Introduction

The view that religious faith need not be based on belief but can also be non-doxastic is currently on the rise (Alston, 1996; Audi, 2011; Buchak, 2017; Howard-Snyder, 2013; McKaughan, 2016; Schellenberg, 2005). I am largely sympathetic to non-doxasticism concerning religion,Footnote 1 which I regard as an eminently reasonable stance given our current epistemic situation concerning religious matters.Footnote 2 Also, I do find the arguments in favour of the possibility of non-doxastic faith convincing.Footnote 3 However, when it comes to spelling out the details of non-doxastic faith, I think that much work remains to be done. The aim of this article is to formulate some desiderata for any future account of rational, non-doxastic faith. In the process, I hope to pinpoint some of the major weaknesses of current accounts.

In the literature, there is an unfortunate tendency to treat faith in isolation from the closely related notions of hope and trust. Two of my three desiderata concern the connection between faith on one side and trust and hope on the other. These desiderata build heavily on current philosophical literature on trust and hope respectively. Thereby, I hope to take some steps toward amending the common oversight.

The first desideratum I will suggest is that faith must have the ability to meet the demands of epistemic rationality. While not denying that there are many irrational instances of faith, as a rational attitude faith is of philosophical and epistemological interest. Perhaps because non-doxasticism implies a lack of belief, there is a tendency to underestimate what epistemic rationality requires (Buchak, 2012; McKaughan, 2016). Building on John L. Schellenberg’s (2009) account of the demands of reason, I show how this underestimation ultimately leads to a failure to meet a plausible standard of rationality. I substantiate this claim by showing how this is true even of a philosopher like Lara Buchak (2012, 2017), whose explicit interest is to show how faith can be rational.

The second desideratum is about faith and trust. Faith is generally thought to imply trust, which has led some philosophers (Audi, 2011; McKaughan, 2016) to give accounts of non-doxastic faith in terms of trust. Turning to the literature on trust (Holton, 1994; McGeer, 2008; Simpson, 2018), this seems to be a mistake. Much like in the literature on faith, there is a major debate whether trust should be considered doxastic or non-doxastic. The problem is that non-doxastic trust is standardly considered irrational and it is therefore hard to see how rational, non-doxastic faith could be understood in terms of it.

The third desideratum is about how faith relates to hope and despair. Hope is important for faith since, according to a widespread intuition, faith implies hope. Inspired by the literature on hope (Bovens, 1999; Martin, 2014; Meirav, 2009), I suggest that a suitable test for any notion of faith is that it should imply hope, while being able to exclude despair. Important accounts of non-doxastic faith (Buchak, 2017; Howard-Snyder, 2017) seem unable to do this.

The concept of faith implied by these desiderata differs in some respects from the everyday notion, and I suggest that to reach a stable notion of rational, non-doxastic faith, we need to employ a reconstructive methodology.

This introductory section is followed by the "On Analysing Rational, Non-doxastic Faith" section, a brief presentation of non-doxastic faith and my approach to the subject. In the "The Demands of Epistemic Rationality", "Faith and Trust" and "Faith, Hope and Despair" sections, the three desiderata for an analysis of rational, non-doxastic faith are presented in detail along with their implications for important existing accounts. In the "Is It Really Faith I am Looking for?" section, I consider what kind of faith account these desiderata suggest, with the focus on their methodological implications. The "Concluding Remarks" section offers some concluding remarks.

On Analysing Rational, Non-doxastic Faith

The word ‘faith’ can be used in a grand variety of ways. The following observation from Jonathan L. Kvanvig illustrates the phenomenon well:

We talk about faith and use the language of faith in a multitude of ways: faith in humanity, being a person of faith, having faith that the economy won’t collapse, the catholic faith, etc. Faith is sometimes used as a synonym of trust, sometimes of belief. Such diversity prompts methodological reflection: is systematic treatment even possible here. (Kvanvig, 2018, 1)

Since such broad use of a word means that one can almost always find counterexamples to every proposed definition, many philosophers have come to reject the idea of providing a conceptual analysis covering all instances (Kvanvig, 2016, 8–10; 2018, 1–8; Whitaker, 2019, 155). Instead, Kvanvig has suggested that we should concentrate on what is most fundamental and primary about faith (Kvanvig, 2016, 6) or alternatively on what it is that makes faith valuable (Kvanvig, 2018, 1–4).Footnote 4

While Kvanvig sets out to find what is most fundamental or valuable about faith, I have a somewhat different approach. I am interested in a particular kind of faith, meeting some specific requirements. While I do think the kind of faith I am interested in holds great value, I would not deny the possibility that there might be some other form which is valuable in some other way, perhaps even more so than the kind I am after (if there is salvific faith, this would surely be the case). It might also be that there are more fundamental kinds of faith.

My interest lies in faith that is both non-doxastic and rational. Philosophers working with non-doxastic faith standardly accept that there is doxastic faith as well (Alston, 1996; Audi, 2011; Howard-Snyder, 2013), and this is not the place to question that assumption. I also take it to be trivial that faith is sometimes held in an irrational manner. In short, there are many instances of faith not covered by these specific requirements.

The core idea of non-doxasticism is that one might well make a religious commitment on some cognitive attitude weaker than belief. Most often, this attitude is spelled out in terms of non-doxastic faith (Audi, 2011; Buchak, 2017; Howard-Snyder, 2013; McKaughan, 2016; Schellenberg, 2005). On other accounts, it is rather hope (Muyskens, 1979; Palmqvist, 2019; Pojman, 1986)Footnote 5 that plays the lead role. However, there seems to be a common core to most accounts, which I have summarised in terms of two necessary conditions:

We can say that for S to have a non-doxastic pro-attitude towards some proposition p, the following two conditions must be met:

ND1: S desires that p or judges that p is an overall good thing.

