2.1 Generic drugs

Generic drugs are pharmaceuticals with the same active ingredient (e.g. Mefenamic Acid) as the original brand-name drug (e.g., Ponstan), but often with different additives or fillers. As the UK’s National Health Service (NHS) website tells us, “The generic versions will be the same as the branded medicine because they contain the same active ingredients. They are used more often by the NHS because they’re just as effective but cost far less” (NHS 2020). Generic drugs make up the majority of prescriptions in the UK and over 80 percent of prescriptions in the USA (Oxera 2019; McGregor Reference McGregor2020). Because the active ingredient is the same and has already been approved for patient use, in the USA (and EuropeFootnote ⁶ ) generic drugs undergo “bioequivalence” testing rather than full trials:

Women and gender minorities are frequently not included in these studies.Footnote ⁷ This omission results in a gender data gap, because the so-called “inactive” ingredients—colours, fillers, and preservatives, known collectively as “excipients”—can have a different effect on how the body absorbs and processes the medication. McGregor uses the example of polyethylene glycol (PEG) as a filler in generic ranitidine, which in initial bioequivalence studies was shown to “increase bioavailability” by 63 percent. On this basis, researchers concluded that the generic product could produce equivalent results to Zantac (the original brand-name drug) with less of the active ingredient. However, these studies didn’t include women, for whom it was later found PEG decreases bioavailability by 24 percent: “So generic ranitidine produced with PEG was actually only about half as effective for women as it was for men at an equivalent dose” (McGregor Reference McGregor2020, 89). This isn’t an isolated case (either with PEG or generic drugs more generally; see for instance Chen et al. Reference Chen, Lee, Ng, Schuirmann, Lesko and Williams2000; Koren et al. Reference Koren, Nordeng and MacLeod2013). Yet the standard model of young adult, healthy, usually male, volunteers is “regarded as adequate in most instances to detect formulation differences and to allow extrapolation of the results to populations for which the reference medicinal product is approved (the elderly, children, patients with renal or liver impairment, etc.)” (European Medicines Agency (EMEA) 2010).

2.2 Implantable devices

Surgery to replace worn or damaged human joints—such as hips and knees—with artificial joints is offered to patients to reduce pain and increase their range of movement, with artificial joints expected to last for “at least 15 years” (NHS 2019a). Advice to patients on the NHS website identifies a range of risks associated with surgeries, together with the probabilities of these risks occurring. This information is a crucial step in enabling patients to make informed choices about their surgeries. However, none of the risks is differentiated by gender.

Katrina Hutchison (Reference Hutchison2019) discusses a significant difference in outcome success of hip replacements between male and female patients. Metal-on-metal hip replacements, once popular for their longevity, had higher failure rates when implanted into women than men (Haughom et al. Reference Haughom, Erickson, Hellman and Jacobs2015). Complications from device failure included pain and inflammation, pseudo-tumours and raised metal-ion levels in blood, in turn leading to further risks to heart health, thyroid function and negative effects on both cognitive and nerve function (Hutchison Reference Hutchison2019, 571–72). The devices had been tested and recommended by the manufacturer, DePuy Orthopaedics, as being more suitable for younger patients, on account of the anticipated longevity. What went wrong? Hutchison uses DePuy’s metal-on-metal design to illustrate a wider problem within medical testing and research: “Gender biases can be built into devices when the possible impact of such differences is ignored in the design, benchtesting, clinical testing, and regulatory approval processes” (2019, 572). Put simply, the differences between how different people walk, sit, stand, and have sex are all relevant factors to consider in testing. The ways in which people do these things can be affected by their gender group membership in virtue of average anatomical differences, social norms, or both. Where implants are tested only on one gender group, a data gap results. The NHS website’s section on risk has nothing to say about how sexual position, carrying an infant on one’s hip, or wearing certain kinds of shoes might increase wear and tear on the device (NHS 2019a). Changes in hormone levels, for example through menstruation, pregnancy, or hormone therapy, may also affect joint stability. Data on the effects of hormone changes on the stability, success and longevity of devices are—perhaps unsurprisingly—not available.

2.3 Car safety

GDGs are not restricted to medical research; they are ubiquitous. Car safety tests, used as marketing tools and required for the protection of drivers and passengers alike, provide another instance (Criado Perez Reference Criado Perez2019). These GDGs occur because cis men are taken to be representative of humans in general. For example, as drivers, women on average tend to sit further forwards to reach the pedals than men do, making them more vulnerable to internal injury in head-on collisions (Gendered Innovations Research Group n.d.). Additionally, sitting in car seats which have been designed in reference to the spinal shape and muscle strength in the neck representative of a cis man results in many women being more likely to suffer whiplash in a rear end collision.

