controversial issues, undermining the credibility and authority of scientific consensus and influencing public attitudes (van der Linden, 2019). In controversial domains, such as vaccines or climate change, one of the most effective misinformation strategies is to cast doubts on scientific consensus (Maertens et al. 2020). This includes tactics, such as false media coverage and the use of fake expert accounts (where individuals without relevant expertise or credentials falsely claim themselves as scientific authorities in a given field) (Cook et al. 2016; Kortenkamp and Basten, 2015). In the era of algorithmically mediated environments, fabricated yet seemingly professional misinformation exerts considerable influence. Its prevalence across online and social media platforms has significantly amplified its harmful societal impacts (Shin, 2024). For examples, misinformation, such as “vaccine trial participants died after inoculating COVID-19 vaccine”, spread rapidly and widely on social media, resulting in immediate and dramatic declines in public acceptance of COVID-19 vaccines (Loomba et al. 2021).

Misinformation is not only easily propagated through social media interactions (e.g., likes, reposts, and comments) but is also often perceived as credible and disseminated by uninformed users (Van der Linden and Roozenbeek, 2021). According to the information deficit model, individuals with greater familiarity with a topic tend to exhibit more positive attitudes and are more likely to accept information uncritically (Miller, 2004). Furthermore, studies on disinformation highlight the concept of “echo chambers” (Diaz Ruiz and Nilsson, 2022), which are cognitive environments where individuals are repeatedly exposed to information that aligns with their preexisting beliefs, thereby reinforcing their convictions and driving groups toward more extreme positions (Diaz Ruiz and Nilsson, 2022). Because information credibility significantly affects acceptance (Purcell et al. 2023), public resistance toward recycled water may partly reflect the effects of misinformation-driven “echo chambers”. These phenomena may involve psychological biases or entrenched stereotypes that resist change. However, the impact of misinformation on public attitudes toward recycled water remain largely unexplored. Given this gap, it is essential to explore resistance against misinformation in recycled water domain and to identify effective intervention strategies to mitigate its influence.

According to the cultural cognition thesis, conveying scientific agreement on contested societal issues may intensify attitudinal divides rather than reduce them (Kahan et al. 2011). Efforts to correct misinformation can also backfire, especially when the corrections conflict with the prevailing worldview of a polarized group (Ecker and Ang, 2019). Once established, initial beliefs formed through exposure to misinformation can persist, even after the misinformation has been corrected (Walter and Tukachinsky, 2019). In brief, traditional “debunking” approaches often prove ineffective in correcting misinformation (Walter and Tukachinsky, 2019). In contrast, a more effective strategy is “inoculation” or “prebunking”, which involves building resistance to misinformation prior to exposure, thereby preemptively mitigating its impact (van der Linden, 2022). This approach will be discussed in detail in the following section.

Prebunking: Inoculation theory

Inoculation theory (McGuire, 1964) suggests that, similar to how immune system triggers the generation of protective antibodies to combat viruses, individuals can build resistance to persuasive information attacks by prior exposure individuals to weakened persuasive information. This early exposure stimulates the development of cognitive-attitudinal “antibodies”, enabling individuals to resist stronger persuasive information in the future (van der Linden, 2022). Inoculation interventions have proven effective in countering misinformation, even in the context of AI-generated misinformation related to public health issues. Research indicates that inoculation messages can activate individuals’ vigilance toward misinformation, strengthening their resistance and ultimately reducing acceptance of misinformation (Shin, 2023).

The inoculation messages typically consist of two key components, forewarning of an impending attitudinal attack and refutational pre-emption (Amazeen et al. 2022). Forewarning serves as a cognitive cue that alerts recipients to forthcoming attempts to challenge their existing views and beliefs (Amazeen et al. 2022). Refutational pre-emption provides specific rebuttals to anticipated counterarguments, equipping individuals with preemptive tools to defend their attitudes (Kyle et al. 1997). This strategy presents weakened counter-messages alongside pre-constructed rebuttals that can be deployed against future persuasive attacks (van der Linden, 2022). The combination of forewarning and refutational pre-emption can create a psychological sense of threat, reinforcing resistence. Specifically, refutational pre-emption can trigger “intrinsic threat”, while forewarning generates “extrinsic threat”, together enhancing attitudinal defense (McGuire, 1964). In an experimental study, Pfau et al. (Pfau, 1997) incorporated forewarning and refutational pre-emption as moderating variables, confirming that both components are effective in promoting attitudinal immunization. A more recent research using both refutational pre-emption and forewarning as indicators of inoculation message strength validated the immunization effects of both components (Shin and Akhtar, 2024). Specially, it demonstrates that the immunization effect of refutational pre-emption was stronger than forewarning (Shin and Akhtar, 2024). While some scholars suggest forewarning is a crucial step in triggering threat (Richards and Banas, 2018), others argue it may not be a key factor in forming attitudinal resistance (Banas and Rains, 2010). Given this lack of consensus, employing a combined inoculation strategy is advisable. Therefore, this study adopts a “forewarning + refutational pre-emption” paradigm, which has been empirically validated as a robust approach to achieving attitudinal immunization (Lewandowsky and van der Linden, 2021; van der Linden et al. 2017).

From a cognitive perspective, individuals process information through dual-process theory, which divides thinking processes into two cognitive processes (Moravec et al. 2019). We adopt the terminology commonly used by previous studies (Stanovich, 1999; Kahneman, 2011), referring to System 1 and System 2. System 1 is intuitive, rapid, unconscious, and autonomous, relying on emotions, memories, and experiences to make quick judgments (Dennis and Minas, 2018); System 2, by contrast, is analytical, slower, and abstract, relying on logical reasoning within cognitive constraints (Dennis and Minas, 2018). Integrating these perspectives, Shin and Akhtar (Shin and Akhtar, 2024) proposed the heuristic-systematic (HS) processes model of inoculation which extends the inoculation theory by showing how both systems contribute to attitudinal resistance. This model suggests that when engaging System 1 thinking, individuals rely on heuristic cues or simple decision rules to form attitudinal resistance; while in System 2, individuals systematically evaluate inoculation messages to develop immunity against misinformation. Shin and Akhtar (Shin and Akhtar, 2024)’s study confirm that both heuristic and systematic cues can activate individuals’ cognitive immunity to misinformation, and that forewarning and refutational pre-emption - key components of inoculation messages - are processed through both systems, validating the dual-process basis of inoculation theory. However, a critical question remains: Do the immunization effects of inoculation messages manifest at the level of individuals’ cognitive processing when they are re-exposed to misinformation or contradictory messages? That is, while inoculation messages function as preemptive interventions, how do individuals cognitively respond when tater confronted with misleading or contradictory messages? Exploring this intrinsic mechanism may strengthen the explanatory power of dual-process theory within the framework of inoculation theory and provides further insights into the HS model of inoculation.

