ISSA Proceedings 1998 – Wilson On Circular Arguments
In his paper “Circular Arguments” Kent Wilson (1988) argues that any account of the fallacy of begging the question based on epistemic conditions is inadequate and suggests grounds on which a more satisfactory analysis can be provided. He does this by criticizing the epistemic attitude in the fallacy analysis and showing how this has led to an unacceptable analysis of the fallacy of begging the question. I will concentrate on Wilson’s two main points. First of them is Wilson’s argument against the epistemic condition: that we should not overemphasize the assumption that an argument should prove its conclusion. Wilson admits that it is an important function of the argument, but thinks that we should recognize other purposes as well, such as refuting a proposition or undermining confidence in it. Understanding these other purposes would then contribute to the study of the fallacy and point us to a better analysis. I will try to show that Wilson’s ideas on argument’s functions are compatible with the epistemic analysis and that they do not therefore improve our understanding of the fallacy.
The second point I wish to comment on is Wilson’s argument against the division of the fallacy of begging the question into two types: the equivalence and the dependency type. According the equivalence type, a fallacy is committed when the conclusion is equivalent with some premise. In the dependency type some premise is dependent on the conclusion: its acceptability somehow depends on the conclusion’s acceptability. This dependency is often analysed in doxastic or epistemic terms, as for example Sanford and Biro have done. Wilson argues that the dependency view of the fallacy of begging the question is not adequate for several reasons and assumes the equivalence view. I will argue for the dependency view of the fallacy. I do not believe we can subsume it to the equivalence type. Wilson’s critique in fact coincides with Biro’s views on some points. In conclusion, I argue that the epistemic version of dependency can adequately analyse the fallacy of begging the question.
2. On the functions of the argument
Wilson argues that to understand the fallacy of begging the question better, we must widen our view of the functions of the argument from the epistemic emphasis. This epistemic approach has assumed the status of background assumption. It shows in the fallacy theory as an attitude that the primary purpose of an argument is to prove some proposition. He quotes two writers that he thinks have especially emphasized this function, David Sanford and John Biro. Sanford’s formulation of this idea is that an argument should increase the degree of reasonable confidence which one has in the truth of the conclusion (Sanford 1981: 150). According to Biro, an argument should make us know that something, which we did not know to be true, is true, because of something which we do know to be true (Biro 1977: 264). Wilson points out that Biro even seems to go as far as to say that knowledge is the sole aim of the argument when he writes that “Someone that has seen Socrates die would not need an argument for the proposition that Socrates is mortal.”(Biro 1984: fn. 5, 243). Wilson criticizes these views. He thinks that we must recognize other purposes as well:
“For example, arguments are offered on occasion to refute some proposition, or to undermine confidence in it by giving a counter argument against it or by showing that an argument that has been given for it is not valid. Arguments are also given in contexts where one wants to understand better a passage of a text or a discourse – perhaps even a novel or other fiction – to unfold the implications of a plot or of a theory, for example. Somewhat further removed from proving paradigm of argumentation as the marshalling of evidence, arguments are sometimes given in order to explain, to understand, and to predict … arguments may be given in order to answer correctly a puzzle or a problem, where “real knowledge” or even belief need not be involved.” (Wilson 1988: 39.)
I believe that Wilson is quite right in saying that knowledge cannot be the sole aim of the argument. Still, even if knowledge is not the sole aim of the argument, its primary function may still be to show that its conclusion is true. The examples Wilson puts forward clearly qualify as legitimate cases of argumentation. But he should establish more clearly that they are not designed to prove their conclusions. Let us consider for example the case of refuting some proposition. In such a case, the conclusion of the argument is that ’it is not the case that p’. However, this is obviously a new proposition that the argument tries to prove. It might not be proved, but the point of the argument is to achieve it. Refuting a proposition means that we know it is false. Another example was “undermining confidence in a proposition”. The conclusion would perhaps be something like ’We should not believe that p’ or ’it is not certain that p’ or ’it is not probable that p’. The conclusions of these arguments are however propositions that the argument tries to prove. Another example was the case of understanding a novel. A person takes some pieces of information given in a novel, such as the characters, their motives etc. and tries to explain why they behave as they do. These explanations hold only in the context of that novel and are objects of a constant revaluation. The author might have meant that the novel contains information about some real life situations, but it is not necessary. However, arguments given to understand the novel, aim to prove something in that context. The fact that the premises of these arguments do not hold in the actual world does not change the way an argument is supposed to function. This applies as well in solving a puzzle or for different problems where ’real knowledge’ or even belief need not be involved. The conclusions proved are ’true’ only in the given context[i], but this does not undermine the function of the argument: to prove its conclusion.
