1 The Question

In this article, I seek to discuss some putative answers to the question ‘Why be rational?’ The purpose of ‘Sect. 1’ is to clarify my understanding of that question, and ‘Sect. 2’ outlines who I take to be asking it and discusses its intelligibility. In ‘Sect. 3’, I critically assess three different answers, each bearing enough of a resemblance to suggest that they belong to the same family, and in ‘Sect. 4’, I offer a brief statement of my conclusion and its scope.

The question ‘Why be rational?’ is ambiguous. We may usefully distinguish two readings of it. On the first reading, we concede that we should act on reasons, or, at least, certain types of reasons, namely ‘normative reasons’. A normative reason is a consideration that counts in favour of, or against, doing something — this is distinct from a ‘motivating reason’ which may explain behaviour but need not justify it. But we may wish to distinguish, with Niko Kolodny, between objective rationality, which one exemplifies when one acts as one is required to by reasons which obtain on the basis of some feature of one’s actual situation, and subjective rationality, which one exemplifies when one’s attitudes conform to rational requirements governing the relations between them (for example, of coherence) — in abstraction from any reasons we might have for or against those attitudes (Kolodny, 2005, 509–510). Having made this distinction, we might further wish to say, again with Kolodny, that though we may accept that we should act in accordance with the requirements imposed on us by objective rationality, we need not accept that we should act (nor even be disposed to act) in accordance with the requirements of subjective rationality. To say that we should (be disposed to) act in accordance with the latter requirements calls for further argument, and so the question ‘Why be rational?’ needs to be answered, even having already conceded that there is some sense in which we should be rational.

On the second reading, we make no such concession. We do not just question whether we should (be disposed to) adhere to the various strictures on action that comprise subjective rationality, such as consistency (in which, typically, the injunction not to so adhere is motivated by a reason of its own). We ask why ‘normative reasons’, obedience to which is constitutive of objective rationality, should have a claim on our allegiance; why should we act in accordance with them? Note that this is not the same thing as to claim that there just are no such things as normative reasons. The position from which we ask this version of the question ‘Why be rational?’ is not that of a nihilist about normative reasons. Unless further considerations are advanced to the contrary (and I will examine an example of such a further consideration in the next section, namely that the second reading of this question is just unintelligible), one can admit the existence of normative reasons within objective rationality, but nevertheless question why one should follow them.

In my clarifications so far I have been speaking of reasons to act, of practical rationality. But the question ‘Why be rational?’, in the sense that I wish to examine it, would put into question whether we should obey epistemic reasons, reasons to believe, as well. So, in the same way that some have said that we need not be epistemically (or, to use a different word, theoretically) subjectively rational, but we may nevertheless grant epistemic normative reasons within objective rationality force, my interest is in refusing to grant that — even if we still grant that epistemic reasons exist.

One final distinction that is worth making is between what we might (following Olson, 2011, 77–80) call ‘categorical’ and ‘non-categorical’ reasons. I may have reason not to use the fish knife for the salad course, and I may also have reason not to use the fish knife to decapitate the waiter. But while the former of these is non-categorical — that is, the reason only has force for me on the condition that I have certain ends or desires (such as the desire to adhere to a particular type of dining etiquette) — many philosophers would want to say that the latter reason has force for me irrespective of any ends or desires I might have. So it might be possible to refine the interpretation that interests me of the question ‘Why be rational?’ further, to ‘Why should we act, or believe, on the basis of categorical objective normative reasons?’.

2 Who Is Asking the Question?

Who is asking this question ‘Why should we act, or believe, on the basis of categorical objective normative reasons?’ (henceforth, unless I specify otherwise, my references to ‘reasons’ will refer to such technically specified reasons), or, for short, ‘Why be rational?’Footnote 1 Anyone can ask it, but I have a specific questioner in mind (at least for the moment). Suppose someone has not ruled out the possibility of not acting or believing on the basis of such reasons. You may even be that person. Let’s call this person the arationalist, in contrast to the rationalist, who has ruled out that possibility — at least in those cases where one is not making some kind of error. The arationalist is not non-rational, like a stone or an amoeba, which don’t have sufficient cognitive capacities to have any attitude towards reasons. On the other hand, I prefer not to use the term ‘irrationalist’ as this may suggest someone who is consistent in his rejection of being responsive to reasons. But, if an individual is genuinely not committed to being responsive to reasons, and that fact would be (to anyone committed to respond to reasons) a reason, in any given case, to not respond to that case’s associated reason, then the aforementioned individual may choose, in any given case, not to respond to the reason his reason-non-responsiveness gives him not to respond to the associated reason. In short, I think the term ‘irrationalist’ may suggest a consistency that someone who is not committed to reasons-responsiveness need not have, and so I prefer the term ‘arationalist’. (Of course, we may worry here that, in saying that the arationalist need not even be consistent in not being reasons-responsive, we are trying to make the arationalist more consistent than the irrationalist in some second-order way. To seek to quiet that worry by noting that the arationalist can be consistent in this way as he is not beholden to be consistently inconsistent seems to call for the same worry and the same response to apply at the third-order level, and so on. This, I suspect, suggests a problem (which I will call the ‘situating problem’) in trying to rationally explain arationalism, as I am doing here, and which I hope to examine more thoroughly another time.) The arationalist/irrationalist distinction is quite an important one, as we’ll see later.

