3. Harmful expert testimony

Understanding mediators’ responsibility in cases of harmful expert interviews requires two main moves: illustrating how expert testimony can harm audiences epistemically and providing an account of mediators’ obligations toward the public. This section addresses the former issue, while the next focuses on mediators’ epistemic duties.

Expert testimony can harm lay audiences in several ways. First, experts may propagate misinformation or make unjustified claims because they have gone rogue (Jäger Reference Jäger2024; Croce and Marsili Reference Croce and Marsili2025) or through genuine errors while acting in good faith (Croce Reference Croce, Brossel, Eder and Grundmannforthcoming). Andrew Wakefield infamously exemplifies a rogue expert for his attempt to establish a causal link between the MMR vaccine and autism. Regarding genuine errors, during the COVID-19 pandemic, for instance, expert pronouncements about mask ineffectiveness and transmission modes often created public confusion and undermined trust in scientific guidance (Williams et al. Reference Williams, Hooker and Gilbert2023).

Second, experts may abuse their epistemic authority by giving testimony qua experts outside their domain of competence without qualifying their claimsFootnote ⁷ – which philosophers consider to be problematic instances of epistemic trespassing (Gerken Reference Gerken2018; Ballantyne Reference Ballantyne2019; DiPaolo Reference DiPaolo2022). Such testimony harms audiences when they are epistemically entitled to assign greater credibility to the expert than she deserves. Consider a pure mathematician invited to a TV show to discuss cryptocurrencies who is asked about trade policies and defends the importance of tariffs without specifying that she is no expert on the matter. If the audience cannot realize that the mathematician’s competence does not extend to economics, the trespassing testimony causes epistemic harm.Footnote ⁸

Third, public expert testimony can generate or amplify the effects of expert disagreement, negatively impacting public understanding of scientific matters. To mention but a few detrimental consequences (de Melo-Martín and Intemann Reference de Melo-Martín and Intemann2018: 3; De Cruz Reference De Cruz, Baghramian and Martini2022), publicly displayed expert disagreement may lead to confusion and false beliefs about the current state of scientific knowledge or evidence supporting particular claims. Additionally, it can create a false impression that expert views are equally divided when most experts actually agree against a minority position, thereby fostering false balanced reporting (Gerken Reference Gerken2020; Terzian Reference Terzian2025). Finally, by forcing expert communities to invest time and energy to handle dissenting voices, it creates delayed debunking. Research on memory-based mechanisms such as the “continued influence effect” shows that once misinformation circulates, corrections – even if promptly provided – often fail to prevent its spread (Lewandowsky et al. Reference Lewandowsky, Ecker, Seifert, Schwarz and Cook2012). Hence, retracted misinformation can continue to shape beliefs long after a correction is issued.

These considerations illustrate how the harms of expert testimony can erode public trust in science and motivate the importance of countering them. What sorts of duties do mediators have to prevent this harm?

4. The right to information and the duty to inform

There is broad agreement within journalism ethics and social epistemology that journalists have a duty to inform the public (Vos Reference Vos, Mellado, Hellmueller and Donsbach2016). In journalism ethics, this duty directly attaches to the journalistic profession and encompasses truth-telling and harm minimization norms, alongside special responsibilities arising from journalists’ power to influence public opinion (Ward Reference Ward, Wahl-Jorgensen and Hanitzsch2020: 307).

Philosophical analysis offers deeper insight into the grounds of this duty. In Hohfeld (Reference Hohfeld and Cooke1919) rights theory, duties are the correlatives of rights: if someone has a right, that is a claim to demand something from another, then someone else has an obligation to act accordingly (Graham Reference Graham1996). Understanding the duty to inform within this framework requires introducing the correlative right to information: a person’s obligation to inform another stems from the other’s right to receive relevant information. This applies to doctors who must inform patients of the results of their blood tests, station clerks who are responsible for explaining monthly passes to their clients, and news reporters who must share daily news with their audience.

