ISSA Proceedings 2002 – On Reasonable Question-Begging Arguments

1. Introduction
This paper will criticize the claim that arguments that beg the question can, in some special cases, yield reasonable belief to the conclusion, made by Kasper Lippert-Rasmussen (2001). Lippert-Rasmussen presents and examines two possible cases of arguments that appear to beg the question, but arguably give the addressee a reason to believe the conclusion. Based on these cases, Lippert-Rasmussen puts forth the following criterion:

A question-begging argument is reasonable if:
1. the addressee of the argument has reasons independent of the conclusion to accept the premises of the argument;
2. the addressee of the argument fails to conduct his reasoning on the basis of these reasons; and
3. the reasons for which the addressee rejects (or accepts) the conclusion are bad ones (Lippert-Rasmussen, 2001, 126).

What does this criterion amount to? Lippert-Rasmussen builds on the view of David Sanford (see 1972, 1981, 1988, 1989). The details of Sanford’s account need not concern us here, but two essential elements on which Sanford builds should be noted. Namely, according to Sanford, whether an argument begs the question should be decided based on the 1) the logical relations between the propositions involved and 2) the way belief in these propositions has been acquired, namely that the belief in the premise is not due to the belief in the conclusion. In essence, if one has the right reasons for holding a belief and uses these beliefs appropriately, one reaches conclusions worthy of belief. Lippert-Rasmussen argues that a comparison between the content of belief-set and arguments is what decides the value of the argument, not the manner in which the contents of belief-set are actually used: if the addressee of the argument has good reasons to accept the conclusion, the argument is reasonable, irrespective of the way these reasons affect the process of reasoning.
The intuitively good thing about Lippert-Rasmussen’s cases is that they look like a situation where a person sees the error of his or her ways and says: “Okay, I admit it now: I did actually have good reason to accept the premises (I just didn’t remember them), and the reasons why I originally rejected the conclusion were bad ones (I never should’ve trusted that fellow anyway). It was indeed a good argument.” It is this way of looking at arguments that we need to discuss. Is it tenable? Is it acceptable to evaluate arguments solely by the content of the arguer’s belief-set?
I will argue that Lippert-Rasmussen’s argument can be challenged on two main points. First challenge objects to his argument against the division between being justified in believing (situationally justified) and justifiedly believing (doxastic justification). I will argue that this division is important for the explication of fallacies. Second, and closely related, challenge is that it can be questioned whether it is these arguments that actually provide reasons for believing the conclusion. At the end of the paper, I will also briefly consider the nature of second-order conditions in argument evaluation. My conclusion is that Lippert-Rasmussen fails to show that arguments that beg the question can make the addressee’s belief in the conclusion reasonable.