ND2: S believes p to be epistemically possible and neither believes nor disbelieves p. (Palmqvist, 2021, 52)

In what follows, I will take it for granted that ND1 and ND2 apply to all accounts of non-doxastic faith. ND1 is about motivation. Since a non-doxastic commitment is commonly thought to be entirely voluntary, some motivation is needed if the subject is to take it on. In contrast to a believing subject, the non-doxasticist needs to desire p or at least think it is an overall good thing.

ND2 is about epistemic possibility. It delimits non-doxasticism not only against belief-based religion, but also against non-realist approaches such as religious fictionalism. Non-doxasticism is about relating to a religious view that, for all the subject knows, just might turn out to be true. While some philosophers hold that all that is needed is that the subject neither knows nor disbelieves p (Schellenberg, 2009, 8), others demand that p must have a non-negligible probability of being true (McKaughan, 2013, 113).

It should also be noted that on the standard account, hope is analysed as being constituted by ND1 and ND2. Therefore, hope must be considered the smallest common denominator of non-doxasticism. In other words, faith implies hope, and if you have any non-doxastic pro-attitude at all towards p, you also hope that p.

I concentrate on non-doxastic faith because the notion holds great promise as a previously neglected way of being religious, a way wholly compatible with being an intellectually honest agnostic.Footnote 6 Schellenberg (2007) has claimed that the proper response towards religious claims given our current epistemic situation is agnosticism and that we all should be agnostics. While I think there is much truth to that claim, it is enough for present purposes to note that agnosticism is indeed where a great many epistemically sincere people tend to end up after making a thorough religious investigation. And while traditional belief-based faith is surely unavailable to the agnostic, for non-doxastic faith agnosticism is a requirement. As Schellenberg (2009, 3–5) has noted, if one pursues non-doxastic faith, agnosticism is not the end but the beginning of one’s religious life.

My concern with the connection between non-doxastic faith and agnosticism also explains why my interest is limited to epistemically rational faith. Since it is rational considerations that have led the intellectually sincere agnostic to assume her position, it seems natural to suppose that only a faith which can be had without interfering with one’s epistemic endeavours could be relevant for her. To be sure, ‘epistemic rationality’ can mean a great many things, which is why my first desideratum is dedicated to spelling out these requirements of epistemic reason in detail (see the "The Demands of Epistemic Rationality" section).

Since I am interested in non-doxasticism in relation to religion, should I not also delimit my claims to religious rational, non-doxastic faith? I find no reason to do so, since I do not think that the attitudes employed in religion are in any way special. Religious hope and faith are but instances of the same attitudes one might express in a secular, everyday setting. Therefore, if non-doxastic faith can be rational in a religious setting, there is no reason to suppose that the same attitude cannot be reasonably entertained in other circumstances as well.

The desiderata I suggest are intended for any account of non-doxastic faith coming with claims of epistemic rationality. In the literature, it is common to distinguish between different kinds of faith, most often between faith-that or propositional faith, on one hand, and something like faith-in, operational faith or relational faith one the other (Howard-Snyder, 2013, 358; Schellenberg, 2005, 126). While I mostly have propositional faith in mind, the desiderata are supposed to be equally relevant for any account of non-doxastic faith for which epistemic rationality is claimed.

The Demands of Epistemic Rationality

The first desideratum for rational non-doxastic faith is about adherence to the demands of rationality.Footnote 7 At face value, it might seem like a trivial demand, since surely everyone would agree that any account of faith worthy of the label ‘rational’ must meet the demands of reason. However, it seems to me that several philosophers underestimate the relevance of such demands for non-doxastic faith, settling for some over-relaxed standards of epistemic rationality and severely downplaying its importance on non-doxasticism. The assumption seems to be that since epistemic rationality is primarily about belief-regulation; it is not a very relevant notion for non-doxasticism. This is especially true for authors working with an action-centred notion of faith, like Lara Buchak or Daniel McKaughan:

One can clearly meet the requirements of epistemic rationality, as I’ve stated them, while having faith – whether one has faith is completely separate from whether one is epistemically rational because it is separate from whether one has appropriately evaluated the evidence one has. (Buchak, 2012, 237, my italics)

Epistemic opinion (assessment with respect to considerations that bear on likely truth or falsehood) and action are subject to very different criteria of appraisal. Epistemic rationality, on this view, is solely a matter of finding yourself with a credence level or subjective probability judgement that fits the evidence… Femia’s [McKaughan’s recurring exemplary subject] action-centred faith, as characterized in the previous section, leaves her entirely free to follow the arguments and the evidence… there is nothing inherent in action-centred faith that leads her into epistemic irrationality. (McKaughan, 2016, 84–85, my italics)

Buchak and McKaughan are surely right when they claim that epistemic considerations of evidential probability have little bearing on the rationality of non-doxastic faith (except, of course, to establish that one’s object of faith is truly an epistemic possibility in the relevant sense). However, the demands of reason are more extensive than they assume. It is not enough to demonstrate faith’s independence from considerations of evidence; one must also demonstrate that one can have faith without it interfering with one’s epistemic endeavours. There must be no conflict between faith and inquiry. One philosopher working with this deeper understanding of the demands of reason is SchellenbergFootnote 8:

