34.34112edited © Lilian Bermejo-Luque. Informal Logic, Vol. 34, No.3 (2014), pp. 311-336. Deduction without dogmas: the case of moral analogical argumentation LILIAN BERMEJO-LUQUE Department of Philosophy University of Granada Campus de Cartuja, Granada-18071. Spain lilianbl@ugr.es Abstract: In a recent paper, Fábio Perin Shecaira (2013) proposes a defence of Waller’s deductivist schema for moral analogical argumentation. This defence has several flaws, the most important of them being that many good analogical arguments would be deemed bad or deficient. Additionally, Shecaira misrepresents my alternative account as something in between deductivism and non- deductivism. This paper is both an attempt at solving this misunderstanding and an analysis and criticism of Waller and Shecaira’s forms of deductivism. Résumé; Dans un article récent, Fábio Perin Shecaira (2013) défend l’approche déductiviste de Waller pour évaluer des arguments par analogie traitant de sujets moraux. Cette défense a plusieurs défauts, le plus important d'entre eux est que plusieurs bons arguments par analogie seraient jugés mauvais ou faibles. En outre, Shecaira dénature mon approche alternative en la décrivant comme quelque chose entre un déductivisme et un non déductivisme. Je tente à la fois de résoudre ce malentendu et d’offrir une analyse et une critique des formes de deductivisme de Waller et de Shecaira. Keywords: Analogical argumentation, a priori analogical argumentation, deductivism, defeasibility, linguistic normative model. 1. Introduction In a recent paper published in this journal, Perin Shecaira (2013) proposed a defence of Waller’s (2001) deductivist schema for moral analogical argumentation against Trudy Govier’s (1989 and 2002), Marcello Guarini’s (2004) and my own (Bermejo- Luque, 2012) proposals. As I see it, this defence has several flaws, the most important of them being his contention that many good analogical arguments should be deemed bad or deficient on that standard. Additionally, Shecaira misrepresents my proposal as something in between deductivist and non- deductivist accounts of this type of argumentation. This paper is Lilian Bermejo-Luque © Lilian Bermejo-Luque. Informal Logic, Vol. 34, No.3 (2014), pp. 311-336. 312 both an attempt at solving this misunderstanding and an analysis and criticism of Waller and Shecaira’s form of deductivism. To this end, I will start by characterizing analogical claims as a very special type of claim in that they do not have semantic content of their own. This account, which is in line with the mainstream view in philosophy of language and linguistics, is meant to explain the peculiarities of the use of analogies in argumentation, as argued in Bermejo-Luque (2012). Then, I will distinguish between analogies and analogical arguments and, within each of these categories, between quantitative (or a posteriori) analogies and qualitative (or a priori) analogies, on the one hand, and between deductive analogical arguments and inductive analogical arguments (sections 2 and 3), on the other. Following these distinctions, I will argue that the expression ‘a priori analogies’, as used by Govier and her followers, is misleading in several ways. I will then show that both Shecaira and Waller’s deductivism and Govier and Guarini’s non-deductivism regarding conclusive analogical arguments share a formalist conception of deductive validity – which is the reason why Govier and Guarini do not regard conclusively valid analogical arguments as deductively valid, and also the reason why Waller and Shecaira do (section 4). I will criticise this view by explaining the problems of evaluating natural language argumentation by using argument schemas which have not been properly backed by a theory of interpretation (sections 4 and 5). I will also criticise the particular problems of Shecaira’s proposal, and explain why the model proposed in Bermejo-Luque (2012) was a way to avoid these problems and the problems associated with deductivism, in general (sections 5 and 6). 2. Quantitative analogies and qualitative analogies As a basis for our discussion, it seems appropriate to start from an overall characterization of analogies themselves. Analogies are powerful cognitive tools: in thinking of unfamiliar objects as similar to familiar ones, we enable working hypotheses for dealing with novelties, paths to become familiar with new things and phenomena.1 But what does it mean to say that two things are “similar”? Similarity is a concept that is difficult to define non-circularly. For example, dictionaries bring you from 1 On this view, analogies are not particularly argumentative. For example, as Waller (2001: 200) pointed out, they may have a merely figurative use. (For a survey of uses of analogies, see Bermejo-Luque, forthcoming-a) Deduction without Dogmas © Lilian Bermejo-Luque. Informal Logic, Vol. 34, No.3 (2014), pp. 311-336. 313 “similarity” to “alikeness”, “comparability” or “resemblance”, and from these, to “similarity” again. I myself will not try a definition as such, but I will make a few distinctions. Current literature on analogies – and specifically, on analogical argumentation and reasoning – seems to be torn between two conceptions of similarity. On the one hand, there is a quantitative view, according to which two things, facts, situations, etc. are similar when they have enough relevant properties in common. For example, you are about to buy a new car, and you observe that the one you are considering is like another one that you have already refused. It is about the same size, both engines are basically the same; the quality of the finish is on a par, similar shape, and so forth. In these conditions, we deem the analogical judgment to be very sound: quantitative analogies are better or worse depending on the amount of “relevant properties in common” – whatever that means, as we are going to see. Of course, you probably don’t care about things like the similarities between their plate numbers or the first names of the owners of the companies that manufacture each car. Remarkably, because anything has an indefinite number of properties, in principle, the amount of properties that any two things have in common is as indefinitely big as the amount of properties that they don’t. This is why we must always judge analogies by taking into account the context and purposes for which they have been put forward. Regarding quantitative analogies, such context and purposes implicitly determine the set of properties that counts for making the comparison, as well as the amount of properties in common that is enough for making the corresponding analogy sound. Whether or not two things have enough relevant properties in common is a contextual matter in this double sense; yet, it is also critical because not finding enough relevant properties in common would render the analogy (more or less) unsound. At any rate, in this quantitative sense, whether or not an analogy is sound is an empirical, or better, an a posteriori matter – that is, something to be discovered and checked by further inspecting both analogs. Consequently, for a quantitative analogy to be appraisable – and, therefore, meaningful – independent information about the corresponding characteristics of both analogs must be available.2 2 For example, a quantitative analogy like “the organs of complex creatures living on planets with high atmospheric pressure would be like those of the animal fauna living in the deep ocean environment” makes sense as a quantitative analogy because we can figure out the characteristics of the organs of complex creatures living in a planet with high atmospheric pressure, despite the fact they do not exist, as far as we know. Lilian Bermejo-Luque © Lilian Bermejo-Luque. Informal Logic, Vol. 34, No.3 (2014), pp. 311-336. 