Our next point of clarification should be: ‘In what way does the arationalist, when he does so, not respond to reasons?’ I suggested earlier that the arationalist need not be a nihilist of any kind about reasons; he need not think that there just are no such things as reasons. Nor (and I appreciate that the relation between the concepts of ‘reason’ and ‘normativity’ is controversial) need the arationalist be a nihilist about normativity, where this is either restricted to epistemology or universal. This affirmation of normativity could be a full-blooded ‘categorical’ normativity, or an affirmation of a merely ‘non-categorical’ normativity (cf., for example, Husi, 2011 for the latter). The arationalist I want to consider is one who may accept both the existence of objective reasons and their categorical normative force, but still ask why one should be responsive to them.

To those who affirm our duty to be responsive to reasons, the rationalists, such a position may have an air of incoherence about it. Spencer Case suggests, in reference to Husi’s view, that ‘[w]e could imagine an epistemic counterpart of [a] psychopath, who knows that the epistemic norms forbid undisciplined pattern-seeking and indulgence in confirmation bias, yet who thinks that these norms are arbitrary, like the rules of a game that, at any time, he can opt out of playing’ (Case, 2020, 97). Case rejects this view; he believes that such epistemic norms cannot be sensibly seen as merely arbitrary, and so one must be either a nihilist or a believer in categorical norms.Footnote 2 Suppose he’s right, and — for sake of argument — forget nihilism. Can’t we conceive of an epistemic counterpart of a kind of super-psychopath, who knows that the epistemic norms forbid certain cognitive moves, and doesn’t think these are arbitrary, but still feels able to opt out of compliance with them? No, the rationalist will say, we can’t conceive that. For if the norms in question are truly categorical, one cannot opt out, as one can with arbitrary ones. (Benjamin Kiesewetter comes close to making this point, in saying in a note that ‘once the existence of a reason has been accepted, its normative authority cannot coherently be questioned’ (Kiesewetter, 2017, 4). However, Kiesewetter’s main text suggests that his reason for thinking this is not simply because there is a conceptual connection between reasons being categorical and our being obliged to obey them, but rather because the question ‘Why should I do what I have decisive reason to do?’ asks for reasons, and so its asker presupposes the normative authority of reasons. This is a different response to the arationalist, one which I will consider in ‘Sect. 3.2’.) Of course, one might not act in accordance with these norms, if say, one is mentally compromised, or lacks the capacity to understand them — but to the extent that one does this one would be a non-rational being, not an arationalist.Footnote 3 The arationalist has asked the question ‘Why be rational?’, and, we might think, any arationalist who asks this question and yet is happy to concede that there are truly categorical epistemic norms which provide reasons simply fails to understand the conceptual relationship between the existence of such norms and the reasons they provide on the one hand and the necessity that we obey them on the other — if the norms exist, the necessity of our obedience (as recognition, if not in our practice) to the reasons they provide is ‘baked-in’, so to speak. That arationalist is like the person who says ‘I know I should do these things, but why should I do them?’, or, to use an alternative formulation suggested by an anonymous referee, ‘Why ought I to do or believe what I ought to do or believe?’ Such a position is, the same referee worries, empty, if it is intelligible at all — it may simply be incoherent.Footnote 4 In the face of this important concern I believe there are a couple of points worth making.

On the level purely of intellectual sociology, it is at least de facto the case that some committed rationalists who have considered this intelligibility concern about the asking of the question ‘Why be rational?’ have ultimately dismissed it, and sought instead to offer an alternative response by either rejecting the arationalist’s question on some other basis or attempting to successfully answer it. These rationalists include some whose alternative responses to that question will be examined later, such as Harvey Siegel and Nicholas Rescher. Hence, even if a given rationalist personally thinks that the intelligibility concern is dispositive, it is instructive to see whether there is any response to the question ‘Why be rational?’ for another rationalist who is not of a like mind.

That said, a route into a more substantive response might be to observe that my earlier talk of ‘a psychopath’ and ‘a super-psychopath’, following and expanding on Case, is misleading. For such talk patterns the arationalist on a certain type of amoralist, but there is an important difference. Before he talks about an epistemic psychopath, Case talks about a regular moral psychopath, who might believe there are moral norms but thinks he can opt out of them. He thinks such a person would have a ‘defective’ idea of morality — I take it that this means that his conception of morality would not be adequately supported by epistemic reasons, where this lack of reasons-responsiveness is what leads to a concept of morality that is different from the non-psychopath’s insofar as it countenances the possibility of opting out of moral norms. And I suspect Case would say the same a fortiori of a moral super-psychopath, who believes there are categorical moral norms but nevertheless asks why he should obey them — though here the irrationality of the position is even more evident as it is more obviously contrary to reason, more obviously a violation of epistemic norms, to claim that there are categorical moral norms but then to go on to ask why one should obey these. But what it’s important to note here is that it is the idea of morality that is defective in the case of the moral (super-)psychopath. The problem with his position is not that he is disobeying moral norms, but that he is disobeying epistemic ones. And that makes for a significant disanalogy with the epistemic (super-)psychopath. For the moral (super-)psychopath is (we assume) committed to an observance of epistemic norms and the reasons they provide, rejecting fealty only to moral ones and the reasons they provide, and so he can be convinced by his opponent, using epistemic reasons, that he is making a mistake — that his conception of morality is defective insofar as it is contrary to reason. Not so the epistemic (super-)psychopath. The problem that is being leveled at the epistemic (super-)psychopath is not that leveled at the moral (super-psychopath): that his opting out of following one set of norms and the reasons they provide (moral ones) conflicts with his own and his interlocutor’s common decision to follow a different set of norms and concomitant reasons (epistemic ones). It is that his opting out of obedience to one set of norms and their reasons conflicts with his interlocutor’s decision to obey that same set of norms and their reasons (epistemic ones in both cases). If an arationalist is like the person who says ‘I know I should do these things, but why should I do them?’, then the arationalist is clearly not responsive to (at least some) reasons, for his question is not rational. Now we are clearer on exactly what the arationalist is doing, we can be clearer on the rationalist’s concern, which is that, in opting out of obedience to epistemic norms and concomitant reasons and asking a question that is not rational, the arationalist is making an unintelligible challenge.