The justification of these correlative rights and duties varies across theoretical approaches. Non-instrumental approaches hold that the value expressed by rights is final rather than derivative: right-holders possess inherent value (Watson Reference Watson2021: 10–11) – hence, the label status-based views (Nagel Reference Nagel1995; Kamm Reference Kamm2007; Cruft Reference Cruft2010). An individual’s final value allows them to make certain claims over other people, who therefore have correlative obligations toward them. Watson’s recent work (Reference Watson2021) offers a non-instrumental account of epistemic rights and duties. In her theory, epistemic rights are complex entitlements that provide “justification for the performance and prohibition of actions and omissions concerning epistemic goods” as opposed to other kinds of goods (15, my italics). Epistemic rights constitute a distinct class within the broad realm of moral rights: if right-holders’ moral status justifies the possession of moral rights, the same status justifies the possession of epistemic rights. Hence, epistemic rights are as final and non-derivative as other moral rights. Within this framework, the duty to inform is an epistemic obligation protecting the right-holder’s claim by mandating specific actions (and omissions) concerning information. Mediators have this duty to the extent that their audience has a corresponding right to receive information from them.

Instrumental accounts offer a different justification of correlative rights and duties based on consequentialist considerations (Mill Reference Mill and Sher2002). Here, their value derives from some other good they promote: on contemporary instrumentalist theories like Raz’s (Reference Raz1995), rights are justified by the interests of the right-holder. Applied to the epistemic domain, epistemic rights and duties are justified because they promote our epistemic wellbeing, while the justification of moral rights and duties derives from their contribution to moral flourishing. This distinction echoes Lackey’s point about different sources of normative pressure (Section 2). The public’s right to information and mediators’ correlative duty to inform are justified by our epistemic interest in acquiring knowledge, which can exist independently from more basic interests. This correlative right and duty may also be justified by a distinct moral interest in acquiring knowledge, which is derivative from political concerns such as the need to be sufficiently informed to make political decisions (Raz Reference Raz1995; Buchanan Reference Buchanan2013) and the right to make autonomous democratic decisions like voting and demonstrating (Scanlon Reference Scanlon2013; Levitsky and Ziblatt Reference Levitsky and Ziblatt2018).Footnote ⁹

Two points deserve further attention. First, I remain neutral about the justification of these epistemic rights and duties: what matters for this paper is that mediators’ duty to inform is not merely a professional duty – or, a role ought, broadly speaking (Feldman Reference Feldman and Steup2001) – but also a genuine epistemic obligation.

Second, like rights, duties have both positive and negative dimensions. Incorporating the aforementioned remarks from journalism ethics into this framework, mediators’ duty to inform encompasses a positive obligation to provide truthful and newsworthy information and a negative obligation not to propagate fake news, bullshit, or other forms of misinformation.Footnote ¹⁰

This general characterization of the duty to inform manifests concretely in what information theorists and social epistemologists call gatekeeping. Social epistemologists (Greco Reference Greco2020, Croce Reference Croce2022, Dormandy and Grimley Reference Dormandy and Grimley2024) understand gatekeeping as the function of checking information quality and filtering out low-quality content to ensure only reliable information goes through community gates. Besides this concern with the epistemic quality of information – or news accuracy in journalism ethics’ terms (Ekström and Westlund Reference Ekström and Westlund2019) – information scholars consider gatekeeping to also include selectivity: the criteria through which journalists reduce potentially unlimited information to a finite selection of newsworthy content (Shoemaker and Vos, Reference Shoemaker and Vos2009: 1).Footnote ¹¹

Applying these considerations to public expert testimony, Mascitti et al. (Reference Mascitti, Ausili and Bina2025) define “epistemic gatekeeping” as the set of norms and methods allowing mediators to select and manage expert testimony in public debates. Drawing from their analysis, I distinguish between two gatekeeping norms: credentials monitoring – a quality assessment of which experts deserve to be heard, evaluating their authority to speak on particular topics – and content monitoring – an assessment of experts’ statements to ensure that accurate and relevant information reaches the public. These norms ground two potential answers to the central question of this paper. In the next sections, I explore whether mediators’ duty to inform in the context of expert interviews encompasses monitoring the credentials of expert interviewees and the content of their testimony.