314 As I argue in Bermejo-Luque (forthcoming – b), this quantitative conception of similarity is particularly at stake in law courts, when one of the parties’ strategy is to show that the case under consideration is analogous to another one already judged. However, we frequently establish similarities that do not fit this quantitative conception. On the one hand, we say things like “these colours are alike”, “our kids are about the same size”, “their voices are very similar”, etc. Being similar in this sense is not a matter of a set of relevant properties in common, but rather of reference to a particular quality and the idea of proximity of the things compared to something like a point on a scale of this quality (i.e., a colour spectrum, human size, sets of harmonics, etc.). This conception of similarity is qualitative in the following sense: it is not the amount of properties in common that makes the similarity claim more or less sound, but the proximity between the properties at stake within a given scale. Of course, which properties are at stake is a contextual matter, as it is the idea of “proximity” in each case. For example, because we are talking about kids, and not about microorganisms, a difference in microns happens to be irrelevant. And when the properties at stake are not close enough to each other on the relevant scale and in the relevant amount, the similarity claim can be dismissed. Certainly, this is another way of establishing similarities; yet, I think we should regard claims stating this kind of similarity as comparisons, rather than as analogies proper. The need to distinguish analogies from comparisons comes from the fact that there seems to be something peculiar in saying that claims such as “our kids are about the same size” are analogies, despite being similarity statements indeed. On the other hand, we also establish similarities between things, facts, situations, etc. when they share something very significant and remarkable within the context and for the purposes of the conversation. Such analogies – e.g., “a life without friends is like a garden without flowers”, “character is like a tree and reputation is like a shadow”, “love is like a butterfly”, “love is like a rock”, “love is like a bottle of gin” – also pivot on a qualitative conception of similarity in the sense that it is not the amount of relevant properties in common which makes the comparison more or less sound, but rather it is how significant and striking the property that both analogs have in common is. Noticeably, in formulating this type of analogy, we somehow predetermine the way in which it makes sense to compare the two analogs. We do not state this kind of similarity Deduction without Dogmas © Lilian Bermejo-Luque. Informal Logic, Vol. 34, No.3 (2014), pp. 311-336. 315 as an empirical matter – i.e., as something to be discovered by empirically inspecting the properties of both things, as in the case of quantitative analogies – but as an a priori one. That is, we conceive it as a matter of both analogs being members of a category that is settled, a “new” category for which the very analogy stands. This category may be difficult or just impossible to specify in so many words without losing part of the analogy’s insight. This is why qualitative analogies are characteristically vague and inexplicit, and closer to metaphors than quantitative ones. However, qualitative analogies may also be dismissed or resisted by pointing out that there is some very relevant feature which does not really apply or should not be applied to one of the analogs. This feature would be something that would be distinctive of the category if the analogy really made sense. For example: “I think of love rather as a path than as a butterfly; it’s not something to contemplate, but something to be actively involved in.” Finally, it is also important to distinguish similarity statements (analogies and comparisons) from sameness statements because, apart from statements of co-referentiality, sameness statements can also be statements of an attributive relationship between two different things. For example, my piece of lemon pie is not the same as your piece of lemon pie; but if I say that you and I have chosen the same dessert, you will not understand that statement to mean that I have chosen your particular piece of lemon pie. There is a sense in which “sameness” is a matter of inclusion in a specific category. In this sense, my dessert and your dessert are “the same” as far as they both can be described as “a piece of lemon pie.” Which category is the relevant one is something implicit in the context of the conversation. In one context, my piece of Gino’s lemon pie is not the same as your piece homemade lemon pie; in other contexts, they can be said to be the same, e.g., pieces of lemon pie, lemon desserts, pieces of a pie, etc. Because we have characterized qualitative analogical statements as statements that pivot on the soundness of making a category out of the two analogs, it may seem that they are a type of sameness statements. But there is a key difference between them: qualitative analogical statements do not appeal to already available categories. Rather, they create the relevant categories in themselves; the category is constituted by the very comparison and this is its entire specification. In this sense, sameness statements work as classifications whereas qualitative analogies have a more prospective character. In dealing with their argumentative uses, we will have to be very clear about whether we are considering a sameness Lilian Bermejo-Luque © Lilian Bermejo-Luque. Informal Logic, Vol. 34, No.3 (2014), pp. 311-336. 316 statement or an analogy, because their corresponding criteria of correctness are very different from each other. As pointed out, two different things are “the same” – always, in some respect– if they both may be described, within the context, as members of a certain category, and this is an all-or-nothing property. In turn, quantitative analogies are analogies whose (more or less) soundness is a matter of the number of relevant properties that are shared by both analogs, and qualitative analogies are analogies whose (more or less) soundness depends on the significance and relevance of the comparison between the analogs, and the adequacy of establishing a category consisting of the two analogs. As argued in Bermejo-Luque (forthcoming – b), the distinction between quantitative analogical statements and sameness statements is crucial for distinguishing juridical argumentation from precedent (which is deemed to be conclusive) from juridical analogical argumentation (which is deemed to be non-conclusive). 3. Inductive and non-inductive So much for the distinction between comparisons, analogies (quantitative and qualitative) and sameness statements. Now, this distinction between quantitative and qualitative analogies will bring to memory Trudy Govier’s distinction between inductive analogies and a priori analogies. However, Govier’s distinction is not on analogies, but on analogical arguments: Following Wisdom and Barker, I use the term "inductive analogy" for those arguments by analogy in which the analogue used is a real (that is, non-hypothetical) instance, and the features it is said to have are attributed to it on the basis of observation or other empirical means. (Govier, 1989: 141). Basically, quantitative analogies (our terminology) would correspond to the “reasons” in the sort of analogical arguments that Govier calls “inductive analogies.” However, as we have seen, our distinction between a posteriori or quantitative analogies and a priori or qualitative analogies is not a matter of whether one of the analogs is real or invented, which is Govier’s criterion for distinguishing inductive analogies from a priori ones: The difference between a priori analogy and inductive analogy, as I’m employing the terms here, is that in an a priori analogy, the analogue need not be a real case. It can be entirely hypothetical and may, in fact, be positively fanciful. (Govier 1989: 142) Deduction without Dogmas © Lilian Bermejo-Luque. Informal Logic, Vol. 34, No.3 (2014), pp. 311-336. 317 In our account, quantitative analogies are not necessarily empirical, as they may involve one or two non-existent analogs; but they are always a posteriori. They pivot on the possibility of further inspecting and comparing both analogs – possibly, by inferring their (would-be) properties, not necessarily by empirically checking them. In turn, qualitative analogies are a priori because their criterion of soundness is a matter of the adequacy of seeing one of the analogs in terms of the other. If an a priori analogy is taken to be sound, i.e., if we agree in seeing one of the analogs in terms of the other, then that cognitive strategy precedes our approach to the phenomena. It is genuinely a priori. As pointed out before, this feature makes a priori analogies very close to metaphors.3 However, it is not only qualitative analogical statements that are far from ordinary descriptions. Both quantitative and qualitative analogical statements involve, as a condition of meaningfulness, an intrinsic appeal to that which is relevant to the case. Certainly, any communicative move involves a certain appeal to relevance in order to make sense as such; but for analogies this appeal is a necessary condition for having content at all, precisely because any two things are both similar and dissimilar in indefinitely many ways. This intrinsic appeal to relevance – i.e., the relevance of the features that matter (in the case of quantitative analogies), or the relevance of the feature that is the ground for making a category out of both analogs (in the case of qualitative analogies) – is