What many will consider the strongest challenge to faith claims that the aim of staying true to reason should all things considered be pursued… and that this aim can only or best be pursued by not having faith. (Schellenberg, 2009, 75)

How exactly is faith supposed to conflict with staying true to reason? Schellenberg elaborates the point in the following passage:

It may be said that in faith one must portray the world to oneself, and act, in a manner quite at odds with what the intellect has revealed… the person of faith emphasizes only the possibility of truth, setting aside the possibility of falsehood. (Schellenberg, 2009, 76, italics in original)

Another way of putting this is that faith that p might make the subject insensitive to the evidence, especially evidence pointing to the falsehood of p. As I have pointed out in an earlier publication, if faith that p is inconsistent with conducting an inquiry whether p, and if one makes a faith-commitment, one must postpone further inquiry indefinitely. Such a postponement is deeply problematic on non-doxasticism, since the truth-value of p is always unknown, and at the same time something we standardly deeply care about. Postponing further investigation indefinitely on such a matter is surely a great price to pay, and the resulting faith would be clearly at odds with the demands of reason, being in this sense irrational (Palmqvist 2019, 323).

Note that this understanding of the demands of reason does not require any advanced notion of epistemic rationality. The basic notion employed by Buchak and McKaughan, where epistemic rationality only requires that one’s beliefs and epistemic attitudes are properly sensitive to any relevant evidence, is sufficient. The problem with Buchak’s and McKaughan’s accounts is that they fail to see the latent conflict between this notion of epistemic rationality and some forms of non-doxastic faith.

As an illustration of how this conflict might play out, consider how everyday faith might prevent inquiry. Bruce, who works in a small company in crisis, has faith that the company will pull through. Having faith, Bruce ignores any signals that the company is close to bankruptcy and refrains from investigating the matter further. When the evidence is too strong to ignore, like when he overhears his boss crying alone in his office, Bruce reinterprets what he experiences in light of his faith, focusing on the possibility that his boss cries over something other than the state of the company. Had Bruce not had faith, he would soon have discovered that the company was done for. However, with faith preventing any serious investigation, he never learnt that. Clearly, Bruce has not done his best from an epistemic point of view. He has failed to meet the demands of reason; he has been irrational due to having faith.

I am not saying that faith always works this way. If it did, rational non-doxastic faith would be impossible. I am only saying that when faith works this way, it is epistemically problematic, and, more importantly, faith does seem to work this way given some of our present-day accounts of non-doxasticism. Any account of rational, non-doxastic faith must do things differently and offer an analysis of faith which does not come with these consequences. Thus, my desideratum of epistemic rationality runs as follows.

Desideratum of epistemic rationality (DR): Any account of rational, non-doxastic faith must be able to meet the demands of rationality so that there is no conflict between the subject having faith and her engaging in rational inquiry.

Several present accounts of non-doxastic faith fail to meet DR. It is beyond the scope of this paper to address all such accounts, so I will focus my attention on the well-known non-doxastic faith advanced by Buchak (2012, 2017). According to Buchak, the hallmark of faith is that one does not look for further evidence and that one likewise remains ‘steadfast in the face of counter-evidence’. This (along with the fact that there must be some risk involved) is the defining feature of faith according to Buchak:

A man who has faith that his wife is true to her vows commits to his marriage without reading her private correspondence. An individual who has faith that a particular bridge will hold his weight doesn’t test the bridge before stepping onto it. Not only do individuals with faith not need further evidence, they will choose not to obtain it if it is offered to them. (Buchak, 2017, 114)

Given that Buchak considers epistemic rationality largely irrelevant for faith, it is perhaps unsurprising that her account fails to meet DR. When it comes to the kind of rationality she does find relevant, instrumental or practical rationality, she goes to considerable length to demonstrate how faith can meet its demands (Buchak, 2012, 235–246). I have no intentions to contend her conclusions in that regard. A faith which fails to meet our epistemic standards could still be beneficial for the individual.

Buchak’s considerations regarding practical rationality lead her to conclusions softening her claims regarding faith and evidence. While according to Buchak faith is always a matter of ignoring new evidence, she does not claim that such faith is always practically rational. Two further conditions must be met. The subject needs to have a high degree of confidence in p before stopping the search for evidence, and the evidence already gathered should suggest that no potential piece of evidence would tell conclusively against p (Buchak, 2012, 246).

These qualifications will have an impact on epistemic rationality as well. The requirement that the subject needs to start with high confidence sets Buchak apart from other writers on non-doxastic faith, and it clearly seems less irrational to stop looking for evidence for something you believe than something you are agnostic about. However, consider what will happen in the long run. Even if we grant that it might be epistemically permissible to stop looking for evidence occasionally, to make more room for action, it seems to me that as a long-term commitment, Buchakian faith will become increasingly problematic the longer it lasts. After having had faith for a long time, you will have no idea how the evidential situation regarding p has developed during your commitment, why you will also be ignorant whether holding on to p is rationally sustainable or not. Clearly, holding on to p under such conditions violates DR.

The second condition does even less to make Buchakian faith epistemically rational. That one’s body of evidence may suggest that no potential piece of evidence is missing does not, in any way, establish that this is the case. Significant unknown evidence can often shatter what one thought was the whole picture, forcing the subject to rethink her conclusions when it appears. As long as the possibility of significant unknown evidence remains, a faith which requires an indefinite postponement of further investigation will not be able to meet the demands of reason.

