On the Nature (and Irrationality) of Non-religious Faith

Abstract

My main aim in this paper is to contribute to the elucidation of the nature of non-religious faith. I start by summarising several well-known arguments that belief is neither necessary nor sufficient for faith. I then try to identify the nature of the positive cognitive attitude towards p that is involved in having faith that p. After dismissing some candidates for the role, I explore the idea that faith and hope are similar attitudes. On this basis, I then advance a new characterisation of faith. Finally, I turn to the question of the rationality of faith. I argue that faith is intrinsically irrational because it is an intrinsically incoherent propositional attitude, but that there is nonetheless a sense in which faith is neither intrinsically epistemically irrational nor intrinsically practically irrational.

1 Introduction

Epistemological debates in the past few decades have largely focused on belief and its norms. However, things have recently begun to change, as other members of the family of cognitive attitudes, such as credences and suspension of judgement, are now receiving more attention, though some other cognitive attitudes, such as doubt and certainty, remain, by and large, neglected despite their manifest centrality in our doxastic lives. In this paper, I shall focus on another cognitive attitude, faith, the fate of which in contemporary philosophical discussions is intermediate. While the nature and rationality of faith have been much discussed in philosophical debates about religious faith, non-religious faith, and its relation to other cognitive attitudes and their rationality (that of belief, in particular), have been much less studied. (Notable exceptions include Lara Buchak (2014, 2017), Daniel Howard-Snyder (2013a, 2019) and Elizabeth Jackson (2021, 2022a), to mention a few).

My aim in this paper is to contribute to the elucidation of the nature of non-religious faith by bringing out the various different conditions that must be satisfied for this attitude to be instantiated.¹ On this basis, I shall then address questions about the rationality of non-religious faith and, more specifically, questions about whether faith is an attitude that is intrinsically irrational. By “non-religious faith”—which I shall henceforth simply designate as “faith”—I shall mean the kind of attitude at issue in the following cases.²

Mary, the Liverpool FC supporter the day before the match. “Though AC Milan are the favourites for tomorrow’s match, I have faith that Liverpool will win. But, for sure, I don’t believe they will win in the way I believed last week they would win against that little third-division team whose name I forgot.”

Steven, the Liverpool FC player twenty minutes before the end of the match (and his disheartened teammate). “We’re three-nil down and they’re better than us. Steven, do you really think we can still come back to win?—Yes, I have faith that we will. And you too have to believe that we will win!”

Esther, the candidate before her mirror (before or after her audition). “True, there are five other excellent candidates, but I have faith that I can get this position. I will do it!”

In order to elucidate the nature of the attitude of faith that is involved in these three cases, I start by summarising some well-known arguments for the idea that belief is neither sufficient (§ 1) nor necessary (§ 2) for faith. I then try to identify the nature of the positive cognitive attitude towards p involved in having faith that p—supposing, on the basis of the arguments advanced in the previous section, that this positive cognitive attitude cannot be that of believing that p. After dismissing some candidates for the role (§ 3), I turn to the idea that faith and hope are similar attitudes. I explore this idea and, after suggesting a new way of distinguishing hope from despair (§4), I advance a new characterisation of faith (§ 5). After indicating how this characterisation can account for the distinctive features of faith mentioned in §1, and can resist the objections raised in §3 against other accounts of faith, I turn to the question of the rationality of faith (§ 6). I argue in this final section that faith is intrinsically irrational because it is an intrinsically incoherent propositional attitude, but that there is a sense in which faith is neither intrinsically epistemically irrational nor intrinsically practically irrational.

2 Why Belief is Not Sufficient for Faith

As they are usually characterised, outright or categorical beliefs (which I shall henceforth simply designated as “beliefs”) are “attitude[s] you form by settling for yourself positively the question of whether p”, so that “your mind is made up that p” (Lee, 2023: 1494), and you “simply take p for granted” (Wedgwood, 2012: 312) or “treat p as given” (Jackson, 2019: 518). Believing that p, in other words, is having a positive cognitive stance on whether p that can be described as taking it as a fact that p.

It is also commonly admitted that belief has a practical dimension, in the sense that believing that p implies being disposed to act, talk, and reason as if it were a fact that p. The same goes for faith: having faith that p also involves having a positive cognitive stance on whether p and being “disposed to live in light of that attitude”, as Howard-Snyder puts it (2017a: 57). According to Buchak, having faith that p “requires that one is willing to take risks on p” (Buchak, 2017: 58)—that is, according to Buchak, it requires that one does not look for, and even “refrain from gathering”, further evidence as to whether p before committing to acting on p (Buchak, 2014: 55). However, since belief also seems to lead subjects to take risks on the truth of p and to render, from the perspective of the subject, further inquiry into whether p superfluous, one may object that it is not clear that faith really differs from belief in this respect. In Sects. 6 and 7 I will argue that the practical dimension of faith does differ from that of belief, but in this section I shall stick to the more obvious major differences between the two attitudes.

First, as Robert Audi writes, “if I do not have a favourable attitude toward something’s happening, I cannot have faith that it will” (Audi, 2008: 97), whereas belief does not require such an attitude. In other words, unlike believing that p, having faith that p requires having a positive conative and/or evaluative attitude towards the fact that p—typically, desiring, or judging it to be desirable, that p. Howard-Snyder has advanced many convincing arguments in favour of this idea. Here is, first, a linguistic argument: “I have faith that Liverpool will win, but I want Barcelona to win” sounds very odd.³ Second argument: “one cannot have faith that something is so without at least some tendency to feel disappointment upon learning that it’s not so” (Howard-Snyder, 2013a: 360). Now, if one tends to be disappointed upon learning that something does not obtain, this indicates that one wanted it to obtain, or judged its obtaining to be desirable. By contrast, “one can believe something even though one has no tendency at all to feel disappointment upon learning that it’s not so” (ibid.). Third argument: “[l]ike fear and hope, faith motivates behavior; that’s why it explains it. But cognition alone cannot motivate behavior; desire is required. Like propositional fear and hope, therefore, propositional faith has desire built into it” (ibid.: 363).