the reason why bare analogical claims – i.e., claims of the form “A is like B” – do 3 In the article of mine that Shecaira criticizes, I argue further for this view, pointing out that the idea that (qualitative) analogies are like metaphors in not having a propositional content as such is the mainstream position in the philosophy of language, and that this is the reason why analogical argumentation is particularly powerful as a rhetorical device. But Shecaira does not get into this point, insisting on treating all types of analogical statements as mere descriptive claims that may be readily generalized by making explicit that which makes the two things compared “similar”. Contrastingly, in my view, the way in which qualitative analogies work as proposals for “seeing as” would be the key to understand their widespread use in moral reasoning. Actually, our account would explain why moral particularists can find in analogical reasoning a preeminent way of making conclusive moral arguments. But Shecaira’s strategy is rather to dismiss particularism altogether. At any rate, the fact that analogies do not have a semantic content of their own would be the reason why, when used as reasons, we have to reconstruct their propositional content, so as to have something on which we can agree or disagree. In Bermejo-Luque (2012) I claimed that analogical argumentation is a type of non-literal argumentation, and I provided a proposal for reconstructing qualitative analogies when used as reasons. Lilian Bermejo-Luque © Lilian Bermejo-Luque. Informal Logic, Vol. 34, No.3 (2014), pp. 311-336. 318 not work as ordinary, literal descriptions. In other words, there is no real property in the world for which “similarity” stands. “Similar” is not a real predicate, it does not have a semantic content of its own; rather, it is a suggestion to consider things in a certain way. Consequently, analogical claims, instead of being properly true or false statements, behave as more or less sound cognitive proposals. It might be argued that it is possible to turn bare analogical claims of the form “A is like B” into real descriptions by specifying in which sense or senses A is like B. That would be making ampliative analogical claims of the form “A is like B in sharing features a, b, c.” However, there are two reasons why this suggestion cannot really work. On the one hand, there is the problem of determining what “sharing a feature” is. Is it a matter of having “the same” feature in the above sense (so that, again, a reference to the relevant category for making the classification is necessary) or is it a matter of having a “similar” feature (so that we are back to the problem of determining their similarity)? On the other hand, the role of such specifications cannot be that of describing the similarity between A and B because, for any A and B, there are indefinitely many ways in which they can be said to be similar just as there are many other ways in which they can be said to be different. Rather, such specifications behave as reasons for showing the plausibility of the bare analogical claim. After all, in order to reject an ampliative analogical claim, we do not necessarily have to refuse that A and B “share” properties a, b, c; we can also reject that sharing these properties makes A and B similar – that is, similar in the relevant sense. In other words, we can reject an argument for the similarity between A and B either by refusing the reason – namely, that A and B share properties a, b, c – or by refusing the warrant – namely, that this is a good reason for saying that they are alike. Now, the other reason for not adopting Govier’s distinction between inductive and a priori analogies is that it seems ill-founded. In contrast with “deductive” and “inductive,” “a priori” and “a posteriori” are not types of inferences4 but types of judgments, and it is somewhat misleading to identify the a priori or a posteriori nature of the analogical claim with its foreseeable use as a reason or backing for an inductive or a 4 Or types of validity, as suggested by Hitchcock (1980) or types of standards for evaluating arguments, as suggested by Govier (1980). As I explain below, in our model it is safe and coherent to think of deductiveness and inductiveness as types of inferences, for this model does not conflate the concepts of validity and deductiveness. Deduction without Dogmas © Lilian Bermejo-Luque. Informal Logic, Vol. 34, No.3 (2014), pp. 311-336. 319 deductive argument, respectively. For, in principle, both quantitative (or a posteriori) analogies and qualitative (or a priori) analogies can be used as reasons for both deductive and inductive arguments. It depends, among other things, on which other reasons constitute these arguments and the analogies for which they are a reason.5 On our account, analogical argumentation is, basically, argumentation in which an analogy is adduced in order to support a further claim. What we have in these cases is, paradigmatically, arguments of the form “A is like B, therefore A is Z”. In this type of argument, the analogy happens to be what Toulmin calls the data in his model, i.e., the evidence that is pointed out in support of a conclusion. For example, in “the political situation is like in the past elections; the party in office will probably win again”, a quantitative analogy is used as a reason, whereas in “the news has dropped on the Peace Conference like a bomb; there will probably be retaliatory actions”, it is a qualitative analogy which does the work.6 However, as pointed out in Bermejo-Luque (2012), arguments of the form “B is Z, therefore A is Z” may also be analogical in some cases, namely, when the inference is drawn in virtue of an alleged similarity between A and B. In these cases, the analogy would correspond to the backing in Toulmin’s model, i.e., the fact that stands behind the warrant of the argument and justifies the step from reason to conclusion. In these cases, the context can make it clear that there is an analogy adduced as a reason for the warrant “if B is Z, then A is Z”. “— 5 James Freeman (2013) has also entered into the discussion about the definition of a priori analogies. He agrees with Govier that the distinction between a priori analogies and inductive analogies regards arguments, not judgments, but disagrees with her definition of a priori analogies as analogies based on the possibly hypothetical status of the analogs. In Freeman’s view, “the hallmark [of a priori analogies] concerns the epistemic status of the warrant, not whether we may reflectively compare analogue and primary subject for similarities argued from.” (Freeman 2013: 178-179). Actually, in his account, there is another hallmark of a priori analogies, which is that they are about moral issues. According to Freeman (2013:180), the epistemic status of the warrant is a priori because it is based on moral intuitions, which, following Ross’ form of intuitionism, he takes to be a priori but synthetic (giving rise to the possibility of admitting the defeasibility of the corresponding arguments). However, I see no reason for restricting the use of the term “a priori” for analogies grounding moral arguments. The analogy in “don’t cheat the taxman, it’s like stealing from society” does not seem more a priori than the analogy in “don’t mix the salad dressing like that, it’s like dancing single legged”, which is not particularly moral. 6 Noticeably, this is an example of a non-deductive argument having as its reason an a priori analogy. Lilian Bermejo-Luque © Lilian Bermejo-Luque. Informal Logic, Vol. 34, No.3 (2014), pp. 311-336. 320 Why?” “— Because A is like B”. For example, in saying something like “That strategy didn’t work last time; so, it won’t work now”, we are reasoning analogically. That is, we conclude that a given strategy won’t work from the fact that it didn’t work earlier because we take both situations to be (quantitatively) alike. In turn, in “don’t ask Jane to pay her bill; you wouldn’t ask such a thing of a goddess”, it is the (qualitative) comparison of Jane with a goddess which justifies the inference from “you wouldn’t ask a goddess such a thing” to “don’t ask Jane to pay her bill”. In sum, within the argumentative uses of analogies, we must distinguish between using analogies as data or premises, and using analogies as backings or reasons for the inference- claim. And both quantitative and qualitative analogies can be used as data or backings for both deductive and non-deductive arguments. In what follows, I will only focus on conclusive analogical arguments having a priori (qualitative) analogies as their reasons, because these are the type of arguments Govier, Guarini, Waller and Shecaira have discussed. Contrary to them, I will argue that some of these arguments can be deductively valid without involving further premises. However, this will amount to a defence of Govier’s account of a priori analogies – or, at least, of her view that analogical arguments can be conclusively valid without involving universal claims as premises. 