Given this, I’m not sure that the rationalist’s concern disposes of the arationalist’s question. It certainly appears possible to ask intelligible questions which are nonetheless not rational to ask, in addition to it being possible to ask unintelligible questions which are not rational to ask. So, let us return to the rendering of the question as ‘I know I should do these things, but why should I do them?’, or ‘Why ought I to do or believe what I ought to do or believe?’ Is there any intelligible reading of these?Footnote 5 It is tempting to say that ‘Why ought I to do what I ought to do?’ does have an intelligible meaning insofar as it’s interpreted as ‘Why ought I to do each those specific things which I ought to do?’ But this reading of the question buys its intelligibility at the expense of its philosophical interest; the question is easily answered piecemeal by giving the corresponding reasons for doing each of the specific things that I ought to do. To give the question philosophical interest, the reading required would be: ‘Why ought I to do whatever it is that it turns out I ought to do?’ where ‘whatever it is’ refers to nothing specific. The concern is that this philosophically interesting reading will lack intelligibility, but I’m not convinced that this is so. It certainly seems that this reading of the question is susceptible to being answered, even if only trivially — the answer would be: ‘Because whatever it is that it turns out you ought to do just is what you ought to do’. If whatever it was that one ought to do turned out, on further reflection, also not to be what one ought to do (in exactly the same sense), reality would be contradictory, but, as reality is not contradictory (pace dialetheists), this fact gives us reason (explains why we ought) to do what we ought to do.Footnote 6 But the fact that a question can be answered trivially does not show it to be an unintelligible question; indeed, the fact that the question can be answered at all shows it precisely to be intelligible. So the question is intelligible even if the answer to it is so evident to the rational mind that it seems strange to give it. Of course this answer, insofar as it gives a reason, will, for all that it is so evident to the rational mind, be unlikely to move the arationalist, who will regard it as question-begging on the part of rationalism.

But the rationalist may continue to press his point. He may demur from the claim that one can ask intelligible questions that are not rational to ask (at least in cases were these violate categorical objective normative reasons), saying that such questions can always be rephrased more lucidly to reveal their unintelligibility. Furthermore, the rationalist may attempt to enlist certain philosophical theses, of varying controversy, to demonstrate that this specific question is unintelligible — perhaps he will say, in response to the considerations advanced in the previous paragraph, that a negation of a contradiction is merely a senseless tautology and a senseless proposition cannot be an answer to a question, and so cannot show the question to be intelligible. Discussion of the internal details of such a strategy, which will rest on a number of contentious claims, would require more space than I have here and so I lay them aside. But an immediate ‘external’ rejoinder to this response is that if one claims that the question ‘Why be rational?’ is not merely trivially answered but rather simply unintelligible then it is still the case that a commitment to rationality cannot be shown to be justified. Of course, it’s equally true that a commitment to rationality cannot be shown to be unjustified either. If it simply makes no sense to ask for reasons to be reason-responsive — rather than it being meaningfully tautologous to do so — then the rationalist cannot look askance at arationalism (although the arationalist can look askance at rationalism, as his looking askance need not be reason-responsive). This result is surely at least somewhat uncomfortable for the rationalist.Footnote 7

Still, I’d like to suggest a more radical reply: why might the arationalist, in the face of the claim that the question is unintelligible, not nevertheless continue to ask his question? Obviously, to take this position will be unintelligible, a conceptual confusion to the rationalist who obeys epistemic norms, but to consider this an objection to the arationalist’s question or his asking of it will merely amount to nothing more than calling the arationalist an arationalist; as such, it is difficult to see what role this can play on its own in assessing the question ‘Why be rational’? when it is genuinely asked (where, by ‘genuinely asked’, I mean that the asker is not presupposing rationalism). This being so, the arationalist is within his rights to advance his challenge to the rationalist and to request a response.