The internal validity of inoculation messages has been confirmed across multiple domains. Bither et al. (Bither et al. 1971) were among the first to apply inoculation theory to the field of marketing, demonstrating its ability to shape consumer attitudes. Subsequent studies have proved that inoculation messages play a crucial role in protecting public attitudinal beliefs (Banas and Rains, 2010) in contexts ranging from marketing and corporate advertising (Banas and Rains, 2010), media dissemination (Burgoon et al. 2016), and political campaigns (Meirick, 2002). Over time, the scope of inoculation theory has expanded from straightforward truisms to more controversial topics. Scholars have applied the theory to areas, such as climate change communication (Maertens et al. 2020), news media resilience (Lewandowsky and van der Linden, 2021), hygiene and health awareness (van Der Linden et al. 2020), and AI-generated misinformation (Shin 2024). These applications demonstrate the theory’s efficacy as a social intervention tool capable of producing large-scale resistance to misinformation in various fields. However, water-related topics - especially recycled water - presents unique challenges compared to other subjects. As water is essential to life, the public tends to be particularly sensitive to health-related risks associated with its use (Fu et al. 2022). Negative information related to water quality or safety tends to be quickly noticed, rapidly disseminated, and amplified through public discourse. Consequently, it remains uncertain whether inoculation theory can effectively counter misinformation in this domain. Although one study attempted to apply inoculation theory to recycled water perceptions, it failed to demonstrate a measurable inoculation effect, possibly due to the recency effect - when more recent messages (e.g., misinformation) overshadow prior inoculation efforts (Kemp et al. 2012a). Another study examined the therapeutic benefits of general inoculations versus specific inoculations (Amazeen et al. 2022), yet the precise cognitive mechanisms driving the differences remain poorly understood. To address this uncertainty, it is essential to examine not only whether inoculation can generate resistance against misinformation in the recycled water domain, but also how it does so. By exploring the cognitive mechanisms that underlie attitudinal resistance, this study seeks to determine whether inoculation theory can be applied effectively to an issue marked by high public sensitivity and scientific controversy. Understanding this process will contribute to the design of more effective, targeted intervention strategies, ensuring that misinformation related to recycled water safety can be countered and mitigated more systematically and successfully.

Theoretical framework and research hypotheses

Previous research has predominately examined information reception effects via questionnaire-based studies (Fielding and Roiko, 2014), including investigations into the effectiveness of inoculation message interventions (Basol et al. 2021; Maertens et al. 2020). However, such methods have inherited limitations as they often failed to capture intuitive reactions and authentic cognitive responses when individuals are first exposed to new information. Eye-tracking technology - a tool tooted in cognitive neuroscience - can effectively address these issues by providing real-time, objective measures of attention and cognitive engagement during information exposure (Rayner, 1998).

According to the elaboration likelihood model (ELM), attention is a strong indicator of information processing intensity. Information is processed through a central and more effortful route, where the degree of central processing reflects an individual’s information reception effects (Yu et al. 2022). Greater central processing enhances information reception effects, leading to stronger persuasive effects on belief change (Allen et al. 2013). In complex information processing tasks, such as reading, attention is strongly linked to fixation (Rayner, 1998) -the act of keeping eyes steady on an object (Skaramagkas et al. 2023). The eye-mind hypothesis suggests that fixation is directly connected to cognitive information processing, meaning individuals simultaneously browse and engage in information stimuli (Just and Carpenter, 1976). Specifically, fixation duration and fixation count can serve as two primary oculomotor indicators to measure cognitive information processing (Rahal and Fiedler, 2019; Atalay et al. 2012), representing their reception effects to informational stimuli (Pickering and Gambi, 2018). In this context, fixation duration (the amount of time the eyes remain focused on a stimulus) reflects the cognitive effort invested in processing the information. Longer fixation duration indicate higher information reception effects (Fu et al. 2020). Fixation count (the number of times the eyes fixate on the stimulus) reflects the depth of cognitive engagement. A higher fixation count signifies greater information processing (Rayner, 1998). Based on these, this study employs eye-tracking indicators - fixation duration and fixation count - to quantify individual cognitive information processing and measure information reception effects (evidenced by longer fixation duration and higher fixation counts) exert greater influence on belief formation and change.

As previously stated, the dual-processing model distinguishes cognitive information processing into two systems. System 1 operates continuously and automatically, generating rapid conclusions after processing new information. Intuition primarily largely relies on System 1 (Dennis and Minas, 2018). When individuals encounter new information, they first unconsciously activate System 1 to search past memories for familiar patterns and form judgments rapidly - typically within less than one second (Moravec et al. 2019). This process is unintentional and unavoidable. If the new information aligns with prior beliefs, System 1 generates a “feeling of rightness (FOR)” (Bago and De Neys, 2017). Conversely, if the new information conflicts with prior beliefs, FOR generates a sense of incongruity (Bago and De Neys, 2017), which can trigger System 2 - a deliberate, effortful, and slower cognitive system - to resolve the conflict through deeper cognitive processing (Kahneman, 2011). Generally, negative stimuli, threat, surprises, or the sense of incongruity from FOR can trigger System 2 (Moravec et al. 2019). Because System 2 requires cognitive effort, individuals generally minimize its use and it is only engaged when individuals are motivated or when cognitive dissonance cannot be ignored (Kahneman, 2011; Taylor and Fiske, 1978). Eye-tracking research provides key insights into how individuals process information within these two systems. Shorter fixation duration and fewer fixation counts are indicative of System 1 processing, where individuals engage in quick, automatic evaluation of information (van der Gijp et al. 2017). By contrast, longer fixation duration and higher fixation counts reflect System 2 engagement, signifying deeper, analytical information processing (Moravec et al. 2019).

Scientific consensus messaging about recycled water has been shown to enhance public acceptance by improving understanding and perceived credibility (van der Linden et al. 2019). However, influential misinformation can neutralize or even reverse these positive effects (Walter and Tukachinsky, 2019). Despite corrections being provided, the continued influence of misinformation may persist, affecting individuals’ thinking - a phenomenon known as the continued influence effect (Lewandowsky et al. 2012a). Studies on climate change have demonstrated that the positive effects of scientific consensus can be undermined or even entirely invalidated by misinformation (van der Linden et al. 2017), especially when both claims appear mutually exclusive. In such cases, System 1 may struggle to resolve the conflict, leading to cognitive dissonance (Moravec et al. 2019). In response, individuals may either ignore the cognitive dissonance (Nickerson, 1998) or activate System 2 to invest more time and cognitive efforts to assess which claim is more credible (Kahneman, 2011; Moravec et al. 2019). Notably, people tend to trust familiar information by default (Pennycook et al. 2018), an automatic System 1 response. Moreover, people exhibit heightened sensitivity toward negative information, tending to pay more attention on it and reinforce this experience over time (Baumeister et al. 2001). This phenomenon, known as “negative preference”, reflects a strong cognitive tendency to prioritize negative information (Bebbington et al. 2017).