4. The debate Remarkably, Govier, Guarini, Waller and Shecaira agree on the idea that inductive analogical arguments are structurally different from non-inductive ones.7 They also agree on naming non-inductive analogical arguments a priori. As already pointed out, this would be a category mistake; “a priori” should be used for naming types of analogical claims, not types of arguments. Yet, for the sake of following their naming and line of reasoning, I am going to assume that it makes sense to establish a category called ‘a priori analogical arguments’ which consists of non-inductive analogical arguments having qualitative analogies as reasons. In principle, non-inductive arguments 7 This is something I have criticised in Bermejo-Luque (2012: 3-4) as a merely ad hoc claim having to do with their corresponding assumptions on argument analysis as a matter of adopting one or another argument schema. I go back to this discussion in sections 5 and 6. Deduction without Dogmas © Lilian Bermejo-Luque. Informal Logic, Vol. 34, No.3 (2014), pp. 311-336. 321 might have been regarded, simply, as deductive arguments – that is, as arguments whose conclusions are meant to follow by necessity and whose validity is a matter of whether or not this is actually so.8 However, neither Govier and Guarini, on the one hand, nor Waller and Shecaira, on the other, were willing to call their target objects “deductive analogical arguments.” For they all seem to assume that, if these arguments are deductive, it is because they are formally valid (i.e. they include certain universal premises turning them into valid syllogisms or instances of modus ponens) and their point of discrepancy is on whether a priori analogical arguments involve universal claims among their premises and, thus, whether or not they are deductive (in that sense). Because Govier and Guarini think that a priori analogical arguments do not involve universal claims as premises, they refuse to name them “deductive.” In turn, because Waller and Shecaira wish to defend the position that such arguments involve universal claims as premises, they cannot prejudice the issue by naming these arguments “deductive.” Certainly, since Trudy Govier has championed the informal logic approach to argument evaluation, such identification of deductiveness with formal validity seems strange indeed. But the reason for this incidental identification is that, in distinguishing between deductive, inductive and analogical arguments, Govier, and the discussants following her, were adopting the Aristotelian distinction between arguments from the general to the particular, from the particular to the general, and from particular to particular, respectively. Thus, Waller and Shecaira would be rejecting the view that a priori analogical arguments are arguments from particular to particular after all; or better, they would be holding that if such arguments were arguments from particular to particular, they could never be good. However, this Aristotelian distinction is old-fashioned for good reasons. Notably, it is too influenced by the syllogistic, according to which, from two particular premises, no conclusion 8 The idea that some properties of arguments depend on the possibility of presenting arguments as being thus and so is consubstantial to the very concept of argument. Indeed, we all are familiar with the standard definition of arguments as “any group of propositions of which one is claimed to follow from the others, which are regarded as providing support or grounds for the truth of that one” (Copi and Cohen (2002: 6), my italics). In my view, this is so because arguments are, first of all, speech-acts –that is, propositional contents put forward as being thus and so and having such and such properties— even though the standard definition makes no reference to their pragmatic nature. Lilian Bermejo-Luque © Lilian Bermejo-Luque. Informal Logic, Vol. 34, No.3 (2014), pp. 311-336. 322 follows of necessity. This is why it happens to be misleading out of this realm. What should we say of arguments like “so it’s false that she didn’t come; therefore, she came indeed” or “John is either at home or working, and he is not at home; so, he is working”? Are they deductive, or arguments from particular to particular, or both? Saying that they are deductive in this Aristotelian sense by merely positing some implicit general claim seems quite ad hoc, and saying that they are conclusively valid but not deductively valid raises the problem of explaining what this means.9 According to Govier, a priori analogical arguments are conclusive;10 but, she says that they cannot be taken to be deductive because, in general, we cannot take universal claims be part of them. At least, she argues, such universal claims should not be taken to be implicit premises; for, otherwise, our interpretation of a priori analogical arguments would make mere syllogisms of them, rendering the analogies themselves redundant and leaving out of our analyses all which is characteristic of this way of arguing. In her account, a priori analogical arguments amount to a special type of inference precisely because they can be conclusively valid in absence of 9 This is, after all, Govier’s main debt: to explain what is being conclusively valid if not being deductively valid. 10 Consider this remark by Govier: If argument A is invalid and argument B is "just like" it, then B is invalid too. The point would seem to be established entirely conclusively. But of course no two argument or cases are ever just alike. What is meant, in effect, is 'alike in all relevant respects'. And whether two cases are alike in all relevant respects is something that is always open to further discussion. (Govier, 1989: 143) Because whether or not “all relevant respects” have been taken into account depends on whether or not arguments A and B can be said to be alike in the relevant sense (a sense that determines the meaning of “just like” after all), this argument is certainly conclusively (deductively) valid: its conclusion cannot be false if its premises are true. Yet, because the arguer does not predetermine the way in which A is like B, it might be the case that her addressee is willing to take for granted the premise (by adopting a different sense of “just like” – for example, as “they both are really difficult to follow,” “they are the same rubbish,” “they are written with a similar typography,” etc) and still resist the conclusion. As argued before, the meaning of analogical claims is characteristically vague, and this is why they leave open the possibility of taking them to be true and yet taking them to imply (even necessarily imply) contradictory conclusions. As I see it, with this case Govier would be pointing at something Freeman (2013) and myself (Bermejo-Luque, 2012) have more explicitly argued for, namely, that analogical arguments can be both deductive and defeasible. I go back to this question in section 5. Deduction without Dogmas © Lilian Bermejo-Luque. Informal Logic, Vol. 34, No.3 (2014), pp. 311-336. 323 universal premises –which, on the other hand, she admits, may be “implied by the argument” (Govier, 1989: 148). Contrastingly, Waller (2001) contended that a priori analogical arguments should be interpreted and analyzed in terms of the following schema: 1. We both agree on case a 2. The most plausible reason for believing a is acceptance of principle C 3. C implies b (b is a case that fits principle C) 4. Therefore, consistency requires accepting b This schema would not make the analogy redundant, as we would need it in order to ascertain the universal claim that is part of the argument, according to Waller. Yet, it would render a priori analogical arguments deductive.11 In response to Waller’s article, Govier (2002) pointed at what I take to be the key issue in this discussion, namely, the role of the relationship between argument analysis (i.e., the activity of portraying the structure of arguments as logical devices) and argument interpretation (i.e., the activity of ascertaining the meaning of arguments as pieces of communication). In Bermejo-Luque (2012), I pointed out that unless our models for argument analysis are based on our models for argument interpretation, the question of whether or not a certain type of arguments has this or that structure will only get petitio principii answers: -­‐ “These arguments are/aren’t deductive because we have to analyze (and eventually reconstruct) them according to this/that logical schema” -­‐ “Why should we?” -­‐ “Because they have this/that logical structure” -­‐ “How do you know?” -­‐ “Because they have to be analyzed (and eventually reconstructed) according to this/that logical schema” If the task of portraying the logical structure of a certain type of argument is a matter of following our intuitions about what is structurally going on in such arguments, then we will have nothing, apart from our confronted intuitions, to appeal to in order to grant our claims that these arguments have/haven’t a 11 See Guarini (2004: 164, footnote 2) for an explanation of why this argument schema can be taken to be deductive after all. Lilian Bermejo-Luque © Lilian Bermejo-Luque. Informal Logic, Vol. 34, No.3 (2014), pp. 311-336. 