The rationalist might reply at this stage — feeling, one imagines, somewhat annoyed — that although he is obliged to answer rational challenges to his position, it is rather too much to ask him to answer irrational or rationally unintelligible challenges. But the seeming obvious plausibility — indeed, rational incontrovertibility — of this response (and of the idea that a given challenge just making no sense is dispositive of that challenge), especially to the rationalist, may devolve from the fact that typically any response that a rationalist will have been asked to give to a certain question has, up until now, taken place against a background of a mutual acceptance of rationalism.Footnote 8 The difference is that here we are precisely not inhabiting a context of such mutual acceptance. To rule out answering such a question on the basis of its rational unintelligibility would be analogous to the follower of an oracle saying that we cannot ask questions that place into doubt the unquestionable word of the oracle as there just cannot be an unquestioningly obedient answer to such question. Of course, the rationalist is liable to respond that we can at least make sense of, make intelligible, such a question asked of an oracle, whereas this is not true of a question asked of rationality in general. But this merely restates that the dogmatism in question pertains to rationality rather than to oracular pronouncement or somesuch. To discount unintelligible challenges to rationalism, where unintelligibility is understood in terms of rationalism, is to beg the question. Maybe this seems crazy. But ‘craziness’ is understood in terms of rationalism and so to accord this seeming any weight will be question-begging as well.

The frustrated rationalist may be tempted to riposte by saying that, if he is to respond to an unintelligible challenge, he can respond however he likes and interpret the challenge however he likes. This is true, but in responding to a rationally unintelligible challenge, the rationalist will be acting against the dictates of reason which declare that a rationally unintelligible challenge cannot be given a response. Consequently, he will have conceded his rationalism. Of course, there is a legitimate further concern that the rationalist may have that he is unable to force himself to understand the question, even if he decides to be arational, as it will just make no sense to him — whether it does or not is a fact the obtaining of which is not under his control. But here the arationalist can be accommodating: it is only from a rationalist’s perspective that the question needs to make sense to be answered. If the rationalist is prepared to disavow his commitment to rationality, it is perfectly possible, from an arational perspective, to answer a challenge which makes no sense.

A rather different concern about the figure of the arationalist may be that the concept of such a person himself, rather than his question, is incoherent. For, to use the words of an anonymous referee, ‘many... person-constituting features seem to be tied with the possession of reason-responsive attitudes’; if this appearance accurately reflects reality, then we cannot attribute practical or cognitive agency to the arationalist, nor personhood. This concern, an application of a specific type of constitutive argument (cf. Tubert, 2010), is a pertinent and serious worry for the line of thought that I am developing, and I thank the referee for urging it on me. My inclination is that the questions raised by constitutive arguments — on which there is a considerable literature — both in general and in this specific context are sufficiently deep and wide-ranging that I will be unable to do justice to them here given my other aims. This being the case, I hope the reader will forgive me if I offer a promissory note that I intend to take them up in a future publication dedicated solely to the matter.

Before I finally move on, I’d like to deal with a couple of complications. I’ve said that I would, at first, like to consider the arationalist as the asker of the question, which I render for short, ‘Why be rational?’ But S. F. Sapontzis has attacked this rendering of the question — or at least, the near-enough-equivalent ‘Why should I be rational?’ — on the basis that, in contrast to the question ‘Should I be rational?’, it presumes that we should be rational; a presumption that an arationalist would scarcely want to make (Sapontzis, 1979). Sapontzis bases this claim on the fact that ‘No, you shouldn’t be rational’ is an answer to ‘Should I be rational?’, but not to ‘Why should I be rational?’ I am not convinced by this point. All that ‘Why should I be rational?’ presumes is that being rational is a viable option, is ‘on the table’ — but then, ‘Should I be rational?’ presumes that too. The difference between the two questions is one of specificity: a response to ‘Should I be rational’ can be a simple ‘No, you shouldn’t’, but a response to ‘Why should I be rational?’ will have to either specify a reason or say that none exist.Footnote 9 But if we say (as it seems we ought) that the ‘should’ in ‘Why should I be rational?’ is associated with the same kind of normativity as it is in ‘Should I be rational?’ then to say that no reasons exist in this case amounts to answering ‘Should I be rational’ with a simple ‘No, you shouldn’t’. So I don’t think Sapontzis really marks out an important difference between these two questions, although there may be a rhetorical motivation for the questioner to use one form rather than the other.

I think the insignificance in difference between the two versions of the question is borne out by Sapontzis’ treatment of ‘Should I be rational?’, as he goes on to say that only reasons can answer ‘Should I... ?’ questions. But he says himself that ‘No, you shouldn’t be rational’ is a ‘relevant answer’ to ‘Should I be rational?’, and that does not give a reason. To avoid this contradiction, Sapontzis will either have to try to distinguish between two types of ‘answers’ to questions (which he nowhere tries to do) or will have to say that ‘No, you shouldn’t be rational’ is elliptical for ‘No, you should not be rational, because there is no reason to be’, where the ‘because’ indicates that a reason is being given in answer to the question. And, if, as I’ve suggested, the ‘should’ in both versions of the question is doing the same normative work, then the negative answer to ‘Why should I be rational?’ is ‘There are no reasons to be’ which must then be elliptical for ‘There are no reasons to be and so that’s why you shouldn’t’, which is logically equivalent to the answer to ‘Should I be rational?’ Of course, Sapontzis might say that this conclusion is the worse for the arationalist, who then has no way of formulating his question without presuming rationality; once again I shall postpone discussion of this worry until the next section.

