Microsoft Word - IL36.3 Bex BSEDIT.doc © F.J. Bex & V.M. Bex-Reimart BexInformal Logic, Vol. 36, No. 3 (2016), pp. 349-370. Evidence Assessment in Refugee Law with Stories and Arguments F.J. BEX1 & V.M. BEX-REIMERT2 1 Department of Information and Computing Sciences, Utrecht University, The Netherlands 2 Department of Constitutional Law, Administrative Law and Public Administration, University of Groningen, The Netherlands Abstract: In this article, we aim to analyse whether a systematic meth- od for reasoning with evidence in legal cases – the hybrid theory of stories and arguments – can be ap- plied to a novel legal domain, name- ly European asylum law. This analy- sis serves as a case study for testing the applicability of the hybrid theory outside of the context of criminal law. Furthermore, the analysis will provide insights on how the hybrid theory can be used to improve the normative framework for evidence assessment in asylum cases. Résumé: Dans cet article, nous vi- sons à analyser si une méthode sys- tématique de raisonner avec des preuves dans les affaires juridiques - la théorie hybride des histoires et des arguments - peut s’appliquer à un nouveau domaine juridique, à savoir le droit d'asile politique en Europe. Cette analyse est une étude de cas qui cherche à tester l'applicabilité de la théorie hybride hors du contexte du droit pénal. En outre, l'analyse permettra de mieux comprendre comment la théorie hybride peut être utilisée pour améliorer le cadre nor- matif employé pour évaluer des preuves dans les cas d'asile poli- tique. Keywords: evidence, argumentation, asylum cases 1. Introduction Argumentation in legal cases is as much concerned with the common sense reasoning about evidence and facts in a case as it is with legal reasoning about rules and precedents. One of the aims of the study of so-called rational proof is to provide a ra- tional, systematic method for reasoning with evidence in a legal context. More specifically, we are looking for a normative theo- ry that tells us how to rationally reason with evidence whilst re- specting the relevant practical and legal constraints. In the literature, there are three important approaches to ra- tional proof: arguments, statistics and narratives. Two of these, arguments and narratives, have been combined in the hybrid theory (Bex 2011, Bex and Verheij 2012, 2013). According to the hybrid theory, reasoning with evidence involves constructing F.J. Bex & V.M. Bex-Reimert © F.J. Bex & V.M. Bex-Reimart BexInformal Logic, Vol. 36, No. 3 (2016), pp. 349-370. 350 narratives, or scenarios, about “what happened” in a case, which is arguably easiest from a cognitive point of view (Pennington and Hastie 1993, Wagenaar et al. 1993). The hybrid theory fur- ther allows for the critical analysis of scenarios using arguments, and it gives critical questions that point to common counterar- guments (Bex and Verheij 2012). Many of these critical ques- tions are based on common mistakes people make when reason- ing about, for example, human memory (Wagenaar et al. 1993). Thus, the hybrid theory takes ideas from psychology on how people reason with evidence and puts them into a normative framework of structured, dialectical argumentation (Anderson et al. 2005; Bex 2011). Various researchers have investigated the connection be- tween theories of rational proof and the law, putting legal con- cepts such as relevance (Anderson et al.; 2005; Bex and Verheij 2013), burdens and standards of proof (Prakken and Sartor 2011; Bex and Walton 2012) and the presumption of innocence (Godden and Walton 2007) in an epistemological framework. Most of this existing research on combining evidential reasoning with legal reasoning focuses on criminal law and its related con- cepts. An important question that hence remains is whether and how the hybrid theory can be applied to other legal domains in which reasoning with evidence plays an important part. One legal domain in which decision makers have to deal with complex questions regarding evidence is refugee law. In 2014 more than 600.000 people applied for asylum in Europe1. In most of these cases, there is very little evidence to corroborate the story of an asylum seeker and their claims about, for exam- ple, their identity and the situation in their country of origin. Decision makers have access to general information about the situation in the country of origin, but this is not always complete or trustworthy, especially for countries with totalitarian regimes or in the midst of a war, which means that finding “the truth” is not feasible in many cases. Furthermore, an asylum decision is ultimately not based on the applicant’s story of past events but on the assessment of future risk that the applicant would face if returned to the country of origin. All these factors lead to an ev- idence assessment process and evidence assessment criteria which are quite different from those in criminal law. With respect to so-called “credibility assessment” in asylum cases – where credibility stands for both the credibility of the refugee as a person and the credibility of the facts and other evi- dence presented – a number of documents have been drafted by organizations such as the European Asylum Support Office 1 http://ec.europa.eu/eurostat/statistics-explained/index.php/Asylum_statistics Evidence Assessment in Refugee Law © F.J. Bex & V.M. Bex-Reimart BexInformal Logic, Vol. 36, No. 3 (2016), pp. 349-370. 351 (EASO 2015), the International Association of Refugee Law Judges (IARLJ 2013) and the Dutch Immigration Office (IND 2014). Some of these have a formal status as soft law (e.g. IND 2014), meaning that whilst they do not have the binding force of traditional law, they are accepted as a part of the legal system and decision makers have to abide by them. Others (e.g. EASO 2015, IARLJ 2013) are purely informative, providing guidelines to interpretation and practical application of the law. All of them, however, provide a normative framework for how deci- sion makers should reason in asylum cases. The main aim of this paper is to explore the relations be- tween current guidelines for credibility assessment and the hy- brid theory. We will focus on the EASO (2015) guidelines or evidence assessment, and the IARLJ (2013) report containing judicial criteria and standards for the assessment of credibility. Our main question is then: which concepts and requirements do these guidelines use, and can they be captured in the hybrid the- ory? A secondary question is whether the hybrid theory can be used to clarify or add to the guidelines as presented by EASO and IARLJ. 