Another major difference between faith and belief is that faith demonstrates more “tenacity”, “perseverance”, or “resilience” than belief (Howard-Snyder, 2013a; Buchak 2014, 2017; McKaughan and Howard-Snyder 2022b): faith that p is more resistant in the face of evidence that not-p than belief. Faith is also more resistant than belief in the face of practical reasons to cease having faith or believing. For instance, if, in the seventieth minute of the match, Steven has faith that his team will win, it is likely that he will still have that faith five minutes later even though it will then be (a) less probable than five minutes before that his team will come back to win, and (b) more difficult than five minutes before to make the physical efforts on the field that having faith involves. On the other hand, if Steven has no such faith at the seventieth minute, but instead believes that they will win—e.g. on the basis of statistics indicating that if they score in the next five minutes their chances of victory are high (but low if they do not) and on the basis of his belief that they are about to score—it is likely that, at the seventy-fifth minute, after no goal from Liverpool, he will no longer believe that they will come back to win. Furthermore, even if, after these five minutes, he takes his team’s chances of victory to remain the same, the increased difficulty of making the physical efforts on the field that having the belief in question involves might well causally lead him to cease to believe that Liverpool will win.

If the foregoing is correct—if belief and faith do differ in the two ways just indicated—it follows that belief is not sufficient for faith. I shall now turn to three arguments to the effect that belief is not even necessary for faith. Whether or not each argument is, by itself, compelling, it is sufficient that one of them succeeds. At the very least, it seems safe to advance that this series of converging and independently plausible arguments strongly speaks in favour of the conclusion that belief is not necessary for faith.

3 Why Belief is Not Necessary for Faith

Let us start with what can be called the doubt argument. It goes like this:

• (1) Having faith that p is compatible with being in doubt as to whether p.

• (2) Believing that p is not compatible with being in doubt as to whether p.

• (3) Therefore, there can be faith that p without belief that p.

Premise (2) seems clearly correct. Indeed, as Howard-Snyder puts it, being in doubt as to whether p is “to be inclined neither to believe p nor not-p” (Howard-Snyder, 2013b: 189). Hence, “if one is in doubt [as to whether p], one will lack tendencies that one has if one believes [that p]” (Howard-Snyder, 2013a: 361), and that constitute the “dispositional profile” of belief according to William Alston (1996: 4). These tendencies include the tendency to answer that p when asked whether p, the tendency to feel it to be the case that p when considering whether p, and the tendency to use p as a premise in one’s theoretical and practical reasonings. Premise (1) is however much less obvious. Consider Mary, Steven, and Esther: in none of these cases do the subjects involved appear to be in doubt as to the truth of the relevant propositions. And if, at some point, these subjects apparently started to be in doubt, we would certainly be inclined to think that they have lost the faith they previously had. Having some doubts as to the truth of a proposition p seems, on the other hand, to be perfectly compatible with having faith that p, but this does not clearly distinguish faith from belief because it is not clear that belief is incompatible with having such doubts.

Nonetheless, the doubt argument points to what seems to be an important difference between faith and belief: faith seems to be compatible with a higher degree of doubt, or a lesser degree of confidence, than belief. As Elizabeth Jackson writes:

faith is compatible with more doubt than belief. Even if belief is compatible with some doubt—as it seems fine to say, ‘I believe p but there’s a chance I’m wrong’—it seems like faith is compatible with even more doubt—more counterevidence or lower credences. (Jackson, 2022a: 204)

For instance, Mary and Steven both have faith in Liverpool’s victory while taking Milan to be the favourites—i.e. while having a degree of confidence in Liverpool’s victory that is below 0.5. But it is generally admitted that, for one to believe a proposition p to be true, it is necessary that one’s degree of confidence that p is true is superior to 0.5.⁴

A second argument for the claim that belief is not necessary for faith can be called the surprise argument. Here I shall simply quote Howard-Snyder, who states it most clearly:

one can have faith that p but lack a tendency to be surprised upon learning it’s not so; disappointment, yes, but not surprise. However, one cannot believe p while lacking a tendency to be surprised upon learning it’s not so. Thus, one can have faith that p without belief that p. (Howard-Snyder, 2013a: 361)

The third argument I shall mention here is the incompatibility argument. It states that faith is “positively incompatible” with belief (Schellenberg, 2005: 132): if one believes that p, one cannot have faith that p. Supposing it is true that one cannot judge that one’s evidence is sufficient to establish that p without believing that p, it follows that, if faith and belief are incompatible, one cannot have faith that p without judging one’s evidence for p to be insufficient to establish that p. the incompatibility argument is partly based on the linguistic observation that it would be very odd to say things like “I have faith that Roger will soon be in front of us on the platform since I just saw him saying hello by the window of the train”, “I have faith that winter has come since it’s December and temperatures have been freezing for three weeks now”, or “I have faith that there’s a computer in front of me”. The reason such statements would be odd is, plausibly, that in such cases I do not really have doubts as to the truth of the relevant propositions but rather categorically believe them.

Howard-Snyder and McKaughan object to the incompatibility argument as follows:

If you have faith that a friend will keep your secret, then, if faith entails uncertainty, you cannot put your faith in those who are most worthy of your faith, namely those with respect to whom you are certain that they are worthy of your faith – which […] seems implausible. (Howard-Snyder & McKaughan, 2022a: 314).

