Perpetrators’ Knowledge: What and How Can We Learn from Perpetrator Testimony?

Testimony is not only a ubiquitous source of evidence in everyday life but a natural institution which plays an important role in jurisdiction, historiography, religion, and cultural tradition. The current discourse on testimony distinguishes broadly between two types of witnesses: the eyewitness as an impartial bystander — a figure which is traditionally important in the legal context but also shapes the paradigmatic figure of the witness in historiography — and the survivor witness, who has experienced a remarkable appreciation in memorial culture, and in the general upgrading of victimhood in the globalized world. What is missing in this dichotomous typology is the figure of the perpetrator as witness. This paper aims to fill this gap by exploring the specific hermeneutic and moral problems that emerge when dealing with perpetrators’ accounts.


I. Introduction: Perpetrator Testimony -A Neglected Concept?
T he witness is a key figure in law, religion, history, and culture. This figure has many faces: the confession of the martyr (from μάρτυς, the ancient Greek translation for 'witness') is probably the most dramatic kind of bearing witness, giving not only her word but her life to state a conviction. Furthermore, eyewitness reports are central pieces of evidence in court or historiography -despite their disputed reliability. Last but not least, we rely on others' testimonies in a plethora of ways, whether we listen to today's news, learn about mundane facts, or ask someone the way to the next tram station. While the figure of the witness has been subject to many cultural changes and appears in diverse configurations, the act of bearing witness itself seems to be a universal 'natural institution' 1 of human culture: testimony is a fundamental social practice to convey or to establish common knowledge.
In the twentieth century, the accounts of survivors of genocide and massive political violence had provoked a reconsideration of this natural institution, and a reflection upon its scope and limits. The very notion of testimony has undergone radical I would like to thank the two anonymous reviewers and the journal editors for their valuable critical comments.
re-examination, reflecting the difficulties and impossibilities of bearing witness to a traumatic event -I refer here to Shoshana Felman's and Dori Laub's well-known phrase of the Holocaust being 'an event without witnesses', 2 and to Giorgio Agamben's declaration that survivor testimonies about the concentration camps entail a lacuna, indicating a specific impossibility of bearing witness. 3 However, or precisely for this reason, testimony is a key cultural term and booming phenomenon todayespecially the testimony of victims and survivors of political violence. Memorial projects such as the Fortuno� Video Archive at Yale or the Shoah Foundation have collected thousands of survivor testimonies from all over the world and made them accessible online for scientific and pedagogical purposes. The voice of the survivor witness is present in literature and art, in documentary film projects, and on the theatre stage. Historiographical works have increasingly included the victims' testimonies into their accounts. Last but not least, the role of victim witnesses in criminal procedures has been significantly strengthened over the last decades. 4 Truth commissions which were established around the world in order to come to terms with political violence in the past are essentially built on the idea of bearing witness as a social form of establishing the truth and overcoming the past and are also a device for democratization and political participation. 5 The ambivalence of testimony -it is both a source of evidence and an ethical and political act -is mirrored in the dual nature of the witness. Giorgio Agamben among others has suggested distinguishing between two figures of bearing witness. 6 On the one hand, there is the figure of the eyewitness, in Latin testis, who is strongly connected to the legal sphere and called upon to establish factual truth. The eyewitness is neither victim nor perpetrator, neither acting nor suffering: he or she speaks from the position of a third party (the word testis comes from terstis, Latin the third), and his or her authority is rooted in impartiality and distance. On the other hand, there is the victim or survivor witness, Latin superstes, who has lived through a catastrophic experience. The authority of the survivor witness is rooted in the fact that he or she has experienced an event first hand and completely. While these accounts are less valuable to establish facts, they are, as Avishai Margalit has pointed out, 'more valuable at telling it like it felt, that is, telling what it was like to be subjected to [...] evil'. 7 Thus, they articulate an appeal to the community and, ideally, trigger an ethical learning process. This dichotomous scheme is quite dominant in the current discourse on testimony. What is completely left out in this scheme, however, is the figure of the perpetrator: the auctor delicti.
Indeed, there seems to be something deeply problematic about the very notion of 'perpetrator testimony' -especially when considering the significance which the concept of testimony has reached in our contemporary culture. Firstly, the positions of the witness and of the perpetrator generally seem to be mutually exclusive: in court, the testimony of an accused person is subject to a different assessment than that of a third person. Generally, any person who is called for as a witness in court has an obligation to appear and give his or her testimony and can be compelled to do so -except the accused person, 8 who has a right to remain silent, that is, a right to refuse to give evidence. The idea behind this is to protect the accused person from self-incrimination (and after all, this person has a legal representative who can articulate his or her views and interests in the trial).
