1 Introduction

Epistemic justice, or rather its negation, epistemic injustice, is one of the philosophical topics of the times. It illuminates a type of injustice that has long been known but rarely lucidly conceptualized: People can be treated unjustly by being disbelieved, or more concretely, by their testimony not being accorded the level of credibility that it deserves. This form of injustice should be of utmost interest to the law, especially criminal law, as evaluating testimony is one of its central practices, and associated normative problems among its key theoretical concerns. The concept of epistemic injustice may reconstruct some long-standing criticisms of the criminal justice system regarding the treatment of witnesses and assessments of their credibility. Fictitious legal cases often serve as examples of philosophical debates, which, however, often fall short of capturing the complexity of procedural law and its underlying considerations.

In its current form, the philosophical concept of epistemic injustice does not directly speak to legal discussions. It must be translated and adapted to the law. This paper wishes to contribute to this task by examining one of the most contentious topics in this area, witness’ testimonies following trauma-focused psychotherapy, in light of epistemic justice. Since the 1980s, courts around the world have grappled with accusations stemming from false memories induced by suggestive therapeutic interventions. As a result, post-therapy testimonies are discounted in one way or another in many jurisdictions. These legal cases also sparked a subfield of psychology—false memory research—and the subsequent “memory wars” questioning the very possibility of forgetting or repressing and subsequently recovering memories of sexual abuse. While these debates have lost some of their momentum, central aspects are yet unresolved; the controversy in psychology is continuing,Footnote 1 courts are still confronted with such testimonies, and the modi vivendi of legal systems to deal with them give rise to complaints about unfair treatment of witnesses.

These concerns tie in with the broader criticism that legal systems often fail to fully hear people’s voices and appreciate their testimonies in an unbiased manner. This includes, but is not limited to, complaints of witnesses, predominantly women, about testimonies of sexual abuse not given the attention they deserve. Whether these claims are true is of course partly an empirical question that requires empirical investigation. But the prevalence of such complaints, in combination with cognitive and systemic biases disfavoring women and minorities, substantiates the suspicion that a problem exists which the law better addresses than represses (Boyle 2009; Tuerkheimer 2017). These complaints also have a normative side with respect to the accurate assessment of testimonies by the trier of facts, and possible legal reforms to improve the situation. This normative side is the topic of the following. More precisely, it examines whether legal rules or established practices of evaluating testimony based on memories that resurfaced after or were potentially altered by pre-trial trauma therapy or similar interventions (“post-therapy testimony”) are epistemically warranted and legally just. The concept of epistemic injustice provides the lens for the examination; it seems well-suited to reconstruct complaints, provides a critical angle on legal practices, and connects the sometimes-narrow intralegal justifications of procedural practices with broader political and ethical considerations. Conversely, applying the philosophical concept of epistemic injustice to an established epistemic practice, such as law, may also test the concept’s capacity to capture real-life problems and contribute to their improvement. Law and philosophy may learn from each other.

The first section briefly introduces the concept of epistemic injustice in its influential formulation by Miranda Fricker, along with several aspects in need of clarification, leading to a modified expanded version that will be used in the following. The second section briefly examines how the law relates to epistemic injustice and points to convergences and divergences. The third section sketches the psychological controversy over recovered memories and the unclear epistemic situation in which the law operates. Seven ways in which legal systems may deal with post-therapy testimonies are presented in the fourth section and analyzed in light of epistemic injustice subsequently. In a nutshell, it will be argued that some ways of assessing testimonies are epistemically unwarranted and may ground concerns of epistemic injustice. However, these assessments may follow from, or are at least consistent with, general principles of criminal procedural law, particularly the paramount aim to avoid false convictions. Unwarranted epistemic treatments might thus be justified by overriding legal principles. However, the concept of epistemic justice, more broadly understood, draws attention to potential harms which witnesses may incur and which the law should strive to avoid. This may inspire legal reform. In particular, preventing deformations of epistemic subjects through legal proceedings and improving the quality and the assessment of testimony may be two foundations for a future theory of epistemic justice in legal contexts.

2 Epistemic and Testimonial Injustice

Situated at the intersection of ethics and epistemology, the notion has received its influential formulation in Fricker’s Epistemic Injustice: Power and Ethics of Knowing (2007). Since its publication, it has inspired a wealth of writing that applies the idea to numerous social domains. The central proposition is that there is a distinct type of injustice done to persons as epistemic subjects, or as Fricker writes, in their “capacity as knowers” (Fricker 2007, 1). The basic idea is intuitively appealing. Some people are unjustly treated as knowers, i.e., as believers, informants, testifiers, or more broadly, as participants in epistemic practices. Epistemic injustice can manifest in various forms and has been decried by different names for a long time (Pohlhaus 2017). For example, lacking access to education hinders people from fully developing their epistemic capacities and from thriving as epistemic subjects. Likewise, excluding or silencing voices, or misrepresenting what they say, may be detrimental to people in their specific role as epistemic subjects (Kidd et al. 2017, 1; Wanderer 2012).

The following focuses on the specific account of epistemic injustice developed by Fricker, the central reference point of contemporary debates. She distinguishes between two forms: testimonial and hermeneutical injustice. The latter concerns impoverished conceptual resources for understanding one’s experiences. For instance, in a world devoid of the concept of sexual assault, people may struggle to make sense of their experiences. The former, testimonial injustice, occurs when hearers “give a deflated level of credibility to a speaker’s word” (Fricker 2007, 1). This type of injustice is of interest to the law, given the significance of testimony as a source of evidence in legal proceedings. It is therefore no surprise that many writings on epistemic injustice refer to historic or fictitious legal cases; a central example of Fricker concerns the testimony of Tom Robinson, a Black person whose testimony is disbelieved due to prejudice. This is a paradigmatic case of testimonial injustice. Importantly, unduly discounting credibility may cause several negative effects that should be kept apart: It obstructs the epistemic objective of the court—accurate fact-finding-, and it may also disadvantage the testifier, as witness or defendant, if the discount causally contributes to a false judgment, e.g., convicting an innocent person. The crucial point of Fricker’s account, however, is that there is an additional injustice to the testifier, independent of the outcome of a case, a distinct epistemic harm to the epistemic agent. This is the basis for epistemic injustice.Footnote 2

While the basic idea of epistemic injustice is appealing, its conditions and finer details provoke many questions and have spurred numerous papers suggesting clarifications or modifications. These debates must be largely left aside here, but some aspects merit further remarks in the legal context.

