42

Default‐Reasonable Beliefs An increasingly influential thought these days runs along the following lines: It is a mistake to suppose that a belief is unreasonable or unjustified merely because it is not supported by some observation or argument. Some beliefs are simply ‘default reasonable,’ reasonable in and of themselves, without any supporting justification. In particular, the fundamental logical beliefs have this feature.9 It is reasonable to believe them, but not because there is some positive ground by virtue of which they are reasonable. So although it is true that no inference or observation supports our fundamental logical beliefs, it does not follow that they are unjustified, and so the potentially paradoxical conclusion is blocked. I am not implacably opposed to the idea that there might be some beliefs that (p.239) are reasonable on the basis of nothing, especially if this is understood to mean simply that they are presumptively but defeasibly justified. It is possible that this will prove to be the best description of the epistemology of our first‐person knowledge of the contents of our own minds. What I do not see is how this idea could plausibly apply to the case at hand, to the generalization that all inferences of a certain form are necessarily truth‐preserving. If a particular class of beliefs is default reasonable, then, it seems to me, there must be some explanation for why it is. This insistence does not contravene the root idea that, in the case of a defaultreasonable belief, there is no ground that makes it reasonable; for it is consistent with a belief’s having that status that there be a criterion by virtue of which it has that status and an explanation for why it has it. All that the insistence amounts to is that there be some principled way of saying which beliefs are default reasonable in the relevant sense and why. What I think we can’t obtain, in the case under consideration, is any sort of satisfactory explanation for why belief in the validity of MPP should count as default reasonable. One popular suggestion has it that a default‐reasonable belief is any belief which, by virtue of being presupposed in any justification that a thinker might have, is neither justifiable nor refutable. But this has two implausible consequences. First, it entails that what is default reasonable has to be relativized to individual thinkers, for different thinkers may build their epistemic systems around different claims. Second, it has the consequence that some very implausible claims would come out as default reasonable for someone if they happened to be presupposed by that person’s epistemic system. For example, suppose that someone takes as basic the negation of the law of non‐contradiction; on this view, we would have to say that the negation of that law is default reasonable for him, because, by assumption, it will be neither justifiable nor refutable for that person. A second suggestion has it that the beliefs that are default reasonable are those beliefs that a thinker finds ‘self‐evident’—that is, that he is disposed to find plausible simply on the basis of understanding them and without any further support or warrant. But this proposal, too, would seem to be subject to the previous two objections. Once again, it is entirely possible that two people will find very different propositions ‘self‐evident’, and that some of those will include propositions that are intuitively highly implausible. Nor would it help to strengthen the requirement so that it concerns those beliefs that actually are selfevident, as opposed to those that merely seem self‐evident. Here the problem is that no one seems to me to have shown how this notion is to be spelled out. In particular, no one has supplied a criterion for distinguishing those propositions that are self‐evident from those that—like the parallel postulate in Euclidean geometry or the proposition that life cannot be reduced to anything biological—merely seemed self‐evident to many people for a very long time. By contrast, there is one form of explanation that seems to me have some (p.240) promise. There may be beliefs that are such that, having those beliefs is a condition for having one of the concepts ingredient in them. Any such belief, it seems to me could plausibly be claimed to be default reasonable. For if it really were part of the possession condition for a given concept that to possess it one had to believe a certain proposition containing it, then that would explain why belief in that proposition is at least presumptively (though defeasibly) justified. It would certainly not make sense to insist—in the way that it would with the belief that all swans are white—that one justify the proposition before one came to believe it. Unfortunately for the problem that concerns us, it is not remotely plausible that anyone possessing the concept of conditional would have to have the belief that MPP is a valid rule of inference. Surely, one can have and reason with conditional without so much as having the concept of validity or of logical implication. At most what the theory of concept possession would license is that inferring according to MPP is part of the possession condition for conditional, not the belief that MPP is valid. But what we are after now is the justification for the belief.10 Non‐Factualism About Justification Another thought that has also been gaining in influence recently is even more radical. It denies not merely that there always has to be a ground in virtue of which a belief has the property of being reasonable; it denies that being reasonable is a property in the first place. How are we to think about this? We are assuming that logic is factual, so there are facts about what logically follows from what. What we are contemplating is the idea that there is no fact of the matter to about what it would be reasonable to believe, given a specification of the available evidence, however exactly the notion of evidence is to be understood. (p.241) Now, belief, we may all agree, constitutively aims at the truth, so that anyone who was in the business of inquiry and belief in the first place would have to subscribe to the norm that one ought to believe only what is