2. The hybrid theory of stories and arguments 2.1 Arguments The argumentative approach to evidential reasoning has its roots in Wigmore’s charting method (Wigmore 1937), a more mod- ernised version of which was presented by Anderson and col- leagues (2005), which was in turn captured as logical defeasible arguments (see e.g. chapter 3 of Bex 2011). Arguments are often based on evidence, such as witness statements made in court, forensic expert reports handed to the jury and so on. Given this evidence, we can make consecutive inferences, thus reasoning towards the factual claims in a case. Figure 1 presents an argu- ment in which evidence, namely claims by a refugee, Adnan, and Country of Origin Information (COI) is used to infer that Adnan faced a risk of serious harm had he stayed in Aleppo (Syria). As in logic, the inferences in arguments are based on (de- feasible) inference rules. Such inference rules represent stereo- typical patterns of human reasoning – in essence, they are based on our knowledge of the world. For example, in general we be- lieve a witness who testifies about something they are in a posi- tion to know about. So, for example, we can say that the general rule “If a witness says p then p is true” can be used to infer con- F.J. Bex & V.M. Bex-Reimert © F.J. Bex & V.M. Bex-Reimart BexInformal Logic, Vol. 36, No. 3 (2016), pp. 349-370. 352 clusions from witness statements, as in Figure 1. In addition to commonsense reasoning, we can also capture other forms of reasoning using inference rules, such as reasoning with legal rules or reasoning about the legal interpretation or qualification of facts (Bex and Verheij 2013). For example, in the case of Adnan we could say that the facts allow us to infer that there was a serious threat to Adnan’s life (Figure 1) by using the rule “If someone lives on the front line of a civil war and their house is hit by grenades, then this counts as a serious threat to their lives”. Article 15 of the EU Qualification Directive, which says that “serious harm consists of a serious […] threat to a civilian’s life due to indiscriminate violence in situations of internal armed conflict”, then allows us to infer the final conclusion about the risks Adnan faced. Adnan  lived  in   Aleppo Government  and   rebel  forces  fought   in  Aleppo Adnan’s  house  was   hit  by  grenades   There  was  a  serious  threat  to  Adnan’s  life  due  to  indiscriminate   violence  in  a  situation  of  internal  armed  conflict Adnan  faced  a  real  risk  of  serious   harm  in  Syria Adnan  claims  he   lived  in  Aleppo witness  evidence COI  information   on  Aleppo Adnan  claims  his  house   was  hit  by  grenades witness  evidencedocumentary  evidence qualification  step EU  Qualification  directive   article  15 evidence facts legal   conclusions Figure 1: Argument based on evidence An important feature of argumentation is that it is adversar- ial: not only arguments in favour of a particular conclusion have to be considered, but also possible counterarguments that attack these arguments. If we consider inferences as stereotypical pat- terns of reasoning, then we can provide some typical sources of doubt for each type of inference. For example, a witness argu- ment generalization can be attacked by arguments questioning the person’s truthfulness, or their observational abilities. If, for example, Adnan cannot answer basic questions about Aleppo, we can say that Adnan might be lying about his hometown and thus counter the witness inference in Figure 1. Here it is not the premise or conclusion that is denied, but rather the inference from the premise to the conclusion. Another possible way of at- tacking an argument is to argue for the opposite conclusion. For Evidence Assessment in Refugee Law © F.J. Bex & V.M. Bex-Reimart BexInformal Logic, Vol. 36, No. 3 (2016), pp. 349-370. 353 example, an argument for “Adnan did not live in Aleppo” at- tacks the conclusion that “Adnan lived in Aleppo” in Figure 1. It is understood that conclusions that are at one point ac- cepted can be rejected at a later stage. So, even though given the argument in Figure 1 we might say Adnan can be given a resi- dence permit, if we subsequently find counterevidence (e.g. evi- dence for the fact that Adnan is not from Syria or for the fact that Adnan is lying), we may reject the conclusion about Adnan’s nationality and hence the final conclusion about the residence permit. Given a complete set of arguments and their mutual attack relations, it is possible to determine the conclu- sions that follow from this set of arguments using any of the formal argumentation semantics that have been proposed by Dung (1995). 2.2 Stories The story-based or narrative approach to evidential reasoning stems from legal psychology (Pennington and Hastie, 1993; Wagenaar et al., 1993), and it has only relatively recently been further specified (Bex, 2011; Bex et al., 2010). This approach focuses on stories about what happened in a case that explain the evidence. Such a story is defined as a coherent sequence of events connected by (often implicit) causal relations. For exam- ple, Adnan might tell the following simple story “I lived in Aleppo, Syria, when government and rebel forces started fighting. My neighbourhood was really on the front lines, and my house was hit by grenades. I feared for my life so I fled the city”. Causal relations between the events can be made explicit: for example, we can say that ‘the fact that Adnan’s house was hit by grenades caused Adnan to flee the city’. In much of the more formal work on stories and scenarios, such explicit causal rules play a large role: often stories are rep- resented as fully connected causal networks (see e.g. chapter 3 of Bex 2011). In contrast, a more holistic perspective on story coherence is provided by story schemes, abstract scenarios that act as a scheme for specific stories (Schank 1986, Bex 2011). Such story schemes consist of generalized events or event types, abstract renditions of how things generally happen in the world. For example, Pennington and Hastie’s story scheme for inten- tional actions gives some initial states of affairs, which together with a motive may lead to an action with certain consequences (Pennington and Hastie 1993). A more specific scheme is con- cerned with, for example, fleeing from Syria to Europe, viz. F.J. Bex & V.M. Bex-Reimert © F.J. Bex & V.M. Bex-Reimart BexInformal Logic, Vol. 36, No. 3 (2016), pp. 349-370. 354 Person p lives in Syria – p fears for his life in the civil war – p flees to Turkey – p takes a boat to Greece – p travels via the Balkans to Austria – p takes a train to the Netherlands Such story schemes often follow from general knowledge: we know that people usually have motives for their actions, and in 2015, of the many refugees coming to Europe in 2015 tell us that the route via Greece and the Balkans is a standard route. Like the argumentative approach, the scenario approach has an adversarial element: different stories should be constructed that fit the evidence and these stories should be compared. For example, a witness might tell the alternative story that Adnan is not really from Aleppo but that he is in fact a Jihadist who was trained in Syria to commit a terrorist attack in Europe. Pennington and Hastie (1993) provide several criteria for judging stories. The most important ones are evidential coverage (how much and which evidence is explained by a story?), completeness (does a story have all the elements of the story scheme for intentional actions?) and plausibility (does the story fit with our ideas about how things happen in the world?). Note that both completeness and plausibility are about the quality of the story itself, irrespective of the evidence: a story can be very plausible and complete but not be backed by any evidence (cf. the risk of choosing a “good story” over a “true story”, Anderson et al., 2005). Novels of fiction, for example, can present very plausible stories for which there is no evidence; the same goes for our example, in which a quite plausible story about Adnan is presented, but, this being an example case, there is no real evidence for this particular story. 