Wilson’s examples emphasize that there can be various kinds of arguments: arguments about an argument, or arguments about the relative position of a proposition in our belief system, or arguments that start from premises accepted only for the game’s sake. Nevertheless, they do not show us that argument as such changes in any essential way in these situations. So I would conclude that Wilson does not succeed showing that these different uses of argument go against the widely held background assumption that an argument is supposed to prove some proposition, namely its conclusion.
Yet Biro is not saying that arguments have no other functions than proving some proposition[ii]. He is emphasizing knowledge-acquisition for at least two reasons: First, to make clear what is the difference between him and Sanford, and secondly, to give us a clear criterion by which we can evaluate arguments. Biro thinks that if the judgement of an argument is connected to the beliefs of the proponent of the argument the whole process of argument evaluation becomes radically relativistic (Biro 1984: 246). Later, Biro wrote in a joint article with Harvey Siegel:
“Argumentation is a complex phenomenon with disparate aspects and functions: persuasive, communicative, social, logical, etc. Argumentation theory, consequently, is properly interdisciplinary; we theorize about argumentation in rhetorical, philosophical, logical and social scientific terms.” (Biro & Siegel 1992: 85)
Biro emphasizes that the central purpose of the argument is to prove a new proposition, but he does not deny the other functions. The knowledge-acquisition, however, is for him more than a background assumption, it is the central norm by which we should judge arguments. He argues that only such normative account can capture what is essentially wrong in the fallacy of begging the question. Proving some proposition belongs to the hard nucleus of the argumentation. It is its primary function. It has others, some of which were mentioned earlier, but they are secondary aspects: many of them are often ruled out from the domain of argumentation theory as unessential. Incidentally, it seems that the argumentation theorists can widely agree on the idea that there can be no real argumentation without a difference of opinion between the arguers and that there are at least two parties[iii] involved. If there is no difference of opinion, there is nothing to be proved, and consequently no real argumentation.
I am not claiming that studying argumentation in different contexts and with several purposes is not useful. Argumentation varies and we should study this variation to understand fallacies. Quite another question is, whether it should differ in quality as much as it does. The different contexts do not, however, alter the central task of the argument: to prove its conclusion. I do not think that these examples succeed in showing that this emphasis should be given up.
3. Defence of the dependency view of the fallacy
The second point of Wilson’s arguments I wish to comment on is his argument that the dependency notion of the fallacy of begging the question is not satisfactory. I will first give a textbook example of the dependency type[iv]:
A: God exists!
B: How do you know that?
A: Because the Bible says so.
B: How can we trust the Bible?
A: Because it is the word of God.
God’s existence is A’s first conclusion and the issue to be proved. B is in doubt of this and asks for further evidence. A offers the word of the Bible as grounds for believing the existence of God. B has doubts on the trustworthiness of the Bible as evidence, and asks why he should trust it. A offers as evidence the original conclusion, the question at issue. B is offered no other reason than the original question. The reliability of the Bible is dependent on the existence of God. Yet the conclusion, God’s existence, is dependent on the Bible. The conclusion and the premise seem to depend on each other, and since no other evidence is offered for them, the argument can be judged as question begging.
Wilson does not accept the dependency conception: “Conceptions of the fallacy formulated in terms of premises being evidentially dependent on the conclusion are too indeterminate to be of much use” (1988: 43). He gives two versions of the dependency criterion for the fallacy of begging the question. The first version states that an argument begs the question if the conclusion is evidentially relevant to some degree to at least one premise. The second version classifies an argument as question-begging in the case that the conclusion is the only evidence for one or more of the premises. I have several objections to Wilson’s approach.
The first version is clearly too wide definition to be acceptable. But I do not believe that anyone propounding the dependency version thinks that the conclusion’s evidential relevance to premises alone makes some argument fallacious. It can be considered as a necessary condition for the fallacy of begging the question, but not as a sufficient one.