2. Cases of Reasonable BQ-arguments
First, we need to make two qualifications. Lippert-Rasmussen (2001, 123-124) argues that there are question-begging arguments that may give the addressee a good reason to believe the conclusion. This should be separated from the claim that the addressee of a question-begging argument may have good reasons to accept its conclusion that are independent of the premises of the argument. His claim is that a question-begging argument itself may give the addressee a good reason to accept the conclusion. Second, Lippert-Rasmussen does not want to claim that all BQ-arguments are reasonable, only that reasonable BQ-arguments are possible.
The following case should be an example of a reasonable question-begging argument. First, let us suppose that I believe that Smith is not in the dining hall. I also believe that he is not in the library. I infer that it is not the case that Smith is either in the dining hall or in the library. An acquaintance of mine thinks that Smith is in the dining hall. To convince me of this, he offers the following disjunctive syllogism: “Smith is either in the dining hall or in the library. He is not in the library. Thus, he is in the dining hall.” This argument would seem to beg the question, since I would accept the disjunctive premise only if I were to accept the conclusion. Lippert-Rasmussen notes that the disbelief in the disjunctive premise is grounded on the disbelief in the conclusion (Lippert-Rasmussen, 2001, 124).
Although the argument seems to beg the question, it might be reasonable, according to Lippert-Rasmussen (2001, 124-125), if the following two things were true. First, my reason for accepting that Smith is not in the dining hall might be a bad one. Second, I might have good reasons to accept the disjunctive premise, but fail to bring these reasons to bear on the issue. Both could happen for a number of reasons. I might just very much want it to be the case that Smith is not in the dining hall and base my belief on unreliable witnesses. Even when I am being offered the argument, I might fail to recall that Smith in fact told me yesterday that he would spend the day alternating between the library and the dining hall. Thus, even though I have good reasons to accept the premise, and a good reason to discard my earlier belief that Smith is not in the dining hall, I might not use either one of them to my benefit. In this case, the argument would be reasonable one to accept, even though it begs the question against me.
The second example Lippert-Rasmussen (2001, 125-126) puts forth is a case where I believe, independently of the conclusion, the premises and have good reason to do so. Still, I might fail to base my acceptance of the premises on these good reasons: I might have just visited the library and the dining hall. I believe to have observed Smith in the dining hall, and infer from this that Smith is either in the library or in the dining hall. Now my friend offers the same argument as before. Again, it seems that the argument begs the question. Still, this might be reasonable if the following two claims were true. First, my observation that Smith was in the dining hall might be an unreliable one. It could be that I know that Smith’s look-a-like twin brother is working in the dining hall, but failed to recall this when making the observation. Second, it could be that I am justified in believing the premises independently of the conclusion. I could have known that Smith will be alternating between the library and the dining hall, visited the library, and despite seeing everyone there, saw no one looking like Smith. Again, my failure to make these considerations relevant in connection with the argument could be due to several reasons that are similar to the first case. Yet, the argument may give me reason to accept the premise.

3. Argument for the Reasonability
Lippert-Rasmussen’s case for reasonable question-begging arguments turns on whether it is the case that if an argument is to give you a reason to accept its conclusion, you must be justified in believing, and actually believe, the premises(i). This challenge stems from the distinction between being justified in believing (situational justification) and justifiably believing (doxastic justification). If one holds that one must actually believe the premises for them to count as justification, one could then argue that Lippert-Rasmussen’s examples are not cases of justification. Lippert-Rasmussen argues that this would only affect the first example, because in the second, he does believe the premise, but he formulates the challenge in a stronger form:
“One way of doing this is to insist that for an argument to provide one with a reason for accepting the conclusion not only must one’s belief in the premises be present in consciousness: one must also believe the premises for the right reasons. This I do not do in my second example. Certainly, there is a clear sense in which I could see the arguments as providing me with a reason for accepting the conclusion: I just have to recall the strong independent evidence I have in favour of the premises and accept them on that basis. But this, so the present challenge goes, is irrelevant to whether the argument in fact gives me a reason to accept the conclusion, since I did not reason in light of this evidence. Hence, in neither of my two examples does the argument in question give me a reason to accept the conclusion. Neither example qualifies as a reasonable question-begging argument” (Lippert-Rasmussen, 2001, 128).

Lippert-Rasmussen finds this complaint unpersuasive for two reasons. First reason is that although we could say that I am justified in accepting a conclusion inferable from premises, which I justifiably believe only if my acceptance of the conclusion is grounded on these premises, I am not justified simpliciter in accepting the conclusion. For this would imply I am justified in accepting the conclusion even if my acceptance of it were generated in an intellectually dubious way. Lippert-Rasmussen follows William Alston(ii) in saying that although this is an acceptable concept of epistemic justification, it is not the only one. Instead, he would hold that having adequate grounds for a belief is the acceptable concept (Lippert-Rasmussen, 2001, 128). This argument contains a small mishap. I will construe the doxastic justification Lippert-Rasmussen argues against as:
(DJ) S is justified in accepting conclusion c inferable from acceptable premises p1,…, pⁿ  only if S’s acceptance of c is based on premises p1,…,pⁿ (and S lacks sufficient overriding reasons to the contrary)(iii).

We notice that basing the acceptance on the premises is a necessary condition, not a sufficient one. Thus, there might be other elements involved and we do not have to admit that we have justification when the acceptance is generated in an intellectually dubious way. Nevertheless, let us overlook this for now and formulate situational justification as:
(SJ) S is justified in believing c only if S has adequate grounds for the belief that c (and S lacks sufficient overriding reasons to the contrary).

