30.4Bottingfinal © David Botting. Informal Logic, Vol. 30, No. 4 (2010), pp. 413-434. A Pragma-Dialectical Default on the Question of Truth DAVID BOTTING Institut de Filosofie da Linguagem FCSH Universidade Nova de Lisboa LISBON, PORTUGAL Email: davidbotting33@yahoo.co.uk Abstract: The problem with the pragma-dialectical view, it has been argued, is that it takes argumentation as aiming at consensus rather than truth or justified belief. The pragma- dialecticians often imply that an ar- gumentative process aiming at con- sensus in a way constrained by the “Ten Commandments” will in the long run converge on epistemically favourable standpoints. I will argue that they are right provided: (i) pragma-dialectics is construed, as they say, as a theory of criticism; (ii) pragma-dialectics and the other theo- ries of argumentation have in common the requirement to account for the fallacies. Résumé: Keywords: dialectical models; epistemic normativity; agreement; truth; critical rationalism 1. Introduction The pragma-dialectical theory makes three claims: The Functional Claim: The function of argumentation is to resolve disputes on the basis of the better argument. The Instrumental Claim: Following the pragma-dialectical rules conduces to the reasonable resolution of disputes. David Botting 414 The Normative Claim: Standpoints that have the unqualified con- sensus of all participants in the dispute will generally be epistemi- cally sound.1 The Normative Claim, I should say, is more implied than claimed, the strength of this implication varying among the members of the Amsterdam School. I wish to make this claim in a relatively strong way—not only are the norms of pragma-dialectics and epistemic norms not necessarily in conflict and even often in collaboration (a weak form of compatibilism) but the norms of pragma-dialectics have epistemic normativity inherently (a strong form of compati- bilism), which is to say that the rules put forward in pragma- dialectics (the Ten Commandments) are truth-conducive. I do not make the even stronger claim that the Ten Commandments are even more truth-conducive than any other possible set of rules such as those offered by an approach oriented towards justified consen- sus (Lumer 2000, 2010) or that what I have to say will be in any way decisive in determining whether the pragma-dialectical theory is better than the others on offer. In this sense my intentions are more modest and will consist of discussing objections to the three claims and presenting a way of looking at the pragma-dialectical theory that nullifies these objections. Irrespective of whether pragma-dialectics is the best theory, it is not that bad, at least, not in the ways that it has been recently criticized; whatever its defects, they are not epistemological. 2. Critical discussions, rational resolutions, and reasonableness (The Functional Claim) Resolution of a dispute being taken to be the principal aim of argumentation, critical discussion is an ideal model against which any argumentation can be normatively com- pared. What is the nature of this normativity? Critics of pragma- dialectics say that norms of argumentation should be epistemic, and that pragma-dialectics is deficient for this reason; the aims of unqualified consensus and justified consensus are, Lumer (2010) argues, incompatible, there being no necessary connection between 1 The Functional and Instrumental Claims can be found in any introduc- tory text to pragma-dialectics, e.g., en.wikipedia.org/wiki/Pragma- dialectics. The Normative Claim is difficult to find explicitly stated, but I believe it to be the strongest, most interesting and most consistent way of interpreting van Eemeren’s notion of a “rational judge” (see Huss 2005 for this claim and Huss’s rejection of this kind of normativity, and also Lumer 2000, esp. footnote 2). At the very least, I think van Eemeren would approve of pragma-dialectics having this kind of epistemic norma- tivity if such could be defended, and it is my intention to defend exactly this view. Question of Truth 415 agreement and truth. The choices open to defenders of consensual views of argumentation are to deny that norms of argumentation should be epistemic—saying, in effect, “so much the worse for justification” as in Huss (2005)—or to deny Lumer’s claim that these aims are incompatible. An insufficient because trivial re- sponse would be that resolution has some kind of normativity, since for any procedure governed by rules those rules can be turned around and used normatively as a standard of correctness. Thus, in their recent defence of pragma-dialectics, Garssen & van Laar (2010, 138) expend a great deal of effort in refuting what nobody denied; they argue only that the pragma-dialectical rules are nor- mative (which, as I said, is trivially true) and leave unanswered the complaint that this normativity is not epistemic. On the contrary, they distance themselves from the epistemic optimism that informs Popper’s critical rationalism, content instead to limit their norma- tive claims to the management of disagreement that may adventi- tiously have positive epistemic effects; in other words, what I de- scribed as a weak form of compatibilism. I think that a stronger case can be made for genuinely epi- stemic normativity. In this paper I defend the view that critical discussion is an epistemically normative model, that is to say, to uphold the Normative Claim. It is conceded that the fact that many or all people agree to something does not mean that the thing agreed to is true or even justified, but the picture changes when we consider critical discussion as a process or mechanism. The ration- ale of pragma-dialectics—in particular its concepts of a critical discussion, rationality and reasonableness—is based on a version of Popper’s critical rationalism. The question is whether these concepts have epistemic normativity. I will consider each in turn and hopefully show that, considered together, they do. Critical discussions model the critical rationalist procedure of conjecture and refutation: standpoints are put forward and sincere attempts are made to falsify those standpoints. This may not be altogether obvious, since the respondent in a critical discussion asks for reasons in support of the standpoint and require the stand- point to be retracted if those reasons cannot be given or are found inadequate, without necessarily providing a positive reason for the standpoint’s being false, i.e., a falsifier. The rules governing criti- cal discussion are constructivist, derived from the constructivism of the dialogue logic of the Erlangen school (Lumer 2010). There are two things that I wish to say about this. One is that there is a reformulation of the Ten Commandments in 2004 that phrases the rules negatively, as prohibitions of argumentative moves, that is more transparently consistent with critical rationalism (see Zenker 2006 for a description of this change and its epistemological sig- nificance). The second is that persuading a respondent that a stand- point is to be accepted is always a matter of showing that the com- David Botting 416 mitment-store is inconsistent otherwise, catching out the respond- ent in the performative contradiction of incompatible commit- ments, or in Walton’s (1989) phrase that the standpoint is in the commitment-store’s “dark side”. The important point is that the requirement