Negative information generally undergoes more extensive cognitive processing than positive information. In decision-making contexts, negative information is often perceived as a more reliable indicator of quality, resulting in greater visual fixation (Yu et al. 2022). The processing preference suggests heightened visual attention and cognitive effort toward negative content (Skaramagkas et al. 2023). Based on dual-process theory, we hypothesize that when individuals receive only scientific consensus messaging, the information is primarily processed via System 1 (Moravec et al. 2019), leading to lower information reception effects. However, as exposure to scientific consensus increases, information familiarity builds over time, which in turn leads to higher trust levels (Miller, 2004), thereby enhancing acceptance of recycled water. In contrast, when individuals are sequentially exposed to both scientific consensus and misinformation, the contradictory nature of the information induces cognitive dissonance, triggering System 2 engagement for deeper cognitive evaluation (Nickerson, 1998). Due to negativity preference, individuals demonstrate greater information reception effects when processing misinformation, ultimately leading to lower acceptance of recycled water. Accordingly, we propose the following hypotheses:

H1a: Scientific consensus, when presented alone, has a greater positive effect on public acceptance of recycled water compared to the combined presentation of scientific consensus and misinformation;

H1b: Misinformation elicits higher reception effects than scientific consensus, thereby leading to negative effects on the acceptance of recycled water.

Inoculation messages function as a protective mechanism against misinformation (van der Linden, 2019). Previous studies on inoculation theory consistently found that when people are exposed to persuasive information attacks, those who receive only one-sided information are more significantly influenced by opposing viewpoint. In contrast, individuals who receive both positive and negative information tend to exhibit greater resistance, with their attitudes remaining relatively stable despite subsequent exposure to conflicting messages (McGuire, 1961). Based on these foundations, this study hypothesizes that individuals exposed only to scientific consensus about recycled water would be more negatively affected by subsequent exposure to misinformation. Conversely, those who are first presented with both scientific consensus and inoculation messages would exhibit greater resistance to misinformation, thereby maintaining more favorable attitudes toward recycled water.

Inoculation messages influence the reception of information and, in turn, impact attitudes toward the acceptance of recycled water. When individuals identify with a message, their cognitive system instinctively defends it against conflicting messages. According to confirmatory bias theory, people tend to actively seek, favor, and process information that aligns with their preexisting beliefs, while disregarding information that contradicts them (Nickerson, 1998). Exposure to both scientific consensus and inoculation messages strengthens emotional and cognitive connection to the endorsed viewpoint. Consequently, when later confronted with misinformation, they tend to dismiss it, reducing its influence on their beliefs. This selective avoidance leads to lower cognitive engagement, resulting in reduced visual attention to the contradictory information (Rayner, 1998).

From an information processing perspective, inoculation messages contain forewarning and refutational pre-emption, equipping individuals with counterarguments and rebuttals to defend against opposing arguments (Lewandowsky and van der Linden, 2021). This dual structure introduces s sense of threat and presents two conflicting information types, both of which are well-established triggers of System 2 processing (Moravec et al. 2019). Therefore, inoculation messages motivate individuals to engage in System 2, facilitating deeper cognitive evaluation of information. Hence, we hypothesize that scientific consensus establishes an initial belief framework (Lewandowsky et al. 2012b), followed by the introduction of inoculation messages, which aim to activate System 2 for critical evaluation. Individuals then process information systematically, leading to belief reinforcement aligned with their initial stance. As a result, when misinformation is subsequently introduced, the cognitive conclusion has already been formed through System 2, reducing the likelihood of reactivating System 2 due to cognitive dissonance. Instead, System 1 automatically dismisses the conflicting information, leading to cognitive avoidance of misinformation. This reduced engagement with misinformation ultimately protects the individual’s existing viewpoint from persuasive attacks. Accordingly, we propose the following hypotheses.

H2a: Compared to the combination of scientific consensus and misinformation, the addition of inoculation (i.e., scientific consensus + inoculation messages + misinformation) exerts a greater positive effect on public acceptance of recycled water;

H2b: Inoculation messages reduce the reception effects of misinformation, thereby positively influencing public acceptance of recycled water.

Many stereotypes and false prejudices arise from hasty reasoning due to limited and one-sided information. Therefore, eliminating prejudice requires exposure to more comprehensive and balanced information (Kunda, 1990). Inoculation messages, which present both counter-arguments and their refutations, offer a two-sided informational approach that promotes critical thinking and resistance to persuasion. According to the motivated reasoning hypothesis, further reasoning - including information collection, belief construction, and decision-making - can be influenced by biased perceptions (Druckman and McGrath, 2019). Motivation refers to the cognitive preferences and mental cues that guide information-seeking behavior and influence subsequent reasoning (Kunda, 1990). These cognitive processes determines which information individuals prioritize and how they further regulate their evaluations and behavior (Frey, 1986).

Scientific consensus can function as a motivational trigger, activating System 2 and encouraging individuals to engage in deep thinking and reasoning when subsequently exposed to inoculation messages, including misinformation. When scientific consensus aligns with individuals’ prior motivation, they are more likely to prioritize it, investing greater cognitive efforts in processing it. This heightened engagement leads to stronger information reception (Rayner, 1998), ultimately shaping cognitive conclusions and behavioral decisions that align with their initial motivation. However, when scientific consensus is presented repeatedly in isolation, it can unintentionally provoke psychological resistance, potentially triggering the “backfire effect” - wherein individuals defend their preexisting beliefs more strongly when repeatedly exposed to same information (Swire-Thompson et al. 2020). Therefore, we propose the following hypotheses.

H2c: Compared to presenting scientific consensus alone, the addition of inoculation (i.e., scientific consensus + inoculation messages + scientific consensus) exerts a greater positive effect on public acceptance of recycled water;

H2d: Inoculation messages have a positive effect on the reception of scientific consensus, leading to a positive effect on the acceptance of recycled water.

Methods

Based on the literature review, this study identifies three key information types - scientific consensus, misinformation, and inoculation messages - that may affect public acceptance of recycled water. Additionally, to establish a blank control and ensure the consistency in experimental materials across different groups, an irrelevant placebo has been introduced as a fourth information type. To simulate real-world scenarios where individuals encounter multiple sources of complex information, the four information types are combined into four sequential message block types. These experimental conditions are outlined in Table 1.

Table 1 Overview of experimental conditions.

Two experimental studies were conducted to test the proposed hypotheses. Study 1 adopts MESSAGE BLOCK TYPE 1 and MESSAGE BLOCK TYPE 2 to explore the effectiveness of scientific consensus and misinformation in shaping public attitudes toward recycled water in the absence of inoculation messages. This serves as a baseline condition for Study 2. The theoretical framework for Study 1 is presented in Fig. 1. Study 2 introduces inoculation messages, utilizing MESSAGE BLOCK TYPE 3 and MESSAGE BLOCK TYPE 4 to form a between-groups comparison with Study 1. The objective is to investigate the mechanism through which inoculation messages interact with scientific consensus and misinformation to influence public acceptance of recycled water, and examine whether inoculation messages exert an immunization effect, mitigating the impact of scientific consensus and misinformation on public acceptance. The theoretical framework for Study 2 is presented in Fig. 2. To evaluate these effects, this study employs a questionnaire survey to measure changes in participants’ acceptance of recycled water. Additionally, eye-tracking technology is utilized to capture eye movement data, providing objective insights into participants’ information reception effect.

Fig. 1

The framework of Study 1.

Fig. 2

The Hypothesis in Study 2.