It does not seem, however, that this case (let us call it friend) is an instance of the attitude that is at issue in this paper—i.e. the attitude instantiated in Mary, Steven, and Esther. Indeed, in friend my attitude vis-à-vis the proposition that my friend will keep my secret looks more like a categorical belief that she will do so. This belief is probably based on my taking her to be trustworthy about such matters, which would nicely explain why I would clearly be surprised if she did not. Admittedly, faith that p does not exclude being surprised upon learning that not-p. But it seems to me that in friend my cognitive attitude vis-à-vis the truth of the relevant proposition is much more similar to the attitude I have when I believe that Roger will soon be in front of us on the platform than it is to Mary’s faith that Liverpool will win.⁵

4 What is the Positive Cognitive Attitude Involved in Faith ?

Uncontroversially, faith involves taking a positive stance on the truth of p, such that one's attitude would be incorrect if p were not the case. But if having faith that p does not involve believing that p, what exactly is the nature of this positive stance?

In the recent literature, two of the main candidates for replacing belief in this role are acceptance and assumption, with Alston defending the former option and Howard-Snyder the benefits of the latter.⁶ I shall focus in this section on their respective views, as I take them to capture two fundamental aspects of faith: (i) the fact that faith has an agential dimension that belief lacks and (ii) the fact that faith that p (generally) involves acting as if p were true while not taking p to be true. In that sense, their analyses can be seen as paving the way for the account of non-religious faith I shall advance in Sect. 7. My account will be based on what I take to be fundamentally correct in Alston’s and Howard-Snyder’s accounts while avoiding the objections I raise against them.

Let us start with acceptance. For Jonathan Cohen, acceptance consists in taking a proposition p for granted “for deciding what to do or think in a particular context” (Cohen, 1992: 4). For Alston, it consists in including p “in one’s repertoire of (supposed) facts on which one will rely in one’s theoretical and practical reasoning” (Alston, 1996: 8). This involves performing “a voluntary act of committing oneself to [p], to resolve to use [p] as a basis for one’s thought, attitude, and behaviour. (And, of course, it involves being disposed to do so as a result of this voluntary acceptance)” (ibid.: 17).

An important point here is that accepting that p is not identical to merely acting, talking, and reasoning as if it were a fact that p, because doing this is compatible with believing that not-p. Now, one is “unable to accept” or to have faith that p if one takes p “to be obviously false” (Alston, 2007: 132): Esther cannot accept or have faith that she will get the position if she believes she will not. But note that for one to accept or to have faith that p, it is insufficient that one does not believe that not-p while intentionally acting, talking, and reasoning as if it were a fact that p. If this were sufficient, one could have faith that Liverpool will win while simply not believing that Milan will win, which is intuitively insufficient for having faith that Liverpool will win. For Alston, accepting that p requires “really taking seriously the idea that [p is] true”, accepting p “as true” (Alston, 1996: 17). This means that I can accept or have faith that p only if my “evidence and arguments pro and con [p]” make it that p “seems most likely to me”, compared to not-p (Alston, 1996: 10–11). In short, “to be able to (really) accept p I must find p to be somewhat credible, enjoying some support though not enough to make it something I find myself believing” (Alston, 2007: 132).

On Alston’s view, accepting that p also requires that one inwardly assents to p’s truth. One of the main differences between Alston and Howard-Snyder when it comes to the cognitive attitude involved in faith is that Howard-Snyder rejects this condition of inner assent. This rejection follows from his view that the cognitive attitude involved in faith is not acceptance but assumption. In order to illustrate the nature of this attitude, Howard-Snyder uses this thought-experiment:

Consider an army general facing enemy forces. She needs to act. Her scouts give some information about the disposition of the enemy but not nearly enough to settle whether they are situated one way rather than several others. So she assumes that they are situated in the way that seems the least false of the credible options given the available information, say, that they are scattered throughout the boulder field on the peak. Then, acting on that assumption, she disposes her forces in the way that seems most likely to be an effective means to thwarting the enemy. (Howard-Snyder, 2019: 122)

This kind of assumption, he explains, neither consists in taking something for granted in the way in which we do “when we assume the world is more than five minutes old”—which “refers to a cognitive attitude too much like belief”) (ibid.: 123)—nor in admitting a proposition into one’s reasonings just to see what follows from it. The kind of assumption that p at play in the army general case rather consists in “tak[ing] a stand on behalf of the truth of p” (ibid.: 124) resulting from one’s taking p to be “the most likely, the most strongly indicated, or the least false of the options [one] deems credible” on whether p (Howard-Snyder, 2017b: 161). Crucially, though, the kind of assumption at play does not require that one “has settled” for oneself the question whether p, and so does not necessarily go along with a tendency to “assent inwardly” to the truth of p or a tendency to “to be surprised upon learning not-p” (Howard-Snyder, 2019: 124; 2013a: 366).

In my opinion, Alston’s and Howard-Snyder’s views face three common difficulties. I am not certain that all of them are fatal to their respective views, but they certainly put pressure on them. The first of these difficulties is that taking p to be the likeliest option on whether p does not seem to be necessary for having faith that p. Indeed, as Steven illustrates, it seems clearly possible to take oneself (or one’s team) to be the underdog in a sports match—and hence not to be surprised by one’s defeat if one is defeated—while having faith in one’s victory. Second difficulty: satisfying Alston’s or Howard-Snyder’s different conditions for acceptance and assumption, plus the conative and/or evaluative condition(s) for faith, does not seem to be sufficient for having faith. It could be for instance that two friends, Paul and Peter:

• (1) Both equally desire that a horse A wins the race and judge that A’s victory would be a good thing,

• (2) Both believe that A’s chances of winning are higher than those of the other two horses in the race (let us suppose that they both believe that the odds are 55% for horse A, 25% for horse B, and 20% for horse C), and

• (3) Both have, as a result, resolved to base their actions, discourses, and reasonings on A’s victory, and tend to do so as a result (which led them both to bet on A),

Yet Paul, unlike Peter, is constantly anxious before and during the race about A’s possible defeat, and envious of Peter’s obvious faith in A’s victory, which he kindly tries to convey to his friend by saying “you’ll see Paul, it will win!”. It does not seem that, if Paul were to “inwardly assent” to the proposition that A will win the race, it would necessarily follow that he would cease to be anxious about A’s possible defeat and would start to have faith in A’s victory.

