CAPÍTULO III. MODELO DE GESTIÓN DE RIESGO OPERATIVO PARA
3.1. Diseño del modelo de gestión de riesgo operativo para instituciones
Argument schemes are usually framed by and develop within dialectical/dialogical contexts: conceptually, there are a proponent and an opponent, the first asserting a claim, the second variously challenging it through critical questions. Establishing a claim, weakening its supporting reasons through counterarguments and excep-tions, and defending and reaffirming the opening claim, all require for balancing several elements.
Consider the simple version of the argument from expert opinion scheme (Wal-ton et al. 2008, p. 14):
Major Premise: Source E is an expert in subject domain S containing proposition A.
Minor Premise: E asserts that proposition A (in domain S) is true (false).
Conclusion: A may plausibly be taken to be true (false).
Compared with other versions of the same scheme, this form has the merit of showing “how argument from expert opinion works as a fast and frugal heuristic in everyday thinking” (Walton 2010a, p. 164): in ordinary situations, we often jump to a certain conclusion through shortcuts of many kinds, such as the idea that if it is an expert opinion, it is necessarily correct and we can safely base our deci-sions or actions on it. The reasoner, who is putting forward the argument from ex-pert opinion, needs to pay attention not to incur the fallacy of the argumentum ad verecundiam, or appeal to authority, in the form of inappropriate argument from expert opinion. For instance, this may occur when the argument is proposed in such a way that the opponent feels threatened and does not advance the needed critical questions, consequently treating the argument as a deductively valid argu-ment (Walton and Koszowy 2016), or when the cited authority is not an expert or his/her expertise pertains to another field.
The presented version of argument from expert opinion hides an implicit condi-tional premise (Walton and Reed 2002, p. 2):
Conditional Premise: If source E is an expert in subject domain S containing proposition A and E says that A is true then A may plausibly be taken to be true (false).
It is thus possible to see the “structural resemblance” (Verheij 2003, p. 170) be-tween this argument scheme and the deductive scheme of modus ponens presented above: as modus ponens is expressed in the form of premises and conclusion, so is the argument from expert opinion. Argument schemes are generally seen as rules and correspond to rules of inference in logics. However, argumentation schemes are often defeasible, context-dependant, and only pragmatically valid. For exam-ple, in the argument from expert opinion, experts are fallible and may be wrong regarding the concrete case, or their opinion may be falsified by new information:
its conclusion is always just tentatively drawn. So, the argument from expert opin-ion, as many other argument schemes, follows the pattern of defeasible modus po-nens (Walton 2010a, p. 165):
If A then (defeasibly) B A
Therefore (defeasibly) B
Besides, pragmatic validity implies that the argument schemes that do not re-produce deductive forms of reasoning are acceptable depending on the context and on the specific, even contingent (Verheij 2003) circumstances of their application.
This property meets the needs of modelling concrete argumentative situations that are very common in human reasoning but that “have proved troublesome to view deductively” (Walton 2010a, p. 160). Many lists of argument schemes have been presented over the years,24 but none of them can claim to be exhaustive. In this re-gard, it does not even exist a standard criterion for establishing if a pattern of rea-soning is acceptable as argumentation scheme. As a result, an infinite range of ar-gument schemes could be enumerated, at least theoretically, and their acceptability
24 See, for example, Walton et al. (2008).
verified from time to time on the basis of their contextual relevance (Verheij 2003). As open tools of analysis, argument schemes are thus potentially applicable to whatever knowledge domain. Additionally, they can either be used as patterns for reconstructing arguments (e.g., they can detect implicit premises, as the next subsection shows) or as generators for constructing new arguments (Gordon and Walton 2009).
3.1.1. Critical Questions
Argument schemes are combined with a set of critical questions, which are usually included in the definition of argument scheme.
Consider what critical questions Walton et al. have identified for the cited ar-gument from expert opinion (2008, p. 15):
CQ1–Expertise Question: how credible is E as an expert source?