2.3 Connecting arguments and scenarios: A hybrid theory One perspective on reasoning with evidence is provided by a combination of stories and arguments, and hence Bex and col- leagues (Bex et al. 2010; Bex, 2011) propose a hybrid theory of arguments and stories. In the hybrid theory, stories can be used to provide natural explanations of “what happened” in a case, and arguments can be used to discuss whether a story conforms to the evidence and whether a story is plausible. As an example, consider Figure 2, in which arguments are used to support and attack Adnan’s story. In this figure, there are three types of ar- rows. The arrows with a closed head represent, like in Figure 1, inferences based on evidence. Note how the evidence can sup- port a single conclusion, but also the story as-a-whole: Adnan’s testimony, for example, contained his whole story. The arrows Evidence Assessment in Refugee Law © F.J. Bex & V.M. Bex-Reimart BexInformal Logic, Vol. 36, No. 3 (2016), pp. 349-370. 355 with an open head are explicit causal links, which were men- tioned by Adnan in his story. Finally, the arrows with a cross as a head are attack relations, representing counterarguments to the story. Adnan  is  from   Syria Adnan  lived  in   Aleppo Government  and   rebel  forces  fought   in  Aleppo Adnan’s  house  was   hit  by  grenades   Adnan  fled   the  citycauses causes Adnan’s   passport COI  documents   on  Aleppo Adnan’s   testimony Adnan’s  story Witness  A   testimony Adnan  did  not     live  in  Aleppo evidence  for  evidence  for  evidence  for   attacks evidence  for   Figure 2: Adnan's story, supported and attacked by arguments Once one or more stories have been constructed, their quality has to be judged. Thus, it can be assessed whether one story is better than another, or whether a story meets a certain standard of proof (Bex and Walton 2012). There are a number of criteria that can be used to judge a story’s quality. Perhaps the most im- portant one is evidential coverage, the amount of evidence in a case that supports a story. For example, our story in Figure 2 has an evidential coverage of three: Adnan’s story is supported by his testimony, his passport and the COI (Country of Origin In- formation) documents. Related to evidential coverage is eviden- tial contradiction, which is the number of arguments based on evidence that contradict a story. For our example story, this would be one: witness A’s testimony is the basis of an argument that contradicts the story. Note that the two evidential criteria do not give an absolute measure of how good or strong a scenario is, but rather provide relative measures to compare scenarios and guide the search for further evidence; if a plausible story has low evidential coverage it might make sense to search for evi- dence that supports the story. Note that only arguments that have not been attacked themselves count towards the evidential coverage and contradiction. If, for example, we have an expert testimony that Adnan’s passport is a forgery, we can attack the argument in Figure 2 from “Adnan’s passport” to “Adnan is from Syria” – in other words, the existence of this passport is not evidence for the fact that Adnan is from Syria. This would thus lessen the evidential coverage from three to two. Another way to compare stories is by looking at their co- herence irrespective of the evidence in a case. In other words, is the story coherent given our general knowledge about the world? With respect to coherence, there are three central con- F.J. Bex & V.M. Bex-Reimert © F.J. Bex & V.M. Bex-Reimart BexInformal Logic, Vol. 36, No. 3 (2016), pp. 349-370. 356 cepts: completeness, plausibility and consistency. Completeness, which was already mentioned above, is about whether a story has all its parts. Here, story schemes play an important part, as we have to look whether a story fits a particular scheme such as, for example, the intentional action scheme. Is there a motive? Does the story mention the relevant consequences of the central actions and the causal relations between the actions and conse- quences? For example, if in Adnan’s story he says that he fled the city but makes no mention of what happened afterwards, that is, the consequences and story of his flight, we might say the story is incomplete. The consistency of a story is about whether one part of the story contradicts another part. For example, if Adnan at one point claims he was in Aleppo on the 11th of December and later on in his story says he was in Istanbul on the 11th of December, the story as a whole is clearly inconsistent. The final criteria for story coherence, plausibility, concerns the number of arguments that are based on general knowledge that support or attack a sto- ry. The idea behind this is that explicit arguments that demon- strate why we think something is plausible or implausible, help us to ultimately determine whether a story is plausible. This is similar to Wagenaar and colleagues’ (1993) anchoring process, in which the assumptions in a story are made ever more explicit until the point that one cannot sensibly deny them. As an exam- ple, say that an Abu Dhabi airport stamp is found in Adnan’s passport, dated somewhere between when Adnan fled Aleppo and when he arrived in Europe. As an explanation of this stamp, Adnan tells the following story: “After I fled Aleppo I went to Istanbul. I wanted to take a plane to Amsterdam, but I boarded the wrong plane and ended up in Abu Dhabi, where they stamped my passport. I flew back to Turkey, from where I took a boat to Greece with some other refugees”. Now, one argument based in general knowledge that could be brought against the plausibility of this story is that “It is implausible that Adnan first tried to take a plane to Amsterdam and then took a boat to Greece, because it is general knowledge that refugees usually do not change their route and mode of transport during their jour- ney”, bringing the plausibility of the story to minus one. How- ever, this argument can be countered by saying that “it is not general knowledge that refugees do not change their route and mode of transport during their journey – in fact, changing cir- cumstances and available resources often require refugees to make on-the-fly changes of plan”, bringing the plausibility back to zero. A stronger argument to lower the plausibility of Adnan’s story is that “It is implausible that Adnan took the wrong plane in Istanbul, as it is general knowledge that people Evidence Assessment in Refugee Law © F.J. Bex & V.M. Bex-Reimart BexInformal Logic, Vol. 36, No. 3 (2016), pp. 349-370. 