In an insightful paper, Harold Brown, in a different way, takes issue with the arationalist as the raiser of the question ‘Why be rational?’ He distinguishes between asking the question without assuming rationality (the arationalist way), and asking it when assuming rationality (the rationalist way). But he takes the difficulties in answering the question without assuming rationality (the raising of which without such an assumption he does not dismiss) as suggesting that ‘the way to raise it’ is to raise the question with the assumption of rationality, with the further understanding that failure to answer the question satisfactorily on that basis ‘could count against the question’s presupposition’ (Brown, 1978, 241) — presumably, the presupposition that we should be rational. But I don’t see how the difficulties in answering the question without assuming rationality suggest this. Answering the presuppositionless question is its own investigative endeavour, and so it seems immaterial whether the question with a presupposition can be satisfactorily answered or not. What would be needed to show that we should tackle the question with a presupposition first is to show that a satisfactory answer to the question with a presupposition obviates the need to answer the presuppositionless question, and Brown doesn’t give us any reason to think that is the case.

3 Assessment of One Set of Putative Answers to the Question

3.1 Rescher

I have now set out what I mean by the question ‘Why be rational?’, and who I take (at least initially) to be asking it. In the remainder of this article, I want critically to assess an influential family of answers to the question. These are often advanced by self-confessed rationalists. What qualifies each of these answers to be members of the same family is their common appeal to the idea that reasons-responsiveness can justify itself. My conclusion will be that all such responses are wanting; to the task of attempting to show this I now turn, dealing with members of the aforementioned argumentative family as propounded by Nicholas Rescher, Harvey Siegel, and Max Black.

The obvious concern will be that responses to ‘Why be rational?’ that purport to show that reasons-responsiveness is self-justifying in the face of such a question will be question-begging or circular in some way in their reliance on reasons. Rescher’s contention is that it is not question-begging or viciously circular (he seems to see the latter as at least a species of the former) to claim that we can justify a determination to respond to reasons on the basis of reasons (for example, on the basis of the reason that being responsive to reasons is the best way of achieving certain goals). He appears to claim that basing rationality on reasons exhibits a virtuous circularity, showing how ‘things are connected together in a well-co-ordinated and mutually supportive relationship’ (Rescher, 1988, 43). Of course, this is not going to be much help to the genuine asker of ‘Why be rational?’ Rescher himself doesn’t appear particularly comfortable with this answer either, as he admits at first that it ‘has an appearance of vitiating circularity’ as the argument’s force rests on an appeal to reasons, and then, even more damagingly, suggests that this is not merely an appearance in his next sentence: ‘[T]his sort of question-begging is simply unavoidable in the circumstances’ (Rescher, 1988, 43). So Rescher moves from saying that justifying rationality with reasons is not question-begging to saying that it has the appearance of question-begging to saying that it is a form of begging the question in the space of two paragraphs.

His next move is to say that this sort of question-begging is what just we want and need, as there is nowhere else to look for a rational validation of rationality than to reasons. But no follower of reasons wants or needs reasons that beg the question, as their question-begging nature cancels their utility as reasons (I’ll say more on this in the next paragraph). It may be that what we want and need here just can’t be had (some kind of reason that can validate the use of reasons yet not beg the question), but if that’s the case, then so much the worse for rationalism. The fact that what we want and need can’t be had does not license use of something that we can have (question-begging reasons) but that can’t do the job. The genuine asker of ‘Why be rational?’ has no brief to defend rationality, and so need not expect (as Rescher seems to assume) that the answer must specify a reason which does ground obedience to reasons — the answer ‘There is no reason to be rational’ is a perfectly good answer, and has the virtue of not being question-begging. To a genuine questioner, the claim that the only way of justifying rationality is via reasons is not going to lead to him being happy with circularity, but is rather going to lead him to the belief that rationality is unjustified (how he reacts to this revelation is another matter, of course).

Rescher’s problem may be that he is assuming that the question is being asked by someone who holds to rationality and just wants reassurance (it is not, to use my parlance, being genuinely asked): ‘Already embarked on the sea of rationality, we want such assurance as can now be made available that we have done the right thing... [a]nd such reassurance can indeed be given’ (Rescher, 1988, 44). But why assume that we already have so embarked on the sea of rationality? In other words, how does this help with the arationalist’s question? Moreover, and as I will develop in more detail in the next section, even if one is seeking reassurance it is suspect to use a methodology which one has not yet given reason to use to reassure oneself about one’s use of it — there is no assurance to begin with for one to reassure oneself about! Of course, one might be psychologically assured of it, as one’s using it is based on an unexamined assumption insofar as the question is not even entertained, but this (i) is hardly satisfying for the rationalist once he realizes that the assurance is merely psychological, and (ii) means that no virtuous circular grounding actually takes (or has taken) place from the point of view of the rationalist as an epistemic agent.