5. The Viola case

To ground our analysis of mediators’ epistemic obligations in concrete practice, let us examine a recent case of expert testimony. In 2023, immunologist Antonella Viola, Professor of General Pathology at the University of Padua (Italy), co-authored a book on intermittent fasting with medical doctor Antonio Paoli, Professor of Exercise and Sport Sciences at the same institution. Following the book’s publication, several Italian scholars, including some of Viola’s colleagues at the University of Padua,Footnote ¹² publicly challenged claims made in the book. Critics argued that the authors overemphasized the need to avoid obesity and minimized the risks and side effects of intermittent fasting. One nutritionist explicitly accused Viola of sharing her personal experience with intermittent fasting as a marketing strategy and committing epistemic trespassing.Footnote ¹³

In January 2024, Viola was interviewed by Corriere della Sera, Italy’s most widely read newspaper, about intermittent fasting, semaglutide, diabetes, alcohol’s relationship with cancer, and longevity science. Here are two relevant exchanges between the journalist and Professor Viola:

The Viola case exemplifies potential harms of expert testimony discussed in Section 3. By presenting intermittent fasting – a debated topic in nutrition studies for which we still lack robust evidenceFootnote ¹⁵ – as settled science with clear benefits and minimal risks, the interview contributed to misinforming the readers. Predictably, it generated backlash from other experts who cited recent studies showing cardiovascular risks associated with intermittent fasting. This public display of expert disagreement created additional epistemic harm, as laypeople lack the time and competence to consult scientific studies and assess specialized disputes.Footnote ¹⁶

6. The credentials monitoring solution

What epistemic obligations did Viola’s interviewer have? The first approach to mediators’ epistemic obligations understands their gatekeeping duties in terms of credentials monitoring: interviewers ought to confirm that their expert interviewees possess the epistemic authority to address the topics at hand. This answer looks backward: when an expert harms the audience, the interviewer has already failed to protect them and hence to comply with their negative duty not to misinform the public.

The credentials monitoring solution draws support from the literature on the “credentials problem” (Cholbi Reference Cholbi2007) in the epistemology of expertise. How can laypeople identify experts in a domain if, by definition, they lack competence in such domain? Epistemologists have offered several strategies to address this problem following Alvin Goldman’s seminal insight (2001): while laypeople lack access to first-order evidence for expert opinions, they nonetheless may access indirect or second-order evidence of an expert’s trustworthiness – credentials or markers of expertise.Footnote ¹⁷

These considerations clarify the rationale behind the credentials monitoring solution: mediators may lack competence in the subject matters they discuss with experts, but this does not discharge them from ensuring that their interviewees possess relevant expertise. Much like laypeople seeking trustworthy sources of information, mediators should verify that potential interviewees meet standard expertise markers.

Among several accessible dimensions of expertise that mediators should assess, reliability and reputation play a central role. The most straightforward strategy to measure reliability is evaluating an expert’s track record – i.e., seeking evidence of past success and accurate predictions within their domain (Collins and Evans Reference Collins and Evans2007; Goldman Reference Goldman2001; Levy Reference Levy2022, Reference Levy2024; Grundmann Reference Grundmann, Lackey and McGlynn2025). For instance, knowing that a nutritionist has accurately predicted the effects of various diets in controlled studies provides evidence of their reliability. This marker has the advantage of not requiring detailed understanding of the reasons behind the expert’s success, only evidence of their performance.

Reputation, by contrast, measures whether peers consider an expert a reliable source (Goldman Reference Goldman2001) and includes multiple parameters. The most objective indicator is the alleged expert’s formal qualifications such as education degrees, certifications, awards, and institutional affiliations. Scientific metrics like citation-indexes and h-index also indicate expert reputation as they reflect peer recognition of work quality (Anderson Reference Anderson2011). Less formal parameters include work experience and committee appointments. Another key marker related to reputation is expert consensus – the fact that many scientists in a domain acknowledge someone as their peer. This marker is effective only insofar as the judgment about an expert’s standing comes from genuine experts independently assessing their colleague’s scientific merits (Goldman Reference Goldman2001: 101–103). Finally, mediators can monitor how fellow scientists respond to an expert’s public statements. This method typically works in the negative: public attempts by peers to correct an expert’s claims or, even more significantly, to signal that her testimony spreads misinformation may indicate that her reputation is questionable.