At this point, it might be worth pointing out that an indefinite postponement of further investigation for the sake of faith can be rational in some extraordinary circumstances.Footnote 9 I am referring to the situations described by William James, where the truth of p ‘cannot by its nature be decided on intellectual grounds’ (James, 2014, 13). However, such situations of principled evidential ambiguity are rare indeed, and it is notoriously difficult to establish beyond doubt that the conditions James mentions obtain. Principled ambiguity can never be assumed without substantial argument, since a lack of progress in settling the truth-value of p can be due no numerous other reasons.Footnote 10

To postpone investigation indefinitely, we need compelling evidence that a situation of principled ambiguity obtains. Buchak’s requirement that the evidence should ‘suggest’ that no further evidence is missing is not nearly strong enough. While, of course, Buchak could rework the second condition and restrict her account to situations of principled ambiguity, she would presumably not be interested in doing so. Buchak aims to provide a general account of faith, and situations of principled evidential ambiguity are simply too rare for it to be reasonable to suggest that one must be in one in order to have faith.

Since Buchak holds that not being sensitive to new evidence is a defining feature of faith (Buchak, 2012, 232–233), it is hard to see how her account could be developed to accommodate for DR. It is conceivable that she might contest DR, but that would only reveal that she is not interested in the same kind of rational, non-doxastic faith my desiderata aim at. If she is not, it would be better to acknowledge this from the beginning, by claiming that her account concerns a specific kind of non-doxastic faith, a practically rational kind one can take on in situations where action is more important than epistemic concerns.Footnote 11

Faith and Trust

It is sometimes said that faith implies trust. True to this intuition, several philosophers working on non-doxasticism maintain that some important kinds of faith are to be analysed as containing trust as a core element (Audi, 2011; McKaughan, 2016). Other philosophers, however, reject such an analysis (Kvanvig, 2016). While a close connection between faith and trust might seem intuitive, I will argue that this intuition must be abandoned if we want non-doxastic faith to be an epistemically rational attitude.

In recent philosophical literature on trust, there is a debate between doxasticists (Fricker, 1995; Keren, 2014; Simpson, 2018) and non-doxasticists (Baier, 1986; Holton, 1994; McGeer, 2008) which echoes the debate over whether faith requires belief. At surface level, the two dominant views in this debate seem to fit well with corresponding views on faith. If one is a non-doxasticist regarding faith, surely one should also be a non-doxasticist regarding trust, keeping the intuition that faith implies trust intact. However, there is a major difference between how non-doxastic faith and non-doxastic trust are standardly conceived. As far as I am aware, no proponent of non-doxastic trust claims that this is an attitude that meets the demands of epistemic reason. Non-doxastic trust can no doubt be practically rational, but as will be explained below, it is exceedingly hard to interpret as epistemically rational as well. Thomas W. Simpson, proponent of doxastic trust, claims that the major point of disagreement between the two camps (which he calls cognitivist and non-cognitivist) is whether trust is primarily a matter of epistemic or practical rationality:

The division between cognitivists and non-cognitivists is well understood as a disagreement about which of the following intuitions is basic.

Interpersonal. Trust is an attitude embedded in and expressive of interpersonal relationships. For instance, to trust someone is to respect them, and to distrust is to slight. It is properly responsive to the reasons I have for respecting another person. These reasons are practical.

Rationality. Trust is vindicated when I have trusted the trustworthy and not trusted the untrustworthy. It is properly responsive to the reasons I have for believing the trusted to be trustworthy. These reasons are theoretical.

(Simpson, 2018, 447, italics in original)

It is outside the scope of this paper to take a stand on the correct account of trust, so I will somewhat pragmatically assume that both non-doxastic and doxastic trust are viable attitudes, as most think is the case when it comes to faith. I only deny that non-doxastic trust can be epistemically rational.

Philosophers who claim that trust is an epistemically rational attitude standardly take this to require that trust is properly based upon belief. In other words, they favour a doxastic account of trust. I find no reason to reject this conclusion, especially since trust is standardly taken to imply reliance. It seems to me that if I am to trust another person to do something for me, that trust would be irrational if I did not believe that person to be reliable and capable. A bit more formally, it seems reasonable to suggest that if A trusts B to ɸ, for A to be epistemically rational, A must believe that B is reliable and that it is within B’s capacity to deliver ɸ. Without such beliefs, or some similar beliefs to the same effect, how could trust be considered a rational attitude? If I trust you with my car, even though I know you are a terrible driver and will presumably wreck it somehow, I cannot rationally expect that I will get the car back in one piece. But if I do not expect that outcome, I cannot be said to trust you. If I do not believe I will get my car back, any expectation of getting the car back must be irrational. Therefore, non-doxastic trust must be conceived of as an irrational attitude as well.

At this point, an objector inspired by James might ask if I have not forgotten the possibility of there being faith (and trust) contingent truths—‘cases where a fact cannot come at all unless a preliminary faith exists in its coming’ (James, 2014, 19). If we return to the example with the car, can it not be the case that even if you are a terrible driver, you will never get good at driving unless I have faith in you and trust you with my car even though I do not believe I will get it back undamaged?Footnote 12 I do not contest that trust might be practically rational in this way, but we must not conflate this with epistemic rationality. This kind of trust is epistemically irrational unless we have robust evidence that it will bring about the desired fact. It only becomes rational if we have compelling evidence that it will indeed bring the fact about, but in such case, trust will be doxastic rather than non-doxastic. The conclusion that non-doxastic trust is always epistemically irrational is hard to avoid.