357 usually do not take the wrong plane on a big airfield like Istan- bul” – this can be based on general knowledge of boarding pass checks at airfields. 2.4 The hybrid theory as a method for legal argumentation The hybrid theory combines different insights from argumenta- tion theory, legal psychology, legal philosophy, law and artifi- cial intelligence. The theory has a clear normative aim (cf. Bex 2011, Verheij et al. 2016), in that it tells us how we should rea- son with evidence. The hybrid theory provides a by providing a “logic of evidence and proof” based on stories and arguments. Such a logic can be used to, for example, find out which scenar- io is of the best quality, or which arguments are acceptable giv- en the evidence, thus telling us why a particular story is bad. However, the basic hybrid theory does not tell us explicitly how we should reason in order to come to a good story. This implicit procedural character of the hybrid theory was made more clear in Bex and Verheij (2012), in which six critical questions based on the hybrid theory were proposed. (1) Are the facts of the case made sufficiently explicit in a good story? (2) Is the story sufficiently supported by evidence? (3) Is the support that the evidence gives to the story suffi- ciently relevant and strong? (4) Has the story itself been sufficiently critically as- sessed? (5) Have alternative stories been sufficiently taken into account? (6) Have all opposing reasons been weighed? The first question ensures that there is a complete and coherent story. Question 2 concerns the evidential coverage of the story. Question 3 is about the inferences from the evidence to the facts in the story: are they correct? Have they been called into ques- tion? Question 4 concerns the evidential contradiction and the plausibility of the story, that is, whether any counterarguments based on evidence or general knowledge have been considered. Question 5 is important: it negates the risk of confirmation bias or “tunnel vision”, where one story is taken as the main hypoth- esis and possible alternatives are not considered or set aside without due consideration. Finally, question 6 concerns the ac- ceptability of the arguments and stories in the case: which choices and decisions do we make? If, for example, we have two arguments for opposing conclusions, in the end we have to F.J. Bex & V.M. Bex-Reimert © F.J. Bex & V.M. Bex-Reimart BexInformal Logic, Vol. 36, No. 3 (2016), pp. 349-370. 358 choose which conclusion we accept given all the information we have. Similarly, if we have two alternative stories, we have to choose one on which to base our decision. Apart from the critical questions about the case as a whole, we can also ask critical questions for individual arguments. Wal- ton and colleagues (2008) have defined a large set of argumenta- tion schemes and critical questions, many of which can be asso- ciated with specific types of evidence. For example, a witness’ truthfulness (“Is there a reason to believe the person is lying?”) or observational abilities (“Could the witness have seen the event from where they were standing?”) can be questioned. Sim- ilar questions are possible for other types of evidence, such as documents (“Is there a reason to believe the documents are forged?”) or expert testimonies (“Is the expert really an expert in the field he is testifying about?”). Another type of critical ques- tion, introduced in (Bex and Verheij 2012), concerns critical questions for stories or, more specifically, for story schemes. For example, given the basic story scheme for intentional ac- tions, we can ask “Is the motive a credible motive for the ac- tions?”. The idea of critical questions makes sense in a dialectical setting, where there are multiple parties that engage in a dia- logue. Bex and Prakken (2008) have proposed rules for a dialec- tical dialogue in the hybrid theory, in which dialogue, partici- pants can each propose a story and critically assess these stories by proposing pro- and counterarguments to these stories. Some of the rules in a dialogue concern the burden of proof (Bex and Walton 2012, Prakken and Sartor 2011). One such burden is the burden of persuasion, which is set for the entire case and deter- mines which party has to prove their story according to some standard of proof. For example, in a criminal trial the burden of persuasion lies on the prosecution, as they have to prove the sto- ry that shows the defendant’s guilt beyond a reasonable doubt. The defendant – in theory at least – does not have to present or prove a story, but only has to cast enough doubt on the prosecu- tion’s story. Related to the burden of persuasion is the tactical burden of proof: if we would decide in the case now, what would the decision be? In other words, given the stories and ar- guments we currently have, who would “lose” the case – this party has the tactical burden of proof. The final type of burden is called the burden of production. This burden, which may shift during a case, specifies which party has to offer an argument based on evidence on some specific issue during the trial. Evidence Assessment in Refugee Law © F.J. Bex & V.M. Bex-Reimart BexInformal Logic, Vol. 36, No. 3 (2016), pp. 349-370. 359 3. Assessing credibility in refugee and asylum law Refugee law is a relatively young field of law, and legal defini- tions are often not as precise and worked-out as in, for example, criminal law. Furthermore, because of the international nature of the field, there is no single supervising judicial organ that de- cides on matters of interpretation; this is left up to countries’ in- dividual legal systems. These aspects reflect on credibility as- sessment in asylum cases, as there is no clear tradition of evi- dence theory like in criminal law (cf. Wigmore 1931, Anderson et al. 2005). Credibility assessment has been discussed, though often quite summarily, by researchers in the field of refugee and asylum law (e.g. Noll 2005, Staffans 2012, Hathaway and Foster 2014). However, often no explicit reference is made to theories of rational proof. As Noll (2005, pp. 6) says: “evidentiary theo- rists have typically drawn on criminal law or other areas within the domestic domain, which begs the question to what extent these finding can be transposed to the atypical setting of asylum law”. Outside of academia reports and guides on the process of credibility assessment have been drafted by organizations such as the European Asylum Support Office (EASO 2015) and the International Association of Refugee Law Judges (IARLJ 