A common method of criticising an argument is to attack one or more of its premises. However, if we know that a valid argument has one or more false premises, this does not allow us to say anything at all about the truth or falsity of the conclusion. Is the procedure of criticising the premises of an argument, therefore, pointless? I show that an attempt to provide a rationale for this strategy, based on a traditional understanding of rationality, fails to adequately explain its value, but I then argue that there is a place for this tactic in argumentative discourse. However, in order to appreciate its value we need to consider how arguments are used to get people to believe things in the context of a dialogue and not just the formal properties of those arguments. (It should be noted that, in this paper, I assume that the strategy of attacking one or more of the premises of an argument is only used to criticise valid arguments.)
2. The strategy in practice
Many authors either advocate or use the strategy of attacking an argument by criticising one or more of its premises. I will mention several of these in order to illustrate how widespread the acceptance of this tactic is. Gilbert (1996) gives a lot of useful advice about how to improve your argumentative skills. Rather than talking about the conclusion and the premises of an argument, he talks about the claim that someone puts forward and the reasons that that person has for asserting that claim. In a situation where the person you are arguing with puts forward a claim that you disagree with, Gilbert (1996, p. 32) gives the following advice, ‘Always attack the reasons for a claim, not the claim itself’.
Fogelin and Sinnott-Armstrong (1997, p. 366) are not as categorical as Gilbert. They acknowledge that there are several ways in which an argument can be criticised. However, they write:
The second main way to attack an argument is to challenge one of its premises. We can argue that there is no good reason to accept a particular premise as true, asking, for example, ‘How do you know that?’ If there is no way to justify a premise, then the argument usually fails to justify its conclusion. More strongly, we can argue that the premise is actually false. In this second case, we refute an argument by refuting one of its premises.
Like Fogelin and Sinnott-Armstrong, Schopenhauer, in an essay only published after his death, states that there are several ways in which an opponent’s thesis can be refuted. One of these he calls the direct method. He writes that the ‘direct course attacks the reasons for the thesis’ and that, if successful, this ‘direct refutation shows that the thesis is not true’ (Schopenhauer 2005, p. 174).
Some writers do not explicitly state this way of attacking an argument, but just use it. Reinard is one of these. He considers someone who ‘is opposed to the current welfare system because he or she believes benefits go to many who are capable of working’ (Reinard 1991, p. 308). The argument used here has an unexpressed premise and the conclusion is not fully expressed. We can take the conclusion to be the statement that the current welfare system is unfair and the unexpressed premise to be the proposition that, in a fair welfare system, benefits do not go to those who are capable of working. (Whether or not this is the best way to reconstruct this argument is not the issue here.) Reinard says that one way of criticising this argument is to ‘cite evidence that fewer than five percent of welfare recipients could work if jobs were available in their vicinities’. He is attacking the claim that the current welfare system is unfair by showing that one of the reasons given for this claim is false.
Shaw also uses the strategy of criticising an argument by attacking one of its premises without explicitly stating it. One of the arguments he analyses is, ‘Most of those who say they believe in capital punishment do not really believe in it. For if they were called upon to carry out an execution themselves they would not be able to bring themselves to do it ‘ (Shaw 1997, p. 61). He first notes that there is an unexpressed premise in this argument and he says that this is the statement, [Anyone] who sincerely believed capital punishment was right would be prepared to carry out the punishment himself’. Whether or not this is the most likely unexpressed premise is not the issue here. What is relevant is that Shaw criticises the argument he has reconstructed by showing that the unexpressed premise is suspect. He does this as follows, ‘A person might surely fail to carry out an execution because he regarded executions as unpleasant rather than wrong. It is only too possible to shirk unpleasant tasks one thinks ought to be done, leaving it to others to perform them’.
3. An attempted rationale for the strategy
Gilbert is one of the few writers who attempts to give a rationale for the strategy of attacking an argument by criticising one or more of its premises. He is a rationalist and he puts forward the following principle of rationality, ‘Always assume that people have reasons for their beliefs’ (Gilbert 1996, p. 35). For him, in the context of argumentation, ‘being rational means providing reasons for beliefs’ (Gilbert 1996, p. 34).
The conception of rationality that Gilbert is here assuming is very old. It has been called uncritical or comprehensive rationalism (Popper 1966, p. 230) and also panrationalism (Bartley 1984, p. 85). Bartley (1984, p. 93) says that comprehensive rationalism consists of two requirements:
These are (1) that any position which can be justified or established by rational argument is to be accepted; and (2) that only positions which can be justified or established by rational argument are to be accepted.