(SJ) is, in my opinion, too wide to be an acceptable criterion of justification. I admit that we find it natural to say that a person has a reason to accept even the remotest consequences of his or her adequate grounds. But we can still question whether one is justified in believing all the consequences of one’s beliefs, even if one has no idea how one can prove those consequences. If mathematical proofs proceed through relations of equivalence, we would then be justified in believing all the so far unproven theorems. The trouble is, of course, that we do not know what will be proven, thus the interest in, for example, begging the question. This criterion would seem adequate only to a deductive omniscient, which is what Lippert-Rasmussen (2001, 127) explicitly denies we are.

As Lippert-Rasmussen notes, Alston (1985, 75) recognizes that both of these concepts ((SJ) and (DJ))(iv) are concepts of epistemic justification, and concedes that there is something to be said for both sides of the issue. Yet, Alston thinks that even though we might be unable to decide which concept is the concept of justification, we can still consider which concept is more fundamental to epistemology:
“On this issue it seems clear that the [DJ](iv) concept is the richer one and thereby embodies a more complete account of a belief’s being a good thing from the epistemological point of view. Surely there is something epistemically undesirable about a belief that is generated in an intellectually disreputable way, however adequate the unutilized grounds possessed by the subject. […] So if we are seeking the most inclusive concept of what makes a belief a good thing epistemically, we will want to include a consideration of what the belief is based on” (Alston, 1985, 75).

Alston then considers an example. Person x has excellent reasons to suppose that person y is trying to discredit person x professionally, and x does believe that y is trying discredit x professionally. Nevertheless, he argues that if x bases this belief not on the excellent reasons, but on paranoia in such a way that x would believe this even though he did not have these reasons, it is undesirable from the viewpoint of the aim at truth for x to form the belief that y is trying to discredit x professionally. Thus, the basis of a belief counts in epistemic justification. This leads Alston to consider (DJ) as the favored formulation of epistemic point of view.

I submit that we in fact need to supplement even (DJ), as there are two things that can go wrong here. As in Lippert-Rasmussen’s examples, one can have good reasons to believe that c, and base one’s acceptance of c on the wrong reasons. But further, one can also have good reasons to believe c and base one’s acceptance on the good reasons one has, but generate the acceptance of c from these good reasons in an intellectually dubious way. For example, I may have good reason to believe that the sides of my office are two meters, and from this belief, I may form the belief that the surface area is four square meters. However, if I form this belief by adding the sides instead of multiplying them, there should be something intellectually disreputable about my belief. Still, my belief is based on good reasons: I have correctly measured the lengths of the sides and one should use the length of the sides in determining the surface area. Therefore, we should note that we have only paid attention to the source conditions of the belief, not to the transmission conditions of the belief. We might then formulate the doxastic justification again, this time with equivalencies:
(ADJ) S is justified in accepting conclusion c inferable from acceptable premises p1,…,pⁿ if and only if S’s acceptance of c is based on the premises p1,…, pⁿ in an acceptable manner and S lacks sufficient overriding reasons to the contrary.

‘A’ is added to specify that we are dealing with argumentative justification her. It might be objected that including ‘acceptable manner’ makes the criterion circular. I disagree, because when we are studying begging the question, we are explicating what this acceptable manner means. I prefer to use ‘acceptable’ rather than ‘valid’ for it seems unreasonable to limit the criterion to deductive arguments. Naturally, Lippert-Rasmussen may still hold on to (SJ)-view on argumentation, but I believe his position can be strongly challenged.

