Glass Hospitals: Transparency and Trustworthy Interpretation in Medical and Healthcare Expertise

1. Introduction

In their recent article “Expertise, Disagreement, and Trust in Vaccine Science and Policy,” Alberto Giubilini, Rachel Gur-Arie, and Euzebiusz Jamrozik (hereafter GG&J) argue that there is more to expertise than individual doctors’, nurses’, or other health professionals’ knowledge of their fields. They depart from veritist conceptions of expertise in terms of reliable access to truth¹ and align with more social conceptions of expertise² though epistemic considerations remain relevant. To be an expert is in some significant sense to be recognized as a credible authority. This requires more than skilled know-how: as GG&J see it, to be a credible authority requires trust. If that is right, then those occupying positions of medical and healthcare expertise should ask themselves not just “What do I really know?” but also “How can I be trusted?”³

When it comes to credible expertise, epistemic and moral dimensions of trust are both relevant and, for GG&J, inseparable. “Having trust in someone’s expertise typically requires trusting their commitment to relevant ethical principles which are essential both for acquiring knowledge and for putting knowledge into practice.”⁴ Among the relevant principles they identify for trusted experts are humility, honesty, and transparency.

I would agree that trust matters for expertise; I am especially compelled by the relational account that GG&J offer on trust in health expertise generally and vaccine policy specifically. Trustworthy experts should be worthy of the trust actually and potentially placed in them, worthy of specific entrusted objects (for example, the grounds of our beliefs, the safety of our children, etc.) valued by specifically situated epistemically dependent people. Of whose trust are experts worthy or not? Whose trust is being exploited; whose distrust is actually well founded? Perhaps most importantly, what can medical and health experts do to demonstrate continuing or renewed credible authority for specifically situated trusting or distrusting publics? As GG&J see it, the value of transparency is particularly relevant to this last question. They argue that trustworthy health experts must be not only scrupulous but also transparent in their decision-making, and they take recent examples of vaccine science and policy to confirm the importance of expert trust and transparency.

In what follows, I aim to affirm these authors’ linkage of trust and expertise by decoupling both from transparency. My claim is not that health experts should lie or deceive, but that articulating their credible authority in terms of transparency misframes things. We can see this in several ways: through the negative epistemic effects of a general norm of expert transparency, the importance of discretion for healthy trust, and the need for relationally responsive interpretation in how medical and health experts communicate with patients and publics across social-epistemic difference.

2. A Case for Expert Transparency

What must trusted experts be transparent about? We can begin to answer this question by attending to what sorts of failures of transparency – rather than failures of expert knowledge directly – GG&J take to undermine expert status and authority. The authors repeatedly identify transparency about scientific uncertainty, about ignorance, about expert disagreement, and about the value judgments applied in evaluating evidence.⁵ For example, GG&J discuss 2021 public health recommendations on childhood vaccination against COVID-19. Their view is not that the wrong decision was made, nor that it ran contrary to established scientific procedures, evidence, or majority expert opinion, but that such recommendations should have been accompanied by clear acknowledgement of the minority expert opposition and the value judgments at work in the decision.⁶

I would not disagree that uncertainty, ignorance, value judgments, and peer disagreement are all relevant factors when it comes to expert judgments and recommendations. But what does it mean for experts to be transparent about these things? GG&J repeatedly frame transparency in terms of acknowledgement: to be transparent about something is to acknowledge it. Perhaps the idea is that if experts acknowledge uncertainty, disagreements, and value judgments, trusting publics will not lose trust when they turn out to be mistaken. Perhaps the idea is that if experts don’t acknowledge these things, then they will lose public trust when they are wrong. Or perhaps we are meant to see a biconditional relationship: that health experts will not lose public trust and so lose expertise itself if and only if they are humble, honest, and transparent. “Expert status and expert authority are not necessarily undermined by lack of knowledge,” GG&J argue, “but they will be undermined by failures of transparency in the acknowledgement of scientific uncertainty, absence of knowledge, and expert disagreement about scientific knowledge.”⁷ The fact of an expert being wrong about the safety of a new vaccine may or may not threaten their perceived expertise, not so much due to insufficient knowledge itself as because of what it means for what trusting publics ascribe to them. If experts have consistently, clearly acknowledged uncertainty on the matter in question, the thinking goes, then if and when they get things wrong, trusting publics need not see it as a mark against their credibility. Experts are human, after all; so long as they do not pretend to be perfect, they need not lose our trust when reminded of their human fallibility.

