ISSA Proceedings 2006 – On Justified Belief In The Link Of An Argument
A natural requirement for justifiably believing the conclusion of an argument is that the arguer or an inferer must justifiably believe the premises. This paper studies the question whether we should also require that an arguer or an inferer must justifiably believe the link of an argument in order to justifiably believe the conclusion. I will first draw some theoretical divisions and then present an intriguing argument by Andrew D. Cling (2003) that appears to show that this requirement is necessary for justifiably believing the conclusion. I will then present four different arguments that challenge Cling’s argument. Finally, I discuss the implications of rejecting the requirement.
An argument is an ordered pair of a set of propositions, the premises, and a proposition, the conclusion. We acquire indirect beliefs, i.e. beliefs based on other beliefs through the use of arguments and inferences, and the propositions of the abstract argument structure express the content of those beliefs. The uses of arguments may have varying purposes but here I will only discuss the core case of using the propositional structure of an argument with the intention of becoming justified in believing the conclusion. Obviously, not every use of an argument results in justified beliefs. We should also note the difference between being justified in believing that C and justifiably believing that C. The former is evaluative in the sense that it merely means that a person S has justification for believing that C, has a good reason to believe C, but it does not imply that S actually believes that C. If S justifiably believes that C, S has good reason to believe C and actually has the belief that C. It is thus evaluative and factual. Further, S can have justification for C, and believe that C, without justifiably believing that C. This would be the case if S were to base her or his belief not on the justified belief that P, but on some other belief R that is not justified. In this case, P would not be the reason for which S believes that C.[i] In what follows, I will discuss cases where the arguer or reasoner bases his or her belief that C on the premises. But even if S is justified in believing the premise P and bases her or his belief that C on P, S is not necessarily be justified in believing that C, if S’s belief that C was generated from P in an intellectually dubious manner, for example, by a fallacious argument).
3. The argument for justified belief in the link of an argument
It seems reasonable to accept the following principle:
(JP) Necessarily an argument P therefore C is justification-affording for S only if S justifiably believes P. (Cling 2003, p. 286)[ii]
If the arguer is not justified in her or his belief in the premises, the argument cannot make the belief in the conclusion justified. But Cling argues that we also need the following requirement:
(JCC) Necessarily, an argument P therefore C is justification-affording for S only if S justifiably believes if P, then C. (Cling 2003, p. 287)
Cling argues that if one does not include the (JCC), we end up drawing arbitrary distinctions between different arguers. Namely, there can be an argument
(1) P therefore C.
that can be justification-affording for persons who do not justifiably believe if P then C. Given the principle (JP), the corresponding argument
(2) (P and if P then C) therefore C.
– the argument that results from taking (1)’s corresponding conditional as a premise – will not be justification-affording for the same persons, if only because they do not justifiably believe a crucial premise of (2). Among those for whom (1) is but (2) is not justification-affording will be some persons whose predicaments are such that the only non-epistemic differences between (1) and (2) are those that result from the fact that the corresponding conditional of (1) is a premise of (2). Since such differences are epistemically irrelevant, the epistemic distinction that is drawn between (1) and (2) is arbitrary. (Cling 2003, pp. 299-300)
We need to look at this argument a bit more closely. The idea is that there are persons who can use the argument (1) to reach a justified belief C even though they are not justified in believing the corresponding conditional of the argument (1) but for same persons (2) is not justification-affording, given the principle (JP), that is, because they do not justifiably believe its premises. The only difference is that (2) has (1)’s corresponding conditional as premise and that (1) and (2) have different corresponding conditionals. In particular, some S for whom this applies, can be such that neither S’s internal states nor any external features of S’s world provide a basis for distinguishing between (1) and (2) beyond the role of if P then C in the two arguments. So, S’s predicament can be such that S’s beliefs and the logical and epistemic relations among S’s beliefs and other internal states provide no basis for distinguishing (1) from (2). Furthermore, S can be such that the conditional reliability of the processes of inference that S would bring to bear on (2) is at least as high as the process S would bring to bear on (1). According to Cling, the point is general: specify internal or external conditions as you will, the same problem can always reproduced. Since both (1) and (2) depend upon S’s commitment to if P, then C, it is arbitrary to claim that (1) is justification-affording but (2) is not.