CQ2–Field Question: is E an expert in the field that A is in?
CQ3–Opinion Question: what did E assert that implies A?
CQ4–Trustworthiness Question: is E personally reliable as a source?
CQ5–Consistency Question: is A consistent with what other experts assert?
CQ6–Backup Evidence Question: is E’s assertion based on evidence?
Critical questions depend on the content and type of the sentences that compose the reasoning pattern, in turn defined by the context in which the parties are argu-ing. In our example, the argument is reasonable if certain “contextual factors”
(Wagemans 2011, p. 333) concerning the expert, the delivered opinion, and the re-lation existing between the expert opinion and the claimed field of expertise are actually present. In particular, the expert should be a credible source (CQ1) in the specific field of expertise of A (CQ2) and personally trustworthy (CQ4), his or her assertions should imply A (CQ3), he or she should deliver an opinion that is both consistent with what other experts maintain (CQ5) and based on evidence (CQ6).
CQ1, CQ2, CQ3, CQ6 could be modelled as implicit premises, or assumptions, of the argument (Walton and Gordon 2009), and are considered sufficient to refute the claim if not adequately replied to, shifting the burden of proof from the re-spondent to the proponent. Differently, CQ4 and CQ5 requires that the rere-spondent, questioning both the expert’s trustworthiness and the consistency of his or her as-sertions, backs them with proper evidence. So, he/she should present evidence that the expert is untrustworthy and indicate the different opinions other experts in the field have expressed (Walton 2010a).
Notably, these critical questions only refers to the argument from expert opin-ion. In this respect, consider also the different argument from analogy together with its critical questions (Walton et al. 2008, p. 56, version I, and p. 315):
Major Premise (or Similarity Premise): Generally, case C1 is similar to case C2. Minor Premise (or Base Premise): Proposition A is true (false) in case C1. Conclusion: Proposition A is true (false) in case C2.
CQ1: Is A true (false) inC1?
CQ2: Are C1 and C2 similar, in the respects cited?
CQ3: Are there important differences (dissimilarities) between C1 and C2?
CQ4: Is there some other case C3 that is also similar to C1 except that A is false (true) in C3?
Critical questioning arguments from analogy means to investigate whether sim-ilarity and differences exist among the cited cases and whether it is possible to es-tablish comparisons with other, ignored cases: the abovementioned questions on experts and their opinions are absolutely irrelevant. Identifying the types of sen-tences that characterise an argument scheme is therefore a necessary step for ana-lysing argument schemes in general (Verheij 2003).
Three further observations naturally follow.
Firstly, the field-dependency of critical questions grounds the concept of validi-ty of argumentation schemes (Prakken 2005): validivalidi-ty is not understood in a strict-ly logical sense, but, broadstrict-ly, in terms of reasonableness and persuasive strength of the argument. That is why different arguments require different standards of evaluation. Critical questions, thus, play a key role in evaluating arguments. If evaluation generally intends to establish whether an argument, fitting a precise ar-gument scheme, is strong or weak, how to better accomplish this task is object of a lively debate in the AI and Law community. A feasible solution is represented by the reformulation of argument schemes in an argumentation framework, so that a computational model is then available to evaluate the argument-scheme approach.
As for informal logic, the usual approach to argument evaluation consists of inves-tigating three main aspects (Walton et al. 2008): a) whether the premises are ac-ceptable; b) whether the premises are relevant in sight of the conclusion; c) whether the premises provide for sufficient support for accepting the conclusion;
c1) if other and better reasons exist in favour of the opposite conclusion.
Second observation is that, even if field-dependency boosts difference between critical questions pertaining to different argument schemes, some degree of simi-larity can be nevertheless distinguished. Each argument scheme is challenged in its premises and conclusions: the former always need to be true, well supported, and justified, whereas the latter always need to resist to contrary reasons or differ-ent conclusions (Verheij 2003). Thus, there exist questions that are common to every argument scheme as for the function they accomplish within the scheme, and this is linked to the fact that critical questions fulfil many tasks with reference to argument schemes: some represent premises of the scheme otherwise left im-plicit, some identify exceptions that could prevent the scheme from operating, some indicate other possible reasons or arguments against the conclusion of the argument.