One ambiguity about transparency understood in terms of acknowledgement is whether trusted experts must speak to the matters in question unprompted or only respond openly when (and if) asked. It is also unclear whether fulfilling a norm of transparency so understood means giving unfiltered access or active guidance to relevant information: as discussed below, these are not necessarily equivalent.

3. Transparency in the Sausage Factory

In “Epistemic Trust and the Ethics of Science Communication,” Stephen John pushes back against the norm that scientists should be transparent. John, much like GG&J, cares about how scientists and other experts communicate their ideas, not only how they generate and justify them. His argument here is based not in scientists’ right to privacy or the irrelevance of public access to scientific knowledge but the worry that transparency may produce more confusion than understanding. “Unfortunately, just as publicising the inner workings of sausage factories does not necessarily promote sausage sales, so, too, transparency about knowledge production does not necessarily promote the flow of true belief throughout the population (and so on for honesty, sincerity, and openness).”⁸

Good communication between scientists and the public matters, but transparency is too blunt an instrument to guide this interpretive work effectively. For one thing, John is pessimistic about the underlying assumption that nonexperts will consistently draw the right lesson if and when the curtain is pulled back on how an expert community comes to its conclusions. Why assume that a non-expert will have an accurate grasp of the internal social-epistemic dynamics of an expert field? John worries that widespread “folk” philosophies of science mislead outsiders to think they know what makes for a well-functioning scientific community regardless of the specifics. “For example, they might believe that the social structures of science should constantly encourage debate and discussion when, in fact, dogmatism is epistemically useful.”⁹

If, along with an important public health recommendation, experts also announce that there is outstanding expert disagreement on the matter, what sort of lesson are non-expert recipients of this announcement supposed to take? Is it the relatively narrow implication that this recommendation could be wrong but nevertheless should be followed? With John, one might ask: why we should assume that non-expert recipients will consistently give expert acknowledgements of uncertainty or expert disagreement the precise social-evidential weight that the medical / healthcare expert community intends for it to have?

Attitudes of trust can be warranted and robust, but also warranted and fragile, John reminds us. For fragile yet warranted trust, a default norm of expert transparency might not bolster but actively undercut it. Indeed, he cautions, “sometimes being transparent may be positively (epistemically) harmful to nonexperts. If we care about the promotion of true belief, we should not demand that scientists are transparent and open.”¹⁰

4. Trust and Discretion

One might think that John’s skepticism about transparency gives too much emphasis on successful transmission of true beliefs and not enough on maintaining good trust relationships. Consider then how a norm of transparency fares in terms of the ethics of trust. In “Trust and Anti-trust,” Annette Baier provides a relational, affective account of trust and trustworthiness. Many otherwise healthy, morally defensible relationships are not completely transparent, Baier reminds us. In relationships of interdependency, we must extend one another some discretion in how we deal with entrusted matters. Demanding transparency undercuts this discretion. Indeed, an insistence on transparency may be evidence that one doesn’t really trust the other after all. “One leaves others the opportunity to harm when one trusts, and also shows one’s confidence that they will not take it,” Baier says; rational trust requires “good grounds for such confidence in another’s good will, or at least the absence of good grounds for expecting their ill will or indifference.”¹¹

The discretion Baier is talking about here means that we are not only dependent but vulnerable in our trust relationships, a vulnerability that GG&J themselves emphasize in their argument for the importance of trust in expertise. How can we diagnose a betrayal of trust, then, if it is not about making mistakes or failing to be fully transparent? Baier is not naïve about trust relations, after all.

“There are immoral as well as moral trust relationships, and trustbusting can be a morally proper goal.”¹² If an expectation of transparency is at odds with trust as Baier describes it, what makes for an immoral trust relationship? It is not so simple as whether both parties see the other as trustworthy or not. The very discretion Baier pinpoints as the critical difference between trust and mere reliance brings with it vulnerability to betrayal that one may never see coming. She proposes instead an expressibility test: “to the extent that what the truster relies on for the continuance of the trust relation is something which, once realized by the truster, is likely to lead to (increased) abuse of trust, and eventually to destabilization and destruction of the relation, the trust is morally corrupt,” Baier argues. “A trust relationship is morally bad to the extent that either party relies on qualities in the other which would be weakened by the knowledge that the other relies on them.”¹³