to provide reasons does not mean that this is not a fal- sificationist enterprise: what is being falsified is, so to speak, what is in the commitment-store.2 It is a necessary but not, of course, sufficient condition for a set of propositions to be true that they are consistent. Popper obviously takes this process of conjecture and refuta- tion to be epistemically normative; if he is right—as van Eemeren & Grootendorst (1988) suppose—and given that critical discus- sions model this process, then the rules regulating critical discus- sions (whatever these ultimately turn out to be—arguments must be and will be given that the Ten Commandments do actually regu- late) are also epistemically normative. Popper would not call the norm in question “justification”, but this does not mean that it is not an epistemic norm nevertheless. The epistemic norms in play in falsificationism are not those of traditional foundationalist episte- mology because we simply do not have the infallible basic beliefs required for a concept of justification that would allow us to say that the standpoint is justified or more likely to be true tout court, since unless some beliefs are certain, or at least have a determinate probability, one cannot apply the axioms of the probability calculus to them or say what probabilities beliefs based on them should have. It should be noted in this regard that, unlike the “basic be- liefs” of the foundationalists, Popper’s “basic statements” are sim- ply statements that the scientific community all agree to and are fallible. This lends scientific methodology to the possibility of be- ing modelled dialectically as a critical discussion. This critical discussion is an epistemically normative process that begins and ends with a consensus. Rationality is concerned with intersubjective agreement—in the opening stage to the propositions and inference types that can be used (starting points), and in the closing stage to the proposition that only these propositions and inference types have been used (intersubjective testing procedures)—and a resolution is described as a rational resolution or equivalently as convention-valid if it passes these tests. If it is decided that a proposition or an inference type was used that was not among those agreed to then the arguer has committed a fallacy, since the proposition or inference type in question is as much in doubt as the standpoint and is therefore powerless to make the standpoint more acceptable to the respond- 2 Thanks go to an anonymous reviewer for pressing me to respond to this issue. Question of Truth 417 ent (van Laar 9-12). This implies that an argumentation can only be criticized relative to its own rules of procedure. However, it is not- able that this is not an objective criterion—the fact that all partici- pants agree that premises and inference types have been used le- gitimately does not mean that this is true. Nonetheless, in line with its concept of rationality, pragma-dialectics is committed to the view that in such a situation the resolution is convention-valid and rational. I will argue later that although such mistakes leads to in- dividual resolutions that are not epistemically normative, it does not follow that an iterative process of critical discussion is not epistemically normative. Reasonableness is characterized by the attitude of the partici- pants: they intend to be swayed by logos and not by ethos or pa- thos; this is what pragma-dialectics calls the critical aim and what Popper calls the ‘rational attitude’.3 It is a presupposition of the critical discussion that participants have this attitude. This presup- position is embedded in the felicity conditions for a speech act complex of argumentation, in that one cannot argue (perform some part of this speech act complex) if one does not have the rational attitude. If one is trying to win the argument by any means possible then one is simply not arguing, by definition.4 This speech act is modelled dialectically, as we have seen, as the critical discussion. 3 Popper (1986, 225-31) points out that someone who is not prepared to take such an attitude cannot be persuaded to do so by use of argument or experience, because in order to be so persuaded he must, contrary to the assumption, take argument and experience as persuasive. There must be a faith in reason that cannot be supported rationally, or to put it another way, the truth-conduciveness of reason must be an intersubjectively shared assumption for which no further argumentation can be offered. 4 This does not rule out the possibility of the arguer having other attitudes as well, e.g., the rhetorical aim to be discussed shortly. What is meant to be excluded by this definition is arguer’s appealing in argumentation to premises or inference types they do not believe but think may be persua- sive nonetheless. I do not think there to be any inconsistency between this and the dialectical model’s preference for commitments rather than beliefs: I take commitments to be engendered not by mere utterances but by assertives, or whatever (not necessarily linguistic) act is analysed as an assertive. If I lie, then I do not commit myself to the truth of my utter- ance. On the contrary, on the correct analysis of the argumentation, which is to say, by analysing the speech act committed as a lie or equiva- lently as an assertive asserting the negation of what I literally say, I am committed to the falsity of my utterance. If I lie as part of a deliberate strategy of deception, as opposed to an oblique way of making assertions where I let it be known that I am lying, then I am not arguing. Lying is not the same as committing a fallacy. When a fallacy is committed, then argumentation is taking place, but is taking place incorrectly, as will be argued later. David Botting 418 Critical discussion is a normative model in one sense of the word “normative” because it models these speech acts, in performing which participants commit themselves to having the rational atti- tude in much the same way that Searle argues that people who make promises commit themselves to having the intention of fulfil- ling the promise (van Laar 2003, 9); the speech act ‘misfires’ if they do not have the relevant attitudes and psychological states. This in itself does not establish the Normative Claim because one can aim at epistemic goals while remaining arbitrarily far from ever achieving those goals, having chosen the wrong means to achieve them, e.g., fallacious forms of inference. These conditions or rules for performance of the speech act complex fall into two groups: identity and correctness. Identity rules are constitutive rules and dictate which attitudes and psy- chological states are relevant (one of which is, in the case of argu- mentation, the rational attitude). If and only if the identity condi- tions are satisfied, then argumentation is taking place, but when the correctness conditions are not satisfied, then although argumenta- tion has been offered it has fallen short of the ideal of “correct” argumentation, which is to say that it violates regulative rules of the critical discussion. The Ten Commandments are proposed as correctness/regulative rules (van Laar 2003, 21-22). There are two kinds of validity that a regulative rule can have: conventional va- lidity (which, as we have already seen, is just a matter of its being agreed to or adopted as