Third difficulty: even though it is true that faith, unlike belief, seems to involve a form of commitment from the person who has faith, faith does not seem to be voluntary in the way it would be if it basically consisted in the acceptance or assumption of the truth of a proposition. For instance, it does not seem that Esther—the candidate who does not have faith that she will get the position despite taking her chances of getting it to be of the same as those of the other candidates—can decide to have faith that she will get it in the way she can decide to accept or assume the truth of a proposition p she judges to be at least as probable as not-p on the question whether p.

Note that to say that faith is not voluntary is not to deny the widely accepted idea that one’s agency is involved in one’s faith (an idea that is very much in line with the idea that faith partly consists in a commitment). To say that faith is not voluntary is rather to say that one cannot have faith at will in the way in which one can act at will (e.g., raising one’s arm, going to church and praying every morning). Here is an argument for this, in my view, very intuitive claim. Suppose that two friends, John and Mohamed are both single parents of ten years old daughters. Suppose that, sadly, their daughters both have cancer. Suppose also that John and Mohamed equally desire their own daughter to survive, and that both friends equally believe that (i) there is, statistically, a 35% chance that their own daughter will survive, and that (ii) when parents of children who have cancer have faith that their children will survive, then, for some medical reason, the chances that they will survive increase. Suppose, finally, that John says to Mohamed “I can see, I can feel, in the way you talk, in the way you think, in the way you act, that you have faith that your daughter will not die of her cancer. But I don’t. I’ve tried so hard, but I just can’t. And that’s horrible, because I know that as a result her chances of survival are not as high as they could be”. Intuitively, it would be wrong to judge John to be at fault here on the grounds that it is up to him to have faith or not, in much the same way that it is up to him to raise his arm or not. On the other hand, if it was possible to have faith at will in the sense in which it is possible to raise one’s arm at will, then responding to John by saying “What you said just doesn’t make any sense, because having faith that your daughter will survive is as up to you as raising your arm” would be harsh but correct, which is clearly not the case. What one can do at will however is cause oneself to have faith by acting in certain ways, for example, “acting as if [one] believed, taking the holy water, having masses said, etc.”, as Pascal famously argued in the case of religious faith. Another thing that one can do at will is—provided certain evidential conditions are satisfied—accept or assume that something is the case.

If neither acceptance nor assumption can be the positive cognitive attitude involved in faith, what about the idea that faith is an attitude that simply involves desiring, or judging it to be desirable, that p, plus having a degree of confidence that p is true that is inferior to the degree of confidence required for believing that p? The main problem with this idea is that it is generally admitted that one’s degree of confidence that a certain proposition p is true is, or corresponds to, one’s subjective estimation of the probability that p is true. Now, as Mary, Steven, and Esther clearly illustrate, having faith that p seems to consist in taking a positive stance on the truth of p that goes beyond one’s subjective estimation of the probability that p is true. For example, we would not expect someone who, like Mary, desires and has a certain degree of confidence (let us say 0.4) that Liverpool will win, but who, unlike Mary, does not have faith that Liverpool will win, to confidently assert that Liverpool will win, like Mary does. So, even if, as I shall argue in §5, faith that p involves the cognitive attitude of having an intermediate degree of confidence that p (to put it briefly), and a positive conative and/or evaluative attitude towards p’s obtaining, having these attitudes is not sufficient for having faith that p.

5 Is Faith Similar to Hope ?

In order to capture the nature of faith, in this section I shall explore the idea that faith is, if not identical, at least similar, to hope. This view is defended by Louis Pojman (1986), for whom religious faith is a “profound hope” that requires, as Buchak summarises, “a particularly intense desire” for the obtaining of the object of one’s religious faith (Buchak, 2017: 58). For Carl‑Johan Palmqvist, faith and hope are not identical, but faith “implies” hope—which fits with the fact that faith appears to be “as inconsistent with despair as hope is” (2022: 513).⁷ More generally, because faith and hope intuitively are (and are generally taken to be) very similar, it seems prima facie plausible that, by clarifying the nature of hope, we can clarify the nature of faith.

Let us start from this standard minimal definition of hope: “to hope is to desire an outcome we deem neither certain nor impossible” (Han-Pile, 2017: 176). In other words, hoping that p consists in desiring that p, and in neither dismissing the possibility that p nor the possibility that not-p. As Jackson writes, “hope is consistent with high credences […] but not maximally high” (as “it seems odd to hope for things in which we are certain”) and “with very low credences—as long as they are non-zero” (Jackson, 2022a: 205). For her part, Claudia Blöser rightly underlines that “rather than entailing a belief in the possibility of the object, hope […] entail[s] the absence of a belief in its impossibility” (Blöser, 2019: 209). Indeed, one can hope that p while being in doubt as to whether p really is possible, but one cannot hope (but merely wish) that p while believing p to be impossible.

One difficulty with the idea of identifying faith with hope as it has been defined is that, as Jackson writes, “the cognitive component of hope is much weaker than faith’s” (Jackson, 2023: 41). Indeed, one cannot have faith that p without a sufficiently high degree of confidence that p is the case, whereas for one to have hope that p it is sufficient that one does not believe that it is impossible that p. Thus, while “I believe that there’s almost no chance now that Liverpool will win this match, but I hope they will win” seems to make perfect sense, “I believe there’s almost no chance now that Liverpool will win this match, but I have faith that they will win” does not make that much sense. Similarly, it is conceivable that, when floating on a raft in mid-ocean without beacon or food, an individual might keep on hoping to be rescued until the very end, even when they know that the chances of rescue are becoming smaller and smaller. But it does not seem conceivable that, in this situation, they could have faith that they will be rescued until the very end.