true. Furthermore, it seems reasonable to suppose that this norm will, by itself, ground various other norms: for example, that one ought to believe that which is likely to be true and not believe that which has no chance of being true; that one ought not to believe p if some alternative proposition incompatible with p has a higher likelihood of being true; that one ought to believe p only if its likelihood of being true is high enough, given the sort of proposition it is.11 Plausibly, then, all of this will be common ground between believers, however their epistemic principles may otherwise differ. So what we are imagining, when we imagine that there is no fact of the matter about what it would be most reasonable to believe on the basis of a given specification of the evidence, is that there is no fact of the matter which rules of evidence are correct—no fact of the matter, that is, about which hypothesis a given piece of evidence confirms. To put matters crudely, but starkly: we might have one epistemic system incorporating a rule of simple induction, and another a rule of simple counter‐induction, and there would be no saying that one system is more correct than the other. But this sounds like an impossible view. If every epistemic system can be said to be as correct as any other, does not it follow that every proposition is as justifiable as any other, since every proposition can be justified relative to some evidential system or other? And is not such a gross relativism about justification not only false but self‐undermining? For if every proposition is just as justifiable as any other, then that would appear to imply that for any proposition p, p and its negation are equally justifiable. In particular, it would seem that the claim that not every proposition is as justifiable as any other is just as justifiable as the claim that it is, if the claim that it is is true. And that would appear to leave us lacking a reason for believing anything in particular. Now, this objection would have been decisive if the view that there are no facts about the correctness of evidential systems had to be accommodated in overtly relational terms. The contemplated objection assumes, in other words, that on the view in question, the predicate ‘x is justified’ denotes a relational property that obtains between a belief and an evidential system. That assumption, then, in conjunction with the non‐factualism about evidential systems, generates the relativism that we found to be problematic. But, as is well known, and as we have just seen in the case of logic, non‐factualism does not have to be elaborated in this way. Instead, it is better to try to understand it along expressivist lines. Allan Gibbard has developed just such an expressivist theory of judgements of rationality; adapted to the present case it would yield something like the following view: When someone says ‘x is a justified belief’ they are not attributing any (p.242) sort of property to it at all, relational or otherwise; rather, they are expressing their acceptance of a system of norms that permits that belief in those circumstances.12 Does this formulation evade the self‐undermining relativism just described? Well, since in saying that ‘x is justified’ I am not stating anything at all, but merely expressing my acceptance of a system of norms that permits that belief in those circumstances, and since, naturally, I do not accept an evidential system that permits believing just anything, I do not have to agree that every proposition is just as justifiable as any other, even if I am a non‐factualist about justification. So it appears—prima facie anyway—that a Gibbard‐style expressivism about justification is not subject to the relativistic worries just outlined. That makes it worth while to ask whether it can help with the problem about logic with which we are concerned. The following thought might seem tempting. On a non‐factualist view, nothing is really justified as such, for there is no such property for things either to have or to lack. Rather, talk about a belief’s being justified is just a way of expressing one’s acceptance of a system of norms that permits believing L. Since there is no further question about whether L is really justified, apart from its figuring in an epistemic system that I accept, our problem with the justifiability of logical beliefs disappears. I am entitled to say that a given logical principle is justified if I accept an epistemic system that permits it. And, naturally, I do accept an epistemic system that permits the core principles of logic. So it can seem as if this is the answer we were looking for: a viable reconstrual of the notion of justification that refutes the claim that the core logical beliefs are unjustifiable. Unfortunately, these appearances are deceptive. Ultimately, an expressivism about justification is just as unpalatable as a relativism about it. Or so I will argue. Let us imagine that I come across someone—call him AR—who holds a view I consider utterly unjustified: for example, that there is a spaceship trailing the comet Hale‐Bopp that is going to come down and swoop him away. What can be my attitude towards such a person, given a Gibbard‐style expressivism? I can express my acceptance of a system of norms that forbids that belief, all right, but that seems to leave something important out. If I tell AR that his belief that p is irrational and unjustified, I am not merely expressing my acceptance of a system of norms that forbids it; I am claiming to see something that he is not, namely that p ought not to be believed, given the available evidence. I am saying (roughly): I do not believe p; you should not either. Gibbard tries to account for the normativity of such judgements by invoking a classic expressivist resource: the conversational demand. In saying that x is unjustified, he says, I am expressing my acceptance of a system of norms that forbids x and adding: Do so as well! (p.243) In and of itself, however, this does not capture the claim that I appear to be making when I claim that I am justified and he is not, for even someone who is simply browbeating his interlocutor can issue a conversational demand. So the question is: with what right