Thus, in comprehensive rationalism the only way in which a belief can be justified is by showing that it follows logically from other justified statements. There is no way of justifying a belief other than by showing it to be the conclusion of a valid argument whose premises have themselves been established.
Comprehensive rationalism can easily be shown to be untenable. Consider some position, claim or standpoint. For the comprehensive rationalist to accept this claim it has to be justified or established by rational argument. That is to say, it has to follow logically from other justified positions. The only way in which these positions can themselves be established is by means of rational argument. That means that they have to follow logically from other established positions which themselves have been justified by means of rational argument and so on. Such a view of rationality entails that no position whatsoever can ever be justified, because the attempt to justify any belief would lead to an infinite regress.
Gilbert (1996, p. 34) is well aware that his principle of rationality cannot be universally valid and his discussion brings out the untenability of comprehensive rationalism in another way:
Someone who believes something without reason is being irrational. In terms of argument, being rational means providing reasons for beliefs. In the end all of us may be irrational, since sooner or later we reach a point of ultimate beliefs (for which it is impossible to provide reasons).
In order to prevent an infinite regress of reasons, Gilbert accepts that people must have some beliefs which they have acquired directly in some way and not as a result of argumentation. Because of his principle of rationality, this acquisition must have been irrational. In fact, the situation is even worse than this because Gilbert’s acceptance of the existence of ultimate beliefs shows that his comprehensive rationalism is inconsistent. He begins by assuming that a rational person has reasons for all his beliefs and ends by acknowledging that that person must have beliefs for which reasons cannot be given. As Gilbert’s comprehensive rationalism is inconsistent it cannot be used to provide a rationale of the strategy of criticising an argument by attacking one or more of its premises.
A common response to the realisation that comprehensive rationalism is untenable is to adopt a different account of rationality which has been called limited rationalism (Bartley 1990, p. 232). A limited rationalist accepts that some beliefs cannot be justified by means of rational argument, but he then insists that every other rational belief must be ultimately justified in terms of some collection of justification-terminating beliefs. That is to say, the limited rationalist accepts that all the leaf nodes of the tree of justificatory reasons for every one of his beliefs consist of statements belonging to a privileged class of justification-terminating beliefs which are established in some non-argumentative manner. Limited rationalists disagree about the nature of the justification-terminating beliefs and also about the ways in which those beliefs are established extra-logically. I will mention three varieties of limited rationalism for illustrative purposes, but there are many more.
A very popular version of limited rationalism is empiricism in which the collection of justification-terminating beliefs consists of observation statements which are established on the basis of sense experience.
Williams (1999, p. 180) presents a ‘picture of human knowledge as an evolving social phenomenon’. The collection of justification-terminating beliefs in his account consists of those beliefs that are generally accepted by some community. He writes, ‘at any given time we must have some stock of beliefs which are not thought to be open to challenge’ (Williams 1999, p. 83).
Wittgenstein can be interpreted as saying that our justification-terminating beliefs are established by our form of life. In On Certainty he is much troubled by the practice of asking for and giving reasons. He writes that some people behave ‘[as] if giving grounds did not come to an end sometime. But the end is not an ungrounded presupposition: it is an ungrounded way of acting’ (Wittgenstein 1969, p. 17e). Thus, the class of justification-terminating beliefs are grounded, not by other beliefs, but by a particular form of life.
The main problem with all varieties of limited rationalism is that no rational account can be given for the choice of justification-terminating beliefs. That choice has to be irrational. This is because, for the limited rationalist, being rational means providing reasons for all beliefs except justification-terminating beliefs. Reasons, therefore, cannot be given for those beliefs. Limited rationalism can, therefore, be seen to be relativistic and fideistic: the collection of justification-terminating beliefs is chosen by an irrational act of will and, being irrational, any such choice is as good as any other.
To overcome the difficulties of comprehensive and limited rationalism Bartley has proposed a new version of rationalism which he calls pancritical rationalism. Rather than thinking that ‘being rational means providing reasons for beliefs’ (Gilbert 1996, p. 34) or that being rational means providing reasons for all beliefs except a privileged class of justification-terminating beliefs, Bartley holds that being rational means that you are willing to allow any of your beliefs to be criticised. He stresses that there are many different ways in which a claim can be criticised (Bartley 1984, pp. 126-136). However, there is at least one method of criticism that he outlaws and that is the strategy of criticising an argument by attacking one or more of its premises. Bartley outlaws this strategy indirectly.