Another problem with the idea of using the definition of hope above to elucidate the nature of faith is that this definition is in fact unsatisfactory. Indeed, it is “unable to correctly exclude all situations where hope is absent” (Palmqvist, 2021: 587). In particular, it is unable to exclude those situations in which one’s attitude is not that of hope but, on the contrary, that of despair. For instance, suppose that there are two persons on the raft, that they equally desire to be rescued, and that their estimation of the chances of being rescued is similar (and is neither 0 nor 1). It could be that one of them will hope until the end that they will be rescued, while the other will never have that hope.

Different additional conditions on hope have been advanced in the literature so as to exclude the possibility that despair satisfies the conditions for hope. I shall focus here on the idea that we can account for the difference between the subject who hopes that p and the subject who despairs that p by arguing that the former, but not the latter, is disposed to act in a way she takes to increase the chances that p obtains. The main problem with this condition is that its satisfaction does not seem to be necessary for an attitude to be one of hope, at least if it is admitted that hope can be “directed to past states of affairs or to future states of affairs for which the subject is conscious that they are not in his or her power” (Blöser, 2019: 211). Maybe this objection can be partly overcome by arguing that, when one hopes that p but believes that p’s obtaining is not within one’s power, having the disposition in question just requires that if, at some point, one took p’s obtaining to be within one’s power, then, ceteris paribus, one would act in a way one takes to increase the chances that p obtains. However, there are some cases of hope that do not involve any such disposition.

Consider for instance the case of Günther and his brother Helmut, who have both recently discovered archives that clearly indicate—though not conclusively—that their beloved grandfather Klaus was in fact an active Nazi militant. Suppose that, after this discovery, Günther keeps hoping that grandpa Klaus was not an active Nazi, whereas Helmut does not. Even if Günther were offered a time machine that could enable him, if necessary, to prevent his grandfather from becoming an active Nazi, his hope does not seem to involve any disposition to do this. Günther’s hope is rather the hope that, if he could go back to the late thirties, he would discover that it is unnecessary for him to do anything to prevent grandpa Klaus from becoming an active Nazi. Suppose that Günther, by using a time machine, discovers that his grandfather was an active Nazi from January 1936 until the end of the war. Günther’s hoping, before using the machine, that grandpa Klaus was not a Nazi would be perfectly consistent with his having, at the same time, no disposition to act (in December 1935 for example) so as to prevent his grandfather from being a Nazi. In other words, it could be that, before using the time machine and while still hoping that his grandfather was not an active Nazi, Günther is in the following state of mind: “If I discover that grandpa Klaus was a Nazi, I won’t try to prevent him from being one. He can go to hell. If it turns out that he was someone who, absent my intervention before January 1936, would then become an active Nazi, he was not the sort of person I thought he was, and definitely not a person for whom I would want to do anything”.

Another problem with the disposition to action condition is that its satisfaction, even in combination with the conative and cognitive conditions summarised in the claim that “to hope is to desire an outcome we deem neither certain nor impossible”, does not seem sufficient for there to be hope. Indeed, this disposition to do “what needs to be done” to increase the chances of something we desire obtaining sometimes leads us to do this à la désespérée only—that is, when we despair of the desired outcome obtaining. In particular, this happens in situations in which our other desires or interests do not immediately call for any action from us, so that “we might as well”, as we would ordinarily say, do what increases the chances of the desired outcome obtaining.

The attempt to account for the difference between a subject who hopes that p and a subject who despairs that p by adding a disposition to action condition for hope is however on the right track. To be satisfactory, this condition should, in my opinion, be revised as follows: when two subjects S1 and S2 both desire that p and take p to be neither certain nor impossible, but S1 hopes that p while S2 despairs that p, this is because S2 in fact categorically believes that not-p, which leads her to tend to act, talk, and reason as if it were a fact that not-p, unlike S1, who does not categorically believe that not-p.

The difference between our two shipwrecked people can then be explained by the fact that the one who hopes to be rescued does not categorically believe that they will not be, whereas her companion in misfortune does categorically believe – i.e. has no doubt – that they will not be rescued. Both of them categorically believe that their rescue is metaphysically possible, and hence that there is, in this sense, a chance that they will be rescued. But, while the chance that they will be rescued is an epistemic possibility for the one who hopes they will be rescued (since she does not categorically believe that they will not be rescued), this chance is not an epistemic possibility for the one who does not hope that they will be rescued. It is not in epistemic possibility for this second subject in the sense that the question is settled for her: she categorically believes that they will not be rescued, and takes the fact that the available evidence does not exclude their being rescued as insufficient for making the question unsettled.⁸ Her doxastic state of mind could be expressed as follows: “Yes, it’s still objectively possible that we will be rescued, but this won’t happen”. Mary’s state of mind is similar when she despairs that Liverpool will win when they are three-nil down five minutes before the end of the match: while believing that Liverpool’s victory is still objectively possible, she believes it is not something that will happen, and takes it as a fact that Milan will win.⁹ (Note that this state of mind is not incoherent: believing that p is objectively or metaphysically possible but will not happen is not believing something incoherent since, as is widely admitted, it is metaphysically possible that a metaphysically possible thing will never become actual.) Being in this state of mind, Mary will be disposed to act as if it were fact that Liverpool will not come back to win. For instance, if she believes she could avoid the tiresome traffic jam around the stadium after the match finishes by leaving it right now, she might well be strongly tempted to do so.