do I insist that someone accept my view and abandon his, on non‐factualist views of justification? Could not AR insist, with equal right, that I abandon my view in favour of his? Indeed, as a non‐factualist, would not I have to recognize that our claims to normative authority here are perfectly symmetrical, thereby undermining any hold I might have had on the thought that I am justified and he is not? And is not this a version of the sort of relativism expressivism was supposed to avoid? Now, AR’s belief about alien spaceships may arise in a number of different ways. He may share all my epistemic norms on the fixation of belief and he may be very good at reasoning from those norms and the available evidence to the relevant conclusions. He may simply not be aware that there is not a scintilla of evidence that there is a spaceship trailing Hale‐Bopp. In that case, there is no difficulty accounting for my demand that he give up his view in favour of mine. Knowing that his problem stems simply from an ignorance of the relevant facts, I can coherently ask that he take my reasoning as proxy for his own. And he, for his part, would be entirely reasonable in taking me up on my invitation. Then, again, AR’s curious belief may derive not from his ignorance of any item of evidence but from his poor abilities at reasoning: he may be bad at moving from the epistemic norms that we share and the evidence to the appropriate conclusions. Here, again, there is no difficulty accounting for the normative authority that I claim. Given that we share the relevant norms, I can again ask him to take my reasoning as proxy for his own. But suppose that the difference between AR’s beliefs and mine stems not from such mundane sources but rather from a deep‐seated difference in the fundamental epistemic norms to which we subscribe, norms for the fixation of belief that are not derived from any others. In calling his view irrational, then, I am in effect demanding that he give up his fundamental epistemic norms in favour of the ones that I employ. And the question I am asking is: With what right do I do this, on a non‐factualist view? As a realist, I would have no trouble explaining my attitude here. Since, as a realist, I take there to be a fact of the matter which fundamental norms are correct, and since I take myself to know what those facts are, I can easily explain why I am insisting that my interlocutor give up his norms in favour of mine. Of course, my interlocutor, convinced of the correctness of his own norms, may make a similar demand on me. If the norms are fundamental, this may well result in an impasse, a disagreement from which neither of us can be budged by argument. But it would at least make sense that there is a disagreement here, and it would make sense why we are issuing (potentially ineffective) conversational demands on each other. But what explanation can the non‐factualist offer of these matters? The non‐factualist may reply that there is no difficulty here. After all, he will (p.244) say, the rules of evidence that I accept are unconditional: they apply to someone whether or not that person is inclined to accept them. There seem to me to be two problems with this reply, however, one with the assumption that I accept unconditional norms in the first place, the other with my insistence that someone else also accept them. First, if a non‐factualism about justification is correct, with what right do I accept epistemic norms that are unconditional, so that they apply to someone whether or not they accept them?13 If there really are no perspective‐independent facts about which epistemic norms are correct, with what right do I accept norms that apply to people whether or not they accept them? Should not an appropriate sensitivity to the fact that there is nothing that makes my norms more correct than anyone else’s result in my being hesitant about accepting norms that apply to others regardless of whether they are also inclined to accept them? Second, and putting this first problem to one side, on what basis do I insist that AR give up his unconditional norms in favour of mine? I accept a particular set of fundamental norms. He accepts another. By assumption, the norms in dispute are fundamental, so there is no neutral territory on which the disagreement can be adjudicated. Furthermore, on the non‐factualist view, there are no facts about which fundamental epistemic norms are correct and which ones are not. So, on what basis do I insist that he give up his norms in favour of mine? The expressivist thinks he can evade the clutches of a self‐undermining relativism by claiming that talk about a belief’s being justified expresses a state of mind rather than stating anything. But this stratagem does not long conceal the view’s inevitable relativist upshot, which can now be restated in terms of the problem of normative authority. If no evidential system is more correct than any other, then I cannot coherently think that a particular belief is unjustified, no matter how crazy it may be, so long as that belief is grounded in a set of fundamental epistemic norms that permit it, no matter how crazy they may be. To repeat: the point here is not about suasive effectiveness. I do not mean that the realist about justification will have an easier time persuading anyone of anything. In fact, it is quite clear that there are numerous extreme positions from which no one can be dislodged by argument, whether confronted by a realist or an expressivist (this is a point to which we will have occasion to return). The issue is rather about having the resources with which to think certain thoughts coherently. By virtue of believing that there are objective facts about what justifies what, the realist can coherently think that a particular epistemic system is mistaken. The non‐factualist, however, cannot. In a sense, the difficulty should have been evident from the start. For the root problem is with the claim with which the expressivist about justification must begin, that there is nothing that epistemically privileges one set of epistemic principles over another. Once that thought is in place, it becomes impossible to evade (p.245) some sort of relativist upshot. It does not