Second complaint Lippert-Rasmussen has is that this form of doxastic justification involves a rather sweeping indictment of the power of argumentation to render conclusions reasonable. He notes when capitulated, this challenge leads to the view that one has a reason to accept a conclusion based on an inference only if one already accepts it. To explicate, Lippert-Rasmussen puts forth two cases. First, consider the case that you accept and have reason to accept ‘ØP’ and ‘PÚQ’. You are entitled to infer ‘Q’, but you do not infer this and do not accept ‘Q’. Suppose also that your not accepting ‘Q’ is not due to you having strong reasons to accept ‘ØQ’, but simply that you fail to follow the inference through. As Lippert-Rasmussen notes, there is a sense in which you have a reason to accept ‘Q’. Denying this, he holds, would commit one to the view that you have a reason to accept a proposition based on an inference only if you already accept a proposition. (Lippert-Rasmussen, 2001, 128-129.) I would argue that such commitment does not follow, provided we may differentiate between having a reason and being justified. I believe in the present case one has a reason to believe the conclusion, but one is not justified in believing it before one has proven it from acceptable premises, in an acceptable way. One needs not say that an inference can justify a proposition to person A only if A already accepts the conclusion, but that an inference can justify conclusion to person A only if A makes the inference in question. Possible inferences from the premises to conclusion are reasons to accept the conclusion, but do not make the belief in the conclusion justified to a specific person before the inference has been made. As noted already, there is indeed a sense in which we have a reason to believe a statement for which we have justification. Alston (1985, 75) notes that ‘[…] one is better of in believing something for which one has adequate grounds than believing something for which one doesn’t.’ I do not disagree with this, but I think a necessary condition for an inference justifying anything to a person is that the inference is actually made.

Next Lippert-Rasmussen (2001, 129) asks us to reconsider the first example of reasonable question-begging. You accept ‘ØQ’, and see that ‘ØQ’ and ‘PÚQ’ entails ‘P’, but fail to recall the evidence for ‘PÚQ’. Compare this case with the former. Lippert-Rasmussen argues that if one agrees that one had a reason to accept the conclusion one failed to infer in the former case, then it is hard to see how one could deny that the argument gives you a reason to accept the conclusion in the present case – given, of course, that one has a reason to accept the premises. Lippert-Rasmussen argues that it is incoherent to accept the following two claims.

A. Failing to recall evidence in favor of a proposition prevents you from having a good reason to accept a further proposition you know it entails.
B. Failing to make an inference you are entitled to make, from premises you believe and have reason to believe, cannot prevent you from having a good reason to accept the conclusion (Lippert-Rasmussen, 2001, 129).

He notes that it is not inconsistent to uphold this distinction, but the division seems to him ad hoc, and in conjunction with the claim that this position implies a commitment to the view that one has a reason to accept a proposition on the basis of an inference only if one already accepts it, he concludes that this second challenge has little force to it.

Given that the distinction between situational and doxastic justification is tenable, we can agree with (B). We will just add that failing to make an inference does not prevent you from having a good reason to accept the conclusion, but it can prevent you from being justified in accepting the conclusion: having a good reason is not equal to being justified. In the same spirit, we can discard (A). One can have good reason to accept a belief with which one is no longer able to associate a ground, and its implications. Yet, it does not mean that one is necessarily justified in believing all the implications of a belief one has.
If we could assume that one always needs to be aware of adequate grounds to uphold a belief, Lippert-Rasmussen’s claim would be easier to dismiss: the arguer is not basing his belief in the conclusion on the right grounds, but accepts a question-begging-argument. However, we cannot assume this. As the theorists of belief revision have convincingly argued, it might be rational to uphold a belief for which one is no longer able to produce grounds(vi). Yet, the question here is the legitimacy of ways to arrive at conclusions. I submit that the introduction of a new belief needs to be done in an acceptable manner, in order for the new belief to be rational. Transferring justification from premises to conclusion is a different matter than having a reason (whatever that reason may be) to accept the conclusion.
A further argument can be brought against Lippert-Rasmussen from the point of view of fallacy theory by asking what implications his view would have on fallacy theory in general. Namely, the cases Lippert-Rasmussen puts forth are not specific to the fallacy of begging the question but can be reproduced with any fallacy. All one has to do is to assume that the addressee of the argument has reasons independent of the argument to accept the conclusion, but does not, for some reason, bring them to bear on the case. Then, any argument whatsoever is a reasonable argument, as long as the addressee has adequate (unused) reasons for conclusion. I believe this would completely deflate the concept of fallacy. For this reason, I do not think it is in any way ad hoc to insist that having a good reason to believe the conclusion is not yet the same as being justified in believing it. Yet, I must admit that this is not the only of looking at the situation: different emphases may result in different choices. This discussion comes down to how strict requirements we are willing to place on justified belief, and how we see the role of argument in this justification. I am arguing that for the fallacy theorist, the division between doxastic justification and situational justification is a reasonable one to uphold. Perhaps we are here on a point where epistemology, theory of belief revision and fallacy theory part company. Traditionally, epistemology considers whether one is justified in believing one single belief. Theory of belief revision has other goals, namely, to study how to manage larger sets of beliefs. For fallacy theorists, this might be a point of divergence as well.