a convention) and problem-validity (see van Laar 2003, 11-12 where problem-validity is called problem- solving validity). A regulative rule is problem-valid if it does in fact regulate, i.e., conduces, or belongs to a system of rules that conduces, to a reasonable as opposed to merely rational resolution. Van Laar (2003, 2) describes a reasonable resolution as follows: the stand- point is either conceded by the respondent or retracted by the pro- ponent and (1a) each has argued by giving what they consider their best argu- ments; (1b) the arguments have been critically tested by being subjected to critical questions; (1c) participants cooperate in helping each other perform (1a) and (1b); (2a) the opponent does not try to hinder the proponent in any other way; (2b) the proponent does not hinder the asking of critical questions; (2c) participants cooperate in helping each other perform (2a) and (2b). Question of Truth 419 Resolution can be reached without these conditions being satisfied, e.g., by not critically testing the standpoint. For a purely rhetorical theory of argumentation, this would be enough; rhetorical theories do not even require that the arguer believe the standpoint or the arguments she gives to support it. The pragma-dialectical theory certainly does not deny that arguers hope to win their arguments and that in addition to the critical aim arguers have this as their rhetorical aim. Nonetheless an argumentation that ended with no consensus but had been critically tested is preferable to a consensus that had been given a free ride.5 The claims that argumentation is correct and that the consensus is a reasonable consensus thus re- duce to the claim that argumentation is guided by rules that are both convention-valid and problem-valid. The Instrumental Claim is that the pragma-dialectical rules are problem-valid; the task, then, is to get the rules intersubjectively agreed to, after which they will be convention-valid also.6 What grounds can be given for acceptance of the pragma- dialectical rules as problem-valid? van Eemeren & Grootendorst (1988, 282) suggest the following: The problem-validity of the system of discussion rules as a whole can be rendered plausible only by illustrating that each rule fulfils a specific function in connection with fur- thering the resolution of a dispute . . . [I]t is possible with each of the formulated discussion rules to indicate pre- cisely which classical fallacies can be controlled through these rules. Methodically speaking, this seems to us the best test of the dialectical system of rules presented. In other words, they seem to be saying that the best way of show- ing that the rules lead to correct discussion is, using logic as a theory of criticism, to show contrapositively that incorrect discus- sion is a violation of the rules, and we know independently what constitutes incorrect discussion, namely the fallacies. Ultimately this is an empirical test. But in the meanwhile acceptance of the rules can be argued for pragmatically by showing that each type of fallacy can be correlated with a particular rule and stage of the dis- cussion. Although whether a speech act has a certain inescapable presupposition is a conceptual truth about it and whether what fol- 5 This is expressed (not altogether helpfully) by van Eemeren & Groo- tendorst (1988, 285-86) when they say that the function of the rules of critical discussion is not to maximize agreement (which could always be achieved trivially) but to minimize disagreement. This is also stressed by Garssen & van Laar (2010). 6 It is not clear whether the rules need to be agreed to because they are problem-valid, but it seems natural to take it this way. David Botting 420 lows from such presuppositions follows a priori, the pragma- dialecticians nevertheless try to provide a posteriori corroboration and pragmatic argumentation rather than involve themselves in interminable theoretical disputes with philosophers who do not share their assumptions—it would be arguing in a circle and a vio- lation of their own principles to provide more argumentation of a theoretical or philosophical sort. In other words, even a priori theo- ries have consequences that can be empirically tested, and often the best way of persuading another that your theory is the correct one is to show that it passes those tests, or putting it more succinctly, to show that it works. If an occurrence of a fallacy is found that does not seem to be a violation of the rules, then the system of discus- sion rules is at best incomplete and at worst mistaken.7 7 Interestingly, Habermas takes a similar approach. Showing that a par- ticular presupposition is inescapable is a matter of corroboration through counterexamples and the assertion that there is no alternative to a given presupposi- tion . . . must be checked against individual cases. To be sure, the intuitive knowledge of rules that subjects capable of speech and action must use if they are to be able to participate in ar- gumentation is in a certain sense not fallible. But this is not true of our reconstruction of this pretheoretical knowledge . . . we have to put our reconstructions up for discussion in the same way in which the logician or the linguist, for example, presents his theoretical descriptions. (Habermas 1990, 97) It is not completely clear here whether Habermas means by corroboration what the critical rationalists do; putting reconstructions “up for discus- sion” seems to be making the point that they must be actually accepted, which is to say in pragma-dialectical terms that they are convention- valid. It does seem to me that such acceptance would qualify as corrobo- ration and as epistemically normative. However, he says later that a reconstructive theory is dependent on indirect validation by empirical theories (he uses Kohlberg’s theory of moral consciousness) that are consonant with it. He uses the word cor- roboration again (Habermas 1990, 118) when he says: The empirical corroboration of an empirical theory that presupposes as valid the fundamental assumptions of a norma- tive theory cannot . . . pass for an independent corroboration of the normative theory. But independence postulates have been shown to be too strong in several respects. For example, the data used to test an empirical theory cannot be described inde- pendently of the language of the same theory. Similarly, two competing empirical theories cannot be evaluated independ- ently of the paradigms furnishing their basic concepts. On the meta- or intertheoretical level, the only governing principle is that of coherence. . . . [L]ooking for an independent proof is a Question of Truth 421 When Lumer (2010, 63-64) contends that pragma-dialectics lacks a theory of the function of argument and states the Functional and Instrumental Claims without argument I think that he mis- understands the order of explanation in the pragma-dialectical scheme of things. They do provide an argument, albeit a pragmatic rather than a theoretical one. One can imagine a critical discussion between Lumer and van Eemeren going like this: Lumer: Why do you think that the function of argumentation is to achieve unqualified consensus? van Eemeren: Let’s put that to one side, since this is not a shared commitment between us. Let us find something that we can agree on. Irrespective of what argumentation is, we can at least agree that we