When first introduced, crash test dummies were based upon an average male body of 1.77m height and 76 kg weight. A call to model crash test dummies on females was ignored for decades (Criado Perez Reference Criado Perez2019, 187). When crash test dummies resembling female bodies were eventually introduced in Europe, the car safety rating agency EuroNCAP used sex disaggregated data “where data [wa]s available” (188). But since the data are frequently unavailable, the safety ratings awarded may completely ignore average anatomical differences.Footnote ⁸

In some cases, EuroNCAP used “scaled down male dummies.” But females are not mini-males: many females have breasts and—as noted—differently shaped spines, different upper-body muscle strength, and different seating positions. Pregnancy is also ignored in standard seatbelt design (Criado Perez Reference Criado Perez2019, 188).

In other cases, where anthropometrically female crash test dummies were used, it was only in the passenger position, yet car safety ratings were based upon the outcomes for the crash test dummies in the driver’s position. This kind of research generates two data gaps for women by failing to collect data on women as drivers and by failing to publish data on women as passengers. When these data are published, the safety ratings for cars can drop dramatically (Criado Perez Reference Criado Perez2019, 189).

2.4 Identifying injustice

Common to each of the case studies is the presence of a gender data gap which hides certain important facts, for example that the risks and effectiveness of a product or procedure are not the same for cis men as for women and gender minorities.Footnote ⁹

It is clear that something is going wrong as a result of GDGs. In Generic drugs and Implantable devices, women and gender minorities are vulnerable to unexpected and worse treatment outcomes; in Car safety women and gender minorities are likely to come off worse in an accident whether they are drivers or passengers.Footnote ¹⁰ Along with authors such as Criado Perez, Hutchison and McGregor, we think these GDGs lead to injustice. However, while those writing on GDGs have deemed the status quo and the resulting inequalities as unjust, there has been insufficient attention paid to the nature of the injustice.

Intuitively we might judge the injustice to be socio-political, particularly where it arises from subconscious biases, prioritizing ease or profit, legislative or regulatory oversight or omission, or similar. Furthermore, there might be many instances of GDGs which do not lead to actual harm. For example, some of the women and members of gender minorities using implantable devices might be sexually active in ways that do not lead to increased wear and tear. Many people will not be in motor accidents and so not be caused significant damage due to the seat design. Nonetheless, we recognise that exposing women and gender minorities to higher risks than cis men—and to potentially worse outcomes—without justification is a social injustice.

However, we shall show that, even where socio-political injustice results from GDGs, it does not exhaust the kinds of injustice present; data gaps such as those identified in our case studies also give rise to a form of injustice that is distinctly epistemic in nature. This epistemic injustice arises for agents (in our examples women and members of gender minorities, both individually and as a group) when they are denied important knowledge, relevant to the proper exercise of their epistemic agency, for bad reasons. As we shall argue, GDGs inhibit the formation of self-regarding action-guiding beliefs; to borrow a turn of phrase from Fricker (2017, 53), they represent a systemic rigging of the epistemic economy.

4.1 Epistemic deprivation as conceptually distinct

Discussions of epistemic injustice often focus on the interpersonal. When Fricker and her interlocutors talk about the wrong done to someone in their capacity as a knower, there is a heavy focus on communication, and the challenges which members of marginalised groups face in contributing to, or acquiring, knowledge (e.g. Fricker Reference Fricker2007, Reference Fricker2010; Hookway Reference Hookway2010). Carel and Kidd (Reference Carel and Kidd2014) offer particularly evocative examples of the epistemic injustice faced by ill people in the context of interactions between patients and medical professionals, such as the following conversation between a patient who has just given birth and their doctor: ‘‘That hurts. Are you using anaesthetic?’’ ‘‘No’’ the doctor replies calmly, ‘‘there is no need to. I’m nearly finished.’’

The paradigmatic cases of epistemic injustice are taken to be those where a person, or group of people, have knowledge to communicate but something–frequently prejudice against that group or a paucity of shared social concepts—prevents that communication. But, as has been well-demonstrated in recent years, such cases do not exhaust the injustice one can face in one’s capacity as an epistemic subject. Indeed, communication isn’t at the heart of any of Generic drugs, Implantable devices, or Car safety. Rather, our case studies reveal injustices related to self-regarding knowledge acquisition and decision-making, not interpersonal exchange.

We acknowledge that, at least sometimes, other forms of epistemic injustice are present in GDG cases; sometimes GDGs arise as the result of—or result in—these other varieties. Nonetheless we contend these other forms do not exhaust (or fully capture) what’s going wrong in our case studies.