2.1 Culpable Wrong/Prejudice

According to Fricker, not every epistemic failure suffices to constitute testimonial injustice. It must be a culpable wrong. In her account, culpability primarily arises from discriminatory prejudice (Fricker 2007, 27). Although undoubtedly a concern of paramount importance, restricting the concept to it narrows its scope and creates gaps. For instance, implicit biases of well-meaning hearers may unduly discount credibility but are often not culpable.Footnote 3 This raises the question of the larger aims of a theory of epistemic injustice. If it is about defining conditions for fairly blaming people, culpability should be required, and this is at least part of Fricker’s project (e.g., 2007, 42). Alternatively, if it is about highlighting the types of harm that people may incur through the justice system and the conditions that bring them about, culpability may not be required. A legally interesting account of epistemic injustice encompasses the latter, it is less interested in blaming courts or individual judges, and more in establishing fair general practices ensuring that involved parties get their due, even in the absence of culpability. Culpability is thus a too strong prerequisite for a legally interesting account of epistemic injustice. Moreover, in Fricker’s account, epistemic mistakes must result from specific distorting factors described as “identity prejudice” or “identity-based discrimination such as stereotyping”. While the meaning is clear with respect to race and gender, it becomes less clear with respect to other characteristics such as class, occupation, personal features and proclivities, or previous convictions, which Fricker at least partly explicitly excludes (2007, 4). However, the law demands that testimonies are assessed free from any unwarranted discounts, not only those based on identity prejudice; all unreliable generalizations from groups to individuals about legally relevant attributes are concerning. The following thus adapts a broader version of testimonial injustice that also comprises the non-culpable, unwarranted deflation of the credibility or probative value of testimony because testifiers belong to, or appear to belong to, a group defined by social, bodily, biological, or other criteria.

2.2 Epistemic Mistakes

Another key point, sometimes passed too quickly in the epistemic justice literature, is that discounting credibility must be based on an epistemic mistake to qualify as an epistemic injustice, i.e., it must violate standards of epistemic rationality. If discounting is warranted for epistemic reasons, say, because the testifier is a compulsive liar, she is not treated unfairly as an epistemic subject. Fricker’s account presupposes this condition because it is inherent to the idea of prejudice and implicit in her exemplary cases that feature contrastive assessments of one person being believed to a lesser degree than another for non-epistemic reasons. Conferring credibility to speakers unequally is an epistemic mistake unless epistemic reasons warrant doing so; discounting testimony because of race is an epistemic mistake because no valid epistemic rule warrants it.Footnote 4 To Fricker, the general obligation of hearers is “to match the level of credibility she attributes to her interlocutor to the evidence that he is offering the truth” (2007; 19). This notably does not imply that speakers must be believed simpliciter. Rather, prejudice—or for present purposes, every epistemically distorting factor—must result in the speaker “receiving less credibility than she otherwise would have—a credibility deficit.” (Fricker 2017; 21). A deficit presupposes the existence of a normative standard, namely the level of credibility demanded by the evidence. Therefore, a necessary condition of epistemic injustice is that testimony is discounted in an epistemically unwarranted way. In short: Without epistemic mistake, no epistemic injustice.

This has a few noteworthy implications. For instance, defendants in criminal trials often provide only one-sided exculpating testimony. This is their good right. In some jurisdictions, defendants are not even prohibited from lying. But it may then be a useful epistemic strategy of courts to factor this into deliberations and diminish the weight accorded to defendants’ testimonies. In medicine, epistemic practices privileging physicians’ over patients’ views are lamented as instances of epistemic injustice (Kidd and Carel 2017; Scrutton 2017). Indeed, if patients’ reports about symptoms such as experiencing pain are unduly discounted, this may often be epistemically fallacious (Gallagher, Little, and Hooker 2021). But it is also true that physicians are epistemically privileged with respect to diagnosing and treating disorders. Given their expertise and experience, they are often warranted in discounting testimonies. Discounting testimonies of groups like patients cannot be dismissed as unjust without giving thought to the logically prior epistemic question. Medicine might be more compassionate and less objectifying if physicians listened more often, but the epistemic injustice of not listening to disordered or delusional thought is not evident.

2.3 Harms to Epistemic Subjects

At the heart of the concept of testimonial injustice lies the contention that insufficiently conferring credibility harms and wrongs the testifier, over and above harms to hearers (including courts) and obstructions of their epistemic aims. Without such harms to testifiers, testimonial justice would collapse into epistemic mistakes of hearers. Although harms to testifiers seem plausible, it is worth asking wherein they lie more precisely. Fricker distinguishes between primary and secondary harms. Secondary harms are downstream negative effects to testifiers such as them losing a court case. Secondary harms also encompass psychological harms to epistemic subjects, such as loss of confidence in their beliefs, which may result in loss of knowledge, distrust in their epistemic abilities, or a general sense of exclusion from collective epistemic. This can impair self-trust and self-esteem and lead to reduced motivation and engagement in epistemic activities. These secondary psychological harms are empirically observable entities that may (or may not) exist in specific cases, and further research should be devoted to examine them. As a first approximation, one may ask oneself how it feels to be disbelieved. It has a distinct quality, it is a devaluation of the worth of one’s assurance (Ferzan 2021). As Anscombe once remarked, “it is an insult and it may be an injury not to be believed” (Anscombe 1979, 150).

Fricker’s account, however, is mainly interested in primary harms, the “essential harm that is definitive of epistemic injustice” (2007, 44). It is explained in several loose descriptions. Credibility discounting is allegedly an intrinsic injustice that disrespects testifiers, degrades and dehumanizes them, may constitute objectification (2007, 132), or with Kantian tones, treats others as a “mere source of information” (2007, 134). These concepts and their supposed wrongness require more explication than Fricker delivers. Objectification is a notorious multifaceted concept, and people often treat each other merely as a source of information without raising concerns. That basic cases of credibility discounting qualify as wrongs of this gravity is not self-evident. In a legal understanding of degrading treatment or violations of dignity, the great majority of credibility discounts in courts would not attain the required level of severeness. The most promising suggestion to elucidate primary harms is that people owe others recognition as full epistemic agents and that unwarranted discounting fails to live up to this (Congdon 2017; Fricker 2017). The concrete harm is then what attributably results from the hearer’s breach of his epistemic duty and the corresponding disappointment of speakers’ expectations. This argument is plausible, but presupposes equal participation in an epistemic community, in which everyone owes, and is owed, full recognition as an epistemic subject. But which domains of society are, or should be, structured in this way? Academia might be an example, but hardly serves as a model for other domains. It needs to be shown for each context why people owe treating others as full epistemic subjects (Maitra 2010). Accordingly, defining primary harms is an open challenge, they are not as evident as they might appear (this is often noted in the literature, cf. Congdon, 2017; Pohlhaus, 2014; Wanderer, 2012). For this reason, the following focuses on the supposedly secondary psychological harms to epistemic subjects, including negative effects on epistemic capacities. They are empirically verifiable and may suffice to turn a mere epistemic wrong into a harm and thereby, into a form of injustice.

Herewith, we have arrived at a modified and broader understanding of epistemic or testimonial injustice that may speak to legal issuesFootnote 5: the epistemically unwarranted discounting of testimony, due to group generalizations, causing psychological and possibly further harms to the testifier, irrespective of harms to hearers or the aims of the epistemic practice. It might be schematized like this:

Testimony → epistemic mistake by hearer due to group generalizations → reduced credibility → harm to testifier as epistemic subject → potential further harms to testifier.

With this understanding in place, we can turn to the law.