There is also a further reason non-doxastic trust must be considered epistemically irrational. In arguments paralleling Buchak’s view on non-doxastic faith, proponents of non-doxastic trust standardly advance the idea that trusting p is incompatible with inquiring whether p, i.e. with searching for evidence regarding the truth or reliability of p (McGeer, 2008, 240). The examples advanced are also similar to Buchak’s concerning faith, like that it is hard to maintain that you trust your husband to keep his marital vows while at the same time searching his private email for signs of infidelity. To really trust him, you must also disregard evidence that could be interpreted against him, like phone calls from unknown females. However, if this kind of trust is not based upon belief, it is surely irrational epistemic behaviour in violation of the considerations leading to DR.

With these considerations in mind, it seems straightforward to suggest that rational, non-doxastic faith cannot be analysed in terms of trust. It is easy to see that non-doxastic faith cannot be combined with doxastic trust. Non-doxastic faith and doxastic trust concerning the same proposition are mutually exclusive attitudes since doxastic trust requires belief and non-doxastic faith is inconsistent with belief. This means that the rational form of trust is unavailable to the non-doxastic.

Both ways in which non-doxastic trust must be considered irrational prevent it from figuring in an account of rational, non-doxastic faith. First, consider that non-doxastic trust means that the subject will have to expect outcome ɸ even though she does not believe that ɸ will obtain. She is expecting ɸ even though she lacks a sufficient evidential ground. One of the points with non-doxastic faith is that it is not supposed to be reducible to believing on insufficient evidence. But if you do not believe on insufficient evidence, you cannot very well expect anything on insufficient evidence either. The subject of rational, non-doxastic faith cannot expect ɸ and remain rational, since that would amount to expecting ɸ on insufficient evidence. Therefore, rational, non-doxastic faith is inconsistent with non-doxastic trust.

Secondly, that non-doxastic trust that p implies that one should end one’s inquiry and ignore evidence regarding p is another source of inconsistency between the two attitudes. Since DR for rational faith demands the opposite stance, namely that one must not ignore the evidence regarding p, the attitudes seems to be mutually exclusive, since one cannot both ignore and respect the evidence at the same time.

Based on these considerations, we get the following argument to the effect that rational, non-doxastic faith cannot be understood in terms of trust:

  • Premise 1: Rational non-doxastic faith is by definition an epistemically rational attitude (as explicated in DR) incompatible with belief (according to ND2).

  • Premise 2: Trust is either doxastic or non-doxastic.

  • Premise 3: Doxastic trust requires belief, and it is therefore incompatible with rational, non-doxastic faith.

  • Premise 4: Non-doxastic trust is not an epistemically rational attitude, and it is therefore incompatible with rational, non-doxastic faith.

  • Conclusion: Rational, non-doxastic faith is incompatible with trust.

Reformulating the conclusion into a second desideratum, we get the following.

Desideratum of faith and trust (DT): Any account of rational, non-doxastic faith must not require the attitude of trust since this attitude is incompatible with such faith.Footnote 13

Like DR, DT is at odds with several accounts of non-doxastic faith. Once again, I will focus on showing how one of the more prominent account fails to meet the desideratum, leaving the implications for other accounts to the reader. I will focus on the faith-analysis presented by McKaughan in ‘action-centred faith, doubt and rationality’. McKaughan’s account of non-doxastic faith is based on trust in a testimonial source:

Femia finds herself confronted with a testimonial source (TS) that proclaims some religious teaching (K) and invites response (follow a religious way, W)… Femia has significant doubt about the reliability of the testimonial source with respect to central parts of the core teachings. She sees no convincing non-question begging reasons for the reliability or trustworthiness of the source, through this remans a live possibility. Suppose then that Femia responds in two ways: she trusts (or relies on) the testimonial source and she follows the Way…I claim that Femia can do all this without either first person epistemic or practical irrationality. (McKaughan, 2016, 73–74, italics in original)

McKaughan never gives any explicit account of trust. However, since he explicitly claims that one can trust p without believing p to be reliable, it seems safe to assume that he has a non-doxastic notion in mind. He also claims that such trust is not epistemically irrational. Given the considerations above, this is a claim we have good reasons to reject.

First, as we saw in the previous section, McKaughan understands epistemic rationality in an overly narrow way, without regard to the wider epistemic considerations that led us to DR. For ease of reference, here is McKaughan’s view again:

Epistemic rationality, on this view, is solely a matter of finding yourself with a credence level or subjective probability judgement that fits the evidence… Femia’s action-centred faith, as characterized in the previous section, leaves her entirely free to follow the arguments and the evidence. (McKaughan, 2016, 84–85)

Since trust is generally conceived as requiring an end to inquiry, and since a trusting subject does not take any potential further evidence into consideration (at least not counter evidence), it seems obvious that the subject is not as free to follow the evidence as McKaughan supposes. To the contrary, trust is supposed to prevent such investigations, a fact about trust not accounted for by McKaughan.

Secondly, and as stated above, it is hard to maintain that you trust someone to ɸ if you do not expect ɸ to happen. The same holds for testimonial trust, but instead of expecting an outcome, you must believe what the testimonial source tells you. McKaughan seems to claim that non-doxastic trust does not lead to belief-formation, but how can you be said to trust a source if you do not believe what it says? Contrary to McKaughan, also non-doxastic trust in testimony is supposed to lead to belief, and such belief will not be epistemically rational, since it will be based on trust in a source not believed to be reliable (Holton, 1994).