Our use of eye-tracking methodology to investigate the cognitive processes underlying inoculation interventions is grounded on the assumption that fixation within the foveal region of eyes or sustained visual focus on an object reflects active cognitive processing of information stimuli (Meißner and Oll, 2019). To quantify this cognitive engagement, we define the information reception effect as the extent to which participants cognitively process information, measured by fixation duration and fixation count within areas of interest (AOI) - the specific regions of stimulus material on which participants’ gaze is concentrated (Skaramagkas et al. 2023). Fixation duration consists of two key measures: average fixation duration and total fixation duration (Rayner, 1998). Longer fixation duration and higher fixation counts indicate stronger information reception effect within an AOI, reflecting deeper cognitive engagement with the presented information.

Stimuli materials

To ensure that the scientific consensus materials are both representative and scientifically valid, we adopt a structured approach, following methods used in previous studies (van der Linden et al. 2017). The materials are presented in a standardized format, such as “xx% of field experts conclude that xxx”, and incorporate excerpts from authoritative literature on recycled water to ensure that the content is both accessible and scientifically grounded. To simulate misinformation in a realistic manner, trending statements about recycled water are collected from social media platforms, with a specific focus on health-related misinformation, as prior studies show that the public is particularly sensitive to negative messages concerning health (Bass et al. 2022). To enhance perceived credibility, the misinformation is deliberately framed to mimic the structure of scientific consensus. For example: “Several journalists jointly stated that using recycled water for landscaping could lead to infections and diseases in individuals exposed to the vapor from the recycled water.” The inoculation messages consist of two critical components: (a) a forewarning alerting participants that their views may be challenged; and (b) refutational pre-emption containing a counterargument followed by a refutation. For example, participants first receive a forewarning: “Non-internal medicine doctors and nurses used misleading strategies to convince the public that recycled water is harmful to human health.” The forewarning builds on previous findings about emotional threat effect (van der Linden et al. 2017; Amazeen et al. 2022). This claim is then debunked through explanation, highlighting the fallacies in the misleading argument (Ecker et al. 2022): “However, internal medicine doctors and scientists indicate that only drinking recycled water that does not meet national drinking water standards can be harmful to human health.” The development of these refutations follows guidelines by Ecker et al. (Ecker et al. 2022), ensuring that inoculation messages effectively foster cognitive resistance to misinformation. Finally, placebo materials are designed using neutral, unrelated content (e.g., “Math is xxx”) to serve as a control stimuli without affecting participants’ attitudes toward recycled water.

All stimulus materials are presented in the form of “Recycled Water Tips” images, designed using Photoshop. A total of 12 tips were created, containing six scientific consensus, three misinformation, three inoculation messages, and three placebo messages. The text areas within each image serve as AOI for eye-tracking analysis. To enhance participant immersion, each tip features the slogan “Recycled Water Tips” and a background designed thematically aligned with recycled water. To ensure stimuli uniformity, all images follow the same layouts, differing only in text content while maintaining consistent background, composition, and text formatting across all message types. The word count is proportionally balanced across message types, with an average of 40 words for scientific consensus, 40 words for misinformation, 81 words each for inoculation and placebo messages. The stimuli materials are provided in the Supplementary Material.

Participants

To ensure that early stage inoculation interventions address future decision-making dynamics, careful consideration was given to participant selection. This study focuses on current undergraduate and graduate students, who represent an ideal demographic for early inoculation efforts. In the next 10 to 20 years, this group will become key decision-makers shaping policies, consumption patterns, and implementation strategies related to recycled water.

The eye-tracking experiment in Study 1 and Study 2 followed a completely between-group design, where each participant was exposed to only one message block type. The experiment in Study 1 was conducted from 2 March to 29 March 2023, involving 108 participants (50 males and 58 females) with normal vision, recruited from Xi’an University of Architecture and Technology. The experiment in Study 2 was conducted from 30 March to 30 April 2023, involving 107 participants (51 males; 56 females) with normal vision, recruited from the same university. All participants were between 18 and 25 years old. To ensure the validity of eye movement data, the following data cleaning procedures were applied. First, samples with excessive eye blinks resulting that sampling rates below 80% were removed (Yu et al. 2022). Second, participant data were also excluded if the average total fixation duration within the AOIs of each scientific consensus/misinformation stimuli material was less than 1 second or abnormally high (defined as a natural logarithm of the total fixation duration exceeding sample mean) (van der Lans et al. 2021). After data cleaning, 99 valid samples remained in Study 1 and 98 valid samples in Study 2. The removal rate of samples falls within acceptable limits as established by previous eye-tracking research (Atalay et al. 2012).

Procedure

This experiment utilized the Tobii Pro Fusion eye-tracking device to collect eye movement data. The display resolution was set to 1920 × 1080 with a 100% scaling ratio to maintain visual consistency across all trials. The eye-tracking device was connected to a laptop through a Type-C port, and Tobii Pro Lab software - compatible with the eye tracker - was used to manage all experimental tasks, including stimulus programming, eye movement data acquisition, and visualization. Additionally, to ensure standardized viewing conditions, the distance between the screen and participants’ eyes was controlled at approximate 60 cm during the experiment.

To ensure that the stimulus material effectively influenced participants’ attitude change and eye movement behavior in the formal experiment, a pre-testing phase involving 32 participants was conducted over three days prior to the formal experiment. Research indicates that a sample size of 30 participants is sufficient for a pilot study (Shin and Akhtar, 2024). The pre-testing procedure followed the exact same procedure as the formal experiment, thereby validating the expected effects of the stimulus materials under consistent conditions. The formal experiment followed the procedure outlined below:

Before the eye-tracking experiment began, participants received introductory information to enhance the effectiveness of the information stimuli. This included: a) A brief 30-second introduction of recycled water, providing participants with basic background knowledge about recycled water; b) A description of the experimental procedure, explaining that the experiment would last approximately 6 min and involve viewing 9 information pictures. Participants were instructed to read the presented materials carefully and follow the on-screen prompts throughout the procedure; c) An explanation of experimental safety, followed by presentation of a “Consent Form for Participation”, which outlines the study’s purpose, potential benefits and risks of participating in the study, future use of data, and participants’ rights to withdraw at any time without any penalty. Upon obtaining informed consent, participants were taken to the eye-tracking laboratory, which was kept quiet and free of distractions.

Prior to the eye-tracking experiment, participants were asked to complete a questionnaire measuring their acceptance of recycled water (pre-test). Then, a pre-experiment was conducted with the participants to familiarize them with picture switching operations during the main experiment.