In a judicial context, certain constraints are placed on the use of the testimony of an accused person or even a sentenced perpetrator, but it is not an impossibility as such -and the same may be said in the case of historiography. In contexts in which the notion of testimony is equipped with a normative value, the case is different: bearing witness has become a keyword not only for a social process of truth-finding, but also for ethical learning and political emancipation. 9 Witnesses are endowed with moral authority. Thus, it seems morally highly problematic and even conceptually contradictory to attribute the capacity of bearing witness to the perpetrators, regardless of whether they were torturers or desktop murderers: the emphatic notion of testimony is, explicitly or implicitly, reserved for the victims.
Still, perpetrators do bear witness. They write memoirs, give interviews, they attest in court or truth commissions, and they produce a variety of written or au-dio-visual material to document, testify, or confess their deeds. Nowadays one can observe how the act of testifying and confessing has become an essential part of the perpetration itself, thus perversely mirroring the economy of testimony in the public discourse. 10 Indeed, the sheer amount of perpetrator testimonies which are publicly accessible has made it increasingly necessary to take a closer look at the hermeneutic and ethical problems raised by this material: how do we deal with these utterances, in the contexts of sociological research or teaching, or even generally in the media? As Anneleen Spiessens has observed, 'the very possibility of the killer's testimony is disquieting. How can a person be capable of telling this kind of story and voicing the ultimate transgression?' 11 So far, there have been very few attempts to approach this problem systematically. This is surprising, since perpetrators' testimonies have been widely used as evidence in jurisdiction and historiography, 12 and, for a certain period of time, they have even been favoured over victims' accounts for being more reliable, informative, and accurate. Even for the purposes of pedagogy and the teaching of ethics after the Holocaust, authors have emphasized the value of perpetrators' accounts over those of the victims. In 1966, Theodor W. Adorno famously pointed out the importance of investigating the perpetrator mentality: 'The roots are to be sought in the perpetrators, not in the victims, who were murdered under the most miserable pretences.' 13 Gitta Sereny follows this path in her introductory remarks to her book on Franz Stangl, the commander of the Sobibor and Treblinka Death Camps. Sereny writes: [It is] essential […] to try at least once, as far as possible unemotionally and with an open mind, to penetrate the personality of a man who had been intimately involved with the most total evil our age has produced [and] to assess the circumstances which led up to his involvement, for once not from our point of view, but from his. It was a chance, I felt, to evaluate, through examining his motivations and reactions as he described them rather than as we wished or prejudged them to be, whether evil is created by circumstances or by birth, and to what extent it is determined by the individual himself or by his environment. 14 The newly emerging interest in the figure of the perpetrator and its manifestations follows in the footsteps of these statements. It goes without saying that in order to understand the motivations and intentions of people who have engaged in genocide and massive political violence, the focus is not only on official documents and observations, but also and especially on the testimonies of the perpetrators themselves. The point I want to make and develop in what follows is that dealing with perpetrator testimony requires particular reflections from an epistemological as well as ethical perspective -something which has not yet been systematically addressed, although one can find approaches to diverse aspects of this problem in literature. 15 In my view, the accounts, statements, and memoirs of perpetrators are valid forms of testimony, since they are acts of speech addressed to an audience, and the speakers claim to offer a truth which is rooted in a personal experience. They do not differ from victims' accounts in being fallible sources of historical knowledge. Still, dealing with perpetrator testimony requires specific reflection. As Catherine Coquio points out, the criminal is neither testis nor superstes -he or she belongs to a third category of testimony. 16 This category has been, until now, almost completely undetermined. It is this failure to delineate and reflect upon the concept of the testimony of the auctor delicti that I want to address here. I will concentrate on three aspects that are fundamental to testimony as a social epistemic practice, namely trust, truth, and authority, and argue how these aspects pose specific hermeneutic and moral problems in the case of perpetrator testimony. Let me add three general remarks before starting the discussion. Firstly, thinking about perpetrator testimony requires one, to a certain extent, to put aside the 'morality' or ethics of testimony as a concept, and to take on the stance that testimony is not per se 'good' or ethically valuable but normatively ambivalent, depending on what purpose it serves. Secondly, the recording of testimony is, as I will argue, different from analysing documents and inferring facts from statements. Testimony is a dialogue in which the speaker and the listener both play an active role. Therefore, it is fruitful to examine the nature of testimony as a speech act when working with perpetrator testimonies, since it brings us to reflect upon how the audience affects what kind of knowledge is revealed. Thirdly, the following analysis refers to testimony as a speech act in a very general way, and, as most current philosophical approaches to the phenomenon, it takes the situation of face-to-face testimonies as a paradigmatic case. Yet, one could argue, testimonies obviously occur in many different forms. They are transmitted by different media such as written texts, images, audio-visual recording, and framed by different genres such as documentary film, oral history, memoir, and so on. These modes of presentation surely shape the attitudes and expectations of speaker and audience and thus have a major influence on what is being testified. For a full analysis of the nature and meaning of a concrete empirical testimonial speech act, these media and material aspects have to be taken into account. For instance, it makes a difference whether we are listening to a witness in a face-to-face situation, where they directly address us and invite us to trust them, or whether we are watching a recorded interview or reading a written document, which opens up more possibilities to assess the witness' words critically. Mediated forms can help to create a (critical) distance, but they also can create proximity, for instance, when they give a face to people who are otherwise referred to as anonymous others. 17 To describe the role of different media in the shaping of testimonies and their epistemic and ethical message is an important task, which I will not be able to do justice to here. However, it is not my aim to define all elements that shape the epistemic and ethical meaning of a testimony, but to set out on a more fundamental level some pivotal concepts which govern the giving and receiving of testimony as a social practice. I then proceed to show how a perpetrator bearing witness might contrast or undermine this social practice of telling and trusting. In this context, this article should be regarded as a preliminary sketch, which sets out the cornerstones but leaves a lot of space for more nuanced analysis.