2013). Both documents place great emphasis on a so-called structured approach, reflected in checklists, summary charts and lists of criteria based on which an asserted fact can be accepted or re- jected (EASO guide) or the credibility of a claimant can be de- termined (IARLJ). However, no explicit reference is made to theories of rational proof. Furthermore, while the guides use fa- miliar concepts – for example the IARLJ argues that the refu- gee’s story should be coherent and plausible – these concepts are only explained summarily and not put into a comprehensive framework for rational legal proof. In this section, we want to give an overview of how credi- bility is assessed in asylum claims, and how this related to the hybrid theory. In other words, how do decision makers reason in asylum cases, and can this be captured with the hybrid theory? We do this by following the process of credibility assessment as set out in the EASO guide: (i) determine the material facts (ii) collect evidence (iii) assess credibility (iv) apply the “benefit of the doubt” principle and perform a risk assessment. 3.1 Determining the material facts The first stage of the assessment process according to the EASO guide is the gathering of information, which starts with the iden- F.J. Bex & V.M. Bex-Reimert © F.J. Bex & V.M. Bex-Reimart BexInformal Logic, Vol. 36, No. 3 (2016), pp. 349-370. 360 tification of the material facts, that is, those facts that can be di- rectly linked to the definition of a refugee or to the definition of a person eligible for subsidiary protection. (EASO 2015; IARLJ 2013; Hathaway and Foster 2014). The 1951 Geneva Conven- tion (Article 1.A.2) states that a refugee is someone who has a well-founded fear of being persecuted for reasons of race, reli- gion, nationality, social group or political opinion, and is outside their country of nationality or former habitual residence and un- able or unwilling to avail himself of the protection of that coun- try. In cases where someone does not qualify as a refugee, the European Qualification Directive (Article 2.f, Article 15) states that they may still be subject to subsidiary protection when the person would face a real risk of suffering serious harm when returned to their country of origin because of, for instance, a civ- il war. The question is then how to identify the material facts. In practice, a decision maker is almost always initially faced with the story of the person who applies for asylum: who are they, where are they from and why do they seek asylum? The EASO guide argues that only the facts of the story that can be linked to the definition are material facts – exactly how this “linking” should be done is left implicit. Anderson et al. (2005) treat link- ing as a question of relevance, that is, which facts of the story are directly relevant for ultimately inferring an element of a le- gal definition? Take, for example, the argument in Figure 1. No- tice how, at the level of “facts”, we have many elements of Adnan’s story from Figure 2. Thus, we can reason from the ref- ugee’s story to the elements of the legal definition in the same way that we reason from intermediate, “factual” claims to legal conclusions. Note that in the argument in Figure 1, a conclusion is drawn about whether Adnan faced a risk in the past. Any future risk, which is necessary for the correct legal definition of subsidiary protection, will be assessed in a later step (see section 5.4). 3.2 Collecting evidence and the burden of proof Once the material facts have been determined, the evidence can be gathered. The burden of persuasion lies with the applicant, who has to present and prove a story relevant to the material facts, that is, how and why they fled their country of origin. The applicant also has to support their story with sufficient evidence, or at least have made a genuine effort to provide this evidence. Furthermore, the evidence should be presented as soon as possi- ble, and explanations should be given in case the applicant is unable to support parts of their story with evidence. In terms of Evidence Assessment in Refugee Law © F.J. Bex & V.M. Bex-Reimart BexInformal Logic, Vol. 36, No. 3 (2016), pp. 349-370. 361 the hybrid theory, this means that the applicant has the initial burden of production. The decision maker, however, also has a so-called duty of investigation. This means that they also have a burden of producing the relevant country of origin information (COI) and evidence that cannot be expected to be given by the applicant, such as DNA-tests or language reports (Staffans 2012, IARLJ 2013). Gathering evidence can be challenging in asylum cases. Very often, not much more than the testimonies from the appli- cant are available, which are further complicated by multi- lingual communication involving interpreters. Furthermore, ap- plicants may be traumatized and in most cases a significant amount of time has elapsed between the events that caused the applicant to seek refuge and the actual assessment of the case. If documents are available, their authenticity has to be assessed, but the authorities of the country of origin cannot be involved, since they are possible persecutors. The question here is whether the existing argumentation schemes (Walton et al. 2008) suffice. The EASO evidence and interview guides (EASO 2014, 2015) provide explicit guidelines for assessing documents, COI and applicant stories, which can be recast as critical questions. The duty of investigation of the decision maker also entails that the decision maker has to ask the applicant to address issues that remain unclear due to, for example, lack of information or inconsistencies in the applicant’s story. This is not directly relat- ed to the burden of proof, but concerns the roles of the different participants in the dialogue about stories and arguments. While the applicant has a more persuasive role, they have to persuade the decision maker of their central claim, the decision maker has a more inquisitive role, they are more interested in gaining knowledge about the applicant’s case. Different types of dia- logue goals, such as persuasive and inquisitive, have been pro- posed by Walton and Krabbe (1995), but there is not much work on combining different types of dialogue agents in one dialogue. Bex and Prakken (2008) propose a dialogue which is both per- suasive and inquisitive, but do not assign specific roles to partic- ipants. 3.3 Credibility assessment The next phase in the assessment of an asylum claim is the cred- ibility assessment, in which the decision maker needs to decide which facts are established and which facts are rejected. Credi- bility is a term that is used extensively in the literature on evi- dence assessment in asylum cases (REFs). It can mean the cred- ibility of a story, or a collection of facts (e.g. “this story about F.J. Bex & V.M. Bex-Reimert © F.J. Bex & V.M. Bex-Reimart BexInformal Logic, Vol. 36, No. 3 (2016), pp. 349-370. 