matter whether the basic thought is embedded in an expressivist or a non‐expressivist frameworkRule‐Circular Justification: Two Problems If this is correct, then only one possibility remains: that we were too hasty in assuming that a rulecircular justification of a logical belief cannot confer genuine justification upon it. Let us, then, reopen that question. What intuitive constraint on justification does a rule‐circular argument violate? It will be useful to approach this question with another: what intuitive constraint on justification does a grossly circular argument violate? For it is plausible that a rule‐circular argument will be problematic to the extent to which it approximates a grossly circular argument, an argument that explicitly includes among its premises that which it is attempting to prove. There are at least two things wrong with a grossly circular argument. First, it assumes that which it is trying to prove and that, quite independently of any further consequences, seems wrong. An argument is being put forward with the intent of justifying—earning the right to believe—a certain claim. But it will only do so if it proceeds from premisses that are justified. If, however, the premiss is also the conclusion, then it is simply helping itself to the claim that the conclusion is justified, instead of earning the right to it. And this manoeuvre offends against the very idea of proving something or arguing for it. As we are prone to say, it begs the question. A second problem is that, by allowing itself the liberty of assuming that which it is trying to prove, a grossly circular argument is able to prove absolutely anything, however intuitively unjustifiable. Let us call the first problem the problem of ‘begging the question’ and the second that of ‘bad company’. Is a merely rule‐circular justification subject to the same or analogous worries? It is not obvious that a rule‐circular argument begs the question, for what we have is an argument that is circular only in the sense that, in purporting to prove the validity of a given logical law, it must take at least one step in accordance with that law. And it is not immediately clear that we should say that an argument relies on its implicated rule of inference in the same way as we say that it relies on its premisses.14 (p.246) Well, perhaps not in the same way, but it is not difficult to motivate a worry on this score. One clear way of doing so is to look at the role that a rule‐circular argument might play in a dialectical context in which it is being used to silence a sceptic’s doubt about its conclusion. Suppose that you doubt some claim C and I am trying to persuade you that it is true. I offer you an argument A in its support. In general, in such a context, you could question A’s cogency either by questioning one of its premisses or by questioning the implicated rule of inference R. If you were to proceed by challenging R, then I would have to defend R and my only option would appear to be to try to defend my belief that R is truth‐preserving. Now, suppose that the context in question is the special case where C is the proposition that R is truthpreserving and my argument for C is rule‐circular in that it employs R in one of its steps. Here it very much looks as if I have begged the question: I have certainly begged your question. You doubt MPP. I give you an argument in support of MPP that uses MPP. Alert enough to notice that fact, you question my argument by reiterating your doubts about MPP. I defend my argument by asserting that MPP is truth‐preserving. In this dialectical sense, a rule‐circular argument might be said to beg the question. At a minimum, then, the sceptical context discloses that a rule‐circular argument for MPP would beg a sceptic’s question about MPP and would, therefore, be powerless to quell his doubts about it. In doing this, however, it reveals yet another sense in which a worry might arise about a rule‐circular argument. An argument relies on a rule of inference. As the sceptical scenario highlights, one’s reliance on such a rule might be questioned. But, quite apart from whether it is questioned, in what does one’s entitlement to rely on that rule consist, if not in one’s entitlement to the belief that the rule is truth‐preserving? And if it does not consist in anything else, how can a rule‐circular argument in support of belief in MPP confer warrant on its conclusion? In relying on a step in accord with MPP, in the course of an argument for MPP, one would be leaning on the very conclusion one is allegedly trying to prove. Under the general heading of a worry about begging the question, then, I want to distinguish two problems: first, to say in what the entitlement to use a rule of inference consists, if not in one’s justified belief that that rule is truth‐preserving; (p.247) second, to say how a rule‐circular argument can confer warrant on its conclusion even if it is powerless to move the relevant sceptic. What about the problem of bad company? Prima facie, anyway, there looks to be a big difference between a grossly circular argument, on the one hand, and a rule‐circular argument on the other, so far as their potential to positively rationalize belief is concerned. A grossly circular argument is guaranteed to succeed, no matter what proposition it is attempting to rationalize. A similar charge could not be made against a merely rule‐circular argument: the mere licence to use an inferential step in accord with modus ponens, for example, does not in and of itself guarantee that a given argument will succeed in demonstrating the validity of modus ponens. Appropriate permisses from which, by (as it might be) a single application of MPP, we can get the general conclusion that MPP is truth‐preserving, may simply not exist. In general, it is a non‐trivial fact that a given rule of inference is self‐supporting in this way. While this point is strictly correct, however, the fact is that unless constraints are placed on the acceptability of rule‐circular arguments, it will nevertheless be true that we will be able to justify all manner of absurd rules of inference. We must confront the charge that unconstrained rule‐circular justifications keep bad company.