In “Minding the Gap: The Absence of Pennsylvania’s Involuntary Commitment Data,” Gluckman et al. draw on frameworks from agnotology and absence studies to persuasively argue that unreported information about psychiatric commitments results in the disenfranchisement of vulnerable populations.Reference Gluckman 1 They call for more and better data regarding individuals affected by civil commitment. If states and agencies are to heed this call effectively, however, an awareness of additional agnotological considerations can help to ensure that new data initiatives do not inadvertently perpetuate the problems they are meant to address. In employing the frame of undone science to examine ignorance, the authors primarily focus on omission and absent data; they use granularity, scale, and intentionality to describe the contours of ignorance as an object through the negative space around it. In this essay, we present a complementary agnotological framework that focuses on the processes, active and passive, through which ignorance is generated and maintained.Reference Kourany, Carrier, Kourany and Carrier 2 We first examine four elements passively producing ignorance: frameworks, community, methodologies, and incentives. We then discuss active agnotological forces, both harmful and virtuous, as they shape the field of what is knowable, with concrete applications to the call to create data dashboards measuring involuntary psychiatric commitments.
Background
Agnotology, the study of ignorance, is a recently developing field, the term having been coined in 1992,Reference Proctor, Proctor and Schiebinger 3 with published work increasing significantly after 2015.Reference Smithson, Gross and McGoey 4 An interdisciplinary subfield, ignorance studies draw from disciplines like philosophy, sociology, psychology, anthropology, and criminology, 5 with the goal of creating social science that “takes the ‘regularity’ of ignorance seriously.”Reference Gross, McGoey, Gross and McGoey 6 Agnotology and undone science intersect with applied interdisciplinary fields like Science and Technology Studies (STS) and the health humanities, in which social sciences and humanities contribute to the projects of physical science and medicine.Reference Hess, Gross and McGoey 7 Indeed, historians of medicine have at times uncovered evidence of intentional ignorance production, providing templates for further analysis.Reference Proctor 8 An important thread within agnotology “analyzes how our knowledge has been influenced by struggles determining who is included and who is excluded, … whose experiences are validated and whose are not, and who stands to gain in terms of wealth or well-being.”Reference Schiebinger, Kourany and Carrier 9
Agnotology is also a philosophical inquiry, complementing epistemology and philosophy of science by asserting that “knowledge is frequently what is natural, and ignorance is frequently what needs to be explained.” 10 This more expansive examination of ignorance has created taxonomies to distinguish different types of unknowns, ranging from absence to misinformation to situations in which someone is “incapable of even understanding the question” being posed.Reference Hoyningen-Huene, Kourany and Carrier 11 Scholars of undone science look for omissions generated by “broader cultural, historical, and macrostructural conditions,” including conflicts between public good and industrial or institutional priorities. 12 They, along with scientists, tend to focus on “a knowable unknown that they see as positive or desirable to study.” 13 Gluckman et al. use precisely this absence-focused model in their call for data on involuntary commitment. 14
Typing ignorance as either passively or actively generated shifts our gaze from ignorance-as-object to the conditions that shaped it, encouraging us to consider ignorance creation an ongoing process, similar to knowledge production. 15 This approach allows investigators to focus not on empty space, but on visible, knowable processes, patterns, and problem areas, as well as suggesting avenues for intervention.
Passively Generated Ignorance
Any research program makes choices that preclude other avenues from being explored. Scientific inquiry can be described as a “searchlight” that “illuminates certain aspects of experience and thereby leaves other features in the dark.”Reference Carrier, Kourany and Carrier 16 Although researchers do their best, gaps are inevitable, and the areas left unexamined are neither neutral nor predestined. Agnotology helps make them visible. Guided by Kourany and Carrier’s typology, this section examines frameworks, community, methodologies, and incentives in passively shaping fields, with applications to Gluckman et al.’s work on involuntary commitment data.
Frameworks
Conceptual frameworks provide the resources with which people approach problems and the boundaries within which they inquire. They define the key terms, mental models, and important goals of different vocational communities and scholarly disciplines. While each system ignores something vital, they likewise reveal something unique. Employing a pluralist approach to epistemologies, “in which more than one theory (or classificatory system) can be the best way to represent the world, even when some theories (or classificatory systems) disagree with one another”Reference Solomon, Kourany and Carrier 17 allows us to integrate different types of knowledge. Conversely, using only the perspective of public health, biomedical science, or law enforcement may create critical omissions in understanding, or prevent public servants and health workers from cooperating well.
For example, biomedical and public health paradigms approach similar problems through different lenses. A biomedical, disease-centered research program encourages researchers to study biological mechanisms of sickness and test the efficacy of potential treatments. Other disciplines understand health and illness as arising from social determinants and material conditions, like access to food and sanitation. Either framework determines the direction of inquiry, forcing researchers to choose to “focus on the social conditions of the problem or seek technological solutions; either choice produces selective ignorance.” 18 However, public health and biomedical approaches synergize in interdisciplinary approaches to mental health.