know about particular cases when they are cases of bad argumentation. Lumer: How do we know that? van Eemeren: Because we know what the fallacies are and can identify when they have been committed. We do not need a theory of the function of argumentation for that. Lumer: OK. I’ll concede that. van Eemeren: Good. Let’s see what else we can agree on. Would you agree that a theory of argumentation should provide a sys- tematic and unified account of why the fallacies are fallacies? That by identifying norms that fallacies fail to aim at we expli- cate what we mean by good and bad argumentation? Lumer: Yes.8 van Eemeren: Well . . . if we consider argumentation as an instance of a critical discussion guided by the Ten Commandments, I conjecture that all instances of fallacies will turn out to be vio- lations of those commandments. It’s true that I haven’t shown that, but only done a bit of armchair analysis that shows that fallacy types can be associated with particular rules and stages of argumentation, but that is enough to make it a serious can- didate for further critical testing.9 waste of time. It is only a question of whether the descriptions produced by several theoretical spotlights can be integrated into a relatively reliable map. This once more emphasizes that evaluation of theories is always relative to the rules (the basic concepts) accepted. Although “several theoretical spotlights” may be useful, I do not think that they are as crucial as Habermas seems to be implying here. Competing theories can be evalu- ated so long as they have something in common, as I will show. 8 Not everybody agrees on this point. For a dissenting voice, see espe- cially Woods (1994). 9 This marks the important point that although the space of possible con- jectures is huge, perhaps infinite, only a few of these are deserving of David Botting 422 Lumer: There are two points I would like to say about that. The first is that the pragma-dialectical explanation of the fallacies is not the only explanation that can be inferred to. I myself of- fered a systematic and unified account of the fallacies in 2000. Even if your conjecture turned out to be corroborated, this would not show that it was the best explanation, or that it was better than mine. The second is that the critical discussion is an ideal construction that is probably never really instantiated. When we identify a fallacy we identify it as a fallacy in the ac- tual dialogue in which it occurs—we do not say that it is a fal- lacy because it would be erroneous in some counterfactual dia- logue that is not occurring and perhaps never does occur, but because it is erroneous in this, the factual, dialogue. Also, if Walton is correct, then in some dialogue-types what we have identified as fallacies are not really fallacies. van Eemeren: As for your first point, relative to our respective theoretical assumptions, each of our theories is correct and are equally good for as long as they both account for the data. Only time and severe critical testing will tell which is better. There is no independent standard of correctness or “better- ness” by which we may compare our explanations. This rela- tivism is unavoidable and therefore unobjectionable. As for your second point, the critical discussion is of course an ideal, but this does not mean that it does not function as a norm, as something that our procedures aim at. Consider the deductive closure of belief. This is a norm that is impossible to satisfy, but if we are shown that two of our beliefs are logically incon- sistent, we are obliged to give one of them up. As for Walton, what he has really shown is that there are cases when fallacies are harmless in the context of the aims of the dialogue that they occur in. It does not follow that they are not fallacies. Lumer: “Harm”, then, is characterized entirely in terms of whether or not it achieves unqualified consensus. But is this what we normally mean by harm? Don’t we consider a rule or an argu- ment harmful if it conduces to false beliefs? Isn’t that at least part of what we mean when we call it fallacious? Nowhere in your theory is consensus systematically linked to truth. The participants may indeed be sincerely aiming at truth, yet never- theless remain arbitrarily far from it because they have agreed to rules that are epistemically non-normative. The pragma- dialectical theory has no means to criticize this—only epi- stemic theories can do so. critical testing, first and foremost those that have already passed some critical tests, but secondly those that clearly identify a set of critical tests for it to pass, that is to say, potential falsifiers. This is how the critical rationalist construes epistemic progress. Question of Truth 423 In this way the Functional and Instrumental claims are defended. It is time to address the Normative Claim. 3. The epistemic non-normativity of the Ten Commandments Here is an example of such an epistemically non-normative rule. Suppose that there are a group of gamblers and all of them believe that the gamblers’ fallacy is a sound rule of inference and on that basis reached a consensus on what the result of the next coin toss should be. Such a consensus would obviously be non-normative, epistemically speaking, and it is only by appealing to epistemic relations that such a consensus can be criticized (Huss 2005, 265- 68). Argumentation aims, say proponents of the epistemic theory, primarily at truth or justified belief, or to put it another way, the epistemic theory is a theory of what argumentation is; this does not mean to say that consensus is not a worthwhile aim, or that rules designed to bring about consensus are not valuable. Feldman (2005, 278-79) seems to say that, understood in this way, the epi- stemic and dialectical theories are compatible, but only in a weak sense that they have distinct sets of norms that are not necessarily in conflict. This weak kind of compatibilism also seems to fall out of Garssen and van Laar’s (2010) paper. I want to show a stronger compatibility thesis that reasonableness is an epistemic norm; only this will suffice to establish the Normative Claim. Reasonableness is stronger than rationality in requiring a con- sidered use of the mind. Toulmin and Perelman reject any concept of reasonableness based on formal validity, but van Eemeren & Grootendorst (1988, 275-78) point out that what Toulmin and Perelman are really rejecting is a strong foundationalist episte- mology, and wrongly suppose formal validity and foundationalism to be inseparable. They are not. We are right to reject infallible basic beliefs—this is the lesson that should be taken from Albert’s Münchhausen Trilemma—but we can still use logic critically. Something is reasonable if it has been subjected to and passed critical tests; this is to use formal logic negatively as a theory of criticism. Lumer (2010) and Siegel and Biro (2008) take issue with this apparently defeatist attitude towards positive justification.10 Siegel 10 Here are a couple of examples: The Münchhausen Trilemma is simply false. It rests on a hid- den and false premise, namely that deduction from true prem- ises is the only form of acceptable justification . . . there are forms of justification that do not rely on already justified prem- David Botting 424 and Biro correctly separate within critical rationalism the issue of whether there are infallible grounds