These considerations leave McKaughan with at least two options. Either he can revise his account to give the right characterization of non-doxastic testimonial trust, so trust ends inquiry and implies belief. Or he can abandon trust and replace it with some other attitude which is easier to combine with the demands of rationality, like acceptance (basically understood as the voluntary act to think and act as if p is true, where p is an epistemic possibility).Footnote 14 Since McKaughan is keen on showing that non-doxastic faith can be both rational and belief-less, and non-doxastic trust implies the very opposite, he would presumably reject the first option. The second option seems more in line with McKaughan’s project. Even though McKaughan talks about trust, his account fails to capture key features of the attitude. On a charitable interpretation, this might suggest that the attitude in question is only labelled trust in lack of a better name. Be that as it may, McKaughan could easily avoid the problems of non-doxastic trust by replacing it with acceptance. Doing so would satisfy both DT and DR.

Faith, Hope and Despair

As mentioned briefly in the "On Analysing Rational, Non-doxastic Faith" section, according to the standard account, hope is analysed in terms of desire and epistemic possibility. Since these conditions are also the smallest, common denominator of non-doxasticism (as captured in ND1 and ND2), hope should be considered the most basic non-doxastic pro-attitude, which is why all accounts of non-doxastic faith should be taken to imply hope.Footnote 15

In recent literature on hope, the standard account has come under fire because it seems unable to rule out what should be considered the opposite of hope, namely despair. There are several well-discussed examples in the literature, like the two prisoners Andy and Red from the movie Shawshank redemption:

Suppose Andy and Red have been discussing the chances of attaining freedom, reviewing all options of escape, early release on parole, general amnesty, etc. Suppose they agree on all the details, that is, what conditions are on which each of these options depends and how likely these conditions are. Perhaps Red summarizes the picture by saying that the chances are one in a thousand. And Andy does not disagree…Red will say, ‘I grant you it is possible, but the chance is only one in a thousand!’ whereas Andy will say ‘I grant you the chance is only one in a thousand, but it is possible!’ The whole of the contrast between hoping and despairing seems to be expressed in the subtle difference between the intonations of the word possible in these two sentences. (Meirav, 2009, 223, italics in original)

In examples like this, we have two subjects assigning the same probability to p and desiring p with equal strength, yet one hopes and the other despairs. If such a case is indeed possible, something must be wrong with the standard analysis of hope. Many philosophers are now looking for some third condition, which combined with desire and epistemic possibility is sufficient for hope (Bovens, 1999; Calhoun, 2018; Kwong, 2018; Martin, 2014; Meirav, 2009; Palmqvist, 2020).

This development in the philosophy of hope is relevant for non-doxastic faith in two major ways. First, according to a common intuition, faith implies hope. Hope is undeniably a non-doxastic attitude, perfectly consistent with a lack of belief. This is problematic for a doxastic understanding of faith, since the belief-component of faith will in many cases be inconsistent with hope.Footnote 16 On non-doxasticism, the intuition that faith implies hope can be easily preserved, which constitutes a strong reason to prefer such an account over a doxastic. Therefore, any philosopher working on non-doxastic faith should pay attention to what happens in the philosophy of hope so that the intuition is preserved and the resulting account of non-doxastic faith is compatible with at least some major account of hope.

Secondly, the despair-problem has hitherto been overlooked in the literature on faith. This is a grave oversight, since surely faith must be at least as inconsistent with despair as hope is. If you have faith that p, you cannot very well despair over p, at least not at the same time. Consider the parent to the lost toddler. If the parent has faith that the toddler will be found alive and well, she will not at the same time despair over the prospects of finding the toddler alive and well (she might, however, be ‘torn between faith and despair’ and switch between the stances).

These considerations can be reworked into a third desideratum:

Desideratum of hope and despair (DH): Any account of rational, non-doxastic faith should imply hope and exclude despair.

Since there is nothing approaching a new consensus in the debate over the proper analysis of hope, for now we might be content if our account of faith at least implies the standard account of hope.

The demand that rational, non-doxastic faith should exclude despair is more pressing. It is especially of concern for accounts of faith where the cognitive side of faith is underdefined or intentionally left vague, as in more action-oriented analyses. Once again, Buchak’s account is among the more problematic ones. Consider her definition of faith (from her 2017 paper where the idea that faith requires a postponement of investigation is elaborated so that faith also requires risk-taking):

Act A is an act of faith that X, for S, if and only if:

(i) S performing A constitutes S taking a (subjective) risk on X

(ii) S chooses (to commit to A before he examines additional evidence) rather than (to postpone his decision about A until he examines additional evidence). (Buchak, 2017, 115)

Now, even if one also have ND1 and ND2 in mind (which one should since Buchak (2017, 114) explicitly thinks these two conditions must hold), the cognitive side of faith remains curiously underdefined, and there is nothing in this account that stops a subject that fulfils its conditions from despairing. Consider, for example, a dutiful general fighting a superior enemy invading his home country. The general desires victory, and even though the enemy is much stronger, he believes it possible but not very likely that he might outwit the enemy by using the terrain to his advantage. It is an epistemic possibility, but the probability is certainly not high enough to warrant belief. Doing his duty to his country, the general produces a cunning plan and launches a risky sneak-attack on the enemy forces, a decision he takes instead of gathering more evidence concerning enemy strategy and movement. The general satisfies all the demands of ND1 and ND2 and Buchak’s (i) and (ii). However, it seems clear to me that the general might well be a defeatist, who fights for the sake of duty while he despairs over the outcome of the war and lacks any faith that his sneak-attack will turn the tide.