What then about subjects who have hope that p? Since they neither believe that p (just like those who despair that p) nor believe that not-p (unlike those who despair that p), they neither tend to act, talk, and reason as if it were the case that p nor as if it were the case that not-p. They rather tend to act, talk, and reason in a way that is compatible with the truth or falsity of p. By this I mean tending to act, talk, and reason in such a way that the rightness or wrongness—whether epistemic or practical—of so acting, talking, and reasoning does not depend on whether p is true or false. For instance, if, after the discovery of the archives, Günther still hopes that his beloved grandpa was not a Nazi, he will not tend (unlike Helmut) to burn his pictures of his grandfather, to speak badly of him, or to base his reasonings on the premise that grandpa Klaus was a Nazi. But Günther will also not tend to hang new pictures of his grandfather on the wall of his living room, to praise publicly “the deep humanity” of grandpa Klaus (which is something he used to do before the discovery of the compromising archives), or to base his reasonings on the premise that his grandfather was a just man. In short, those who merely hope that p, and hence neither believe that p nor that not-p, will tend not to engage in courses of action, discourses, and reasonings the rightness or wrongness of which depends on the truth-value of p.

6 The Nature of Faith

According to the foregoing, belief, despair, and hope can be distinguished, in part, through the different tendencies to act, talk, and reason that they respectively imply. Believing that p implies the tendency to act, talk, and reason as if it were the case that p; despairing that p implies the tendency to act, talk, and reason as if it were the case that not-p; and hoping that p implies the tendency to act, talk, and reason in such a way that the rightness or wrongness of so acting, talking, and reasoning does not depend on the truth-value of p. This way of understanding the nature of hope suggests, and makes it natural, to also analyse the intuitively similar attitude of faith in terms of such tendencies, and their relation to one’s cognitive and conative attitudes vis-à-vis the proposition in question. I would then advance the following analysis of the nature of faith (henceforth designated as faith):

For a subject S to have faith that p, it is necessary and sufficient that:

• (1) S has a positive conative and/or evaluative attitude towards p’s obtaining—typically, desiring that p.

• (2) S has a degree of confidence that p that is not sufficient for S to believe that p, i.e. to take it as fact that p (as the incompatibility argument indicates), and that can even be inferior to S’s degree of confidence that not-p (as Mary and Steven illustrate)—though not too inferior, as the fact that it would be odd to say “I believe there’s almost no chance now that Liverpool will win this match, but I have faith in their victory” illustrates.

• (3) S confidently commits herself to the truth of p in her actions, discourses, and reasonings—at least in circumstances in which S knows that, if it turned out that not-p, the negative consequences of doing so would be insignificant or negligible—which involves acting, talking, and reasoning (in these circumstances at least) as if it were a fact that p.

Some remarks about faith are in order.

First, the qualification that appears after the dash should not be read as a proviso: the condition (3) does not say that S confidently commits herself to the truth of p in her actions, discourses, and reasonings provided the negative consequences of doing so would be insignificant or negligible if it turned out that not-p. Such a proviso would clearly be overly restrictive. The meaning of the qualification included in condition (3) can then be clarified as follows:

For S to have faith that p,

• (i) it is necessary that S confidently commits herself to the truth of p in her actions, discourses, and reasonings in circumstances C in which S knows that, if it turned out that not-p, the negative consequences of doing so would be insignificant or negligible;

• (ii) but it is not necessary that S so commits herself in circumstances C’ in which more is at stake (i.e. in which the costs or risks associated with being so committed are higher).

• (iii) Supposing that S does not so commit herself in C’ (i.e. ceases in C’ to have faith that p), one should not infer from this that S does not so commit herself in C.

In Sect. 7, when discussing the rationality of non-religious faith, I shall further clarify condition (3) and the qualification it contains by contrasting two cases illustrating them.¹⁰

Second, faith fits with the different features of faith indicated in Sects. 2, 3 and 4. faith is obviously in line with the doubt argument, the surprise argument, the incompatibility argument, as well as the positive conative and/or evaluative attitude requirement. faith can also easily account for the fact that faith is more resilient than belief: because belief requires a higher degree of confidence in the truth of p than faith, more evidence that not-p will often be necessary for one to cease to have faith that p than for one to cease to believe that p. Moreover, the fact that faith that p involves a commitment to the truth of p can explain the distinctive resilience of faith when faced with practical reasons to abandon it.

faith also fits nicely with the observation, made in §3, that having faith that p seems to consist in taking a positive stance on the truth of p that goes beyond one’s subjective estimation of the probability that p is true: according to faith, having faith that p involves confidently committing oneself (in certain circumstances) to the truth of p, and hence involves acting, talking, and reasoning (in these circumstances) as if it were a fact that p even though this does not reflect one’s doxastic state about whether p. This means that, when it comes to non-religious faith, “the leap of faith” is, according to faith, a practical leap, not a doxastic leap: the idea that “faith seems to ‘go beyond the evidence’ in a way that belief doesn’t” (Jackson, 2022a: 204) must not be interpreted (at least in the case of non-religious faith) as meaning that, when we have faith that p, our degree of confidence in the truth of p goes beyond our evidence for p. Rather, “the leap of faith” consists in the fact that having faith that p involves committing oneself (in certain circumstances) to the truth of p, and hence involves acting, talking, and reasoning (in certain circumstances) in a way that goes beyond one’s degree of confidence in the truth of p.