3 Epistemic and Procedural Justice

How does the law, especially criminal procedural law, relate to the idea of testimonial justice? There is of course not the law but various legal systems operating by different rules. Because inquisitorial and adversarial systems may differ in this regard, the following remains on a more general plane: The law does not explicitly endorse concepts of epistemic or testimonial justice but contains numerous related ideas. The first point to note is that many legal systems address the key philosophical example—discounting testimony because of race or gender—as instances of discrimination due to protected characteristics. With varying details, many human rights treaties and constitutions prohibit discrimination on enumerated grounds such as race or gender.Footnote 6 Credibility discounting then appears as one instantiation of a much broader phenomenon, which is generally considered to be wrong.Footnote 7 Cases of discrimination may therefore fall short of demonstrating the need or theoretical usefulness of introducing another distinct type of injustice (although it may be useful for practical purposes to mark the problem with a special term). These cases are thus left aside in the following. More interesting are cases of testimony discounting due to non-protected characteristics, from attractiveness or unfortunate facial features of testifiers to social properties, such as class or having undergone trauma treatment, or practices such as introducing the past criminal record of a person to lower her credibility in front of a jury. Whether they are epistemically warranted could be examined for every such practice. Legal systems, however, rarely do so directly because they have their own epistemology and resulting norms for establishing “legal truths”. Epistemic justice is not among its explicit concerns. Applying it to the law requires to show how it relates to established legal objectives, norms, or principles.

The main objective of criminal procedures and trials is epistemic: finding the truth for grounding correct legal decisions or correct outcomes. For this reason, triers of facts such as juries or courts (for simplicity henceforth “courts”) are required to adequately assess evidence, including testimony. This means: in the way most likely to reveal the truth, or in philosophical parlance, according to standards of epistemic rationality. Numerous procedural principles serve these legal epistemic objectives: Judicial impartiality, due process, fair trial, audiatur et altera pars, the general right to equal treatment, and more. The general duty of courts to find truth entails the more specific duty to avoid unwarranted deflation of credibility and testimony. Accordingly, one may say in this sense that criminal procedural law is already committed to the idea of testimonial justice because, and insofar as it aligns with its own epistemic objectives;  testimonial justice is one aspect of procedural justice.

However, the epistemic objectives of the law are restricted by countervailing normative considerations such as fairness, e.g., illegally obtained evidence might be inadmissible although it would advance epistemic ends. Among those considerations, the motivating concerns of testimonial justice—primary and secondary harms to testifiers—are not very pronounced and are only partly covered. Especially in criminal law, fairness considerations predominantly pertain to defendants, not witnesses (fair trial guarantees such as Article 6.1 of the European Convention on Human Rights refer to defendants only). The interests of ordinary witnesses play a negligible role in the principles of criminal procedural law. The situation of victim-witnesses is usually better, as they enjoy some procedural rights and various protective norms apply to them. For instance, some jurisdictions allow hearing the testimony of alleged victims in the absence of defendants to prevent retraumatization. Likewise, the US Federal Rule of Evidence 412 prohibits evidence of previous sexual conduct to avoid embarrassment. These rules may also have beneficial epistemic effects as distortive influences on witnesses are screened off. Nonetheless, one may say that in general, procedural justice furthers testimonial justice of witnesses rather as a side-effect in the pursuit of its own epistemic objectives, and provides only limited protection to witnesses against harms resulting from testifying.

However, that might not be the whole story; a stronger case might be made at least in some jurisdictions. Some procedural principles such as the right to a fair hearing may apply to witnesses as well. For instance, Article 10 of the recent EU Victim Directive guarantees victims a right to be heard (Directive 2012/29/EU). A similar provision can be found in Article 5 of the US Crime Victims’ Rights Act (18 U.S. Code § 3771). From there, it is only a small step towards testimonial justice in a weak form. A right to be heard supposedly entails a right to an epistemically adequate assessment of what one says. To this degree, the law accepts testimonial justice for witnesses. And this small step might be supported by general principles of procedural law. An impartial, unbiased court that hears every side but favors none, audiatur et altera pars, and the careful and correct evaluation of all testimonies before decisions are made seem to be guiding guarantees of the law, enshrined in its imaginaries and symbols such as Justitia. At this level of principles, both defendants and witnesses are entitled to an epistemic correct assessment of their testimonies. As a consequence, testimonial justice, at least in a wek sense, may be considered to be an implicit principle of procedural law, if only in an Dworkinian sense because it provides the best explanation of various established procedural rules such as the right to be heard. Evaluating this argument would require further examination of concrete norms in specific jurisdictions, but it enjoys some plausibility.

The scope of testimonial justice as a procedural principle, however, is much more limited than its philosophical progeny. It is only one among many competing procedural principles. Violations do not provide grounds for legal actions such as challenging decisions or granting specific remedies, as they usually require showing that outcomes of decisions are wrong. Testimonial justice is thus subordinate to the epistemic aims of legal procedures: if outcomes are correct, it does not matter. In other words, the law does not conceptualize testimonial injustice as a wrong in itself, as an independent matter, distinct from the correctness of the outcome of a legal decision.

It is further worth noting that the law often treats witnesses as “mere sources of information”, a key aspect of Fricker’s concept of testimonial justice. While the Kantian phrase is concededly vague, the way legal systems treat witnesses may qualify. Witnesses can neither freely decide whether, when, or how they testify, nor what their testimony is about. With the coercive powers of the state, they are summoned and obliged to testify. However, they might not be listened to at all if courts prefer other sources of evidence. Their credibility is assessed without their involvement; whether their testimony is accepted and how it feeds into the general knowledge production process (deliberation and decision) is not directly explained to them and is only contingently addressed at all, depending on its relevance to the decision. The only form of recognition that witnesses receive is typically a note of thanks when they leave the stand. This may-well amount to treating a person as a mere source of information. Put plainly, the law is only interested in witnesses to the extent that they provide useful information. However, this treatment is not generating major ethical concerns. This casts doubt on Fricker’s construction of the primary harm of epistemic injustice along Kantian lines as disrespect for epistemic subjects. The legal treatment of witnesses might, of course, be wrong, but the widely established practice does not indicate this. The large majority of people summoned as witnesses in trials accepts that testifying is not about them but part of an epistemic endeavor to which their contribution is merely instrumental. The wide acceptance of the practice of trials thus provides a counterexample to the idea that being treated as a mere source of information is inherently wrong.

In conclusion, the idea of testimonial injustice is not alien to law because it is committed to avoiding discrimination and epistemic mistakes as part of its search for truth. Insofar as the right to be heard for witnesses exists, it supposedly includes a fair evaluation of what they say. To this extent, the core of testimonial justice—the epistemically rational assessment of testimony—may be considered an implicit principle of procedural law. But its scope is much smaller than fully respecting witnesses as epistemic agents as envisioned by Fricker. More generally, the law’s epistemic aspirations are restricted by other considerations. It primarily aims at delivering correct decisions, its central institutional obligation, and seeks to avoid detrimental consequences arising from false decisions for affected parties, with a strong bent towards defendants. The law does not pursue its epistemic aims at all costs, and to the degree it does not, the possibility of epistemically mistaken assessments, and thereby, epistemic injustice, emerges.