324 deductive structure. Contrastingly, if our model for argument analysis is based on a model of argument interpretation, our claims on the alleged structure of such arguments will be supported by an independent theory – a theory that, on the other hand, is the first instrument we must use when dealing with natural language argumentation, as it is needed in order to get the meaning of argumentative speech-acts as actual pieces of communication. Govier is sensitive to the problems of reconstructing arguments in order to make them fit our schemas. But her strategy is just to insist that (many) a priori analogical real arguments do not make any explicit reference to principles or general claims that could turn them into deductively (formally) valid arguments.12 On the other hand, as Govier pointed out, the fact that in arguing by analogy we do not have to appeal to general principles explains, at least in part, the characteristic charm and rhetorical power of using analogies for drawing conclusions. Often, we use analogies when we do not have a principle at hand or do not want to commit ourselves to one such principle in so many words. Consequently, according to Govier, if we impose an argument schema for analyzing the type of argument that requires the inclusion of general principles, we will probably be unfaithful to the speaker’s words–or think good a priori arguments not fitting such schema to be bad ones. In other cases, interpreting these arguments as deductively valid will make them less cogent (not only less persuasive) than what they really are. Besides, at times, such unfaithfulness will amount to imposing reconstructions that will turn the dialectic into a straw man fallacy. In Govier’s words, Waller’s case “is flawed due to his failure to distinguish between an argument per se, as articulated by its author, and a philosophical reconstruction of that argument” (Govier, 2002: 155). In her view, principles or universal claims are not implicit premises in (most) analogical real arguments, and we should not force our analyses of analogical real arguments in order to make them fit our schemas, but rather look for schemas that properly deal with their special characteristics. According to Govier, then, the evaluation of analogical argumentation would depend on the ARG (acceptability, 12 Certainly, such strategy does not work for convincing those, like Shecaira, who have opposite intuitions. This is why an important part of my argument in the paper that Shecaira criticises was to point at the need of linking our theories of argument analysis to our theories of argument interpretation. However, Shecaira’s paper does not address this question in his defence of deductivism. Deduction without Dogmas © Lilian Bermejo-Luque. Informal Logic, Vol. 34, No.3 (2014), pp. 311-336. 325 relevance and good ground) criteria that characterize her approach to argumentation appraisal in general: The analogy will be a good one insofar as the analogue and the primary case share all the features which are logically relevant to the conclusion. Whether this is the case is something we can determine a priori, from reflective examination of the cases (Govier 2013, p. 328) But Shecaira (2013) finds this model wanting. In his view, the point of Waller’s endeavour was to avoid what he takes to be a consequence of Govier’s approach: namely, that “a debate prompted by an analogical argument will amount to no more than a battle of intuitions” (Shecaira, 2013: 422). This is why he provides a defence of Waller’s model against Govier’s, Guarini’s and my own criticisms. This defence involves a reinterpretation of Waller’s proposal. For, in his view, “Waller’s schema does not represent analogical arguments simply as deductive inferences, but rather as complexes of two inferences only one of which is deductive” (Shecaira, 2013: 407). In addition, Shecaira suggests that Waller’s formulation involves some rhetorical features that are dispensable. Thus, according to him, Waller’s schema should be outlined in the following way: Premise: A Inductive (partial) conclusion: Plausibly, Principle C Premise: if C, then B Deductive (final) conclusion: B13 Against Govier’s criticism that most everyday analogical argumentation does not involve principles as explicit premises, Shecaira’s main point is to insist that his schema – just as any other argumentative schema, to be fair – is meant to fulfil a double function: firstly, to provide a starting point for analyzing the corresponding arguments, and secondly, to provide a basis for evaluating them. Insisting on this latter function, Shecaira points out that his defence of Waller’s account involves “a normative claim to the effect that analogical arguments that do not fit Waller’s schema are defective or sub-optimal instances of their kind.” (Shecaira, 2013: 407-408). Thus, according to Shecaira, (his amendment of) Waller’s schema: 13 Actually, Shecaira’s model corresponds pretty exactly with what Aristotle called an argument from example (paradeigma), as described in the Rhetoric (Rhetoric 1402b15) and the Prior Analytics (Pr. An. 69a1-15ff), but not to an argument from likeness (homoiotes) proper. (Cfr. Bartha (2013) for aditional references). Lilian Bermejo-Luque © Lilian Bermejo-Luque. Informal Logic, Vol. 34, No.3 (2014), pp. 311-336. 326 (…) is not meant just as an accurate representation of analogical arguments crafted by professionals; it is also a model informed by normative assumptions about what makes an analogical argument a good argument (or at least a better argument than other possible instances of its kind). (Shecaira, 2013: 423) 5. Problems with Shecaira’s account According to Shecaira, his amendment of Waller’s schema “is an accurate representation, not of analogical arguments generally, but of analogical arguments as they are characteristically formulated by professional moral philosophers and judges” (2013: 430). This is an empirical claim that he supports with some evidence from law court argumentation – which, in my view, is not compelling and, at times, is frankly misleading, (see Bermejo-Luque (forthcoming – b) for details). Yet, together with this empirical claim, there is the normative claim that a priori arguments not fitting the Waller-Shecaira schema are defective in some sense. But what does this latter claim mean? Even if it were true that professional moral philosophers and judges use only a priori arguments fitting this schema, this is not enough to conclude that arguments that do not fit it are wrong. At best, that fact could be a symptom of their falling short of being right, but we would still need a proper explanation of why they are defective. Apart from “defective,” Shecaira calls these arguments not only ‘unprofessional’ but also ‘sub-optimal’. So, what does it mean for an argument to be sub-optimal in Shecaira’s sense? Is sub-optimality a type of rhetorical flaw? Well, not really. In principle, a priori arguments not fitting Waller-Shecaira’s schema are frequently successful at persuading their addressees, and they can be deemed optimal in that they are both persuasive enough and short enough for slogans, quick advices, introductions or summaries, for example. For advertisers, politicians – or even lawyers, at least, at times – this is the most professional and optimal an argument can be. Is, then, “sub-optimal” a type of dialectical flaw like that of making a straw man, insufficiently responding to criticisms, or illegitimately shifting the burden of proof? As I’ve pointed out, Shecaira’s main complaint regarding a priori analogical arguments not fitting his amendment of Waller’s schema is that, in his view, a debate prompted by any of them “will amount to no more than a battle of intuitions.” But, as I see it, there must be something wrong in this line of reasoning. After all, debates Deduction without Dogmas © Lilian Bermejo-Luque. Informal Logic, Vol. 34, No.3 (2014), pp. 311-336. 