II. Trust
Testimony is a source of knowledge -were it not so, we would hardly 'know' anything about the world, its history and places to which we have never been. However, this knowledge is most fundamentally built on trust: since we are not able to find out everything on our own, with our senses or logical reasoning, we find ourselves dependent on others' narratives in many ways. This epistemic dependence has a social dimension. By giving and taking testimonies and thus sharing knowledge within a community, we are not just compensating for our cognitive imperfection; indeed, the fact that a great part of what and how we learn is based on communication shows how our epistemic life is deeply interwoven with social and ethical dimensions. 18 The fact that trust plays a key role for testimony as an epistemic practice has been a pivotal point in the contemporary debate on the epistemology of testimony. It would be too great a task to outline even the main controversies within this broad and increasingly differentiated debate. For the issue at stake here, however, it is illuminating to pick out one thread of thought which has evolved in recent years and sheds new light on the nature of testimony: the so-called Assurance View or Second-Person-View on testimony. 19 Contrary to the opinion of most epistemologists, and also contrary to the concept of testimony as it is classically conceived in historiography, 20 law, and other disciplines, this view emphasizes that testimony is not only some form of evidence, but a social practice which is based on trust and personal commitment. Testimony is not a piece of information which we assess independently, but a sort of dialogue with normative social implications. Philosophers who advocate for this theory argue that testimony as a source of knowledge 'works' fundamentally different to perception or logical reasoning, since it is based on communication, or to be more precise: on the interpersonal relation of telling and being told. Richard Moran, for example, points out that testimony is a speech act and involves a special commitment on the side of the speaker: '[t]elling someone something is not simply giving expression to what's 18 For a philosophical analysis of testimony as an epistemic source, see C. A. J. Coady, Testimony: A Philosophical Study (Oxford: Clarendon, 1992 . 20 One could object that this narrow understanding of testimony as merely a form of empirical evidence is long overcome in historiography, especially since the oral history approach has brought into view the inter-subjective dimension and the value of subjective memories in testimony. Still, historiography ultimately sets the focus on testimony as a historical source (aiming at knowing and understanding the past) and less on its inter-subjective and performative dynamics per se.
on your mind, but is making a statement with the understanding that here it is your word that is to be relied on'. 21 By making himself or herself accountable for the truth of what is being said, the speaker gives the audience a reason to believe, 'conferring a right of complaint on his audience should his claim be false'. 22 Testimony entails a promise. However, Moran does not only highlight the fact that an ethical commitment on the side of the speaker is crucial for his or her utterance to be a valuable source of knowledge. He also emphasizes that the hearer plays a constituent part in the language-game of testimony: if testimony is to 'work' as a successful communication of knowledge, the hearer also has to engage in a certain way. He or she has to accept the promise that the speaker offers, thus acknowledging the speaker's sincerity, competence and intention to speak the truth: 'when the hearer believes the speaker, he not only believes what is said but does so on the basis of taking the speaker's word for it'. 23 The Assurance View highlights the fact that testimony has a dialogical character: testimony is an act 'addressed to another person'. This dialogue has a normative dimension: in telling and being told, accepting or refusing someone's word, both speaker and hearer engage in a normative relationship of trust. 24 He who trusts also risks disappointment -it is above all in the cases of failure that the normative and ethical dimension of testimony becomes evident. For instance, upon finding out that my colleague told me things about the last meeting which were not true, I will not simply accept this 'misinformation' with a shrug. I might be disappointed, even feel disrespected, and I will probably call him to account. On the other hand, if my colleague experienced me rejecting his testimony about the last meeting, he would feel irritated or even angry since my refusal to believe him amounts to a refusal to acknowledge his competence and authority. The interpersonal relation of telling and being told is fallible and fragile; nonetheless, it constitutes a great part of the web of beliefs we hold about the world.