Subsequently, the main eye-tracking experiment was conducted. The experimental procedure is shown in Fig. 3. Participants were randomly assigned to one of the message block types and subsequently exposed to 9 informational pictures. In Study 1, the presentation sequence was Scientific consensus → Placebo → Misinformation / Scientific consensus. In Study 2, the presentation order was Scientific consensus → Inoculation messages → Misinformation / Scientific consensus. To ensure controlled conditions, the three scientific consensus stimuli were presented in the same order and with identical content across all groups. Similarly, the final presentation of scientific consensus stimuli was kept identical within MESSAGE BLOCK TYPE 1 and 3, and within MESSAGE BLOCK TYPE 2 and 4, respectively. During the experiment, each picture was displayed individually. Participants viewed all three pictures within a message block before progressing to the next block. Each picture was viewed only once during the experiment. The presentation time for scientific consensus and misinformation stimuli was not fixed, allowing participants to advance by pressing the space key after reading the material. However, the presentation time of placebo and inoculation messages stimuli was fixed to a duration of 60 seconds. Before each picture was presented, a calibration page featuring a cross-shaped logo (“+”) was shown in the center of screen for 800 milliseconds to ensure precision in the eye-tracking calibration. The eye-tracking experiment concluded after participants had viewed all 9 pictures sequentially.

Fig. 3

Stimulation Procedure of Eye-tracking Experiment.

Following the eye-tracking session, participants completed the questionnaire again (post-test), which was identical in content to the pre-test questionnaire but presented in a different item order. At the end of the experiment, all misinformation presented during the experiment was debunked, and participants were given corrected information to support accurate public understanding. To acknowledge their participation, each participant received a small gift worth approximately $2.70. The entire experiment lasted approximately 20 min.

The study obtained ethics approval from the Research Review Board of Laboratory of Neuromanagement in Engineering, Xi’an University of Architecture and Technology. All procedures involving human participants adhered strictly to the principles outlined in the Declaration of Helsinki, ensuring the experiments were non-invasive and had no harm to participants. Participants’ personal information was kept confidential, and they were advised of the data protection measures prior to the commencement of the experiments.

The questionnaire assessing public acceptance of recycled water was designed based on previous studies (Portman et al. 2022). The full questionnaire is provided in Supplementary Note. The information reception effect was measured using three key eye-tracking indices: average fixation duration, total fixation duration, and fixation count in the text area (AOI). To quantify these effects, eye movement data were drawn from the final three scientific consensus stimuli in MESSAGE BLOCK TYPE 1 and MESSAGE BLOCK TYPE 3 to represent the information reception effect under those conditions. Similarly, eye movement data from the final three misinformation stimuli in MESSAGE BLOCK TYPE 2 and MESSAGE BLOCK TYPE 4 were used to represent the information reception effect under those respective conditions.

Data handling

Eye movement data were collected using Tobii Pro Lab software and processed with SPSS 26.0 in this study. To test whether differences between experimental groups are statistically significant, analysis of covariance (ANCOVA) was employed to assess the main effects across conditions. Additionally, structural equation modeling (PLS-SEM) was performed using Smart-PLS 2.0 to test the theoretical model proposed in this study. PLS-SEM is known for its greater predictive power compared to Covariance-based SEM (CB-SEM) which primarily focuses on explanation. Unlike CB-SEM, PLS-SEM not only enhances predictive accuracy but also contributes to theoretical interpretation (Guenther et al. 2023). In addition, PLS-SEM demonstrates strengths in handling complex multi-indicator models, making it well-suited for studies requiring sophisticated data structures (Hair et al. 2019).

Given the relatively small sample sizes in this study, PLS-SEM was deemed the most appropriate analytical method. It is generally suitable for sample sizes ranging from 24 to 100, provided that the sample size exceeds 10 times the number of structural items (Tenenhaus et al. 2005). This study meets these criteria, with 99 valid samples in Study 1 and 98 valid samples in Study 2, while both models contain 5 constructs. To ensure robust statistical inference, bootstrapping with 5000 resamplings was applied to test the statistical significance of path coefficients. In addition, the reliability of both the pre-test and post-test questionnaire items was confirmed by Cronbach’s alpha values of 0.912 and 0.974, respectively, indicating high internal consistency of the measurement instruments.

Results

The direct impact of scientific consensus and misinformation on the acceptance of recycled water

In Study 1, paired sample t-tests were conducted to assess whether a significant difference existed in the acceptance of recycled water before and after participants were exposed to the informational stimulus. The results, presented in Table 2, indicate a clear impact of both scientific consensus and misinformation on attitudinal change. As expected, exposure to scientific consensus led to a significant increase in participants’ acceptance of recycled water (Mdiff = 0.455, SD = 0.708, p < 0.001). However, when misinformation was introduced in MESSAGE BLOCK TYPE 2, it had a significant negative effect on acceptance (Mdiff = −1.017, SD = 1.188, p < 0.050).

Table 2 Descriptive overview of mean (pre-post) differences in public acceptance of recycled water.

To further test whether the differences between the two groups were statistically significant, an analysis of covariance (ANCOVA) was conducted, with the post-test scores as the dependent variable and the pre-test scores as the covariate. The ANCOVA revealed a significant main effect for each group, confirming that the acceptance of recycled water varies significantly by message block type (F(1,96) = 53.638, p < 0.001). Specifically, the post-test mean for MESSAGE BLOCK TYPE 1 (M = 5.704) was significantly higher than that of MESSAGE BLOCK TYPE 2 (M = 4.303), with a mean difference of 1.401 (P < 0.001).

In summary, the results indicate that the positive effect of scientific consensus on the acceptance of recycled water was significantly diminished by the introduction of misinformation. Specifically, participants exposed only to scientific consensus messaging (MESSAGE BLOCK TYPE 1) demonstrated a significantly greater acceptance compared to those exposed to both scientific consensus and misinformation (MESSAGE BLOCK TYPE 2), thereby providing empirical support for Hypothesis H1a.

The indirect effect of scientific consensus and misinformation on the acceptance of recycled water through information reception effects

SEM analysis was performed in Study 1 to examine the indirect effect of scientific consensus and misinformation on the acceptance of recycled water, with the acceptance of recycled water as the dependent variable while message group type as the independent variable. MESSAGE BLOCK TYPE 1 (scientific consensus + placebo + scientific consensus) was set as the reference category (dummy variable: 0), and MESSAGE BLOCK TYPE 2 (scientific consensus + placebo + misinformation) as the comparison category (dummy variable: 1).

The results from 5000 bootstrapping iterations show that MESSAGE BLOCK TYPE 2 (scientific consensus + placebo + misinformation) had significant positive effects on the information reception effects (B = 0.787, SE = 0.029, p < 0.001). This indicates that participants exhibited stronger information reception effects with misinformation compared to scientific consensus in MESSAGE BLOCK TYPE 1 (scientific consensus + placebo + scientific consensus). In line with the findings from the eye-tracking and questionnaire data, these results indicate that misinformation stimuli attracted more attention and cognitive processing, thereby amplifying their influence. Moreover, the analysis reveals that the information reception effects had a significant negative effect on the acceptance of recycled water (B = −0.319, SE = 0.134, p < 0.05). In other words, greater cognitive engagement with misinformation was associated with reduced acceptance of recycled water. Accordingly, Hypothesis H1b was supported. As illustrated in Fig. 4, the structural model maintained strong explanatory and predictive power, with R² = 0.608 and Q² = 0.327.

Fig. 4

SEM Results in Study1.