Together with indicators of character and professionalism (Anderson Reference Anderson2011), these markers support mediators’ gatekeeping efforts by filtering out potential interviewees lacking the competence to speak on the interview’s topic. A notable advantage of this account of mediators’ duty to inform is its consistency with a traditional understanding of journalism as guided by impartiality (Clayman and Heritage Reference Clayman and Heritage2002: 29). According to this norm, mediators’ professional obligations begin with thoroughly checking the expert’s credentials and determining that their voice deserves public attention, and end with providing experts the platform they need to share relevant knowledge. In short, mediators must let “the facts speak for themselves” (Munoz-Torres Reference Munoz-Torres2012: 571), treating expert testimony as one fact amongst many that should not be interpreted or manipulated.

In this approach, mediators share the epistemic duties protecting the public’s right to information with both the expert interviewee and the scientific community. The expert ought to inform the public to the best of their knowledge, and the scientific community is expected to intervene and rectify any false or controversial claims made by the interviewee. Incidentally, this is exactly what happened in the Viola case, with colleagues and other scientists challenging her claims about intermittent fasting, arguing that she minimized potential health risks and exaggerated benefits.

Upon closer inspection, however, the credentials monitoring approach falls short as a solution to the problem of harmful expert testimony. The first problem is that checking credentials has several limitations as a gatekeeping strategy. Track records may not be publicly available or easily accessible to mediators (Guerrero Reference Guerrero and Peels2017; Levy Reference Levy2022). Formal qualifications might help establish that someone is more competent than most people in a domain, but this might not suffice to determine whether they can authoritatively address the specific matters at issue in the interview (Grundmann Reference Grundmann, Lackey and McGlynn2025). (For instance, a cardiologist with excellent medical credentials may lack the specific expertise needed to comment on AI-powered cardiac diagnostic tools.) Institutions may encourage their researchers to cite one another to boost citations (McCook Reference McCook2017). Rogue experts typically increase their informal reputation by becoming celebrities within dissenter networks (Croce and Marsili Reference Croce and Marsili2025), and public sensitivity to expert disagreement may be weaponized for personal disputes (Croce Reference Croce, Brossel, Eder and Grundmannforthcoming).Footnote ¹⁸ Determining expert consensus is often complicated by fake experts whose opinions receive media coverage and dubious journals posing as legitimate sources (Brennan Reference Brennan2020; Levy Reference Levy2022: 113). Finally, mediators frequently fail to identify instances of epistemic trespassing and distinguish them from innocuous trespassing testimony regarding interdisciplinary questions (see footnote 9). All these limitations demonstrate the inherent difficulties mediators face when relying on credentials as a gatekeeping mechanism.

Beyond these limitations, a second and more fundamental problem undermines credentials monitoring as a strategy to counter the epistemic harms of public expert testimony. In a nutshell, even experts who brilliantly pass all credentials checks can harm their audience by propagating misinformation, presenting contested scientific views as settled, or generating public expert disagreement that erodes trust in science.

The Viola case vividly illustrates this problem. Her interview with Corriere della Sera brought about several epistemic harms. For starters, it misinformed readers by presenting intermittent fasting – a subject still debated in nutrition science – as settled science with clear benefits and minimal risks. This led nutrition experts and physicians to publicly object to her portrayal of intermittent fasting, thereby creating what appeared to the public as disagreement between equally qualified experts. Furthermore, the audience was exposed to her claim that intermittent fasting has “no contraindications unless you have specific medical conditions.” Many readers, particularly those with undiagnosed eating disorders, likely internalized this information and acted on it before encountering subsequent criticisms from other experts.

The Viola case shows that while credentials monitoring may be a necessary component of mediators’ due diligence, it is certainly insufficient to protect the public’s right to information from the harms of expert testimony. Even if mediators perform this function scrupulously, they cannot guarantee that genuine experts will provide testimony that is accurate, properly contextualized, and reflective of the current state of scientific evidence.

Thus, we must consider whether mediators’ epistemic duty to inform requires something more than mere credentials monitoring. If identifying trustworthy experts is not enough to prevent the public from being harmed by what experts say, does this duty also extend to monitoring and potentially challenging the content of expert testimony? The next section examines this possibility.

7. The interference solution

The second approach, which I label the interference solution, is grounded in what Clayman and Heritage (Reference Clayman and Heritage2002) call adversarialness: according to this journalistic norm, in order “to achieve factual accuracy and a balance of perspectives, journalists should actively challenge their sources, rather than being simply mouthpieces of ciphers for them” (29).