If we were to accept the second version, we would, according to Wilson, be forced to accept several fallacious arguments as non-fallacious:
“For example, consider any argument having as a premise a conjunction of two (distinct) propositions, one of which occurs as the conclusion. Generally there will be evidence supporting the conjunct of the premise that is distinct from that expressed by the conclusion. This general shortcoming affects the more specific versions of the dependency conception that follow.” (Wilson 1988: 43.)
Since this is a general shortcoming, we can consider it before going on to the specific versions of the dependency. We can do that by looking into the following dialogue-game that represents the situation Wilson describes:
n A: p.
B: why p?
n+1 A: p&q.
B: why p&q?
n+2 A: if r, then q. r.
B: q, why p?
Wilson’s claim is that the criterion does not hold because there is evidence for the other conjunct q. But the propositional device of linking propositions to conjuncts holds to the other direction too: the premise p&q can be divided to two different premises p and q. This is B’s tactic in n+2. She accepts q, because she accepts r and that q follows from r, but insists for further evidence for p. This tactic exposes A’s error. He has evidence for q, but not for p. But p&q does not follow from r so B’s tactic pinpoints the fact that A’s only evidence for p is p, and he can be accused of begging the question. This is what the criterion states: the argument begs the question because the conclusion is the only evidence for the premise.
Having dealt with this general shortcoming, we must now turn to the specific objections Wilson raises against the dependency analysis given in epistemic or doxastic terms. He asks us to consider the following formulations:
“(a) in order to know or reasonably believe that one or more of the premises are true, one must know or reasonably believe the conclusion to be true; or
(b) knowledge or reasonable belief that one or more of the premises is true requires inference from the conclusion (i.e. knowledge of the premises is not independent of knowledge of the conclusion)” (Wilson 1988: 43).
Wilson has several objections to these formulations and I will deal with them one by one. I believe that his objections give us grounds to decide how we should define begging the question. First, Wilson remarks that it is often difficult to decide why someone accepts some proposition. He thinks that the consideration of the parties involved in the argumentation is irrelevant. It introduces a relativistic consideration that precludes us from obtaining a satisfactory analysis. Since there are indeterminately large number of different ways of coming to know a proposition, it becomes difficult to see how arguments could be found to beg the question. (Wilson 1988: 44.) I agree with Wilson that it may be very hard to decide why someone accepts a certain proposition. In this connection, it might be useful to separate between why someone believes some proposition and what grounds one gives in the argument. There are obviously indeterminately large number of ways of coming to believe something, and one can have indeterminately large number of reasons to believe something. I may believe that the earth is a geoid, because in my dream, I saw God taking earth to his hands and squeeze it from the poles. Yet if one reasonably believes or knows something, it means one must have some grounds, some warrant for one’s belief. If questioned, one must be able to produce this warrant, and an argument is a tool for producing it. If my belief on earth’s shape is caused by my dream, others come to know this after I produce the argument. So, why someone believes some proposition is often hard to know, but, given the requirements of openness and intra personality of knowledge, what grounds one has for believing, is something that can and should be conveyed through arguments. In an argument one gives the one’s grounds for believing something. They are what we are judging when we judge an argument. We are not judging the personal reasons one may have for believing something. Nevertheless, because we are judging the argument, not some person’s beliefs, we should not attach the explanation of the fallacy to these beliefs. That is why I feel that Wilson’s argument does have some force in it, and it guides us in choosing between the epistemic and the doxastic version.
Wilson’s critique does more damage to Sanford’s doxastic version than to Biro’s epistemic one. This is so because only Sanford ties his explanation to the belief’s of the individual (see for example Sanford 1972: 198). He argues that whether an argument a person proposes begs or does not beg the question depends on the fact why that person believes the premise. The question is begged, if the premise is believed by the person only because he or she believes the conclusion. But since it is very hard to decide why someone believes some proposition, we could hardly ever give judgement on arguments. Sanford would have to stipulate that a person has given all his or her reasons for the conclusion, but this is troublesome for there might always be some arguments we have not heard. Also, I would prefer to treat the argument as a set of propositions that may at least sometimes be evaluated on its own. We could not do this if we tied the evaluation of the argument to the arguer’s beliefs. As I said, we are judging the argument, not personal beliefs.