(Interestingly, faith also fits with the common observation that there is a sense in which faith seems to be an all-or-nothing affair, something one either has or lacks, something one can either keep or lose entirely—as suggested by the fact that it sounds rather odd to say that one does not fully have faith but just has an intermediate degree of faith. Indeed, there is a sense in which one cannot more or less act, talk, and reason as if it were a fact that p; either one does these things, or one does not. But faith is also compatible with the seemingly opposite idea that one person’s faith that p can be stronger than someone else’s, and so faith can admit of degrees: one’s confidence in the truth of p and one’s positive conative and/or evaluative attitude towards p both admit of degrees. The higher these degrees, I would suggest, the more solid or resilient one’s faith—the more one will continue to confidently commit oneself in thought, word, and deed to the truth of p when evidentially challenged, or when the costs or risks associated with being so committed rise.)

Third, faith escapes the difficulties raised in §3 for Alston’s and Howard-Snyder’s accounts of faith: (1) faith is clearly not subject to the objection that having faith that p does not require taking p to be likeliest option on whether p; (2) Esther’s anxiety that not-p is incompatible with the satisfaction of condition (3) of faith, which states that faith that p requires confidently committing oneself to the truth of p; (3) according to faith, faith requires a commitment but is not voluntary in the way in which acceptance and assumption are, because one cannot confidently commit oneself to the truth of p voluntarily.

This last point may look tricky, but the idea behind it is in fact rather simple: satisfying conditions (1) and (2) of faith—e.g. by desiring that p and by having a degree of confidence of 0.4 that p is true—plus judging that one should confidently commit oneself in one’s actions, discourses, and reasonings to the truth of p, given one’s practical reasons to do so, can be sufficient for non-confidently doing so, but not for confidently doing so. Suppose for example that I am watching the match between Liverpool and Milan with my friend Mary, and that I have, like her, a middling degree of confidence that Liverpool will win, plus a strong desire that this happens, but that I do not have, unlike her, faith that they will win. Knowing that I would make her very happy by having, when watching the match, the attitude of confidently committing myself in thought, word, and deed to the truth of “Liverpool will win”—and so, according to faith, by having faith that Liverpool will win—cannot be sufficient for me to have this attitude. In other words, I cannot have the attitude of confidently committing myself in thought, word, and deed to the truth of “Liverpool will win” on the basis of my practical reason to have this attitude, though I can, on that basis, non-confidently commit myself to this in thought, word, and deed—i.e. act, talk, and reason as if it were the case that p.

What could make me have the both practical and affective attitude of confidently committing myself to the truth of “Liverpool will win” in thought, word, and deed? Typically, the causal influence of my strong desire that Liverpool will win, combined with my sufficiently high degree of confidence that they will win. So, the fact that one cannot have faith that p at will is not incompatible with faith involving a commitment. This fact rather means that the practical/affective commitment in question cannot directly result from one’s will to follow practical reasons one may possess for having this attitude, but typically results from the causal influence of one’s desire that p is true.

One last remark about faith: faith should not be understood as the claim that having faith that p is believing that p out of a desire that p. Indeed, faith does not claim that one’s degree of confidence in the truth of p gets higher due to one’s positive conative and/or evaluative attitude towards p, leading one to believe that p. Faith then is not a form of wishful thinking, where one’s doxastic state on whether p is the causal result of one’s desire that p. According to faith, having faith that p rather involves a practical/affective attitude towards p (that of confidently committing oneself in thought, word, and deed to the truth of p, in certain circumstances) that does not match—and more precisely goes beyond—one’s doxastic attitude towards p (one’s not taking it as a fact that p).

Let us now turn to the question of the rationality of faith.

7 Is Faith Intrinsically Irrational?

If faith is correct, does it follow that faith is necessarily or intrinsically irrational? If so, in what sense? Consider the following case, recently discussed by Jonathan Jenkins Ichikawa:

Suppose you are starting a business, and considering including Ranjit as a business partner. Bringing in a business partner is a momentous decision—a poor choice will result in tremendous future stress and negative financial implications. […] Because you take these considerations seriously, you investigate Ranjit thoroughly. […] After a thorough process, you’ve gathered a lot of evidence, and—let’s stipulate—that evidence rationalises proceeding with Ranjit as a business partner. We don’t have to suppose that the evidence justifies certainty that he will work out well. But let’s stipulate that you have enough evidence to justify the risk. […] Under the circumstances, you probably won’t feel sure about whether you should go into business with Ranjit. But let’s suppose you do it anyway. Good decision! [...] You’ve proceeded as the evidence dictated; you’ve done the rational thing. […I]t is easy and proper to describe this as an exercise of faith. (Ichikawa, 2020: 3)

In this scenario it seems true that, by including Ranjit as a business partner but not categorically believing that he will be a good business partner, we adequately respond to the evidence. But it also seems true that, by having faith that he will be a good business partner, we go beyond the evidence, though according to Ichikawa this does not prevent our faith from being rational. How are we to understand this?

Let us first consider whether the fact that faith is more resilient to counterevidence than belief is sufficient to make faith irrational—and, more specifically, epistemically irrational. This would be the case if faith were more resilient than belief because it is less doxastically sensitive to counterevidence. But, as I already indicated, the fact that faith is more resilient than belief can simply be explained by the fact that faith that p does not require as high a degree of confidence in the truth of p as belief does. Indeed, if so, a relatively small amount of counterevidence will often be sufficient for one’s degree of confidence in the truth of p to fall below the level required for belief, but not sufficient for it to fall below the level required for faith. So, faith’s superior resilience to counterevidence does not necessarily imply that it is epistemically irrational.

But what about its practical rationality? According to Finlay Malcolm, faith’s resilience in the face of counterevidence or practical reasons to abandon it “makes faith valuable” because it means that “faith involves a disposition to retain one’s long-term commitments to plans and projects, and […] to persist with these commitments in the face of difficulties in achieving them” (Malcolm, 2020: 137). In a more or less similar spirit, Jackson (2021) holds that, when the rationality of one’s long-term projects and commitments requires one to take it as a fact that p, having faith that p is better suited for the role than believing that p due to faith’s superior resilience.