for some of our beliefs— agreeing with the critical rationalists that there are not—from the issue whether there is any possibility of positive justification; ac- cording to them, beliefs can be justified on the basis of evidence irrespective of whether that evidence is infallible. In their response to Siegel & Biro, Garssen & van Laar (2010, 139) seem to make a straw man of Siegel and Biro’s claim and confuse being justified on the basis of evidence, which was all that Siegel and Biro were claiming, with being justified on the basis of other beliefs and be- ing justified tout court. The justificatory relation between evidence and the content of a belief is a purely objective one—more specifi- cally, an epistemic support relation between propositions— irrespective of whether anybody ‘has’ or questions that evidence. Garssen and van Laar confuse this, I suspect, with one belief being based on another belief, where the evidence must not simply exist but be believed by, i.e., be a commitment of, the arguer, and as such is a subjective relation. Garssen and van Laar’s claim that whatever the belief is based on must be starting points that have been agreed upon in a dialectical model (or else it could be ques- tioned) talks past Siegel and Biro, talking about one of these basing relations while Siegel and Biro are talking about the other. Ques- tioning a commitment amounts to questioning whether the evi- dence for the commitment is believed justifiably and in no way questions the relation between the evidence and whatever is based on it. It is true that in the absence of starting points and dialectical rules against questioning them such questions can be asked ad infi- nitum, but justificationism in the advertised sense does not require being justified tout court (and, hence, a regress-stopper such as a non-inferentially justified belief) but only being justified relative to the evidence. ises, in particular observation, and there are ampliative forms of justification, in particular inductive reasoning. (Lumer 2010, 51.) r strongly supports q, which, in turn, strongly supports p. Of course, r could itself be challenged; in that case the discussion, and the chain of justificatory reasons and evidence, might be extended further back to some further consideration s. Absent some such further challenge, however, p is justified, and we are well within our epistemic rights to so take it, on the basis of evidence supplied by q (and, if needed, r). No regress is neces- sary in order that a given standpoint or claim be justified . . . absent a good reason to query q, it is not arbitrary to stop the justificatory chain there; and if there is such reason, but there is no good reason to query r, then it is not arbitrary to stop the justificatory chain at the latter. (Siegel & Biro 2008, 200.) Question of Truth 425 Even so, I think that Lumer, Siegel and Biro are scarcely any more charitable to the argument of van Eemeren and Grootendorst, who I think need not deny that there is positive justification in the sense illustrated. Their point is simply that such justification is relative to the system of rules and premises—which include rules about how to deal with observational evidence, as I will show later—agreed to in the actual discussion, acknowledging that such rules and propositions cannot be infallibly given any determinate probability- or truth-value. This clearly does not mean that I would be irrational or “unjustified” in an everyday sense of the word in believing, for example, the logical consequences of other things I believe.11 Irrespective of their views on the Münchhausen Tri- lemma, which I think is in this sense something of a red herring, all parties can agree that it is epistemically preferable (in a relative, not absolute sense) to believe one thing over another, and that for- mal logic plays a role in this preference. Arguers can only establish their standpoint via propositions and types of inference that that the other participants are explicitly or implicitly committed to, even though these types of inference may be non-normative, such as the gamblers’ fallacy. It is this rule, common to the dialectical approaches, that is the real bone of con- tention behind the dispute over the Münchhausen Trilemma. The rule that leads to this situation derives, as Robinson (1971) points out, from the Aristotelian game of elenchus, which in his view is as irrelevant to the question of ascertaining the truth as would be ob- serving the Marquess of Queensberry’s rules when attacked by a murderer. However, Hintikka (1987) claims that Robinson is wrong to dismiss elenchus and that it is a good model even of sci- entific enquiry; it is only when an anomaly comes to light, e.g., the next coin toss is not what was expected, that our beliefs or our sci- ence has to be revised. Perhaps there is thought to be a critical ra- tionalist rationale for this rule: appealing to propositions that the other participants are not explicitly or implicitly committed to would only be permissible on the proviso that belief in those propositions is infallible, and since there are no such propositions the rule conduces to justified belief and to consensus equally. I find such a thought unconvincing; fallibilism does not require such a rule and can be dealt with adequately by the epistemic approaches that Lumer, Siegel and Biro prefer. Even propositions that have been justified non-inferentially need not, on that account, be con- 11 This seems to be largely a verbal dispute over the meaning of “justifi- cation.” All parties seem agreed on rejecting infallibilism and strong foundationalism: Popper, Perelman, Toulmin, van Eemeren, Grooten- dorst, Lumer, Biro and Siegel. The issue is whether the meaning of “jus- tification” is so closely tied to these that rejecting them amounts to the rejection of justification also. David Botting 426 sidered infallible—there are instances when we rightly consider ourselves unjustified in believing, for example, what we are seeing. The epistemic approaches are not smuggling in strong foundation- alist assumptions when they claim as an advantage of their ap- proach that they reject this rule, justifying this preference by con- sidering the following two kinds of situations. The first is when the arguer has a way for arguing for his standpoint from the shared commitments but this way does not reflect what he himself believes to be the strongest reasons for be- lieving the standpoint. This seems to be the case considered by van Eemeren, Grootendorst, Jackson and Jacobs, when they say never- theless that using the shared commitments is the better way to ar- gue because it is more likely to lead to resolution—“the reasons (motives) people may have for holding a belief are not always the same as the reasons (grounds) they will offer and accept in defense of a claim” (van Eemeren, Grootendorst, Jackson & Jacobs 1993, 12)—but this should not be construed as some kind of error in ar- gumentation, or else people would be performing such errors all the time; to think it an error, they seem to say, involves a confusion between argumentation and reasoning. However, it implies that what is correct as a piece of argumentation may be incorrect as a piece of reasoning and thereby creates a potential problem for the Normative Claim, since epistemic norms are norms of reasoning. But it seems to me that arguing from