DH is also somewhat problematic for Daniel Howard-Snyder’s well-known account of non-doxastic propositional faith, which runs as follows:

For you to have faith that p, for some proposition p, is (a) for you to have a positive cognitive attitude towards p, (b) for you to have a positive conative orientation toward the truth of p, (c) for you to be disposed to live in light of that attitude and orientation, and (d) for you to be resilient in the face of challenges to living in that way. (Howard-Snyder, 2017, 57)Footnote 17

Howard-Snyder’s account is broad and intended to capture all instances of propositional faith. He holds that faith is something that can be instantiated in a lot of diverse ways, with different attitudes playing the key roles in (a) and (b). According to Howard-Snyder, (a) can be satisfied by cognitive stances as different as belief, acceptance and mere assuming (Howard-Snyder, 2013, 367). In this way, the account is close to being an ‘umbrella-account’, acknowledging many of the attitudes different philosophers have suggested to be the bearing attitude of faith as valid instances of faith’s cognitive side.

The strength of Howard-Snyder’s account is that it is wide enough to incorporate all the different stances that can reasonably be labelled ‘faith’. While I have delimited my interest to a specific kind of faith, it seems reasonable to suppose that if a unified account of faith is at all possible, it must be developed along such permitting lines.

However, an account this wide will also have problems with exclusion.Footnote 18 If, in spelling out the details, we let some sufficiently weak positive cognitive attitude play the role of (a), we will get non-doxastic faith that is consistent with despair. For example, if we go for acceptance, we can see that the defeatist general who satisfies the requirements of faith on Buchak’s view can also be an example of the faith according to Howard-Snyder.Footnote 19

The defeatist general satisfies (a) since he accepts that ‘we will win the war’ as a basis for his life and action. He also satisfies (b) since he desires victory and (c) since he lives in the light of ‘we will win the war’, constantly making sophisticated battle plans to that end. The fact that the general does not surrender but continues to fight the superior foe makes him satisfy also (d). In short, the general can satisfy (a) to (d) and still despair, doing his duty to his country while awaiting his doom.

Presumably, we will get the same result if we choose Howard-Snyder’s preferred positive cognitive attitude, which is belief-less assumption. It seems perfectly consistent with being in despair regarding p, to assume p for some practical reason. Consider Howard-Snyder’s example of Captain Morgan, the hiker who is bitten by a poisonous snake and stumbles on ‘in the assumption that help lies ahead’ without really believing this to be the case (Howard-Snyder, 2013, 364). The captain need not be brave or faithful for stumbling ahead on that assumption; he might very well despair over being saved.

At this point, an objector might claim that if the captain and the general truly despair, should they not just give up? Do not their actions imply that they do not despair? To this, I would reply that despair is supposed to be a mental state, and as such, it seems to make little sense to claim that it is inconsistent with any action. Since we are creatures capable of acting in a way that does not reveal our feelings or our true assessment of the situation, there is no direct link between actions and cognitive attitudes. Surely one can act and still despair over the prospects of success if one has at least some kind of motivation for carrying on. For example, the general might be motivated by duty or by how he will be remembered after the defeat he envisages. Captain Morgan might despair over being saved but still hope that his corpse will be found and buried rather than eaten by vultures.

I want to make clear that I do not think that choosing any positive cognitive state as (a) yields these results. Belief clearly does not. However, I do think that there are non-doxastic alternatives that can handle this objection as well. One example might be Schellenberg’s ‘imaginative assent’, where the subject persistently pictures the actuality of p to herself (Schellenberg, 2005, 129–139). Another is propositional trust, where the subject trusts that p (presumably on testimony). Both these cognitive attitudes seem inconsistent with despair. The persistent imagining of p as real, or the trusting that p, both suggest mental states that seem to leave no space for despair. However, as the results of the previous section made clear, there are strong reasons to reject trust in an account of rational, non-doxastic faith, since the attitude is hard to reconcile with the demands of reason. Likewise, I have previously argued that Schellenberg’s ‘imaginative assent’ meets similar difficulties (Palmqvist, 2019).

It seems the desiderata point in different directions. To satisfy DH, we need a strong cognitive attitude which leaves no room for despair. However, such strong cognitive attitudes standardly become problematic in the light of DR, since they tend to interfere with the demands of reason. A cognitive attitude strong enough to block out despair is often also strong enough to block out some evidence. To satisfy DR, we should rather go for a weaker attitude, like acceptance, which allows for full appreciation of the evidence. But then our faith will fail to exclude despair as demanded by DH.

This dilemma not only concerns Howard-Snyder’s account. To the contrary, it poses a great challenge for any account of rational, non-doxastic faith. How can we find a middle road here, a positive cognitive attitude strong enough to exclude despair, but one which does not interfere with the demands of reason?

Is It Really Faith I am Looking for?

I want to end this discussion by addressing a major objection to the account of faith my proposed desiderata point towards. The first two desiderata go against some plausible intuitions regarding faith: that faith should imply that one stops looking for further evidence and that faith should imply trust. It therefore seems to be some major differences between the kind of faith I am after, and the common usage of the term. In the light of these differences, I think one might legitimately ask if it is really faith my analysis aims at. Could the attitude suggested by my desiderata ever be called ‘faith’ in more than a stipulative, technical sense of the word?