This approach holds that mediators’ epistemic and professional duties extend beyond credentials monitoring to include content monitoring. On this view, mediators should actively object to expert statements that they justifiably believe to be false, scientifically unsupported, or otherwise misleading. Unlike the backward-looking approach of credentials monitoring, this answer is primarily forward-looking: mediators must intervene during the interview itself to counter potentially harmful testimony before it reaches the public unquestioned.

Philosophical grounding for content monitoring can be found in Jennifer Lackey’s recent work on the duty to object. According to Lackey (Reference Lackey2020), the duty to object is an epistemic obligation targeting problematic statements – specifically, false, unwarranted, or harmful assertions. Objecting amounts to making assertions that aim to correct the record and contribute to conversational contexts. The intrinsic end of this duty is epistemic, concerning the protection or promotion of epistemic value – whether knowledge, evidence, or truth. Thus, this duty maintains its normative force even when an objection is unlikely to be accepted by the interlocutor (Reference Lackey2020: 36).

Lackey conceives the duty to object as a collective duty “where each member’s individual share of the responsibility is determined in large part by (i) the distribution of goods, especially social and epistemic statuses, and (ii) whether the other members of the collective do their share. The greater (lesser) the goods, the greater (lesser) the duty; the more (less) that some members do, the less (more) that other members have to do” (Reference Lackey2020: 53).

These conditions illuminate the imperfect nature of the duty to object: unlike perfect duties such as promise keeping demanding specific action (i.e., the duty to do X), imperfect duties allow discretion in fulfillment – latitude, in technical jargon (Schroeder Reference Schroeder2013) – typically understood in disjunctive terms (i.e., the duty to do X or Y or Z). Condition (i) specifies that the distribution of goods – including epistemic goods like authority, knowledge, and credibility – influences one’s responsibility to object. Those with greater epistemic and social standing have increased responsibility because their objections are more likely to be effective. Condition (ii) addresses how others’ actions affect one’s own duty. If several community members have already objected effectively to a false claim, my duty to do so diminishes. Conversely, if no one has objected, my responsibility increases.

At first glance, mediators’ responsibility in the context of expert interviews seems to fall within the scope of the duty to object: not only mediators are part of the relevant information community, but they also have professional obligations to protect the public’s right to information. However, the extent to which this duty applies to mediators interviewing experts depends on whether they satisfy conditions (i) and (ii) – hence, whether mediators have the appropriate social status and epistemic authority to object, and whether other members of the community have already intervened.

Mediators clearly satisfy condition (ii). As the first link in the transmission chain that brings expert testimony to the public, they are also the first audience exposed to potentially harmful testimony – whether live, recorded, or written. In practice, nobody has the chance to object before them, nor is anyone better positioned to do so from the standpoint of communication dynamics. As for condition (i), things are more complicated. One might argue that mediators lack the epistemic authority to challenge harmful expert testimony during interviews. On this view, the responsibility should fall instead to better epistemically positioned community members – such as fellow experts.

However, the fact that fellow experts are better positioned does not in itself show that mediators fail to meet condition (i). While experts have greater domain-specific authority than journalists, this doesn’t necessarily make them more effective at protecting the public from misinformation in the short term. As argued in §3, relying on post-hoc correction by the expert community increases the public display of expert disagreement and leads to delayed debunking. Thus, experts’ superior epistemic authority doesn’t guarantee their testimony will have a more positive epistemic effect on the audience than timely intervention by mediators.

Furthermore, condition (i) concerns one’s social status, and mediators have a professional duty qua information gatekeepers to protect the public’s right to information. This places them in a uniquely advantageous position to fulfill condition (i), even if they lack domain-specific expertise. Interestingly, Lackey admits that under certain conditions – such as obligations tied to professional roles – an imperfect duty to object may become perfect (51). Thus, although Lackey does not address journalistic obligations, and the nature of mediators’ duty to object may vary with context, the proponent of the interference solution can still argue that mediators are bound by this epistemic norm.

Applied to the Viola case, this view plausibly implies that the journalist had a perfect duty to challenge Viola’s claims about intermittent fasting. Letting misinformation circulate before the expert community can respond risks causing significant harm. For instance, when Viola claims there are “no contraindications” except in specific medical conditions, the journalist should object by noting that this is oversimplified and potentially misleading, as intermittent fasting may be inappropriate for many populations including pregnant women or individuals with a history of disordered eating. This diagnosis might not apply if Viola made the same testimony in a different setting – say, a live public event involving nutrition scientists. In such a scenario, the journalist’s duty to object would be imperfect and severely weakened by the presence of other experts under greater epistemic and professional pressure to challenge her testimony.