In the case of data concerning individuals who have experienced commitment, some applicable frameworks may be less commensurable with each other. Health-focused frameworks still construct the individuals in question as patients who need medical care and bioethical protections. On the other hand, public safety and law enforcement frameworks have different assumptions and obligations; they construct the same people, not as patients, but a population defined by the risks they pose. Legal frameworks also have their own nomenclature and mandates. For instance, although a new rule refining the language is under review, as of this writing, Pennsylvania defines individuals with mental illness as “persons who have been adjudicated as mental defectives or been committed to a Mental Institution,” a subset of “the Categories of Persons Prohibited from Receiving Firearms,” for the purposes of enforcing the Gun Control Act of 1968. 19 Pennsylvania’s statewide background check system, the PICS database, was built to enact gun control legislation. It functions not as a record of patient care, but a gatekeeper on the legality of selling firearms.
All of these conceptualizations necessarily coexist with disability justice perspectives that understand the same people as community members and stakeholders in this conversation. The absence of any one conceptual framework creates corresponding dimensions of ignorance.
Community
The different frameworks above may be ascribed to different vocational communities. However, agnotology also considers socially-determined communities as they are prioritized or delegitimized, and the detrimental impact of such absences on scientific norms and knowledge. 20 Epistemic injustice and less accurate science are both consequences of marginalizing perspectives.
When Gluckman et al. analyze granularity, they assert that “the lack of patient perspective in administrative documents means that the data are complicated by a kaleidoscope of imbalanced power relations.” 21 This problem was passively generated, not just by paperwork, but by the perspectives missing during the paperwork’s design. In the analogous case of harms created through the exclusions of women from medicine, solutions have required rethinking foundational assumptions and reshaping practices; including women in drug trials lessens pernicious side effects, but improving care requires that women actively participate in selecting problems and defining desirable outcomes. 22 Similar work would integrate lived expertise into research about involuntary commitments.
Methodologies
Different research methodologies result in fundamentally different kinds of available information. From many potential methodological questions, we select two relevant to Gluckman et al.’s article: quantitative versus qualitative data, and dashboard construction.
The community-based omission Gluckman et al. identify in their analysis of granularity is mixed with concern for “the richer qualitative data that may provide a better picture of the patient’s experience.” 23 Employing methodologies from a broader range of epistemologies does shine the “searchlight” on new types of information. However, gathering qualitative data at intake is not equivalent to addressing epistemic justice concerns, as disambiguating community-based and method-based ignorance shows. Further, ensuring such data are rigorous and meaningful requires careful, potentially resource-intensive interdisciplinary design, implementation, and analysis. Qualitative data presentation might also complicate the construction of data dashboards.
Designing dashboards is likewise not neutral. While the rapid expansion of data dashboards seems beneficial for public health decision-making, data experts are still assessing their usability and generating recommendations for best practices in design and stakeholder alignment.Reference Stahlman 24 Usefulness is impacted by “significant differences in how, for whom, and for what purpose public health dashboards are created” as well as designs lacking theoretical grounding to “link usability and usefulness affordances to user experience and learning.”Reference Yanovitzky 25 Improved data availability depends on the theory and methods used in dashboard design, as well as thoughtfully selecting by and for whom the data are collected and displayed.
These choices have powerful consequences. Critical data studies scholars and experienced dashboard builders have described a data dashboard as “a complex socio-technical assemblage,” noting that “the production and maintenance of a dashboard unfolds contextually, contingently and relationally.”Reference Kitchin, Lauriault and McArdle 26 Further, dashboards are not reflections of reality so much as “translators and engines,” and using them “frames how data are visualized and thus what the user can see and engage with, and what questions can be asked and how the answers are displayed.” 27 Data workers themselves must grapple with “data’s epistemological ambiguities” especially when missing or partial data for marginalized populations complicate both collection and meaning.Reference Kim 28 Advocates might usefully center which stakeholders commission data dashboards, how they are designed, and, as addressed below, what financial incentives mandate.
Incentives
Considering incentives highlights an assumption that is usually in the background: economic and interpersonal incentives determine what knowledge is collected, how it is shaped, and its eventual utility. More formally put, non-epistemic, sociocultural values play a formative role in knowledge creation. 29 Often these external values support epistemic values, which “appreciate knowledge and understanding.” 30 For example, when knowledge workers are rewarded for accurate research, their social and economic goals align with epistemic values.Reference Klintman, Gross and McGoey 31 Insofar as possible, then, the knowledge goals of stakeholders must align with incentives impacting individuals employed in data collection and dashboard design.