Biro, on the other hand, does not tie his explanation to what the persons actually believe in. He thinks that a question-begging argument is not epistemically serious (see Biro 1977: 264). This means that it does not, as an epistemically serious argument should, show us that something, which we did not know to be true, is true, by showing that it follows from something we know to be true:
“…they [examples] show clearly that nothing turns on the beliefs of any individual, either in terms of their temporal order, causal connectedness or relative strength. The features on which the epistemic seriousness of an argument – and thus the justice of BQC[v] – depend, are in no way psychological or relative to the arguer (or addressee)” (Biro 1977: 266-267.)
Begging-the-Question-Criticism depends on the essential relativity involved in Biro’s terms. The premise should be more knowable than the conclusion. The term ’more knowable’ means that p is more knowable than q if one can know p without knowing q, and as Sanford (1981: 156) later added, one cannot know q without knowing p. So, Biro’s explanation is dependent on the general epistemic situation, not on individual arguers, and Wilson’s critique does not seem to apply.
Wilson gives the following example to support his thesis that it is hard to decide why someone believes the premises:
A very reliable source (Johnson-Laird, 1983, 180) tells us that ordinary individuals who have never been taught logic do not make use of rules of inference to make valid deductions. From this proposition I infer that no individual who has been taught no logic uses rules of inference to make valid deductions. Does this inference beg the question? I (now) assume that the same proposition is involved, yet my evidence in one instance was the testimony of the source and in the other instance an inference. (Wilson 1988: 44.)
He admits that the assumption that the same proposition is involved is incorrect, but the reader should perceive from this that “.. there is a wide variety of possible evidence matched by paths for coming to know; consequently it will be unusual to find a proposition occurring as a premise the sole evidence for which will be the conclusion. (Wilson ibid).” I am not sure whether this example proves anything. Let us see if they could be presented better. There are two different arguments here, the first one being:
(3) Johnson-Laird claims that ordinary individuals who have never been taught logic do not make use of rules of inference to make valid deductions.
Ordinary individuals who have been taught no logic do not make use of rules of inference to make valid deductions.
The second argument would be:
(4) No individual who has been taught no logic makes use of rules of inference to make valid deductions.
Ordinary individuals who have been taught no logic do not make use of rules of inference to make valid deductions.
Now, I do not think we would claim that (3) is case of begging the question, it is an argumentum ad verecundiam, the proponent is referring to an authority in the field and there is nothing wrong in that. (4), on the contrary, would be considered as question-begging in a normal dialectical situation. Its conclusion is reached by applying the replacement rule of the universal quantifier. Such a replacement is not inherently wrong, but in a dialectical situation it would do no good to the proponent of the conclusion. I do not see how this example discredits the view held by Biro that (4) would not be an epistemically serious argument since the premise is not more knowable than the conclusion, and one could not know the premise and at same time not know the conclusion, stipulating that the context is not a lecture on logic. Still, Wilson’s example does have some force against Sanford’s position. Knowing when an argument begs the question can be difficult, when its assessment is tied to the beliefs of an individual proposing the argument. Actually, this difference is what Biro and Sanford have discussed in several articles [see Biro 1971, 1977, 1992 and Sanford 1972, 1981]. This would seem to show that Wilson’s argument do not apply to Biro’s position.
Before concluding I wish to discuss few minor objections Wilson raises against Biro. He argues that Biro’s explanation of begging the question runs into difficulties when a person’s commitments form an inconsistent set. In that case, those commitments cannot be more knowable than the conclusion since an inconsistent set of propositions cannot be known. This is an interesting problem. What sort of criteria should we impose on a person’s commitments, for example in dialogue-games, or should we limit the classical logic somehow? A player may prove any proposition from his opponent’s inconsistency, which does not seem represent actual discussions very well. Yet in relation to Biro’s analysis, it suffices to remember that his explanation is tied to true propositions. We can claim, with some credibility, that true propositions do not form inconsistent sets of propositions, so Biro needs not concern himself with this situation. If on the other hand one’s explanation is tied to the beliefs of the arguer, as Sanford’s is, this objection would have some force, since beliefs do form inconsistent sets.