According to this view, knowledge and acknowledgement are deeply interwoven in the practice of bearing witness. As Miranda Fricker has pointed out in her work on 'Testimonial Injustice', the experience that one's own capacity to give knowledge 21 Moran,p. 8. 22 'When all goes well in testimony, a speaker gives his audience a reason to believe something, but unlike other ways of influencing the beliefs of others, in this case the reason the audience is provided is seen by both parties as dependent on the speaker' s making himself accountable, conferring a right of complaint on his audience should his claim be false' (Moran,p. 21). For a further development of this thought see McMyler, who argues that, by accepting testimony, a hearer entitled to 'pass the justificatory buck' back to the authority. 23 Moran,p. 2. 24 '[I]n testimony in particular the kind of reason for belief that is presented is one that functions in part by binding speaker and audience together, and altering the normative relationship between them.' (Moran,p. 20) is denied 'can cut deep'. 25 According to Fricker, the capacity to give knowledge -to be a good informant for others -is part of our capacity for reasoning, and thus a significant part of what it means to be human. The witness' appeal to be believed can thus also be understood as an appeal to be acknowledged socially in a certain way. Especially when it comes to survivor witnesses testifying about atrocities perpetrated against his or her own body, we are reluctant to reject or even question these reports, since we are aware that this would deprive the speaker of a sort of social acknowledgement. However, viewing testimony as a social practice does not necessarily imply abandoning the epistemological, critical attitude. To acknowledge the survivor witnesses in this way does not mean to regard them as living traces and monuments of the catastrophe whose testimonies are above criticism, but rather it means to take them seriously as givers of knowledge, and this does not preclude a critical attitude towards their accounts.
While the view of testimony as a dialogical speech act may make sense inside the philosophical ivory tower, one might reasonably ask how it applies to the exigencies of working with testimonies in different institutional contexts, such as court hearings or the critique of written or oral testimonies in historiography? In court or in history, testimony is (usually) conceived of less as a dialogue carried by mutual personal acknowledgment, but rather as a piece of evidence which affords critical questioning and a hermeneutics of suspicion. From the perspective of historians or jurists, the Second-Person-View might appear as an ideal but unrealistic conception of testimony, which is hardly applicable to the needs of evaluating testimonies as pieces of judicial or historical evidence. The critical scrutiny needed in these contexts stands in contrast to the conception of mutual, basic trust as depicted under the Assurance View.
As convincing as this objection may be, one can challenge it by pointing out that acknowledging the interpersonal dimension of testimony does not necessarily result in adopting an uncritical stance towards it. The Assurance View describes an elementary linguistic and social practice -that of telling and being told -which is the foundation of learning about anything beyond the realm of our perceptual senses or autonomous logical reasoning. Trust in the words of others is a necessary precondition not only for a great part of our knowledge but also for any critical assessment of testimony. As Ludwig Wittgenstein put it, doubt is only possible on the grounds of 25 'The capacity to give knowledge to others is one side of that many-sided capacity so significant in human beings: namely, the capacity for reason. certainty and acknowledgment. 26 Moreover, being aware of the intersubjective dynamics and normative implications of testimony does not preclude critical reflection, but rather enriches it: a critical hermeneutics of testimony should take into consideration how aspects of recognition and loyalty come into play in the act of telling and being told.
Let us now turn to the case of perpetrator testimony and to the specific problem that arises when this type of testimony is examined in light of the Assurance View. According to the Assurance View, testimonial knowledge is knowledge that is essentially based on an interpersonal relation of trust. As Moran states, the act of giving and receiving the speaker's words binds 'speaker and audience together, [...] altering the normative relationship between them'. 27 In this respect, perpetrator testimony obviously poses a problem: when listening to a perpetrator's testimony, the listener is confronted with a troubling responsibility. If telling is indeed 'inviting to trust', as Edward Hinchman puts it, then the listener faces the dilemma that he or she cannot receive the utterance as a testimony, whilst at the same time maintaining a neutral, objective standpoint towards it. Objectivity is, in the words of Thomas Nagel, the 'view from nowhere' 28 -but receiving a testimony implies taking up a position, entering a reciprocal normative relationship with this person and accepting a certain ethical bond, which is closer than any liaison we have with other random people. Listening to a perpetrator's testimony thus opens up a dilemma: if we decide to accept the invitation to trust and believe the perpetrator-testifier, we risk believing a false testimony. Testimony, of course, is always fallible, yet the risk that a perpetrator gives a false statement is even higher since he or she has a direct motivation to make themselves appear in a more favourable light. Believing in and accepting a perpetrator's false testimony can result in (unintended) complicity: for instance, if the false report is accepted as a piece of historical evidence, it contributes to historical misrepresentation.