Deception often but not always satisfies the conditions of Baier’s expressibility test. After all, lies and other forms of deception are often used to hide the truth from others, such that liars are relying on others’ ignorance of the truth for their trust relationship to continue on. Revealing a particularly cutting lie might well destabilize and destroy the relationship. By contrast, sometimes revealing a bit of deception does not threaten the relationship at all, as the one doing the deceiving depends on keeping it hidden for another purpose, not for the relationship to continue. Sometimes the deceived party would actually appreciate the liar’s efforts to keep a truth from them. We can similarly find cases where lack of transparency need not be a red flag for morally corrupt trust. A trusted person might not fully inform a trusting person of how they are attending to some entrusted object, yet not actually rely on the latter’s ignorance of these facts. Even when a trusted person is intentionally less than forthcoming, and is actively relying on a trusting person’s ignorance, they may be doing so to achieve another good, one the trusting person may well endorse and appreciate. Consider a surprise party, for example. Its planning depends on less than full transparency toward the trusting guest of honor. It will ruin the surprise if they learn about the party in advance – but it would not destabilize or destroy their relationship with those who kept the secret from them.

5. From Transparency to Trustworthy Interpretation

At this point some might reply that while Baier’s account of healthy trust and John’s account of science communication make sense elsewhere, they fall short when it comes to medical expertise specifically. I would argue, however, that relationally responsive expert interpretation rather than transparency is as important for medicine and healthcare as for any field of expertise.

Ironically, the metaphor of transparency can serve to obfuscate the interpretive work required in testifying effectively across epistemic difference. It suggests a plane of glass that lets outsiders simply see for themselves what is inside.¹⁴ The expert’s role so conceived is a passive one: do not hide anything, do not filter, step aside so the processes and results of your work show themselves. And this may well be appropriate when you have no reason to trust how I would choose to present my work to you, or at worst, reason to think I would do so misleadingly. But even then, conducting an unguided audit of expert work will not necessarily be a simple process. Transparency does not show us how to make sense of what has been made visible. By their nature, expert processes and results do not speak for themselves to any and all who would listen, clearly grasped for what they are if only the experts in question would be transparent. Even an outside review of untrusted expert work will need to call on trusted auditors with their own expertise to interpret what they find and the relevant implications for trusting nonexpert audiences.

If we think of medical and health experts as simply highly skilled, knowledgeable practitioners, perhaps we could see why a more passive duty of transparency might be less demanding and thus more palatable than an active duty of interpretation. Experts so characterized would be compelled to periodically open their books with no obligation to help anyone make sense of what they might find there. But I take it that a strength of how GG&J theorize expertise, authority, and trust is that it avoids this sort of picture. If what it means to have medical or health expertise is to be trusted as an authority on medical and health matters of public importance, then such experts must at least sometimes attend to how their work is taken up, grasped, and critically engaged by others. This is what I mean by relationally responsive expert trustworthiness,¹⁵ that trusted experts’ relationships to dependent others make a difference to them in how they pursue and present their work. Simply stepping to the side and allowing others to see for themselves fails to acknowledge the significant interpretive work needed for successful understanding and engagement of those occupying other social-epistemic locations. This is not to assume a one-way street of expert-public communication, a top-down transmission of established truth from experts to a passive general public. Nor do I have in mind a binary picture, with all experts on one side and the nonexpert public on the other. There are of course experts specializing in many different areas of medicine and healthcare, and likewise, different publics concerned with particular medical and health decisions. The interpretive work needed to communicate within and across these different social-epistemic positions is varied indeed.

Effective communication concerning individual medical decisions and general health policy recommendations is itself important work of what Casey Johnson calls epistemic care:¹⁶ care work that is not easy, but which trustworthy medical and health experts are specifically well-positioned to do for vulnerable patients and publics who rely on them. Given the social-epistemic differences and complexities involved, my concern is that a norm of expert transparency is inadequate for this sort of careful intermediary work. Uncertainty, ignorance, disagreement, and value judgments are relevant to vaccine policies and other public-health decisions; but how to attend to them requires nontrivial interpretive judgments by skilled, trustworthy experts. Take for example the existence of expert disagreement. There is the primary question of how significant a minority position needs to be in order for health experts to refrain from taking a settled position; there is also the question of how significant a minority position needs to be in order to justify its explicit acknowledgement in individual medical or public-health recommendations. The second question not only requires expertise to evaluate, but also admits of different answers for different trusting (and not so trusting) publics.