4. Begging the Question and Second-Order Conditions
Lippert-Rasmussen believes that Sanford’s account of begging the question is basically correct, but disagrees with Sanford on why question-begging arguments are defective. Sanford’s (1972, 198) first formulation is:
“An argument formulated for Smith’s benefit, whether by Smith himself or by another, begs the question either if Smith believes one of the premisses only because he already believes the conclusion or if Smith would believe the one of the premisses only if he already believed the conclusion.”

We can see that both of Lippert-Rasmussen’s cases beg the question by this account. According to him, it is important that we notice that there are (at least) two different desiderata that need to be used in evaluating an argument. First one is conclusion-acceptability desideratum: Does the argument give the addressee a reason to accept the conclusion? Second is the inferential-route desideratum: is the inferential route from the premises of the argument to conclusion non-fallacious? The answer to the second question depends on (a) the addressee’s second-order beliefs about what makes it reasonable for him to believe, or disbelieve, the premises and (b) whether the way in which his belief in the premises is actually grounded, cohere with the route (rationalized by the argument) between his pre-inferential and post-inferential beliefs. Lippert-Rasmussen argues that question-begging arguments per se do not fail the conclusion-acceptability desideratum, but fail the inferential route desideratum. As these desiderata have not been separated, accounts of begging the question have failed to explain what makes the question-begging arguments defective (Lippert-Rasmussen 2001, 138). In both of Lippert-Rasmussen examples, the argument brings out what the arguer has reason to believe.
I agree that if every valid argument gives you a reason to believe the conclusion, so does an argument that begs the question. In question-begging arguments, this should always be the case as the conclusion and the premise might be the same. Then, by Lippert-Rasmussen’s view, we need to study the inferential route desideratum more carefully as every time one has reason to believe the premises, one does have reason to believe the conclusion. Yet a vexing question in respect to Lippert-Rasmussen’s argument is this: if he believes that one needs not actually use the premises one has in order to be justified in believing the conclusion, but to merely to have them, how can he claim that the argument in question makes the conclusion acceptable? Surely, he must admit that the arguments in question have nothing to do with the justification of the conclusion, if the justification stems from premises one has and needs not use in order to be justified in believing the conclusion.
As noted, we now need to study the second-order condition. Lippert-Rasmussen notes that if a person believes the premises of an argument, because he or she believes the conclusion, normally such a person possesses a second-order belief that part of what makes him or her believe the premises is the conclusion. Let us then assume that the addressee in reasonable question-begging arguments does have such appropriate second-order beliefs. Lippert-Rasmussen reminds us that rationalization is not an unfamiliar phenomenon: surely it possible for someone to believe the premises because he or she believes the conclusion and yet believe that premises make it reasonable for him or her to believe the conclusion. According to Lippert-Rasmussen, such a case begs the question on Sanford’s account, although it should not. Therefore, Lippert-Rasmussen believes that the criterion should be supplemented with a clause ‘provided that the addressee does not believe that the premises make it reasonable for him to believe the conclusion’ (Lippert-Rasmussen, 2001, 132-133).

I am hesitant about bringing in the second-order beliefs into the analysis(vii), but for Lippert-Rasmussen’s position, it seems incoherent. He argues that it is not necessary to use even the correct first-order beliefs one has in reaching justified conclusions. Why would the second-order beliefs be more important? I agree that one may arrive at beliefs one has reasons to believe via dubious inferences, but hold that one is not justified in believing such conclusions based on dubious inferences. Yet, Lippert-Rasmussen insists that if an arguer believes the premises because of the conclusion, but is able to develop a second-order belief that the premises make it reasonable to believe the conclusion, one is not begging the question. At first glance, this looks like the classic case of begging the question: conclusion is grounded on a premise and the premise on the conclusion. Could mere second order beliefs justify beliefs? Yet, perhaps this is hasty. We should first take a closer look at the second-order criteria.