If we decide against accepting the perpetrator-testifier's invitation of trust, it is ultimately impossible to learn from his testimony as such, since we are not able to make sense of the account without investing a minimum of trust in the speaker's words. We can analyse it, we can compare it to other sources, but while we do this, we do not treat the utterance as a speech act, but as a symptom: we draw our conclusions from it, but we are not able to learn from it.
Christopher Browning's writings on Adolf Eichmann's memoirs are a striking contribution to the issue of dealing with perpetrator testimonies. Browning makes a case for using Eichmann's testimonies as historical sources and not disregarding this potentially insightful material. However, he obviously feels the need to justify this approach and to propose a strict methodology for dealing with such testimonies. He emphasizes that using Eichmann's various testimonies as historical sources requires a special process of justification and verification since they are very probably a 'conglomeration of faulty memories on the one hand and calculated lies for legal defence and self-justification on the other'. 29 The concrete measures proposed are four test questions which the historian should apply: 1. Is it in the speaker's interest to tell the truth? 2. Does the testimony show 'unusual attention to details of visual memory?' 3. Is the content of the testimony plausible, and are there any other sources that contradict the details? 4. Is it probable, that is to say, are there other sources which support it? 30 Historiography has a tradition of critical hermeneutics when dealing with testimony. Browning's example shows that, to some extent, the testimony of perpetrators requires a more uncompromising application of the traditional methodology. This very cautious approach could not be more different from the emphatic acknowledgement of the speaker's commitment as described in the Assurance View. From the latter perspective, one could even state that Browning does not treat Eichmann's memoirs as testimony at all, but as evidence which has to be carefully analysed and interpreted. Browning's decision not to take Eichmann's word for it, but to counter-check his account with any other possible source, has, of course, to do with the fact that in this case, the testifier is a perpetrator. Browning suggests that we consider Eichmann's statements as traces rather than testimonies: they have to be analysed and deciphered properly. He creates the impression that one can deal with testimony through independent observation and inference, thus avoiding the problem of placing trust in the perpetrator. Nevertheless, the dilemma of trust does not completely disappear when using the historical method, and Browning is very well aware of this. His detailed justification serves only to highlight his unease at the very fact that, in order to make use of the testimonies as historical sources, he has to believe Eichmann to a certain extent.
Testimony is ultimately a social practice, binding speaker and listener together in a special way. This has two implications. Firstly, testimonial knowledge is essentially second-hand knowledge: to learn something via testimony requires accepting the speaker's authority, to rely on it, and to give credit to it. Secondly, the listener plays a constituent part in the acceptance of testimony as a source of knowledge: it is only through his or her accreditation and authorization that the act of testifying is completed. To reflect upon the dialogical nature of testimony is to reflect upon the fact that both speaker and listener are responsible for the 'knowledge' which is transferred or even newly produced in a situation of testimony.

III. Truth
To put it simply, we can gather knowledge and facts about the world through testimony. However, testimony is a discursive practice which is subject to historical and cultural change and is shaped by different institutional frameworks. Diverse cultural forms of testimony offer different sorts of truth: this much is clear when we consider different 'types' of testimony such as the religious martyr, the historical witness, or the survivor witness. Although it is certainly possible to identify a wide variety of testimonial types, each corresponding to a certain type of truth, I want to propose here a very basic, two-fold distinction: the distinction between the external and internal truth of the testimony. The report of an eyewitness in court, for example, is drawn upon to reconstruct true facts (the accident happened at four o'clock at the big junction, when the silver car went through the red light). This kind of truth, which consists of perceptual knowledge based on external events only, corresponds to what I call the 'external truth' of a testimony. At this point, it should be mentioned that eyewitness reports rate as highly fallible sources of factual knowledge -they are usually overwhelmingly inaccurate and unreliable, even when they are given by honest witnesses to their best knowledge and belief. Still, eyewitness reports are often indispensable in order to ascertain factual or external truths. On the contrary, survivor testimonies, as they are recorded in oral history databases, recordings from therapy sessions, or as acts of commemoration, give expression to another sort of truth: they do not only articulate the mere external facts of a historical situation but the internal truth of experience. Avishai Margalit has highlighted the function of recounting internal truth as especially morally relevant: 'a seismograph does not tell us what it is like to be in an earthquake. For that we need a moral witness'. 31 What survivor testimony speci�cally expresses, Margalit emphasizes, is an inner, existential experience. In the following, this shall be named the 'internal truth' of a testimony.