shared commitments cannot be epistemically inferior either. Suppose that the commitment-store does contain what the arguer believes to be his best reasons—it does not follow that using those reasons is epistemically superior to using any other set of reasons that have the same result. Using rea- sons that are intersubjectively agreed to but epistemically weaker and believed to be epistemically weaker by the arguer in order to argue for a standpoint certainly seems like a non-normative rule for any discussion aiming at being epistemically virtuous. But if rea- sons and standpoints are taken as commitments then this is an illu- sion. As far as commitments go, a commitment either follows from other commitments or it does not, and whatever sets of commit- ments it does follow from it follows from equally; there are not degrees of commitment in the sense that epistemologists often talk about degrees of justification. This, a critical rationalist might say, is not a defect in the commitment-based models but an epistemo- logically good thing, because there are no degrees of justification; in the final analysis, those reasons are better that pass critical tests. At a later date some of those sets of reasons may be falsified, but until then there is no sense to the question of which is better any more than there is sense to the question of which valid logical proof is better. The second case is when the arguer does not have a way for arguing for his standpoint from the shared commitments. Although Question of Truth 427 all his beliefs are justified and his inferences logically valid Lumer (2010, 47-48) points out that he could be forced to retract his standpoint in favour of beliefs that are unjustified. It should be noted that being forced to retract the standpoint in no way means being forced to accept the negation of the standpoint or any other belief the arguer believes to be unjustified, since these are not part of the shared commitments. However, the arguer is in the unhappy position of either not being able to try to persuade his audience at all and withdrawing as a participant, or proceeding regardless, knowing full well that if the Ten Commandments are observed he will be forced to retract the standpoint. It seems to me that the rule is too strict in this situation: it should be at least permissible to ad- vance propositions that have not been previously agreed to when there is no other way of arguing; an audience is more likely to be persuaded by an arguer who can advance reasons, even reasons they are not committed to, than one who cannot, and are likewise less likely to be persuaded by an arguer whose reasons, even if they were true, do not seem to adequately support his standpoint. It seems to me that an arguer who follows this course of action is embracing the rational attitude more than one who is content to remain silent. It could be countered that this assumes that the arguer occupies an epistemically privileged viewpoint; although as a matter of fact the arguer’s beliefs and inferences may be epistemically sound, the arguer is not in a position to know this and can do no more than offer it up for discussion. Without even the possibility of such a viewpoint, it is pointless to pursue epistemic goals and it is better to pursue consensus. In this way, Huss (2005, 265-68) rejects epi- stemic goals. He asks the epistemic theorists what advice they would give to the gamblers. If it is “Use types of inference that are epistemically strong,” then this advice is useless because the gam- blers are already acting in accordance with this advice and so can- not follow it; they use the gamblers’ fallacy because they believe, albeit mistakenly, that this is an epistemically strong inference, or otherwise they would not be in a critical discussion. From their own point of view they cannot distinguish what is epistemically strong from what they believe to be epistemically strong.12 If the 12 An anonymous reviewer remarks that the advice is not useless if it is to use ‘types of inference that are strong according to some external standard of strongness’. It seems to me that this is the same thing as before—the arguer takes his own standard to correspond to an external standard, which is only to repeat that it is epistemically strong. Perhaps the reviewer has in mind the kind of situation where the arguer has it on some authority that the gambler’s fallacy is fallacious and the arguer defers to that authority even though the gambler’s fallacy seems episte- mically strong to him. But then it is for the arguer to decide which set of reasons is the stronger. If he does accept the authority, then it seems to David Botting 428 advice requires showing why the gamblers’ fallacy is a fallacy, then the only way it is possible to do so is to show that it is incon- sistent with the gamblers’ other commitments, which is exactly what the consensual theory says.13 Huss claims that giving useful advice is the right kind of normativity. Although I think that Huss’s advice is good, it does not follow that consensus lacks epistemic normativity. Huss’s analysis is too quick and sets up a false di- lemma: even if epistemically privileged viewpoints are impossible it does not follow that pursuing epistemic goals is pointless or deny that pursuing consensus is to pursue an epistemic goal. A very similar debate occurred between Felix Kaufmann and Ernest Nagel. I will present Kaufmann’s view, Nagel’s critique, and Kaufmann’s response. My aim is to show that a suitably guided consensus is an epistemic norm, so there is not this conflict between different kinds of normativity. I do not wish to claim that the Ten Commandments are more conducive to epistemic goals than other systems of rules. I leave it an open question whether argumentation would be more truth-conducive without the conten- tious rule, but a system of rules containing it does not, because it contains it, fail to be truth-conducive, and this is for the following reason: each critical discussion, I maintain, does not start with a fresh slate, but with the accumulated wisdom of experience, and I furthermore maintain that van Eemeren can quite happily admit that epistemic theorists such as Feldman (2005)—who strangely thinks this is some kind of problem for the pragma-dialectical the- ory—can write books to help people to argue better and not to commit logical or tactical mistakes. Normally, we do not take the gamblers’ fallacy to be sound, because firstly we have instances where experience contradicted what the gamblers’ fallacy predicts, and secondly because we realize that there is no causal connection that could make the outcomes of previous tosses affect the out- comes of future tosses (and so the probability axioms requiring independent trials are applicable, but if the probability calculus led to results that were equally contrary to experience, we would con- clude that so much the worse for the probability calculus). me that he no longer accepts the gambler’s fallacy as epistemically strong although he may not know why it is a fallacy. 