Thirdly, critical questions determine whether an argument scheme is appropri-ate in a concrete case by investigating whether its conditions are relevant to its use and ends. Relevance is a sort of transverse issue: argumentation theory wonders
about its definition,25 AI studies if and how it is possible to express this property formally and treat it computationally, law also has extreme interest in better defin-ing what can be labelled as relevant for the purposes of well-supported judicial opinions and lawyer’s argumentative strategies. So, even if analysing the concept of relevance is not the object of the present thesis, relevance is not exempt from our discourse, for its implicit role in identifying conditions of argument schemes.
3.1.2. AI and Law Perspective on Argument Schemes Formalisation
Argument schemes are, technically, rules of inference of a logical system (Verheij 2003, Prakken 2005, Walton et al. 2008, van Eemeren et al. 2014): informal ar-guments can be instantiated into logical inference rules by seeing the relation be-tween premises and conclusions in terms of conditional rules. It is worth noting that the obtained rules are generally defeasible since deductive arguments are one of the several possible types of arguments, and not even the most recurring in or-dinary argumentation. Prakken (2005) refers to Horty’s attempt to formalise some argument schemes as instances of the defeasible modus ponens (Horty 2001):
P
If P then usually Q Therefore (presumably), Q
However, the defeasible modus ponens is still excessively abstract with its log-ical language and it does not preserve the vocation of argument schemes of main-taining nuances of meaning while reproducing the reasoning process (Prakken 2005). Moreover, when approaching an informal argument scheme, there are parts that are general, and basically function as ordinary rules of logical inference, but there are also parts that are strictly context-specific and, as such, determined by the content of the sentences they consist of: it is on this blurred line that abstract logic and contextual logic would distinguish from each other (Verheij 2003).
Argumentation schemes hence show an inner contradiction. On the one hand, their underlying logical form allows for expressing them in logical language, on the other hand, that same formalisation puts in danger their most significant fea-ture, i.e., the capacity to transmit meaningful reasons. Formalisations run the risk of losing those details that field-dependency of argument schemes offers, for ex-ample in the form of distinct critical questions. Once more, logic has the oppor-tunity to rethink itself, its methods, and its purposes, for encompassing the enrich-ing specificity of argumentation schemes, and not stumblenrich-ing upon the shortages of formalism.
In line with this vision, argumentation logic makes use of non-monotonic tech-niques that allow to express exceptions, disagreement, and uncertainty. Argument schemes can thus be embedded in such argumentation framework (Prakken 2005),
25 See for example Walton (2004).
where, defined as trees including both deductive and defeasible inferences, argu-ments can be attacked also on the defeasible inference step. Rebuttals, undercut-ters, and undermining defeaters are adequate tools for the formalisation: according to Prakken, who has further improved this AF in ASPIC+ (Prakken 2010, Prakken et al. 2015) by formalising argument schemes for factor-based reasoning, a) argu-ment schemes can be modelled as the “prima facie reasons” dear to Pollock; b) whenever the application of a scheme results in contrary conclusions is nothing but an instantiation of rebuttal; c) negative replies to critical questions corresponds to undercutters, which actually highlight exceptional conditions.
Feasible path, argumentation logic clearly expands upon the issue of distin-guishing what relations form and content have within argument schemes. Addi-tionally, such an abstract AF is capable of dealing with conflicts among argu-ments, first establishing criteria to assess which argument is strong enough to prevail in the dialectical relation (e.g., by modelling preference rules), then defin-ing the defeasible validity of arguments, often understood in game-theoretic terms:
an arguer has a defeasibly valid argument, whenever he or she has a winning strat-egy. “Tree of trees” in Prakken’s words, argument-based logic gives a useful rep-resentation of ordinary argumentation.