Again, Rescher shows a certain self-awareness about the problems with his position, as he considers a last ‘desperate’ objection: ‘So rationality speaks on its own behalf... But why should I care for rationality? Why should I set myself to do the intelligent and appropriate thing?’ Far from being some last, desperate roll of the dice for the questioner, though, this is exactly what the genuine questioner’s query has been all along. Given that, what does Rescher have to set against it? Unfortunately, not a great deal: ‘At this point there is little more to be said’. After re-iterating his point that, if one is already committed to rationality, one can use reason to support the use of reason, which we have already evaluated, Rescher simply says ‘trying to reason with someone who stands outside the range of rationality to convince them to come into its fold is clearly an exercise in... futility’. But the genuine questioner is merely asking why he should care for rationality (concerns about what that questioner’s use of ‘why’ may betoken will be addressed in the next section), rather than asking to be reasoned into it — he would be perfectly happy with the answer ‘There is no reason to be rational’. Given Rescher’s concession that any attempt to reason with such a person is futile because such an attempt will beg the question (and this is what I take Rescher’s ‘clearly’ to refer to), this should reveal to the rationalist that, were he to genuinely pose the question ‘Why should I care for rationality?’ to himself, he can give no answer to it, and so his past commitment to rationalism was ungrounded.

3.2 Siegel

Siegel believes that rationality can be self-justifying in a non-circular and non-question-begging way. He firstly attempts to show that a self-reflexive strategy of justification can work in other cases. His most pertinent example concerns theories of epistemic justification, like foundationalism or coherentism; the proponents of such accounts, Siegel claims, offer those accounts hoping they will be successful, and these will be justified in their own terms — if a theory is a theory of epistemic justification then, if it is successful, and thus presumably justified, it will have to apply to itself.

But this observation does not serve to show that self-reflexive strategies of justification can work in a non-question-begging manner. For, while the proponent of, say, foundationalism, may expect that particular account of epistemic justification to be ultimately shown to be justified in a foundationalist way, this can only be discovered after the account has been argued for using premises and principles which can equally be affirmed by foundationalist and non-foundationalist alikeFootnote 10 (otherwise the non-foundationalist will indeed find the foundationalist’s arguments question-begging), and so foundationalism is not used to justify itself. So we need to distinguish between the capacity of an account of epistemic justification to apply to itself, and the capacity of an account of epistemic justification to justify itself — the former is a harmless post-hoc discovery, the latter betokens a vicious circularity. So it has not been shown that is a successful self-reflexive strategy of justification in general.

Let’s look at how Siegel proposes such a strategy would work in the specific case of rationality justifying itself.Footnote 11 His key argument is that, in asking the question ‘Why be rational?’, one presupposes rationalism insofar as one is asking for reasons to establish whether one is justified or not to accept it, where asking for reasons in this way commits one to acknowledge the potential force of reasons. ‘In this sense rationality is self-justifying: one cannot question it except by accepting it, for acceptance is a precondition of the serious posing of the question’ (Siegel, 1997, 82).Footnote 12

The immediate response the arationalist might have when hearing such an argument is that it gives them no reason to be rational, for it is an argument and only the rational need take the reasons provided by arguments to believe a proposition to have any claim on their obedience. Of course Siegel will reply that the arationalist, in showing himself to care about the potential force of reasons (i.e. to be willing to respond to at least some reasons), has revealed himself not to be an arationalist after all, and so cannot use his status as an arationalist in this way. But here the arationalist can counter that the fact that he is an arationalist does not mean that he need not care about the force of reasons, and so his care for the force of reasons in asking the question cannot be taken as evidence for him not being an arationalist. He may point out that, as an arationalist, he feels no need to be consistent in his attitude to the claims that the force of reasons have upon him, that is, to disregard them always or to regard them always, and so the fact that, on the occasion of asking his question he demonstrates an acceptance of the force of reasons, does not mean that he will accept them at any other time or in any other context (even were he to ask the very same question two minutes later). His lack of acceptance of the force of reasons at any other time or context can include even reasons related with the utmost conceptual propinquity to his acceptance of the force of them in asking the question ‘Why be rational?’ — for example, he can happily acknowledge that he accepts the force of reasons when asking the question ‘Why be rational?’ but equally happily reject any claim that this acceptance gives him a reason with force that he must obey to accept the force of reasons in general or in any other case. It may even be that he would reject the claim that his acceptance of the requirement to obey the force of reasons on his occasion of asking the question gives him a reason to accept the requirement to obey the force of reasons on that same occasion of his asking the question. This may not be a rational position (to say the least!), but obviously the arationalist will not be concerned with that. The great advantage the arationalist has here is that the rationalist is obliged to obey certain dialectical rules, but the arationalist need only do so when it suits him.Footnote 13 Siegel’s self-justification argument for rationalism, in attempting to show that any serious asker of the question must presuppose rationalism, is shot through with words like ‘commits’, ‘cannot’, etc., which are all elliptical for ‘rationally commits’ and ‘rationally cannot’, etc. (just as the ‘must’ earlier in this sentence is elliptical for ‘rationally must’). But this will only be plausible if used to preach to the choir; such words draw their normative status from their appeal to the force of certain reasons, and the arationalist need not heed such force. The moral is: an arational position will be impervious to attempts to bring it under rational rules (although it should be noted that this moral is precisely an attempt to do just this — part of the ‘situating problem’ I referenced earlier).