13 Feldman (2005) responds to Huss that in taking participants’ beliefs to be believed to be epistemically strong Huss is helping himself to a notion of positive justification that he rejects, and that anyway we often argue without the belief that our beliefs are justified. This seems true, but I think that Huss has simply expressed himself too strongly and can de- mand more modestly only that participants do not agree to or propose what they believe to be unjustified. Obviously, not every premise and type of inference can be supported by further argument; it is only if one is questioned that it must be either supported with reasons or retracted. Question of Truth 429 Let us look at Kaufmann’s view of science. A science has rules of procedure and a corpus of propositions currently accepted by the science. All currently accepted propositions are fallible or ‘controlled’, which is to say that there is a potential scientific situa- tion where the accepted rules demand the elimination of the propo- sition. This—the Principle of Permanent Control—is a property of the system of rules as a whole. Reasons must be given for the way any scientific situation is resolved this is called the Methodological Principle of Sufficient Reason—and the reasons given for the deci- sion on how to resolve the situation must always appeal and only appeal to the rules of procedure, which means that “correctness” vis-à-vis the decision is always relative to what the rules and the corpus are at the time the decision is taken. Any criticism of the scientific decision, that is to say, showing that it is incorrect, must also appeal and only appeal to the rules of procedure, i.e., it must show that they were misapplied—the scientist may think that the rules demand the acceptance of his proposition, but the critic may show that the rules do not demand it. The normative terms “cor- rectness” and “incorrectness” are defined in terms of the rules of procedure and corpus indexed to the time of the decision; it may well turn out later that the proposition in question once rejected is now accepted, perhaps because some member of the corpus with which it previously conflicted has been subsequently eliminated by a control. It is in this way that scientific methodology differs from deductive logic: a deduction is not valid relative to some time or situation. But like deductive logic, it is an entirely internal matter solved by linguistic analysis of the terms involved what decision is correct or incorrect and not the satisfaction of some external condi- tion imposed from some epistemically privileged viewpoint (Kaufmann 1944). Kaufmann’s theory of criticism disagrees with van Eemeren’s on this subtle distinction between using deductive logic itself critically and using some methodological analogues (called by Kaufmann “procedural correlates”) of the logical oper- ations. He also makes clearer than van Eemeren does that the cor- pus is an organic entity that grows, shrinks, and persists from one scientific situation to the next. This, I will show, is vital to estab- lishing its epistemic normativity, and by extension the epistemic normativity of the Ten Commandments. Nagel objected in the following way. If decisions are always relative to rules of procedure and their correctness is always de- fined in terms of those rules of procedure, how is it possible to criticize, e.g., magic? According to the rules of procedure for magic, which includes what is to be taken as evidence, magic is in every way just as correct as modern science. But quite clearly it is not, and the reason it is not is because magic fails against the ex- ternal criterion of describing reality as it is. A purely internal cri- David Botting 430 terion of correctness denies what should be obvious (Nagel 1944b, 62-64). However, magic is inferior to science because, Kaufmann (1944, 71) says, magic and science both share the rule that proposi- tions be held up against experience: “Presupposing rules of the observational test of predictions as basic rules of both magic and science we can judge that science is preferable to magic in terms of the rate of fulfilled predictions pertinent to both fields.” We have, so to speak, crucial experiments to decide between the two sys- tems. Indexed to a time before such a crucial experiment is per- formed it is true that both systems can give different, but equally correct, decisions, but both magic and science are cumulative en- deavours and over time science will nonetheless prove itself epis- temically preferable to magic. Nagel (1944a, 76-77) persists, giving the example of someone consulting two physicians: Let us assume . . . that the facts accepted by both physi- cians are the same, but that the rules of procedure they ac- cept are different. It is small comfort to me that each phy- sician advances a conclusion which is warranted by the “rules” he happens to accept as his standard of criticism . . . at least one of the physicians has adopted rules of pro- cedure which are unsuitable for the type of problem at hand . . . there is a way of deciding—at least in princi- ple—whether each set of rules . . . merits confidence . . . those rules are more adequate as norms, which achieve a greater proportion of successful resolutions of the prob- lems . . . . . . [I]f the problem of criticism involves more than the question whether a given standard is accepted, if criti- cism may legitimately deal with the problem whether specified critical standards are adequate or valid, it . . . be- comes relevant to note that in the history of scientific in- quiry standards do change as a consequence of appraising their adequacy. . . . [T]he enterprise of science is not con- stituted by a self-contained system of mutually determin- ing definitions. Nagel’s rejoinder in the first of the paragraphs above seems odd, since Kaufmann has explicitly said that the set of rules is preferable that fulfils the greater number of predictions or solves the greater number of problems. The issue seems to be over exactly how ex- perience is to be used. For Kaufmann, the status of the rule of ob- servational test is a rule of procedure; even if it were common to every system of rules, it would still have this status—it would still be internal and not external to the system of rules—and every deci- Question of Truth 431 sion would still be relative to the system of rules as a whole. When we prefer one set of rules or another it is not because there is some- thing standing external to the system of rules, e.g., a non- inferentially justified “immediate experience” to which we can appeal, but because there are rules by which what we putatively call our immediate experience can be taken as grounds for proposi- tions and these rules are referred to by rules of a higher-order. His- tory shows that what we consider rules are later rejected, but Kaufmann’s point is that this rejection is always relative to such a system of presupposed higher-order rules. If there is a disagree- ment at the highest-order between two scientists, then this is tanta- mount to a disagreement about how the word “science” is to be defined, or the meaning-in-use of “knowledge”. Beyond such a point philosophical analysis cannot go; what analysis can do is to show the consequences of various alternative definitions between which the community must make a decision.14 Relative to that de- cision, its consequences follow by an a priori necessity. Any scien- tific decision presupposes such a hierarchy of rules simply because they follow in this way from the definition of “scientific”. Pragma-dialectics can defend itself along much the same lines. The mistake that the epistemic theorists make is in supposing that the propositions and types