4 Pre-Trial Therapy and Recovered Memories

This divergence between epistemic aims and other considerations shall be examined at one example, testimony following trauma therapy (for an overview, Marzillier 2014). Since the 1980s, courts worldwide have dealt with spectacular criminal cases of alleged childhood sexual abuse, many of which resulted in acquittals. As it turned out, the testimonies of (alleged) victim-witnesses were often unreliable. They genuinely thought to remember events that never happened. Given the gravity of the accusations, this was hard to believe. Witnesses testified bona fides, did not intentionally lie, and were often confident in their memories. Nonetheless, many of them were apparently false, created through suggestive interactions in therapy or police interviews, sometimes involving hypnosis. As a consequence, courts grew suspicious of testimonies based on memories affected by therapeutic methods explicitly addressing the traumatic event (“trauma-focused” methods, in contrast to stabilizing methods).

This seemingly bizarre phenomenon sparked the field of false memory research, still thriving today (Loftus 2017). It demonstrates the malleability of human memory, often through surprisingly simple or subtle influences. In laboratory experiments, people can be made to falsely “remember” autobiographical events such as getting lost in a mall as a kid in the famous experiment by Loftus and Pickrell (1995), recently replicated by (Murphy et al. 2023), or stealing something (Shaw and Porter 2015, but see Wade, Garry, and Pezdek 2018). False memory research has changed the general view on memory. Memories are not stored in a fixed and static engram, but transform over time, also at the neurobiological level. Simply recalling memories may alter them as they are updated in light of current knowledge (“reconsolidation” thesis, Lee, Nader, and Schiller 2017). This carries the unsettling implication that many of our memories might be false to some degree, although it is currently not possible to reliably estimate their prevalence (Brainerd and Reyna 2005).

The same is true for the potential of trauma therapies to induce false memories. To some, “memory editing” is about to become a general tool of psychotherapy to be applied to various conditions and disorders (Phelps and Hofmann 2019). However, memory-distorting potential varies among treatment modalities. While some studies suggest that implanting is relatively easy (Shaw and Porter 2015), reviews estimate a success rate between 15% (Brewin and Andrews 2017b) and 30% (Scoboria et al., 2017). Key factors contributing to the formation of false memories are imagining fictitious events (“imagination inflation”), by giving encouraging feedback, providing false information about such events from authorities, or showing doctored evidence (“misinformation paradigm”), together with mental vulnerabilities, the need to explain mental symptoms, and the relief that even false narratives and “therapeutic truths” may bring. In the heyday of the controversy in the 1990s, methods that suspected dark past events needed to be uncovered through memory works were apparently popular (Lindsay and Read 1995), but likely contravene today’s professional standards. However, the memory-distorting potential even of common current methods such as the Eye Movement Desensitization and Reprocessing (EMDR) technique is disputed (Otgaar et al. 2021; Van Schie and Leer 2019). A recent study using imaging methods failed to detect alterations in healthy volunteers (Ganslmeier et al. 2022). It is further not clear how deeply people believe in the truth of implanted memories, or whether they may be reversed (Oeberst et al. 2021). Above all, it is not clear whether laboratory studies that seek to create false memories are generalizable to current therapeutic methods and real-life trauma (McNally 2017).

For the law, two constellations are especially problematic. The first concerns reappearing details of events which people remember in other aspects all the time, e.g., memories of a different perpetrator, further instances of abuse, specific practices; in these cases the details, not the occurrence of the event, are in question. The second constellation concerns memories about events that were supposedly forgotten or could not be remembered in total for a long time and have become accessible subsequently in the context of therapeutic interventions or through other triggers, so-called “discontinuous recovered memories” (sometimes called “false memory syndrome”, Kihlstrom 1998). They are the main topic of contention and the present focus. In contrast, it is not controversial that people may actively avoid thinking about events and evade associated triggers to prevent involuntary remembering, or that people may reinterpret the meaning of remembered events and thereby realize that they had experienced abuse only many years later.

Even after three decades of the so-called memory wars, fundamental questions about discontinuous recovered memories remain unresolved to a surprising and unsettling degree. The empirical discussion is too complex to rehearse here, and experts are sometimes bitterly divided (for an overview, see Belli 2012; Brewin and Andrews 2017b; Scoboria et al. 2017; Patihis et al. 2018).Footnote 8 To cut a long story short: It is not even established whether discontinuous recovered memories of sexual abuse exist (apart perhaps from a few exceptional cases). In (unrepresentative) public surveys, a small but significant percentage of people reports having experienced recovered memories of abuse themselves (6.3% in France, Dodier and Patihis 2021; 5% in the US, Patihis and Pendergrast 2018). Most of these people never underwent therapy, and their memories often reappeared in prosaic moments such as watching documentaries or playing with children. These self-reports might of course be largely mistaken. A sizable part of researchers denies the existence of recovered memories because, so goes a pointed slogan, “sexual abuse is not forgotten”, and even if it were, it is unlikely that it is truthfully retrievable through therapy. In this perspective, repressed memories appear like a “myth” (Loftus and Ketcham 1996), or psychological “folklore” (McNally 2004), and most self-reports supposedly originate from memory distortions. It is commonly found among experimental memory researchers who point to many unsuccessful attempts to create recovered memories in the laboratory (Patihis et al. 2018; Patihis, Ho, et al. 2014a, b; Loftus and Ketcham 1996). By contrast, clinicians and therapists who work with patients report prima facie plausible cases of truthfully recovered memories from their practice (Brewin and Andrews 2017a; Patihis and Pendergrast 2018; Van der Kolk 2015; Yovell, Bannett, and Shalev 2003). Courts have accepted the existence of truthfully recovered memories, and corroborating evidence has been found in several cases.Footnote 9 The traditional explanation for recovered memories is that painful or burdensome memories can be blocked or repressed by subpersonal mechanisms. The unconscious or the body “keep the score” and enable retrieval under safer conditions, which trauma therapy aims to create (“where Id was, Ego shall be”). The Freudian idea of repression is highly contested (Erdelyi 2006; Patihis, Lilienfeld, et al. 2014a, b; Brewin 2021; Akhtar 2020), but it is not the only explanation. An alternative suggests that a subset of supposedly recovered memories may not have been inaccessible in a strict sense; they simply were not accessed by persons although they could have been (McNally and Geraerts 2009). People suppress rather than repress memories, and report this imprecisely as an inability to remember. This indicates that the phenomenon of recovered memories may encompass several subtypes, which speaks against generalizations.

In the background of the dispute stand competing schools of psychology; differences among experts extend to views on related topics such as dissociation or dissociative amnesia (Dalenberg et al. 2020; Merckelbach and Patihis 2018). All sides invoke sympathetic higher-order narratives: insisting on the primacy of science over ideology and self-deception on the one hand; ending the distrust of women and the silencing of abuse victims on the other. Both narratives often provide useful epistemic heuristics but tend to polarize debates in this case and seemingly create myside biases, making proponents fail to see valid arguments of opponents.