Incentives are part of both passive and active agnotology. Systems that incentivize ignorance or inaccurate information pit epistemic and anti-epistemic, or pro-ignorance, values against each other. As a result, even when individuals are being passively influenced, incentives actively shape what it is profitable to know. One example involves American chemical safety regulations. Unlike pharmaceuticals or pesticides, which are safety tested before being marketed, many chemical types are governed by post-market regulations, allowing products to be sold until sufficient evidence of risk surfaces.Reference Cranor, Kourany and Carrier 32 Regulatory structures thus incentivize companies to avoid learning about harms their products might pose.
Regardless of intent, there is “a general principle that historians and philosophers need to pay more attention to: problem selection and funding shape what kind of science gets done.”Reference Galison, Proctor, Kourany and Carrier 33 In the tobacco industry’s campaign to avoid regulation by obscuring connections between cigarettes and health problems, the Tobacco Industry Research Committee funded research that led to Nobel Prizes and thousands of peer-reviewed articles. 34 Individual scientists were likely unaware they were being incentivized to ignore problems.
Active
Actors can also shape ignorance purposefully, both for harmful purposes, or through well-meaning, virtuous intentions.
Harmful
Active agnotology case studies focus on groups like the tobacco industry that demonstrably conspired to avoid regulations and legal liability, 35 or individuals like Andrew Wakefield and his Lancet article tying vaccines to autism, in which “not only methodological blunder, but intentional forgery was established.” 36 Much of the field of agnotology elucidates harmful activity through archival proof and explication. Since, as Gluckman et al. note, no such proof exists in the case of commitment data, 37 we briefly summarize some major patterns to provide context and highlight the potential role of incentives.
Sometimes active agnotology multiplies information rather than creating gaps. Public opinion that contradicts scientific consensus frequently results from agnotological efforts to control public policy for economic or ideological gain. 38 “Manufacturing uncertainty” can involve denying scientific consensus, raising the standard for proof, insisting “more research is needed,” or creating rival causal theories, strategies requiring more publications.Reference Michaels, Proctor and Schiebinger 39 Along with tobacco, agnotological campaigns created public doubt regarding problems created by “beryllium, lead, mercury, vinyl chloride,” and other deadly substances, often led by scientific product defense firms and lobbyists working for businesses or interest groups. 40
The funding effect has also impacted academic publishing. Peer-reviewed biomedical journals adopted publishing policies mandating transparency, accountability, and research integrity, in response to a demonstrated pattern of distortions: Studies evaluating the efficacy and safety of medicines, when funded by pharmaceutical companies, were frequently impacted by methodological flaws that supported market-readiness and downplayed dangerous side effects. 41 One example concerns Vioxx, an NSAID pulled from market for unacceptable adverse consequences after Merck-funded studies misinterpreted the data. 42 While we cannot know the motivations of any individual scientist, journals have measured the demonstrable impact of financial incentives on medical science. 43
Thus, the funders involved, and the funding mandates themselves, will determine whether commitment data will be useful and to whom.
Virtuous Ignorance
While many types of virtuous ignorance exist, privacy, in which ignorance is intentionally constructed for the protection of others, is most relevant to the current discussion. 44
Interestingly, in the case of commitment data, the virtue of privacy is contested. Gluckman et al. argue that enumeration and visibility are critical for advocacy and improved services, 45 drawing on work connecting data omissions to reduced “claims-making potential” for marginalized groups.Reference Cheong 46 Concern for privacy can thus deprive vulnerable people of something they need, creating tension between protective ignorance and harmful ignorance that requires careful and thoughtful solutions. When ignorance-creating policies are genuinely intended beneficently, and the dangers they address are real, more vigilance, not less, is indicated to ensure that they neither obscure needed data, nor excuse the violation of necessary protection.
The passive elements mentioned above can help thread this needle. For example, the cognitive frameworks dominating calls for data will determine whether the enumerated are primarily constructed as stakeholders, patients, or adjudicated persons, shaping data availability and protection. Involving the community in design and evaluation is essential to their privacy and advocacy needs being met. The methods used to construct data dashboards and gather data will ultimately determine whether essential privacy is maintained, breached, or ignored, as well as how information is understood. Finally, incentives shape outcomes. RFPs and funding proposals can limit potential economic or political benefit from medical datafication and surveillance, while providing data workers access to sufficient resources to create and maintain highly effective projects.
Conclusion
Agnotology has more tools than this brief commentary can consider, so we have left significant omissions in our discussion of ignorance. At the same time, an epistemological lens that names and normalizes the inevitability of such knowledge gaps improves one’s ability to manage and minimize the problems they produce. Gluckman et al.’s call for more data is warranted, but as they note, more information does not ensure better outcomes. Ultimately, understanding how ignorance is created strengthens data initiatives’ potential to improve outcomes instead of replicating harms.
Disclosures
The authors have nothing to disclose.
Open access