Let us admit that this property of faith can confer on it a form of practical rationality. Isn’t it nonetheless the case that acting as if it were a fact that p, even though one does not take it as a fact that p, makes the actions in question irrational? Here I think we should distinguish two different situations. The first is a situation in which one acts as if it were a fact that p but, if it turned out that not-p, nobody would be adversely affected. The second is a situation in which one acts as if it were a fact that p and, if it turned out that not-p, somebody would be adversely affected.

Suppose for example that Mary is at the stadium with a friend during the match between Liverpool and Milan. Her having faith that Liverpool will win despite the odds (let’s say her degree of confidence that they will win is 0.4) means, according to faith, that she confidently acts, talks, and reasons as if it were a fact that Liverpool will win—e.g. she enthusiastically cheers Liverpool on, reassures her anxious friend, thinks about the next Champions League match against FC Barcelona, etc. Suppose that Mary does this while knowing that, if Liverpool were to lose the match, acting, talking, and reasoning in these ways would have no negative consequences for her or anybody else. It does not seem that her behaviour would be irrational. Similarly, in virtue of her faith in Liverpool’s victory, Mary would be ready to bet her friend a beer that Liverpool will win, or to pay more than 40 pence (let’s say, 70 pence) for a ticket that will get her £1 if Liverpool win and nothing if they lose. This is because, in both cases, she knows that nothing dramatic would follow if Liverpool were to lose: the cost of the beer, or 70 pence, is negligible for her. And it would be rather incongruous to say that taking these little bets would be an irrational thing to do. More generally, it seems that acting, talking, and reasoning as if it were a fact that p, even though this does not reflect one’s doxastic state on whether p, p is not irrational in situations where one knows that one has (almost) nothing to lose by doing so—i.e. in situations where one knows that, if p were false, the negative consequences of doing so would be insignificant or negligible.

Suppose now that Mary is offered instead the chance to pay £700 for a ticket that will get her £1000 if Liverpool win, and nothing if they lose, but that she does not take the bet because she knows that in case of defeat she will be in serious financial trouble. In such circumstances, Mary would not demonstrate faith that Liverpool will win. But this does not mean that Mary did not really have faith in Liverpool’s victory when, before being offered to pay £700 for the ticket in question, she bet a beer, or 70 pence, on Liverpool’s victory. According to faith, she was really committing herself confidently and sincerely to the truth of “Liverpool will win” when going beyond her 0.4 degree of confidence in their victory by betting a beer, or 70 pence, on it. In other words, one should not infer that Mary never had faith in Liverpool’s victory from the fact that someone else’s faith in Liverpool’s victory is more solid or resilient than hers—so resilient that it leads them to commit themselves in deed to the truth of “Liverpool will win” by paying £700 for the ticket in question, even though (just like Mary) they have just a 0.4 degree of confidence that Liverpool will win, and know they will be in serious financial trouble in case of defeat. Their faith is just more solid than Mary’s (perhaps due to her having a less intense desire that Liverpool win), which also, given the stakes, makes their betting behaviour irrational.¹¹

As Mary’s case indicates, this does not mean that the mismatch between one’s way of acting (i.e. acting as if it were a fact that p) and one’s doxastic state (i.e. not taking it as a fact that p) that is definitional of faith according to faith necessarily renders one’s way of acting irrational. It is not so in situations where one knows that, if p turned out not to be the case, no, or no significant, negative consequences would follow from acting as if it were a fact that p. What condition (3) of faith requires is that, at least in such situations, one acts, talks, and reasons as if it were fact that p despite not taking it as a fact that p.

Let us come back now to Ranjit’s case, as described by Ichikawa:

• (1) We adequately respond to the evidence by not categorically believing that Ranjit will be a good business partner (henceforth “GBP”),

• (2) But we also go beyond the evidence by having faith that he will be a GBP—and so, according to faith, by (in certain circumstances) acting, talking, and reasoning as if it were a fact that he will be a GBP —,

• (3) However, this does not prevent our faith from being rational.

If faith is correct, for this description of the case to be coherent our practical attitude vis-à-vis the proposition that Ranjit will be a GBP should be as follows: in circumstances where we know that, if Ranjit turned out not to be a GBP, we would have much to lose by acting, talking, and reasoning as if it were a fact that he will be a GBP (e.g. when we have to decide, at the bank, how much money to invest in the business activities in which he will be involved), we act, talk, and reason in a way that is in line with our degree of confidence that he will be a GBP—let’s say, 0.75. But, in circumstances in which we know that, if he turned out not to be a GBP, we would have nothing or not much to lose by acting, talking, and reasoning as if it were a fact that he will be a GBP (e.g. when interacting with him in a non-professional context, when talking about him to our friends, or when thinking of plans for our next holiday on the premise that we will make good money together), we act, talk, and reason as if it were a fact that he will be a GBP, and not just as if our degree of confidence that he will be a GBP is 0.75. Since having faith that Ranjit will be a GBP is compatible with having this practical attitude vis-à-vis the proposition that Ranjit will be a GBP, and since none of the ways of acting, talking, and reasoning just mentioned would be, in themselves, irrational, it could be argued that faith is neither intrinsically epistemically irrational nor intrinsically practically irrational.

There is however an obvious sense in which faith is intrinsically irrational: since having faith that p involves having a doxastic attitude towards p (not taking it as a fact that p) that does not match with the practical/affective attitude that having faith that p involves (confidently committing oneself in thought, word, and deed to the truth of p in situations where there would not be much to lose if it turned out that not-p), faith is an intrinsically inconsistent propositional attitude. And if one cannot have faith without being, more or less consciously, aware of this inconsistency, it could be argued that faith never comes without a soupçon of bad faith.¹²