Despite the attempt to grant plausibility to the interference solution, this approach fails for three main reasons. First, it is epistemically overdemanding for mediators. Challenging misinformation from an expert during an interview requires fact-checking – something that takes time in written or recorded formats, and more importantly epistemic authority, particularly in live settings where mediators must verify complex scientific claims on the spot.

In our case study, suppose that Viola had stated that “intermittent fasting is safe for individuals with preexisting cardiovascular conditions under 50 years old.” Fact-checking this claim requires assessing whether there is a causal link between 8-hour eating windows and increased cardiovascular mortality, which is clearly too heavy a burden on the journalist’s shoulders. If the journalist already knew all the relevant facts, there would be no need to interview Viola. The purpose of expert interviews is precisely to give the public information it could not obtain from the mediator alone.

Second, this content monitoring norm is also impractical. Restricting science communication to those with scientific expertise in the relevant fields would require media outlets to rely on an army of highly specialized mediators (Fairbank Reference Fairbank2025: 7). This scenario could hardly be more distant from the reality of science communication: as widely argued (Goldman Reference Goldman2001; Figdor Reference Figdor2017; Gerken Reference Gerken2022; Fairbank Reference Fairbank2025), science journalists are often more akin to laypeople when it comes to reporting on specialized scientific issues.

Third, even in situations where a mediator might justifiably object to general, non-specialized aspects of an expert’s harmful claims, such objections could prove counterproductive for the public’s understanding. When a mediator challenges an expert’s claims, this creates what appears to be a non-peer disagreement between an expert and a journalist who has some background in the domain but presumably knows less than the expert. The audience would then be prima facie justified in granting more credibility to the expert, especially since the mediator selected them as an authoritative source in the first place.

The proponent of the interference answer might respond that a mediator’s objection still provides the audience with higher-order evidence against the expert’s words, given that mediators are expected to possess some basic understanding of the relevant domain. While I remain doubtful of this move’s epistemological soundness, evaluating its theoretical merits falls outside the scope of this paper. What matters for present purposes is that practical concerns persist. It is unrealistic to suppose that lay audiences could effectively assess the epistemic authority of a mediator who disagrees with their chosen expert source. If expert disagreement can erode public trust in science, non-peer disagreement between journalists and experts might prove equally problematic – albeit for different reasons.

If this analysis is correct, then the interference answer cannot deliver on its promise to protect the audience’s right to information through direct objections to harmful expert testimony. These considerations motivate searching for a different solution to the central problem of this paper – one that navigates between the problematic extremes of impartiality and adversarialness. Such an approach would satisfy three key desiderata:

1. 1. Alerting the public when expert testimony is epistemically problematic.

2. 2. Avoiding backfire effects such as the non-peer disagreement problem.

3. 3. Imposing feasible content gatekeeping measures on mediators.

In the next section, I propose an intermediate answer that fulfills these requirements through strategic questioning rather than direct objections.

8. The good questioning solution

The preceding analysis reveals significant limitations in both the credentials monitoring and interference approaches. The former fails to protect the public from harmful testimony even from genuine experts who pass all relevant checks, while the latter places unrealistic epistemic demands on mediators. The alternative solution I propose in this section is grounded in what Lani Watson (Reference Watson2021) calls “good questioning.” At its core, questioning amounts to eliciting information. Good questioning, according to Watson, involves eliciting worthwhile information competently. Two aspects of this definition require careful analysis.

First, worthwhile information is information whose content has epistemic significance for the audience. Depending on the specific features of the situation as well as the epistemic and practical needs of the audience, worthwhile information amounts to true information, justified information, or both. To borrow an expression from Zagzebski (Reference Zagzebski1996), it’s information that enhances our “cognitive contact with reality.”Footnote ¹⁹ Second, eliciting information competently concerns the questioning context. It involves selecting appropriate questions to ask a given source, deciding when to raise these questions during the interview, and how to articulate them effectively. Notess and Watson (Reference Notess, Watson, Fox and Saunders2023) identify two types of questions central to good questioning in the context of media interviews: expanding questions and contesting questions.