I assume that each and every testimony has both aspects -indeed, it seems to be an essential trait of testimony that it is always a combination of external and internal truths, connecting an outer event to an inner experience and vice versa. However, there are of course types of testimony that serve better for the establishing of external truths, and others that are more apt to express the internal truth of an experience. Victims of political persecution and violence are often less able to reconstruct the precise course of events or depict the details of a crime than persons who were not directly trapped in the violence. For example, journalists can often move more freely in an area of conflict and are generally in a better position to gain information about 31 Margalit,p. 163. certain events and have an 'overview'. However, their reports are less valuable for expressing the internal truth of an event. According to Margalit, this is essentially tied to the perspective of the victim. The interruptions, the silence, and even the flaws within Holocaust survivor testimonies are symptomatic expressions of the internal truth of the events to which they are bearing witness. The victims' voices articulate the characteristic difficulties and even impossibilities to translate experiences of violence and trauma into discourse. An extreme example for this are the 'testimonies' of chronically hospitalized Holocaust survivors, who were recorded by Dori Laub in the first phase of the Holocaust Trauma Research Project conducted at Yale University. 32 What kind of truth can a perpetrator's testimony provide? What kind of truth are we looking for when dealing with perpetrators' testimonies? This question, again, leads to a dilemma, since in both respects perpetrator testimony appears as a specifically problematic source of truth. First, it is a precarious source of 'external truth': it is not only as unreliable as any eyewitness reports, but it is considered to be even more unreliable insofar as the perpetrator usually has a direct reason to conceal the factual truth or certain details. In the judicial context, the accused's right to remain silent is the right not to incriminate him or herself, but it also denotes how the testimony of an accused person has to be treated differently to the statements of bystanders or victims. Is the testimony of a perpetrator generally less reliable than that of a bystander or victim? This conclusion seems unjustified -after all, perpetrators very often dispose of factual knowledge which the victims lack, or to which they are too traumatized to attest. It is not least because of this fact that perpetrator sources have been generally preferred to victim testimonies as evidence in court and in historiography. However, there is a special kind of risk involved when taking perpetrator testimony as a source of factual knowledge, and with reference to Browning's reflections, it seems only rational that one should apply an especially diligent historical or forensic methodology to evaluate it. Some approaches to perpetrator testimony, among them documentaries, films, scientific or journalistic interviews, or exhibitions at memorial sites, aim to capture 'internal truth'. Their focus is not set on the reconstruction of historical facts, but rather on identifying elements in the statements which provide an insight into the attitudes, moral values and motives of the perpetrators, whose actions often seem so motiveless. Moreover, when it comes to capturing 'internal truth', testimonies  icine, 24.2 (2005), 253-65. Laub emphasizes that Holocaust testimony is essentially marked by an absence of narrative. The testimonies of psychotic, hospitalized survivors represent for him ' the ultimate in the absence of narrative' (Laub,p. 257), the '"extreme" on the spectrum of speechlessness and silence' (ibid., p. 259), which in Laub' s opinion is not just caused by schizophrenia but ' an extreme case of the speechlessness of trauma that afflicts victims, witnesses, and those who attempt to be its chroniclers' (ibid.).
of perpetrators appear to be especially opaque, particularly with those involved in violence on a large scale. Catherine Coquio, who argues for applying the emphatic concept of bearing witness exclusively to those who are bystanders or victims, vehemently denies that perpetrators of genocide can be regarded as witnesses of an internal experience at all. 33 Referring to the testimonies of Adolf Eichmann and Rudolf Höss, she states that these men 'testified to no "catastrophe" whatsoever', but rather only reported in an impersonal manner on their 'difficult and demanding job', 34 without ever imagining an internal perspective on the destruction. According to Coquio, Eichmann and Höss's reports lack something which is essential for testimony, namely an authentic first-person perspective on what is going on. For Coquio, this lack of inner perspective is a characteristic feature of genocide itself, for a person who can suppress any kind of emotional reaction in the face of genocide is by definition not capable of living through it herself. Does perpetrator testimony lack the inner truth of testimony altogether? Taking into consideration the many examples of critical field research that is being done, including qualitative interviews with perpetrators, a general rejection does not seem justifiable. There is some kind of inner truth expressed in these accounts. More precisely, perpetrators' accounts may reveal social and psychological truths about themselves. How do they imagine and depict themselves? How do they recount their own personal development? To what kind of moral (or aesthetic) values do they subscribe? What is their psychological and physical cartography of the world like, how do they describe their bodily interaction with the outside world? In stating this, however, I am also suggesting that we must take Coquio's critique seriously: genocidal crimes are essentially linked to denial and destruction of witnesses. This structure of denial and non-witnessing must also affect the subjectivity of the perpetrators and their relation to what they see and do: their lack of inner perspective and truth is not accidental, but a constitutive element of genocide. Thus, Coquio's considerations actually open up a third perspective on the question of truth in perpetrator testimony: the speci�c truth of these testimonies is neither an external, nor an or internal truth, but a negative variable -in other words, a specific lack or absence in their reports. While testimony is always selective, perpetrator testimony specifically requires one to reflect upon what is being said in light of what is being silenced.