Before we can attack the reasons that someone has for a claim we have to ascertain what those reasons are and that is achieved by asking a question like ‘How do you know?’ Bartley (1984, p. 113) sees no place for this question. He writes, ‘The traditional demand for justification – the ‘How do you know?’ question – would not legitimately arise [in pancritical rationalism]’. He explains why as follows:
Any view may be challenged by questions such as ‘How do you know?’, ‘Give me a reason’, or ‘Prove it’. When such challenges are accepted by citing further reasons that justify those views under challenge, these views may be questioned in turn. And so on forever. Yet if the burden of justification is perpetually shifted to a higher-order reason or authority, the contention originally questioned is never effectively defended. One may as well never have begun the defence: an infinite regress is created (Bartley 1990, p. 231).
I have a great deal of sympathy for pancritical rationalism, but I think that Bartley is wrong to forbid the use of the question ‘How do you know?’ and thereby outlaw the strategy of attacking premises as a legitimate method of criticism. His argument is flawed because asking the question ‘How do you know?’ does not necessarily lead to the creation of an infinite regress. Above I discussed the claim made by Gilbert (1996, p. 35) that people always have reasons for their beliefs. This does lead to an infinite regress, because those reasons are themselves beliefs. Bartley’s claim is significantly different because he is talking about one person asking another for reasons. No one can ask another an infinite sequence of questions. Furthermore, having asked the question ‘How do you know?’ once, the questioner does not have to ask it again. He could change to a different kind of challenge. He could say, for example, something along the following lines, ‘But that contradicts what you said earlier’ or ‘Your claim has these unacceptable consequences’. In fact, he does not have to challenge everything his opponent says. The opponent may actually say something that the questioner agrees with.
Not only is Bartley’s stated reason for outlawing the strategy of asking for and criticising reasons flawed, he also implicitly endorses the use of this strategy! For example, in discussing the critical method that he employs in his book The Retreat to Commitment he says, ‘I do not pretend to give an exhaustive critique of the thinkers or the systems of thought which I discuss and criticize’ (Bartley 1984, p. xxvi). He continues, ‘I have tried to aim my criticisms at only the most basic assumptions of these systems of thought, their feet as it were, without which they cannot stand.’ Thus, concerning each system of thought that he discusses, Bartley’s strategy is to attack the premises (which he calls ‘feet’ or ‘basic assumptions’) from which the statements that make up that system follow. The result of this is to show that all those statements are false. Those statements are all conclusions of different arguments having the same premises. Thus, Bartley uses the strategy of attacking premises to criticise a set of conclusions and not just a single conclusion, but he uses it nonetheless.
As already mentioned, I have a lot of sympathy for Bartley’s pancritical rationalism, but I also think that the strategy of attacking the premises of an argument is an important argumentative tactic. Thus, the task to which I now turn is that of legitimising that strategy in a way that can be accommodated within pancritical rationalism.
4. Legitimising the strategy
In order to appreciate the strategy of asking for and criticising reasons we have to stop thinking of arguments as abstract objects to be studied without regard to their use. The legitimacy of the practice of asking for and challenging reasons can only be explained by looking at arguments that are being used to try and persuade someone of the truth of some claim. Consider, for example, the following argument, which I have already mentioned and which is based on (Shaw 1997, p. 61), ‘If a supporter of capital punishment was called upon to carry out an execution, he would be unable to do so. Anyone who sincerely supported capital punishment would be prepared to carry out the punishment himself. Therefore, supporters of capital punishment are insincere.’ When such an argument is presented without any consideration being given to its use, then showing that one or more of its premises is false tells us nothing at all about the truth-value of the conclusion. There is a role for the study of arguments as abstract objects without consideration being given to their use. However, it is a mistake to think that studying arguments in this way exhausts everything interesting that can be said about them. No doubt, arguments can be put to many different uses. It is when they are used to influence people’s beliefs in the context of a dialogue that the strategy of attacking reasons makes sense.
Many rules would appear to govern argumentative discourse, but two in particular are especially relevant to understanding the strategy of criticising an argument by attacking one or more of its premises. The first applies to much more than just argumentative discourse and was called, by the eighteenth-century philosopher Thomas Reid, the principle of credulity. This ‘is a disposition to confide in the veracity of others, and to believe what they tell us’ (Reid 1997, p. 194).
He adds that this principle ‘is unlimited in children, until they meet with instances of deceit and falsehood: and it retains a very considerable degree of strength through life.’ Because of this principle, we do not need a reason in order to believe what we are told, but we do need a reason to reject what we hear.