362 Adnan fleeing from Aleppo because of the civil war is credi- ble”) as well as the credibility of a person (e.g. “Adnan seems a credible person”). Both the EASO guide and the IARLJ guide mention a number of criteria to judge the evidence and the ap- plicant’s story, namely consistency, plausibility, sufficiency of detail, specificity and personal involvement. Consistency is defined by the EASO guide as compromis- ing a lack of discrepancies, contradictions and variations in the material facts. A difference is made between internal and exter- nal consistency (IARLJ 2013), where the internal consistency concerns the evidence and story presented by the claimant itself, and the external consistency concerns consistencies or discrep- ancies between the evidence and statements of the applicant and other evidence, such as COI reports. Internal consistency is part- ly captured by the consistency criterion in the hybrid theory (section 2.2.). If, for example, if Adnan in the same story claims they were in Aleppo and Istanbul on the 11th of December, the story contains two logically inconsistent elements. Consistency, both internal and external, is also captured by evidential support and contradiction in the hybrid theory. Evidential contradiction clearly points to inconsistencies: for example, if the applicant first tells a story about being in Istanbul and later simply denies this, his later statement effectively contradicts an element of his earlier story, Similarly, an external witness claiming that Adnan was never in Aleppo is inconsistent with Adnan’s earlier story and thus contradicts this story (see Figure 2). While, logically speaking, evidential support and consistency need not be the same – a story can be consistent with evidence, but not be sup- ported by that evidence – the text of the EASO and IARLJ guides does show that when a story is consistent with evidence, it is meant that the story is supported by evidence: under “exter- nal credibility”, it is mentioned that the external consistency should be checked by determining whether there is, for example, other evidence “supportive […] to the applicant’s statements” (EASO 2015, pp. 12). With respect to consistency, only the credibility of the facts that are directly relevant to the material facts should be checked, and that any credibility findings on statements that do not per- tain to the material facts should not be taken into consideration. This can be captured in the hybrid theory by the concept of rele- vance explained earlier. It is further stressed that the applicant should be given the chance to explain or clarify any inconsisten- cies. Like the decision maker’s duty of investigation, this “rule” of the EASO guide can only be fully captured in a dialogical setting: if one participant in the dialogue (the decision maker) Evidence Assessment in Refugee Law © F.J. Bex & V.M. Bex-Reimart BexInformal Logic, Vol. 36, No. 3 (2016), pp. 349-370. 363 finds an inconsistency, they should ask the other participant (the applicant) to resolve this inconsistency. With respect to the weighing of the different (internal, ex- ternal) consistent and inconsistent pieces of evidence, the EASO guide clearly states that the credibility assessment should be positive if there are no inconsistencies, that is, if according to the hybrid theory the evidential contradiction of the applicant’s story is low and the story itself is consistent. So while support- ive or corroborative evidence will help an applicant’s case, the main criterion is that the claim is at least consistent with what we know from external evidence such as COI. However, caution is needed in situations were claimants tailor their claims to be consistent with COI: too much consistency can also point to the fact that, for example, the applicant has rehearsed a certain sto- ry. Plausibility concerns the stories in a case, and is defined by the EASO guide as follows: “to be plausible the sequence of events has to have the quality of being likely and seeming pos- sible to a reasonable person” (EASO 2015, p. 12). This can therefore be more or less equated with the criterion of plausibil- ity in the hybrid theory: does the story follow our knowledge about how things typically happen in the world? Or, in more technical terms, can we find arguments based on general knowledge for the elements of the story? These explicit argu- ments against the plausibility of a story can be found in the IARLJ guide: if a decision maker decides to reject a claim be- cause it is deemed implausible, this should be fully reasoned. Because plausibility is based on the general knowledge of the decision maker, it is always secondary to findings on con- sistency. This is similar to the hybrid theory, where it is argued that evidential support should always take precedence over the plausibility of a story, lest we run the risk of preferring a plausi- ble but false story (i.e. a plausible story with not much eviden- tial support) over an implausible but true story (i.e. an implausi- ble story supported by evidence). Furthermore, it is important to note that, contrary to cases in criminal law, the decision makers are dealing with general knowledge from other cultures and countries. Take, for example the situation were a woman testi- fies that she has no idea where here husband was for the last half year. From a Western point of view, this might seem wholly im- plausible – however, in some cultures in Africa, for example, women are not supposed to ask their husband where they are and what they do, and it is normal for men to be away from home for longer periods. Sufficiency of detail is about the detail in which the appli- cant presents about their situation in the country of origin and F.J. Bex & V.M. Bex-Reimert © F.J. Bex & V.M. Bex-Reimart BexInformal Logic, Vol. 36, No. 3 (2016), pp. 349-370. 364 their flight story. This is captured by the completeness criterion in the hybrid theory: does a story have all its parts, that is, is there a sufficiently detailed motive for the actions of the appli- cant, and does the story clearly state how and why things hap- pened as they did? With respect to detail or completeness, the EASO interview guide (EASO 2014) mentions the elements of a story that need to be discussed with the applicant in order to have a complete picture. Interestingly, the guide provides a number of critical questions that could be used to check the completeness of a story, such as “What happened?” and “Where did it happen?” (EASO 2014, pp. 15-17). A sufficiently detailed story is not enough, however. The statements of the applicant should also have a certain level of specificity (EASO 2015) or personal involvement (IARLJ 2013). Some asylum claims tick all the boxes, including sufficiency of detail, but later on decision makers find out that the applicants rehearsed their story with smugglers. In those cases, it might be helpful to ask personal questions and not stick too much to the core elements of the claim (EASO 2015). 