IV. Authority
Testimonial knowledge is second-hand knowledge: it is not based on independent inference from someone's words but implies acknowledging the speaker's authority on the attested matter. Accepting testimony is different from accepting a reason or a piece of evidence since it means to authorize and to give credit to the witness. To describe this, Paul Ricoeur has proposed to use the notion of accreditation (French: accréditation), 35 which signifies more than mere acknowledgment, since it has a performative, productive aspect. To give credit is not only a form of assent: it is an act of empowerment, allowing the speaker to exercise a form of rational agency. 36 In the discourse on survivor testimony, the ethical implications of accreditation have been widely discussed: listening to the victims' voices and acknowledging their authority has been conceived as an act of recognition, which can help heal individual trauma. As long as we, as audience, are listening to victims of political violence, we seem to be on the morally right side. But what if it is perpetrators who, by bearing witness, invite us to believe them and to authorize them? There is something disquieting about the accreditation of perpetrators, about giving them the opportunity to articulate their version of history and, above all, about accepting their epistemic authority. In the following, I carve out what exactly is so discomforting here, and whether there is a moral problem with accepting perpetrator testimony. I argue that there is indeed a moral problem, and that this is connected with the notion of authority and its ethical implications.
One Latin word for perpetrator is auctor delicti. Indeed, whether the perpetrator is a Schreibtischtäter ('desk-murderer') or a torturer, this person is the author of a violent transformation of the state of the world, be it through his or her actions or omissions. The deed carries, so to speak, his or her fingerprint. Perpetration has to do with authorship -even if structural political violence seeks to deny and conceal individual responsibility. A comparatively subtle example for this claim of authorship and authority is given by Adolf Eichmann, who, in his fifth and final account following his capture, writes in the style of a historian, citing and quoting documents at length and claiming scientific authority over history. 37 If taking a testimony means to take in someone's point of view and accept this person's authority over a certain matter, then taking a perpetrator's testimony is accompanied by the risk of taking in a dehumanizing, degrading view of the victims -and thus victimizing them twofold. A striking 35 'La certification du témoignage n' est alors complète que par la réponse en écho de celui qui reçoit le témoignage et l' accepte; le témoignage dès lors n' est pas seulement certifié, il est accrédité.' Paul Ricoeur, La Mémoire, l'histoire, l' oubli (Paris: Seuil, 2000), p. 205. 36 For a detailed analysis of epistemic authority, see McMyler. The ethical and political meaning of epistemic authorization is extensively explored in José Medina, The Epistemology of Resistance: Gender and Racial Oppression, Epistemic Injustice, and Resistant Imaginations (New York: Oxford University Press, 2013). 37 See Browing, p.8. example of this problematic implication of listening to perpetrators can be found in Joshua Oppenheimer's documentary film The Act of Killing. 38 The protagonists, among them a man called Anwar Congo, are perpetrators of the massacres in Indonesiathey are the filmmaker's main interviewees and give lengthy testimonies about the 'good old times', while the victims, still frightened of political repression, only appear ghostlike at the periphery of the film's scenes. Invited by Oppenheimer to produce a film about the events in 1965, the perpetrators enthusiastically take on the roles of leading actors and directors. Recollecting their memories, they re-enact scenes of torture, murder, and the destruction of a whole village, in order to bear witness to their historical deeds. In one of these staged film scenes, the audience watches one man being 'directed' to re-enact the role of a tortured victim. Later, this man proves to be a real victim of the political cleansing: his Chinese stepfather was deported and killed, the family was displaced, and the man himself was never able to attend school. After his nightmarish report, he hurries to add that it is just an interesting story that might be integrated into the documentary. The killers react with a shrug of the shoulders and state that not everybody's life story can be represented. The scene becomes unbearable when that very man appears in the re-enacted scene as the victim of torture. Obviously, he suffers a re-traumatization.