If some gender groups are excluded from participating in research because they are thought to be unreliable reporters (of, say, their experience of a drug), their exclusion would cause a GDG arising from testimonial injustice. However, testimonial injustice is not a good explanation for the injustice in our case studies. The GDGs in Generic drugs, Implantable devices and Car safety give rise to injustice even where there is no interlocutor and therefore no credibility deficit. Suppose a non-binary person were to look up car safety statistics online when choosing to buy a new car. Encountering data skewed by GDGs, the beliefs they form about the safety of particular vehicles for themself, as a driver, may be untrue. The same applies for women or members of gender minorities reading the patient information sheet accompanying generic drugs. Even in cases where testimonial injustice does cause a GDG, it may not fully explain the resulting injustice: women and gender minorities remain epistemically deprived as the result of their exclusion, because the requisite data was not gathered to enable them to form true beliefs or avoid false ones, for no good reason.

Likewise, some GDGs can be understood, at least in part, in hermeneutical terms. For example, it has become apparent that women’s experiences of heart attacks can be quite different from men’s experiences (Davidson Reference Davidson2017) and are often called “atypical.”Footnote ²¹ However, prior to the recognition of these “atypical” symptoms (that is, prior to addressing the data gap), it’s plausible that the women suffering them—and the doctors in a position to make a diagnosis—might have lacked the concept not of the experiences themselves, but of these experiences as being symptomatic of a heart attack. In other words, there was conceptual paucity around “heart attack.” This is paradigmatic hermeneutical injustice: women suffering heart attacks were unable to understand their experience or make it intelligible to others. ^{Footnote 22} Again, hermeneutical injustice does not fully explain what goes wrong in Generic drugs, Implantable devices, or Car safety; in these cases, the injustice does not stem from women failing to be understood or their perspectives not being considered; indeed, in order to tease out the distinction between hermeneutical injustice and epistemic deprivation we have deliberately chosen cases where people’s perspectives are not the relevant (missing) data.Footnote ²³

It is common to characterise hermeneutical injustice as a lack of (access to) concepts, but the injustices encountered in our three case studies concern a lack of (access to) particularly valuable facts: facts that allow women and gender minorities to form self-regarding action-guiding true beliefs, and avoid false beliefs; facts that enable women and gender minorities to have the same degree of epistemic agency as cis men. What is missing in GDG cases is often not conceptual understanding to make sense of our lived experience, but factual information pertaining to risks and efficacy. Indeed, we frequently do have mutual understanding across genders of the relevant concepts: risk, effectiveness, pain, disaggregated data. In Generic drugs, for example, even if one did lack understanding of what it is for a drug to be effective, or what is meant by a side effect, this would not exhaust what is missing; women and gender minorities with the aforementioned concepts would still lack access to relevant information such as the degree of active ingredients their bodies would likely absorb. In Car safety, the injustice arises because of a lack of data regarding facts about the nature and chances of potentially harmful outcomes. Will someone other than a cis man survive a particular kind of crash? How is their body affected by the material a given car’s seats are made of?

We also contrast epistemic deprivation with Sophia Dandelet’s recent work on epistemic coercion, in which she discusses self-gaslighting as a form of injustice. Dandelet makes space for the notion that an agent’s higher-order epistemic decisions can be coerced when they decide how to structure their epistemic inquiries in response to threats that place unfair pressure on them (Dandelet Reference Dandelet2021). As we note in section 4.3, epistemic deprivation may sometimes lead to epistemic coercion. However, one can conceive of cases of epistemic deprivation without coercion, and thus we consider them conceptually different.

Finally, we take epistemic deprivation to be distinct from epistemic exclusion, as introduced by Dotson (Reference Dotson2014). Pohlhaus Jr notes that exclusion involves “pre-emptively regarding a person as unreliable or incapable of participating in particular epistemic systems” and so impedes “epistemic agency by barring them from full participation in those systems” (2020: 235). GDGs may arise when researchers or policymakers view women or gender minorities as unreliable or incapable of participating but that needn’t be the case. GDGs can (and seem often to) arise because of an unquestioned assumption that women and gender minorities are just like cis men in ways relevant to that product, procedure, or policy. This assumption skews the epistemic landscape for women and gender minorities in two ways, as outlined in section 3: they are less able to avoid false beliefs, and to form true beliefs, about important self-related propositions. For example, the false belief that the effects and risks associated with generic drugs are identical for oneself to those associated with branded drugs, and the true beliefs concerning the risks and effectiveness of the generic drugs simpliciter. Similarly, women and members of gender minorities cannot accurately weigh up the risks and benefits of an implantable device without knowing how implants are likely to perform in a life like theirs. The epistemic agency of women and members of gender minorities is thus impeded in ways that cis men’s epistemic agency is not. While there is a meaningful sense in which women and gender minorities have been excluded or their differences elided in research, we take the phenomenon we are focusing on to be importantly distinct from Dotson’s sense of exclusion.