The second rule that governs argumentative discourse and is relevant to understanding the strategy of attacking premises is a restricted version of, what Harman (1986, p. 12) has called, the logical closure principle. Harman’s principle states that a person’s beliefs should be closed under logical implication: if a proposition follows logically from your beliefs, then you should believe that as well. The logical closure principle is controversial. Since every statement has an infinite number of logical consequences (to be precise, it has countably many logical consequences), the logical closure principle would mean that anyone with at least one belief would have infinitely many beliefs. For my purposes I only need a restricted version of this principle which states that if you believe that a proposition follows logically from your beliefs, then you should believe that as well. In the original principle you do not need to know or believe that the proposition in question follows from your beliefs, whereas in the restricted version you have to believe that the inference from your beliefs to the proposition in question is valid.
I will show the usefulness of the strategy of attacking reasons by considering a concrete example of its employment. In this I make use of the argument from Shaw (1997, p. 61) that has already been mentioned. Consider a conversation between two people. One of them, the protagonist, asserts, ‘Supporters of capital punishment are insincere.’ (Following van Eemeren and Grootendorst (1992, p. 35), I call the person who makes the assertion the protagonist and the hearer the antagonist.) We may assume that the antagonist does not believe this assertion at the beginning of the dialogue. Because of Reid’s principle of credulity, people generally believe what they are told, so, if the antagonist does not challenge this assertion in some way, he should accept it. If he does not criticise the protagonist’s claim, then the protagonist would be right in thinking that he had convinced the antagonist that supporters of capital punishment are insincere. There are many ways in which the antagonist could reply to the protagonist’s assertion which would indicate that he did not accept it. One of these is to ask ‘How do you know that?’ or ‘Why do you believe that?’ Such a question has several purposes in argumentative discourse. One of them is to indicate to the protagonist that the antagonist has not believed the assertion and it can, thus, be interpreted as a challenge, ‘Convince me of the truth of your statement.’ It also, as Norman (1997, p. 487) puts it, ‘temporarily suspends the claimant’s right to use the claim questioned. The implicit understanding is that, if entitlement to the claim is to be redeemed, adequate grounds must be provided.’ The protagonist’s ultimate aim might be to convince the antagonist that capital punishment should be abolished and his argument for this position may depend on the claim that supporters of capital punishment are insincere. Thus, he needs to convince the antagonist of this claim. The question ‘How do you know?’ prevents him from using this claim as a premise until he has convinced the antagonist to believe it.
In reply to the antagonist’s question ‘How do you know?’ the protagonist might reply, ‘Anyone who sincerely supported capital punishment would be prepared to carry out the punishment himself. Furthermore, if a supporter of capital punishment was called upon to carry out an execution, he would be unable to do so.’ If the antagonist were to accept these reasons, then, because of the restricted logical closure principle, he would be committed to accepting the claim. If he does not challenge the reasons given, by Reid’s principle of credulity, he would be assumed to have believed them. And believing the reasons, he would be forced to believe the claim. (He can challenge the reasons given for a claim in many different ways. He could ask again ‘How do you know?’ However, he could adopt some other strategy, like showing that the reasons had consequences that the protagonist found unacceptable.)
Reid’s principle of credulity helps to explain why the strategy of attacking premises has a place in argumentative discourse. In an argument one person tries to convince another of a claim, that is to say, he tries to get the other person to believe that claim. One way of getting someone to believe something that they are reluctant to believe is to get them to accept reasons from which that claim follows. If the antagonist does not attack those reasons, the protagonist can assume he has accepted them and thus, by the restricted logical closure principle, that he has been forced to accept the controversial claim. That is why the antagonist must challenge one or more of the reasons given for a claim that he does not want to accept.
The strategy of attacking premises comes into its own when arguments are used to influence beliefs in the context of a dialogue. In argumentative discourse, if the antagonist does not challenge an assertion made by the protagonist, it can be legitimately assumed that he accepted that assertion. Challenging an assertion is a way of signalling to the protagonist that the antagonist does not believe it. And not believing one of the reasons given for a claim shows that he does not believe the claim. In argumentative discourse, if I do not believe one or more of the reasons given for a claim that I have challenged, then it can be correctly assumed that I do not believe the claim. Contrast this with the formal properties of the same argument: if one or more of its premises are shown to be false, then the conclusion can be either true or false. The falsity of one or more of the premises of a valid argument is not transmitted to its conclusion, but, in argumentative discourse, my not believing one of the reasons given for a claim I have challenged is transmitted to that claim. Thus, the strategy of attacking premises makes sense in argumentative discourse.