There is also a further argument. There could be a situation where S does not justifiably believe corresponding conditional of (1) if P, then C, but does justifiably believe if (P and if P, then C), then C, the corresponding conditional of (2). Theories of justification that do not include (JCC) imply that although S could come to be justified in believing C by reasoning to C through P by means of a conditional statement that S does not justifiably believe – if P, then C – S could not come to be justified in believing C by making more of S’s commitments explicit as premises and reasoning to C through P and if P, then C, by means of justified true proposition that S does justifiably believe – if (P and if P, then C), then C. Such theories imply that even though (2) makes more of S’s commitments explicit as premises and even though S justifiably believes (2)’s corresponding conditional, S cannot acquire justified belief in C by means of (2) but can acquire justified belief in C by means of (1). This is clearly arbitrary. (Cling 2003, pp. 300-301)[iii]
4. Arguments against justified belief in the link of an argument
There have been some influential epistemologists who have held that the requirement (JCC) is too strict. For example, according to William P. Alston (1989, pp. 164-165) (JCC) is not a reasonable requirement, mainly for two reasons. First, requiring (JCC) would make the indirect beliefs, that is, beliefs based on other beliefs, of animals and preverbal children unjustified. Arguably, both preverbal children and animals infer, but they do not possess the concepts of deduction, induction or argument scheme. Hence, they cannot justifiably believe anything what the principle (JCC) requires. In addition, consider someone reading the local newspaper and unthinkingly (but truly) assuming that the newspaper is a reliable source of local news. Does this prevent her or him from coming to know about what has happened in the community?
The second reason is Alston’s famous level confusion argument in regards to mediate knowledge. He argues that the requirement of justified belief is tempting because we are not careful to differentiate between being mediately justified in believing that C and being justified in believing that one is mediately justified in believing that C. If this distinction is not upheld, Alston argues, one will naturally suppose that what is required for the latter is also required for the former.
Yet, Cling’s argument seems compelling. Perhaps it is then that we should hold that the unreflective justification of animals and preverbal children is not the same kind of justification that we are after when discussing ‘full-fledged’ epistemic justification of mature adults. Bearing in mind the proneness of humans to argue fallaciously, we might want to raise the bar for justifiably believing an indirect belief.[iv] However, this might lead into difficult questions on where to draw the line between reflective and unreflective justification (Cf. also second argument below). Alston’s second reason is not so much an argument as it is an explanation of why we end up requiring (JCC). An independent argument for (JCC), which Cling appears to have, should defuse it. But there are also independent arguments for challenging Cling’s position. This is where I will turn now.
First, there is the case of the Tortoise (Carroll 1895) and the looming regress. If the arguer must justifiably believe that ‘if P, then C’ in order to justifiably believe C, must the arguer not also justifiably believe ‘if (if P, then C) and P) then C’ and so on ad infinitum. Cling (2003, pp. 293-294) responds to this by arguing that the Tortoise’s point is doxastic, not epistemic. The puzzle only shows that one cannot force the acceptance of any argument on a person who refuses to accept the conditional. This blocks inferential justification and is naturally quite compatible with (JCC). However, one can accept that Tortoise’s point is at least doxastic, but this does not imply that it might not still be epistemically problematic as well. I will come back to this in the fourth argument below.
Second, as Robert Audi (1993, pp. 238-241) has argued, there needs not to be a belief ‘If P, then C’ at all, when someone advances an argument ‘P, therefore C’. We may accept that in every case, where S believes C based on a reason P, there is an argument, an abstract propositional structure, and accept that every such belief is structurally inferential. The abstract structure indicates how the belief is grounded, but such a structure does not necessarily imply that the resulting belief is episodically inferential. For example, I may infer that ‘There is wind out there’ from ‘The trees are swaying’ without conceptualizing the connection. Audi (ibid.) notes that an indirect belief need not arise from an internal recitation of that structure in any way that deserves the name of ‘inferring C from P’. There could thus be de re-beliefs that do not require that one believe ‘If P, then C’ but only that one takes P to support C.
Third, it can be argued that as the acquisition of indirect beliefs is a case of belief-basing, we should pay attention to the fact that the starting belief and the end belief are belief states, but the move from the premises to the conclusion is an action. The argument Cling advances assumes that these can be treated equally. But thinking that the abstract propositional structure that represents the beliefs used in arguing fully describes the act of arguing or inferring is dubious. Cling argues that the there should be persons whose predicaments are such that the only non-epistemic differences between (1) and (2) are those that result from the fact that (2) has (1)’s corresponding conditional as a premise and that (1) and (2) have different corresponding conditionals. But if I can merely take P to support C, and having seen that P, move right away to C, then the difference between (1) and (2) is not epistemically irrelevant: in the latter I have a state of belief which I lack in the former.[v]
Fourth, there seems to be several types of arguments that we take as being able to justify beliefs in their conclusion, yet the theorists have not agreed why exactly they do so, for example inductive arguments and arguments from analogy. If we take the requirement of justified belief in the link of an argument seriously, we might want to ask if anyone is really justified in using these arguments. Similar points seem to apply to several arguments about the coherence of a theory in respect to its competitors or to arguments about simplicity. This relates to Tortoise’s point. For example, I may justifiably believe that theory T1 is more coherent (simpler) than T2 and infer from this that theory T1 is more preferable than T2. But assume that this inference is done in the context of a relatively undeveloped field of study. A critic might then quite reasonable launch two different attacks. First, the critic could challenge my belief in the premise that T1 is more coherent than T2 and the implication that this coherence (simplicity) should lead to difference in preference. Second, the critic might admit the premise and the implication, but still, quite reasonably, ask why I accept this argument (in toto) in this case, i.e. why should the lesser coherence of T2 be a sign of its falsity in this relatively undeveloped field. Perhaps it is only due to our lack of further knowledge about the field that makes T1 seem more attractive. I do not think that Tortoise’s point about the corresponding conditionals can be swept under the rug so easily. We need further argumentation to show that the point is only doxastic.