From a distance, the situation might be summarized as follows: Beyond doubt, false memories exist and are implantable via therapy, including memories of childhood abuse (Otgaar et al. 2022); particularly suggestive methods can be identified. Discontinuous recovered memories were repeatedly proven false. However, this does not seem to warrant the generalization that no truthfully recovered memories exist: the falsity of some does not entail the falsity of all. Prima facie plausible case reports suggest the contrary. The existence of recovered memory is not refuted by lab studies; in fact, it might never be, as it may require proving the non-existent. This leads to the broader question about the status of different types of evidence or “knowledge” in the law. Controlled experimental studies, the gold standard in science, fail to deliver proofs of recovered memories. However, people assert that they have experienced them, corroborated cases exist, and experts in the field accept the construct and its explanatory power. Surely, history of science has shown that such constructs might turn out to be non-existent entities. Conversely, however, it might be the case that recovered memories touch on methodological limits of experimental studies of the human mind. Mental states and processes are naturally hard to capture due to their peculiar metaphysical status, and this elusiveness is aggravated for consciously inaccessible elements. Prospective surveys encounter the problem that repression and dissociation might be rare phenomena that occur only under grave conditions, which makes them hard to detect and prohibits to recreate experimentally for ethical reasons. Retrospective corroborated reports might then be the best evidence possible. Such reports exist, but the corroborating evidence is always susceptible to doubts.Footnote 10 This is the messy epistemic situation in which the law operates.

It is further worth noting that it creates a dilemma for trauma therapy with the mental health of patients as one horn and justice and credibility as the other (Ellison and Munro 2017). Should witnesses delay therapy until the end of the legal proceedings (months, years)—or should they undergo pre-trial therapy, compromising their chances of a favorable outcome and legal remedies? The sooner therapy commences, the higher the chances of preventing trauma-related mental health problems such as post-traumatic stress disorder (PTSD). The law usually does not bar witnesses from pre-trial therapy, but their prospects of being believed diminish. For instance, the guidance on Provision of Therapy for Vulnerable or Intimated Adult Witnesses by the UK Crown Prosecution states (2002, 3.4.): “The key issue regarding pre-trial discussions of any kind is the potential effect on the reliability, actual or perceived, of the evidence of the witness and the weight which will be given to in court. Pre-trial discussions may lead to allegations of coaching and ultimately, the failure of the criminal case.” This dilemma has implications for the ethics and methods of psychotherapy. Suffice it here to note that best practices to avoid memory distortions in therapy should be developed and that patients be informed about potential detrimental effects of therapy on their credibility (Patihis and Pendergrast 2018).

5 Legal Assessment of Post-therapy Testimony

Consider the following case: Witness W testifies about having been the victim of a criminal offense. She did not think about these events for many years and apparently did not know that they had occurred. She came to realize them after memories became accessible to her following trauma-focused therapeutic interventions. Her testimony meets usual standards of credibility. The records documenting the therapeutic sessions do not indicate severe manipulative or suggestive interventions by therapists, but do indicate the repeated use of standard methods suspected of having memory-altering potential.

In trials, numerous further aspects may play a role, such as the intensity and vividness of the memory or how it came back, but none of them can provide a clear indication of whether memories are veridical or false. Suppose no corroborating evidence exists. A court must then determine whether the offense happened based solely on the testimony. In assessing it, many factors become relevant; “credibility” describes some of them but is not used uniformly. It may narrowly refer to the trustworthiness of the testifier as a personal characteristic, or more broadly, as to whether the testimony is believable given its properties (content, comprehensiveness, contradictions, unexpected errors, richness, etc.). This may change in light of other evidence. The latter sense is relevant in the following. Credibility is a gradual concept. The task of the court is, as Fricker writes, to “match the level of credibility … to the evidence that [the witness] is offering the truth” (Fricker 2007, 19). While the direction is clear, the route is vague because a “precise science” for doing so is lacking (Fricker 2007, 18).

Most legal systems do not have binding rules for the evaluation of evidence, but there are helpful illustrative models, among them Bayes Theorem (Fenton, Neil, and Berger 2016). Details aside, the present question can be put in Bayesian terms as the degree to which a piece of evidence—W’s testimony—should change the belief of the court about the proposition that the event happened. Belief is here understood as a gradual concept, since propositions can be believed to stronger or lesser degrees. Call the level of belief “LB”. Testimony can increase or decrease LB, or leave it unaffected. An increase in LB through W’s testimony may suffice to pass the beyond reasonable doubt threshold, as criminal convictions in word-on-word cases based on a single testimony are possible. However, the increase must be very high (and not offset by countervailing decreases). Let us examine seven idealized options for how courts could deal with W’s testimony through this lens:

  1. 1.

    No credibility/inadmissible: Courts may hold post-treatment testimony to be inadmissible, especially regarding discontinuous recovered memories, with the effect that such testimony is not heard. Sometimes supposedly tainted testimonies are excluded in pre-trial “taint hearings” (Cheit 2014). Some legal scholars (Malone, 2020) as well as memory researchers demand that “testimony based on repressed memories (or ‘dissociated’ memories related to so-called dissociative identity disorder) should be banned universally from all courtrooms, without the need for any preliminary hearings” (Pendergrast 2017, 406; also Berkowitz and Loftus 2013). According to this option, W’s testimony cannot affect LB; absent other evidence, the case fails.

  2. 2.

    Reduced credibility: Alternatively, courts may hear the testimony but accord it lesser credibility and weight than ordinary testimony because of its potential distortion. It may thus increase LB, but less than testimony without previous therapy would. This effect can be attained on different procedural routes. Courts can simply discount testimony, or juries be given special instructions to be more suspicious (Loftus 2018, 6).

  3. 3.

    Presumption of falsity. A third way is to presume that post-therapy testimony is false by default. The presumption is defeasible if the testimony stands up to special scrutiny or other pieces of evidence support it. This is roughly the approach of German Criminal law; and it might be considered a corollary of the presumption of innocence of defendants.Footnote 11 Unless the presumption of falsity is defeated, LB remains the same or decreases (a false testimony of a witness may lower court’s belief).

  4. 4.

    No positive verification (no high credibility). The main method for credibility assessment in numerous European civil-law countries is Statement Validity Assessment. It presupposes nuanced differences between testimonies that are made up or based on actual lived experience; experts may detect these differences by systematically screening them (Volbert and Steller 2014). If testimonies pass this test, LB increases considerably. However, the screening method cannot detect false or recovered memories arising from suggestions or misinformation, because people experience them as genuine memories. W’s testimony cannot be verified, and thus it does not significantly increase LB.

  5. 5.

    Unfettered credibility: The testimony is accorded the same level of credibility as the testimony without treatment. LB increases to the ordinary degree, and therapy is effectively disregarded.

  6. 6.