The direct impact of scientific consensus and misinformation on the acceptance of recycled water under inoculation condition

As in Study 1, paired sample t-tests were conducted in Study 2. The results, presented in Table 3, indicate that the combined presentation of scientific consensus and inoculation messages led to a significant increase in acceptance of recycled water (Mdiff = 0.904, SD = 0.575, p < 0.001). Similarly, the combined presentation of scientific consensus, inoculation messages, and misinformation also resulted in a positive change in acceptance of recycled water (Mdiff =0.384, SD = 0.772, p < 0.001). These findings suggest that inoculation messages effectively mitigated the negative impact of misinformation on public attitudes toward recycled water.

Table 3 Descriptive overview of mean (pre-post) differences in public acceptance of recycled water.

To compare the results of Study 2 with those of Study 1, an analysis of covariance (ANCOVA) was conducted using the data from both studies. The ANCOVA used post-test scores as the dependent variable and pre-test scores as the covariate. The analysis revealed a significant main effect across four experimental conditions, indicating statistically significant differences in acceptance level of recycled water between message block types (F(3,192) = 53.894, P < 0.001). Further analysis demonstrated that the post-test mean for MESSAGE BLOCK TYPE 4 was 1.460 points higher than that for MESSAGE BLOCK TYPE 2 (M4 = 5.834, M2 = 4.374, P < 0.001). Similarly, the post-test mean for MESSAGE BLOCK TYPE 3 was 0.547 points higher than that for MESSAGE BLOCK TYPE 1 (M3 = 6.308, M1 = 5.761, P < 0.05). Moreover, the post-test mean for MESSAGE BLOCK TYPE 3 was 0.474 points higher than that for MESSAGE BLOCK TYPE 4 (M3 = 6.308, M4 = 5.834, P < 0.001). These results indicate that MESSAGE BLOCK TYPE 3 had the most significant effect on changing public acceptance of recycled water among the four message block types. MESSAGE BLOCK TYPE 3 produced a greater positive effect on acceptance than MESSAGE BLOCK TYPE 1, while MESSAGE BLOCK TYPE 4 showed a stronger positive effect than MESSAGE BLOCK TYPE 2. Therefore, Hypotheses H2a and H2c were supported.

The indirect effect of inoculation messages on the acceptance of recycled water through their impact on the reception of scientific consensus and misinformation

In Study 2, SEM analysis was conducted to examine the indirect effect of inoculation messages on the acceptance of recycled water, with the acceptance of recycled water as the dependent variable and message block type as the independent variable. The first SEM model analyzed data collected from MESSAGE BLOCK TYPE 4 and MESSAGE BLOCK TYPE 2. In this analysis, MESSAGE BLOCK TYPE 2 (scientific consensus + placebo + misinformation) was set as the dummy variable. The SEM results indicate that message block types (MESSAGE BLOCK TYPE 2 = 0; MESSAGE BLOCK TYPE 4 = 1) had significant negative effects on information reception effects (B = −0.836, SE = 0.022, p < 0.001). Specifically, compared to the combined presentation of scientific consensus and misinformation (MESSAGE BLOCK TYPE 2), the addition of inoculation messages (MESSAGE BLOCK TYPE 4) significantly reduced information reception effects. This suggests that inoculation messages had negative effects on the information reception effects of misinformation, thereby reducing its influence.

Moreover, the results revealed that the information reception effects of misinformation negatively affected the acceptance of recycled water (B = −0.209, SE = 0.101, p < 0.050); as shown in Fig. 5a. These findings support the conclusion that inoculation messages diminish the information reception effects of misinformation, thereby exerting an indirect positive effect on public acceptance of recycled water. Accordingly, Hypothesis H2b was supported. The structural model maintained strong fit and predictive relevance (R² = 0.605, Q² = 0.358).

Fig. 5

SEM Results in Study2.

The second SEM analysis was performed using data from MESSAGE BLOCK TYPE 3 and MESSAGE BLOCK TYPE 1. In this analysis, MESSAGE BLOCK TYPE 1 (scientific consensus + placebo + scientific consensus) was set as the dummy variable. The results showed that message block types (MESSAGE BLOCK TYPE 1 = 0; MESSAGE BLOCK TYPE 3 = 1) had significant positive effects on the information reception effects (B = 0.636, SE = 0.053, p < 0.001). Specifically, compared to scientific consensus alone (MESSAGE BLOCK TYPE 1), the inclusion of inoculation messages (MESSAGE BLOCK TYPE 3) led to significantly higher information reception effects, indicating that inoculation messages had positive effects on the reception of scientific consensus.

Furthermore, the results revealed that the information reception effects of scientific consensus positively affected the acceptance of recycled water (B = 0.816, SE = 0.041, p < 0.001); as shown in Fig. 5b. These results that inoculation messages have a positive effect on the information reception effects of scientific consensus, which, in turn, positively influences public acceptance of recycled water. Therefore, Hypothesis H2d was supported. The structural model maintained strong fit and predictive relevance (R² = 0.826, Q² = 0.311).

The interaction effect of scientific consensus, misinformation, placebo, and inoculation messages on the acceptance of recycled water

The analysis further revealed a significant interaction effect among scientific consensus, misinformation, placebo, and inoculation messages (F(1,193) = 15.663, p < 0.001). As shown in Fig. 6, inoculation messages significantly mitigated the differences between the effects of scientific consensus and misinformation on the acceptance of recycled water (F(1,193) = 9.336, p < 0.01), whereas placebo messages amplifying this difference (F(1,193) = 75.398, p < 0.001). These results indicate that inoculation messaging effectively moderated the influence of misinformation, thereby enhancing the resilience of public acceptance of recycled water.

Fig. 6

The interaction effect of message types on the public acceptance of recycled water.

Discussion

The findings from Study 1 indicate that while scientific consensus positively influences public acceptance of recycled water, this effect is significantly diminished when misinformation is introduced. In other words, the beneficial impact of scientific consensus is neutralized in the presence of misinformation. To conceptualize these effects, we define the effect of scientific consensus on the acceptance of recycled water as A, and the effect of misinformation as -B. The experimental results, showing the pattern A + A > A-B, logically imply that A < B. Therefore, we conclude that the negative effect of misinformation on the acceptance of recycled water is greater than the positive effect of scientific consensus alone.

The eye movement data in Study 1 further supports this conclusion. Participants exhibited significantly higher level of information reception effect on the recycled water when exposed to misinformation compared to scientific consensus. Given that information reception effect reflects the degree of information processing (Rayner 1998), a lower information reception effect suggests reliance on System 1 (automatic, heuristic-based thinking), while a higher information reception effect indicates the engagement of System 2 (deliberate, analytical thinking) (van der Gijp et al., 2017; Purcell et al. 2023). This pattern suggests that the public is more inclined to engage cognitively with misinformation rather than scientific consensus. The underlying mechanism of this preference may be rooted in differences in cognitive processing. When scientific consensus is presented alone, individuals lack motivation to activate System 2 processing, leading individuals to rely on quicker, less effortful System 1 judgments. In contrast, when individuals encounter both scientific consensus and misinformation, the conflicting information inputs and negative stimuli provide sufficient motivations for them to activate System 2 to process information more deeply. This cognitive engagement manifests as longer fixation duration and higher information reception effects.