So the arationalist may seriously ask the question ‘Why be rational?’ without committing himself to obedience to the force of reasons, that is to say, without revealing himself to be a rationalist. If this is so, then we cannot say that anyone who seriously asks the question must see that it is self-justifying, and so this version of the self-justification strategy fails (unless we qualify ‘anyone’ here as ‘anyone rational’ in which case the strategy begs the question). It is worth noting at this point a potential ambiguity in what it is to ‘seriously’ ask this question. For Siegel, to seriously ask the question is to take seriously putative reasons for being rational. My inclination is that to seriously ask the question is to ask the question from a point of view that does not assume rationality. I’ve used ‘genuinely’ as a synonym of ‘seriously’ for my understanding of this hitherto, and I will continue to use this to mark the distinction between Siegel and myself. Now there seems to be at least a tension between what it is to genuinely ask the question and what it is to ask it (in Siegel’s sense of) seriously, although to my mind this tension is resolved when we observe, as I have done above, that the arationalist, in virtue of being arational, need not assume rationality in order to take seriously putative reasons for being rational (the tension is, after all, a rational one). But Siegel’s self-justifying strategy to answer the question is vulnerable in another way, given his understanding of ‘serious’, in that it may be perfectly possible for the arationalist to non-seriously ask the question ‘Why be rational?’ to the rationalist in the form of an argument from commitment as a way of embarrassing the rationalist.Footnote 14 The arationalist might deny that there is any need for him to obey the force of reasons, yet nevertheless ask the question, ‘non-seriously’ in Siegel’s sense, and ‘genuinely’ in my sense, in order to show to the rationalist (and this ‘in order to’ need not necessarily be backed by any reason) that the only non-question-begging answer to the question can be ‘There is no reason to be rational’, and that, therefore, the rationalist’s commitment to rationality is ungrounded by his own lights.Footnote 15

The rationalist can respond to this point, though, by inverting the defence I gave of arationalism above. He can say that although the arationalist questions in the hope of producing an argument from commitment, when the rationalist hears the question he will ask it of himself ‘seriously’ in Siegel’s sense — he is committed to rationalism, after all, and so his asking of the question ‘seriously’ in this way will show an implicit commitment to rationalism, and thus prove rationalism to him to be self-justifying. So we seem to be at an impasse; rationality cannot be shown to be self-justifying to the arationalist, but can be shown to be so to the rationalist.

This deadlock can be broken (which would, in turn, show that there would be value in the arationalist advancing an argument from commitment) by observing that, though one might show oneself to be committed to rationality if one asks the question ‘seriously’, it does not follow that, in asking the question, one shows rationalism to be justified. To seriously question rationality, one may indeed need to be committed to obeying the force of reasons used in asking the question,Footnote 16 but, as soon as the question is asked, rationality logically comes into question, and so cannot be assumed — that is, it rationally cannot be assumed, as this would be question-begging. So rationalism leads us out of rationalism. The fact that one was committed to a position in the past, and it was only that commitment, now no longer held, that led one into the conceptual problem one faces in the present, does not confer justification on the position to which one was formerly committed which can then be used to solve that conceptual problem. Moreover, the necessity to suspend one’s commitment to rationalism must appear logically prior to the application of the self-justification argument which makes retrospective appeal to one’s former commitment to rationalism in asking the question (which appearance thus renders the self-justification argument ineffective) as otherwise the question has been begged; the requirement to obey the force of reasons was assumed throughout.

Suppose someone had committed to the position that he would only believe what the tea leaves he read told him to believe and nothing else. One day, he asks the tea leaves what to do, and they tell him to suspend his commitment to tea leaf guidance. But, having so suspended his commitment to tea leaf guidance, he cannot then say that, because it was tea leaf guidance that led him to suspend his commitment to tea leaf guidance, he is still committed to tea leaf guidance and will continue to take his direction from them. As soon as he sets foot outside his commitment to tea leaf guidance as mandated by the tea leaves, which is what he must do in order to genuinely suspend his commitment to them and so to actually follow their guidance in the first place, he can longer appeal to their guidance to form his beliefs — he must use some neutral form of guidance, if there is any. The rationalist is in an analogous position, when, through the application of rationality, we put rationality itself into question, but the problem is more serious insofar as, with absence of guidance by reasons following reason’s dictating that one puts responsiveness to reasons itself into question, there are no neutral sources of guidance to be had.

If we say that to ask seriously ‘Why be rational?’ acknowledges the potential force of reasons, insofar as one is now on the lookout for reasons that may justify our commitment to rationalism, this does not mean that, in asking this question one will not also acknowledge that, in fact, one must suspend one’s previous acknowledgement of the potential force of reasons. This is because it is precisely this further acknowledgement that our previous acknowledgement mandates, insofar as we as yet have no reason to maintain that previous acknowledgement. So we are taken outside of rationalism, and, having moved there, are in no position to support by reasons the acknowledgement of fealty to the force of reasons (not that it will need support by reasons if one has moved outside of rationalism). Indeed, the question ‘Why be rational?’ at one and the same time both acknowledges and rationally calls for suspension of commitment to obedience to the potential force of reasons.