Some might worry about conceptual inflation, given that a thriving literature continues to explore the limits of Fricker’s account and the conceptual possibilities of epistemic injustice. Some authors seek to extend Fricker’s framework by, for example, extending the definitions of hermeneutical injustice (Simion Reference Simion, Carter and Bondy2019); others, like Dandelet, argue for new kinds of epistemic injustices. Bratu and Hänel (Reference Bratu and Hänel2021) claim that Fricker offers a concept of epistemic injustice and that additions to the framework can be understood as different conceptions of that concept. We think that, as the landscape currently stands, epistemic deprivation is plausibly conceptually distinct from other forms of epistemic injustice. However, should the future lead to a complete understanding of epistemic injustice, and hermeneutical injustice in particular, one might find that epistemic deprivation is best understood as a conception of hermeneutical injustice, or as a variety of a heretofore unrecognised form of epistemic injustice. Such a future may be unlikely, however, as Dotson notes:

However, regardless of its place in the ultimate conceptual topography of epistemic injustice, we propose that epistemic deprivation be treated as distinct on pragmatic grounds, as we go on to discuss below.

4.2 Epistemic deprivation as practically distinct

Dotson (Reference Dotson2014) argues that different forms of epistemic exclusion can be identified by thinking about what would need to change for it to be remedied. While we focus on a different variety of epistemic injustice than Dotson, we follow her in distinguishing between varieties by focusing on what needs to change. In section 4.1 we considered the conceptual distinctness of epistemic deprivation, but we also think it’s pragmatically distinct; it requires its own remedy. Here we discuss why remedies for other varieties won’t be enough, and in section 5 make suggestions targeted specifically at GDGs and epistemic deprivation.

Testimonial injustice arises when prejudice causes a hearer to give a deflated level of credibility to a speaker’s testimony. Fricker’s (2007) solution naturally focuses on developing virtues to avoid listener prejudice. But as we have argued, there need not be a credibility deficit for GDGs to emerge. Furthermore, addressing listener prejudice will not on its own prevent epistemic deprivation. Suppose that a doctor prescribes generic medications as a matter of course for all patients, genuinely believing—and thus reporting to patients—that they work as effectively as brand-name medications. Suppose also that the doctor tends to disbelieve women, but not men, when they report experiencing negative side effects. The doctor coming to realize they are prejudiced, while beneficial, will not go all the way to improving the epistemic position of their women patients (as shown in Generic drugs).

To resolve hermeneutical injustices, Pohlhaus Jr (Reference Pohlhaus2020) considers the suggestion that a system of collective interpretive resources must be further—and more inclusively—developed so that all can make sense of their social experiences. It is likely that such development might help to resolve some GDGs, since some gaps could be the result of deficiencies in our collective hermeneutical resources that make it harder for women and members of gender minorities to articulate their experiences. However, these solutions would not eradicate epistemic deprivation in our case studies. Imagine that a car manufacturer ran focus groups to understand how women and gender minorities use their cars, and how they conceive of safety and risk in relation to driving and being a passenger, and then that the car manufacturer updated their working understandings of safety and risk accordingly. These findings would be insufficient for addressing epistemic deprivation without access to data as yet unproduced: data relating to how safe women and gender minorities are in cars.

Women and members of gender minorities could have maximal access to the data in circulation concerning car safety, drugs, and devices but would still lack epistemic agency; they would be disadvantaged when it comes to making choices, forming true beliefs, and avoiding false beliefs about matters of importance.

Resolving epistemic deprivation requires distinct remedies, which we discuss in section 5. First, however, we briefly discuss an important interaction between epistemic deprivation and other forms of epistemic injustice.

4.3 Vulnerability

Although we argue that epistemic deprivation is distinct from other forms of epistemic injustice, epistemic deprivation can make one more vulnerable to these other varieties. For instance, the untested and thus undocumented effects of Generic drugs on women and members of gender minorities can make them vulnerable to testimonial injustice when they report said effects. As McGregor writes:

Similarly, women and gender minorities may be more prone to pernicious self-doubt, what Dandelet calls “self-gaslighting”: “In cases of self-gaslighting, the subject worries that other people will be skeptical of one of her beliefs … Prompted by this worry, she scrutinizes her original belief and ultimately gives it up” (2021: 489). It’s not difficult to imagine how epistemic deprivation might contribute to or exacerbate hermeneutical injustice either. For instance, in Implantable devices, we noted that women and gender minorities are epistemically deprived when choosing the implant that is right for them, due to the gap in relevant data. A woman or member of a gender minority might additionally suffer hermeneutical injustice when considering findings pertaining to how an implant performs in “normal life,” if that conception of “normal life” fails to help them to make sense of their own experiences.

What is going wrong in GDG cases need not be a lack of credibility or faulty conceptual apparatus, but these hallmarks of more familiar varieties of epistemic injustice may nevertheless be the result of GDGs. This increased vulnerability provides additional motivation to address GDGs, over and above the injustice of epistemic deprivation.