Another problem for Biro is, according to Wilson (1988: 45), the case of strategic planning. This case is very similar to the case I discussed earlier in relation to arguments given to understand, for example, a novel. The strategic planners are not using true propositions as premises in their arguments. They are toying with hypothetical statements. If there is no real knowledge, i.e. no true propositions involved, does not that make Biro’s account of argument in terms of knowability inadequate? I believe that this problem can be solved by looking into the status of the propositions acquired from strategic planning. These propositions are not considered as true statements as such, but only in relation to the hypothetical statements about the enemy’s moves. They are hypothetical statements or recommendations such as “If the enemy attacks with this type of force and from these directions, the following measures would most likely be the most efficient…”. In this context, the premise that concerns the enemy’s moves is surely more knowable than the conclusion about the measures that should be taken since the measures taken can be unsuccessful. Strategic planning is in this sense analogical with the case of trying to understand a novel. The information that the author gives about the characters and their motives is surely more knowable than our speculation from those premises.
The important thing to notice is that Biro’s criterion speaks of relative knowability, not absolute knowability. His position is not affected by the fact that the premises are not true.
I believe that I have presented enough reasons for us to decide that Wilson’s critique against the dependency notion and the epistemic version of it is not adequate. At the end of his article, Wilson brings forward an example to note the problems of assuming only the equivalence analysis:
(5) Nixon realized that he was dishonest; hence he was dishonest.(Wilson 1988: 51)
This example can be analysed with the epistemic version. In (5) one could know the conclusion without knowing the premise, but one could not know the premise without knowing the conclusion. Therefore, the argument is not epistemically serious, but begs the question. The same can be stated about the example (1), the discussion on God’s existence. These arguments do not seem to fit to the equivalence type of analysis.
Wilson’s critique of the epistemic and doxastic version’s of the fallacy of begging the question had force against Sanford’s view, but was compatible with Biro’s critique of the doxastic version. I believe that Wilson did not succeed in showing that it applied to the epistemic version as well. If we can accept the dependency view, as I think we can, this seems to work towards the acceptability of the epistemic version of the fallacy of begging the question. Biro’s version is not without problems though. For example, the term ’relative knowability’ needs to be explicated further.
i. Problems relating to specific contexts have been examined by for example David Lewis in his article ‘Truth in fiction’ (1978).
ii. Nor is Sanford: his version speaks of the primary purpose of the argument (see for example 1972: 198).
iii. In the case of a solitaire arguer, the other party would be nature, from which the arguer elicits answers by tests.
iv. Even though this is a textbook example, I cannot claim that it is good example of begging the question in the sense that it would be an example of someone’s actual argument. It could also be improved by adding some other independent premises, which would make it plausible. But I do claim that in this form, it does beg the question against B.
Biro, J. (1977). Rescuing ’Begging the Question’. Metaphilosophy 8, 257-271.
Biro, J. (1984). Knowability, Believability and Begging the Question: a Reply to Sanford. Metaphilosophy 15, 239-247.
Biro, J. & Siegel, H. (1992). Normativity, Argumentation and an Epistemic Theory of Fallacies. In: F.H. van Eemeren, R. Grootendorst, J.A. Blair, C.A. Willard, (Eds.), Argumentation Illuminated.. (pp. 85-103, ch. 6), Amsterdam: SICSAT.
Cohen, M. & Nagel, E. (1934). An Introduction to Logic and Scientific Method. New York: Harcourt, Brace & World.
Johnson-Laird, P. N. (1983).Thinking as a Skill. In: J. St. B. T. Evans (Ed), Thinking and Reasoning: Psychological Approaches (pp. 164-196). Routledge and Kegan Paul.
Lewis, D. (1978). Truth in Fiction. American Philosophical Quartely 15, 37-46.
Sanford, D. (1972). Begging the Question. Analysis 12, 197-199.
Sanford, D. (1981). Superfluous information, Epistemic Conditions of Inference, and Begging the Question. Metaphilosophy 12, 145-168.
Walton, D. (1984). Logical Dialogue-Games and Fallacies. Lanham: University Press of America.
Wilson, K. (1988). Circular Arguments. Metaphilosophy 19, 38-52.
Woods, J. & Walton, D. (1975). Petitio Principii. Synthese 31, 107-127.