My legitimisation of the strategy of attacking premises in the context of an argumentative dialogue depends on the acceptance of only two rules, namely Reid’s principle of credulity and the restricted logical closure principle. These two rules are not inconsistent with the tenets of pancritical rationalism. Thus, they allow us to legitimise the strategy of criticising premises while at the same time embracing pancritical rationalism. However, those two rules are not specific to pancritical rationalism. Therefore, my legitimisation of the strategy of attacking premises can be accepted by anyone who accepts those two rules no matter how they conceive of rationality.
I have argued that there is a place for the practice of asking for and criticising reasons in argumentative discourse. Most people accept this, but a common account of why this strategy is used is faulty. This says that you show that the conclusion of an argument is false by showing that one of the premises of that argument is false (Schopenhauer 2005, p. 174). This is straightforwardly wrong as a valid argument can have a true conclusion even if all of its premises are false. To understand the strategy of attacking reasons we have to look at how it is used in argumentative discourse involving a protagonist and an antagonist. I have indicated several of the consequences of using this strategy both on the protagonist and the antagonist. By asking for the reasons given for a claim made by the protagonist, the antagonist signals that he does not believe that claim. Furthermore, it prevents the protagonist from using that claim as a reason for a further claim until he can provide acceptable reasons for it. By criticising the reasons that the protagonist gives for his claim the antagonist again signals that he does not believe them and, if the protagonist cannot rebut those criticisms, he cannot make further use of his original claim in his argument with the antagonist. If he cannot, he may excuse himself from the conversation or try a different approach to getting his message across. The truth or falsity of the reasons given for a claim and how these may affect the truth-value of the claim is not the issue when the strategy of attacking reasons is used in argumentative discourse. What matters in an argument that takes place in the context of a dialogue is whether or not the antagonist accepts what the protagonist claims and whether the protagonist can make further use of a claim he has made. It should be noted that I am not suggesting that truth and falsity are irrelevant to argumentative discourse. They are crucially important as all of us want to have true beliefs rather than false ones. What I am saying is that you cannot understand the strategy of attacking premises just in terms of the truth-values of the component statements of the argument being criticised as falsity is not generally transmitted from the premises to the conclusion of a valid argument. To understand this strategy you need to take into account what the antagonist, as the user of the strategy, believes. This is because his not believing one of the premises of an argument, used by the protagonist, is transmitted to the conclusion of that argument in the sense that he is not compelled to believe that conclusion.
Bartley, III, W. W. (1984). The Retreat to Commitment. Second edition. La Salle: Open Court.
Bartley, III, W. W. (1990). Unfathomed Knowledge, Unmeasured Wealth: On Universities and the Wealth of Nations. La Salle: Open Court.
Eemeren, F. H. van and R. Grootendorst (1992). Argumentation, Communication, and Fallacies: A Pragma-dialectical Perspective. Hove and London: Lawrence Erlbaum Associates.
Fogelin, R. J. and W. Sinnott-Armstrong (1997). Understanding Arguments: An Introduction to Informal Logic. Fifth edition. Fort Worth: Harcourt Brace College Publishers.
Gilbert, M. A. (1996). How to Win an Argument. Second edition. New York: Wiley.
Harman, G. (1986). Change in View: Principles of Reasoning. London: MIT Press.
Norman, A. (1997). Regress and the doctrine of epistemic original sin. The Philosophical Quarterly 47, 477-494.
Popper, K. R. (1966). The Open Society and its Enemies: The High Tide of Prophecy: Hegel, Marx, and the Aftermath, volume 2. Fifth edition. London: Routledge & Kegan Paul.
Reid, T. (1997). An Inquiry into the Human Mind on the Principles of Common Sense: A Critical Edition. Edinburgh: Edinburgh University Press. This book was first published in 1764.
Reinard, J. C. (1991). Foundations of Argument: Effective Communication for Critical Thinking. Dubuque: Wm. C. Brown.
Schopenhauer, A. (2005). The Art of Always Being Right. London: Gibson Books.
Shaw, P. (1997). Logic and Its Limits. Second edition. Oxford: Oxford University Press.
Williams, M. (1999). Groundless Belief: An Essay on the Possibility of Epistemology. Second edition, Princeton: Princeton University Press.
Wittgenstein, L. (1969). On Certainty. Oxford: Basil Blackwell.