327 always provide a good field for battles of intuitions. For, any of the reasons that we offer in support of our claims can be further argued for, but this process must eventually come to an end; if the outcome is agreement, such end is the point in which arguer and addressee share intuitions, and if the outcome is disagreement, the end is the point in which arguer’s and addressee’s intuitions just clash. Regarding analogical argumentation, agreement of intuitions may happen as soon as a bare analogy is put forward or as late as numerous subsequent reasons have been adduced in its support or in support of the view that such analogy supports the conclusion (that is, in support of the warrant of the argument, following Toulmin’s model). Certainly, moral principles may do the trick for the latter; sometimes, they can provide a rationale for showing our inferences from certain analogies to certain moral claims to be right. However, in order to avoid a battle of intuitions, both reasons for the analogy and reasons for the warrant (including principles) will require either further reasons or a later agreement of intuitions. Consequently, it seems excessively demanding on a priori analogical arguments to state that they can only be good as long as they do not provide fields for battles of intuitions; no argument would be good on such strict dialectical standard. We are then left with the idea that, by “sub-optimal,” Shecaira is naming a logical defect. In his account, a priori arguments not fitting his amendment of Waller’s schema would be logically flawed. However, the plausibility of this claim depends on the conception of logical goodness that we endorse. And at this point, we can see that Waller and Shecaira’s defense of a deductivist account of conclusive analogical arguments against Govier’s proposal should have been backed by a proper defense of formal logic as the right standard for logical goodness. For, in their view, we have seen, a priori analogical arguments can only be conclusively valid if they are formally valid.14 For her part, Govier just agrees that for these arguments to be deductively valid, they must be formally valid. However, she contends that they can be conclusively valid without being deductively valid – that is, “formally” valid in her approach, which is in line with her informal conception of logical validity. In turn, as I tried to show in Bermejo-Luque (2012), if we adopt a substantial conception of logic like the one Stephen Toulmin 14 Actually, Shecaira’s point is that the part of a priori analogical arguments that is deductively valid is formally valid (i.e., the part corresponding to the second sub-argument in his amendment of Waller’s schema) Lilian Bermejo-Luque © Lilian Bermejo-Luque. Informal Logic, Vol. 34, No.3 (2014), pp. 311-336. 328 developed in The Uses of Argument (1958), we can deem some conclusive analogical arguments to be deductively valid without conceding that they involve universal premises that make them formally valid. Actually, my proposal was meant to be more than that, as it was based on a normative model for argumentation that links a model of argument analysis to a theory of argument interpretation. This model is the pragmatic- linguistic reconstruction of Toulmin’s model developed in Bermejo-Luque (2011), which characterizes argumentation as a second order speech-act complex by means of Bach & Harnish’s Speech Act Schema (Bach & Harnish, 1979).15 But why should we prefer such substantial conception of logical goodness and the corresponding model of argument analysis based on a theory of argument interpretation? In my view, there are practical reasons having to do with the problems of reconstructing natural language arguments in order to make them fit (either formal or non-formal) argument schemas, and more theoretical reasons having to do with the very project of dealing with logical goodness independently from a theory of argument interpretation. The peculiarities of conclusive analogical arguments make these reasons particularly striking. Let me start by the former kind of reasons. As we have seen, Shecaira does not deny that there are a priori analogical arguments that do not fit his amendment of Waller’s schema. Rather, his point is to insist that, unless we can find a universal principle – i.e., the conclusion of an allegedly implicit inductive sub-argument – for them to fit it, such arguments will be defective. But, where is the gain in condemning these arguments? In principle, bare analogical arguments – i.e., arguments of the form “A is like B; so, you’d better not do it,” “A is like B; so, there is nothing wrong with it,” “A is like B; so, it’s up to you,” etc – are quite frequent. And this must be so 15 As explained in more detail in Bermejo-Luque (2011, Chs. 3 and 4) and (2012), this characterization of acts of arguing as second order speech-acts by means of Bach & Harnish’s SAS enables us to deal with both indirect and non-literal argumentation (like analogical argumentation). In this model, any act of arguing would have the following structure: (omr)Premise _________therefore___ (emx)(omt) Conclusion | since (omi)Warrant: “if R, then C” And argumentative discourse will be a series of acts of argumenting, more or less related to each other according to this schema (See for example Bermejo-Luque (2012) for a brief explanation of this model and the theory of evaluation that results from it) Deduction without Dogmas © Lilian Bermejo-Luque. Informal Logic, Vol. 34, No.3 (2014), pp. 311-336. 329 precisely because they seem perfectly fine to many people; otherwise, they would be a bad option, even from a merely rhetorical point of view. Should we then say that these people (including me) are stupid or unreasonable? Well, it seems appropriate to refrain from the temptation of declaring a good deal of people unreasonable – at least, whenever we have an option not to do so. Shecaira might respond that we take at least some of these arguments to be good because we take them to implicitly involve the corresponding principles. However, as Govier pointed out, when an arguer does not explicitly commit herself to one of such principles, supplying one of our own may involve an uncharitable account of her actual views. Consider, for example, this short version of Judith Jarvis Thomson’s famous argument on abortion (Jarvis Thomson, 1971): “Becoming pregnant as a result of rape is like becoming physically connected by force to someone who would certainly die if disconnected. Therefore, there is no obligation to carry on a pregnancy resulting from rape.” Many different principles could be used in order to make this argument fit Waller-Shecaira’s schema. Among them, these three: (1) “There is no obligation to stay linked to a human being, even if that person will die otherwise, if we have been forced into that situation” (2) “There is no obligation to save another human being’s life for whose situation we are not responsible” (3) “We have a right to free ourselves from whatever we have been forced to” Yet, whereas (1) seems quite plausible as a general principle, (2) and (3) seem much harder to defend. Do we really have no obligation to save a drowning stranger, for example? Can you leave your little sister alone at the fair just because your mom obliged you to take her there? As Govier might argue, taking Jarvis Thomson to be arguing in consonance with (2) or (3) would be quite unfaithful to her point of view and would make her position much less cogent than what it actually is. In fact, even principle (1) may be taken to misrepresent the point of this argument. After all, the analogy pivots on only two particular cases and what they are supposed to have in common, whereas Lilian Bermejo-Luque © Lilian Bermejo-Luque. Informal Logic, Vol. 34, No.3 (2014), pp. 311-336. 