These findings provide empirical evidence for the cognitive processing mechanisms underlying the public’s negative preference on misinformation about recycled water. By measuring visual fixation metrics, the study offers cognitive neuroscientific evidence for long-standing psychological assumptions about negative preference theory (Baumeister et al. 2001) in the field of recycled water acceptance. The analytical results of the SEM model in Study 1 validate the mediating effect of information reception effects, demonstrating scientific consensus and misinformation influence recycled water acceptance through information reception effects. Specifically, the negative effects of misinformation are mediated through a higher degree of information processing, highlighting the pivotal role of cognitive engagement in shaping public perceptions.

Study 1 extends previous research, which has primarily focused on the positive affects of scientific consensus on the acceptance of recycled water (Fielding and Roiko 2014; Price et al. 2015), by introducing misinformation as a critical influencing factor. Through a direct comparison of the influence effects of misinformation and scientific consensus, this study uncovers an essential cognitive processing mechanisms underlying the influence of misinformation about recycled water: unlike scientific consensus, misinformation possesses an inherent capacity to activate System 2 processing, prompting individuals to engage in deeper cognitive evaluations. Moreover, Study 1 validates the negative preference theory from the perspective of cognitive neuroscience, uncovering how negative information about recycled water is processed more deeply than neutral or positive information. In doing so, it contributes novel insights into the cognitive mechanisms that underlie negative preference in the domain of recycled water acceptance, thereby expanding the theoretical scope of negativity bias research.

Since this study did not examine corrective interventions or post-misinformation correction data, it cannot directly test the mechanisms underlying the continued influence effect of misinformation. However, based on the validated cognitive mechanism associated with negative preference for misinformation, it is reasonable to hypothesize that the continued influence effect of misinformation may stem from erroneous conclusions formed through System 1 or System 2 processing, which often exert a greater impact than factual corrections (Moravec et al. 2019). The results of Study 1 highlight that the public’s negative preference for misinformation, demonstrating its potential to substantially reduce public acceptance of recycled water. Given the often complex and persuasive nature of real-world misinformation, further investigation into intervention strategies to mitigating the influence of recycled water misinformation is urgently needed.

The findings of Study 2 demonstrate that inoculation messages significantly increased public acceptance of recycled water, with participants exposed to inoculation exhibiting higher acceptance levels than those in the non-inoculated groups. Consistent with previous studies on inoculation theory across various research fields (Lewandowsky and van der Linden 2021; Bolsen and Druckman 2015), these results affirm that inoculation messages were effective in increasing individuals’ positive attitudes toward certain events. Additionally, we found that inoculation messages function as an information immunization mechanism, effectively increasing participants’ acceptance of recycled water by decreasing the information reception effects of misinformation and increasing the information reception effects of scientific consensus. This interpretation is supported by three key findings from Study 2: (a) lower information reception effects - indicative of System 1 processing - were observed for misinformation in the inoculation group (MESSAGE BLOCK TYPE 4); (b) higher information reception effects - indicative of System 2 processing - were observed for scientific consensus in the inoculation group (MESSAGE BLOCK TYPE 3); and (c) the mediating role of information reception effects in shaping acceptance was successfully validated in Study 2. Together these findings support the cognitive mechanism proposed in our initial assumptions. The following elaborates on these three key findings, interpreted through the lens of dual-process theory.

(a) The inoculation messages, containing forewarning (extrinsic threat) and refutational pre-emption (intrinsic threat) (Amazeen et al. 2022), present individuals with conflicting or threatening information, thereby prompting individuals to trigger System 2 to invest more cognitive efforts into critical evaluation, which leads to formed conclusions by weighing new information against their prior cognition (Bago and De Neys, 2017; Kahneman, 2011). Subsequently, when misinformation is reencountered, the inoculated individuals experience less cognitive dissonance or a sense of threat. Instead, they rely on System 1 to swiftly dismiss misinformation, as reflected by lower information reception effects the misinformation and higher acceptance of recycled water in the inoculation group (MESSAGE BLOCK TYPE 4) compared to the no-inoculation group (MESSAGE BLOCK TYPE 2). The eye movement data further support this interpretation, demonstrating that inoculation messages reduce cognitive engagement with misinformation;

(b) Conversely, when participants encounter scientific consensus, the prior System 2-driven engagement induced by inoculation messages reinforce motivated reasoning (Moravec et al. 2019). This facilitates deeper cognitive processing of information aligned with prior cognition (Carpenter 2019). The observed higher degree of information processing (information reception effects) for scientific consensus validates the reinforcing role of inoculation messages in scientific consensus processing.

(c) Further analysis in Study 2 demonstrates a significant interaction effect between inoculation messages, scientific consensus and misinformation. Inoculation messages can simultaneously restrain cognitive processing of misinformation while strengthening motivated reasoning on scientific consensus. Inoculation messages are able to introduce a perceived threat, create prior attitude disposition, and guide individuals toward defensive information processing (Amazeen et al. 2022). When people process information defensively, they tend to accept information that is consistent with their preferred attitudes and reject information that contradicts them (Amazeen et al. 2022). Thus, inoculation messages about recycled water plays a positive role in narrowing the attitudinal gap divided by misinformation and scientific consensus, thereby improving public acceptance of recycled water.

Study 2 expands the application of inoculation theory by confirming the internal validity of inoculation theory in the context of recycled water. Consistent with the results of previous inoculation studies across various fields (Maertens et al. 2020; van der Linden, 2022), the results demonstrate that inoculation messages related to recycled water can effectively mitigate the influence of misinformation. While there are studies suggesting that inoculation theory failed to influence recycled water attitudes, attributing to a strong recency effect generated from the final defense messages overriding the inoculation effect (Kemp et al. 2012b), this study, in contrast, found that the inoculation effect was significant, regardless of whether the final message presented was misinformation or scientific consensus. These discrepancies may stem from prior studies employing suboptimal inoculation message designs that failed to evoke the necessary cognitive responses for the inoculation effect to take hold.

Previous studies have also highlighted the strong influence of prior beliefs on the immunization effect of inoculation messages (Amazeen et al. 2022). To account for this key factor, the present study exposed all experimental groups to scientific consensus, thereby helping to equalize participants’ baseline attitudes. The results of this study further confirm that combined presentation of scientific consensus and inoculation messages consistently exert an immunization effect, irrespective of participants’ prior attitudes toward recycled water, leading to a measurable positive shift in recycled water acceptance. Accordingly, this approach provides a systematic strategy for mitigating the influence of uncontrollable prior attitude biases. It also demonstrates that scientific consensus followed by inoculation messages represents an effective strategy for fostering resistance to misinformation.