3.4 Benefit of the doubt and risk assessment Because it is often difficult to find reliable external evidence, decision makers can reach the conclusion that they do not know if a material fact is credible or not. In those cases, the principle of the “benefit of the doubt” can be applied in order to deter- mine whether a material fact (i.e. an element of the applicant’s story) can be accepted or rejected. For the benefit of the doubt to be applied, five conditions have to met: (i) the applicant has made a genuine effort to substantiate his claim; (ii) a satisfactory explanation has been given regarding any lack of evidence; (iii) the applicant’s statements are coherent and plausible and corre- spond with generally available information; (iv) the applicant has applied for asylum at the earliest possible time, unless there is a good reason for not having done so; (v) the applicant has been credible on other material facts. Conditions (i) and (iv) concern reasoning about the behav- iour of the applicant during the asylum claim process, and are thus not captured in the hybrid theory. Condition (ii) concerns the explanations of the applicant on, for example, they were not able to hand over their passport. Such explanations can be cap- tured and assessed in the hybrid theory. Whilst the coherence and credibility in (iii) and (v) can be captured in the hybrid theo- ry, an additional standard defining, for example, how many of the other statements and material facts have to be credible could be incorporated in the hybrid theory. Evidence Assessment in Refugee Law © F.J. Bex & V.M. Bex-Reimart BexInformal Logic, Vol. 36, No. 3 (2016), pp. 349-370. 365 Once the credibility of the material facts has been estab- lished, either in the usual way or through application of the ben- efit of the doubt, an assessment of a future risk needs to be made. Even if we accept the material facts as being credible, this still might only mean that the applicant in the past had a subjec- tive fear of persecution or that the risk of harm was subjective. So the fear of persecution or the risk of serious harm has to be supported by objective evidence. Furthermore, whilst past per- secution is a strong indication for future persecution (EASO 2015), it might be that the situation in the country of origin has changed after the applicant has fled and no risk of future perse- cution exists. But it can also be the other way around, when there is no indication of past persecution but future persecution is likely, for instance when there is a regime change or war breaks out in the country of origin while the applicant was not there. Although the risk assessment might be the hardest part of the assessment of an asylum claim, not to mention the enormous implications if misjudged, both the EASO and IARLJ guides pay much more attention to the credibility assessment than to the risk assessment. This is probably because the risk assess- ment is based on the material facts that are established when as- sessing the credibility, and a mistake in the credibility assess- ment hence has direct implications for the risk assessment. While the hybrid theory could support additional argumentation about the future risk based on findings of past risks, this has not yet been explored in depth. The main focus of the hybrid theory lies on providing evidence for what happened in the past and drawing legal conclusions based on this. That said, arguing about future risks is, from a logical point of view, perfectly pos- sible using a combination of stories and arguments (Hovestad and Bex, 2016). 4. Conclusions and discussion The main aim of this paper has been to explore if the hybrid theory can be applied to the assessment of evidence in asylum cases. We have described the process of evidence and credibility assessment in asylum cases as laid out in the EASO guide, and looked at which elements and criteria used in this process are captured by the hybrid theory as presented in the literature (Bex et al. 2008, 2011, 2012, 2013). Table 1 presents the correspond- ences between the credibility assessment process and the hybrid theory. F.J. Bex & V.M. Bex-Reimert © F.J. Bex & V.M. Bex-Reimart BexInformal Logic, Vol. 36, No. 3 (2016), pp. 349-370. 366 As can be seen in the table, there are a number of areas where the hybrid theory can be expanded. The first area con- cerns different types of critical questions, which need to be for- mally defined in order to be fully incorporated into the hybrid theory. Some of these questions can be adapted from the EASO guides (EASO 2014, 2015), which present critical questions for different types of evidence as well as for applicant stories. Credibility Assessment in Ref- ugee Law Hybrid Theory Determining the material facts Infer the legal definitions from the facts of the refugee’s story through (legal) rules Decision maker’s duty of inves- tigation - Burden of proof (production) - Participant-specific rules for a dialogue game* Reasoning about types of evi- dence (e.g. traumatized witness- es, reliability of COI) Argument schemes and critical questions for types of evidence* Internal consistency of the story Story consistency External consistency of the story with other evidence - Evidential contradiction - Evidential support (to a lesser extent) Plausibility Plausibility Sufficiency of detail - Story completeness - Critical questions for stories aimed at checking detail* Specificity/personal involve- ment - Critical questions for stories aimed at checking personal in- volvement* Benefit of the doubt - Rules concerning the (dialogi- cal) moves of the applicant* - Stories explaining (lack of) observations - Standards of proof* Risk assessment Argumentation of future risks based on past events* Table 1: Elements of the credibility assessment process and the corresponding elements of the hybrid theory. A star * indicates a possible future development of the hy- brid theory. Evidence Assessment in Refugee Law © F.J. Bex & V.M. Bex-Reimart BexInformal Logic, Vol. 36, No. 3 (2016), pp. 349-370. 