    Other additional screening methods. Post-therapy testimonies can be screened with additional methods and subjected to a deeper level of scrutiny than ordinary testimonies, effectively treating them differently. Changes in LB depend on the outcome of additional scrutiny.

  7. 7.

    Expert witnesses. A variation of option 6 compatible with the foregoing options is to hear additional evidence on the testimony from forensic experts. It seems that virtually all jurisdictions resort in some way or another to experts in difficult cases to which recovered memories belong.

Actual judicial practices in many countries likely resemble these idealized options but have further detailed rules and practices, such as the US Federal Rules of Evidence, which must be left aside for the sake of trans-jurisdictional dialogue. How courts should assess and weigh evidence may also vary between legal systems, but many differences can be bracketed for present purposes. These options have to be examined in light of the idea of testimonial justice—the epistemically correct assessment of testimony. In the absence of a worked out theory of epistemic rationality, we can draw on some uncontroversial epistemic assumptions.

Conferring no credibility, option 1, seems epistemically unwarranted. It results in disregarding any post-treatment testimony. However, given the expert dissensus and prima facie plausible reports of truthfully recovered memories, it cannot be assumed that all memories—or even the great majority—appearing post-treatment are wrong. Accordingly, not hearing testimony is epistemically unwarranted and may constitute a form of epistemic injustice, even of an aggravated kind, a form of silencing witnesses.

Conversely, unfettered credibility, option 5, is equally implausible. Courts must acknowledge that a piece of evidence was subjected to potentially distortive influences. In the wording of process reliabilism (Goldman and Beddor 2021): The process that generated the memory on which the testimony is based is unreliable; its potential to distort outcomes has been proven many times. The unreliability must be reflected in the assessment of evidence. Not doing so would increase LB too much (credibility excess).

Options 2 to 4 negatively affect the credibility of post-treatment testimony in different ways. The lacking positive verification, option 4, only thwarts chances of increasing LB, whereas both options 2 and 3 lower LB. Importantly, the regular consequence of these three options in cases in which the word of one stands against another is that defendants are acquitted because LB falls short of the reasonable doubt threshold. However, recall that outcomes of trials are not the primary concern for epistemic justice at least in the philosophical sense; its main interest is the correct (non-discriminatory) assessment of testimony, independent from further downstream effects. Assessing options 2 to 4 is challenging without a fuller theory of epistemic rationality. The inability to positively verify testimonies (option 4) is grounded in practical-methodological problems. Should they indeed be insurmountable with respect to false memories, option 4 seems not to be epistemically mistaken.

A defeasible presumption of falsity (option 3) may work similarly, depending on how easily it is defeated. German law deploys a special credibility assessment procedure (partly akin to option 4): if testimony passes it, the presumption is defeated, LB increases significantly; if it fails, the presumption remains. In general, however, a presumption of falsity for recovered memory testimony is epistemically unwarranted, under the arguable assumption that post-therapy testimonies are not mostly incorrect.

Some discounting (option 2) of post-treatment testimony in relation to ordinary testimony is epistemically warranted. The question is how much. It is important to avoid discounting too strongly (Puddifoot 2020). Two remarks: False memory research has demonstrated the susceptibility of memory due to its reconstructive (rather than reproductive) nature (Loftus 2018). This motivates caution about every memory-based testimony. Emphasizing therapeutic implantation might downplay ordinary malleability of memory.

The rational updating of beliefs in light of evidence is described by Bayes’ Theorem. It seeks to establish the conditional probability of an event given specific evidence. But data to establish the necessary base rates are unavailable. However, a central element—the likelihood ratio—illustrates the applicable logic: The odds under which an event such as recovered memory testimony is observed under different conditions are set in relation to each other. Suppose for simplicity that only two mutually exclusive hypotheses explain such testimonies: H1, recovered memories have formed in response to a real experience of abuse; H2, recovered memories have formed in response to suggestive therapeutic interventions. To determine the likelihood of the hypotheses, one needs to know which percentage of people who have experienced abuse develop recovered memories and testify, and which percentage of people undergoing trauma therapy do so, multiplied by the probability of experiencing abuse or undergoing therapy, respectively (i.e., the absolute frequency of abuse and therapy in a given population). The result is the likelihood ratio (e.g., “the truth of H1 is twice as likely as that of H2”). Courts’ beliefs regarding H1 and H2 should change accordingly. The logic is intuitively plausible: If more people undergo suggestive therapy than suffer abuse, or if the probability of developing recovered memory is higher in this way than in the other, the chances that a given testimony originates from it is also higher, and of course vice versa. Unfortunately, the individual elements are hard to ascertain.Footnote 12 Likelihood ratios would provide practically relevant information for courts and a standard of epistemic rationality to measure court assessments.

Some final words about special scrutiny methods (option 6) and expert testimony (option 7). They lie on a different plane than foregoing options because they provide additional inputs into court assessments. A potential problem with special measures (option 6) is that they may lead to unequal treatment of ordinary and post-treatment testimonies. Unequal treatment is a prime candidate for epistemic injustice. However, epistemic reasons—the increased risk of distortion—suggest and may even demand heightened scrutiny, including through special methods provided they are epistemically beneficial. Such measures (option 6) are then warranted.

Expert witnesses (option 7) are in principle welcome as their expertise increases the epistemic capacities of courts and reduces risks of mistakes. Denying expert testimony with the effect that judges and juries must assess unfamiliar and counterintuitive pieces of evidence on their own is epistemically inferior and increases risks of mistakes. However, the deep disagreement among experts in this specific case creates two problems. The overemphasis on particular positions creates epistemic risks. One-sided testimony must be countered by other experts. If courts hear only one expert, they should appoint persons who are not heavily committed to a particular position in the unresolved dispute. Moreover, some jurisdictions, e.g. in the US, seek to exclude expert testimony that is unreliable under the Daubert admissibility standards. How they are precisely defined and applied with respect to recovered memories cannot be pursued here (see Grove and Barden 2000; Piper, Lillevik, and Kritzer 2008; Dalenberg 2006; Dillhoff 2011), but the main problem is clear. Expert testimony has the power to impress juries and raise LB considerably. As gatekeepers, courts must seek to exclude pseudoscience. This is epistemically correct. The question is what counts as such. Criteria such as “general acceptance in the relevant field” are hard to apply to fields of deep disagreement between competing schools, which all understand themselves as scientific, broadly understood, but differ in their commitments to experimentalism, and positivistic standards of natural science versus hermeneutical science. These questions ultimately relate to ongoing controversies in philosophy of science about the status of psychoanalysis (Grünbaum 1984), connected to unresolved metaphysical questions in philosophy of mind about the nature of the mind and the accessibility and powers of non-conscious processes (Hassin, Uleman, and Bargh 2005; Hassin 2013), which are continuously debated also in light of novel empirical findings e.g. from the neurosciences (Solms 2017). Tentatively, the law should avoid taking an a priori position on such controversial complex matters. Practically speaking, it bears emphasizing that the phenomenon of people experiencing the (re-)appearance of memories might have many causes, some of which are uncontroversial (brain damage, intoxication, forgotten or suppressed rather than repressed memoires, etc.). The debate overly focuses on the status of repression, but it is only one mechanism; its dismissal does not directly speak to the believability of such reports (Brewin 2021). Forensic experts therefore must examine every case of witnesses claiming recovered memories, and the results need to be presented and explained especially to juries in any case, which seems to require expert testimony.