The problem is how to make use of these second-order beliefs in analysis. We should then amend the criteria Sanford suggested in his 1981 paper for evaluation of an argument ‘P, therefore Q’:
1. Constitutive Conditions: (i) p; (ii) p implies q.
2. Epistemic Conditions: (i) S believes that p; (ii) S believes that p implies q; (iii) S does not have either of these beliefs because he already believes that q. (Sanford, 1981, 149.)

The constitutive conditions are propositional conditions for argument analysis; only the epistemic conditions involve beliefs. The first and second clause of (2) are about the beliefs one has about the propositions, they can be properly called as first-order beliefs. The third clause of (2) is not similar to the previous two as it is a factual clause. As Lippert-Rasmussen adheres, to certain extent, to Sanford’s account, let us try to formulate a second-order clause that would be in line with these clauses:

3. Second-order Condition: S does not believe that he or she believes that p, only because he or she believes q.
Is this a reasonable condition? Let us first imagine a case where S’s inference passes the evaluation on other clauses, but it turns out that S believes that he believed the premise p, because he believed the conclusion q. It would indeed seem that S has failed to fulfill some epistemic obligation to investigate his beliefs, but it seems questionable to claim that the actual argument begs the question. In fact, it seems somewhat unnatural to suppose that S could even pass the evaluation on other clauses but still believe that he believes the premise only because he believes the conclusion(viii). Second, if S fails the condition (iii) of (2), he seems to beg the question, regardless of his position in respect to (3). Third, if S has no second-order beliefs, perhaps due to inattention, we can still decide whether S’s argument begs the question. (3) seems redundant.

Yet, maybe we just need to rephrase the second order condition. Lippert-Rasmussen discusses reasonable belief. The third clause should also be phrased in conditional form:
(3*) Second order Condition: If S believes that p makes it reasonable for him or her to believe q, the argument does not beg the question.

This formulation makes the appropriate second-order condition as a sufficient condition for the argument not to beg the question. It would also allow for the case that S does not believe the premises make it reasonable for him or her to believe the conclusion, but the argument would still not beg the question, provided the argument passed the other clauses. This would allow the argument to be reasonable even if S had not formed any second-order beliefs.

Next, we need to define the relation of this clause to the two other conditions. There is reason to suppose that it acts as a ‘rider’ that overrules certain arguments that fail to qualify as reasonable arguments by clause (iii) of (2), for otherwise it is redundant. Yet, such a situation seems very odd. It would allow the arguer to make the argument legitimate merely by the act of believing: an arguer may believe that premises make believing in the conclusion reasonable, but also that the conclusion makes believing in the premises reasonable(ix).  Perhaps there are situations where this is possible, especially in complex argumentative situations where one is searching for reasons and conclusions in a network of inferences, but in general it seems suspect that an argument that does not pass clause (2), could be made non-question-begging by the (3*). In order to be able to rule out networks of arguments as justifying each other altogether, one should be able to claim that justification is always unidirectional, something that always proceeds from premises to conclusion, and only from premises to conclusion. However, this cannot be single-mindedly upheld as the ongoing debate between foundationalism and coherentism (in the traditional senses of the terms) shows. But even the hardest coherentist agrees that a circular argument cannot justify a conclusion, if there is no independent line of support for the conclusion (which is what this discussion presupposes). If we were to allow (3*), we could not always rule such arguments as fallacious. In fact, it would legitimize several question-begging arguments. For example, if the arguer believes in God, and therefore starts to believe certain premises that he would not have believed, had he not already believed in God, he begs the question, but (3*) would legitimize these beliefs. (Naturally, this presupposes that we are discussing arguments meant as proofs, not explanations.) I submit that the contention that second-order conditions may legitimize otherwise question-begging arguments, is not tenable.