367 What is clearly underdeveloped is the dialogical process of reasoning with stories and arguments. Preliminary ideas have been expressed in (Bex and Prakken 2008), but no dialogical context for the hybrid theory has been given. The study of the asylum claim assessment process shows that, for example, dif- ferent roles and strategies of participants must be understood and captured. The decision maker’s duty of investigation in- volves asking the right questions at the right time, and for the applicant to be granted the benefit of the doubt they must have provided the right information at the right time. The only way to fully capture this is as a dynamic dialogue which takes place over time. Another possible addition to the hybrid theory is a standard of proof for refugee law. Bex and Walton (2012) have provided some interpretations of the various proof standards of civil and criminal law (balance of probabilities, beyond a reasonable doubt), and a standard of proof for asylum claims could be de- signed along similar lines. However, a problem is that in refugee law, the standard of proof is not clearly defined. While it is widely acknowledged that the standard of proof applied in crim- inal law (beyond reasonable doubt) is too high (Staffans 2012, Hathaway and Foster 2014), there is no agreement on what the exact standard should be and how this should be expressed. Fur- thermore, the principle of benefit of the doubt also acts as a standard of proof, even though it is not expressed as some kind of likelihood. Recall that this paper also intends to briefly address a se- cond question, namely whether it makes sense to follow the hy- brid theory’s method of story and evidence assessment in asy- lum cases. Because there is often no additional evidence, the asylum process is very much focused on a central story has thus, in contrast to criminal law, led to a number of clear criteria for story assessment in refugee law. Thus, the critical assessment of an explicit story (questions 1 and 4 in section 2.4) is central to the claim process: both the EASO and the IALJ guideline stress that there should be a clear and sufficiently detailed story, which should be assessed on (internal) consistency, plausibility and personal involvement. The lack of external evidence also means that in refugee law, most stories are not sufficiently supported by evidence (question 2, section 2.4). The focus in asylum cases is on wheth- er the story is simply consistent with the other evidence, which is a less strict requirement than requiring the story to be actively supported by evidence. Furthermore, the benefit of the doubt principle shifts the focus from proving the facts with direct evi- F.J. Bex & V.M. Bex-Reimert © F.J. Bex & V.M. Bex-Reimart BexInformal Logic, Vol. 36, No. 3 (2016), pp. 349-370. 368 dence to showing that the applicant has at least a consistent and plausible story. While in the hybrid theory alternative explanations for the evidence are very important (question 5, section 2.4), the EASO and IARLJ guidelines do not mention alternative stories. The claim process focuses on the claim made by the applicant, which can either be rewarded or rejected but there is little room for al- ternative stories. Because so little additional evidence is availa- ble, there is nothing that may point to a possible alternative sto- ry. Furthermore, because the alleged events have taken place outside the country where the application is made, there is little room for the serious investigation of possible alternative ac- counts of what happened, especially given the amount of appli- cations and the time available for each application. However, a possible interpretation of the need for alternatives lies in the risk assessment process. When performing the final risk assessment, decision makers have to not only take into account the (credible) story told by the applicant, but also possible alternative possibil- ities of persecution or risk of serious harm. For example, say that Adnan was persecuted by the government of Syria. Now, if the central government is no longer in control of Adnan’s home town, one could say that there is no fear of persecution anymore. However, if Adnan’s hometown is suddenly on the front lines of a civil war, one could argue that, alternatively, Adnan is eligible for subsidiary protection because there is a risk of serious harm. Because of the focus on credibility, the exact process of risk as- sessment and how to reason in this risk assessment is not made entirely clear in the EASO and IARLJ guidelines, but in our opinion the comparison and construction of alternative risk sce- narios deserves more attention, both in the literature on as- sessing asylum claims and in the hybrid theory. References Anderson, T.J., Schum, D. A., and Twining, W. L. (2005) Anal- ysis of Evidence, 2nd edition, Cambridge University Press, Cambridge. Bex, F.J. (2011). Arguments, Stories and Criminal Evidence: A Formal Hybrid Theory. Dordrecht: Springer. Bex, F.J. and Prakken, H. (2008) Investigating Stories in a For- mal Dialogue Game. Proceedings of COMMA 2008 172:73- 84. IOS Press, Amsterdam. Evidence Assessment in Refugee Law © F.J. Bex & V.M. Bex-Reimart BexInformal Logic, Vol. 36, No. 3 (2016), pp. 349-370. 369 Bex, F.J. and Verheij, B. (2012) Solving a Murder Case by Ask- ing Critical Questions: An Approach to Fact-Finding in Terms of Argumentation and Story Schemes. Argumentation 26:3, 325-353. Bex, F.J. and Verheij, B. (2013) Legal Stories and the Process of Proof. Artificial Intelligence and Law 21(3): 253-278. Bex, F.J. and Walton, D. (2012) Burdens and Standards of Proof for Inference to the Best Explanation: Three Case Studies. Law, Probability and Risk 11 (2-3): 113-133. EASO (2014) EASO Practical Guide: Personal interview. Eu- ropean Union: European Asylum Support Office. EASO (2015) EASO Practical Guide: Evidence Assessment. Eu- ropean Union: European Asylum Support Office. Godden, D. M., and Walton, D. (2007). A theory of presumption for everyday argumentation. Pragmatics & Cognition, 15(2), 313-346. Hathaway, J.C. and Foster, M. (2014) The law of refugee status, Cambridge: Cambridge University Press. Hovestad, B.W. and Bex, F.J. (2016) Making sense of risks: A hybrid argumentative-narrative approach to risk assessment. Technical Report, 2016-006. Utrecht University. IARLJ (2013) Assessment of Credibility in Refugee and Subsid- iary Protection claims under the EU Qualification Directive - Judicial criteria and standards. International Association of Refugee Law Judges IND (2014) Guidelines 2014/10 (Immigratie en Naturalisatie Dienst, werkinstructie 2014/10). Noll, G. (eds) (2005) Proof, Evidentiary Assessment and Credi- bility in Asylum Procedures, Brill: Leiden. Pennington, N., and Hastie, R. (1993) Reasoning in explana- tion–based decision making. Cognition 49:1–2, 123 – 163. Prakken, H. and Sartor, G (2011) On modelling burdens and standards of proof in structured argumentation. Proceedings of JURIX 2011, 83-92. IOS Press. Schank, R.C. (1986) Explanations Patterns: Understanding Me- chanically and Creatively, Lawrence Erlbaum, Hillsdale (New Jersey). Verheij, B., Bex, F., Timmer, S., Vlek, C., Meyer, J.J., Renooij, S., & Prakken, H. (2016). Arguments, Scenarios and Proba- bilities: Connections Between Three Normative Frameworks for Evidential Reasoning. Law, Probability & Risk 15, 35-70. Wagenaar, W.A., Koppen, P.J. van, and Crombag, H.F.M. (1993) Anchored Narratives: The Psychology of Criminal Evidence, St. Martin's Press, New York (New York). Walton, D.N., C.A. Reed, and F. Macagno (2008) Argumenta- tion schemes. Cambridge: Cambridge University Press. F.J. Bex & V.M. Bex-Reimert © F.J. Bex & V.M. Bex-Reimart BexInformal Logic, Vol. 36, No. 3 (2016), pp. 349-370. 370 Wigmore, J.H. (1931) The Principles of Judicial Proof or the Process of Proof as Given by Logic, Psychology, and Gen- eral Experience, and Illustrated in Judicial Trials (2nd edi- tion). Little, Brown and Company, Boston (MA).