Expanding questions “draw out further details and points of clarification during the course of an interview” (Reference Notess, Watson, Fox and Saunders2023: 150) to get to the truth or improve the audience’s understanding of what is being said. For instance, suppose a mayor states during an interview that “the new urban development plan will increase affordable housing in our city.” An expanding question might be: “Could you specify what percentage increase in affordable housing units we can expect, and over what timeframe?” This question invites the official to provide concrete details that enhance the audience’s understanding of the plan’s impact.

Contesting questions “are designed to test the merits of the story” provided by the interviewee by soliciting evidential support or justification for the information conveyed (151). For example, if a finance minister claims that “our new tax policy will benefit all citizens,” a contesting question might be: “Independent economic analyses suggest that households earning below the median income might actually see a net increase in their tax burden. How do you respond to these assessments?” Such questions prompt the interviewee to address potential weaknesses in their claims. As these considerations suggest, the interviewer’s competence involves awareness that they raise questions on behalf of a silent but interested audience.

This account of good questioning provides the foundation for a novel approach to content gatekeeping in expert interviews. As a first step in the argument, consider how this concept applies to our specific context. In expert interviews, worthwhile information encompasses exoteric evidence that can be shared with a non-expert audience. Mediators’ questions should aim not only at disseminating significant scientific news, theories, or open problems to the public but also at providing accessible explanations of such matters.

Regarding eliciting information competently, the questioning context is highly relevant to expert interviews. To illustrate, consider the opening question in the Viola case: “Professor, once and for all, what is intermittent fasting?” By using “once and for all,” the journalist implies that Viola has the epistemic authority to give the final word on a disputed topic. This phrasing is epistemically harmful because it mistakenly signals to the audience that they should grant high credibility to a speaker who lacks the definitive epistemic authority suggested by the phrasing. A more appropriate opening question would be: “Professor, how would you describe intermittent fasting as it’s currently understood in scientific literature?” This alternative phrasing acknowledges Viola’s expertise while avoiding the implication that her explanation should be regarded as the final word on a contested topic. It also encourages her to situate her explanation within the broader scientific context.

The communicative elements of question-asking can also make a positive difference, as demonstrated elsewhere in the interview. When introducing the topic of alcohol consumption, about which Viola has maintained a position that conflicts with other scientists, the journalist asks: “We are in the middle of the holiday season; we have toasted and will toast more. You have long argued that alcohol is carcinogenic and that even the daily glass of wine considered harmless by many doctors should be eliminated. Your statements have sparked controversy. Is it really so?”

This question exemplifies good questioning in several ways. By explicitly acknowledging the controversy surrounding Viola’s position, the journalist provides relevant context to the audience and steers the reply toward providing supporting evidence. Indeed, Viola responds by introducing data from a British study and a WHO report and compares alcohol to smoking’s carcinogenic effects. This questioning strategy helps the audience understand not only Viola’s position but also the evidence behind it and its place within ongoing scientific discourse.

To further illustrate how good questioning could improve the Viola interview, consider her claim that intermittent fasting “has no contraindications, except in cases of specific medical conditions.” An effective follow-up question might be: “Could you elaborate on which medical conditions might make intermittent fasting inadvisable? Are there particular populations who should exercise caution?” This expanding question would require Viola to provide more specific information about limitations and risks, potentially benefiting audience members considering this dietary regimen. Similarly, when Viola asserts that intermittent fasting “works without major sacrifices,” a contesting question like “Several nutrition specialists and health researchers have pointed to challenges with patient adherence to intermittent fasting regimens. How would you address concerns about sustainability and compliance that have been raised in the field?” would allow the expert to qualify their own claims, preventing misleading oversimplifications and signaling to the audience that the matter may be more complex than initially presented.

To generalize, eliciting information competently in expert interviews involves deploying questioning techniques in service of the central epistemic goal: protecting the public’s right to information. As we have just seen, the interviewer’s questions should facilitate the audience’s assessment of the interviewee’s epistemic authority and make available evidence accessible to them.Footnote ²⁰

Having established the concept of good questioning and its application to expert interviews, I now turn to the second step in the argument: demonstrating that good questioning constitutes a sufficient means for mediators to fulfill their duty to inform. I argue that this approach meets the three desiderata identified in Section 7, providing a virtuous mean between the problematic extremes of credentials monitoring and objecting.