To sum up: From the seven options, (1) and (3) are epistemically concerning as they too strongly discount testimony while (5) is not discounting enough; (2), (4), (6), and (7) seem epistemically warranted.

6 Justification & Error Distribution

The follow-up question is whether epistemically unwarranted assessments constitute an epistemic injustice. There are two perspectives to evaluate it, the rich ethical concept in the modified version and the rudimentary legal procedural principle. In light of the former, they at least constitute prima facie injustices. Fricker’s narrower account does not elaborate on whether injustices are justifiable, as it is restricted to presumably never justifiable prejudicial discriminations. As a legal procedural principle, testimonial justice may come into conflict with other principles; above all, the prevention of convictions of innocent defendants. This is the paramount principle of procedural justice in criminal law; burdens of proof, the presumption of innocence, or the maxim in dubio pro reo are all calibrated to avoid mistakes disfavoring defendants. Accordingly, even if testimonial justice is accepted as an (implicit) principle, current law clearly prioritizes competing defendant-oriented principles.

Against this background, all options (1–7) seem prima facie justifiable, including the epistemically unwarranted ones (1 and 3), since they err on the side of the defendant. The exception is option 5, which errs the other way, but we may leave it aside here. The point is this: Discounting testimony to an epistemically unwarranted extent favorable to defendants is justifiable by principles of procedural justice which not only seek the truth, but also ensure the direction of errors. Philosophical and legal versions of testimonial justice might diverge here. However, procedural justice does not necessitate epistemically unwarranted discounting. One may suggest that the law should refrain from options 1 or 3 and replace them with option 2, which would increase epistemic accuracy. However, legal systems must have some leeway in the trade-off between protecting defendants against errors and epistemic accuracy and may prioritize the former through options such as 1 and 3. A thought experiment: Suppose criminal procedural law stipulates that neither witnesses’ nor defendants’ testimonies shall be believed in the absence of corroborating evidence (there are historical examples). While this is epistemically unwarranted, for the same reason as option 1, such a strict stance might nonetheless be justified by the concern of convicting the innocent. Counterarguments would not be based on considerations of testimonial justice but on whether such a system could still meet the overall aims of social institution, social stabilization, and deterrence. This example shows that even a strict evidentiary standard cannot be dismissed out of hand. Therefore, epistemically unwarranted options seem justifiable, although they might not strike the best balance between affected interests, all things considered. To be clear: Testimonial justice for witnesses could be elevated into a stronger distinct procedural principle of criminal law. However, to the extent that it would curtail defendants’ rights, this causes a substantive shift in the general architecture of criminal law and its orientation towards protecting defendants. If one is not prepared for this—and nothing in the foregoing suggests so—the epistemic unwarranted options are justifiable. Perhaps even more, without corroborating evidence, recovered memory testimonies should regularly not suffice to overcome reasonable doubts, or to establish “clear and convincing evidence” (highly and substantially more likely to be true than untrue) for civil cases.

7 Towards a Legal Account of Epistemic Justice

Where does this leave us? Testimonial injustice sheds light on how legal systems deal with recovered memories and reconstructs often-voiced complaints about disbelief. Some options used in some jurisdictions (1 and 3) lower the credibility of testimony to an unwarranted extent, but they may be justified by overriding aims of procedural criminal justice. The charge of epistemic injustice then finds no traction within law. This might provide a general lesson about applying the philosophical account of epistemic injustice to established epistemic practices in specialized domains which operate under specific conditions and follow specific epistemic norms. Assessing epistemic injustice according to those domain-specific norms comes at the cost of losing much of its critical edge. Maintaining independence from domain-specific epistemic norms, however, has the price of lacking applicability to those domains. In the present case, the law might well concede ethical problems with testimonial justice while at the same time denying the need for reforms.

However, this separation of law and ethics should not be the end of the story. The idea of epistemic injustice may inspire legal reforms in two directions: avoiding harms to epistemic subjects through legal procedures and improving the assessment of testimonies as far as possible without detrimentally affecting other principles of procedural justice. Both are sketched in turn.

8 Preventing Deformations of the Epistemic Subject

A premise of epistemic injustice, at least in the modified version, is that unwarranted discounting may harm testifiers. More would need to be said about the type of harms, their intensity, and the conditions giving rise to them. For today, assume that not being believed potentially causes a variety of harms to persons in their capacities as knowers, which may be pointedly summarized as “deformations of the epistemic subject”. Notably, these deformations may occur irrespective of whether discounting is justified by other considerations of procedural justice or not. In all testimonies, it is to be expected that some are correct while others are false. But when all are discounted, some witnesses are falsely disbelieved and thereby exposed to the risk of deformation through unwarranted discounting. Moreover, in the exceptional case of false recovered memories, witnesses may wholeheartedly believe in the truth of their objectively false testimony. It seems likely that they are exposed to the same risks of harms to their capacity as knowers as witnesses with true recovered memories. Deformations and injustice may thus come apart. Witnesses might be harmed as epistemic subjects without necessarily being wronged. Nonetheless, justice systems should strive to avoid harms to the persons they recruit, especially witnesses summoned with a duty to testify. Governmental obligations to reduce harms as much as possible might derive from (underexamined) fundamental rights protecting mental or psychological integrity.Footnote 13 Recognition of these harms motivates a reformulation of the idea of testimonial justice for legal contexts. One aspect, among others, may be this: Participation in epistemic practices in the law should not have detrimental effects on testifiers as epistemic subjects. This principle protects persons as epistemic subjects irrespective of whether they testimony is evaluated correctly epistemically. It might be limited by other procedural principles. While it may sound self-evident, this principle is not commonly expressed explicitly in procedural law. It dovetails with classic concerns about measures in the criminal justice system such as manipulative or coercive interviewing techniques or memory distortion leading to false confessions (Lackey 2023). Introducing prior criminal convictions to motivate the jury to discount testimony seems to be another practice that is not epistemically warranted and may have negative effects. The principle against deformation may be the foundation for a systematic protection of epistemic agents in the legal context.

A practical way to avoid deformations could be a two-stage assessment of evidence that disentangles credibility assessment from the overall verdict. Courts make a first finding about credibility based on epistemic standards and subsequently operate with the presumption of innocence and related norms. Credibility assessments are then free from norms stacking the cards against witnesses (as per options 1, 3, and possibly 2). Established rules ensuring a fair trial for defendants apply subsequently in full strength. Domestic procedural laws would need to allow a two-stage approach, disentangling might work in some jurisdictions but not in others. It seems impossible when the presumption of innocence is understood as an evidentiary rule that applies to the assessment of each individual piece of evidence (rather than a decision rule setting in after all evidence has been assessed).