As I hope became clear in the section "On Analysing Rational, Non-doxastic Faith", I am not aiming at anything like conceptual analysis. I am not trying to find necessary and sufficient conditions for ‘faith’, not even for ‘rational, non-doxastic faith’. Neither do I aim at stipulating a new, technical sense. As pointed out by Timothy Williamson, arbitrary stipulation is not the only alternative to conceptual analysis:

Indiscriminate description of the ordinary use of a term and arbitrary stipulation of a new use are not the only options. We can single out theoretical functions central to the ordinary concept… and ask what serves them. (Williamson, 2000, 194)

My two first desiderata should be understood in line with these methodological remarks. I have singled out a function of non-doxastic faith, namely its compatibility with agnosticism and epistemic rationality, and I asked what serves, or at least preserves, this function. That is the process behind DR and DT.

The method suggested by Williamson lies close to what is commonly known as ‘rational reconstruction’. The basic idea is that in reconstructing a concept for theoretical use, one should aim to strike a balance between being true to the original meaning and working out a definition that is theoretically fruitful.Footnote 20 By combining a desideratum that is clearly in line with ordinary usage (the DH idea that faith should imply hope and be incompatible with despair) with two desiderata serving the rationality function, I aim towards an account along these ‘reconstructive’ lines.

One should also note that my proposed desiderata are not the only result one might reach using a reconstructive method. If one has different priorities, one might wish to preserve other intuitions regarding the original concept, and one might aim for other theoretical values as well. It would be possible to meet the objector’s demand that faith must imply an end to active investigation by rejecting some other common intuition regarding faith. For example, following considerations very much like the ones behind DR, I have previously suggested that we should reject the idea that faith must be an indeterminate, long-term commitment (Palmqvist, 2019).Instead, I argue for ‘occasional faith’, a faith ‘one wears in church, so to speak’. Such faith satisfies DR since it does not imply an indefinite postponement of further inquiry.

I think that it is safe to say that we need a reconstructive methodology if we are to reach an account of rational, non-doxastic faith and that the feasibility of my proposed analysis depends upon it. Personally, I do not perceive this to be a problem, since I agree with Kvanvig that the term ‘faith’ is used in so many different ways that conceptual analysis seems out of the question anyway. However, not everyone will accept the move from conceptual analysis to rational reconstruction. Therefore, I want to end this section by also saying something about how my results should be viewed on conceptual analysis.

Obviously, the desiderata for rational non-doxastic faith will hold the same force on conceptual analysis, and on this approach, the fact that DR and DT conflict with some basic intuitions should presumably make us hesitant to regard such faith as a genuine possibility. In other words, the considerations leading up to DR and DT would constitute compelling arguments against the feasibility of rational, non-doxastic faith. I have two further comments to make here.

First, it is not obvious to me that a conceptual analysis of faith can ever meet DR, even if faith is understood in terms of belief. Belief-based faith is still hard to combine with active investigation if we are to preserve the intuition that one does not look for further evidence regarding that which one accepts in faith. Granted, the problem might be less severe on belief-based faith, but as I argued with respect to Buchak’s account, it will grow over time as faith is held as a long-term commitment. This suggests that it is not non-doxastic faith but faith simpliciter which is problematic in relation to DR.

Secondly, one might ask if the notion of faith is essential to religion, or if there could be non-doxastic religion without faith. If non-doxastic faith is as problematic as I have suggested in the discussion of the desiderata, does it not seem reasonable to explore what alternatives there could be to faith? While most non-doxastic accounts come in terms of faith, there are exceptions, like the early account offered by James Muyskens (1979) which is based on hope rather than faith. Of the considerations behind the desiderata for faith, only the importance of excluding despair from the analysis seems relevant for hope, but that might not be a major problem considering that much work is presently done to improve the analysis of hope. There are also more ‘technical’ attitudes available, such as acceptance. As far as I can see, none of the problems we have encountered for rational faith applies to acceptance. So, if one is willing to give up the basic idea that religiosity must be based on faith, non-doxastic religion might still be available also on conceptual analysis.

Concluding Remarks

I have argued that any account of rational, non-doxastic faith should be able to meet three important desiderata. According to DR, it must not interfere with the subject’s epistemic endeavours. DT requires that it must not include trust since trust is either doxastic or irrational. Finally, following DH, it should imply hope and exclude despair.

Can such faith be had? I am optimistic that it can, but due to the conflict between important intuitions and theoretical demands, it seems to require that we abandon conceptual analysis and adopt a more reconstructive methodology. The important question is how much of our common intuitions such a reconstructed concept of non-doxastic faith will be able to preserve. The latent conflict between DR and DH, where epistemically problematic attitudes seem more well suited to handle despair, might force us to move even further away from common intuitions. If no middle road is forthcoming, we might be forced to abandon DH to keep true to the main ambition of having rational faith.

If we must make such a move, abandoning further intuitions, that would surely actualize the question whether there are any other ways of pursuing a rational reconstruction of rational, non-doxastic faith. Perhaps occasional faith would be a better option, sacrificing intuitions about long-term commitment instead? Or perhaps the best way of combining the demands of rationality with as many of our faith-intuitions as possible in a rational reconstruction of non-doxastic faith remains to be discovered?

If a thorough further investigation were to show what my optimism is misplaced and that no satisfactory rational reconstruction of faith is to be found, another possibility would be to follow Muyskens’s idea that hope might be an alternative to faith, thereby abandoning the notion of faith altogether. Or exchange faith for the more technical notion of acceptance. While such faith-less religion should only be considered a last resort, it would still be able to deliver on the great promise of non-doxasticism, by providing a way of being religious to the intellectually honest agnostic.