of inference initially agreed to drop out of the sky. On the contrary, each participant believes, and has a reason for believing, the premises and rules he agrees to; there is a continuity between one argumentation and the next. The commit- ments one makes in performing a speech act do not become null and void at the end of the dialogue they occur in, and just because an argumentation is no longer taking place does not leave the speaker free to change his mind about propositions he became committed to without sufficient reason. Even if the speaker does change his mind about some proposition, this does not involve a change in commitments until he performs a speech act that engen- ders a logically incompatible commitment.15 So, while it is true 14 This is a tempting approach to take to “argumentation”. The Functional Claim is not, on this approach, a conceptual truth, and the dispute be- tween different functional claims (between those of, e.g., consensual and epistemic theories) is not a philosophical dispute. The ‘competing’ func- tional claims are mere definitions/stipulations and, as such, are not truth- evaluable. This is not, I think, van Eemeren’s view. What, I think, we would both argue, is that what is “correct” relative to one definition is also largely “correct” relative to the other, and this is more than an acci- dent but due to inherent features of the Ten Commandments; the decision is not between two claims that are mutually incompatible grounding two completely unrelated kinds of normativity. 15 An anonymous reviewer tells me that in the OSSA proceedings of 2005 van Eemeren and Houtlosser introduced the concept of argumenta- tion activity types such as scientific inquiry or political debate and “can David Botting 432 that the gamblers cannot be criticized through an external criterion, repeated failures of their predictions soon brings about an internal inconsistency in their commitments. Assuming that it is part of their rules of procedure to believe the evidence of their own eyes, the gamblers’ fallacy is soon revealed as a fallacy, and its falla- ciousness, we could say, is added to our corpus. Because there is an established corpus of fallacies, we can compare one system or rules against another according to their problem-validity, i.e., their ca- pacity to prevent fallacies. It is quite compatible with this defence that some propositions and types of inference appealed to are not agreed to prior to the argumentation. Indeed, Kaufmann makes the distinction between propositions that we are obliged to incorporate and some that we are merely permitted to incorporate. I also leave it open whether some rules could be improved, e.g., whether it is necessary to have both the argumentation scheme rule and the validity rule. The point of this paper is that there is a way of testing a system of rules, showing that the rules pass these tests is a good way of arguing for their acceptance, and acceptance is, in the long run, a reliable indi- cator of verisimilitude. This can be considered in analogy with reliabilist theories of justification. Perception, for instance, is usually taken as a reliable source of true beliefs, despite the fact that we know perfectly well that they can deceive us. Even though any particular perceptual belief may be false, and we may have no way of discriminating from the inside those that are deceptive from those that are not, this does not mean that perception is not a generally reliable source of knowledge. The Normative Claim that standpoints that have the unqualified consensus of all participants in the dispute will gener- ally be epistemically sound should be construed in the same way. It can be conceded that in any particular case our perceptual beliefs could be non-veridical or that our rules of inference are non- normative, but it does not follow that this is, or even could be, true in every case. Only if a Cartesian demon makes us imagine falla- cies that are not there is there a problem, but this is the same prob- lem that every epistemology has, social or not. In this way pragma-dialectics can answer the question of truth. Taken in isolation, a particular unqualified consensus may indeed be epistemically non-normative, but this does not mean that un- qualified consensus fails as an epistemically normative aim of en- be seen as institutional contexts for argumentation that provide norms that supplement the norms for critical discussion (extra norms generated by the additional goals served by the institution).” I am not sure that extra norms beyond those provided by the original framework of the critical discussion are needed to attain the continuity that I mention, but I agree that such activity types may be useful. Question of Truth 433 quiry. It can be considered as a default that any arbitrary unquali- fied consensus is likely to be true unless positive evidence that suggests otherwise is brought forward. To conclude, the pragma- dialectical concept of reasonableness is an epistemic norm. References Garssen, Bart & van Laar, Jan Albert. (2010). A pragma-dialectical response to objectivist epistemic challenges. Informal Logic (30.2), 122-141. Habermas, Jürgen. (1990). Moral consciousness and communica- tive action. Tr. by Christian Lenhardt and Shierry Weber Nicholsen. Polity Press. Hintikka, Jaakko. (1987). The fallacy of fallacies. Argumentation (1.3), 211-238. Huss, Brian. (2005). Useful argumentation: a critique of the epis- temological approach. Informal Logic (25.3), 261-275. Feldman, Richard. (2005). Useful advice and good arguments. Informal Logic (25.3), 277-287. Kaufmann, Felix. (1944). Concerning Mr. Nagel’s critical com- ments. Philosophy and Phenomenological Research (5.1), 69- 74. Lumer, Christoph. (2010). Pragma-dialectics and the function of argumentation. Argumentation (24.1), 41-69. Lumer, Christoph. (2000). Reductionism in fallacy theory. Argu- mentation (14.4) , 405-423. Nagel, Ernest. (1944a). Rejoinder on Mr. Kaufmann’s reply. Phi- losophy and Phenomenological Research (5.1), 75-79. Nagel, Ernest. (1944b). Symposium on meaning and truth III: Dis- cussion: Truth and knowledge of the truth. Philosophy and Phenomenological Research (5.1), 50-68. Popper, Karl. (1986). The open society and its enemies: vol 2. Lon- don: Routledge. Robinson, Richard. (1971). Begging the question. Analysis (31.4). Siegel, Harvey & Biro, John. 2008. Rationality, reasonableness and critical rationalism: problems with the pragma-dialectical view. Argumentation (22.2), 191-203. van Eemeren, Frans & Grootendorst, Rob. (1988). Rationale for a pragma-dialectical perspective. Argumentation (2.2), 271-291. van Eemeren, Frans, Grootendorst, Rob, Jackson, Sally & Jacobs, Scott. (1993). Reconstructing argumentative discourse. Tus- caloosa, Alabama: University of Alabama Press van Laar, J. Albert. (2003). The dialectics of ambiguity. [Available online.] . ]. http://dissertations.ub.rug.nl/faculties/fil/2003/j.a.van.laar. Last accessed 18th May 2010. David Botting 434 Walton, Douglas. (1989). Dialogue theory for critical thinking. Argumentation (3.2), 169-184. Woods, John. (1994). Is the theoretical unity of the fallacies pos- sible? Informal Logic (16.2), 77-85. Zenker, Frank. (2006). Changes in conduct-rules and Ten Com- mandments: Pragma-Dialectics 1984 vs. 2004. [Available on- line.] www.frankzenker.de/downloads/Zenker_ISSA_2006.pdf. Last accessed 7th August 2010.