Relatedly, if to count as an expert at all is not just about individual knowledge but a relational matter (namely, to be trusted as a credible epistemic authority), then the extent to which an expert disagreement on a subject of public health even exists will be a moving target for different trusting (and not so trusting) publics. If and when a testifying public-health expert is to be transparent about the existence of expert disagreement on the issue at hand, whose trust matters in assessing relevant expertise? We might consider the testifying experts themselves, whom they recognize as credible authorities in their field and which voices of opposition to this health recommendation count as expert disagreement or not. Yet there is not just who the expert testifiers themselves trust (or not), but also whom the recipients of their testimony trust (or not) as relevant credible authorities. An individual doctor might need to think about this differently for different patients, depending on whom a particular patient trusts or not as a credible authority. Experts tasked with conveying public health recommendations likewise may need to think about how different publics will engage with their recommendations and the contexts in which they convey it.

As GG&J note, recent trends toward a crisis of expertise are perhaps better understood as shifts in whom respondents recognize and trust as experts. “If we stop trusting certain experts, we will simply move that trust on to others.”¹⁷ Among other things, this means that there is not a uniform fact of the matter about whether and how many public-health experts disagree with a credentialed body’s recommendations. To cite a rather notorious example: should medical and health experts always acknowledge Andrew Wakefield’s continued contention of an autism-vaccine link when they recommend childhood vaccination schedules? Wakefield himself has been loudly discredited; his published research on the matter has been retracted and remains unreplicated by others. All of this may be rather damning for many of us, yet as Maya Goldenberg observes, vaccine-hesitant parents and others continue to put their trust in him in part because he legitimates their concerns.¹⁸ A blanket policy of transparency about Wakefield’s autism allegations as either always relevant expert disagreement or always irrelevant disreputable conspiracy on vaccine science and policy would fail. For some patients, our doctors spending precious appointment time to address vaccine hesitancy may be a waste, or even unintentionally legitimate the position. For others, meanwhile, refusing to address vaccine hesitancy would be an interpretive mistake, not because doctors should simply reaffirm patients’ (or parents’) fears, but because these fears will not simply dissipate if ignored.

All of this calls for trustworthy expert interpretation, responsive to the particular relationships of interdependency that medical and health experts are party to. The same can be said for the value judgments at work in medical and health recommendations and when and how to convey them to different trusting (and not so trusting) publics. GG&J hold that when experts speak on matters of public interest, “their evaluations of scientific evidence are typically grounded in value judgments regarding whether the level of available evidence or knowledge is enough to warrant a claim or recommendation…Value judgments of this sort are unavoidable.”¹⁹ It can be tempting to draw a strict dichotomy on values in science, where ethical and political values arise only at the stages of practical application or public recommendation while the science itself, focused on empirical and epistemic questions, remains value-free.²⁰ But as GG&J recognize, philosophers of science have long debated “whether science is a purely empirical matter or the extent to which it is an activity laden with moral (and other) values.”²¹

When health experts are called to testify to value judgments that they and others have made, the scope of which judgments are relevant or irrelevant to the public recommendation at hand may not be obvious. How far back must one go? How deep into decisions of experimental design, research funding, or peer review? In asking these questions, I do not mean to suggest that they are always appropriate or always inappropriate to include alongside individual medical decisions and health policy recommendations. Nor do I mean to imply that experts should hide their (or others’) value judgments: as Kristen Intemann reminds us: “Obscuring values risks reinforcing false beliefs about how science works, making it more difficult for the conditions of warranted trust to be achieved or maintained.”²² The point is that a norm of expert transparency framed as acknowledging value judgments is too broad to provide good guidance on which value judgments to discuss, and how, for health experts working with different trusting (and not so trusting) patients and publics.

Perhaps the answer is to acknowledge just those value judgments in vaccine science and policy on which the experts disagree. To use GG&J’s example of childhood COVID vaccinations in 2021, one might argue that when the UK government authorized the use of vaccines for 12-15-year-olds, the government and its expert spokespersons should have acknowledged the role values played in weighing the significance of not only vaccine side effects but also disruptions in school attendance and mental health costs of continued social-distancing restrictions. These judgments should have been clearly flagged, the argument goes, because different experts weighed them differently. Even if that is right, however, we then must revisit the points raised above about the need for trustworthy interpretive work in evaluating who counts as a relevant expert and whose disagreement should be discussed, and how, with different audiences. And there is also no reason to assume that only those value judgments over which experts disagree are relevant to lost, gained, or continued public trust. Indeed, the fact that none of the recognized experts thought differently than their colleagues about a particular value judgment informing vaccine science or policy itself might be read as evidence of vicious groupthink for a particularly skeptical patient or public.