5. Conclusions
In conclusion, I argue that Lippert-Rasmussen fails to produce question-begging arguments that give their addressee justified in accepting the conclusion. The arguer has a reason to accept the conclusion, but the argument does nothing to add these reasons. If we were to accept these examples as legitimate inferences, we would have to accept that any fallacious argument whatsoever gives the addressee a reason to accept the conclusion, provided the addressee has suitable (unused) beliefs. The problem is that it is not the argument that does the work in these examples, but the content of the belief-set. If we uphold the distinction between having a reason to believe the conclusion (i.e. being in a situation where one could reasonably believe the conclusion) from justifiedly believing the conclusion, we need not admit that Lippert-Rasmussen’s examples are reasonable question-begging arguments, arguments that in themselves justify the conclusion to the addressee.
Nevertheless, the idea of evaluating argument in respect to the whole content of one’s belief store has its appeal. For example, in discussion, we try to use all the commitments the opponent has ever made to our benefit. Still, this idea has its limits. We know that almost everyone’s belief set is inconsistent. Yet, we do think it is rational that to use this inconsistency to derive whatever proposition we wish, as propositional logic would allow. In the examples Lippert-Rasmussen discusses, the inconsistency of the addressee (or should we say compartmentalized thinking) is used to create the impression that the addressee is justified to believe the conclusion based on the premises. I hold that being justified in believing the proposition must be separated from having a reason to believe the proposition in order to keep the notion of fallacy meaningful.

NOTES
[i] Another challenge is to claim that deductive arguments only point out what you must believe because you believe the premises, on pain of inconsistency, but they do not give justification to new beliefs. Instead, you might just reject the premises. This applies to all valid arguments. I will assume this challenge can be rejected: deductive inferences give us new indirect beliefs.
[ii] William P. Alston, 1985, ‘Concepts of Epistemic Justification’, The Monist, 68, p. 75.
[iii] Lippert-Rasmussen does not discuss cases of over-riding evidence but I am adding this qualification to block certain counter-instances.
[iv] (DJ) and (SJ) are not Alston’s formulations of the concept. Alston tries to ‘flesh out’ what it means that believing something ‘is a good thing from the epistemic point of view’. This fleshing out takes the form of giving various ‘modes’ of the term as in “XVI. Objective – S does have adequate grounds for believing that p.” (1985, 73). The ultimate formulation Alston strives for is naturally an equivalence.
[v] Alston (1985, 73-75) is discussing what he calls ‘motivational’ concept of justification (‘S’s belief that p is based on adequate grounds’) in contrast to ‘source-irrelevant’ concept (S does have adequate grounds).
[vi] E.g. Gilbert Harman, 1986, chapter 4. The downside of this is that a belief may end up being justified with itself.
[vii] For example, Alston (1986, 82-83) requires that one needs to be aware of adequate grounds but he does not require that one is aware of the adequacy of the grounds.
[viii] Still, there can be cases where the second-order belief is not in accord with the first-order belief. One could, for example believe that one would feel in certain way in a certain situation, and then, when confronted with the relevant situation, notice that the expected belief does not emerge. Thus, believing that one believes that p does not always imply that one believes that p. (Cohen, 1992, 37.) Yet, mere second-order belief can hardly make a belief justified. Alston (1986, 72-74) argues for the same point.
[ix] Perhaps interesting cases of this kind could be found in inferences to the best explanation. However, I must, bypass this idea.

REFERENCES
Alston, W. P. (1985). Concepts of Epistemic Justification. The Monist, 68, 57-89.
Cohen, L. J. (1992). An Essay on Belief and Acceptance. Oxford: Clarendon Press.
Harman, Gilbert (1986). Change in View. Principles of Reasoning. Cambridge: The MIT Press.
Lippert-Rasmussen, K. (2001). Are Question-Begging Arguments Necessarily Unreasonable? Philosophical Studies, 104, 123-141.
Sanford, D. H. (1972). Begging the Question. Analysis, 32, 197-199.
Sanford, D. H. (1981). Superfluous Information, Epistemic Conditions and Begging the Question. Metaphilosophy, 12, 145-158.
Sanford, D. H. (1988). Begging the Question as Involving Actual Belief and Inconceivable Without It. Metaphilosophy, 19, 32-37.
Sanford, D. H. (1989). If P, then Q. London: Routledge.