The communicative elements of question-asking already illustrate how a mediator’s questions can alert the public that something in the expert’s testimony warrants scrutiny, as required by the first desideratum. Well-crafted questions can signal that a claim is contested, that evidence may be inconclusive, or that certain qualifications are necessary – all without directly contradicting the expert.

Beyond this, good questioning addresses the second desideratum by avoiding backfire effects such as publicly displayed expert disagreement and delayed debunking. Unlike credentials monitoring, which leaves the burden of correction to the expert community (potentially leading to public disputes after information has already spread), good questioning enables mediators to contextualize and qualify expert testimony in real time. When an expert makes a potentially misleading claim, a mediator employing good questioning would immediately raise contesting questions that prompt the expert to address limitations or counterarguments – as I have just illustrated in the Viola case. Importantly, since these interventions take the form of questions rather than assertion aimed at correcting the record, they avoid creating the appearance of a non-peer disagreement between journalist and expert – a key advantage over the interference approach.

The good questioning approach also satisfies the third desideratum by imposing feasible measures on mediators. Unlike the interference approach, which would require mediators to possess sufficient domain-specific knowledge to directly challenge expert claims, good questioning requires competence in questioning strategies rather than comprehensive subject-matter expertise. This distinction is crucial. A mediator need not be an expert in nutrition science to ask Viola about the current state of research on intermittent fasting or to request clarification about at-risk populations. Such questions require some basic knowledge of the issue at stake – the kind of background information that a well-prepared journalist can reasonably possess, rather than the specialized knowledge needed to directly evaluate domain-specific scientific claims. The feasibility of this approach becomes even clearer when we consider that the skills of good questioning can be systematically taught and developed as part of professional training for mediators.

It’s important to acknowledge that this strategy is not foolproof. There is no guarantee that experts will offer accurate or balanced answers to even the most skillfully crafted questions. An expert determined to fuel a particular viewpoint might double down on misleading claims or evade probing questions. In such cases, good questioning alone cannot fully prevent the spread of misinformation. However, this limitation applies equally to the credentials monitoring and interference approaches. The former offers no recourse against misleading expert testimony, while the latter risks increasing confusion through non-peer disagreement.

In summary, the good questioning approach provides advantages over both alternatives. Unlike credentials monitoring, it enables mediators to actively prevent misinformation without waiting for post-hoc corrections from the expert community. Unlike interference, it avoids creating potentially counterproductive non-peer disagreements while imposing more realistic demands on mediators’ knowledge and skills. By meeting all three desiderata identified earlier, good questioning emerges as the most compelling approach to mediators’ duty to inform in the context of expert interviews.

9. Concluding remarks

This paper addressed the underexamined question of mediators’ epistemic obligations in expert interviews. I argued that their duty to inform extends beyond mere credentials monitoring to encompass good questioning – a strategy that enables them to fulfill their epistemic obligations without requiring domain-specific expertise or creating counterproductive non-peer disagreements.

The analysis has both theoretical and practical implications. Theoretically, it contributes to our understanding of distributed epistemic responsibility in public scientific discourse, showing how different actors’ obligations interlock to protect the public’s right to information. Practically, it provides guiding principles for journalists, anchorpersons, and other mediators conducting expert interviews, suggesting that competent questioning strategies – particularly expanding and contesting questions – can significantly enhance the epistemic quality of these exchanges.

The methodological significance of this work also lies in demonstrating that epistemologists should broaden their focus beyond experts themselves when examining science communication. While much philosophical literature has concentrated on experts’ obligations or the public’s vulnerabilities, this paper highlights mediators as crucial epistemic agents deserving of systematic normative analysis. This comprehensive approach to the epistemology of journalism recognizes that information quality depends not only on knowledge producers but also on knowledge brokers who shape how expertise reaches the public.

Moving forward, this framework can inform cross-disciplinary discussions about mediators’ epistemic responsibilities and how these relate to the public’s epistemic needs. Clarifying this relationship can help us develop more effective strategies for maintaining the integrity of expert testimony in public discourse – an increasingly vital task in today’s complex information landscape, where the boundaries between expertise, journalism, and public communication are constantly shifting.