One further point that should be noted is the nature of belief basing relation. Cases of justifiably believing C based on P are instances of belief basing. There are at least four different theories of this basing: the causal, the counterfactual, the doxastic, and the causal-doxastic. Arguably, none of these four are incompatible with us not accepting (JCC). For the case of causal, counterfactual, and causal-doxastic this seems clear. According to causal accounts (such as Moser 1989), the belief in the conclusion C is based on the belief in the premise P, if it is causally sustained by this belief in a non-deviant manner. According to counterfactual theories (e.g. Swain 1981), the belief that C is either caused by the belief in the premise P or would have been caused by the belief that P in appropriate circumstances. The causal-doxastic account (see Korzc 2000) is disjunctive: either the belief in the conclusion is caused by the belief in the premise or there is the appropriate meta-belief to the effect that P is a good reason to believe that C.[vi] This leaves doxastic theories, which usually include the requirement that S must have the appropriate meta-belief to the effect that P is reason to believe that C. But one notable representative of doxastic theory is Robert Audi (1993, p. 241) who does not require that S conceptualizes the relation between P and C, nor believe that P implies C. Admittedly, the basing relation is a controversial issue and all of these theories have to deal with difficult counterexamples. But we should bear in mind that these theories are compatible with the rejection of (JCC), although strong access internalism would seem to necessitate its acceptance.
If a person can come to believe C based on the belief that P without believing that if P then C, then the difference between
(1) P therefore C.
(2) (P and if P then C) therefore C.
is not arbitrary in respect to normal belief basing, and therefore not epistemically arbitrary either. However, some final remarks need to be made on what the rejection of (JCC) does not imply. First, it does not imply that by the use of the argument ‘P, therefore C’ the arguer does not become committed to the link of the argument. Second, this does not imply that the link of an argument should not be objectively good. The issue is only to what extent the arguer needs to be aware of this goodness when the argument is used. Third, this does not imply that one cannot normally track down one’s reasons and consider them critically, although subconscious reasoning might be an exception.
[i] These terms have become standard philosophical jargon. For discussion, see e.g. Alston 1989, ch. 4 and Audi 1993, ch. 8.
[ii] Cling (2003, pp. 281-282) divides justification-affording further into justification-creating and justification-affording. The former are cases where the argument creates justification for belief in their conclusion and the latter are cases where there is already some justification for the conclusion. Cling limits his discussion to arguments that are supposed to be justification-creating.
[iii] It should be pointed out that Cling’s argument is directed against certain philosophers (nick-named friends of self-support) who argue that we can refute the sceptical challenge against deductive and inductive reasoning by holding that both of these methods can be used to justify themselves, given that we do not require that the arguers must be justified in believing in these methods when using them to support themselves. I am no friend of self-support, even though I have my doubts about the principle (JCC): I would argue that we can reject self-supporting arguments by other means but this cannot be attempted here.
[iv] On empirical research on reasoning and discussion on fallacies, see Perkins 2002.
[v] To make a simile, the premises can be likened to bullets and the inference to a gun. A deductive argument is like a gun that hits the target every time one has good bullets (i.e. true premises). Cling’s position would appear to imply that one could hit the target with just two bullets and no gun. (I beg the reader to pardon this militant simile. Having a taste for knock-down arguments is compatible with believing that argumentation is not inherently adversarial or even competitive.)
[vi] However, Korcz (2000, p. 548) also includes in the doxastic disjunct a causal explanation.
Alston, W.P. (1989). Epistemic Justification. Ithaca: Cornell University Press.
Audi, R. (1993). The Structure of Justification. Cambridge: Cambridge University Press.
Cling, A.D. (2003). Self-Supporting Arguments. Philosophy and Phenomenological Research 66, 279-303.
Korcz, K.A. (2000). The Causal-Doxastic Theory of the Basing Relation. Canadian Journal of Philosophy 30, 525-550.
Carroll, L. (1895). What the Tortoise Said to Achilles. Mind 4, 278-280.
Moser, P.K. (1989). Knowledge and Evidence, Cambridge: Cambridge University Press.
Perkins, D.N (2002). Standard Logic as a Model of Reasoning. In: D.M. Gabbay, R.H. Johnson, Ohlbach, H.J. & J. Woods (Eds.), Handbook of the Logic of Argument and Inference: the Turn towards the Practical (pp. 187-223, Ch. 5), Amsterdam: Elsevier.
Swain, M. (1981). Reasons and Knowledge, Ithaca: Cornell University Press.