Furthermore, Study 2 expands the theoretical framework of inoculation theory (McGuire, 1964) by identifying its intrinsic mechanism - achieving immunization effects through the modulation of information reception effects on other message types (scientific consensus and misinformation). It also complements the heuristi-systematic (HS) inoculation model proposed by Shin (Shin, 2024), which posits that inoculation intervention operates within a dual-processing framework. According to this model, both System 1 and System 2 can trigger individuals to process inoculation messages, thereby contributing to individual’ resistance to misinformation (Shin, 2024). In line with this, and with a specific focus on recycled water, this study applies dual-processing theory to explain the immunization effect of inoculation messages on individuals’ subsequent exposures to scientific consensus and misinformation. The findings reveal how prior exposure to inoculation messages shapes the way people engage with subsequent persuasive messages (influential scientific consensus and misinformation), thereby reinforcing the theoretical integration of inoculation theory with dual-processing models in the context of recycled water. Moreover, this study extends the HS model by providing empirical evidence of the back-end cognitive effects of information-based immunization processes.

In summary, this study makes several theoretical contributions:

First, by introducing three distinct information types - scientific consensus, misinformation, and inoculation messages - this study extends previous research on informational influences related to recycled water (Price et al. 2015; Fielding and Roiko, 2014). It provides both practical guidance and theoretical insights into how various information combinations can be leveraged to reduce public misconceptions and negative stereotypes regarding recycled water;

Second, while previous studies have relied on questionnaires to assess information validity and the immunization effect of inoculation information (Maertens et al. 2021; Lewandowsky and van der Linden, 2021), such self-reported measures are inherently subjective, prone to response bias, and unable to capture participants’ immediate cognitive responses to the information stimuli. Moreover, questionnaires measuring information validity are particularly susceptible to manipulation by personal biases (Rayner, 1998). This study addresses these limitations by employing eye-tracking methodology, which offers a more objective, real-time measure of cognitive processing. Given the strong correlation between information processing and eye movement behavior (Skaramagkas et al. 2023), eye-tracking techniques provide precise insights into how individuals process information at the cognitive level. This methodological advancement allows for a more accurate assessment of how inoculation messages operate within the framework of dual-processing theory, reinforcing their role in shaping resistance to misinformation and enhancing scientific consensus processing;

Third, this study provides empirical support for negative preference theory from a cognitive neuroscience perspective. Using visual fixation metrics, it reveals that misinformation - unlike scientific consensus - trigger System 2 engagement, resulting in higher information reception effects. These findings help explain why misinformation often elicits stronger public responses and support theoretical accounts of the disproportionate influence of negative information (Baumeister et al. 2001);

Fourth, this study advances previous researches by validating the effectiveness of inoculation messages in mitigating the impact of misinformation within the context of recycled water. It reveals the cognitive processing underlying the immunizing effect of inoculation messages through the application of dual-processing theory - showing that such messages simultaneously restrain the cognitive processing of misinformation while strengthening motivated reasoning in favor of scientific consensus. Furthermore, this study identifies key interaction effects between inoculation messages, scientific consensus and misinformation related to recycled water, thereby expanding the research scope of inoculation theory. By addressing previously underexplored cognitive processing mechanism of inoculation theory, this study fills gaps in the literature by providing new insights into how inoculation messages can bridge attitude divergences created by misinformation and scientific consensus, thereby promoting public acceptance of recycled water.

This study also makes empirical contributions by providing practical guidelines for stakeholders involved in recycled water initiatives.

First, the findings of this study confirm a negative preference for recycled water misinformation, demonstrating its strong negative impact on public attitudes. Therefore, public policymakers and relevant regulatory authorities should provide timely and authoritative corrections of misinformation related to recycled water, particularly when it circulates on social media platforms, to mitigate its influence effectively;

Second, the study demonstrates that recycled water inoculation messages are effective in both immunizing against misinformation and strengthening motivated reasoning in favor of scientific consensus. Based on this, we recommend that public communication strategies related to recycled water should proactively incorporate inoculation messages. Such preemptive communication can help shape public attitudes toward recycled water before misinformation is encountered, increasing resilience to deceptive information;

Third, regarding the design of inoculation messages related to recycled water, this study proves an effective inoculation structure that consists of “forewarning + refutational pre-emption”. However, considering the ethical dilemma associated with strong forewarnings (Shin and Akhtar, 2024), we recommend using a mild forewarning approach. For instance, the statement used in this study - “group uses misleading strategy” - presents a mild threat but is sufficient to activate System 2 cognitive processing without triggering a strong defensive reaction or a perception of belief infringement;

Fourth, with respect to the design of refutational pre-emption, this study suggests revealing the techniques used to mislead or elucidate the fallacies embedded in misinformation. Presenting a weakened form of misinformation while inoculating the public against common misinformation and manipulation tactics can significantly enhance resistance to deceptive narratives;

Moreover, this study highlights that sequential exposure to scientific consensus and inoculation messages helps neutralize the effects of prior attitudes on the inoculation interventions. Therefore, we suggest that communication campaigns adopt a “scientific consensus + inoculation messages” format when disseminating information about recycled water, ensuring that the public is sufficiently inoculated against misinformation before encountering deceptive claims.

Collectively, these suggestions aim at enhancing public understanding of recycled water, promoting the adoption of sustainable behavioral norms, and reducing stereotypes related to recycled water. The findings have significant implications for the design of global marketing campaigns related to recycled water, as well as the planning, implementation, and operation of recycled water initiatives and policies.

Research limitations and directions for future research

Like any study, our study has several research limitations. First, the sample is relatively restricted, as it exclusively comprises university students. As a result, the influence of other demographic variables was not examined, and the generalizability of the experimental findings to broader populations remains untested. Future research could include a more diverse sample, such as participants from different regions, age groups, and educational backgrounds, to better understand variations in public acceptance of recycled water across demographic contexts. Nevertheless, it is important to note that, given the current global challenges related to water resources, the effective utilization of recycled water is a growing trend. While early attitudinal inoculation is critical for shaping public perceptions, there appears to be no significant time lag in inoculation effects among university students, who are likely to become key decision-makers in recycled water adoption and policy within the next 10 to 20 years.

Second, although efforts were made to simulate realistic information environments, the study’s ability to fully replicate real-world conditions remains limited. Future research could build upon these findings by conducting experiments in a more ecological valid settings or by employing more advanced technological tools to simulate real-world scenarios. Such tools may include virtual reality (VR) equipment (Wu et al. 2021), electroencephalography (EEG) (Cheng et al. 2022), or functional magnetic resonance imaging (fMRI) (Yu et al. 2023). These technologies have the potential to provide deeper insights into public cognitive responses to recycled water across diverse contexts.

Finally, previous studies have shown that algorithmic inoculation is more effective than non-algorithmic (Shin and Akhtar, 2024). While the present study focuses solely on verifying the immunization effect of non-algorithmic inoculation, it provides novel perspectives and contributes valuable research insights into the cognitive mechanisms underlying resistance to misinformation. Building on this contribution, future studies could explore the integration of inoculation theory with human-algorithmic interaction approaches, assessing how algorithmic interventions can further enhance public resistance to misinformation about recycled water.

It is also worth noting that marketing campaigns for recycled water should not be viewed merely as persuasive tools to convince the public to support recycled water initiatives through one-way and monotonous information output. Instead, such campaigns should serve as interactive platforms that facilitate information-sharing, establish feedback mechanisms, and cultivate critical thinking skills to empower individuals in navigating information in complex informational environments effectively.