Furthermore, the space between credible testimony and reasonable doubts might be exploited more systematically. To lay witnesses’ ears, it might sound self-contradictory when a court states to “believe the witness” who testified against the defendant, but then acquits the latter because of reasonable doubts. The pragmatic implicature of an acquittal, at least in cases where one person’s word is against another’s, is disbelief of the witness. But this might not be the case (Bolinger 2021; Ferzan 2021). Beliefs can be understood as categorical or gradual. In the former sense, consistent with ordinary language, there is belief, disbelief, and perhaps partial belief. In the latter sense (credence), beliefs come in degrees. Testimonies can then be believed in the sense of increasing LB, without necessarily maturing into a full belief beyond reasonable doubt, with a LB ranking above 0.5 but below 0.9 (or wherever the threshold for reasonable doubt lies). Assuming—more empirical research is necessary here—that deformations to epistemic subjects are predominantly caused by being disbelieved, by a credence < 0.5, semantic clarifications may avoid many of them. Of course, the court's level of belief should not be encroached by moral considerations; when courts disbelieve, they disbelieve. But it seems to be the case that everything below the reasonable doubt or clear and convincing thresholds is sometimes conveyed as insufficient to establish beliefs, often understood as disbelief. With careful wording, a gradual understanding can be made comprehensible to witnesses; courts can state that they believe their testimony (in a gradual sense) but fail to rule out alternatives, which is necessary to pass the reasonable doubt threshold. This allows for both acquitting and believing witnesses, and would be consistent with many victims of crimes primarily seeking not retribution but recognition of their experience.Footnote 14 In civil cases with a preponderance standard (> 0.5), this might not be possible, but courts might suspend beliefs.

9 Improving Testimony and Assessment

Finally, returning to Fricker’s theme of group-based disadvantages, there is a structural or systematic dimension that might be put like this: Truthful testifiers suffer disadvantages in credibility because they are treated as part of a group—people who underwent trauma therapy--, in which testimonies of others grounds concerns. The reason for group membership—mental health problems—was neither self-chosen nor deserved. From the perspective of the truthful witness, being treated as part of this group may appear as a form of injustice. However, as courts cannot overcome the epistemic hurdles to identify the truthful testifiers, no improvements benefiting them are possible. However, answers to systematic problems often lie at the collective level. The question is thus whether and how the law could improve the general epistemic situation of all testifiers to advance the value of epistemic justice broadly construed. This aim aligns with the law’s own epistemic objectives. Improvements may unfold, in the phrasing of socio-economic human rights, through progressively realizing a range of measures advancing the quality and assessment of testimony. While no single measure eliminates the problem, several may alleviate it, and new ideas can be developed and tested along the way.

Improvements begin with attitude changes, overcoming the defensive posture the justice system often takes against external criticism by acknowledging what has been argued hereto, that procedural law may—unfortunately, unavoidably, but perhaps justifiably—lead to testimonies being unduly discounted. This and other practices in the justice system may harm witnesses. When stakeholders in criminal justice systems recognize the predicament of witnesses and adjust their behavior towards them, without infringing upon guarantees for defendants, deformations of the epistemic subject might be alleviated. For instance, a more appreciative way of communicating with witnesses or toning down the sometimes-overbearing legal rhetoric to their disadvantage may have beneficial effects. Also, some “character assassinations” painting an overly unfavorable picture of the epistemic character of witnesses as a defense strategy might be reconsidered at the level of professional ethics. At the institutional level, the law may monitor the prevalence of cases in which credibility is negatively affected by pre-trial therapy, and systematically examine how it may improve testimonies, from taping of the first police interview (implemented only in a few countries such as the UK), improving the atmosphere of interviewing (Knowles 2021), addressing the potentially far-reaching problem of biases, to devising novel forms of taking testimony under less burdensome conditions. In particular, novel means to give pre-trial testimony could be created, perhaps supported by digital technologies. Think of an app by the prosecutorial services that enables people to record anonymously video-recorded testimony, which is timestamped and safely stored in a digital archive, accessible to prosecutors only with witnesses' consent (possibly post-therapy). Furthermore, with respect to trauma therapy, research into the memory-distorting potential of specific methods may provide data to better assess post-treatment testimony and to develop better therapeutic practices. It would be desirable that therapists, memory researchers, and legal scholars cooperate to examine whether therapeutic methods that carefully avoid suggestive elements can prevent false memory formation or substantially reduce its risks so that discounting may not be necessary.

These practical suggestions improve the quality of testimony and can be understood as strengthening and expanding epistemic justice as a legal procedural principle. Much might be achieved without undermining rights of the defense or forgoing principles such as judicial impartiality. Most importantly, all of these suggestions align with the central epistemic objective of legal procedure as they improve truth finding. This leads to the second part of epistemic justice in a legal context: Giving and receiving testimony should be improved to enhance the quality of testimony to the largest extent possible without undercutting defendants’ rights or other procedural principles.

10 Conclusion

With substantial adjustments, Fricker’s idea may be profitably applied to the law and might reveal an implicit legal procedural principle, which could be further developed into a future account of epistemic injustice in legal contexts. Its core idea is that people may suffer a distinct form of injustice because their testimonies are discounted to an unwarranted degree. Legal systems wield coercive powers over witnesses and command whether, when, and about what they testify; oftentimes they treat witnesses as mere sources of information. But when witnesses testify, courts have to listen and evaluate their testimony in epistemically warranted ways. This may sound self-evident, but Fricker’s concept makes it salient.

Recovered memories have troubled psychology and the law for three decades and the debate is not settled. From an agnostic perspective, some testimonies about them seem false while others may be correct; the ratio between both is hard to ascertain. Notably, recovered memories are not the only type of memories called into doubt in legal procedures; they serve here as a controversial case example. The law has several options to assess testimonies. Of seven idealized options examined, some are epistemically warranted, others are not. This reconstructs often-voiced complaints of witnesses about not being believed. However, the unwarranted discounting may be justified by overriding principles of procedural law, especially the presumption of innocence and the distribution of risks of errors it expresses. At this point, philosophical and legal versions of epistemic justice may come apart. As long as the basic orientation of criminal law towards avoiding convictions of the innocent is maintained—and nothing in the foregoing provides reasons to the contrary—complaints of witnesses about not being believed are understandable but cannot prevail. Accordingly, under the current premises of criminal law, the idea of epistemic justice may find less traction than some proponents might hope.

However, the examination has brought out two aspects that go beyond Fricker’s account and may serve as starting points for developing a procedural principle of epistemic justice in the law. The first is the avoidance of unnecessary deformations to epistemic subjects through legal procedures, irrespective of their epistemic status. Many problematic forms of influencing witnesses can be captured by this idea. The second is the overall improvement of the quality of testimony and its appreciation by courts through a variety of measures not detrimentally affecting rights of the defense. Several measures are conceivable. In this way, the law might reimagine itself as an institution of epistemic justice without curbing the rights of the defense. Opening this perspective attests to the power and untapped potential of Fricker’s idea for the law.