So, in the face of the question ‘Why be rational?’ rationality requires us to ask the question seriously and genuinely, which is not possible unless either one is arational or one separates what it is to ask the question from the development and consideration of candidate answers, and claims that one needs to ask the question seriously, but to develop and consider answers to the questions genuinely (in which case my claim that we need to ask the question genuinely can be made more modest; rather my claim can be that we need to assess answers to the question, once asked, genuinely — not that this should be taken to exclude the possibility of being able to ask the question genuinely on the part of the arationalist). This ‘sequential’ understanding of the demands enjoined by rationality, however, will render the self-justification strategy inefficacious: it is true that one is committed to rationalism when asking the question, but not when going about answering it (indeed it is a rational requirement not to be so committed), and the self-justification strategy just is a candidate answer to the question.

3.3 Black

Black gives a slightly different argument to the effect that rationality can justify itself in some sense. He offers the following on behalf of the rationalist: ‘You should attend to reasons, because unless you do so, such-and-such (which you don’t want) will be the case’. Here ‘such-and-such’ is a placeholder for some state of affairs which, insofar as the rationalist’s interlocutor is concerned, it is undesirable should obtain, and so provides an effective reason to motivate paying attention to reasons. Obviously, this has an air of circularity about it, and, equally obviously, Black is aware of this concern. His response is to say that the conclusion, ‘One should attend to reasons’ is not needed as a premise in the argument, which would be circular; all that is needed is that ‘such-and-such’ is a good reason for not rejecting the conclusion.Footnote 17 From the point of view of the sceptical asker of ‘Why be rational?’, Black says that he need not believe the conclusion ‘One should attend to reasons’ to see Black’s argument as valid; he merely needs to see that the ‘such-and-such’ given is indeed a good reason for holding that conclusion.

The sceptical questioner as I have characterized him — the arationalist — need not be moved by this. For in genuinely asking why we should obey the force of reasons, he not only will not feel obliged to obey reasons in general, but also in any particular case, that is, in any case that fills in the ‘such-and-such’ placeholder. Black concedes something along these lines — he accepts that if the questioner does not ‘see’ that the particular reason given is a good one for the conclusion, his argument fails. His response is to level a bit of question-begging abuse: ‘No argument... can appeal to a questioner too stupid to recognize a good reason when he sees one – or too obdurate to confess that he does recognize it’ (Black, 1982, 153). Of course, the arationalist does not fail to recognize good reasons, nor is he too obdurate to confess that he sees one when he does. Rather, his response when confronted with a good reason, a reason with force, is to question whether he should obey it. Perhaps Black does not consider such a possible reaction due to its obvious deviation from rationality, and maybe if he had, he would have been able to seek a different abusive synonym for ‘not rational’ to label the arationalist with. But, given that, this synonym would be just that — a synonym, with a question-begging negative connotation to it. And to label an opponent’s position with a label that he himself accepts is scarcely to refute it.

Given this, it seems that Black’s argument would only work for those committed to obeying what they take to be a good reason when confronted with a particular instance of one. But I am not even sure it will do that. For Black concedes that no-one could sincerely give the above argument unless they already believed the truth of its conclusion (i.e. that one should attend to reasons), and, further, that we ‘show that we accept [the conclusion] when we argue in its favour’ (Black, 1982, 152). To his mind, this is not a concern: ‘[T]here is nothing disturbing, from a logical point of view, about this, since we often find it useful to argue for a conclusion that we already believe to be correct, e.g. in order to strengthen an antecedent belief or to establish logical connections between disjointed beliefs’ (Black, 1982, 152). But if this is the only use Black thinks he can put his argument to, it will be of no use in actually answering the question ‘Why be rational?’ when it is asked in the first place.Footnote 18 If all that Black means to do with this argument is to strengthen an antecedent belief or establish logical connections between beliefs, then he is not genuinely asking ‘Why be rational?’, but rather has assumed that we should be rational with no justification — that is the aforementioned antecedent belief. But if that is so, the use of this argument to ‘strengthen’ such an unjustified belief is suspect, for its power to strengthen derives solely from the unjustified assumption which it purports to strengthen our belief in (and which it cannot, as Black seems to implicitly acknowledge, provide post hoc justification — rather than mere strengthening — for). There might be nothing disturbing logically about Black’s manoeuvre, as he says, but that sets the bar rather low. It is, of course, perfectly licit for us to given an argument for a conclusion we already accept to strengthen it, but this license is withdrawn when the conclusion we are arguing for is one which concerns the concepts invoked in the argument, just as the concept of rationality will be (implicitly or explicitly) at some point in the case of Black’s argument.

4 Concluding Remarks

My aim in this article has been rather modest: to raise a question, describe a particular sort of individual who might raise that question — the arationalist — and defend the legitimacy of so raising it, and consider a certain family of answers to that question. There are, of course, other families of answers or idiosyncratic answers which I have been unable to canvass here, and there is much more to say about this figure of the arationalist (including examining the ‘situating problem’ I referenced in ‘Sect. 1’). In particular, the constitutive argument against the coherence of the concept of an arationalist referenced in ‘Sect. 2’ is a significant challenge which demands a response. Here, though, I have sought only to indicate that one family of answers to the question ‘Why be rational?’ has no force, and thus that the question may be more worrying than it first appears — both to rationalists and arationalists.Footnote 19