330 principle (1) is supposed to subsume additional cases, some of which might happen to be less evident.16 Certainly, principles can provide something Shecaira seems to demand from analogical arguments, namely, a rationale, an overall justification for our inferring the conclusion from the analogy. But this is not the only way to provide such justification,17 and principles do so at the cost of turning the arguer’s claims stronger than what they originally were. Besides, general principles may be rather unnecessary if the arguer and addressee agree that the two cases are alike in the relevant respect, whether or not they take other cases to be similar to these as well. At any rate, as argued above, pointing out a rationale may be dialectically required at times, but it cannot be a condition of argument goodness; for, such condition involves a vicious infinite regress, and no argument could really meet it. However, Shecaira thinks that making analogical arguments fit his schema by supplying them with general principles does not pose any problem at all: If two different interpretations are equally compatible with the evidence, then the interpreter ought, charitably, to opt for the interpretation that represents the argument in the most favourable light (Shecaira, 2013: 418). This is certainly an expeditious solution. However, it hides what I take to be Govier’s main point, namely, that the Principle of Charity is a principle guiding interpretation, not a rule of courtesy recommending making the best of the speaker’s argument. Thinking of the principle as a rule of courtesy means that it can only be applied once we have interpreted the speaker’s words as an argument, which in turn means that it is not a necessary requirement for interpretation. Contrastingly, as an interpretative principle, the principle is a necessary requirement for interpretation. We have to be charitable because, otherwise, there is nothing to evaluate, no meaning at all. 16 For example: what about if you only have to stay linked for one second to that person, and the people who forced you into that situation will be properly punished anyway (so that there is no risk that this becomes a legitimate way of saving lives), or if by staying linked you save not only one person but a million people? 17 For example, we may also proceed by reductio: we justify our warrant because its negation is untenable or has untenable consequences. This is why moral particularists are not bounded to irrationality or mere clashes of intuitions when they argue by analogy. Deduction without Dogmas © Lilian Bermejo-Luque. Informal Logic, Vol. 34, No.3 (2014), pp. 311-336. 331 In interpreting arguments, we must appeal to the Principle of Charity if, and only if, the arguer’s actual words are not enough for making herself to be understood. Thus, prior to the application of the principle, there is nothing to appraise, and after its application we have both, the speaker’s actual meaning and a piece of argumentation that can be appraised. This is why an interpretative option such as Govier’s, which limits the amount of words that is necessary to attribute to the speaker which she did not actually say, is preferable. Such option amounts to acknowledging that the speaker’s actual words were indeed enough for conveying her communicative intentions – that is, the piece of argumentative content that she has put forward to be understood and appraised. Why should the appraisal of argumentation require anything beyond this kind of “charity”? How could we justify any other demand? In opting, as Shecaira recommends, for the general principle that seems most promising to us, we are reconstructing rather than interpreting arguments. Such general principle may make the argument under consideration better from our perspective, but it can make it worse from someone else’s, including the arguer herself who may feel misunderstood. This is so because reconstruction involves accessory intake, whereas interpretation is about getting to understand the speaker’s meaning. No more, no less. In order to illustrate this point, let us follow Shecaira’s proposal, and think for example of a person holding that we have an obligation to save human beings whenever it is at our hands. That person will take principles (1), (2) and (3) to be false, but may take something like principle 4 to be true: (4) “There is no obligation to stay linked to an organism (not a human being), even if that organism would die otherwise, if we have been forced into that situation.” Which principle should she adopt for arriving at an instance of Waller-Shecaira’s schema from our version of Jarvis Thompson’s argument? If she opts for (1), (2) or (3), she will be attributing to the arguer a principle that she takes to be false. Yet, she cannot just opt for 4 either, because she will not take it to follow from premise 1 of the reconstruction of the original argument, according to Waller-Shecaira’s schema. Actually, it is difficult to see how she could opt for any fitting principle at all, as she cannot take anything to follow from premise 1 in this schema, which she takes to be false: Lilian Bermejo-Luque © Lilian Bermejo-Luque. Informal Logic, Vol. 34, No.3 (2014), pp. 311-336. 332 Premise 1: We have no obligation to stay physically connected to a person if we have been forced into that (even if that person would die otherwise) Inductive (partial) conclusion: Plausibly, C1/C2/Cn Premise: if C1/C2/Cn, then no one should be obliged to carry on a pregnancy from rape Deductive (final) conclusion: No one should be obliged to carry on a pregnancy from rape Does our interpreter’s inability to find a principle mean that she didn’t really understand the argument, that she has no adequate interpretation of it? Shecaira doesn’t seem to care too much about the answer to this question.18 His point is rather to insist that, because his schema is normative, if the interpreter does not get a reconstruction that she can take to be good, then she must deem bad the original argument. Yet, what does this normative claim mean? That is, by fitting Waller-Shecaira’s schema, is an argument rendered good or just shown to be good? For, if we take arguments to be good in themselves, prior to making them fit the schema, then we may take the schema to be only a means among others for showing their goodness. (Other means could be, for example, Govier’s ARG criteria, Guarini’s or my own.) But if we take this type of argument not to be good until made to fit the schema, then we will be implicitly endorsing the view that it is not these very arguments, but the corresponding reconstructions fitting the schema, that are good. This is, in sum, the problem of transferring the evaluation of natural language arguments to a theory of analysis that is not properly backed by a theory of argument interpretation.19 Besides, consider for example what happens if our interpreter above takes embryos and foetuses to be human beings. In this case, she may be perfectly willing to accept the original premise of our argument, namely, the analogy “becoming pregnant as a result of rape is like becoming physically connected by force to someone who would die otherwise,” even though, from this premise, she would rather 18 However, at least from a cognitive perspective, this question seems relevant: disagreeing is not always a matter of not understanding. 19 Regarding the use of formal schemas for logical appraisal, the dilemma is quite the same: we must either adopt a concept of validity which is not essentially formal, or take arguments not to be logically valid until formalized. This is why some formal logicians think of formal logic as a tool for “discovering” natural language validity, whereas others think of formal systems as (partial) definitions of the concept of validity. I have dealt with the problems of both conceptions of formal logic in Bermejo-Luque (2008). Deduction without Dogmas © Lilian Bermejo-Luque. Informal Logic, Vol. 34, No.3 (2014), pp. 311-336. 333 conclude that we have an obligation to both carry on pregnancies from rape and stay connected to anyone who would die otherwise. To be fair, in such a case, Shecaira’s instructions for reconstructing the original argument prove to be neither an answer to the question of how to interpret the arguer’s actual words, nor able to explain the interpreter’s agreement with the arguer’s original premise. Moreover, as I explained in my former paper (Bermejo- Luque, 2012: 21), the possibility of agreeing on the analogy but refusing the conclusion is an important feature of deductive analogical arguments. As I explained there, analogical arguments can be deductive (in the sense of having conclusions meant to follow of necessity) but defeasible at the same time (in the sense of being subject to counterexamples).20 However, Shecaira criticises my account of this feature and deals with it in the following way: Bermejo-Luque also seems to believe that her account has the advantage of being able to explain how an analogical argument can be deductive yet defeasible. But I think that Waller’s schema fares just as well by reference to this criterion. Take Bermejo-Luque’s example cited above: to deny that having sex with children is impermissible is to deny the arguer’s preferred way of treating the source of the analogy. For our purposes, this is a case of denying the truth of 1 in Waller’s schema. Anyone who holds a deductivist account of an argument type will accept that the relevant arguments can be defeated – in the sense of being proved unsound – if any of their premises are shown to be false. It is unclear that deductivists need to accept the defeasibility of analogical arguments in any other sense in order to accommodate the type of example provided by Bermejo-Luque. Yet, this is a peculiar way of understanding defeasibility, which, in principle, is not a matter of the possibility of having false 20 This apparently paradoxical claim makes sense in our model because, contrary to formal logic, this model avoids conflating the concepts of deductiveness and validity. In our model, a deductive defeated argument is an invalid one, as its conclusion is meant to follow of necessity while it doesn’t. As explained in Bermejo-Luque (2011, Ch. 4), and more succinctly in Bermejo-Luque (2012), in our model, arguments are deductive when their warrants are propositions that, in case of being true, would be necessary truths. For example, because they are a conceptual, a mathematical, a logical matter, etc. And deductive arguments are valid in case their warrants are necessary truths indeed. In turn, inductive arguments are those whose conclusions are meant to follow with less certainty than “necessarily.” For example, “(more or less) probably,” “presumably,” “plausibly,” etc. And the validity of inductive arguments would be a matter of whether or not this is actually so, that is, whether or not the corresponding warrants are more or less probable, presumable, plausible, etc. Lilian Bermejo-Luque © Lilian Bermejo-Luque. Informal Logic, Vol. 34, No.3 (2014), pp. 311-336. 