“Speaking of ‘transparency’ raises connotations of complete observability,” says Kevin Elliott, “whereas value judgments are often somewhat opaque and difficult to identify”.²³ For his part, Elliott allows that even if scientists cannot or should not disclose all value judgments related to their work, “it might still make sense to demand that they provide detailed information about their data and methods so other scientists could scrutinize their work and identify important value judgments for themselves.”²⁴ Here again, notice how even decisions about what information to make public call for nontrivial judgment, and how to do so not just perfunctorily but effectively calls for skilled interpretative work responsive to the particularities of varied audiences and communicative contexts.

Some may object that the sort of interpretive work I have been discussing is too demanding for overloaded individual practitioners, health policymakers, and spokespeople. How can a healthcare worker be relationally responsive in how they share matters of medical or health expertise tailored for specific patients (and their partners, parents, families, etc.)? How can public-health experts be relationally responsive in their policy or other recommendations when their expert testimony is transmitted simultaneously to many differently positioned publics? I would not dismiss or diminish these challenges. Taking them seriously underscores just how valuable and impressive it is when doctors, nurses, and other healthcare practitioners do indeed take patient specificities into account in communication. Making them explicit helps to clarify which patients and which publics tend to be prioritized as the presumed default audience of expert testimony and which patients or publics by contrast tend to be overlooked.

Individual medical experts do not exist in isolation, to be sure. Institutions, legal requirements, professional organizations, and cultural expectations construct social-epistemic environments that support (or impede) individual experts in their interactions with vulnerable patients or publics. As Casey Johnson puts it in discussing epistemic care in communities of inquiry, “someone cannot be properly attentive and meet the needs of those who are depending on her if she cannot rely on others to help her meet her needs… we must attend to who is doing the epistemic labor and how well they are supported.”²⁵ A workplan that confines doctor-patient interactions to short bursts will indeed make it harder for an individual doctor to discern how to communicate effectively with a specific patient. An educational system that values patient-practitioner communication will better equip individual practitioners for success in this respect that one focused on expert knowledge acquisition narrowly construed. And since medical and health experts are themselves epistemically interdependent, trusting and being trusted by each other, their collective epistemic resources build through overlapping divisions of labor. Each individual expert need not be highly skilled in every respect in order to be a highly skilled part of a trustworthy expert team or community. This extends to include not only medical specializations, but also the interpretive skills needed to communicate effectively with different patients and publics.

6. Healthy Expert Trust Revisited

If not a commitment to transparency, what other check might we recognize on trusted medical and health experts? How else might we diagnose whether such experts have betrayed or upheld the trust placed in them by vulnerable patients and publics? I would recommend something like Baier’s expressibility test, which leaves room for discretion in how trusting parties attend to the objects entrusted to them while also recognizing how such trust can indeed rot.

As noted above, one ambiguity about the call for transparency is whether trusted experts must speak to the matters in question unprompted, respond openly when asked, or find some balance between these. It is also unclear whether fulfilling a norm of expert transparency requires unfettered, unfiltered access, guidance to the relevant information, or a mixture of the two. In acknowledging the nontrivial interpretation needed when medical and health experts communicate with patients and publics, by contrast, what bolsters or erodes healthy expert trust looks different. So rather than “Have we been transparent to patients and public who trust us?”, Baier’s test would have us ask, “Have we been relying on continued ignorance of something by patients and publics who trust us in order to sustain this trust relationship?” The answers to these questions do not always converge, and oftentimes it is the latter that really matters for healthy expert trust.

Consider for example a public-health spokesperson who would answer uncomfortable questions truthfully, but to their relief they are never asked. Whether this person satisfies a norm of expert transparency is unclear, but they clearly fail Baier’s test. And so does an expert who acknowledges something inconvenient for their position but in ways that a certain trusting patient or subsection of the public will fail to comprehend (and this expert has put things just this way for just this reason). What sorts of value judgments need to be discussed, and in what ways, with non-expert audiences? What constitutes a minority expert disagreement on a matter of vaccine science or policy that need to be acknowledged, and how? Here Baier’s test would have medical and health experts ask themselves, “Which if any of these value judgments, differences of opinion, or other factors are we relying on this patient or this public not knowing about for their continued trust in our recommendations?” These are the sorts of things that bring moral rot to our trust relationships. Trustworthy experts avoid them by being humble and honest, to be sure, but also by doing the subtle and skilled interpretive work of facilitating patient and public comprehension for individual decisions and policy recommendations. Medical and health experts do this sort of epistemic care work every day. We should recognize and appreciate it as a vital part of what makes them credible authorities for those who rely on them.