As we have seen, the testimonies of mass murderers are often characterized by a certain lack (or denial) of experience and awareness. What is strikingly absent in Eichmann's 'objective' account, as well as in Anwar Congo's boastful stories, is precisely the recognition of their victims as human beings. The potential moral problem in accrediting and authorizing perpetrator testimony seems to be this: by giving former perpetrators the space to recount the story from their perspective, and by giving credit to their authority, there is a risk that the perpetration and the annihilation of the victim are repeated on a symbolic level.
I would like to insist that this moral problem, which can (but does not need to) emerge from listening to perpetrator testimony, should not be 'outsourced' or passed on to moral philosophers. Rather, it is a methodological problem that each researcher has to be aware of and reflect upon when dealing with perpetrators' accounts. How does our listening and believing the perpetrators affect the lives of their former victims? I do not want to exaggerate the morality of the matter, but I am convinced that, ultimately, there is no way of applying a completely neutral approach to dealing with testimonies. 39 This leads to the question of how researchers, teachers, and exhibitors

V. Conclusion
Today, testimony, especially the testimony of victims of political violence, holds important moral and political relevance. Yet, the discourse on the cultural and social relevance of testimony and bearing witness has, so far, considerably neglected the phenomenon of perpetrator testimony. This is a surprising failure since the question of how we should deal with perpetrators' accounts, reports, and confessions is more explosive than ever. Not only has the interest in learning about the perpetrators and their point of view significantly increased in the last few years, but perpetrators have ever more possibilities to express their views, attest their stories, and make these attestations accessible to a wide audience. I have argued that, on the one hand, it is reasonable to put aside the moral aspect when dealing with testimony and to include perpetrators' accounts. Testimony is a dialogue, and looking at perpetrators' accounts as testimonies brings into light the role and the responsibility of the audience regarding what kind of 'perpetrator knowledge' is articulated. On the other hand, I have pointed out that perpetrator testimony requires specific hermeneutic and moral considerations, and I have set out certain dilemmas concerning the issues of trust, truth, and authority. Firstly, I have argued that testimony as an epistemic social practice essentially implies trust in the speaker, and I have outlined in what respect trust in a perpetrator raises a problem which is different from the general problem of testimonial fallacy. When it comes to using perpetrator's testimony as a source of historical knowledge, the listener is confronted with a troubling responsibility. On the one hand, it is most important to maintain a critical standpoint towards this kind of testimony; testimony is always fallible, but a perpetrator's account is regarded to demand special caution since the speaker has a direct motivation to distort the factual truth. On the other hand, the giving and receiving of testimony is a social practice: it is not reducible to the critical examination of evidence or the interpretation of traces. To learn from testimony requires investing some trust in the speaker's words. This position of an addressee of testimony is not easy to bring into accordance with the distanced attitude of an objective investigator. Still, I have argued that the two perspectives are not mutually exclusive. Not only is the critical stance an important antidote 40 to the trusting attitude, but reflection upon the social dimension of testimony is an important element of a critical hermeneutics of testimony. This means that the critical listener should reflect on how aspects of recognition and loyalty play a role in the process of giving and receiving testimony and how the listener is involved in the production of testimony, in order to subject his or her own behaviour to (self-)critical examination.
Secondly, I distinguished between the external and the internal truth value of testimony, and argued that perpetrator testimony is problematic as a source of truth in both respects. When dealing with this kind of testimony, I suggested bringing in a third point of view: one should consider perpetrator testimony as providing insight into the psychological or social structure of denial and of a specific absence of truth and conscious experience.
Thirdly, I have emphasized that learning from testimony presupposes the authorization of the witness and that this act of accreditation is a form of recognition and empowerment. In the case of perpetrator testimony, this brings up a moral problem, since it runs the risk of re-enacting the authoritarian violence and victimizing the victims once again. This moral problem is a methodological problem, and it shows how when dealing with perpetrator testimony, it is particularly necessary to look between the lines for what is covered up, blocked out, obscured, and denied.
These reflections are but a first attempt to sketch some questions about the relationship between perpetratorship and testimony. While both concepts have been intensively discussed, historically framed and systematically contested, their interplay is rather unexplored. This is surprising, considering the fact that perpetrators often bear witness, their testimonies are often used as a source of evidence, and that, in certain respects, these testimonies can indeed be more 'telling' than victims' reports. The specific epistemological and ethical risks of perpetrator testimony need to be analysed further in an interdisciplinary dialogue. This seems urgent more than ever, since the 'era of the witness' has entered the digital age, and expressive violence often goes hand in hand with testimonial acts.
40 I thank one of the reviewers of this article for this conceptual suggestion.