334 premises, but of the possibility of taking premises to be true and not taking the conclusion to follow from them. Actually, the latter is what the interpreter in our example would take our argument to instantiate, and the type of situation that Shecaira’s proposal for argument reconstruction seems unable to properly deal with. For Shecaira, our interpreter would just disagree with the arguer’s preferred way of treating the source of the analogy, but such account leaves no trace of her agreement with the analogy itself. 6. Deductivism At this point, we may ask: “what is the gain of Waller and Shecaira’s deductivist proposal?” Deductivism, as defended by argumentation theorists such as Leo Groarke (1999) or Dale Jacquette (1996, 2009), is the view that in order to determine whether an argument is good or bad, we must make it fit a deductively valid argument schema. As David Godden (2005) has pointed out, deductivism involves both a thesis on the interpretation of natural language arguments (interpretative deductivism) and a thesis on the standard of argument goodness (evaluative deductivism). These two theses are related; for the deductivist, we must interpret natural language arguments so as to make them fit deductively valid schemas because this is a means to show that their conclusions cannot be false if their corresponding premises are true. In principle, this is deemed to be a sound standard of inferential goodness in that it shows conclusions to follow of necessity from premises. According to the deductivist approach, if in order to fit at least one deductively valid schema we must incorporate false premises to the original argument, then the argument as a whole must be deemed bad. Remarkably, this strategy of argument appraisal is a means to determine whether arguments are overall good – that is, not just good in their inferences, but also good in their premises. In other words, deductivism is useless as a tool for determining mere logical goodness – that is, the goodness of the inference independently of the goodness of the premises. After all, by making an argument fit the schema of a deductively valid argument, we are already making it to be valid.21 21Or, maybe, showing it to be valid; this is not clear, as argued before. On the other hand, deductivists do not necessarily take formal validity as the standard for argument goodness, as authors such as Groarke are willing to accept materially valid reconstructions of arguments as means to show their validity. Deduction without Dogmas © Lilian Bermejo-Luque. Informal Logic, Vol. 34, No.3 (2014), pp. 311-336. 335 I have already argued against the idea of reconstructing natural language arguments in order to make them fit argument schemas independently of a model of argument interpretation. At the very least, deductivism makes some sense as a tool for determining argument goodness as it provides a reason to justify deduction as a standard for argument goodness. As I’ve pointed out, arguments that fit a deductively valid schema have conclusions that follow of necessity from their premises and this is a suitable account of logical goodness. Contrastingly, according to Shecaira, One crucial qualification is that the schema represents analogical arguments as complexes composed of one deductive inference (hence “deductivism”) but also of one non-deductive subargument. (Shecaira, 2013: 406 (abstract)) As he explains, this complicated proposal results from the need to avoid the problem that, by adding universal principles, the analogies in analogical arguments are rendered redundant. However, by doing so, the schema is no longer a means of showing that the conclusion cannot be false if the premises are true. That is, Shecaira’s deductivist proposal does not even take advantage of deductivism’s main insight: its plausibility as a standard of logical goodness. Thus, even if we are able to reconstruct an analogical argument following Shecaira’s instructions, we will not have at our disposal deductivism’s rationale for saying that such argument is good. As Govier has taught us, the problem with analogical arguments is that supplying allegedly underlying principles in order to make them fit (partially) deductively valid schemas is unwarranted. And it is also dispensable, I have tried to show, as there are less problematic ways to determine their goodness. Acknowledgements I wish to thank the two anonymous referees of this paper for their very helpful comments. The work presented in this paper has been financed by a Ramón y Cajal Research Fellowship of the Spanish Ministry of Science and Innovation and by the research projects FFI2011-23125 and FFI2011-24414 of the Spanish Ministry of Science and Innovation. Lilian Bermejo-Luque © Lilian Bermejo-Luque. Informal Logic, Vol. 34, No.3 (2014), pp. 311-336. 336 References Bach, Kent and M. Harnish (1979) Linguistic Communication and Speech- acts. MIT Press: Cambridge, MA. Bartha, Paul (2013), “Analogy and Analogical Reasoning,” in The Stanford Encyclopedia of Philosophy (Fall 2013 Edition), Edward N. Zalta (ed.), URL = . Bermejo-Luque (2008), “Logic as (Normative) Inference Theory: Formal versus non-Formal Theories of Inference Goodness”, in Informal Logic. Vol. 28/4 (2008), pp. 315-334 ____ (2011) Giving reasons. A linguistic-pragmatic approach to Argumentation Theory. Springer: Dordrecht ____ (2012) “A unitary schema for arguments by analogy,” in Informal Logic. Vol. 32/1 (2012), pp. 1-24 ____ (forthcoming-a) “The uses of analogies”, in H. Jales Ribeiro and F. H. van Eemeren (eds.) Systematic Approaches To Argument By Analogy, Springer: Dordrecht ____ (forthcoming-b) “Appeals to similarity in legal discourse: arguments from precedent and arguments from analogy” Burke, Michel (1985) “Unstated Premises,” in Informal Logic Newsletter, 7, pp. 107-118 Copi, Irving M. and Cohen, Carl (2002) Introduction to Logic. (11th ed.). Prentice Hall: Upper Saddle River, N.J. Freeman, James (2013) “Govier’s distinguishing a priori from inductive arguments by analogy: implications for a general theory of ground adequacy,” in Informal Logic, 32/2, pp. 175-194 Godden, David (2005) “Deductivism as an interpretive strategy: a reply to Groarke’s recent defense of reconstructive deductivism,” in Argumentation and Advocacy ,41, pp. 168-183. Govier, Trudy (1980) “More on deductive and inductive arguments,” in Informal Logic Newsletter, 2/3, pp. 7-8 ____ (2002) “Should a priori analogies be regarded as deductive arguments?,” in Informal Logic, 22/2, pp. 155-157 ____ (2013) A Practical Study of Argument (Enhanced seventh edition). Wadsworth: Boston Groarke, Leo (1999) “Deductivism within Pragma-dialectics,” in Argumentation, 13, pp. 1-16 Guarini, Marcello (2004) “A defence of non-deductive reconstructions of analogical arguments,” Informal Logic, 24/2, pp. 153-168 Hitchcock, David (1980) “Deductive and inductive: types of validity, not types of arguments,” in Informal Logic Newsletter, 2/3, pp. 9-10 Jaquette, Dale (1996) “Charity and the Reiteration Problem for Enthymemes,” in Informal Logic, 18, 1-15 ____ (2009) “Deductivism in formal and informal logic,” in Studies in Logic, Grammar and Rhetoric, 16/29, pp. 189-216 Jarvis Thompson, Judith (1971) “In Defense of Abortion,” in Philosophy and Public Affairs, 1/1, pp. 47-66 Shecaira, Fábio Perin (2013) “Analogical arguments in Ethics and Law: a defence of a deductivist analysis,” in Informal Logic, 33/3, pp. 406-437 Toulmin, Stephen (1958) The Uses of Argument. Cambridge University Press: Cambridge Waller, Bruce (2001) “Classifying and analyzing analogies”, in Informal Logic 21/3: 199-218